Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Intentions do not take care of themselves : attentional and cognitive abilities required for prospective… Jacova, Claudia 2003

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2003-792285.pdf [ 10.4MB ]
JSON: 831-1.0099731.json
JSON-LD: 831-1.0099731-ld.json
RDF/XML (Pretty): 831-1.0099731-rdf.xml
RDF/JSON: 831-1.0099731-rdf.json
Turtle: 831-1.0099731-turtle.txt
N-Triples: 831-1.0099731-rdf-ntriples.txt
Original Record: 831-1.0099731-source.json
Full Text

Full Text

I N T E N T I O N S D O N O T T A K E C A R E O F T H E M S E L V E S : A T T E N T I O N A L A N D C O G N I T I V E A B I L I T I E S R E Q U I R E D F O R P R O S P E C T I V E R E M E M B E R I N G by C L A U D I A J A C O V A Laurea di Dottore, The University of Florence, 1984 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y in T H E F A C U L T Y O F G R A D U A T E S T U D I E S Department of Psychology ,. We accept this thesis as c^u^olrping to^ie r^dj^^^t^ndard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A March 2003 © Claudia Jacova, 2003 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^ ^ ^ d L - c 7 Q - ^ j The University of British Columbia Vancouver, Canada Date rVPvcSH o?) £ o c ? 3 DE-6 (2/88) 11 ABSTRACT The present research focused on event-cued prospective memory (ProM). ProM is becoming aware that some ongoing event signals the need to act upon an intention formed at an earlier time. Two major goals were addressed. One goal was to identify the attentional resources required for processing the ProM event. This goal builds on current theories that understand ProM retrieval as an attention-demanding process. Two experiments manipulated the attentional demands of the ongoing task to selectively constrain resource availability for ProM. The ProM task was to interrupt the ongoing activity if a picture cue was noticed. The cue was embedded in ongoing task trials. In Experiment One, participants performed either a task that required simple speeded responding, or one that required searching for a target letter and matching it to memory. The second task elicited lower levels of ProM than the first. Experiment Two tested the effects of ongoing searching and matching-to-memory separately, by manipulating distractor and target load in a speeded yes/no decision task. Distractor but not target load reduced ProM performance, implicating perceptual search processes in this type of memory. These results were discussed in terms of a link between ProM and attentional selection. The second goal of the present research was to investigate the involvement of top-down thinking processes in ProM. This goal focused on the relationship between ProM and creativity. In Experiment One, performance on two ProM tasks was correlated with performance on two divergent thinking tests. High levels of originality predicted low levels of ProM. In Experiment Two, productions on a verbal divergent thinking test and nonverbal drawing task were scored in terms of originality and flexibility. High levels of originality again predicted low levels of ProM across tasks, but high levels of flexibility predicted high levels of ProM. The interpretation of these results was that thinking in unusual ways may interfere with ProM whereas thinking broadly, in terms of multiple conceptual categories, may facilitate ProM. I l l TABLE OF CONTENTS ABSTRACT ii LIST OF TABLES vi LIST OF FIGURES vii AUTHOR'S NOTE viii ACKNOWLEDGEMENTS ix INTRODUCTION AND OVERVIEW 1 CHAPTER ONE: DEFINING PROSPECTIVE MEMORY 5 A ConosptmlD^mitimqfPn^p^wMerwry 5 A n Operational Definition of Prcspectize Memory. 9 ProspectiiE Memory Paradigms 11 Paradigms in Human Memory Research 11 Paradigms in Animal Memory Research 18 PART I: T H E ATTENTIONAL RESOURCES REQUIRED FOR PROSPECTIVE MEMORY 27 CHAPTER TWO: THEORETICAL MODELS AND EMPIRICAL EVIDENCE 28 Theoretical Models of Prospective Memory Retrieud 29 Effects of Increased Ongoing Task Demands 33 AgeEffects onProspectkeMemory 36 The Neuropsyahobgy of ProspectiiE Memory 42 Aging and Lesion Studies 43 Neuroimaging Studies 45 Electrophysiological Evidence 47 CHAPTER T H R E E : PROCESSING SPEED AND PROCESSING CAPACITY 49 Experiment One. 51 Method 53 Participants and Design 53 Materials 55 Procedure 56 Results 60 Effects of Ongoing Task Type on Prospective Memory Measures 61 Ongoing Task Performance 65 IV Performance on the Retrospective Memory Measure 67 Correlations between Individual Difference Measures and Prospective Memory 68 Discussion 69 CHAPTER FOUR: SEARCHING AND MATCHINCTO-MEMORY 75 Experiment Two 79 Method 81 Participants and Design 81 Materials 81 Procedure 84 Results 88 Effects of Distractor and Target Load on Prospective Memory 89 Ongoing Task Performance 91 Correlations between Individual Difference Measures and Prospective Memory 96 Self-Reports on Prospective Memory Performance 101 Discussion 103 Conclusions to Part 1 110 PART II: PROSPECTIVE MEMORY AND CREATIVITY 115 CHAPTER FIVE: T H E RELATION BETWEEN PROSPECTIVE MEMORY AND CREATIVE THINKING 116 Defining Creative Thinking 119 Predictions about the Rektionship betwen Prospective Memory and Creative Thinking. 123 CHAPTER SIX: INVESTIGATING THE RELATIONSHIP BETWEEN PROSPECTIVE MEMORY AND CREATIVE THINKING 130 Pilot Investigation 131 First Investigation 133 Method 133 Materials 133 Participants and Procedure 134 Results 136 Scoring 136 Test Performance and Test Reliability 139 Correlations between Prospective Memory and Divergent Thinking 140 Discussion 141 Second Investigation 143 Method 145 Materials 145 Participants, Design and Procedure 146 Results 150 Scoring 150 Test Performance and Test Reliability 153 Correlations between Prospective Memory and Creativity Scores 155 Creativity and Self-Reports on Prospective Memory Performance 159 V Discussion 160 Conclusions toPartll. 164 CHAPTER SEVEN: GENERAL CONCLUSIONS 168 Motiiation, Goals and Strategies of This Research 169 Core Findings of This Research 172 Theoretical Implications 175 Methodological Implications 179 L imitations of This Research 182 Suggestions for Future Research 185 Main Contribution of This Research 187 APPENDICES 198 Appendix A: The Story ProM Task 199 Appendix B: The ProspectiiE Memory Questionnaire 201 Appendix C Common, Unusual, and Unique Responses to the Item "Brick ". 202 Appendix D: Flexibility Categories 203 Appendix E: Common and Unusual Responses to the Item "Rope" 205 Table LIST OF TABLES vi Page 3.1 List of tasks and their order of administration in Experiment One 58 3.2 Mean reaction times in ms for performance on the Simple Reaction and Choice Reaction Task as a function of block 64 3.3 Retrospective memory performance (mean number of words recalled) as a function of task type and prospective memory response 64 3.4 Correlations between prospective memory and resource measures by task type .. 68 4.1 Experimental conditions combining the manipulation of ongoing searching and matching-to-memory demands in a target present/absent decision task 80 4.2 List of tasks and their order of administration in Experiment Two 87 4.3 Mean reaction times in ms and decision accuracy in percentage for performance on the ongoing task as a function of block 92 4.4 Frequency of self-reported reasons for failing to respond immediately or at all to the prospective memory cue, by experimental condition and performance 101 6.1 Performance on prospective memory tasks and divergent thinking tasks in the First Investigation 138 6.2 Correlations between divergent thinking indices and with the North American Adult Reading Test (NAART) scores 138 6.3 Correlations between prospective memory and divergent thinking indices 139 6.4 Combinations of creativity tasks and number of subjects who received them 149 6.5 Performance on prospective memory and creativity tasks in the Second Investigation 157 6.6 Correlations between scores on the Uses and Picture Completion test 158 6.7 Correlations between prospective memory and creativity indices 158 vu LIST OF FIGURES Figure Page 3.1 Examples of ongoing and prospective memory displays in Experiment One 57 3.2 Prospective memory performance as a function of ongoing task type 63 4.1 Example of the ongoing task display in Experiment Two 83 4.2 Prospective memory performance as a function of distractor and target load 93 4.3 Mean reaction times on accurate trials as a function of distractor and target load 94 4.4 Scatterplots depicting the relation between mean reaction times and Cue Size at low and high target loads 98 6.1 Picture completion test, Form B, Page 1 148 V1U AUTHOR'S NOTE The title of this dissertation is an adaptation of a quote by Sir Frederic C. Bartlett, "Actions do not take care of themselves" (UnpublishedLecture, Cambridge, England, 1949). Bartlett meant that actions do not just materialize but require our attention at all times. Schonfield (1982) has argued that this is true especially for older people. I have taken the liberty to substitute the word intention for action in the title to state the major point of my thesis. ix ACKNOWLEDGEMENTS Many people have helped me accomplish this work. I am deeply grateful to my supervisor, Peter Graf, who taught me to think with curiosity and rigor, and to always -always- take a second look at things. I owe to him my excitement about prospective memory and my grasp of the puzzles it holds. I am indebted to my Ph.D. Committee members, J im Enns, D o n Wilkie, and Liisa Galea. I thank J im for asking about prospective memory in ways I should have thought of, and D o n for asking about prospective memory in ways I would not have thought of. I thank Liisa for always reminding me that people are women and men. I owe special gratitude to Beat Meier, Todd Woodward, Kevin Peters, Er in Rennie and Noam Butterfield, who did research in the Graf lab when I did. They offered a mix of contributions I will never forget. They shared their own work with me, helped me think about mine, and lifted my spirits with lab-made coffee and humor. I thank Winnie K i , Karen Ip, and Spencer May, for their help in running the experiments, coding and entering the data, their clear thuiking, and their insights into this work. I am grateful for the financial support I received from the National Sciences and Engmeering Research Council of Canada and the University of British Columbia for my Ph.D. studies. Part of this research was made possible thanks to Bob Uttl's expertise in software programming. I thank him for the many hours of work he spent programming and troubleshooting the experimental software. Without the support of my family, this work would not have come into being. I thank my parents for their wisdom and optimism in accepting my endeavors, and offering encouragement exactly at the times I needed it most. I thank my sisters for insisting I should have some fun as weD. I am immensely grateful to my sons Patrizio and Ruben for mspiring me with their innocence and playfulness, and I dedicate this work to them because they grew up with it. Maybe the excitement and satisfaction they shared with me will help them take on challenges in their lives. Finally, I have no way of thanking the one person who went all the way with me, my husband, companion and friend, Giorgio. I can only grant him that prospective forgetting, in his case, is an irresistibly charming personality trait. 1 INTRODUCTION AND OVERVIEW When we trunk about memory, we contemplate the fragile power (Daniel Schacter's expression; Schacter 1996) of our mind's ability to reach back in time and re-experience things past. We have in mind great literary works, such as Marcel Proust's In Search of Lost Time, or Giacomo Leopardi's Remembrances. We are aware of milestone discoveries in memory research over the last century. For example, we know that memory for past events is shaped by what we know (Bartlett, 1932), and by how we experience the event in the first place (Craik & Lockhart, 1972). We also know that our recollection of the past can be inaccurate (Roediger & McDermott, 1993) and that the past can affect us in ways that are not consciously experienced (Graf & Schacter, 1985). Yet, something is missing in this picture. When we are asked to report an instance of memory failure, we often trunk about a type of forgetting that has little in common with an unsuccessful quest for the past. We are likely instead to describe our failure to follow up on an intention or a promise (cf. Einstein ScMcDaniel, 1996). In recent years, memory research has begun to recognize this gap and to distinguish the ability to remember to fulfill an intention from the ability to remember information about the past (cf. Brandimonte, Einstein, &McDaniel, 1996). The former is known as prospective memory (ProM), the latter as retrospective memory (RetM). An example of ProM is to become aware that now is the time to call a friend as intended earlier. An example of RetM is to recall the telephone number of the friend. ProM is as critical to the tasks of everyday life as is RetM. Caring for others, mamtaining a household, honoring professional commitments and obligations, all entail forming intentions and delaying their execution to a later time when they are remembered and the planned action put into practice. Despite its prominent role in personal and social functioning, ProM has become the object of systematic research only in the past two decades (Brandimonte et al., 1996). From the start, two related areas of enquiry have been at the focus of interest: ProM's fate with normal aging and 2 ProM's attentional demands (Craik, 1986; Moscovitch, 1982; Schonfield, 1982). These two areas were the object of exploration and discovery for the present research. The present research had two goals. The first goal was to detennine which attentional resources must be recruited for ProM. The pursuit of this research goal was motivated by the existence of age-associated declines in ProM (e.g., Jacova, Meier & Graf, 2000; Maylor, Darby, Smith, Delia Sala & Logie, in press; Uttl, Graf, Miller & Tuokko, 2001) and by the theoretically based expectation that age declines in ProM result from a depletion of the resources the cognitive system can make available for processing (Craik, 1986). The present research asked about the type cf resources that the system must make available for ProM. Its aim was to find out what these resources are and by what mechanism they enable ProM. The starting point was the resource notions of processing speed and processing capacity, developed to explain age declines in RetM (Graf & Uttl, 1995; Kausler, 1994). The choice of this starting point does not mean that the present research considers ProM a form or manifestation of RetM. On the contrary, as will be clear throughout the dissertation, ProM is viewed as a distinct type of mnemonic activity, characterized by an attentional configuration that shares little with typical RetM situations (cf. Graf & Uttl, 2001). The second goal of the present research was to investigate ProM's relationship with creativity. This goal was motivated directly by the recognition that ProM poses unique cognitive challenges at retrieval that must rely to some extent on top-down processes of categorizing and problem-solving. The rememberer must discover in some aspect of the current context -a certain event or clock time- the cue to bring back to mind a plan formed at an earlier time. The cue is, as it were, embedded in the "wrong" context -the ongoing situation. To realign it with its "right" context -a plan or intention- some kind of reinterpretation or redefinition must occur. To illustrate, assume the task is to convey a message to our spouse when he/she comes home. At the time our spouse steps into the house, we are straggling with a computer problem. Our mental context of that moment is computer troubleshooting and in this context our spouse represents a potential problem-3 solver. He/she maybe many other things, for example a good cook, a fretful traveler, a movie-lover, and critically, a person with friends and colleagues who leave messages, but we don't think of those things in that moment. How do we make the leap from computer trouble-shooting to friendships and messages? Could it be that we need to have an open mind, see alternatives, in short, consider the situation creatively? The question addressed in the present work was whether the conceptual leap required for ProM is similar to, or different from, creative thinking. The dissertation will begin with a conceptual and operational definition of ProM, and an analysis of existing ProM paradigms in Chapter One. It will then address the two goals of the present research in two separate parts. Part I will focus on ProM's attentional resource requirements. Chapter Two will review the theoretical framework that guided this research, and the existing empirical evidence on ProM and attentional resources. Chapters Three and Four will outline the rationale, method and results of the two experiments designed to investigate ProM's resource requirements. In both experiments the attentional demands of the ongoing task -the activity participants were engaged in at the time the ProM cue event occurred- were manipulated to selectively limit resource availability for ProM. Experiment One, described in Chapter Three, examined the effects that ongoing tasks requiring processing speed or processing capacity had on ProM performance. Experiment Two, described in Chapter Four, was designed to develop the finding that ongoing demands on processing capacity were more damaging for ProM than ongoing demands on processing speed. The manipulation targeted ongoing task demands on two separate processes: the perceptual process of searching through stimuli, and the working-memory process of matching stimuli to targets in memory. Chapter Four will include a final section in which the implications of the findings from both experiments are examined. Part II will focus on the relationship between ProM and creativity. Chapter Five will present an analysis of ProM's cognitive features, briefly review existing conceptualizations of creativity, and evaluate theoretical arguments and empirical evidence supporting or opposing a similarity between 4 ProM and creativity. Chapter Six will describe the two investigations conducted on the relationship between ProM and creativity. The first experiment examined performance on two verbal divergent thinking tasks, and the relationship between performance on these tasks and prospective memory performance. The second experiment aimed at replicating the finding of a systematic relationship between creative thinking and ProM, and at extending this finding to different measures of creativity and ProM. In the final section of Chapter Six an interpretation of the observed relationship between creative thinking and ProM will be proposed. The dissertation will conclude with Chapter Seven, in which the goals and key findings of the present research will be reviewed in the general context of research and theorizing on ProM. CHAPTER ONE: DEFINING PROSPECTIVE M E M O R Y In a special issue of Applied Cognitiw Psychology devoted entirely to ProM, the editors define ProM as "... the skills required to support the fidfillment of an intention to perform a specific action in the future" (Ellis & Kvavilashvili, 2000, p. 1). This is a broad definition suited to the scope of the works published in the issue. It allows latitude with regard to such critical questions such as: What skill domain(s) does ProM involve? What kind of remembering occurs in ProM, how is it different from RetM? Do all situations of intention futfillnient require ProM? For the purposes of the present work, I will develop a more narrow definition of ProM in the next paragraphs by using these questions as signposts. A Conceptual Definition of Prospective Memory What skill domain does ProM involve? Is ProM memory or some other skill? Dobbs and Reeves (1996) explicitly state that ProM is not a type of memory. Rather it is a multidimensional task that recruits planning, monitoring, and action control skills. For Ellis (1996) ProM is a process that supports the realization of delayed intentions and their associated actions. Burgess, Veitch, Costello and Shallice (2000) indicate ProM as one of the cognitive processes underlying the ability to coordinate and execute multiple tasks (multitasking). Winograd (1988) characterizes ProM as one of the ways in which the organism regulates goals and actions in response to the temporal contingencies of the environment. By contrast to these views, in the present work, ProM is strictly understood as memory because its core process is the reinstatement of an earlier mental state -an intention or a plan. ProM tasks are structurally equal to other memory tasks, with a study, retention, and test phase, the latter involving the presentation of cues. It has been recognized for some time that futfilling an intended action involves two distinct forms of remembering. One is remembering that one intended to do something. The other is remembering zehat the intended action was (Ellis, 1996; Kvavilashvili, 1987). The two forms of 6 remembering are referred to in the ProM literature as the prospective and the retrospective component of a ProM task (Einstein &McDaniel, 1996; Graf & Uttl, 2001; but see Ellis, 1996, for a different use of the two terms). The prospective component marks a key difference between RetM and ProM tasks at the time of testing. In RetM tasks, subjects are instructed to work with the task-relevant cues. To perform successfully, they must utilize the cues to retrieve the requested information from memory. By contrast, in ProM tasks, subjects are instructed to perform an ongoing activity-an activity that is not the ProM task. Subjects are not alerted to the ProM-relevant cues that are presented in the course of the ongoing activity. To perform successfully, they must discover the cues, that is, recognize them as a signal to recall a previously formed intention. Once the cue discovery is achieved, the task becomes retrospective, with subjects retrieving the intended action. Thus the prospective component ends with the discovery of the cue, whereas the retrospective component begins with it (Graf & Uttl, 2001). To illustrate, subjects in an experiment might be instructed to date the page when they are asked to copy a certain geometric figure Qacova et al., 2001). When this figure is presented to them at a later time, they must become aware that it is the cue to act upon the received instructions -the prospective part of the task. If they do, a memory search for the assigned activity -the retrospective part- will begin. If they don't become aware, no search will take place, although they might be perfectly able to recollect the assigned activity if prompted by the experimenter. The present research is concerned with ProM as a construct, and therefore focuses on the prospective component as the distinctive and unique feature of a ProM task. Accordingly, ProM is defined as remembering (becoming aware) in the designated context that one intended to do something. This type of remembering is further defined as an act of discovery (Uttl et al., 2001). This definition of ProM is self-contained and does not require nor include, remembering what one intended to do. 7 A final critical issue regards whether P r o M is required in all situations where an intention to carry out an action needs to be fulfilled. There is growing consensus in the P r o M literature that intention fulfillment is a P r o M activity only when there is a delay between the formation of the intention and the opportunity to fulfill it (Burgess, Quayle, & Frith, 2001; Ellis & Kvavilashvili, 2000). Graf and Uttl (2001) propose to distinguish between vigilance and P r o M proper as the P r o M equivalents of short- and long-term RetM. They suggest that what best differentiates these activities is the subjective, conscious stance of the subject (see also Meier & Graf, 2000). In vigilance, the subject actively maintains the intention in conscious awareness and directs all or most attentional resources to it. A t the time of retrieval, awareness is dominated by the expectation of the cue to initiate the intended activity. A n example might be waiting for the water in the kettle to boil. In P r o M proper, the subject stops thinking about the intention at some point after its formation, and directs attentional resources to ongoing activities. In this situation, at the time of retrieval awareness is dominated by ongoing activities and ongoing expectations. There is no conscious expectation of the cue to initiate the intended activity. A n example might be remembering to purchase a new kettle the next time one goes shopping. It is clear that cue recognition is a very different cognitive process in these two situations. In vigilance situations, cue recognition can be compared to target search and detection. In P r o M proper situations, by contrast, cue recognition is best described as serendipitous discovery. In this work, P r o M is identified with P r o M proper or long-term ProM. Cue discovery characterizes episodic but not habitual long-term P r o M situations. Episodic situations refer to intentions to do something on one specific occasion, outside the routine structure of activities, for example picking up one's spouse after work because the car failed him/her. In such cases, some normal everyday context, such as leaving one's own work, must be recognized as a cue to do something non-normal. B y contrast, habitual tasks, such as brushing one's teeth in the morning, or checking one's mailbox upon arrival at the workplace, are integrated in routinely 8 occurring sequences of actions, or scripts (cf. Abelson, 1981). They are part of an individual's generic knowledge about actions normally performed in certain situations. Their activation occurs in the context of a well-learned chain of events rather than as a result of recognizing telltale cues. Episodic and habitual tasks involve very different mnemonic processes that can be linked to the distinction between episodic and semantic retrospective memory (Graf & Uttl, 2001). Thus, habitual remembering is excluded from the definition of ProM in this work. In sum, I use the term ProM in this work to refer to memory that supports the fulfillment of one-off or non-routine intentions that are not in conscious awareness at the time of retrieval. Its contribution to intention fulfillment is the discovery of cues in the ongoing context, that signal the need to bring back to mind the intended action. This contribution does not comprise the retrospective component of recollecting the particular action that was intended, nor does it comprise the actual execution of that action. Throughout this work, I will employ the term ProM with this meaning, and refer to other situations of delayed intention fulfillment with different labels, for example short-term ProM for situations characterized by vigilance. The preceding definition matches our intuitive sense of what ProM is in everyday situations. On a given day we might plan a number of things that cannot be carried out until later in the day, for example returning a book to the library when we leave the house in the afternoon. We might make an entry in our day planner but we know it will not ensure remembering. Then, inevitably, we turn our minds to ongoing tasks until the time comes to get ready to leave the house. Importantly, we expect that other things, not our plan, will be on our minds at the time we ought to remember, and we all have mtimate experience of the unique challenge of having to "remember to remember" (Craik, 1986). We might try to help ourselves by introducing conspicuous perceptual changes in our environment, for example placing the book beside the door, or sticking a post-it note onto the car keys. If, when we are in the process of leaving the house, we become aware that there was 9 something we meant to do, we can consult our planner, knowing that the most difficult part of the task is over. An Operational Definition of Prospective Memory ProM is elicited when subjects are given the task to remember to do something at a later time upon encountering a specific cue, and when, in the time interval between instructions and cue presentation, they stop thinking about the upcoming task. ProM is successful when subjects indicate at the appropriate time their awareness that they must perform the assigned task. However, the translation into exact operational terms is far from simple. One problem is that it is very difficult to operationalize the subjects' subjective state of not thinking about the task that lies ahead. Few ProM researchers have explicitly drawn attention to the need for a clear procedure to ensure this critical feature that differentiates ProM from vigilance tasks (Ellis & Kvavilashvili, 2000; Graf & Uttl, 2001). I suggest in this work that ProM, rather than vigilance, is involved if some time elapses between instructions and test, though no exact duration can define a ProM task as such (Graf & Uttl, 2001), and if subjects are actively engaged in activities other than the intended activity until and at the time the cue is presented (Ellis & Kvavilashvili, 2000). A second problem regards the defining features of the ProM cue. The literature distinguishes between time-cued and event-cued ProM tasks (Einstein ScMcDaniel, 1996; Kvavilashvili & Ellis, 1996). In time-cued tasks, the retrieval circumstances are defined by a certain clock time or by the passage of a certain amount of time. In event-cued tasks, the retrieval circumstances are defined by the occurrence of an event. The focus of the present work was on event cues. Does the distinction between event and time cues matter in the operationalization of ProM? Some researchers have argued that event cues provide external support for retrieval, whereas time cues do not (e.g., Einstein & McDaniel, 1996). According to the definition developed for this work, the critical feature of the ProM cue is its embeddedness, that is, its seamless connection to the ongoing context. The ProM cue needs discovering, and it is argued here that this need exists regardless whether the cue is 10 an event or a clock time. To illustrate, consider the event- and time-based ProM tasks used by Mayior et al. (in press). In the event-based task, the cue was the appearance of animals in a film on the city of Aberdeen, viewed on a television screen, and the task was to say "animal" aloud whenever a cue occurred. In the time-based task, the cue was the passage of time displayed on a computer screen positioned next to the television playing the film, and the task was to press the spacebar on the keyboard whenever three minutes had elapsed. In both cases the cue was a natural part of the ongoing simation and did not explicitly prompt retrieval of the ProM task. There are cues that do not fit the present definition of ProM. These cues either have attention-capmring properties, thereby prompting processing of their significance in the ongoing situation, or they directly prompt retrospective recall of the intended action. An example of the former is the ring of a timer in the Appointment task of the Rivermead Behavioral Memory Test (RBMT; Wilson, Cockburn, Baddeley & Hiorns, 1989). The subject must remember to ask the experimenter for an appointment when the ring occurs. This type of signal does not appear a true ProM cue because it does not naturally fit into an ongoing situation. The ring diverts the subject's attention from ongoing activities and prompts questions about how to respond to it. An example of a cue that prompts retro- rather than prospective recall will be discussed more in detail in the next section. A third problem regards the measurement of ProM. How can ProM be measured if its definition does not include the execution of the planned activity, if indeed its only content is the awareness, upon encountering a specified cue, that something needs to be done (cf. Baddeley, 1990; Winograd, 1988)? As will be seen in the next paragraph, some experimental ProM paradigms address this problem by requiring extremely simple types of action upon encountering the cue (e.g., Einstein &McDaniel, 1990) whereas others attempt to measure ProM independently of RetM. Graf and Uttl (2001) argue that a distinct and directly measurable ProM response exists. I identify this response herein as reacting to the cue in some manner that interrupts the ongoing actizity. Accordingly, when the cue is 11 presented, stopping an ongoing action, mrning away from the location of the ongoing activity, for example the computer screen, saying "Oh yes, I need to do something now" or asking the experimenter "What is it that you wanted me to do now?", are distinct and identifiable ProM responses. Prospective Memory Paradigms What methodological implications does the present definition of ProM have? Do existing paradigms investigate ProM as it is defined here or other skills involved in intention fulfillment, and how do the former differ from the latter? Which paradigm is best suited for measuring ProM? To address these questions this section will review paradigms that have been employed in human and animal memory research to investigate memory supporting the fulfillment of intentions. This section has three purposes. One is to identify the core attributes of paradigms that measure ProM. The second purpose is to describe the paradigm selected for the present research and compare it with other ProM paradigms. The third is to analyze animal paradigms and to relate these to the definitional distinctions developed in the previous sections. To my knowledge, the possibility of links between human and animal behavior has not yet been explored in the area of ProM research and theorization (cf. Brandimonte et al, 1996; Ellis & Kvavilashvili, 2000). Paradigms in H u m a n Memory Research Some investigations, particularly early ones, have used a field paradigm in which subjects are asked to perform some action at specified future times in actual-life settings. Meacham and Leiman (1982) gave subjects a number of stamped postcards to take home. The instructions were to mail the cards back on designated dates over a 32-day period. When mailing the cards, subjects were required to write the date and time on them To evaluate performance, the postmark and the date written on the card were considered. Rendell and Craik (2000) gave subjects ten tasks on each of seven successive days. Subjects received a package with task sheets for each day, and were instructed to read each sheet at the beginning of the appropriate day. Tasks included taking medication at 11 a.m. 12 or checking on the supply of butter when first opening the fridge in the afternoon or evening. Subjects did not actually perform the tasks but recorded time-stamped messages on a micro-recorder, for example "It is 11 a.m., and I have taken my medication". Responses were considered correct when they were on time and had the correct content. Field studies capture ProM as it is defined for the present work, that is, as regaining awareness of an intention upon discovering a cue, to some extent because subjects cannot afford to keep thinking about the upcoming task(s) over an interval that may span hours, days, or even weeks. However these studies do not achieve control over important variables, such as the strategy that subjects use for remembering or the type of activity they are engaged in when the time comes to remember. Subjects might reduce the demands on ProM by setting up external aids or by tying the task to a habitual activity (Maylor, 1996). Also, subjects might plan less demanding activities around the time when they anticipate the need for remembering. Importantly, in this paradigm it is very difficult to measure ProM separately from the retrospective aspects of the task. In Meacham and Lehman's study, the postcard mailed on the correct day indicates successful ProM and successful RetM. Rendell and Craik tried to obtain a more direct measure of ProM by instructing subjects to record a message even when they only knew it was time to do something but failed to remember the content (for example, "It is 10 a.m.; I know I have to do something but can't remember what it is"). However, to meet this request subjects still had to retrieve retrospective information such as the instruction to make this type of entry or instructions on how to operate the micro-recorder. Three main strategies have been employed to investigate ProM in the laboratory. Some researchers have embedded naturalistic ProM tasks in interview or testing sessions (Huppert, Johnson &Nickson, 2000; Kvavilashvili, 1987; Mantyla, 1994; West, 1988). In this paradigm, the general procedure is to instruct subjects to perform a namralistic type of activity at a specified later time in the session, for example upon completing a certain test, or upon leaving a room. Uttl et al. (2001) gave subjects three ProM tasks in a 2-hour test battery. Each task was nestled around an 13 attention test (for example the cancel-H task), with instructions to remember to carry out a simple action when the experimenter said, "This is the end of the task", provided prior to the attention task, and the cue delivered upon its completion. The ProM tasks required different types of action: for example, subjects had to request a pen and a piece of paper, and write their name on the paper. In this study three types of scores were computed. The task scores reflected whether subjects responded to the cue in any ProM-relevant manner, for example by saying, "Oh, there is something I have to do now". The item and activities scores reflected whether subjects correctly recollected the items to-be-requested for each task (e.g., pen and paper) and the action to be carried out (e.g., write one's name), respectively. The task score indexes success or failure on the prospective part of the task, whereas the item and activities scores, when conditionalized on the task score, index success or failure on the retrospective part. Einstein and McDaniel (1990) developed a laboratory paradigm that has been widely used in ProM research over the past decade (e.g. Brandimonte & Passolunghi, 1994; Kvavilashvili, Kombrot, Mash, Cockburn & Milne, 2000; Marsh & Hicks, 1998). The basic strategy consists of instructing subjects some time prior to the beginning of the ProM test portion of the experiment that if they happen to see a particular word, for example "rake", they should press a response key on the computer keyboard. The word then appears several times, embedded in an ongoing activity concurrent to the ProM task. In the original experiment, following the ProM instructions Einstein and McDaniel engaged subjects for approximately 15 min in word recall and word recognition tests. Then, without reminding subjects of the ProM instructions, they administered the short-term memory test in which the cue word was embedded. Each of the 42 short-term memory trials consisted of the simultaneous presentation of 4 to 9 words, followed by the prompt to recall the words in their original order. The cue word appeared on three trials, spaced evenly across the test. The measure of ProM was the number of times subjects remembered to make a key press in response to the cue word. 14 Graf and Uttl (Graf & Uttl, 1999; Uttl & Graf, 2000) designed a different laboratory approach to testing ProM, which has been used for the present research. In this approach, subjects are shown a picture cue some time prior to the ProM test portion of the experiment. They are told that if at anytime in the experiment they notice this picture again, they must interrupt any ongoing activity and recall a list of words. The critical part of the experiment begins after a few filler tasks, with a block of trials of an attention-demanding ongoing task. In this task subjects must respond in some specified way to letter displays presented on a computer screen, for example by sorting them into displays including the letter 'A ' and displays including the letter 'B' . In the second block of trials, pictures of common objects are presented on each ongoing task trial in the four comers of the screen. In the third block, the ProM cue is presented repeatedly, at increasing sizes, among the pictures, until the subject responds or until the ongoing trials are completed. ProM in this paradigm is indexed by two different outcomes: whether or not the subject responded to the cue by interrupting the ongoing task, and which cue size was required to make this interruption. The retrospective part of the task -recalling the word list— does not contribute to these indexes and is assessed independently. Al l three laboratory-based paradigms investigate the type of memory that has been defined as ProM in the present work, that is, becoming aware of an intention of which meanwhile one has not been thinking, upon discovering cues in the ongoing context. It can be presumed that as subjects in the experimental session are presented with new ProM-irrelevant activities and situations, they stop thinking about or expecting to perform the ProM task. The cues are part of these new activities and situations, and they do not perse prompt any ProM-relevant thought or action. ProM is either measured directly, for example by computing task versus item and activity scores, or it is inferred from an action that poses minimal demands on RetM, for example a keypress. \vriile these paradigms all satisfy the operational criteria for assessing ProM, there are important differences. The naturalistic paradigm does not embed the cues in an ongoing task but 15 rather in transitions between tasks, to which subjects may respond in different ways. Some subjects may take transitions as an opportunity to engage in their own thoughts, some to initiate interactions with the experimenter, and others might simply wait for the next instructions. These responses might have very different effects on ProM. By contrast, the non-naturalistic paradigms embed the cues in a specified ongoing task, thereby ensuring experimental control over the type of activity subjects are engaged in. This is a crucial requirement for many types of research on ProM, including the present research on ProM's resource demands. Both non-naturalistic paradigms use cue repetition as a method for testing ProM but their rationale is different. In the Einstein and McDaniel paradigm each cue presentation represents one ProM task -pressing a key each time a certain word is encountered. However, repeated ProM tasks may not measure the same ability. Maylor (1996) has suggested that the first successful trial relies on ProM processes but later trials may involve RetM or working memory abilities. Therefore, for the measurement of ProM it appears much more important to measure what happens prior to the first successful ProM response than what happens after it, and the Graf and Uttl paradigm does exactly this. In this paradigm each cue repetition is a trigger for a single ProM task -stopping and recollecting a list of words when a certain picture is noticed. The size of the cue is increased on subsequent presentations to render the picture more noticeable and intrusive. In this way a measure of ProM is obtained that specifies how many increasingly large cue presentations are needed to bring back into awareness the ProM task. This method does not aim to make absolute statements about critical cue sizes as size is confounded with repetition. It only makes the general assumption that smaller cue sizes, hence fewer presentations, index higher levels of ProM, and that two subjects requiring the same cue size, hence the same number of presentations, for the correct response have an equal level of ProM performance. Some paradigms measure skills involved in delayed intention fulfillment that differ from ProM as defined in the present work. Shallice and Burgess investigated intention realization in a 16 paradigm aimed to test a group of skills referred to as strategy application or multitasking (Burgess & Shallice, 1997; Shallice & Burgess, 1991). This paradigm is known as the Six Elements Test or the Greenwich Test. Variants of this procedure have been employed in recent ProM research with neuropsychological groups and with older adults (Burgess, Veitch, Costello & Shallice, 2000; Martin, Kliegel, McDaniel & Einstein, 2000). The basic procedure is to instruct subjects that they should perform a number of simple open-ended subtasks in a fixed amount of time (e.g., 15 min) not long enough to complete all subtasks, and that their goal should be to maximize their overall score on the test. Subjects are given a set of rules they must follow, the most important of which states that earlier items on each subtask will be given more points than later ones. Therefore optimal test performance requires that subtasks be switched often. Subjects make a provisional plan of how they will go about the test, and then immediately set out to work on the subtasks. The outcome measure assumed to index ProM (Burgess et al, 2000) is the number of subtasks attempted during the test. To perform this test successfully, subjects must generate and implement a strategy of frequent task switching. They must monitor the switching requirements as they perform the subtasks. In other words, the ProM task of switching activities is a requirement built into test performance. Subjects must divide their resources between ongoing and intended activities, which has been earlier identified as the hallmark of short-term ProM. Shallice and Burgess (1991) reported a much smaller number of attempted subtasks for three patients with lesions in the prefrontal structures than for normal controls. It is not likely that this outcome reflects low levels of ProM. Rather, it suggests poor attention or executive control skills. One strength of this test worth noting for the present discussion, however, is that it is the only one, among all laboratory based paradigms, in which subjects are not given a ProM request but generate by themselves a plan for future action, upon receiving a general performance rule. Plan generation is certainly an important component of spontaneous, real-life ProM, and its minimal contribution in laboratory tests could affect ProM performance considerably (see Kvavilashvili & Ellis, 1996, for similar suggestions). 17 Einstein, McDaniel, Smith and Shaw (1998) designed a laboratory task for examining habitual ProM. Participants performed 11 tasks involving a variety of skills, such as perceptual speed, in the experimental session. The habitual ProM task was to remember to press a designated key on the computer keyboard once some time after the start of each task. The core features of this ProM task were that the action to be carried out and the cue designated to trigger it were the same across many similar trials. This can be compared to remembering to take a medication at each meal. The overall performance rate of the young and older subjects in this study did not significantly change from the initial to the later trials but the error pattern did. Forgetting (failing to respond because one does not remember to act) decreased over trials but both errors of omission (failing to respond because one thinks one has done so) and commission (repeated responding because one forgets one has already done so) increased in later trials, especially for the older subjects. These findings suggest that habitual ProM poses challenges that are very different from those associated with episodic ProM. In habitual tasks, the intended action is part of a regularly encountered context. It is often thought about and its activation may occur outside conscious control. The key problem is monitoring actions, not remembering to act. Forgetting can be a failure of ProM but it just as likely can be a failure of RetM. The Rivermead Behavioral Memory Test (Wilson et al., 1989) includes three tasks designed to measure ProM. The designation of one of these tasks, viz. "remembering to deliver a message", as a ProM task has recently been questioned (Cockburn, Keene & Hope, 2000; Maylor, 1996). This task is embedded in the RetM task of "remembering a new route". The testing procedure is as follows: the experimenter instructs the subject to watch her while she walks around the room, and then to do "the same thing" at a later time. The experimenter then traces a certain path through the room, for example from the chair to the table, to the window, back to the table, etc. At the start of the route the experimenter picks up an envelope, and then drops it off in a different spot at the end of the route. After a 10-min interval, the experimenter asks the subject to trace the same path. 18 Picking up and dropping off the envelope is considered a measure of ProM, whereas tracing each stage of the route in the correct order is considered a measure of RetM (Cockburn, 1993; Huppert & Beardsall, 1993). However, there is no reason why the experimenter's request should elicit different types of memory-RetM for the various parts of the route and ProM for the delivery of the message. It is more logical to assume that the experimenter's request prompts retrospective recall of all actions involved (see Cockburn, Keene & Hope, 2000, and Maylor 1996, for similar arguments). The previous review reveals the core dimensions on which tests of ProM differ from other tests of intention realization. (1) Tests of ProM involve a filled interval of ProM-unrelated activities between instructions and test. (2) In tests of ProM, the ongoing activity during cue presentation is ProM-unrelated; it does not require the ProM task for its performance. (3) Tests of ProM utilize cues that are part of the ongoing context. The cues do not explicitly prompt a memory search. (4) In tests of ProM repeated cues are a good method for measuring a single ProM response. Repeated cues probing multiple ProM responses may measure other abilities, particularly habitual memory. Paradigms in Animal Memory Research "In the time run through by a course of action focussed upon a final consummatory event, opportunity is given for instinct, with its germ of memory however rudimentary and its germ of anticipation however slight, to evolve under selection, that mental extension of the present backward into the past and forward into the future which in the highest animals forms the prerogative of more developed mind." Charles S. Sherrington, ThelntegratiwAakncftbeNenaus System, 1906, p. 332 Sherrington speculated that biological selection would favor the emergence of brain structures and cognitive abilities that would allow two types of mental representations. One is the ability to form representations of past events and experiences, in short some variety of RetM. The other is the ability to form anticipations of future events and behaviors, and this can be roughly assimilated to some type of ProM. Sherrington's idea implies that prospective functions of memory are evolutionary as old as retrospective functions and that both functions contribute to guide and regulate animal behavior. For the purposes of the present section, the assumption is that some ProM 19 abilities should exist in animals, including, possibly, long-term episodic ProM, defined in this work as the ability to regain awareness of an earlier intention upon encountering a designated cue event. The purpose of this section is to examine the behavioral tests that may have probed these abilities in animals, and to evaluate what exactly they measure. To anticipate, these tests have largely captured animals' ability to form memories for future events or actions, and recollect these memories when prompted by the circumstances. As has been seen in the previous section, this involves RetM, not ProM. The issue I address in this section thus is not which type of ProM activity is measured, as was the case for the human ProM paradigms, but whether some ProM activity is involved at all, and how, by which means, ProM might be revealed in animals. When searching the literature on animal memory and cognition with the keywords "prospective memory", "prospective coding", or "memory for actions", one finds only a handful of empirical research articles (e.g., Cook, Riley & Brown, 1985) and sections of book chapters (e.g., Honig &Dodd, 1986; Roberts, 1998). Additional search with keywords corresponding to the paradigms that have been employed, for example delayed matching, yields a few more works (e.g., Honig & Wasserman, 1981). These search results indicate that research on animal prospective memory, also referred to as prospection, began in the 1980s as an expansion of research and theories on animal working memory. They also indicate that research on animal prospective memory has neither started nor developed in parallel to research on human prospective memory. Early theories of ariimal working memory assumed that animals tackled tasks involving a delay between a stimulus and the opportunity to respond to it by using retrospective information. In the 1980s animal researchers became interested in prospective working memory as an alternative means by which animals may mediate delays (Roberts, 1998). For example, a pigeon might be trained to perform a symbolic delayed matching task, in which the animal is presented with an initial or sample stimulus and, after a delay, with a comparison or test stimulus. The pigeon might leam that if the sample stimulus is a horizontal line, the green comparison stimulus must be pecked. The 20 animal can do one of two things. It can keep in working memory the sample stimulus "horizontal line" across the delay, and retrieve the relevant rule "peck green when line horizontal" from reference memory when the comparison stimuli appear. Alternatively it can retrieve the rule "peck green when line horizontal" from reference memory upon seeing the sample stimulus and retain only the response "peck green" across the delay. The former represents a retrospective strategy in which the delay is mediated by what the animal has seen. The latter represents a prospective strategy in which the delay is mediated by what the animal anticipates doing. Do animals use the prospective strategy? Honig and Wasserman (1981) addressed this question by training pigeons in two different delayed discrimination tasks that involved a go/no go test. In both tasks, a color sample (red or green) stimulus was followed by a line orientation comparison (vertical or horizontal) stimulus. One group of pigeons was trained to perform delayed simple discrimination. In this task, if a trial began with a red sample stimulus, pecking on the comparison stimulus always led to reward, regardless of line orientation. If a trial began with a green sample stimulus, pecking on the comparison stimulus never led to reward, regardless of line orientation. The other group of pigeons was trained in delayed conditional discrimination. In this task, if the trial began with a red sample stimulus, pecking on the vertical line comparison stimulus was rewarded; pecking on the horizontal line was not. Conversely, if the trial began with a green sample stimulus, pecking on the horizontal line was rewarded; pecking on the vertical line was not. The response rule in conditional discrimination thus is more complex than the response rule in simple discrimination. When both groups of animals had learned to respond accurately, delays from 0 to 25 sec were introduced between sample and comparison stimuli. The animals in the delayed simple discrimination condition displayed almost no forgetting whereas the animals in the delayed conditional discrimination condition showed increasing forgetting as retention intervals became longer. 21 On the assumption that delays of the type used in this experiment affect the ^formation kept in working but not in reference memory (cf. Roberts, 1998), this partem of forgetting provides an important clue about what exactly the two groups of pigeons held in working memory across the delay. If the pigeons in both groups held the color sample they observed (red or green), that is, if they used a retrospective code, the working memory load would have been the same for them Both groups would have then had to retrieve the relevant rule from reference memory when the comparison was presented, and this might have given rise to an overall worse performance but importantly, not to increases in forgetting as a function of delay, in the conditional discrimination group. If, on the other hand, both groups of animals held in working memory the response to-be-made, that is, they used a prospective code, the working memory load would have been greater for the pigeons that performed the conditional discrimination task. Upon seeing the sample, they would have had to rehearse "peck vertical but don't peck horizontal" or "peck horizontal but don't peck vertical", whereas the pigeons performing the simple discrimination task would only have to rehearse "peck" or "don't peck". The greater working memory load of the conditional discrimination group would have entailed a greater loss of information as the length of the delay increased. While Honig and Wasserman used two different tasks to infer the use of the memory strategy, Cook, Brown and Riley (1985) used a single task and introduced a delay after a variable number of trials. They gave rats the radial maze task with a free-choice test. In this task, the animals have to collect food from each of several arms (in this experiment there were twelve). Once an arm has been visited, food is no longer available in it. In this situation, rats try to avoid already visited arms and seek out new arms until they have visited all arms in the maze. To perform this task, rats that are familiar with the layout of the maze, can use a retrospective strategy, that is, generate a memory of the arms they have already visited, or a prospective strategy, that is, generate a memory of the arms still to be visited. • • , 22 Cook and colleagues found evidence that the rats used both codes at different times of the sequence. They removed the animals from the maze for a 15-min delay interpolated in the sequence after visits to two, four, six, eight or ten arms respectively, and then returned them to the task. They observed that the rats' performance decreased from a delay after two visits to a delay after six visits. In other words, rats that were taken out of the maze after having visited two arms, were able to make more correct choices after the delay than rats who were removed after having visited six arms. This trend however reversed when rats were removed after having visited more than six arms. Rats who were removed after having visited eight arms, made fewer correct choices than rats who were removed after having visited ten arms. The authors interpret these results as evidence that the animals switched their memory strategy from retro- to prospective as they progressed through the maze. During the initial trials of the task, they utilized a representation of what they had already done to proceed, whereas in the later trials, they relied on a representation of what they still needed to do. In other words, when rats had already made six or more visits, they tackled the remainder of the task by evoking at each trial the arms still to-be-visited. Had they continued to use a retrospective code, their performance would have dropped steadily as a function of the number of arms visited prior to the delay. Do these studies gauge some type of ProM ability? They show that when animals acquire information about how to respond in the near future, they are able to convert this information into a representation of what to do next. In this sense animals can be said to form response intentions (Honig &Dodd, 1986). If animals maintain response intentions in working memory until they can be enacted, might their activity be analogous to human short-term ProM or vigilance as defined earlier in this chapter? Assume, as a parallel to the maze task for the rats, the task of running a list of errands in a shopping mall. If we guide our navigation through the shopping mall by means of the list of things we still need to purchase, we don't remember prospectively, but retrospectively. Each store visit prompts us to remember which other store (s) we still need to go to. Therefore we do not 23 need ProM though the memories we retrieve are for future rather than past things. Similarly, the animals visiting a maze arm or being presented with comparison stimuli are prompted to remember the correct choice to obtain food. They do not have to become aware that a choice needs to be made. In conclusion, animal prospection studies examine memory for to-be-performed actions but because remembering is prompted, it involves RetM, not ProM (cf. Maylor, 2000). Remembering future actions in the above sense has only been investigated in the context of animal working memory. There are other paradigms in animal memory research that have a possible affinity with ProM paradigms, though they do not go under this label. One is the delayed response paradigm (on this paradigm and its relation to ProM, see Bisiacchi, 1996). This paradigm was pioneered by W. Hunter, who attempted to establish the duration of animal "sensory thought", a notion similar to response intentions (Hunter, 1913, reported in Roberts, 1998). Hunter trained animals to choose from among three doorways the one that was iUuminated to obtain food stored behind it. When training was completed, animals were placed in a delay chamber and watched a light briefly iUuminate on one of three doorways in the adjacent choice chamber. After variable delays they were released to go to the food-baited doorway. Hunter estimated that the delay at which animals could respond correctly ranged from seconds to a few minutes, and that this was the maximum duration of sensory thought. Subsequent work has shown that animals can tolerate much longer delays. For example, Maier (1929) trained rats to climb onto one of three ringstands placed in a small testing room, and run along an elevated pathway to a table where food was placed. A path to the table was in place from one ringstand only at any given time. Rats were made to climb the ringstand with the path three times to obtain a bite of food, and then were placed in their home cages. After delays up to 24 hours, the animals were released at a point in the room distant from the three ringstands. All rats went directly to one of the ringstands, and chose the correct one with a frequency greater than chance even after 24 hours. 24 What abilities does this task measure? Let us assume for argument's sake that the rats removed from the testing room after learning which of the three ringstands led to food formed an intention of the type "get food from ringstand 3 when back". When the rats were reintroduced into the room at the later time, what part(s) of this intention did they have to remember? Certainly not the part about "getting food when back" as this was the only thing they ever did in the room. They only needed to remember the part about "ringstand 3", and therefore, RetM not ProM abilities were tested. At this point a picture begins to emerge of what would be needed to create a ProM requirement. The rats would have to be busy pursuing a goal different from getting food at the time of test (an equivalent of the ongoing task in human ProM paradigms). In addition, the cue would have to be an event or circumstance associated with several courses of action, among which is getting food. In this way, the rats' attempt to climb any of the ringstands would signal successful ProM, climbing the correct ringstand would signal successful RetM. Another type of paradigm in animal research that presents affinity with ProM situations examines animals' ability to regulate their behavior according to time. According to Gallistel (1990), animals make sophisticated use of temporal and spatial information about food availability to schedule future behavior. Scheduling is evident for example in some animals' anticipatory strategy, that is, their tendency to arrive at the food source some time before peak availability, to be "the early bird" (Gallistel, 1990; Wilkie, Carr, Siegenthaler, Lenger, Liu &Kwok, 1997). There are many observations that document timing in animals. Early experiments with honeybees have shown that when a food source -typically a beaker with sugar water- is made available at a certain time of day, bees will visit that source on subsequent days shortly before the original time of occurrence (cf. Gallistel, 1990). Birds seem able to time their feeding behavior according not only to time of day, but also day of the week, conditional time of day or time elapsed. For example, Sibly and McQeety (1983) reported that herring gulls nesting on a British island visited landfill sites at exactly the 25 garbage unload schedule, different for weekdays and Saturday, and with no unload on Sunday. They also observed that the gulls timed their visits to the island's pastures to feed on worms according to time of day and wetness of the ground, that is, they went around sunrise after rainfalls. Gill (1988) found that hummingbirds time their visits to clusters of flowers according to the rate at which the latter replenish. These timing behaviors have been documented in the laboratory. Biebach, Gordjin and Krebs (1989) tested garden warblers in an apparatus consisting of a central living area and four connected rooms. In each room there was a feeder open only during a certain 3-hour period of the day. For example, the feeder in room 1 would be open from 6:00 to 9:00 a.m., the feeder in room 2 from 9:00 a.m. to 12:00 noon, and so on. The birds learned to make the majority of their visits to the correct room at the correct time of day in a 12-hour period. Importantly, in the test sessions all four feeders were kept open for the entire day, yet the birds alternated rooms according to the learned pattern. This shows that time cues, not food availability cues, guided the birds' behavior. Wilkie and Willson (1992) trained pigeons to use time interval information to regulate pecking in a testing chamber with four pecking keys. Each of the four keys yielded food only during a certain 15-min interval in the hour the pigeons spent in the chamber. For example, key 1 delivered food during the first 15 min, key 2 during the next 15 minutes, and so on. The pigeons correctly shifted from one key to the next when approximately 15 min had elapsed. The shifts were internally guided, not prompted by actual food availability and withdrawal (Wilkie & Willson, 1992). The animal behaviors in these paradigms resemble human behaviors on habitual time-cued ProM tasks, for example, taking medication at 8:00 a.m., or checking email messages every hour. But do the animals really require habitual ProM in these time-place learning paradigms? The birds in the laboratory settings did not have to remember that it is time to feed, that is, make use of ProM. Their job was only to determine where to feed, that is, make use of RetM. Once more, it is the absence of an ongoing goal different from the intended one that seems to elude ProM. In naturalistic settings 26 this maybe different. Gill (1988) describes how male hummingbirds spend much of their time in an assembly area, trying to attract females with courtship behavior. But they must interrupt courtship to visit replenished blooms and these visits are not dictated by the birds becoming hungry but by the rate at which the blooms regenerate nectar. In this situation there is a clear requirement for ProM, albeit habitual: the hummingbirds must remember to feed at the appropriate time, that is, they must interrupt a different ongoing activity, and the cue is not a direct prompt such as a state of hunger. In sum, there currently is no laboratory paradigm that measures ProM in animals. The problem is not which type of ProM is addressed, whether short- or long-term, episodic or habitual, but it is at a deeper level: none of the experimental situations appears to require any kind of ProM activity. All situations involve to some extent the ability to form memories of to-be-performed actions, but when the time comes to put these into practice, retrieval of the to-be-performed action is prompted by the situation, not initiated by the animal, because there is no competing activity. This involves RetM not ProM. To imagine a situation that presents animals with a clear ProM requirement, we can refer back to the requirements identified for human ProM paradigms. There would need to be an ongoing task at test, that is, the animal would have to pursue a goal different from the intended one, as was the case for the hurnmingbirds. In addition, the cue should not be a direct prompt to switch to the intended goal, as was the return to the testing room for the rats in Maier's study. It should instead be some event or circumstance that is associated with the pursuit of different goals and this may well turn out to be the trickiest part of designing an animal ProM task. Importantly, to target ProM proper, or episodic ProM as defined in the present work, the cue should signal a single, non-routine opportunity to achieve the intended goal in the test setting. 27 PART I: T H E ATTENTIONAL RESOURCES REQUIRED FOR PROSPECTIVE MEMORY 28 CHAPTER TWO: THEORETICAL MODELS AND EMPIRICAL EVIDENCE The present work assumes that ProM draws on attentional resources, and its primary goal is to identify the type of resources the cognitive system must make available for ProM. The participation of attentional processes in ProM retrieval however is not a foregone conclusion and has been debated since the onset of interest in this type of memory (cf. Craik, 1986; Einstein & McDaniel, 1990; Ellis & Kvavilashvili, 2000; McDaniel & Einstein, 2000; Schonfield, 1982). At the heart of this debate is the question of how prospective retrieval is achieved. Is this process effortful and attention-demanding as is retrieval in a free recall test, or is it automatic and pre-attentive as is, for example, retrieval in an implicit memory test? Decisive for answering this question are assumptions about the role the prospective cue plays in retrieval. Theoretically there are two possibilities. One possibility coincides with the view adopted in the present work that the prospective cue, imlike any other memory cue, places the burden of retrieval squarely with the rememberer, by requiring that it be identified as a cue in the first place. This possibility implicates mental work and utilization of attentional resources. The second possibility is that the prospective cue, like other memory cues, acts as an external trigger that supports and directs retrieval, thereby reducing the need for mental work and resource expenditure. Assume the task is to remember to mail a letter on the way home. Is the mailbox we encounter later a shapeless something that requires our attention to become a signal for the intended action, or is it a direct trigger for the action? Our subjective experience of ProM is consistent with both possibilities. We tend to describe the experience of prospective remembering as sudden and effortless, something that just pops into mind (Einstein & McDaniel, 1996). Yet we usually indicate ongoing attention absorption, being busy for example in a conversation, as the main reason for prospective forgetting. I will begin this chapter by reviewing three theoretical models of ProM retrieval processes. As will be seen, two models view ProM as a resource- or attention-demanding process, the third suggests also the involvement of automatic processes. I will articulate the predictions these models 29 make in the areas of ongoing task interference, aging, and neuropsychology. I will then examine the empirical evidence relating to these predictions, with the goal of estabkshing that ProM is indeed a resource-demanding process. Theoretical Models of Prospective Memory Retrieval The first model of ProM retrieval is based on Craik's characterization of different memory tests in terms of the demands they pose on the cognitive system. Craik (1986) formulated his framework to explain the variability in age declines on different memory tests. His framework served as the conceptual anchorage for the present research on the attentional demands of ProM. The framework explains age effects on memory as a result of the type of processing required for remembering. Craik argued that in every memory test a balance exists between processing supported by the retrieval context and processing initiated by the subject. In some tests, the retrieval context guides test-relevant processing and the need for self-initiated processing is negligible. This is the case for implicit memory tests and recognition tests. By contrast, in other memory tests the retrieval context does not support test-relevant processing, and remembering relies critically on processing initiated by the subject. This is true for ProM tests and free recall tests. Craik arranged memory tests in a hierarchy, according to their relative demands on self-initiated processing. In this hierarchy, ProM ranks highest because no external cue initiates remembering -the subject has to "remember to remember". In Craik's view, self-initiated processing is effortful and requires the availability of attentional resources. By this view, ProM depends on the resources the system can make available at the time of retrieval. If resource availability is restricted, as for example by increasing the demands of the ongoing activity, ProM should suffer. Resource availability is assumed to decline with age. Therefore the model predicts that older people should have difficulty with ProM tests, and that these difficulties should be mediated by declines in attentional resources. Although Craik did not specify neuropsychological correlates for his concepts of self-initiated processing and attentional resources, 30 later proposals link these concepts to frontal lobe structures (cf. Glisky, 1996). Thus the model is consistent with frontal lobe involvement in ProM. The second model of ProM retrieval was proposed by Burgess and Shallice to account for ProM impairments in patients with frontal lobe lesions (Burgess & Shallice, 1997; Shallice & Burgess, 1991). Shallice and Burgess (1991) reported a highly circumscribed pattern of deficits in three patients with focal frontal lobe damage. These patients had no difficulties with tests of intelligence, perception, language, and RetM, or with traditional tests of frontal lobe function, but showed marked impairments in laboratory and real-life situations that required ProM \ To account for the independence of ProM from cognitive and RetM memory deficits, Shallice and Burgess proposed to link ProM retrieval to the Norman-Shallice framework of action control (cf. Norman & Shallice, 1986). This framework postulates the existence of a supervisory attentional system located in the cortex of the frontal lobes. This system interacts with lower-level cognitive processing systems located more posteriorly. Lower-level cognitive processes mediate well-learned courses of action triggered by routine contexts. When non-routine courses of action need to be implemented whose context is defined in opportunistic or flexible terms ("do when circumstances allow"), the supervisory system intervenes to modulate lower-level processes by activating and inhibiting particular routines. Lower-level processing and processing mediated by the supervisory system are qualitatively different: the former is driven by external input and the outcome is rigid, the latter is willed, and the outcome is flexible (Shallice, 1988). Shallice and Burgess (1991; Burgess & Shallice, 1997) claim that ProM retrieval requires the intervention of these supervisory control processes. They suggest that when an intention to act at a future moment is generated, an intention marker or message is set up to process that particular event or time in a non-routine manner, that is, as a signal to perform a different or cuMkiond activity. Such a marker might be for example "When going to lunch, post letter". The triggering of this marker in 1 In the 1991 paper the authors refer to the skills required in these situations as the ability to reactivate intentions after delays. In a later paper (Burgess & Shallice, 1997), they employ the term Prospective memory. 31 the ensuing circumstances invokes supervisory processes that take control over cognitive processing to interrupt the routine activities associated with going to lunch, and to switch over to the non-routine action of posting a letter. This model, like Craik's model, views ProM retrieval as an effortful operation, whose initiation requires internal input. Specifically ProM depends on the integrity of the supervisory attentional functions of the brain, located in the frontal lobes. Thus the model predicts that frontal lobe structures are the neuropsychological correlate of ProM. On the assumption that functions mediated by the frontal lobes decline with age (West, 1996), the model predicts age effects on ProM. McDaniel and Einstein (2000) have put forth a multiprocess framework according to which the cognitive system relies on two main classes of processes to achieve ProM retrieval. One class of processes comprises resource-demanding executive mechanisms activated by the individual, which are directed either at the environment to identify ProM cues, or at periodically bringing to mind the intended action. Another class of processes comprises automatic mechanisms activated by the encounter with the ProM cue, that transfer the intended action to conscious awareness with no resource expenditure. The particular combination of processes involved in ProM retrieval varies across task types and characteristics. Thus, the system will exploit the resource-saving automatic processes if the ProM cue is fully processed, either because it captures attention or because it is a focal part of the ongoing activity. Resource-demanding processes will be required instead if there is little or no processing of the ProM cue, either because it is not salient or because it is not attended in the context of ongoing activities. In practice, this model admits both possibilities prospected at the outset of the chapter. It says that to the extent the ProM cue is processed, it does the work for the rememberer, to the extent it is not processed, it requires work from the rememberer. The latter possibility can be assimilated to Craik's, and Shallice and Burgess' models of ProM retrieval. The former requires further examination. 32 How, by which known mechanisms, could the ProM cue bring back to mind the intended action in an automatic fashion? In an earlier work, McDaniel and colleagues (McDaniel et al, 1998) aligned ProM tasks with RetM tasks of the associative episodic type, and the ProM retrieval process with the activation of an automatic hippocampus-based memory system according to the systems view proposed byMoscovitch (1994). This type of memory retrieval is assumed to occur automatically when an external cue interacts with a memory trace. The product of this interaction is the rapid, unsolicited emergence of material associated with the cue in consciousness. For example, when asked "Have you read War and Peace}", the title of the book brings to mind the associated answer, without requiring a memory search (the example is from Moscovitch, 1994). By analogy, in ProM tasks, the encounter of the event, designated previously as the cue for acting on an intention, automatically triggers the retrieval of the intended action. For example, if the intention is to stop at the comer store on one's way home and buy stamps, the sight of the store brings to mind the associated intention, without requiring a self-initiated identification of this event as a ProM cue. The multiprocess framework makes mixed predictions in the areas of task interference and aging. Effects are expected in some but not other situations, with the extent of cue processing marking the difference between these situations. The variables that define extent of cue processing have not yet been clearly laid out, therefore the only prediction that can be evaluated at present is that there should be studies showing ongoing task and age effects, and there should be studies that fail to show these effects. To summarize, all three models indicate attentional or resource availability as an important variable in the ProM retrieval process. There are however significant differences between these models. Craik, as well as Shallice and Burgess, characterize ProM retrieval as a resource-dependent process in principle, that is, by virtue of the processing it requires. When resource availability is reduced, whether by ongoing task demands or by aging, there should always be a cost for ProM. According to McDaniel and Einstein who describe ProM retrieval in more opportunistic terms, as 33 sometimes relying on resources but sometimes taking advantage of automaticity, there should only sometimes be a cost for ProM under those conditions. Neuropsychologically, Craik's model, as well as Shallice and Burgess, place retrieval mechanisms squarely in the frontal cortex, whereas the multiprocess framework predicts the involvement of different brain areas -frontal and hippocampal- to underscore the variety and flexibility of retrieval mechanisms. In this work, resource mediation is viewed as a defining mechanism of ProM retrieval, in line with Craiks', and to some extent, Shallice and Burgess' model. The following section will examine evidence from empirical studies to establish the ubiquity of ProM resource demands. Effects of Increased Ongoing Task Demands This and the next two sections will focus on studies that have utilized laboratory-controlled tasks consistent with the definition of ProM provided in Chapter One. To briefly recapitulate, these tasks involve a request to remember to do something in response to a specific event, a filled delay between instmctions and test, and ProM cues that are presented as part of ongoing stimuli. In some cases, other types of studies will be included for completeness sake, for example field studies on ProM and aging, or because a valid assessment of ProM has yet to be developed, as in the case of neuroimaging studies. One type of evidence critical for evaluating ProM's resource demands comes from studies that have restricted resource availability for the ProM task in young subjects by adding a secondary attention-demanding ongoing task. Two experiments by Einstein, McDaniel and colleagues (Einstein, Smith, McDaniel &Shaw, 1997, Experiments 1 and 2; McDaniel, Robinson-Riegler, & Einstein, 1998, Experiment 3) have documented reductions in ProM performance in university-aged participants in the presence of increased attentional demands. Einstein et al. (1997) examined young (M age =19 years) and older subjects (M age = 73 years). The ProM task was to press the slash key on the computer whenever a particular word occurred in the experiment. In the standard attention condition, the ongoing task was rating words on their pleasantness, familiarity, or concreteness. In 34 the demanding attention condition, a string of digits were presented on audio-tape simultaneously with the word-rating task, and subjects were instructed to press a hand-held counter each time they heard the digit 9. In their first experiment, the attention manipulation spanned all phases of the ProM task (encoding, retention interval, and retrieval). The addition of the secondary task was associated with a reliable drop in ProM. In their second experiment the attention manipulation selectively targeted either the encoding or the retrieval phase of the ProM task. In this case, ProM performance suffered under increased attentional load during encoding but not during retrieval. The finding that added demands in the ProM retrieval phase do not affect performance, goes against the view that resource-mediation was an important factor in responding to the cues. However a later study by McDaniel et al. (1998; Experiment 3) did obtain evidence that ProM is affected when increased attentional load occurs during retrieval. The ProM task was again to press a key on the keyboard upon encountering a specified target word. In their third experiment, they had young subjects perform a word-rating task similar to the 1997 study. Half of the subjects, in addition, had to monitor a sequence of auditorily presented digits and to report immediately any occurrence of three consecutive odd numbers. Subjects who received the secondary task responded to reliably fewer ProM cues than subjects who only received the primary task. A study by Otani, Landau, Libkuman, Louis et al. (1997) failed to demonstrate a cost for ProM associated with the addition of a secondary task. These researchers conducted three experiments with young adults, in which the basic ProM and ongoing task combination was the Einstein and McDaniel paradigm. Briefly, subjects were instructed to hit a key whenever they encountered a target word (or target words). The target word(s) was/were embedded in trials of a short-term memory task. In all experiments there was a zero-load condition in which no secondary task was required, and two attentional-load conditions in which a secondary task needed to be performed during the word presentation phase of the short-term memory trials. Subjects were required to repeat aloud the syllable "the" or a string of digits. Attentional load did not affect 35 prospective memory performance when there was a single target word (Experiment 1), multiple target words (four or eight) or target words belonging to a category (musical instruments) (Experiments 2 and 3). Can the evidence from this study be taken as an indication that resource-free processes mediated ProM performance? An alternative account could be that the added task of repeating syllables or strings of digits was in fact not attentionally demanding, or at least did not draw on resources critical for ProM. A study by Marsh and Eflcks (1998) provides support for this latter argument by showing that ongoing attentional demands on central-executive resources interfere with ProM, but demands on articulatory attention as in the Otani et al. study do not. Marsh and Hicks investigated the attentional demands of ProM in young subjects from the theoretical perspective of Baddeley's model of working memory (Baddeley, 1986). This model has three components, namely, a central executive component with control and coordination functions, and two storage components -the visuospatial sketchpad and the phonological loop. Marsh and Hicks conducted five experiments, in which the ProM and the primary ongoing tasks were the same. Subjects were instructed to press a key on the computer whenever a type of fruit (for example, apple) appeared during the experimental sequence. The fruit words were embedded in a short-term memory task, in which triads of words were presented for immediate recall. All secondary ongoing tasks had either a low or a high load. Significant ProM decrements, ranging from 25 to 31%, were found in three high load secondary tasks engaging central executive functions. One was a running counting task, in which people had to count forward and backward by variable amounts (Experiment 1). The second was a random number generation task, in which subjects were to produce numbers at the rate of one per sec (Experiment 2). The third was to tap alternate key sequences (Experiment 4). By contrast, two secondary tasks engaging the storage components (rehearsing aloud a set of monosyllabic words, Experiment 3; watching a regularly changing color display on the computer screen, Experiment 5) did not involve costs for ProM. Marsh and Hicks interpreted their findings as 36 evidence that successful ProM relies substantially on the attentional control resources of the central executive, but not on articulatory and visuospatial attention. On balance these studies show that ProM demands attention. When attention is heavily drained by ongoing activities, as in the case of monitoring incoming stimuli or initiating complex response sequences, there is a cost for ProM. Importantly, in the experiments that showed no cost for ProM, the added task was mamtaining material in the phonological storage system. It seems more parsimonious to say that ProM does not require this particular type of attention, rather than saying that ProM in these cases was mediated by automatic mechanisms. Age Effects on Prospective Memory Does ProM decline with age? Early field and laboratory studies (e.g., Einstein & McDaniel, 1990; Moscovitch, 1982) indicated ProM as an exception to the general pattern of age-related decrements in memory. Now, after more than a decade of research, the consensus is that when examined under laboratory conditions, ProM performance does show reliable declines in older people (Mayior, 1996; Mayior et al. in press). If it is true that ProM retrieval draws on resources, then age declines should occur because older people have fewer resources available for processing. But there is a distinction to be made. A ProM situation is by definition a dual-task situation, where ongoing activities draw on available resources, with no regard, as it were, for other needs, such as ProM. ProM, in order to occur, has to compete for resources. On reflection, then, it is possible that older people perform more poorly on ProM tasks only when they are challenged to a greater extent than young people by a difficult ongoing task and therefore have fewer resources left for the ProM task (Maylor, 1996). If age declines are observed only under these conditions, they would provide weak support for the view that ProM perse is resource-demanding as they might only be an indication that in older people the total resource availability is reached or exceeded in certain ProM situations. Alternatively, it could be that age declines in ProM also manifest when ongoing task difficulty is equated for young and old, or even when the ongoing task is designed to be easier for 37 older people. This evidence would be strong support for a resource view of ProM retrieval as it would be a good indication that it is the processing involved in ProM that demands resources and that it is these resources that older people lack. It seems that Craik had in mind this latter possibility when he predicted the greatest age effects for ProM tasks because of their high reliance on self-initiated processing. The question addressed in this section therefore becomes whether age effects occur across a variety of ongoing tasks, some difficult, others relatively easy for older participants. There have been a number of studies on ProM and aging in recent years (for a review of older studies on ProM and aging, see Maylor, 1996). Einstein et al. (1997), in their study on the effect of increase attentional demands of the ongoing task on ProM described in the previous section, also examined age differences between young (M age=19 years) and older subjects (M age =73 years). Briefly, the ProM task was to press the slash key on the computer whenever a particular word was encountered. In the standard attention condition, the ongoing task was a word-rating task. In the demanding attention condition, subjects, in addition to rating words, were also to listen to and monitor a string of digits. In their first experiment the attention manipulation was applied at encoding, during the retention interval, and at retrieval. The results showed reliable age differences in favor of the young only in the demanding condition. In their second experiment the attention manipulation selectively targeted either the encoding or the retrieval phase of the ProM task. The older subjects showed reliably lower levels of ProM than the young when the attention demands were increased at retrieval but not when they were increased at encoding. Einstein and colleagues concluded that older adults were significantly impaired on ProM performance only when they were presented with the ProM cues during a highly demanding ongoing task. Park, Hertzog, Kidder, Morrell and Mayhorn (1997) compared performance of young (M age =19 years) and older subjects (M age =70 years) on a ProM task designed to be particularly difficult, as the cue was not linked to the ongoing task. All subjects were given a demanding short-term memory task in which single words were presented at a rate of 3 sec per word, and recall for 38 the last 3 words was tested at unpredictable intervals. Words were presented on different abstract background patterns. The patterns changed with each word. The ProM task was to press a key on the computer whenever a background pattern of undulated lines appeared (altogether either 6 or 12 times). Older people's ProM performance was significantly lower than young people's performance. So far it appears that age affects ProM only when the processing demands of the ongoing task are elevated. However, a large-scale study by Kvavilashvili, Kombrot, Mash, Cockburn, and Milne (2000) provides evidence that ProM is not spared in older age even when the ongoing task has a low cognitive load and its difficulty is equated between young and old. Kvavilashvili et al. presented young (M age =23 years) and older subjects, the latter grouped into four different age bands (61-65, 66-70, 71-75 and 76-80 years, respectively), with an ongoing task that consisted of answering general knowledge questions. Questions were presented at a rate of 12 sec for the young and 15 sec for the older subjects in the low load condition, and at a rate of 9 sec for the young and 12 sec for the older in the high load condition. Some time prior to commencing the knowledge question task, subjects were instructed to type 6 digits (1-2-3-4-5-6) whenever they received a question about a telephone. There were six such questions in a 19-min question block. A reliable decline in ProM performance was found in the two oldest age groups, evident in both low and high load conditions, but more marked in the latter. Maylor et al. (in press) investigated the effects of ongoing task difficulty and of degree of processing overlap between the ongoing and the ProM task, on age differences in ProM performance. Degree of processing overlap was manipulated as a test of the task-appropriate processing hypothesis. There were two types of ongoing tasks. One was semantic and required subjects to identify a synonym for a target word among six alternative responses. This task was assumed to be easier for the old than the young participants because it relied on semantic knowledge. The other was structural and required subjects to write down the word that follows the target word in alphabetical order. These ongoing tasks were combined with a semantic or a 39 structural ProM task. The semantic ProM task required subjects to circle the trial number if the name of a color appeared whereas the structural ProM task required subjects to circle the trial number if a word with a double letter appeared. The two non-overlap conditions (semantic ongoing task/structural ProM task or structural ongoing task/semantic ProM task) were assumed to be more difficult for the old as they required a shift in processing for correct ProM responses. Older people (M age =72 years) showed lower ProM performance than young people (M age=19 years) in all task combinations. Importantly, ProM was not spared in older people even when the ongoing task was relatively easier for them (older people did outperform the young on the semantic ongoing task) and no processing shift was required to respond to the ProM cue (the semantic/semantic combination). Age deficits were more marked in the non-overlap conditions, where a shift was required to process the ProM cue. Though these results need to be interpreted with caution regarding the test of the task-appropriate processing hypothesis (see Meier & Graf, 2000, for a discussion), they do offer compelling evidence that the old have difficulty with ProM across a variety of ongoing and ProM tasks. In all studies reviewed until now, age differences in ProM emerged when participants had to interrupt a controlled ongoing activity to respond to the ProM events. It could be argued that older people's difficulty with ProM lies in their reluctance to interrupt something they are doing (Schonfield, 1982). What happens when no overt interruption is necessary, that is, when the ProM events occur in natural breaks between activities? According to Kvavilashvili and Ellis (1996), this type of ProM situation is likely to tax attentional resources to a much lesser extent than ProM situations where an activity interruption is required. Uttl et al. (2001) investigated age-related ProM declines in four groups of older adults (aged 65-69, 70-74, 75-79, and 80-95 years). Their ProM tasks are described in detail in Chapter One. Briefly, instructions to carry out a simple action when the experimenter announced the end of a task were given prior to, and the cue delivered upon completion of, three attention tasks in a 2-hour test 40 battery". The task scores, which indexed whether subjects remembered that something had to be done, regardless of whether the requested action was recalled, reliably declined across the four age groups on all three tasks. Jacova et al. (2001) obtained similar evidence when they examined ProM performance in three groups of middle-aged and older adults (50-58, 59-66, and 67-82 years). They gave participants three ProM tasks, embedded in different parts of a 5-hour neuropsychological test battery. The first task was to remember to date the page before beginning the copy trial of the Rey-Osterrieth Complex Figure. The second was to remind the experimenter to take along a lab requisition form when the experimenter announced a break. The third was to remind the experimenter to administer a mystery test when the experimenter announced the end of the last part of a smell recognition test. The interval between instructions and test varied from 15 to 30 min. ProM scores, which did not include whether the requested action was recalled, declined substantially as a function of age on the first and the third task (they were at ceiling for the second task). There are a number of field studies on ProM and aging, conducted outside the laboratory. For example, Moscovitch (1982) instructed older and young people to phone an answering service at a time of their choosing. He found that a greater number of older than young people remembered these telephone appointments. In a second study he decided to make the task more difficult by setting up experimenter-determined telephone appointments but older people were still no worse than young people at remembering these appointments in a timely fashion. He concluded that it would indeed be very hard to "bring older people's memory to its knees" in everyday situations because of their ability to compensate for poorer memory with the use of cues and reminders. Others have observed that a more structured and less active lifestyle may also contribute to preserving everyday ProM abilities in later adulthood (Maylor, 1996; Rendell & Craik, 2001). The conclusion emerging from these studies is threefold. One, there is solid empirical evidence that age affects ProM performance under laboratory conditions. As discussed in Chapter 41 One, evidence from field studies must be considered with caution given the lack of control over performance conditions. Therefore, the observation that older people function quite well in everyday situations does not take away from this conclusion, though it raises interesting questions about the type of mnemonic strategies that can support ProM. Two, age-related declines seem to occur regardless of whether the ongoing task at the time of retrieval is difficult or easy, whether or not there is processing overlap between ongoing and ProM tasks, or whether there is an overt ongoing task at all. The only exception is the standard attention condition in the Einstein et al. study (1997). In this condition only nominal age differences favoring the young were found. This picture strongly suggests that age effects on ProM arise not because older people are selectively penalized by the ongoing task, but because their system is penalized by ProM's substantial demands on their resources. Three, the age-related declines become more marked when the demands of the ongoing task increase at the time of retrieval. In summary, the evidence on age effects strongly supports a view of ProM retrieval as a process that can only be enabled by resource expenditure, and cannot occur via more automatic routes. According to Craik's framework, ProM declines with age because of decrements in the resources that support self-initiated processing. This framework falls within a more general theory of cognitive aging known as the resource-reduction theory (Kausler, 1994). According to this perspective, what changes with age, is not the efficiency of specific mental functions but the pool of available cognitive resources. The principal interest of this perspective is in identifying a small set of primary or component abilities, whose change across age can account for performance declines on a wide range of cognitive and memory tasks (Kausler, 1994; Salthouse, 1991,1996). To my knowledge, only two studies in the ProM and aging literature have directly examined the relationship between age, measures of processing resources, and ProM. Uttl and Graf (2000) measured ProM and choice reaction time in older adults aged 61 to 91. ProM was assessed by means of the computer-administered task described in Chapter One. Subjects were instructed to stop 42 performance on any ongoing activity and recall a word list, if they noticed a picture cue in the course of the experiment. At some point, while subjects were engaged in an A/B choice reaction task, the picture cue was repeatedly presented among distractor pictures. Performance on choice reaction trials without pictures was used as an index of processing capacity. This index accounted for approximately 60% of the age-related variance in ProM. Jacova et al. (2001), in the study described in the previous section, assessed processing resources by means of two tasks: a simple reaction task was used as an index of processing speed, an A/B choice reaction task as an index of processing capacity. They tested the contribution of the two resource factors to the age effects on ProM by means of hierarchical modeling. Age had both direct and indirect effects on ProM. The indirect effects were mediated by age-related decrements in both processing speed and processing capacity. The evidence from these two studies, though scant, is consistent with Craik's prediction that a key mechanism underlying age declines in ProM is a restriction of attentional resources that older people have available for processing. The Neuropsychology of Prospective Memory In 1996, three contributors to the first book devoted to ProM (Bisiacchi, Glisky, and Shallice; in Brandimonte, Einstein & McDaniel, 1996) noted that only single case studies (e.g., Cockburn, 1995; Shallice & Burgess, 1991) had addressed the neuropsychology of ProM, and that there was a strong need for systematic lesion and neuroimaging studies. Meanwhile, a number of such studies have been published. Some evidence exists to address the question whether ProM depends on frontal lobe structures as predicted by Craik's as well as Shallice and Burgess' model, or also on hippocampal structures, as predicted by McDaniel and Einstein's multiprocess framework. McDaniel, Guynn, Glisky, Rubin and Routhieaux (1999) have investigated the relationship between individual differences in brain functioning and ProM in older people. Burgess et al. (2000) have correlated lesion foci with performance on a multitasking test in neurological patients. Two 43 studies (Burgess et al., 2001; Okuda, Fuji, Yamadori, Kawashima et al, 1998) have investigated brain correlates of ProM performance by means of positron emission tomography (PET); one study has done so by means of event-related potentials (ERPs) (West, FJerndon & Ross-Monroe, 2000). The lesion, PET and ERP studies provide suggestive neurocognitive evidence about the localization and time course of ProM processes but the evidence is largely based on tasks in which the formation of an intention and the opportunity to act on it have not been interpolated by intention-irrelevant activities. It is therefore likely that these studies are a more accurate characterization of conscious monitoring for cues, or short-term ProM, than of regaining awareness of an intention to act upon discovering a cue, or ProM as defined in the present work. They represent the current neurocognitive understanding of ProM and will be examined here, but with the caveat that it cannot be assumed that short- and long-term ProM processes are supported by similar neurocognitive mechanisms. Aging and Lesion Studies McDaniel et al. (1999) investigated the effects of age-related declines in frontal and medial temporal functioning on ProM performance. Forty-one healthy older adults (M age =74) were classified as high- or low-functioning on two composite neuropsychological scores. One was a frontal lobe score based on multiple tests sensitive to frontal lobe functions, such as the Wisconsin Card Sorting Test or the Controlled Oral Word Association Test. The other was a medial temporal lobe score, based on multiple tests sensitive to hippocampal functions, such as the Logical Memory and Verbal Paired Associates from the Wechsler Memory Scale. Participants were given an event-cued ProM task, which consisted of pressing an F key on the computer keyboard upon encountering a specified word (e.g., president). Eight ProM target words were embedded in questions presented for an ongoing general knowledge task, in which subjects had to choose the correct response among four response options. Frontal but not hippocampal functioning had a significant impact on ProM, with low-functioning subjects missing twice as many cues as high-functioning subjects. 44 This study links ProM to areas in the frontal lobes, but which areas are these, and which cognitive processes do they mediate? Burgess et al. (2000) conducted a more fine-grained anatomical-behavioral analysis on aspects of ProM involved in multitasking. Sixty patients with focal cerebral lesions in different frontal, temporal, parietal and occipital regions, were given the Greenwich test, which required them to perform three open-ended tasks over a 10-min period. They were given a set of instructions and rules stating that they would not be able to finish any of the tasks, that earlier items in any task would gain them more points than later items, and that their job was to score as many points as possible. Patients could consult a stopwatch that was available under a cover. Several dependent variables reflecting ProM and RetM functions were compared. Measures of ProM were "Plan", that is, the adequacy of the strategy patients intended to follow in the test, "Follow", that is, the extent to which patients applied their original plans, and "Score", that is, the number of times they spontaneously switched subtasks over the test period. Lesions in the right dorsolateral prefrontal cortex were associated with low Plan levels, whereas lesions in the left polar and medial frontal regions as well as in the posterior cingulate and forceps major regions (the latter both regions with important hippocampal connections) were associated with low Follow and Score levels. This pattern of lesion-behavior associations, combined with a cognitive-level analysis, identified the right and left frontal poles as supporting ProM processes of planning and retrieval, respectively, and the posterior cingulate/forceps major as supporting RetM processes of recalling rules and plans. Both studies implicate frontal lobe areas in ProM retrieval. The localization of retrieval processes to the left frontal polar and medial frontal regions must be considered tentative in the context of long-term ProM as this evidence is based on a test that seems to require active monitoring of intentions and actions, that is, short-term ProM. 45 Neuroimaging Studies In a positron emission tomography (PET) study by Okuda et al. (1998), healthy subjects learned 10 target words and were instructed to retain them over the scanning period for a later recall test. During scanning they were presented auditorily with 50 words and their task was to repeat each word aloud. There were two scanning conditions. In the experimental condition subjects were instructed to knock with their hand if they heard any of the target words among the words they had to repeat (only two target words were actually presented). In the control conditions no instructions, other than repeating the words, were given. Subjects succeeded in responding to the target words on 65% of trials. The experimental condition was associated with increased activity in the right prefrontal cortex, the left frontal pole and medial frontal lobe, and the left parahippocampal gyrus. This differential brain activity, according to the authors, reflects holding an intention of future behavior rather than retrieving the intention because of the low target rate and the non-perfect performance. The participation of the parahippocampal gyrus is viewed in terms of detection of novelty. Because only in the experimental condition was there an intention associated with the retained words, this study suggests a role for the prefrontal and polar frontal regions in mamtaining an intention to act. However the study is problematic on several counts. First, as already noted by the authors, the study addresses the period prior to retrieval, not retrieval itself. Second, the task only engaged one cognitive domain and had an unusually high target load. The question arises to what extent the activation reflects only task-specific processing. Third, the time frame of the ProM task (6-8 min) suggests that subjects engaged in conscious expecting and looking out for targets, characteristic of short-term ProM. Burgess et al. (2001) addressed some of these problems in a PET study in which healthy participants performed four ProM tasks involving different types of processing: visuospatial, numerical, or verbal. AH tasks involved an ongoing decision task in which subjects pressed either the 46 leftmost or middle key on a 3-key pad, and ProM instructions were to press the rightmost key on the pad when noticing a cue. ProM instructions were given together with the ongoing task instructions prior to a short practice block. Each ProM task was performed under three different conditions. In the baseline condition subjects did not receive ProM instructions. In the Expectation condition, subjects received ProM instructions but no cues were actually presented. In the Execution condition, subjects received ProM instructions, cues were presented on roughly every fifth trial, and were almost always responded to. Twelve scans, one for each ProM task under each condition, were collected. When contrasting the Expectation and Execution conditions with the baseline condition, increased activity was noted in the following brain regions: the frontal poles bilaterally, the right lateral prefrontal cortex, the right inferior parietal lobe, and the precuneus bilaterally. When contrasting the Execution with the Expectation condition, increased activity in the thalamus and decreased activity in the right lateral prefrontal cortex were noted. Burgess and colleagues concur with Okuda et al. in smgling out the frontopolar structures and the right lateral prefrontal cortex as the sites for specific ProM processes of intention maintenance. They argue that these structures mediate a state of anticipation and retrieval readiness uniquely associated with ProM situations where responding is self-initiated, not contextually driven. They are careful to note that given the short time frame of their ProM tasks, this neurocognitive characterization reflects situations in which people hold an intention in conscious awareness, but may not apply in situations typical for long-term ProM, in which people are engaged in other thoughts or activities. There are several problems in relating the findings from these studies to questions about long-term ProM retrieval. As already noted by Burgess and colleagues, the neural basis for short-and long-term ProM retention may not be the same. Of equal importance, neither study really captures retrieval processes. ProM retrieval includes a chain of events that goes from cue discovery, to recollecting the intended action, to executing it. The increase in thalamic activity observed by 47 Burgess and colleagues in the Execution but not the Expectation condition reflects a combination of these events in which probably response execution prevails. In fact, the authors interpret this activation in terms of producing a novel self-generated response. To identify the neural correlates for the early events of ProM retrieval, evidence from electrophysiological studies is needed. Electrophysiological Evidence To my knowledge, to date only one study has employed ERPs to investigate the components of ProM retrieval. West et al. (2000) presented subjects with various ProM trials embedded in a semantic judgment task, in which they had to make decisions about the relatedness of word pairs, presented in the same or different colors. The ProM task was to press a designated key on the keyboard when the words were printed in a specified matching color. The color to respond to changed from one to the next ProM trial. In this task, missed trials, that is, cues not responded to according to ProM instructions, were relatively frequent (from 25 to 50% of trials). West and colleagues compared ERPs for correct ProM trials, with those for missed ProM trials and for ongoing task trials, to identify the electrophysiological correlates for successful ProM. There were two modulations that differentiated correct ProM trials from other trials. One was a phasic negative effect peaking at about 300 ms after cue onset (N320) broadly distributed over the occipital and parietal regions. The other was a more sustained positive effect over the frontal-central region, with roughly the same onset as the negative effect, but lasting until 800 to 1000 ms after cue onset. West and colleagues attributed the function of perceptual noticing of the cue to the first of these effects, and functions of evaluation and attentional control to the second of these effects. The contrast between correct and missed ProM trials in this study directly addresses ProM retrieval, but retrieval of what kind? That is, did the study measure short- or long-term memory? No information is given about critical methodological features, for example, the time mterveriing between instructions and test, and it can only be speculated that the relatively frequent occurrence of missed cues might indicate the involvement of long-term ProM. If this is true, the study has 48 important implication for the understanding of ProM retrieval. It suggests that cue identification is not an automatic process since its neural correlates manifest only on correct trials. Cue identification seems to result from the interplay between perceptual processes mediated by parietal-occipital regions, and attentional control processes mediated by frontal regions. The latter are involved right from the start, and might modulate the former. Interpreted in this way, the ERP evidence is fully consistent with the view advanced at the start of the chapter that perceiving a particular event as a ProM cue requires attentional resources. Furthermore it fits Shallice and Burgess' claim that cognitive routines fall under executive control in ProM retrieval. On the whole, ProM emerges solidly as a mental operation that draws on attentional resources. Ongoing task effects, age effects, and some neuropsychological findings provide strong arguments to support this view, and do not seem to require explanations in terms of more automatic processes, as claimed by the multiprocess framework. ProM retrieval does appear to require a sizable amount of attention when encountering the cue event. It is the type or form of attention required for this process that remains to be defined, and this is the object of the research described in the next two chapters. 49 CHAPTER T H R E E : PROCESSING SPEED AND PROCESSING CAPACITY The purpose of the experiments described in this and the following chapter was to characterize the kind of attentional resources required for ProM. But what exactly are resources? How are they useful in explaining memory and other cognitive functions? Current resource accounts of human cognition aim at explaining variability in a large number of memory and cognitive functions by means of a small number of resource factors, indexed by performance on simple perceptual tasks. The goal of these accounts is parsimony (Salthouse, 1991), parsimony that in most cases is the hunt for a single general-purpose commodity (Kausler, 1994). Salthouse (1996) identifies this commodity with processing speed whereas other theorists (Hasher & Zacks, 1979;mKausler, 1994) identify it working memory capacity. In this work I adopt a more qualitative approach to the understanding of resources and their role in memory functions. The present research focused on a single memory function (ProM) and aimed at expkining performance as well as revealing underlying mechanisms by means of resource factors. Its goal was not parsimony but precision. Therefore it did not view these resource factors as a single commodity but as made-up of distinct and identifiable components. This view is consistent with recent views in the aging and attention literature that processing resources should not be conceived of as unitary but that they need to be translated into clearly defined and measurable components to increase their theoretical usefulness (Graf & Uttl, 1995; McDowd & Shaw, 2000; Neuman, 1996; Wickens, 1984). To begin with, resources were defined in this work as a system's capacity for attention. This definition is based on Kahneman's assumption (1973) that the cognitive system requires two types of input for processing. One type of input is specific information for the cognitive process to be carried out, for example, the ProM cue. The other input is non-specific energy that enables and sustains the cognitive process. The supply of this non-specific input is limited. Capacity for attention was here understood in a broad, basic sense, as something that is needed to carry out mental work. 50 No a priori assumption was made about a particular function (i.e., searching or selecting) or mental stance (i.e., conscious or voluntary). The aim of this work was to shape this broad notion into some clearly identified components that are useful for understanding ProM. As a first step toward this aim, capacity for attention was broken down into two separate components: processing speed and processing capacity. Graf and Uttl have proposed this distinction in an investigation of age effects on RetM (1995). They used the analogy of a computer system to infer the existence of two factors that independently constrain information processing in the cognitive system. Processing speed is the analog of a computer system's clock-rate, and refers to the rate at which the cognitive system can process information. Processing capacity is the analog of a computer system's register-width, and refers to the breadth of information the cognitive system can operate on simultaneously. In this way, these authors successfully link together the speed (Salthouse, 1991) and working memory resource accounts (Hasher & Zacks, 1979; Kausler, 1994) mentioned earlier. Graf and Uttl found that processing speed and processing capacity make separate and different contributions to performance on an episodic RetM test, with processing capacity being a comparatively stronger factor. I imagined a very similar pattern for ProM. By intuition, the speed at which information can be processed is important in the context of ProM retrieval. If processing of the cue is not carried out within a certain time window, other competing processing demands will take over, and the cue goes unnoticed. The breadth of information that can be processed at a given time appears even more decisive for ProM, given the intrinsically high information load that the retrieval context carries. At the very least, there are two streams of mcoming information: information pertaining to the ongoing task, and information pertaining to the ProM task. This reasoning formed the rationale for Experiment One. 51 Experiment One This experiment focused on ProM's requirements of processing speed and processing capacity. The experimental P r o M task was adapted from the paradigm developed by Uttl and Graf (2000), which has been described in detail in Chapter One. In this paradigm, to begin with, subjects were shown a picture, for example a butterfly, and told that if at any time in the experiment they noticed this picture again, they were to interrupt any ongoing activity and recall a list of words they would learn presently. Subjects were then kept busy with non-ProM activities for some time. At test subjects were engaged in an attention-demanding ongoing task during which the cue picture appeared repeatedly at increasing sizes. Three different performance components were distinguished. Two were prospective and referred to whether or not subjects interrupted the ongoing activity in response to the cue, and to the size of the cue when this interruption occurred. It should be recalled here that in actuality this latter measure is not only an index of cue size but also of number of cue presentations. Its purpose is to provide a graded index of P roM, not to discriminate between the two variables of repetition and size. Only for convenience will it be referred to as cue size. The third component was retrospective and referred to the number of list items successfully recalled. B y necessity, cue size was contingent on the first but, as will be seen the retrospective measure was assessed independently of the prospective measures. The strategy for investigating ProM's requirements of processing speed and processing capacity was to manipulate the demands of the ongoing task and evaluate the impact of this manipulation on P r o M performance. This strategy capitalized on the fact that P r o M tasks are embedded in ongoing activities whose demands can be controlled experimentally. The P r o M test situation was conceptualized as a dual task paradigm. In this paradigm, the P r o M task was the target task whose resource demands are under investigation, and the ongoing task the concurrent task whose processing demands curtail resource availability for the target task. (Note, however, that from 52 the subjects'point ofijew, the ProM test situation was a single task situation, requiring responding to ongoing stimuli, until they became aware of the ProM cue.) In this situation, the effect of concurrent task performance on target task performance can be taken as an index of the resource requirements of the latter (Navon, 1984). By this reasoning, a difference in ProM performance between two ongoing tasks requiring distinct resources, with everything else being equal, would indicate that ProM shares more resources with the task associated with lower levels of ProM performance than with the task associated with higher levels of ProM performance. The method for rmnipulating the resource demands of the ongoing task consisted of providing subjects with different task instructions but with the same materials. The idea was to induce the utilization of distinct types of attentional resources, not to increase attentional load as was done in previous studies (e.g., Einstein et al, 1997; Marsh & Hicks, 1998). This method had two advantages over the previous studies. By avoiding the addition of a secondary ongoing task, it was possible to attribute any effect of the ongoing task on ProM to the demands of the former, rather than to additional demands on dividing attention between two ongoing activities. Also, by utilizing the same stimuli, it was possible to rule out any independent effects that different stimuli might produce on ProM. The experiment had two objectives 2. The first was to compare the effects that two ongoing tasks, one requiring processing speed, the other requiring processing capacity, had on ProM performance. For convenience, the terms speed and capacity will be used herein for processing speed and processing capacity. It was hypothesized that the ongoing task requiring capacity would have a greater cost for ProM performance than the task requiring speed, and that this cost would be specific to the prospective part of the task. The second specific objective of the experiment was to examine the relationship between ProM and individual difference measures of speed and capacity. This was done to corroborate the 2 A third objective was to investigate the relationship between ProM and creative thinking. This objective, and the relevant methodology and results, are discussed in Part II. 53 findings from the ongoing task manipulation with correlational evidence, and to estimate the relative contribution each resource factor makes to ProM. As argued earlier, both speed and capacity should have a role in ProM retrieval but capacity should be the more critical resource. If the task requiring capacity produces a greater cost for ProM performance than the task requiring speed, the expectation is that individual differences in the availability of capacity should be more strongly associated with ProM than individual differences in the availability of speed. Speed was defined as simple responding without searching or making decisions; capacity was defined as responding that required searching and making decisions (Graf & Uttl, 1995). The ongoing task requiring speed was operationalized as a simple reaction task, with instructions to respond to the onset of each new letter display. The ongoing task drawing on capacity was operationalized as an A/B choice reaction task, with instructions to search, and make decisions about, a letter display. The letter displays were the same for the simple reaction and choice reaction tasks. Method Participants and Design The participants were 123 undergraduate students from the University of British Columbia, who participated in the experiment in return for course credit. There were 94 women and 29 men, between 17 and 26 years of age (M age=19.4). The core part of the experiment employed a 2 x 2 design with type of ongoing task (simple reaction or choice reaction) and type of cue (butterfly or helicopter) as between-subject factors. Cue type was manipulated to determine the generality of the ongoing task effects. Fifty-nine participants received the simple reaction task (SRT). Of these, thirty-two were given the butterfly, and twenty-seven the helicopter as the ProM cue. Sixty-four participants received the choice reaction task (CRT). Of these, thirty-three were given the butterfly, and thirty-one the helicopter cue. 54 Materials For both ongoing tasks, identical sets of letter displays were employed. These displays were adapted from a card-sorting task used by Rabbitt (1965) to investigate age effects on visual search. Each letter display was in the shape of a playing card measuring 7x7 cm on the computer screen. Letters were printed in capital characters, in 28-point Helvetica font, bright green against black background. Two sets of 54 cards were prepared. In each set, the target letter 'A' was printed on half of the cards, the target letter 'B' on the other half. The card had nine possible locations. Each of the two target letters appeared three times in each location, under three different conditions: (1) by itself, with the rest of the locations left blank (O-distractor cards); (2) together with four distractor letters, selected randomly and without replacement from the rest of the alphabet and randomly occupying four of the available eight locations (4-distractor cards); (3) together with eight distractor letters, selected randomly and without replacement from the rest of the alphabet and occupying all available locations (8-distractor cards). The two sets of 54 cards were combined to form a block of 108 randomly selected trials. Each trial began with the presentation of a card, which remained on the screen for the duration of the response, plus a randomly determined response-stimulus interval of 500, 750, 1000,1250,1500 or 1750 ms. For SRT, subjects were instructed to press the down-arrow key on the keyboard as quickly as possible each time they spotted the onset of a new card on the screen, regardless of whether it contained the letter 'A' or 'B\ For CRT, subjects were instructed to press the left-arrow key when they saw the letter 'A', and the right-arrow key when they saw the letter 'B' in the display. They were asked to make this decision as quickly and as accurately as possible. For the ProM task, photographic picture stimuli were used. Two pictures showing either a butterfly or a helicopter were used as ProM cues. The shapes in both pictures had hues in the yellow-red spectrum, against a white background. Another 108 pictures of common objects were used as distractor pictures. In the instruction phase of the ProM task, only the cue picture was 55 presented in the left upper corner of the computer screen. In the test phase of the ProM task, quadruplets of pictures, located in the four corners of the computer screen (upper right and left, lower right and left) were presented repeatedly throughout the ongoing task. The distractor pictures were of three different heights and widths, measuring 1.5 x 2, 2.25 x 3 and 3x4 cm respectively. The cue picture appeared as part of the quadruplets for a total of nineteen times, or until it was noticed by the subjects. On its first appearance, the cue measured 1.5 x 2 cm, on its nineteenth appearance 9x12 cm. On each presentation subsequent to the first, its measures increased 0.4 x 0.5 cm. The corner of the screen in which the cue appeared was randomly determined. The ProM materials were combined with the ongoing task materials as follows. In the first block of ongoing task trials only the card displays were presented. In the second block of trials, in addition to the card displays quadruplets of distractor pictures were presented. In the third block of trials, the ProM cue was presented up to nineteen times as part of a quadruplet. Figure 3.1 illustrates examples of the screen displays during Block 1 and Block 3. Distractor pictures were sampled randomly without replacement from the 108-picture pool for each quarter block of 27 trials, thus appearing for a total of eight times in the second and third block of trials. Trials on which the cue was presented were randomly sampled so that the cue appeared on average on every fourth trial. For the ongoing task manipulation it was critical that picture presentations (cue and distractors) be coupled with ongoing task trials in such a way that the resource under investigation -speed or capacity- would be tied up in the ongoing response when the cue appeared. It was calculated that the fastest response times, obtained in SRT, would be no less than 200 ms, that is 2.5 standard deviations below the mean expected reaction time for young subjects (cf. Graf & Uttl, 1995). Picture presentation was set at 200 ms, and each of the following three picture/card onset times was randomly used on one third of trials: -50 ms, 0 ms and +50 ms. When the picture/card onset was -50 or 0 ms, picture presentation would be within the response time to the ongoing stimulus on 100% of CRT and SRT trials; when picture/card onset was +50ms, picture presentation 56 would be within the response time to the card on 100% of CRT trials, and on roughly 66% of SRT trials. Procedure Participants were tested individually in a quiet laboratory room. They were seated at a desk in front of a computer and informed that they would perform a number of tasks, some concerned with attention, some with prospective memory, and some with retrospective memory. Table 3.1 lists all tasks and their order in the experiment. For all tasks administered on the computer, subjects were asked to sit at a distance of approximately 60cm from the screen. All computer materials were presented on a 17 in (43cm) VGA-monitor. At the beginning of the experiment subjects received instructions for a naturalistic ProM task, which required them to tell the experimenter about the task they liked best in the experiment upon receiving their participation credit form at the end of the experiment. They then received the North American Adult Reading Test, which is a measure of verbal knowledge, and the study task for a picture memory test, which consisted of identifying as quickly as possible all the differences between two almost identical picture displays. The experimental ProM task comprised an instruction phase, learning a word list for recall if and when the cue was noticed, a filler task, and the three-block test phase. In the instruction phase, the computer screen displayed the picture cue (a butterfly or a helicopter) in the upper left comer, and the experimenter gave the following verbal instmctions: "If you notice this picture at anytime during this experiment, you must stop immediately whatever you will be doing. This picture may appear at anytime during this experiment. It may appear only once or not at all. It is important that you focus on doing all the tasks you will be asked to do but if you notice this picture, anywhere and at anytime, you are to stop and tell me the list of words you are about to learn." gure 3.1. Examples of Ongoing and ProM Displays. C o m p u t e r s c r e e n R W E D A Ongoing task stimulus only (Block 1) lit:: J i l l 'I'lJll Computer screen ProM cue'appears up to 19'at increasing sizes Picture R W E D A Picture Picture Ongoing task stimulus, ProM cue and distractor pictures (Block 3) • »* to*! Table 3.1. List of tasks and their order of administration (core parts in bold). Tasks Materials Naturalistic ProM Task Instructions North American Adult Reading Test Picture Memory Study Experimental ProM Task Instructions Word Learning Task First Divergent Thinking Task Speed or Capacity Measurement Experimental ProM Task Test Personality Scale Picture Memory Test Second Divergent Thinking Task Speed or Capacity Measurement Naturalistic ProM Task Test Described in Blair and Spreen, 1989 Described in Perrig, Hofer, Kling, Meier etal . , 1994 Picture of a butterfly or a helicopter List of 18 concrete nouns Uses Test or Similarities Test (Wallach & Kogan, 1965) One block of SRT or C R T trials Two blocks of SRT or C R T trials, one with distractor pictures, one with distractor pictures and ProM cue Stimulus Screening Scale (Mehrabian, 1977, 1994) Uses Test or Similarities Test One block of SRT or C R T trials Participation Credit Form 59 After ensuring that subjects understood the task, the experimenter proceeded with two presentations of a list of 18 concrete nouns. Subjects were asked to read aloud each word as it appeared on the computer screen, and at the end of each list presentation, to recall as many words as they could, in any order. ProM instructions were not repeated at the end of the word learning trials. Subjects were then given the first of two divergent thinking tests, described in detail in Part II. The two activities following the ProM instructions required from fifteen to twenty minutes of time. The ProM test phase began with a set of 10 practice trials for the reaction task (SRT or CRT). For SRT, subjects were instructed to respond as quickly as possible to the onset of each letter display. For CRT, subjects were instructed to decide as quickly and accurately as possible whether the display contained the letter 'A' or the letter 'B\ When subjects had had sufficient practice, they began the first block of 108 trials, in which only card stimuli were presented. SRT and CRT performance in this block was taken as a measure of individual differences in speed or capacity, respectively. After the first block, subjects were informed that they would perform the reaction task for another two blocks, in which on every trial, in addition to the cards, they would see four pictures of different sizes in the four comers of the computer screen. They were told that the purpose was to find out how well they could maintain their performance with these additional stimuli, and they were encouraged to continue responding as quickly-and, for CRT, as accurately- as possible to the letter displays. To ensure that subjects devoted their full attention to the ongoing task in the second and third block, the software adrninistering the task was programmed to emit an auditory signal -a beep- after each response that was slower than the median response time on the first block. No additional information was provided prior to the third block in which the picture cue would appear repeatedly. Subjects were only reminded to maintain response speed and accuracy, and trials began. If, following any cue appearance, subjects stopped the reaction task and began reciting the word list, or asked what it was they needed to do now, whether they should be reciting the words, or simply said they saw the butterfly (helicopter), the experimenter pressed the mouse key to 60 record that a ProM response had occurred 3. If necessary, the experimenter prompted the recall of the word list. A small number of subjects -two or three in each condition- slowed down and said that the butterfly (helicopter) had just appeared, but carried on the reaction task. In these cases it was decided that there was not sufficient evidence for a ProM response because subjects did not interrupt the ongoing activity. When a ProM response was given and the word list recalled, subjects were asked to resume and complete the reaction task. If no ProM response occurred, after completion of the block subjects were asked to recall the word list, without mention of the ProM task. In the remainder of the experiment, subjects were given a personality scale, a free recall test of the pictures they had studied in the difference detection task, and the second divergent thinking test. To obtain individual difference measures on the resource factor subjects had not been tested on during the first block of the ProM test phase, one block of SRT or CRT trials was administered, whichever subjects had yet not performed. At the end of the experiment, subjects were handed their participation credit form, the cue for the naturalistic ProM task, and their response was recorded. Scoring of this task and performance will be described in Chapter Six. The whole experiment lasted approximately one hour and a quarter. Results Performance on the experimental ProM task yielded two different ProM performance measures for each participant. One was a binary measure of success or failure on the task, hereafter referred to as ProM response. The overall percentage of ProM responses was 54%. The other was a graded measure, indicating the cue size at which the ProM response occurred, hereafter referred to as Cue size. This measure was conditional on a successful ProM response and is therefore available only for approximately half of the participants. It ranged from 1 (smallest cue size) to 19 (largest cue 3 The mouse keypress recorded the number of cue appearances until the ProM response, and the type of card stimulus coupled with the last cue appearance. 61 size). The RetM measure, of marginal interest here, was the number of words recalled either upon responding to the ProM cue, or at the end of ProM test phase. This section will begin with analyzing by means of t-tests the effects the two ongoing tasks reqvriring speed (SRT) or capacity (CRT) had on ProM performance measures. Next, ongoing task performance will be described to show that subjects allocated their resources to the ongoing task across the three blocks of trials, and specifically in the third block, where the ProM cue was presented. Next, the RetM performance measure will be probed for ongoing task influences as well as influences due to ProM with a 2 x 2 ANOVA with factors ongoing task (SRT versus CRT) and ProM performance (success versus failure). Lastly, correlational analyses will be performed to examine the relationship between individual differences in the speed and capacity resource, and ProM performance measures. For all statistical analyses an alpha level of 0.05 was used to infer significance. E ffeds of Ongoing Task Type on Prospective Memory Measures The first question examined was which effect two ongoing tasks had on the two measures of ProM. ProM performance was significantly lower when the ongoing task was CRT than when it was SRT. To pool the data across cue types, two ANOVAs with factors cue type (butterfly versus helicopter) and task type (SRT versus CRT) were conducted. For the ProM Response variable, a main effect for cue type was found, (^1,119) = 4.19, MSE =.97, p <0.05 but there was no reliable interaction with task type. For Cue Size, neither the main effect of cue type nor interaction with task type approached significance. Figure 3.2 displays the ProM performance data by ongoing task type, pooled across cue type. Sixty-eight percent of subjects performing SRT responded to the cue whereas only 42% of subjects performing CRT did. With SRT, subjects on the average required between nine and ten cue presentations to make a ProM response, whereas with CRT subjects 62 required twelve presentations. Thus CRT subjects required a cue size approximately 20% larger than the cue size required by the SRT subjects. Comparisons by means of t tests confirmed these conclusions. The proportion of ProM responses was significantly higher for subjects engaged in SRT than for subjects engaged in CRT, £(121) = 2.95, p <0.005. There was a tendency for Cue size to be smaller with SRT than CRT, £(65) = 2.25, p O.10. Because Cue size was conditional on responding to the cue, the number of subjects compared was smaller and not equal across task types. Therefore, a corrected Cue size was computed, for which missing values were replaced with the arbitrarily chosen value of 20 -that is, one unit above the highest actual value. For the corrected Cue size variable, the difference between tasks was significant, t{\2\) = 3.46, pO.001. To probe ProM performance during CRT further, the frequency of ProM responses was analyzed by type of card stimulus associated with the cue presentation. As mentioned previously, cards with 0, 4 and 8 distractor letters were presented each on one third of randomly selected trials. ProM cue presentations occurred roughly one third of the time together with each card type. The question was whether the ProM response tended to occur more often with the 0-distractor cards than with the other card types, and this was indeed found to be the case. Fifty-two percent of ProM responses in the CRT task occurred concomitant with a 0-distractor card, whereas only 26 and 22 % did so concomitant with the 4- and 8-distractor cards, respectively. These frequencies are marginally different from the expected frequencies, chi-sqmre(2) = 4.22, p =0.11. No such pattern was found for SRT where the frequency of ProM responses was roughly equal across card types (35% for the 0-distractor cards, 37% for the 4-distractor cards, and 28% for the 8-distractor cards). Because SRT required responding to the onset of a novel card rather than searching for, and identifying a target letter on the card, there was no reason to expect an effect of distractor number, but rather an effect of magnitude of increase/decrease of letters from the previous card. 63 Figure 3.2. ProM Performance (vertical bars indicate standard errors) as a function of ongoing task type (SRT versus CRT). 1 </> 0.9 <D (/) C o a w a> * 0.6 0.8 0.7 0.5 0.4 0.3 o Q. £ o o S- 0.2 o i_ a . 0.1 0 X SRT C R T 64 Table 3.2. Mean reaction times in ms (standard deviations in parentheses) for performance on SRT and CRT as a function of block. SRT CRT CRT CRT O-distractors 4-distractors 8-distractors Block 1 (no pictures) 312.00 (72.44) 503.56 (52.98) 608.57 (69.87) 701.47 (93.00) Block 2 (distractor pictures) 275.54 (50.63) 464.81 (44.19) 553.37 (67.49) 617.01 (85.88) Block 3 (distractor pictures + ProM cue) 268.48 (35.28) 457.22 (41.93) 539.02 (61.67) 607.95 (85.35) Table 3.3. RetM performance (mean number of words recalled) as a function of task type and ProM Response. Task type ProM Response M SD N SRT Response 11.95 2.68 19 No 12.15 3.17 40 response Total 12.08 3.00 59 C R T Response 12.08 2.79 37 No 12.35 3.19 26 response Total 12.19 2.94 63 65 It can be imagined that the greater the increase/decrease, the easier it would be to spot the novel card, such as the onset of a 0- after an 8-distractor card, and that the majority of ProM responses in the SRT task occurred with this type of card transition. Unfortunately, the data needed for this analysis could not be extracted from the software output. Ongoing Task Performance Performance on the ongoing tasks was examined for two related reasons. One was to build a compelling case for the ongoing task effects on ProM performance. These effects are meaningful if subjects allocated all or most of their resources -whether in terms of speed or capacity- to the ongoing task. The second reason was to establish whether subjects' response to the cue was a true ProM response -a non-sought cue discovery— or whether subjects, when the pictures began appearing, engaged in a conscious search for the ProM cue. (The important conceptual distinction between ProM proper and vigilance was discussed in Chapter One). It was found that subjects' reaction times were comparable to published data and did not show evidence of slowing in Blocks 2 and 3, when pictures were presented along with the letter displays. Mean reaction times for SRT and CRT were computed for each subject as follows. First for each task type, extreme values, that is, the bottom and top 0.5% of all reaction times in each block were eliminated. Next, subject means across trials were computed by block, and in the case of CRT, also by card type (0-, 4-, or 8-distractor cards), and values more than three standard deviations above or below the individual means were discarded. Mean individual reaction times were calculated using the rernaining values in each block. Table 3.2 shows the group means for SRT and CRT by block. Reaction times for CRT are grouped by card type to allow for comparison with published data. Block 1 reaction times in all conditions are within one standard deviation of the reaction times on almost identical tasks, that Graf and Uttl (1995) recorded for a sample of young people (16-29 years). All reaction times in this study decreased in Block 2, when the distractor pictures were presented. Most importantly, all 66 reaction times underwent a small further decrement in Block 3, when the ProM cue appeared together with the distractor pictures. Block 3 reaction times are lower in all conditions than the reaction times found by Graf and Uttl (1995). This decrement of reaction times in Blocks 2 and 3 maybe at least in part the result of the auditory feedback subjects received in case of slow responses. These data lend substance to the conclusion that in both types of ongoing task, the ongoing activity was the focus of attention and effort. Moreover, they suggest that ProM, and not vigilance was measured. Subjects' responses gave no evidence of divided attention between the ongoing task and the ProM task; their performance became faster across blocks in all conditions, particularly in the critical Block 3. However, the possibility that subjects allocated a part of their attention to searching for the ProM cue was probed further by examining Block 3 reaction times for those subjects who did respond to it. For these subjects, mean reaction times before and after the ProM response were compared, on the reasoning that if they did search for the cue, evidence of slowing should be found prior to cue detection. No performance differences were found on the CRT task for any card type; for all tests, f/27) < 1.5. However, a small, statistically significant difference was found on the SRT task: reaction times prior to the ProM response averaged 270.51 ms, those following the ProM response averaged 262.56 ms, r(39) = 2.56, p <0.05. Thus it is possible that some subjects performing the SRT task may have looked for the ProM cue in Block 3, and that this may have contributed to the overall ProM advantage found for SRT. However, two factors speak against the latter possibility. One, more subjects in the SRT than in the CRT condition responded to the ProM cue, making statistical significance more likely. It is in fact possible that also some subjects in the CRT condition did allocate some of their attention to the pictures, in order to locate the cue. Two, the number of trials preceding the ProM response is not equal across subjects. On the average, subjects in the SRT condition performed a smaller number of trials prior to the ProM response, and slowing might simply indicate some type of "warming up" after the break between Blocks 2 and 3. 67 Performtrwe on the Retrospective Memory Measure The RetM part of the ProM task was to recall a list of words learned previously, either upon responding to the cue, or at the end of the ProM test phase, prompted by the experimenter. Table 3.3 shows the mean number of words recalled by task type and by ProM response. It is clear that performance on this component of the ProM task is neither affected by the ongoing task nor by whether or not the subject responded to the ProM cue. In fact, a 2 x 2 ANOVA with task type (SRT versus CRT) and ProM Response (response versus no response) revealed no significant effect or interaction. Correlations between Individual Difference Measures and Prospective Memory The ongoing task effects on the ProM measures strongly suggest that the capacity resource recruited for CRT has a role in enabling ProM. Therefore, individual differences in this resource should predict ProM. No a priori statement can be made about the speed resource recruited for SRT, but by intuition, individual differences in this factor should also correlate with ProM. Contrary to these expectations, the only individual difference measure found to correlate significantly with ProM was the search slope in the capacity-demanding task. Each participant's mean reaction time on SRT and on the three types of card stimuli presented in CRT, were taken as measures of the speed and the capacity resource, respectively. These measures were correlated with ProM response and Cue size. Correlations were computed separately for SRT and CRT because they might be affected by task type and because participants' reaction times were assessed at different times in the experiment. Participants in the SRT condition were first assessed on simple reaction times, then on choice reaction times. Participants in the CRT condition received these assessments in the opposite order. Test-retest reliabilities (Block 1 with Block 2; Block 2 with Block 3) were r = 0.58 and 0.65 for simple reaction times, r = 0.69 and 0.73 for choice reaction times with 0 distractors, r = 0.73 and 0.81 for choice reaction times with 4 distractors, and r = 0.77 and 0.79 for choice reaction times with 8 distractors. Choice reaction time 68 reliabilities are respectable and similar to those found by Graf and Uttl (1995). Simple reaction time reliabilities are more modest and lower than those computed by Graf and Uttl for a similar task. Table 3.4 presents the simple correlations between the different types of reaction time and ProM performance. The first four rows of this matrix present a true puzzle. Almost none of the expected correlations are significant. The only correlation that is, has a direction opposite to that expected: slower reaction times to 0 distractors are associated with a higher probability of ProM. Most surprising of all was the finding of a lack of correlation between ProM and the reaction times to cards with 8 distractors. These latter reaction times should be the most reliable index of the capacity resource, given the breadth of information that is processed, and thus should be the most predictive of ProM. Table 3.4. Correlations between ProM and resource measures by task type. Resource Measure ProM Response CRT (n=64) Cue size CRT (n=27) ProM Response SRT (n=59) Cue size SRT (n=40) Simple rt .03 -.26 -.03. -.11 O-distractor rt .28* -.07 .08 -.14 4-distractor rt .13 .12 -.03 -.09 8-distractor rt -.07 -.03 -.17 .03 Search slope . - .30* .00 -.34** .20 (ms/letter) * p<05 **p<.01 At this point, the notion began to take shape that the choice reaction task used in this experiment might measure two distinct processes, and that these might relate differently to ProM. These processes were the two operations required for the task: searching for the target letter, and 69 identifying the target letter/selecting a response. By this logic, reaction times to 0 distractors would reflect for the most part the identification/selection process, whereas reaction times to 8 distractors would reflect a combination of search and identification/selection processes. It is possible that only searching is systematically related to ProM, with higher efficiency in this process predicting higher levels of ProM. To isolate the variance unique to search processes, a search function was computed between 0- and 8-distractor reaction times, and its slope was taken as a direct measure of search efficiency. On the average, the slope was 25 ms/letter. The last row of Table 3.4 shows the correlations between this measure and ProM measures. The smaller the value of the slope, hence the higher the level of search efficiency, the higher was the level of ProM. This relationship was reliable for the binary ProM response measure but not for the Cue Size measure. As pointed out at the beginning of this section, Cue size measures were available only for subjects who did respond to the cue, that is, for 68% in SRT, 42% in CRT. When correlations are computed for corrected Cue size, in which missing values are replaced with the estimated value of 20, the r coefficients became similar in magnitude to those obtained for ProM Response, though of course of the opposite sign. The gain in information however is small, as this corrected scale for the most part reflects the binary ProM Response measure. Discussion The main objective of the experiment was to determine whether two ongoing tasks designed to selectively engage two distinct types of attentional resources, that is, processing capacity or processing speed, produced performance differences in ProM. It was found that ProM performance was lower when the ongoing task was an A/B choice reaction task, assumed to require processing capacity, than when it was a simple reaction task, assumed to require processing speed. The cost produced by the choice reaction task was evident in the two ProM measures, specifically, in a lower likelihood of responding to the ProM cue and a larger cue size required to make this response. No 70 cost was expected, and none was found, for the RetM measure -recalling a word list upon responding to the cue. There was no indication that subjects actively monitored the ongoing task stimuli for the ProM cue, in short, that they engaged in some form of vigilance or short-term ProM. Thus, this experiment provides evidence on the resource demands of ProM as defined at the outset of this work, that is, long-term ProM or ProM proper. Because the resource demands of the ongoing task were manipulated only by means of the particular activity subjects were to carry out with the materials, it can be assumed that the ProM performance differences are a direct result of differences in resource utilization for the ongoing task. To briefly review, capacity refers to the total amount of information the cognitive system can process at any given time whereas speed refers to the rate at which the cognitive system can process information. Capacity was operationalized as searching for, and discriminating, targets in letter displays whereas speed was operationalized as simple responding to the onset of the letter display. What is it about resource utilization in the task requiring capacity that is also crucial for ProM operations? How does resource utilization in this task damage ProM? The results suggest that ProM shares some attentional commodity with the choice reaction task, whereas it does not, or does to a lesser degree, with the simple reaction task. Phrased differently, ProM requires capacity but it does not, or not to the same extent, require speed. This is consistent with the theoretical understanding of capacity and speed as two complementary components of processing resources that each make independent contributions to cognitive operations (cf. Graf &Uttl, 1995). It is also consistent with the finding emerging from the ProM literature on task interference that ProM's resource demands are selective. Certain types of attentional demands, for example momtoring mcoming stimuli, interfere with ProM whereas others, for example rehearsing aloud strings of numbers, do not (Marsh & Hicks, 1998, Experiments 2 and 4; Otanietal., 1997). 71 The preceding interpretation of the results in terms of ProM's requirements of capacity and speed requires a cautionary note. It cannot be concluded from the findings that speed plays no role in ProM or even that in this specific experiment the speed-demanding task did not produce interference with ProM. In the ongoing task trials number of distractors was manipulated. This parameter is known to affect capacity (Graf & Uttl, 1995). However, no parameter known to affect speed, for example stimulus intensity (cf. Grice, Nullmeyer & Schnizlein, 1979), was manipulated. In an earlier section, I suggested that SRT trials consisting of sequences of similar letter displays (e.g., two 8-distractor displays) might have been more demanding than SRT trials consisting of sequences of different letter displays (e.g., 0 to 8-distractor displays). It is possible that the more demanding trials were associated with a lower frequency of ProM responses than less demanding trials but this possibility remains untested because the data needed for this analysis could not be made available. The remainder of the present research will focus on developing the role of capacity, but it must be recognized that the contribution of speed to ProM also deserves further investigation. In the context of ProM, capacity can be understood as the ability to process simultaneously ongoing events and cue events. The supply of this ability is limited and, when multiple ongoing events are processed, the cue event may be processed only in part or not at all. In the choice reaction task, cue processing was limited by ongoing processing in at least two ways. Subjects searched through ongoing events (the letters in the display), and this probably restricted perceptual processing. Furthermore, subjects identified targets and selected responses to the ongoing events, and this probably limited their working memory resources. The question therefore becomes whether the choice reaction task interfered with ProM at the perceptual or at the working memory level. There are two indications in the experiment that the perceptual process of searching through ongoing stimuli affected ProM. One indication is that the number of distractors in the ongoing letter display affected the lil^liliood of responding to the ProM cue. Higher distractor load in the display made responding more unlikely. The other indication comes from the correlational fmdings. High 72 search efficiency was associated with high levels of ProM. Taken together, these findings suggest that when searching through ongoing information is more laborious, whether by virtue of distractor presence or by virtue of individual differences, the discovery of the ProM cue is less likely. There is support from one study in the ProM literature that perceptual search processes are implicated in ProM. Maylor et al. (in press) investigated ProM in children aged 6 to 11 years. The children were presented with photographs of their teachers and asked to name them. The ProM task was to point out photos in which a teacher was wearing glasses, or photos that had a plant in the background. Though the plant in the background was actually larger in size than the glasses, younger children had much more difficulty responding to the background than the foreground cue, and this difficulty decreased as age increased. The effect of cue location in the young children and its disappearance in the older children maybe related to the development of the ability to conduct a systematic search of visual displays over this age range (cf. Vurpillot, 1968). What seems to change most dramatically as visual search abilities develop in children, are operations of disengaging attention from one item and moving it onto the next item (Enns, Brodeur & Trick, 1998). In sum, this experiment raises the possibility that ProM competes for resources at the perceptual level. The mechanism underlying this competition is assumed to be the limited supply of information-processing resources. It is however also possible that the competition is at the spatial-perceptual level. Studies on spatial attention have pointed out that greater perceptual load at the focus of attention tends to prevent attention from being allocated to peripheral areas (Lavie & Tsal, 1994; Williams, 1982). In the paradigm used in this experiment, the ongoing stimuli were centrally located, whereas the ProM cue appeared in the periphery. Subjects in the choice reaction task might have processed fewer aspects in the periphery, and thus missed the ProM cue. The two interpretations, in terms of information processing or spatial processing resources, are very similar and equally likely. However, in the context of the present research, the information processing 73 account holds a stronger interest and will be developed in Experiment Two. Spatial attention factors, though probably influential, will not be further addressed in this work. In this experiment the demands on identification and response selection processes were low and constant. Subjects had to match the target in the display to one of two items in memory, and select one of two responses. It is possible that these demands also interfered with ProM, and that a clear effect would be(found if the number of items in the memory set or the number of possible responses were increased. This possibility needs to be investigated prior to localizing the resource demands of ProM at the perceptual level. An even more important objection to the interpretation proposed here is that the results could be explained in terms of amount of resources not type of resources utilized. It could be argued that the choice reaction task required more resources than the simple reaction task, and that this more substantial resource expenditure left fewer resources available for ProM, thereby damaging performance. Although theoretically capacity and speed are not thought of as "more" or "less" than the other (Graf & Uttl, 1995), their operationalization in this experiment is open to a quantitative interpretation. Subjects performing the choice reaction task had to process more information, and carry out more operations, than subjects performing the simple reaction task, particularly when the distractor load in the letter displays was high. There is however one finding that speaks against this interpretation. In the pattern of correlations between individual differences in resource measures and ProM, there was only one resource measure (search efficiency) that reliably predicted ProM. If both tasks recruited resources from the same general pool, then all resource measures should be similarly related to ProM. This was not the case. The core idea emerging from this experiment is that ProM's resource demands are concentrated at the level of perceptual processing. It seems that ProM retrieval is problematic when people are inspecting a busy scene for a target object. This idea is tempered by the possibility, not adequately addressed in this experiment, that also heavier demands on working memory might make 74 ProM difficult. Moreover, this idea has to be assessed against a more parsimonious, quantitative view of ProM's resource demands. In the next experiment I will address these issues by examining separately the effects of perceptual and working memory load on ProM. To investigate the role of working memory, I will focus on processes of accessing and utilizing information in working memory, not on processes of response selection. 75 CHAPTER FOUR: SEARCHING AND MATCHING-TO-MEMORY Experiment One showed that the ongoing A/B choice reaction task,.assumed to require processing capacity, elicited lower levels oiProM performance than the ongoing simple reaction task, assumed to require processing speed. In addition, in the choice reaction task, the ProM response occurred more frequently when the cue was presented with a 0-distractor than with a 4-distractor or 8-distractor display. These findings implicate processing capacity resources in ProM retrieval and point to the perceptual processing of the letter displays as the likely source of interference with ProM's processing demands. To characterize ProM's resource demands at a more specific level, in this chapter I will focus on the notion of capacity and its constituent parts. The idea I aim to develop is that capacity constrains the processing of external and internal information, and that these constraints can be thought of as two distinct types of attentional resources. Accordingly, I propose to distinguish between resources that enable perceptual searching of the external environment to locate a target, and resources that enable access to the internal environment to match the target to an item in memory. I will use the term searching to refer to scanning external information and the term matching-to-memory to refer to searching and evaluating items in working memory4. This chapter will begin by relating the distinction between searching and matching-to-memory to theoretical views and empirical findings in the attention literature. It will then describe Experiment Two, which was designed to test the effects on ProM performance of ongoing demands on searching and matching-to-memory, and thereby evaluate ProM's demands on these resources. Attentional limits on perceptual search are well established. It has been consistently found that when targets and distractors in a visual search task have shared features, search times increase linearly with the size of the display, indicating that there is a limit in the amount of items that can be 4 The term matching-to-memory is used here to refer to the process of hxkingfor a match between a target and an item in memory, not to the product of finding a match. 76 processed in parallel (e.g., Neisser, 1967). Attentional limits have also been documented for accessing items in working (active) memory. When a memory search is conducted to classify a probe as a member or non-member of a previously memorized list, the response latencies increase linearly with the size of the list, thereby signaling that the items in the list cannot be accessed simultaneously (e.g., Sternberg, 1969). Thus there is evidence that both searching and matching-to-memory are attentional resources that the system can make available only in limited amounts. But are they distinct attentional commodities that sustain processing at different stages? Do they contribute independently to the operations of the cognitive system? From a theoretical standpoint, the idea that there is more than one limit or attentional resource involved in the processing of information from the sensory to the deep cognitive level is not new in cognitive psychology. Miller, in his highly influential 1956 paper, argued that human information processing had two different capacity limits. One was at the level of perception, and was defined by the amount of information people could discriminate amongst at a given moment. The other was at the level of immediate memory, and was defined by the number of items people could retain after these were presented. More recently, Pashler (1998) has suggested that processing is limited by an early perceptual filter and by capacity for late semantic analysis. Johnston, McCann and Remington (1995) have proposed to distinguish between two forms of attention, one operating on the perceptual input and one at the central-executive level. Few studies have directly investigated the relationship between perceptual and working memory resources but these studies provide evidence that searching and matching-to-memory involve distinct types of attention. Johnston et al. (1995) demonstrated that input or perceptual attention operates when central attention is tied up by a concurrent task. They employed a psychological refractory period paradigm in which subjects were given two tasks at various stimulus onset asynchronies (SOAs). Task 1 was a tone identification task, Task 2 a letter discrimination task. For the latter, subjects were required to determine on each trial whether a computer-presented letter 77 was an A or an H. There was an easy version of Task 2 where the letters were presented in their normal shape, and a hard version, where the letters were presented in distorted shapes. At short SOAs (e.g., 50 ms) response times on the easy and hard version were equally slow but at long SOAs (e.g., 600 ms) where response times were overall faster, there was a clear difference in favor of the easy letter discrimination. This indicates that when the two tasks were presented almost simultaneously and there was "slack" due to Task 2 queuing for central attention, input attention could operate on Task 2, and the extra work required by the hard discrimination was absorbed into the slack. However, when there was some time between the two tasks, no slack occurred, and therefore the extra work for the hard discrimination added to overall response times. Johnston et al.'s work indicates that perceptual processing is not affected by constraints on central processing. This evidence is consistent with the argument developed here but it is not sufficiently specific. Central attention comprises a range of processes, among which is memory search, but also response selection and initiation. Woodman, Vogel and Luck (2001) directly addressed the relation between visual searching and working memory load. Subjects had to identify one of two targets among 3, 7, and 11 distractors. On some trials, the search array was presented between a working memory study and test array (four items), such that searching was carried out under working memory load. On other trials, the search task was carried out between two blank screens, so that searching occurred under no working memory load. Working memory load slowed down overall response speed on the search task but it did not interact with search array size, that is, it did not affect the efficiency of the search process. This finding directly supports the idea developed in this chapter, that searching is carried out without the contribution of working memory resources. But there is one possible objection to this reasoning. The working memory manipulation in the Woodman et al. study involved storing items not utilizing them for the search task, and therefore it did not actually require matching-to- memory. The question arises whether the postulated 78 independence between searching and matching-to-memory would also be observed if items in memory would have to be accessed as part of the task. Rabbitt (1965), in a study on visual search speed and aging, manipulated both searching and matching-to-memory processes by varying the number of distractor and target items in a card sorting task. Young and older subjects sorted play cards with one target and 0,1, 4 or 8 distractor letters on them in two different conditions. In one condition, cards had one of two target letters on them and were to be sorted into two piles. In the second condition, cards had one of eight target letters on them and were to be sorted into eight piles. The increase in number of targets affected young and older people equally but the increase in number of distractors slowed older people's sorting times disproportionately, particularly with the small target set. Rabbitt suggested that older people utilize a wider range of cues to process information, and therefore can sample only small groups of distractors. This cue range is needed when searching for many targets, but redundant when searching for few targets as it reduces the ability to suppress irrelevant information. Rabbitt's findings suggest that people's ability to search visual stimuli declines with age whereas their ability to match targets to memory remains intact. Therefore they provide direct support for the idea that searching through distractor information and matching target information to items in memory are distinct abilities that are differentially affected by age. To conclude, theoretical views and empirical findings converge on supporting the distinction between two types of attention, one implicated in searching external information, the other in accessing and evaluating internal information. The interest now turns to how these two types of attention relate to ProM. Do they both have a role in enabling ProM retrieval, or does one contribute more than the other? On logical grounds one should expect ProM to require both types of attention. To regain awareness of an intention, a cue (target) event must be detected in the environment, and matched to an internal representation of the retrieval context. But Experiment One suggested a prominent role for searching. Distractor items in the ongoing task appeared to 79 interfere with ProM. Items in memory were not manipulated in the experiment but when individual reaction times were correlated with ProM performance, those indexing for the most part rmtching-to-memory and response selection processes did not predict ProM whereas those indexing search processes did. On this basis, the hypothesis was made that ProM makes higher demands on searching than it does on matching-to-memory. Experiment T w o Experiment Two sought to explore ProM's dependence on searching and on matching-to-memory by manipulating the demands of the ongoing task. The same dual task paradigm as in Experiment One was employed, with the ProM cues embedded in ongoing task trials making different demands on attention. Two types of manipulation were implemented. One manipulation targeted the searching demands of the ongoing activity, operationalized as the number of distractor stimuli in the search display that subjects must scan to locate a designated target stimulus. The other manipulation targeted the matching-to- memory demands of the ongoing activity, operationalized as the number of target stimuli subjects must hold in memory and respond to when they are in the search display. Subjects performed a target present/ absent decision task, with 75% of target-present trials. On these trials, the manipulation of distractor load was nominally equal to the manipulation of target load. Low distractor load involved the presentation of one distractor, high load the presentation of six distractors. Low target load corresponded to one potential target, high load to six potential targets. On target-absent trials (25% of all trials), low distractor load involved the presentation of two distractors, high distractor load the presentation of seven distractors. The four experimental conditions resulting from the combination of the two manipulations are detailed in Table 4.1. 80 Table 4 . 1 . Experimental conditions combining the manipulation of ongoing searching and matching-to-memory demands in a target present/absent decision-task. Low Distractor Load High Distractor Load Low Target High Target Low Target High Target Load Load Load Load Target present trials (75%) Stimulus Target letter Target letter Target letter Target letter No. of letters (A) and one distr. letter Two (A.B.C.D.E, or F) and one distr. letter Two (A) and six distr. letter Seven (A,B,C,D,E, or F) and six distr. letters Seven Target absent trials (25%) Stimulus No. of letters Two distractor letters Two Two distractor letters Two Seven distractor letters Seven Seven distractor letters Seven Experiment Two had three objectives 5. The first objective was to examine independently the effects of ongoing distractor load and ongoing target load on ProM performance. It was predicted that an increase in ongoing distractor load would reduce ProM performance. It was also predicted that an increase in ongoing target load would not per se affect ProM, but would add to the effect of increased distractor load by requiring a higher degree of serial, effortful searching. 5 A fourth objective will be discussed in Part II, Chapter Six. 81 The second objective was to assess the relationship between ProM and individual difference measures of searching and matching-to-memory. This objective was addressed to corroborate the effects of the ongoing task manipulations with correlational evidence. Accordingly, it was expected that measures of searching but not measures of matching-to-memory would predict ProM. The third objective was to utilize subjects' self-reports to evaluate the ProM test situation and the effects of the experimental manipulations by means of subjects' self-reports. The idea was to probe various aspects of ProM performance by means of a questionnaire administered at the end of the experiment, with a particular focus on subjects' use of strategies for the ProM task and their subjective experience of the repeated cue presentations. Method Participants and Design There were 80 participants, recruited from the University of British Columbia Psychology subject pool as well as by advertisement from other departments of the university. Subjects participated in the experiment in return for course credit or for a small amount of money. Most participants were undergraduate students in psychology or other areas. A few participants were graduate students in areas other than psychology, or people working at the university. There were 57 women and 23 men, between 18 and 57 years of age (M age = 23.6). The experiment employed a 2 x 2x2 design, with distractor load (low or high), target load (low or high), and type of ProM cue (butterfly or helicopter) as between subject factors combined in eight experimental conditions. As in Experiment One, type of cue was manipulated to establish the generality of the ongoing attentional load effects. Ten subjects were randomly allocated to each condition. Materials Materials and procedures in Experiment Two were identical to those employed in Experiment One, with the exceptions discussed below. The letters for the ongoing tasks were 82 presented in capital 28-point Helvetica font, bright green against black background. They were shown by means of a circular computer display measuring 7 cm across. The outline of the circle was not visible. The circular display was chosen to ensure that all potential letter locations were equidistant from a central fixation point. There were eight potential locations around the circle, corresponding to the clock times of twelve, one tliirty, three, four tliirty, six, seven thirty, nine, and eleven thirty. For target-present trials the following materials were prepared. In the low target load condition, the letter A appeared twelve times in each of the eight locations, for a total of 96 trials. In the high target load condition, each of the six potential target letters A, B, C, D, E, or F, appeared two times in each of the eight locations, again for a total of 96 trials. Distractor letters were drawn randomly and without replacement from the remainder of the alphabet, and occupied either one or six random locations in the display, depending on the search load of the condition. For the target-absent trials, distractor letters drawn randomly and without replacement occupied either two or seven random locations on the display, for a total of 32 trials. Figure 4.1 shows an example of an ongoing task display. Each trial began with the presentation of the letter display. The display remained on the screen for the duration of the response, plus a randomly determined response-stimulus interval of 750, 1000, 1250 or 1500 ms. Subjects received three blocks of trials. In each block, 96 target-present (75%) and 32 target-absent trials (25%) were randomly presented for a total of 128 trials. Subjects were instructed to place their right index and ring finger on the left and right arrow keys of the keyboard, and to press the left key if the (any of the) target letter(s) was present, the right key if the (any of the) target letter(s) was absent. The ProM task utilized the butterfly and helicopter pictures from Experiment One as cues. As in Experiment One, in the first block of ongoing task trials no pictures were presented. In the second block of ongoing task trials, quadruplets of pictures were presented with each target-present and target-absent trial. Figure 4.1. Example of the Ongoing Task Display. 83 -c \ Computer sc reen MB Y L K H I B H H | M S A tllHB •IIBSliltillltt 1HH1 •HflHI • H i Ongoing task display with target present in the high-distractor load condition 4 There were four different letter display/picture onset times, each randomly used on one fourth of the trials: +40, +80, +120, and +160 ms. Pictures were sampled randomly without replacement from a pool of 128 pictures representing common objects. Each picture appeared four times in Blocks 2 and 3, with the order in the quarter blocks randomly determined. In the third block of ongoing task trials, the ProM cue was presented up to nineteen times as part of a quadruplet as in Experiment One. Cue presentations began after the first 26 trials, and occurred on either target-present or target-absent trials at 2-, 3-, or 4-trial intervals over a total of 76 trials. There was no cue presentation on the last 26 trials of the block. As in Experiment One, the picture-cue measured 1.5 x 2 cm in height and width, respectively, on its first presentation, and 9 x 84 12 cm on its nineteenth presentation. On each presentation subsequent to the first, its measures increased by 0.4 x 0.5 cm. The ongoing tasks in this experiment were at least as demanding as the choice reaction task in Experiment One, where the percentage of ProM responses was 42%. To guard against the occurrence of floor effects, the presentation time of all pictures including the cue was increased from 200 to 400 ms. As in the previous experiment an important requirement was that subjects were engaged in searching and/or matching-to-memory while the ProM cue was presented. That is, cue presentation had to fall within the response window of the trial. It was estimated that the fastest responses in this experiment would occur in the low-target/low-distractor condition, and that these would be comparable to the responses obtained on the A/B decision task with four distractors in the prior experiment (M = 600 ms, SD =70 ms). According to this estimate, averaged across letter display/cue onset times, cue presentation was expected to end prior to the subject's response on approximately 85% of trials in this condition, and on 100% of trials in all other conditions. Procedure The ProM experiment was conducted in combination with an experiment on learning effects in creativity test performance. For this latter experiment, not reported in this dissertation, subjects received two creativity tasks, separated by a filled interval, and the creativity tasks were either the same or different. The portion of the creativity experiment that overlapped with the ProM experiment was the second creativity task, as this task was one of the filler activities administered between ProM instructions and ProM test. In this interval, subjects received either the Uses of Objects test or the Picture Completion test6. The duration of the two tasks was the same but the nature of the activities they involved was different. Because it is possible that activities intervening between instructions and test may affect ProM (cf. Kvavilashvili, 1987), half of the subjects in each of the four ongoing task conditions in the ProM experiment were given the Uses of Objects Test, 6 These tests will be described in detail in Part II. 85 the other half the Picture Completion test. The subjects receiving the Uses of Objects were assigned to the butterfly cue, the subjects receiving the Picture Completion Test to the helicopter cue. Two changes from Experiment One were introduced in the critical portions of the experimental procedure. One change consisted in mforming subjects at the beginning of the experiment that from time to time they would see pictures during the session and that at the end of the experiment their recognition of these pictures might be tested. This was done to induce in subjects an expectation of pictures as a natural part of the experiment and to prevent that their appearance in the second block of the experimental task might serve as an additional P r o M cue, thereby encouraging a monitoring strategy. The other change regarded the Re tM instructions for the P r o M task. Because Experiment One revealed that recalling the word list was not related to P r o M or affected by the experimental manipulations, the RetM requirement in this experiment was simplified. Subjects were instructed to point to the cue picture should they notice it. It was anticipated that this would make the P r o M response more distinct and avoid ambiguous responses such as saying the butterfly (helicopter) had appeared but not stopping the ongoing task. A t the beginning of the experiment subjects were informed that they would receive a variety of tasks, some concerned with the generation of ideas, some with attention, and some with memory. A l l tasks, in their order of administration, are listed in Table 4.2. After receiving instructions for a possible picture recognition test, subjects performed the first of two creativity tasks and the North American Adult Reading Test. They then received instructions for the experimental P r o M task. The P r o M picture cue (the butterfly or the helicopter) was displayed on the computer screen, and subjects were given the same verbal instructions as in Experiment One, excepting the requirement to recall the word list. They were asked to stop should they notice the picture cue again in the course of the testing session, and they were instructed to show the experimenter where they had seen the picture by pointing to it. To ensure understanding of the instructions, subjects were asked to repeat them in their own words. Instructions were repeated if subjects were unable to report them 86 correctly. FoUowing the ProM instructions, subjects were given two filler activities. The first was a personality inventory (the Dissociative Experiences Scale; Goldberg, 1999), the second either the Uses of Objects Test or the Picture Completion test. The two filler activities required from 15 to 20 min of time. The ProM test phase began with practice trials for the target present/target absent decision task. Subjects were told that they would see a letter display and that their task was to decide as quickly and as accurately as possible whether (one of) the target letter(s) was in the display or not, by pressing one of two keys on the computer keyboard. The exact instructions varied depending on the experimental condition. In the two conditions involving low target load, subjects were told to press the left arrow-key if the letter 'A' was among the presented letters, the right arrow-key if 'A' was not among the presented letters. In the two conditions involving high target load, subjects were told to press the left arrow-key if one of the letters 'A', 'B', 'C , 'D', 'E', or 'F' was among the presented letters, the right arrow-key if none was among the presented letters. When subjects were able to perform the task accurately, the first block of trials was administered. Prior to the second block subjects were informed that on the subsequent trials they would be shown pictures for the later picture recognition test. Specifically, they were told that four pictures would appear in the four comers of the screen along with each letter display, and that their task was to continue to perform the decision task as quickly and accurately as in the first block. To ensure that subjects continued to allocate full attention to the ongoing task, they received auditory feedback on trials where their responses were slower than their median response time in the first trial block. Prior to the third block in which the ProM cue would appear repeatedly among the other pictures, subjects were only reminded to continue their performance on the decision task with the same speed and accuracy. The experimenter recorded a ProM response and discontinued the presentation of the cue, when subjects stopped the ongoing task trials and pointed to the cue, or when they simply stopped the trials by moving their hand away from the keyboard or by turning toward the experimenter. Table 4.2. List of tasks and their order (core parts in bold). Task Type Materials Picture recognition test Instructions First Creativity Task Picture Completion from the Torrance Tests of Creative Thinking (Torrance, 1966) or Uses of Objects Test North American Adult Reading Described in Blair and Spreen, 1989 Test Experimental ProM Task Cue picture (butterfly or helicopter) Instructions Personality Questionnaire Second Creativity Task Experimental ProM Task Test DES (Dissociative Experiences Scale; Goldberg, 1999) Picture Completion or Uses of Objects Test Three decision task blocks, first block without pictures, second block with distractor pictures, third block with distractor pictures and ProM cue Story Task ProM Instructions Verbal Learning Task Two Immediate Recall Trials List of 18 concrete nouns Story Task - ProM Test Verbal Learning Task Delayed Recall Trial Story Task Recall Test Eight 100-word paragraphs adapted from A Study in Scarlet with a ProM cue embedded in paragraphs 2, 4, 6, and 8 (see Appendix A) Sentence-Paragraph Matching Task ProM Questionnaire Eight Items (see Appendix B) 88 In the final portion of the experiment, subjects received instructions for another ProM task, which will be described more in detail in Part II. Subjects were told that if in the remainder of the experiment they came across the names of animals, they should say them aloud. After these instructions, subjects were given two trials of a verbal learning task, in which they were presented with a list of 18 unrelated words, a questionnaire relating to their performance on the two creativity tasks, and a delayed trial in which they had to recall the word list. Next, subjects were asked to read a story adapted from the novel A Study in Scarlet by Sir Arthur Conan Doyle, in which four animal names were embedded (see Appendix A). The last three tasks in the experiment were, in this order, a delayed recall test of the word list, a recall test of the story, and a questionnaire assessing various aspects of subjects' understanding of, and performance on, the experimental ProM task (see Appendix B). The items on the questionnaire probed subjects' retention of the ProM instructions (item 1), their expectations (item 8), their reasons for not responding to the cue (items 2, 3 and 4) and the strategies they may have employed for the ProM task (items 5, 6, and 7). The entire experiment lasted about one hour and a half. Results Subjects' performance on the experimental ProM task was scored on the two measures utilized in Experiment One. One measure was the ProM Response which indexed whether the subject responded or not to the picture cue. The overall percentage of ProM responses was 62%. The other measure was the cue size at which the ProM response was given. The range of this measure was from 1 (smallest size) to 19 (largest size). As discussed in earlier section, the term cue size is used for convenience. The measure combines repetition and size, and therefore does not provide a direct index of the cue dimensions required for bringing back to mind the ProM task. Because the Cue size measure was conditional on a successful ProM response, it was only available 89 for 62% of the participants. For some analyses a corrected Cue size measure was used, with the value of 20 being assigned to those subjects who had not obtained a score on the variable. In the following sections first the effects of distractor and target load on ProM performance will be examined. These effects were analyzed with a 2 x 2 ANOVA with distractor load (low versus high) and target load (low versus high). Next, performance on the ongoing tasks will be examined by means of the same ANOVA model. Reaction times will then be correlated with ProM measures. Lastly, the sek>report items on the post-experiment questionnaire will be discussed. Self-reports were analyzed with chi-square tests and correlation coefficients, where appropriate. For all statistical analyses an alpha level of 0.05 was used to infer significance. E ffects ofDistractor and Target L cad on Prospective Memory The question of primary interest was how distractor and potential target load of the ongoing activity would affect ProM performance. It was found that high distractor load reduced ProM performance whereas high target load did not. A trend was observed for the effect of distractor load to occur with low but not with high target load. To analyze the effects of the distractor and target manipulation, it was first established that the data could be pooled across the two ProM-cue pictures employed in the experiment. Two ANOVAs with cue type (butterfly versus helicopter), distractor load (low versus high), and target load (low versus high) were conducted for ProM response and Cue Size. In both analyses neither the main effect of cue type nor the 2- and 3-way interactions reached significance. Figure 4.2 shows performance on the two ProM measures, pooled across cue type, in the main experimental conditions. When distractor load was low, the proportion of subjects responding to the ProM cue was 0.75 with low target load, and 0.80 with high target load. When distractor load was high, the proportion of ProM responses was considerably lower, with 0.35 with low target load, and 0.60 with high target load. For the Cue size measure, when distractor load was low, subjects on the average required between six and seven cue presentations with low target load, and eight presentations with 90 high target load. "When distractor load was high, average cue presentations needed to make a ProM response were eleven with low target load, and nine with high target load. The results consistently suggest the following pattern across the two measures. Overall distractor load reduced ProM performance, whereas overall target load did not. On the contrary there was a suggestion that ProM fared better with high than with low target load. The detrimental effect of distractor load was more evident in the low than in the high target condition. ANOVAs with between-subjects factors distractor load (low versus high) and target load (low versus high), conducted on ProM Response, Cue Size, and corrected Cue Size, bear out most of these observations. For the ProM Response variable, the ANOVA revealed a main effect for distractor load, i7(l,76) = 8.39, MSE = 1.80, p <0.005. There was no effect of target load, and, surprisingly the interaction failed to reach significance. For the Cue Size variable, the ANOVA yielded a marginally significant effect for distractor load, F(l,46) = 3.0, MSE = 65.37, p <0.10. Neither the effect of target load nor the interaction reached significance. Because the Cue Size measure was available only for subjects who did respond to the ProM cue, cell sizes were reduced and unbalanced. Thus, the ANOVA was repeated with the corrected Cue Size measure. This analysis revealed a significant effect for distractor load, F(l,76) = 11.15, MSE = 456.0, p <0.001, and again, no effect for target load. However, this time the interaction was marginally significant, F(1,7G) = 2.11, MSE = 86.11, p = 0.15. A ftest comparing performance at low and high target load when distractor load was high, failed to reach significance. However a significant performance decrement due to increase in distractor load was found only in the low target conditions, n(38) = 3.63, p <0.001, but not in the high target conditions. Given this finding, the effect of distractor load in the low and high target load conditions was evaluated also for ProM Response and for Cue Size. In both cases, performance differences between low and high distractor load were found when the target load was low but not when it was high (for ProM Response, <38) = 2.71, p O.01; for Cue size, t(20) = 2.41, p O.05). 91 Ongoing Task Performance Each increase in load -distractor and target- produced performance costs in the target present/absent decision task Moreover, the individual effects interacted to produce an overadditive performance cost in the condition that combined high distractor with high target load. Table 4.3 summarizes reaction times and decision accuracy on the ongoing tasks. Reaction times were screened with the same procedure used for Experiment One. For each experimental condition, type of trial (present or absent) and block, the top and bottom 0.5% of values were eliminated, individual means were computed, and values more than three standard deviations above or below the individual mean were discarded. Individual mean reaction times were then calculated on the remainder of values. The analyses will first focus on the effect of the experimental manipulations on ongoing task performance by examining group means for reaction times and accuracy on Block 1 trials. This was done to verify the effectiveness of the manipulations and to infer the workload each ongoing task involved. Next performance will be compared across blocks for all subjects and within Block 3 for those subjects that successfully responded to the ProM cue. These comparisons are critical for estabhshing that subjects did indeed allocate their resources to the ongoing task when the ProM cue began to appear, and did not divide their attention between performing ongoing task trials and monitoring the pictures for the ProM cue. Inspection of the reaction times on target present and target absent trials shows that the latter are overall between 20 and 40% slower than the former, with steeper slopes for target absent responses with increasing distractors (cf. Plude, Enns &Brodeur, 1994). But both sets of responses show an identical pattern of effects across the experimental conditions. For the purposes of the present analyses, the reaction times were pooled across trial type. Figure 4.3 plots mean reaction times for accurate trials and decision accuracy by distractor load and target load. Both types of increased load slowed reaction times to a similar degree. 92 Table 4.3. Mean reaction times (ms) and decision accuracy (percentage) for performance on the ongoing tasks as a function of block (standard deviations in parentheses). Low Distractor Load High Distractor Load Low Target High Target Low Target High Target Load Load Load Load Block 1 Target 501.88 673.10 647.22 979.20 present (56.67) (51.48) (60.75) (163.51) Target 590.07 852.90 805.91 1661.35 absent (62.27) (69.76) (96.79) (376.00) Accuracy 98.94 97.50 98.16 94.02 (3.30) (2.41) (1.39) (4.30) Block 2 Target 439.98 647.84 571.95 881.32 present (47.35) (193.20) (129.67) (251.62) Target 547.56 799.48 732.04 1311.49 absent (68.20) (212.89) (167.71) (461.47) Accuracy 94.92 92.66 93.44 89.26 (3.30) (5.50) (4.71) (7.02) Block 3 Target 411.58 582.79 536.65 782.95 present (47.35) (134.15) (129.67) (216.31) Target 525.28 745.71 710.33 1170.65 absent (70.46) (150.37) (179.75) (447.58) Accuracy 93.75 92.19 91.60 88.59 (4.54) (4.91) (5.09) (8.45) 93 Figure 4.2. ProM Performance (vertical bars indicate standard errors) as a function of distractor and target load. • Low target load • High target load Low distractor load High distractor load 94 F i g u r e 4.3. M e a n r eac t i on t i m e s o n a c c u r a t e tr ia ls a s a f unc t i on of d i s t r a c to r a n d target l o a d . • Low target load High target load 1400 1200 | 1000 </> a E c o o CO a: 800 600 \ 400 200 Low distractor load High distractor load Low target load - A - High target load 100 98 c o o a. > o re 1_ D O U < 96 94 92 90 J 88 Low distractor load High distractor load 95 Compared to subjects in the combined low load condition, subjects were slowed by 181 ms with high distractor load, by 217 ms with high target load. It can be estimated that subjects on the average required 36 ms to search each item in the display, and 43 ms to compare the located target to an item in memory. Moreover, the effect of high distractor load was exacerbated by high target load such that subjects in this latter condition were slowed by 774 ms. Because there is no reason to assume a change in matching-to-memory times, this slowing quite certainly reflects a sharp increase in the time subjects took to search each item in the display. The accuracy data are a mirror image of the reaction time data, with high distractor and target load producing a 1 to 1.5% drop in accuracy, and their combination producing a 5% drop. Two ANOVAs with factors distractor load and target load confirmed these observations. For reaction times, there were significant effects for distractor load, F(1,7G) = 141.95, MSE = 2722232, p O.0001, and for target load, (^1,76) = 171.38, MSE = 3286446, p <0.0001, and a significant interaction, F(l,76) = 36.99, MSE = 709444, p <0.0001. For accuracy, there were a significant main effect for distractor load F(l,76) = 13.17, MSE = 90.64, p O.005, for target load, F{1,76) =22.67, MSE = 22.67, pO.001, and a significant interaction, F(l,76) =5.28, MSE = 36.32, p <0.05. Reaction time and accuracy data converge on indicating that the distractor and target load manipulations successfully induced different ongoing demands across experimental conditions. Compared to the combined low load task, an increase in distractor load or in target load both increased workload and their combined increase did not simply add one workload to the other, but quite certainly produced extra work. As for task performance across blocks, reaction times in all experimental conditions became faster on Block 2 trials, when the distractor pictures were presented, and showed a further increase in speed on Block 3 trials, when the ProM cue was presented together with the distractor pictures. This would indicate that in all conditions, the ongoing task was the focus of attention and effort 96 throughout blocks, including the critical Block 3. However, decision accuracy was found to drop across blocks. The drop occurred for the most part from Block 1 to Block 2. Taken together with the findings for performance speed across blocks, this drop in accuracy suggests a shift from Block 1, to Blocks 2 and 3, in the tradeoff between speed and accuracy that subjects adopted. It is likely that this shift was due to the auditory signal subjects received on Block 2 or 3 trials that were slower than their median reaction time on Block 1. Subjects probably attempted to avoid hearing the signal by speeding up their responses and sacrificing some of their decision accuracy. This performance shift however does not mean that subjects were less engaged in the ongoing task trials in Blocks 2 and 3, or that they divided their attention between the ongoing trials and the pictures. An important additional test of subjects' task orientation comes from the comparison of reaction times prior to and following ProM responses for those subjects that made them. If reaction times are slower before than after responding to the cue, this could suggest that subjects were actively looking out for the ProM cue, and once they responded to it, resumed their original performance speed. This was not the case in any study condition. Performance differences, evaluated by means of t tests, were not significant. Corrections betwmlndhkktdDij^^ andPrnpectiveMcvnory It was found that when target load was low, performance speed on the decision task predicted ProM, with faster responses being associated with higher levels of ProM performance. However, when target load was high, performance speed was not predictive of ProM. Reaction times on Block 1 trials served as individual difference measures of resource availability. Unlike in Experiment One, in this experiment a single mean measure was obtained for each participant. Measures are not equivalent across study conditions, as the underlying rationale was that each condition involved the recruitment of two different resources in different amounts. Because the ongoing task effects clearly indicate that ongoing demands on searching but not on matching-to-memory affect ProM performance, the present analysis focused on identifying 97 individual difference measures most directly indexing searching and on testing their relationship with ProM. The reasoning followed here was similar to that used in Experiment One. It was assumed that reaction times associated with low demands on matching-to-memory would reflect for the most part searching, whereas reaction times associated with high demands on matching-to-memory would index for the most part this latter resource. Thus, reaction times from two experimental conditions -low target load with low or high distractor load- were considered an index of searching, whereas reaction times from the remaining two conditions -high target load with low and high distractor load- were considered an index of matching-to-memory. Test-retest reliability for these indices (Block 1 with Block 2; Block 2 with Block 3) were 0.76 and 0.98 for the index of searching, 0.82 and 0.82 for the index of matching-to-memory. Reaction times indexing searching predicted the ProM performance measures, r (38) = -0.41, p <0.01 for ProM Response and r (38) = 0.49, p O.05 for Cue Size. Reaction times indexing notching-to-memory did not predict ProM, r (38) = -0.14, p XX10 for ProM Response, r (38) = 0.08, p XJ.10 for Cue Size. These correlations are solid as shown by the two scatterplots in Figure 4.4 depicting the relation between reaction times and Cue Size with low and high target load. Since the two sets of correlations were computed for two different samples, there is however no assurance that a similar difference between correlations would be found with one sample assessed on both the searching and the matching-to-memory indices. Thus this finding can only be considered suggestive. Figure 4.4. Scatterplots depicting the relation between mean reaction times and Cue Size at low and high target loads. Low Target Load co E , CO CD E • c o ts ro CD c ro 0 800 700 600 H 500 400 Rsq = 0.2403 Cue Size 2000 High Target Load Rsq = 0.0069 10 13 16 19 Cue Size 99 Self-Reports mPmpcxtiwMerrxjryPcrfcmttnce K Subjects' self-reports on their ProM performance revealed occasional thoughts about the ProM task, a tendency to identify the correct retrieval context, and, for some, the implementation of a monitoring strategy. Most subjects who responded to the ProM cue reported having seen it one or more times prior to the response whereas most subjects who failed to respond to the ProM cue, reported not having seen it at all. The questionnaire administered at the end of the experiment probed variables pertaining to the retention interval as well as variables pertaining to the test situation. Retention interval variables will be analyzed to gain general information on the ProM situation in this experiment. These variables comprise Items 1, 6, and 8 on the questionnaire and address subjects' retention of the instructions, their expectations regarding the context in which the cue would appear, and the frequency with which they reminded themselves of the ProM task. Variables regarding the test situation will be analyzed to gain specific information on the effects of the experimental manipulations. These variables comprise Items 2, 3, 4, 5, and 7 of the questionnaire and reflect subjects' use of strategies for the ProM task, their subjective experience of the repeated ProM cue presentations, and the reasons they reported for not responding immediately (or not at all) to the cue. All subjects, regardless of their actual ProM performance, were able to recall the cue picture shown at instructions as well as the instructions themselves. Approximately half of the subjects (49%) expected the cue to appear during the decision task trials. Surprisingly, despite the accuracy of the expectation, these subjects were just as likely to forget as subjects who had no expectation or inaccurate expectations. Most subjects (66%) reported reminding themselves a few times of the ProM task. Only 34% of subjects reported not thinking at all or rarely about the task. Reminding had no effect on the likelihood of the ProM response itself, but it did predict cue size, r (75) = -0.54, p <0.001. On the whole, it appears that knowing when to expect the cue does not help with ProM 100 but thinking about the task from time to time does, by reducing the cue size required to initiate the appropriate response. Approximately half of the subjects (45%) indicated having used a strategy of some type for the ProM task. Strategies were of two different kinds. Task-general strategies did not focus on the specific ongoing task and comprised such activities such as "Always kept an eye on the computer screen", "Reminded myself of the butterfly prior to each exercise" or "Tried to keep looking out for objects colored orange [the color of the helicopter]". Task-specific strategies identified the decision task as the likely occasion of cue appearance. These strategies comprised responses such as "Moved my head back to see both the letters and pictures", "Divided my attention between the letters and a particular spot of the screen", or "Tried to separate the letters and pictures mentally". Type of strategy was related to ProM performance. Sixty-seven percent of strategy-users who made a ProM response (n=14), indicated adopting a task-specific strategy, whereas only 27% of strategy-users who failed to make a ProM response used this strategy (n=3), chi-square(\) = 4.50, p <0.05. There was no reliable association between type of strategy and experimental condition. The finding that some subjects used task-specific strategies and that this strategy use increased the likelihood of ProM success has important implications. It indicates that a small percentage of subjects (21%) in the study actively monitored the pictures for the cue by dividing attention between the ongoing task and the pictures. This does not call into question the effects of the experimental manipulations as these strategy users were distributed evenly across study conditions. However it does raise the problem of how a monitoring or vigilance orientation to the ProM task manifests, if it does not show up as measurable slowing on speeded ongoing task trials (see last paragraph of Ongoing Task Performance). What was subjects' subjective experience of the repeated cue presentations across Block 3? For those who made a ProM response, did they see the cue on some occasions prior to the one they responded to, and for which reason did they fail to respond immediately? For those who failed to 101 Table 4.4. Frequency of self-reported reasons for failing to respond (immediately or at all) to the ProM cue, by experimental condition and performance. Low Distractor Load High Distractor Load Self-reported Reasons Low High Low High Total Target Target Target Target Load Load Load Load Did not see cue ProM Response 2 0 0 0 2 No ProM Response 4 3 10 6 23 Total 6 3 10 6 25 Did not remember cue relevance ProM Response 1 2 1 1 5 No ProM Response 1 0 0 1 2 Total 2 2 1 2 7 Was unable to interrupt ongoing task ProM Response 0 4 1 1 6 No ProM Response 0 1 2 1 4 Total 0 5 3 2 10 make a ProM response altogether, did they see the cue one or several times, and if yes, what was it that prevented them from responding? Among the participants who made a ProM response, only 30% (n=13) thought the picture had appeared once, on the occasion they responded to. The majority (70%, n=31) thought it had appeared two or more times, with some subjects reporting appearances up to eight times. Among the participants who failed to make a ProM response, most (73%, n=22) claimed that the cue had not appeared at all, 23% (n=7) indicated that it had appeared two or more times, and only one subject thought it had appeared once. Table 4.4 lists the frequency of self-reported reasons for not responding immediately (or not at all) to the ProM cue for all subjects who failed to respond and for those subjects among responders, who claimed to have seen 102 the cue on more than one occasion7. The responses were grouped into three categories. Not seeing the cue comprised reports that the cue had not appeared, as well as reports that the subject was uncertain whether he/she had actually seen the cue. Not remembering the relevance of the cue comprised responses such as "I had forgotten all about the butterfly" or "It took me a while to realize that I had to respond". Difficulty intenupting the ongoing task included responses alluding to being too busy with the decision task, unable to stop the trials, or being too motivated to do well on the decision task. Not seeing the cue was the main reason for people who failed to make a ProM response (77%), being unable to interrupt the ongoing task the main reason for people who eventually responded to the cue (50%), chi-square(2) = 16.52, pO.OOl. There was no reliable association between reported reasons and experimental conditions. Thus, unfortunately, this crucial item of the questionnaire does not provide insight into the effects of the experimental manipulations. The finding that people who failed to make a ProM response claim the cue did not appear, is unexpected and somewhat in contrast with the majority of people successful on the ProM task claiming that multiple cues had appeared. This point will be taken up in the General Conclusions in Chapter Seven. 7Unfortunately only thirteen of these subjects actually indicated a reason for not responding immediately to the cue. 103 Discussion The main objective of this experiment was to examine the effects that ongoing searching, operationalized as distractor load in a target present/absent decision task, and ongoing matching-to-memory, operationalized as target load held in memory for this task, had on ProM performance. It was hypothesized that increases in distractor but not target load would damage ProM performance. It was also hypothesized that an increase in target load would add to the effect of distractor load and augment the cost for ProM. The findings were consistent with the first but not with the second hypothesis. Increased distractor load reduced ProM performance, whereas increased target load did not. However, distractor load tended to disrupt ProM more substantially when target load was low than when it was high. The performance decrement affected the overall likelihood of responding to the ProM cue as well as the size of the cue at which the response was made. For most subjects, performance indexed ProM proper, that is, regaining awareness of a previously formed intention upon encountering a cue, and the ensuing discussion will focus on what this experiment suggests about the resource demands of ProM proper. However, there were a few subjects in each experimental condition that reported dividing their attention between the ongoing task and the ProM task. For these subjects performance indexed a state of vigilance rather than ProM proper. This raises new issues about distinguishing operationally between ProM proper and vigilance tasks, which will be discussed in the General Conclusions in Chapter Seven. The centerpiece of this experiment is the specific nature of the ongoing task demands that interfered with the ProM task. ProM was disrupted when the ongoing task demanded more attention in terms of searching, but not when it demanded more attention in terms of matching-to-memory. This means that with regard to ProM the critical factor in this experiment was the type of attention not the amount of attention recruited for the ongoing task. The interpretation must necessarily focus on the type of resources ProM demands, and does not have to contend with a quantitative account, as was the case for the previous experiment. 104 This experiment found that the presence of distractors in an ongoing search task affected ProM. There is no related finding reported in the ProM literature. However there is corroborating evidence from within this set of experiments. The effect of ongoing distractor load in this experiment dovetails with the observation made in the previous experiment that ProM responses were more unlikely when the concomitant distractor load in the letter display was high than when it was low. The consistently observed relationship between individual differences in search speed and ProM provides converging evidence: Subjects that are faster at searching letters are also more likely to respond to the ProM cue. The whole of this evidence pinpoints perceptual processing as a major source of interference between ongoing activities and ProM. How does interference at the perceptual level come about in the context of ProM retrieval? What type of attention might the ongoing and ProM task have competed for? The thinking that has been pursued to this point focused on the amount of items that can be searched in parallel. This amount is constrained by the limited supply of mformation-processing resources. In the low distractor load conditions, subjects had to search only two items in the context of the ongoing task and sufficient resources were available for processing additional items, including the ProM cue. In the high distractor load conditions, subjects had to search seven items, and insufficient resources were available to process additional items such as the ProM cue. Because the ProM cue could not receive adequate processing in the high distractor load conditions, its discovery was less likely. However, this account encounters a problem. It is known that the set of items that can be processed in parallel becomes smaller when the analysis performed on each item is more detailed (e.g., Rabbitt, 1965). The two high distractor conditions were not equal in this respect: people looking for one of several target items (high target load) had to carry out more extensive processing of each item in the display than people looking for a single target (low target load). By this logic, fewer resources would have been available for the ProM cue in the former than in the latter condition, with a comparatively greater cost for ProM. This was not the case. On the contrary, a trend was observed for distractor 105 load to interfere more substantially with ProM when subjects were looking for a single target than when they were looking for multiple targets. This finding is suggestive of a more complex interference mechanism that centers on the difference between single-target and multiple-target searching. An account of this mechanism will be developed in the remainder of the chapter, but it should be considered speculative as the statistical support for the interaction between distractor and target load was not conclusive 8. Reconsider the key observation Rabbitt (1965) made in his study on age effects on visual search tasks. Distractor load particularly affected older people's card sorting times compared to those of young people when the cards had to be sorted by two, not by eight, target letters. Rabbitt's reasoning was that there was an age difference in the ability to process irrelevant information. Young people adapt the processing to the number of relevant items whereas for older people processing remains the same. When searching for a small number of relevant items, young people conduct the search by suppressing large chunks of irrelevant information without further processing, whereas when searching for a large number of potentially relevant items, they proceed by processing to some extent also the irrelevant items. Older people have difficulty ignoring irrelevant items in the search task and therefore always process to some extent also the irrelevant items. Rabbitt's characterization of search strategies in young people can be applied to the tasks employed in this experiment. It is likely that subjects who searched for an 'A' among six distractor letters, utilized their resources to suppress the distractor letters, thereby making the 'A' pop out. Subjects who searched for an 'A', 'B', 'C , 'D', 'E ' or 'F' among the distractor letters directed their resources to all letters in the display. The substantial slowing of search times in the latter compared to the former condition supports this suggestion. Casual observations collected in the experimental setting are also consistent with it: subjects searching for the single letter seemed to employ one long 8 Analyses of variance yielded a marginally significant distractor x target interaction only for one of the three ProM measures. However, for all three measures t tests of ProM performance at low and high distractor load were significant only in the low, not in the high target load conditions. 106 gaze and then proceeded to the keypress, whereas subjects searching for multiple letters seemed to employ short gazes that followed the perimeter of the display. If the pattern of distractor-related interference with ProM is re-examined in light of the two different search strategies, the foUowing argument can be made. ProM shares a commodity with target searching, and likely to a greater extent with single- than multiple-target searching. This commodity is the ability to suppress mcorning information that is irrelevant to the task. Suppressing information and thereby making the cue stand out might play an important role in ProM retrieval. As argued several times throughout this work, the ProM cue is a feature seamlessly fitting in the ongoing scene. The cue does not stand out, that is, it does not direct attention to itself. Conceptually, it would stand to reason that to recognize the cue as a special and distinctive item in the ongoing context, subjects have to work on the other items that are in it. To become aware of the cue, they must disregard non-cue items and interrupt the processing of these items. They must blot them out. It can be imagined that the suppression of non-cue information is particularly demanding in ProM situations, as it is this very information that is processed as relevant in the ongoing activity. Therefore even small individual differences in the ability to suppress information should matter, and this is consistent with finding of significant correlations between search speed and ProM in a relatively homogenous sample of young subjects. Subjects' self-reports in this experiment also fit this picture. Most successful subjects reported seeing the picture cue several times, before they recognized its relevance and responded to it. Their greatest difficulty was in ignoring the ongoing task stimuli. This type of testimony underscores the different time course of simply seeing and recognizing the cue, the latter being a slow, effortful process. It is also possible that the failure of experiencing the cue as a distinctive element led unsuccessful subjects to report not seeing it at all, though this type of report is open also to different interpretations that will be discussed in the General Conclusions, Chapter Seven. 107 The general idea emerging from the interference between ongoing demands on searching and ProM is that ProM requires selective attention resources. There are several arguments to support this idea. First, it has been proposed that breakdowns in inhibitory functions have a role in age-related declines in cognition and memory (Flasher & Zacks, 1988). If ProM requires efficient inhibitory processes, then older people should have difficulty with ProM tasks, and this has been found to be the case across a variety of ongoing activities, as discussed earlier in this dissertation. Second, clinical populations that are known to have selective attention deficits should do poorly on ProM tasks, and this has indeed been reported for children with attention-deficit/hyperactivity disorder (ADHD) (Kerns, 2000). Third, ProM should be facilitated when cue processing and ongoing task processing overlap as in this case the need for suppressing ongoing information is substantially reduced. This is consistent with the existence of concurrent transfer appropriate processing effects on ProM. Meier and Graf (2000) observed higher ProM performance when the processing required for the cue overlapped with the processing required for the ongoing activity, for example perceptual processing of word attributes, than when it did not overlap -for example semantic ongoing processing but perceptual processing for the cue. While ProM clearly shared resources with searching, there was no indication that it did so with matching-to-memory, that is, the other type of resource examined in this experiment. Matching-to-memory was defined as accessing internal representations and comparing relevant mcoming items to them. It was operationalized as the set of targets to be searched in the present/absent decision task. Increased target load did not affect ProM and in fact there was a nominal advantage for ProM in the high target conditions. The target-load manipulation requires comparison with similar studies in the ProM literature. Two studies discussed in detail in Chapter Two found that attentional load involving working memory resources interfered with ProM performance in young people. Einstein et al. (1998, Experiment 3) found interference when subjects performed a secondary ongoing task monitoring a 108 string of digits for the occurrence of three consecutive numbers. Marsh and Hicks (1998) found interference when subjects performed secondary ongoing tasks requiring (a) monitoring a visual display and counting its items according to specified rules, (b) generating random numbers with values from 1 to 10, and (c) tapping a spatially represented sequence of keys. The Einstein et al. study as well as task (a) in the Marsh and Hicks study involved searching a visual or auditory display as well as matching-to- memory and it could be argued that it was the search component that gave rise to interference with ProM. But this cannot explain the interference from tasks (b) and (c) in the Marsh and Hicks study. Despite obvious differences between those tasks and the task employed in this experiment, to the extent that all rely to some degree on working memory resources, the results are in conflict. One possibility is that the manipulation of matching-to-memory demands in this experiment did not engage working memory resources. In the high target load conditions subjects had to hold in memory a target set that comprised the six initial letters of the alphabet. This target set remained constant across trials. It is possible that because subjects became highly familiar with this set and highly practiced in matching mcoming items to it, on later trials they no longer had to search the target set to determine the presence or absence of target items in the search display. In this way they may have had to rely only minimally on working memory resources in the last block of trials, when the ProM cue was presented. However the effect of target set on subjects' reaction times remained constant across blocks of trials, and this finding speaks against the possibility that demands decreased over trials. The extra matching-to-memory work in the high target load condition was as necessary for early as for late trials. This observation is consistent with Sternberg's early observations on memory scanning (Sternberg, 1969). He found that subjects used the same memory search process regardless of whether a varied-set or fixed-set procedure was employed. Another possible explanation for the discrepancy between the present findings and those reported by Marsh and Hicks (1998) regards the involvement of response selection and execution 109 processes. The ongoing tasks that produced interference in the Marsh and Hicks study were highly demanding in terms of these processes. For example, the spatial tapping task required subjects to type a sequence of digits that formed a diamond-in-a-square, alternating between clockwise and counterclockwise direction. By contrast, in the present experiment, regardless of target load, the target present/absent decision task always required choosing between two possible responses. These observations suggest that the conflict between ongoing and ProM demands on working memory maybe strongest at the stage of selecting and coordinating responses. By this logic, the target manipulation in this study might have produced interference had it included mapping each target to a different response. Up until now the reasoning has been that the matching-to-memory manipulation in this experiment did not tax working memory to a sufficient or critical extent, and therefore did not produce a cost for ProM. There is however a more mtriguing account that builds on the assumption that the matching-to-memory manipulation had effects on the test situation above and beyond diverting resources away from ProM. One effect might have been an increase in the level of effort and watchfulness that subjects brought to the ongoing decision task. It is likely that subjects in the high target conditions found the decision task more difficult than did subjects in the low target conditions. There was for the former a higher degree of uncertainty associated with the task than for the latter. High target load meant that on each trial the display could have one of several target letters in it or it could have none. When one target was not spotted, there was still uncertainty as to the others. Low target load meant that on each trial the display could have the target in it or not. When the target was not spotted, the uncertainty was over. It is therefore possible that subjects in the high target load conditions summoned more resources for processing and that these extra resources also benefited ProM. The matching-to-memory manipulation might have affected ProM also by way of transfer of processing (cf. Meier & Graf, 2000). Subjects in the high target conditions were engaged in searching 110 for internal targets to which the information on each trial could be matched. Their attention was directed to internal events and focused on finding items in memory. It can be said that the need to repeatedly access multiple internal targets maintained a readiness to search memory, in short, it placed these subjects into a retrieval mode (cf. Tulving, 1983). Several authors in the ProM literature have emphasized that the ProM test situation is unlike any RetM test situation in that the most important challenge is for the subject to become aware at all that a memory search must be carried out (Graf & Uttl, 2001; McDaniel & Einstein, 2000; Winograd, 1988). In other words, in the ProM test situation the system is not placed by some external agent into a retrieval mode, and must somehow switch over to this mode by itself (cf. McDaniel & Einstein, 2000). Maybe, the processing stimulated by the matching-to-memory manipulation facilitated switching into retrieval mode for the ProM task, thereby giving subjects in the high-target conditions an advantage over those in the low-target conditions. To summarize, the experiment indicates that ongoing searching interferes with ProM. This finding suggests that the perceptual processes involved in ProM make robust resource demands. These processes seem to require attentional resources to segregate the cue item from irrelevant items in the ongoing scene and to make it stand out. The presence of multiple distractors in the ongoing task seems to curtail the adequate infusion of resources for these processes. What resources are required once the cue is singled out and enters the system is not clear from this experiment. Ongoing matching-to-memory did not interfere with ProM but it is possible that this does not signal ProM's independence from working memory resources. Instead ongoing matching-to-memory may have induced a higher level of watchfulness in subjects, or facilitated similar processing required for ProM retrieval. Conclusions to Part I This part of the present research aimed to characterize the type of resource or attention that must be available for ProM retrieval. There is a solid theoretical and empirical basis for assuming 111 that ProM is an attention-demanding process, but to date only one study (Marsh & Hicks, 1998) has attempted to systematically define these demands within a model of attention (Baddeley's working memory model; Baddeley, 1986). This study found that ongoing task demands on the central executive, but not on the storage components, interfered with ProM. The present work approached ProM's demands on attention from a resource perspective and asked which basic attentional input or fuel sustains ProM retrieval processes. Experiment One investigated the role of processing capacity and processing speed. Experiment Two narrowed the investigation to searching and matching-to-memory. The two experiments consistently showed that ongoing demands on attention interfered with ProM when they imolzed searching through distractor items. This effect was particularly detrimental when the search was for a single target. Both experiments also revealed that individual differences in search efficiency predicted ProM performance. People quick at searching through the letter displays in the experiments tended to be more successful on the ProM task than people who required more time. The collective evidence suggests that ProM requires the intervention of attentional resources for early perceptual processing of the ProM cue amidst ProM-irrelevant items. How can these resources be characterized, and how, by what mechanism, do they contribute to the retrieval of intentions? It was speculated that the resources revealed in the present research coincide with selective attention, and specifically with its inhibitory functions. These resources dampen or suppress the processing of some items so that others become salient. In this way they enable the process Kahneman (1973) described as figural emphasis. The observer perceives some things with a "tiling-character'', others with a "stuff-character". Figural emphasis makes things stand out regardless of their intensity. The observer's subjective experience of something standing out has in fact been referred to with the exquisite term "attensity" (Titchener, 1908; in Kahneman, 1973). A parent can spot a child with high attensity in a school photo of eight hundred people. 112 All this appears very relevant in the context of ProM. It can be imagined that selective attention resources enable the retrieval of intentions by emphasizing the cue event over non-cue information and by producing the psychological experience of attensity. In this way the prospective requirement of intention retrieval is satisfied: the cue is discovered as something special that requires conscious probing, and the retrospective part can begin, for example, asking oneself why this particular event is relevant, what it should bring to mind (cf. Graf & Uttl, 2001). Is the idea of ProM as a perceptually demanding process compatible with the conceptual definition of ProM developed at the outset of this work? Is it consistent with the theoretical models of ProM retrieval that emphasize its attentional demands? The answer to the first question is a definite yes. ProM was defined as the discovery of cue events in the ongoing context that signal the need to recall an intention. It was distinguished from RetM, which involves the recovery of information in the presence of an explicit cue. For ProM the system must engage in operations of looking outward, discriminating, parsing, and categorizing incoming information. This is tantamount to saying that ProM is essentially a perceptual operation. For RetM the system must engage mainly in operations of looking inward, searching and reconstructing encoded information. This coincides with the traditional understanding of memory processes. The aids and reminders we use in our everyday life reflect this distinction. To ensure ProM, we create conspicuous perceptual cues: A post-it note on the computer, a library book sitting on the door knob, a clock time written in red ink on the back of our hand. To ensure RetM, we create a record of the information we need: A line in a notebook, a number in the telephone rubric, a map of a hard-to-locate place. The proposal that ProM demands attention for perceptual processing agrees with the resource-mediation models of ProM retrieval in saying that ProM cannot occur without attention, but it does not map neatly onto these models. Craik (1986) underscored the reliance of ProM on processing that is self-initiated and therefore highly demanding on attentional resources. Though he did not indicate the particular level or levels at which self-initiated processing comes into play, he 113 described this processing as effortful, voluntary and intentional, and viewed it as crucial for ProM and certain RetM tests, such as free recall. In this way he seemed to have in mind strategic, organizing, and self-regulating processes characteristic of the central executive, which is located deeply in the cognitive system. Also Shallice and Burgess (1991, 1997) indicate central executive involvement in ProM when they postulate the intervention of supervisory processes to redirect routine processes toward the ProM task. According to these authors the chain of events is the following. At the time an intention is formed, a marker of a specific future circumstance (for example, when going by the comer store, when seeing Marianne) is set up. The occurrence of this circumstance triggers this marker, which in turn activates supervisory processes. Clearly a marker is what the present work understands as the ProM cue. Thus, according to Shallice and Burgess, attention involvement is triggered by the ProM cue, and not, as the present work suggests, needed to notice the ProM cue in the first place. Their model is very similar to the Noticing + Search model of ProM retrieval, developed by Einstein and McDaniel (1996) prior to the multiprocess framework reviewed in Chapter Two (McDaniel & Einstein, 2000). According to the Noticing + Search, the encounter with the ProM cue automatically elicits a feeling of familiarity: the ProM cue is noticed as something special among ongoing events and thus gamers attentional processing. By contrast the present work has provided evidence that attention is necessary in the very early stages of the cognitive events leading to the retrieval of an intention. The present work suggests that attention is required to notice the ProM cue or marker, and that the mechanism underlying noticing is one of selection or figural emphasis. In this way the ProM cue stands out while other things fade away. The psychological experience associated with noticing the cue is characterized in terms of attensity, not familiarity, as postulated by the Noticing + Search model. This work is not the first to claim that noticing the ProM cue does indeed involve attention. West et al. (2000) conducted an investigation of the electrophysiological correlates of ProM, which 114 was reviewed at the end of Chapter Two. These authors found that there were differences in ERP modulations between successful and missed ProM trials. A phasic negative effect over the occipital and parietal regions 300 ms after cue onset was noted only on successful trials. The cue information (matching colors) presented to the subject was identical on both types of trial, but clearly its processing was not. West and colleagues argued that on successful but not on missed trials there was attentional activation of cue-specific perceptual processes. Interestingly, Marsh and ITicks (1998) also argued that the noticing component of ProM tasks requires attentional input, but attributed this requirement to the activation of a rule ("if fruit name respond") rather than to the perceptual process of emphasizing the word in question. The present work did not find evidence that ProM requires working memory resources. There are good reasons to assume that the task of matching stimuli to multiple internal targets which was given to subjects in Experiment Two, did involve working memory resources but it did not interfere with subjects' ProM performance. As noted earlier, this stands in contrast with the findings reported by Marsh and ITicks (1998). This contrast cannot be settled conclusively in the present work but two types of explanation were developed. It is possible that ProM's demands on working memory conflict with those of ongoing activities for processes of selecting and executing responses. It is also possible that the ongoing utilization of working memory for rnatching-to-memory processes may indirectly benefit ProM rather damage it by curtailing resources. PART II: PROSPECTIVE MEMORY AND CREATIVITY 116 CHAPTER FIVE: T H E RELATION BETWEEN PROSPECTIVE MEMORY AND CREATIVE THINKING ProM is something of a paradox. When the time comes to follow up on an intention, our thinking is focused on ongoing activities. We process events in relation to these activities, not in relation to the to-be-remembered intention. For example, we might plan to pick up an ingredient for dinner on our drive home. Assume the store of our choice is located at an intersection. When we reach that intersection in the evening, we check the traffic lights, signal for pedestrians to cross, keep an eye on the biker to our right and try to get a head start over the driver to our left. In short, we respond with our mind set on traffic. No element in this scene prompts thoughts about food and cooking. How is it possible, then, that more often than not ProM prevails? How do we succeed in thinking about Parmesan cheese when we look at a traffic light? In Part I of the present research I examined ProM from a bottom-up perspective. In Part II, I intended to gain insight into top-down processes involved in ProM. The above example suggests that the way we categorize things and think about them might have an important role in ProM. If we always think of a traffic light as a traffic light, ProM might have little chance. How could the traffic light signal anything but traffic-related actions? If, on the other hand, we are able to think of a traffic light as rolling pin, we might well make the leap to food and cooking, and ProM is achieved. Consequently, in this part of the present research I view ProM as a special type of problem-solving situation, in which the problem-solver must realize that exceptional not conventional measures are called for. It might be that to meet this challenge we have to be open-minded, that is, to think creatively. The goal of Part II is to examine the relationship between ProM and creative thinking. The possibility that ProM might be related to creative thinking has not been explicitly suggested in the literature. However, several authors have characterized ProM in terms of shifting thought or decoupling thought from habit and these characteristics are strongly reminiscent of 117 creative thinking. Graf and Uttl (2001) note that "the capacity to see beyond the obvious" is fundamental to recognizing ProM cues. They compare ProM to the task of a prospector, a professional trained in discovery who monitors the environment for telltale signs of underground mineral deposits. Such a specialist would be alert to the fact that some but not other river beds or rock formations may be worth inspecting further and would be prepared to shift from broadly scanning the grounds to closely probing specific spots. Schonfield (1982) views ProM as internal attention switching between distinct trains of thought. He describes attention switching in terms of moving from one conscious translation of a symbol or a piece of information to another. Interestingly, he suggests that such conscious translations become more discrete with aging, and that therefore the transition from one translation to another may become more difficult for older people. In one incidental ProM experiment (in Welford, 1958), Schonfield asked people to press a button on each trial when they were ready to give a verbal answer. The button press was required to mark the time to come up with the answer. The experiment did not work out because people consistently forgot to press the button and measures of response latency were obtained only on few trials. Schonfield notes that this difficulty arose because people, especially the older participants, had difficulty switching between the two discrete translations of the task: providing a verbal answer and pressing the button. Wilkins (in Wilkins & Baddeley, 1978) developed a "random walk" model to account for the timeliness of ProM responses in a simulated pill-taking study. In this model a train of thought is assumed to be traveling in a statistically random manner through a multi-dimensional semantic space. Different areas in this space represent different types of activity, for example, cooking, going to work, and piU-taking. The closer the train of thought is to the area associated with the intended activity, the more likely that the subject will remember it. It can be speculated that not only random factors, but also ongoing activities and the subject's thmking style may influence where the train of thought is located and how fast it can move from one area to another. 118 Ellis (1996) argues that the ProM retrieval context typically triggers actions other than the intended one, actions that are highly practiced and handled by lower-level cognitive processes. In her view, ProM can only occur when the exceptional demands of the context are recognized and supervisory control processes are activated to enable the execution of non-routine actions. In a similar vein, Reason (1984) draws attention to the detrimental effects of highly habitual action sequences on remembering intentions. He cites the example of intending to take medication when getting ready to leave the house. Once the sequence of leaving the house is engaged, it is likely to run its familiar course. It overrides the intention to take medication not only at the motor level, but also at the level of tliinking and perceiving, which follow a routinized, predictable path. Alan Baddeley (1990) reports a personal instance of prospective forgetting that illustrates Reason's points well. Baddeley was invited to a phone-in radio program scheduled early in the morning on a workday. On the appointed day, he went about his morning routine, and then sat down to read the morning paper. When he looked at the television programs for that evening, he was reminded of radio, and realized he had forgotten the broadcasting appointment. It appears from this episode that for some time on that morning the routine sequence of getting ready for the day controlled Baddeley's actions and thoughts. Only when he began to read the paper, were his thoughts free to wander again, and recover the forgotten radio appointment. In sum, these speculations suggest that certain qualities of thinking may be privileged routes to ProM. One can roughly distinguish between the ability to think flexibly, that is, to consider more than one mental context or category (e.g., Graf and Uttl, 2001) and the ability to think differently, that is, to reject habitual or rote ways of thinking (e.g., Ellis, 1996). These abilities present substantial overlap with abilities thought to underlie creative thinking. The following section will define creative thinking for the purposes of the present work, and articulate the specific components of creative thinking that might be relevant to ProM. 119 Defining Creative Thinking What do we mean when we say a person thinks creatively? Do we mean the person is creative, by virtue of producing something that others would recognize as creative, or do we mean that he/she employs thought processes that may lead to being creative? The former perspective focuses on creativity as an achievement and attempts to distinguish the unique characteristics that make up a creative person, and his/her historical context. This perspective understands creativity as something exceptional, that few people "have". It assumes a discontinuity between the creative and non-creative (e.g., Gruber & Davis, 1988; Runco, 1991). The latter perspective focuses on creativity as a process and attempts to identify the mental abilities that sustain it. It assumes that creative thinking relies on normal cognitive abilities that all individuals have to some extent, and that are continuously distributed in the general population (Guilford, 1950; Torrance, 1988). The present research understands creative thinking from a process perspective, and focuses on the cognitive abilities that contribute to creative tliinking. What are these abilities? Guilford (1959, 1975) identified them with a group of abilities he named divergent thinking. In a broad sense, divergent thinking is thinking in different directions and coming up with alternatives. The divergent thinking construct has intuitive appeal in the context of the reflections on ProM developed in the previous section. Also, its assessment is straightforward and objective (Guilford, 1950; Wallach & Kogan, 1965). For these reasons I chose to consider creative thinking as synonymous with divergent thinking in this work, and to use these terms interchangeably. However I am aware that Guilford himself was cautious not to equate divergent thinking with creativity. He understood divergent thinking as a key ingredient for creativity and so do creativity researchers today. Runco (1991) proposes that divergent thinldng are part of the process that leads to creative achievement and that tests of divergent thinking can provide an estimate of creative potential. Furthermore I recognize that creativity encompasses a broader set of cognitive skills such as the ability to discover problems (Campbell, 1960; Guilford, 1968; Torrance, 1988), to tolerate cognitive ambiguity (Kuhn, 1963), or 120 to evaluate ideas (Guilford, 1959; Torrance, 1988). Moreover, the consideration of other personological as well as environmental factors (Amabile, 1996) would be required for a full understanding of creative thinking. Guilford (1959) coined the term divergent thinking in his structure of intellect model, in which he proposed to distinguish between five major groups of intellectual operations. Cognition and memory operations are concerned with coding and retaining information, evaluative operations with judging information. The operations of interest here are convergent and divergent thinking operations, both concerned with the production of information. Guilford defined the difference between convergent and divergent thinking in terms of scope, direction, and determinateness of the production. In convergent thinking operations, the individual searches narrowly for a single correct response to a problem. He/she looks for a "logical imperative" (Guilford, 1975), an obligatory solution to the problem. In the process known rules or truths are applied. The information that is generated in convergent thinking is determined by the information given at the outset. For example, in searching for the answer to the problem "What tool has three letters and opens doors?", the individual will apply language and knowledge rules to generate the answer "key". The problem is taken on a straight path to its solution, without the addition of new information. By contrast, in divergent drinking operations, the individual searches broadly for possible responses to a problem by seeking variety, and moving in different directions from a single starting point. Known rules and truths are stretched or transgressed. Divergent thinking aims to generate "logical alternatives" (Guilford, 1975). With this qualification, Guilford meant to indicate that divergent (creative) thinking is rational and constructive, and denotes the creative person's readiness to generate, or accept as relevant, alternatives that are remotely connected to the starting point. He cautioned against equating divergent with erratic, irrational, or downright bizarre thinking. 121 Divergent thinking generates information that goes beyond the information provided in the problem For example, in searching for answers to the problem "How could you open a door if you don't have the key?" the individual cannot apply the known solution, and has to explore different paths, that may lead only mdirectly to the solution. He/she could look for a key duplicate (could it be under the newspaper?), probe the lock with a paper clip or a flowerpot shard, abandon the door and inspect other openings, enlist the help of a cat, and so on. The problem is taken in different directions, and new information is added in this process, that is not contained in the initial problem. It should be noted here that it is the approach to a situation that differentiates divergent from convergent thinking, not the type of situation. It is not the case that convergent thinking occurs in everyday situations whereas divergent thinking belongs to the realm of the fantastic and the imaginary. Guilford suggested that convergent thinking, along with the cognitive and memory abilities, is essential for intelligence whereas divergent thinking is essential for creativity. Guilford is not alone in stating that generating a multitude of possible ideas is an important part of the creative process. This notion surfaces with consistency in other theories of creativity. For example, Amabile (1996) identifies response generation as an important stage in the creative process. In this stage, the individual generates response possibilities by searching cognitive pathways as well as the immediate environment. Barchillon (1961) indicates the oogito process as a means of "shaking and throwing things together", to produce raw, uncensored possibilities. Also the testimony of creative people points to the production of ideational variety and multitude as an important part in their achievements. Poincare spoke of ideas rising "in crowds"; Housman described "springs of ideas that bubble up"; Thomas Wolfe "... wrote and wrote and could not give up writing" (cited in Ghiselin, 1963). Operationally each type of thinking is defined by the characteristics of the tests that measure it. Tests of convergent thinking ask for the correct or best answr to a problem Typical examples are the 122 following questions: "From which object could you most likely make a needle? (a) a cabbage; (b) a splice; (c) a steak; (d) a paper box; (e) a fish"; "The absence of sound is " (Guilford, 1959). By contrast, tests of divergent tlnrddng ask for all the answers an individual can think of in response to a problem. Typical examples are the following questions: "List all the uses you can think of for a brick"; "Suggest as many titles as you can for this short story... " (Guilford, 1959). As can be seen from these examples, tests of convergent thmking call for answers that best fit the information provided, whereas tests of divergent thinking call for answers that are possiblefits for the information provided. The distinction between convergent and divergent thinking tests has parallels in the creativity literature. Amabile (1996) anchors her conceptual definition of creativity to the distinction between heuristic and algorithmic tasks: a product or response can be creative only to the extent that the task is a heuristic one. Algorithmic tasks have clearly identified goals. The path to their solution is straightforward and known. Heuristic tasks do not have a known path to their solution. In fact, they might not even have a clearly identifiable goal. In this case it is up to the individual to define a goal and discover an algorithm for the solution of the task. Divergent thinking tests can be compared to heuristic tasks in that the goal is broad and open-ended (e.g., all the possible uses for a brick) and many paths must be sought out in the process of discovering acceptable solutions. Guilford suggested divergent tliinking is made-up of three different component abilities. These abilities are fluency, flexibility, and originality. Fluency is the facility and speed with which the individual produces ideas. It is a quantitative measure that refers to the raw count of responses to a problem. According to Guilford fluency is the mainstay of divergent thinking, such that he observed in regard to creativity that "fluency [... ] is the name of the game" (1975, p. 40). Guilford described flexibility as the ability to make transitions between classes of ideas. For example, in thiriking about the uses for a brick, one individual might only invoke the context of building whereas another might invoke sports (e.g., weight lifting), the household (e.g., making a 123 bookcase), personal defense, or geometry (e.g., studying the volume of a rectangle). In other situations flexibility is required to overturn a response set. For example, in any variant of the Match Problem, where the task is to take away some matches and leave a specified number of squares, the subject has to abandon the idea that the solution entails squares of identical size (Guilford, 1959). The third ability involved in divergent thinking is originality. Originality is generating novel and unusual ideas, ideas that few other people would think of. Original ideation is antithetical to rote (Runco, 1991) or stereotyped ideation (Mednick, 1962) and is arguably the hallmark characteristic of any creative product. According to Guilford, originality is a product of fluency and flexibility. A large output of ideas is virtually a guarantee that some ideas will also be original. Frequent shifts between classes of ideas also make the discovery of novel responses more likely. Intuitively, divergent or creative thmking appears to be the kind of thinking that might facilitate ProM. But what kind of argument or evidence could lend credence to this intuition? What do we know about creative thinking that can be linked to ProM? The next section will systematically examine potential links between creative thinking and ProM, and formulate predictions about their relationship. Predictions about the Relationship between Prospective Memory and Creative Thinking The majority of research on divergent thinking has been in the domains of education (Runco, 1991; Wallach & Kogan, 1965), personality theory (Cattell & Butcher, 1968; Eysenck, 1993) and psychopathology (Green & Williams, 1999; Prentky, 1980). There are a few studies on creativity from a cognitive perspective (cf. Smith, Ward & Finke, 1995), but these have focused on the processes of incubation and insight rather than on processes of active thinking. There are no studies, to my knowledge, on the relationship between divergent thinking and memory processes. I will utilize conceptual analyses as well as indirect lines of evidence to develop predictions about the relationship between divergent tliinking abilities and ProM. 124 From a conceptual standpoint, divergent thinking abilities fit the demands of a ProM situation. As observed in earlier parts of the present work, the ProM cue is a natural and expected part of an ongoing task context. The challenge for the rememberer is to discover the cue's connection to a different context, the context of a plan or an intention. All three divergent thinking abilities might contribute to meeting this challenge. Fluency might be required to produce in rapid sequence multiple thoughts in response to the ProM cue. Originality might be required to pick up the more hidden and remote implications of the ProM cue. In fact, our everyday experience supports this latter contention. When we are thinking about things that are closely related to the to-be-remembered intention, we are much more ready to discover ProM cues than when our thoughts are elsewhere. For example, a first-grader might not remember his mother's request to give a form to the teacher during school activities because his thoughts are busy with current activities. But during recess, when he usually misses his mother a bit, thoughts about her bring back the memory of her request. Flexibility might be required in ProM situations to consider more than one line of thinking or give up one line of thinking in favor of an alternative one. There is an interesting empirical demonstration of how flexibility might sustain creative thinking, which might be relevant also in the context of ProM. Mendelsohn (1976) tested the influence of creativity on the ability to benefit from hints in solving anagrams. He assigned subjects to a highly creative or low creative group on the basis of their Remote Associates Test (RAT) scores. The RAT presents sets of three words from semantically remote clusters and requires subjects to find a common link (for example, the solution to the words "rat - blue - cottage" is "cheese" 9. The high and low RAT scorers in Mendelsohn's study were given an anagram-solving task. Half of the subjects received the hint that some anagrams 9 It should be noted that the validity of the RAT as a creative thinking test has been questioned because of its emphasis on a single, correct answer (e.g., Wallach &Kogan). Though this criticism is well founded and would have to be considered when choosing this instrument for measuring creativity, here it is assumed that the RAT does measure to some extent the ability to think creatively because it forces the test-taker to identify new connections. 125 would be kinds of animals, some kinds of food, and some would not belong to any common category. The other half received no hint. Mendelsohn found that the hint condition facilitated performance in high RAT scorers but interfered with performance in low RAT scorers. When given the hint, high RAT scorers were able to solve more anagrams in both the animal and food group, and solved an equal number of anagrams in the miscellaneous category compared to subjects in the no hint condition. By contrast, the low RAT scorers were able to solve more anagrams only in one category (animals or food), and fewer anagrams in the other category and in the miscellaneous category compared to subjects in the no hint condition. This performance pattern suggests that highly creative subjects can utilize flexibly different categories in their thinking processes, whereas subjects low in creativity stick with a single category. In more general terms, the study suggests that highly creative people have greater control over their mental processes and are less influenced by external stimuli. It can be imagined that this type of flexible control over thought processes might sustain ProM in a very similar manner. The ongoing task and the ProM task can be thought of in terms of separate categories or sets. It might be that subjects low in creativity stick with the ongoing task set whereas subjects high in creativity might be able to apply alternate sets, including possibly the set of "things to do", and thus have a much better chance at responding to the ProM cue. There is some empirical evidence to support commonalities between ProM and divergent thinking. At a very general level, both abilities, along with many other cognitive and memory functions, show age-related declines. For ProM, the evidence of age-related performance decrements has been reviewed in Chapter Two. For divergent thinking, McCrae, Arenberg, and Costa (1987) found cross-sectional and longitudinal performance declines. A large sample of subjects ranging in age from 17 to 101 years received five divergent thinking tasks. Some of these subjects were given the same tests again after a 6-year interval. The divergent thinking tasks were producing synonyms of words, writing down sentences with words beginning with designated letters, producing words 126 that begin with, or contain, a designated letter, and imagining the possible consequences of unusual situations. All tests were scored for fluency (number of responses), except for the Consequences test, which was scored for originality (number of remote or unusual responses). The cross-sectional analyses showed significant performance decrements in the older age groups (aged 60 or above). Interestingly, the Consequences test showed earlier declines, evident already in middle-aged (40-60) compared to young adults (20-30). This pattern was confirmed by the longitudinal analyses, which revealed test-retest declines in all age groups for the Consequences test, and in the older groups for the other measures. Taken together, these findings suggest that fluency and originality of thinking decline with age, the latter at an earlier time in the life span than the former 10. ProM and divergent thinking seem to undergo a similar fate with aging but this is true for many cognitive functions. A more interesting question is whether they also involve similar brain regions that may in fact underlie the age decline? As has been seen in Chapter Two, ProM appears largely mediated by areas in the frontal lobes, in particular the frontal poles and the right prefrontal cortex (Burgess et al., 2001). There are many suggestions in the literature that patients with lesions in the frontal lobes have difficulty handling new tasks and finding new solutions to old tasks (Goldstein, 1944; Shallice, 1988). Zangwill (1966) specifically predicted that frontal lesions were more likely to disrupt divergent than convergent thinking. 1 0 Age-related declines in divergent thinking conflict with the influential idea that peak creativity is reached in different domains at vastly different ages, with philosophy and prose writing peaking as late as in the seventh decade of life (Lehman, 1953). One account of this discrepancy lies in the distinction between normal and gifted creativity, advocated by many creativity theorists (Feldman, 1986; Gruber and Davis, 1988; Runco, 1991). 127 A recent study by Carlsson, Wendt, and Risberg (2000) compared patterns of blood flow as measured by regional cerebral blood flow (rCBF) in subjects high and low in creativity. Both groups were given the Controlled Oral Word Association Test (COWAT) as a control task and the Uses of Objects test as the experimental task. The latter task is a divergent thinking task, which asks subjects to list all the possible uses they can think of for some common objects. Compared to the control task, the highly creative group showed bilateral blood flow increases in the anterior prefrontal and superior regions of the frontal cortex when performing the Uses of Objects test whereas the low creative group showed only lateralized increases in the left regions. This suggests the involvement, in creative thinking, of the right prefrontal and superior frontal areas, the very same areas that are assumed to play a role in ProM, and that are known to show neuroanatomical and neurochemical changes that begin early in the aging process (West, 1996). So far speculation and evidence converge on the idea that creativity overlaps with ProM, and that in fact it might support ProM. On this basis, the prediction could be made that a positive relationship exists between measures of ProM and measures of creative thinking. But there is one claim about creativity that goes against this prediction. Some theorists have proposed that creative people have difficulty with selective attention and that this very difficulty may contribute to the creative process (Andreasen & Powers, 1975; Eysenck, 1993; Prentky, 1980). The general argument goes as follows. Creative people tend to pay attention to a greater number of stimuli than people who are low in creativity. Many of these stimuli are unrelated to current thoughts and activities. They may become interconnected by virtue of having entered conscious awareness at the same time. This can result in unusual associations of ideas and a general "widening of the associative horizons" (Eysenck, 1993). There is some empirical evidence to support the claim of an inverse relation between creativity and selective attention. Stavridou and Furnham (1996) investigated the relationship between creativity and the ability to inhibit to-be-ignored information. They gave subjects five 128 divergent thinking tasks and a Stroop task in which subjects were exposed on each trial to a to-be-ignored prime word which either matched or did not match the ink color of the subsequent probe word. A measure of negative priming was computed by subtracting reaction times on non-matching trials from reaction times on matching trials. Consistent with the claim that creativity is associated with low selective attention skills, the study found a tendency for subjects with high divergent thinking scores to have lower values of negative priming than subjects with low divergent thinking scores. Stavridou and Furnham's findings are not compelling but there is converging evidence from a study on creativity and breadth of attention. Kasof (1997) assessed breadth of attention with the Stimulus Screening Scale (Mehrabian, 1977). This scale measures individual differences in the tendency to screen out irrelevant environmental information. L o w stimulus screening indicates wide breadth of attention. A sample item is "I often become aware during the day of the texture of the clothes I wear." Creativity was measured by means of a poetry-writing task. There was a positive association between breadth of attention and creativity ratings. Moreover it was found that ambient noise disrupted performance on the creativity task to a greater extent in subjects with wide than with narrow breadth of attention. Both findings support the idea that creative people tend to process task- or situation-irrelevant information to a greater extent than less creative people. The notion that creativity is associated with low filtering of information and the related evidence depict creative thiriking as a process involving low control over mental events and creative individuals as adrift in the stream of mcoming stimuli. This picture of creative thinking is dramatically different from the picture emerging earlier in this section, a picture that highlighted the high degree of control over thought processes that creative individuals may have. Which picture is more accurate? There is no definite answer to this question in the existing literature, but if the view of creativity in terms of low filtering and high dependency of mental events on external stimuli is correct, then the prediction regarding the relationship between creative thinking and P r o M must also 129 change. If creative individuals exert low control over their mental responses to external events, they might have difficulty responding to the ProM cue as they cannot stop themselves from attending, and responding to, ProM-irrelevant contextual information. This reasoning would lead to predicting a negative relationship between measures of ProM and measures of creative thinking. The conclusion that can be drawn from this mix of speculations and evidence is that ProM and creative thinking are in some important way related. However, arguments exist to support the prediction of both a positive and a negative relationship between ProM and creative thinking. Predictions about this relationship must remain open-ended but their investigation promises to be of high interest to understanding ProM. 130 CHAPTER SIX: INVESTIGATING T H E RELATIONSHIP B E T W E E N PROSPECTIVE MEMORY AND CREATIVE THINKING The general strategy for investigating the relation between ProM and creative thinking was to examine the correlations between performance on ProM and creative thinking measures in samples of undergraduate student volunteers. The data for this part of the present work were collected in the context of Experiments One and Two described in Part I. One of the ProM measures was the paradigm used for the attentional manipulations. In this paradigm subjects were instructed to interrupt the ongoing task if they noticed the picture of a butterfly or helicopter in the course of the experimental session. This picture then appeared in the third block of an ongoing activity requiring responding to letter displays in specified ways. A detailed description of this paradigm was given in Chapter Three. A second ProM measure was embedded in a separate portion of the experimental session, and was different for Experiment One and Two. The second ProM measure will be described in detail in the relevant Method sections. To examine ProM's relationship with creative thinking, it was important to utilize reliable and valid measures of creative thinking. A variety of measures, among which behavioral assessments, have been used in creativity research. Issues regarding the psychometric adequacy of these assessments have been among the most debated problems in the creativity literature. A moderately pessimistic disposition dominates these debates, and it is recognized that the behavioral measurement of creativity is complicated for more than one reason. One problem frequently reported is that creativity, whether scored in terms of originality or in terms of flexibility, may simply not be a reliably measurable attribute (Runco, 1991). Another problem is that performance on any given instrument is highly sensitive to variables of the testing environment. For example, timing (timed versus non-timed), atmosphere (test versus game), and instructions (produce practical ideas versus wild ideas) were all found to influence performance (see Amabile, 1996, for a comprehensive 131 review). Also troublesome is the finding is that different behavioral measures of creativity are not highly correlated with each other, nor with other assessments of creativity, for example, teacher ratings or checklists assessing creative everyday activities (Baer, 1993; Hocevar, 1981). For the present work I defined creativity in terms of divergent thinking abilities, and hence employed measures known to assess divergent thinking. These measures are no exception to the problems with reliability and validity noted above (see Runco, 1991). The reliability of divergent thinking measures was a primary concern also for this work, given that the statistic of interest was the correlation coefficient. Hence this issue was addressed firsthand in a pilot investigation. The validity of divergent thinking measures as indices of creativity was deemed to be of lesser concern as this work did not aim to make general statements about creativity. Pilot Investigation This preliminary investigation aimed to assess the internal consistency and inter-test correlations for four measures of divergent thinking. It also aimed to establish that these measures probed abilities different from those indexed by measures of verbal knowledge. Forty undergraduate students from the University of British Columbia participated in the study in return for course credit. They received four tasks, each consisting of five test items and a practice item Three tasks (the Uses of Objects test, the Similarities test, and the Instances test) were adapted from the battery of divergent thinking tests for children, developed by Wallach and Kogan (1965). Some items, deemed too easy for young adults, were replaced with more difficult ones: for example the item "tilings that are square" on the Instances test was replaced with "things one can see through". The fourth task (the Consequences test) was developed on the basis of the description provided in McCrae et al. (1987). For the Uses of Objects test subjects were asked to tell the experimenter all the possible uses of five common objects (for example, "brick" and "shoe"). For the Similarities test subjects were instructed to tell the experimenter all the ways in which five pairs of objects were alike (for example, 132 "door - chair" or "radio - telephone"). For the Instances test the request was to tell the experimenter all the items that belong to five class concepts (for example, "liquid things" or "things that make noise"). For the Consequences test, the request was tell the experimenter all the consequences if five unusual events happened (for example "What if food were no longer needed to sustain life?" or "What if all countries of the world spoke the same language?"). The order of administration of these tasks was counterbalanced. The time limit for each item was three minutes, but subjects were not informed of this limit. If subjects indicated that they had no further ideas prior to reaching the time limit, they were encouraged to think some more. Subjects' responses were audio-taped for later scoring. In addition to the divergent thinking tasks subjects were given the North American Adult Reading Test (NAART). This test presents sixty-four English nouns to subjects, who are required to read these words aloud to the best of their abilities. The words increase in difficulty from relatively easy items such as "subtle" to very demanding ones, such as "synecdoche". All divergent thinking tests were scored for fluency, that is, the raw count of non-repeated responses. Inter-scorer agreement for twenty protocols was found to be perfect. All tests were found to have high internal consistency (Cronbach's alpha = 0.93 for Uses, 0.92 for Similarities, 0.89 for Instances and 0.85 for Consequences). Inter-test correlations ranged from r (38) = 0.71 to 0.90. No divergent thinking measure correlated significantly with verbal abilities as measured by the NAART. Generally, the reliabilities found in this pilot investigation were equal to, or better than, similar published estimates (Wallach & Kogan, 1965, for Uses, Similarities, and Instances; McCrae et al., 1987, for Consequences). It was concluded from this pilot investigation that the four divergent thinking measures had adequate internal consistency even when shortened to four items. Given the high inter-test correlations and the absence of a correlation with NAART scores, these tasks were assumed to tap the same type of abilities, which could not be equated simply with verbal aptness. 133 First Investigation The first investigation was intended to explore the relationship between ProM performance and measures of divergent thinking. There were two objectives. The first was to establish the existence of a relationship, and to indicate its sign -positive or negative. The second objective was to pinpoint the type(s) of divergent tliinking abilities that are related to ProM. The investigation focused on fluency, that is, the facility and speed with which ideas can be produced, and originality, that is, the tendency to generate ideas that most other people would not think of. Fluency was measured as the total raw count of responses to each test item. Originality was measured as the total count of unusual responses to each test item. Unusual responses were responses given by less than 5% of all subjects queried. Flexibility was not considered in this investigation as it was assumed, in keeping with Guilford's original assumptions (Guilford, 1959) that this ability would be reflected in the quantitative fluency score and the more qualitative originality score. Method Materials The ProM tasks were the experimental and naturalistic ProM task employed in Experiment One, Chapter Three. For the nataralistic ProM task, subjects were instructed at the begmning of the experiment that they should remember to mention one task they enjoyed in the experimental session upon receiving their participation credit form at the end of the experiment. For the experimental ProM task, subjects were instructed to stop any ongoing activity and recite a list of previously learned words if they encountered a picture cue (butterfly or helicopter) at anytime in the experimental session. The cue was presented in the third block of an attention-demanding task. For this investigation, the two divergent thinking tasks with the highest internal consistency were selected from the battery of tasks employed in the Pilot Investigation. These were the Uses of Objects test and the Similarities test. For the Uses of Objects test, subjects were instructed to generate all the possible uses that they could think of for four common objects. The objects were a 134 brick, an automobile tire, a bottle cork, and a shoe. The subjects were told: "I [the experimenter] am going to name a few common objects. Your task is to tell me as many uses as you can think of for each of these objects. For example, if I say 'newspaper', what could you say? [Subjects were asked to elaborate on this practice item.] Now tell me all the uses you can tliink of for a brick [or other object]." For the Similarities test, subjects were asked to list as many ways as they could think of in which each of four pairs of common items were alike. The item pairs were a cat and a mouse, a train and a tractor, milk and meat, and a grocery store and a restaurant. They were told: "I [the experimenter] am going to name a few pairs of objects. Your task is to tell me all the ways that these pairs of objects are alike. For example, if I say 'apple - orange', what could you say? [Subjects were asked to elaborate on this practice item.] Now tell me all the ways a cat and a mouse [or other pairs] are alike." Participants and Procedure The data for this investigation were collected as part of Experiment One. The participants in this experiment were 123 undergraduate students from the University of British Columbia. There were 94 women and 29 men, between 17 and 26 years of age (M age = 19.4). The order of administration of the ProM and divergent thinking tasks is listed in Table 3.1. The procedures for administering the ProM tasks are described in detail in Experiment One, Chapter Three. The procedure for administering the divergent thinking tasks was the following. The tasks were presented to participants as investigating ways in which people come up with ideas. The first divergent thinking task was adnninistered midway through the experiment, between the instructions and test phase of the experimental ProM task. The second divergent thinking task was given toward the end of the experiment, between the Picture Memory test phase and a block of trials of the simple or choice reaction time tasks. To control for possible effects due to practice and fatigue, the order of adrninistration of the two divergent thinking tasks was counterbalanced. There were two between-subject conditions in Experiment One. These were the two different ongoing tasks (simple or choice reaction task) in which the experimental ProM task was embedded. Half of the subjects in each condition were given the Uses test, the other half the Similarities test, as the first divergent thinking task. For the Uses test, subjects were told that they would be presented with the names of four common objects and that their task was to tell the experimenter all the possible uses they could think of for each object. To begin with, they were given a practice trial in which they were asked to suggest a few uses for a newspaper. If their responses reflected a variety of uses, the experimenter proceeded to administering the test items. If subjects could think only of one type of use (for example, reading world news, reading the sports section, reading job advertisements), the experimenter mentioned a few additional uses (lighting a fire, as a protection from the sun, as lining for a birdcage). For each test item, the experimenter allowed three minutes but subjects were not informed of this time limit to prevent them from fomiing the impression that this was a speeded exercise. This procedure was adopted to follow Wallach and Kogan's (1965) recommendation of creating a task context free of time pressure. If subjects, prior to the three minutes' limit, indicated that they had no further ideas, they were encouraged to try for another while. If no response was provided after another half minute, a note was made of the total time taken for the item, and the next item was presented. The procedure for the Similarities test was identical to that for the Uses test. In the practice trial, subjects were asked to indicate a few ways in which an apple and an orange are alike. If they were able to identify only one type of similarities (for example, fruits, edibles, taste sweet), the experimenter mentioned a few additional similarities (contain vitamin Q are round, grow on trees). The test was then administered in the same way as the Uses test. All responses were recorded in writing by the experimenter as well as audio taped. 136 Results This section will begin by describing the criteria followed for scoring ProM and divergent thinking performance. Next, performance on the ProM and divergent thinking tests will be described and the reliability of the tests will be examined. Lastly, the correlations between ProM indices and divergent thinking indices will be evaluated. Scoring Two ProM measures and two measures of divergent thinking were considered for the analyses. The experimental ProM task was scored for occurrence of a ProM response and for size of the cue at which the response was given. The scoring criteria are described in Experiment One, Chapter Three. These two outcomes are referred to as ProM response and Cue size, respectively. The naturalistic ProM task was scored for the timeliness of the response. Subjects received 3 points if they responded immediately upon taking the participation credit form by mentioning a task they enjoyed, or by remarking that they were supposed to say something at this time. They received 2 points if they gave either response during explanations about how to fill out the form to obtain course credit, 1 point if they gave either response before leaving the room, and a score of 0 if they did not respond in the described way at all. The two divergent thinking measures yielded two distinct scores each. A fluency score was obtained by counting the number of non-repeated responses to each test item. This count included all responses that stated a use (a purpose one could employ an object for) or a similarity. Descriptive responses (for example, "A brick is red", or "Mice are afraid of cats") as well as paraphrased repetitions were excluded. A mean fluency score was obtained for each test by averaging across the four single-item scores. An originality score was obtained by compiling a lexicon of subjects' responses to each test item and by computing the frequency of each response in the total sample. For example, for the item brick, the response "It could be used as a paperweight" was given by 50 subjects, or 41.5 % of 137 the entire sample. The response "It could be used to flatten something'' was given by 3 subjects, or 2.4% of the entire sample. The response "It could be used as a shock absorber" was given by a single subject. Responses were classified as common'^, more than 5% (n <or = 7) of the entire sample had given them. Responses were classified as unusual if 5% or less of subjects (n <or =6 and >or =2) had expressed them. Responses were classified as unique if only one subject had expressed them 11. Sample responses for the three response categories are given in Appendix C. A careful inspection of the unique responses revealed a potential problem with this type of ideas. While some responses clearly had an original content (for example the idea to use a brick as a chopping board), others were best described as idiosyncratic or bizarre (for example, to use a brick to create a universe). It has often been recognized that statistical infrequency alone is no guarantee of creative content, and that value and appropriateness must be considered as well (e.g., Amabile, 1996). It was therefore decided to utilize only the unusual responses given by people as a measure of originality, and to exclude their unique responses from further analyses. Originality scores were computed for each test by averaging across items the number of unusual responses given by each subject. 1 1 Several studies on divergent thinking have utilized the frequency of five percent as the cutoff to distinguish between obvious or common responses, and unusual or original ones (Hocevar, 1978). Some studies have further broken down original responses into unusual ones and unique ones (Runco, 1991). The present work broadly followed Runco's criteria. 138 Table 6.1. Performance on ProM tasks and divergent thinking tasks. Task Index Mean and Standard Deviation Experimental ProM task ProM response Cue size Proportion Unit from 1 (smallest) to 19 (largest) 0.54 (0.50) 10.66 (5.07) Naturalistic ProM task Score on 0 to 3 scale 2.18 (1.14) Uses of Objects Fluency Originality Number of responses Number of responses 8.63 (3.41) 1.65(1.15) Similarities Fluency Originality. Number of responses Number of responses 11.52 (4.09) 1.46 (1.74) Table 6.2. Correlations between divergent thinking indices and with NAART scores (n = 122; * = r coefficients with p values < 0.001). 2 3 4 5 (1) Uses Fluency .85* .78* .64* -.03 (2) Uses Originality .68* .59* -.09 (3) Similarities Fluency 76* -.02 (4) Similarities Originality -.06 (5) NAART 139 Table 6.3. Correlations between ProM and divergent thinking indices (n = 122; * = r coefficients with p values < 0.05) Prom Index U S E S Fluency U S E S Originality SIMILARITIES Fluency SIMILARITIES Originality ProM Response -.03 -.19* -.02 -.10 Cue size -.01 -.11 -.01 -.11 Naturalistic ProM Score -.13 -.23* -.06 -.22* Test Performance and Test Reliability Table 6.1 displays performance on the ProM and divergent thinking tasks. Performance on the ProM tasks was scored differently and cannot be directly compared. However, neither was at or close to ceiling. There were modest differences between the two divergent thinking tasks. Items on the Similarities task elicited a higher overall fluency than items on the Uses task. However, the Uses task elicited a higher percentage of original responses than the Similarities task (19 versus 13%). Fluency scores were normally distributed whereas originality scores presented some degree of positive skewness (~1.0). To reduce skewness, originality scores were transformed into square roots. As each ProM score was derived from a single measurement, it was not possible to establish internal consistency for the ProM measures. Their intercorrelations were the following: Naturalistic task with ProM response r (118) = 0.26 (p <0.05); naturalistic task with Cue size r (62) = -0.08. These single-observation correlations are not impressive, but it must be bome in mind that by definition ProM abilities cannot be assessed repeatedly in a single experiment. For the divergent thinking tasks, internal consistency was assessable and found to be very high for fluency scores on both tasks. For the Uses task, Cronbach's alpha was 0.90, for the Similarities task it was 0.89. Internal consistency was acceptable for originality score, with alpha = 140 0.72 for Uses, 0.68 for Similarities. Table 6.2 reports the correlation coefficients between all divergent thinking scores, and between these and the NAART scores. All correlations between divergent thinking scores had r coefficients greater than .60. No divergent tliinking index correlated significantly with NAART scores. Correlations between Prospectiie Memory and Divergent Thinking It was found consistently across ProM tasks that originality scores were inversely related, and fluency scores not related, to ProM performance. Performance on the experimental ProM task was measured while subjects were perforating either a simple reaction or a choice reaction. The type of ongoing task had a significant effect on ProM. To test the possibility that divergent thinking scores might interact with the ongoing task variable, subjects were assigned to two groups on the basis of their fluency and originality scores: Subjects scoring above the median were assigned to the high-scoring group, subjects scoring below the median to the low-scoring group. ANOVAs with factors ongoing task type (simple reaction versus choice reaction) and divergent thinking level (high versus low) were conducted for ProM response and for Cue size. Neither fluency nor originality level were found to interact with ongoing task type. Thus, for the purposes of the correlational analyses, the performance data for the experimental ProM task were pooled across ongoing task type. Table 6.3 reports the correlation coefficients between the three ProM indices and the four divergent thinking indices. The pattern of correlations was consistent across measures. Fluency did not correlate significantly with any ProM index. Originality was significantly correlated with two of three ProM indices. The correlation was negative, that is, higher originality scores predicted a lower likdihoodqfProM. To probe the relationship with Cue size further, a corrected Cue Size index was computed by replacing the missing values of subjects who failed to respond to the ProM cue with the value of 20 (one unit above the highest actual unit). The correlation of this corrected index with originality was similar to that found for ProM Response, and as expected, of the opposite sign r 141 (120) = 0.17, p <.10). Again, as observed earlier in Chapter Three, corrected Cue size seems to reflect mostly the binary ProM Response variable. Discussion The questions addressed in this investigation were whether performance on divergent thinking tests was related to ProM performance, and, if so, which divergent thinking abilities were responsible for the association. The results consistently indicate across different measures of ProM and divergent thinking that a relationship exists, and that its sign is negative. They also show that this relationship exists for originality of thinking, not for overall fluency of thinking. First of all, it is interesting to note that originality, but not fluency, is implicated in the relationship with ProM. Many authors have identified facility of thought as the cornerstone of creativity (e.g., Guilford, 1975). Consistent with this view, fluent and original thinking are usually highly related (e.g., Wallach &Kogan, 1965), and the results of the present study are no exception. It must be then that the variance that original thinking shares with ProM is unrelated to the variance it shares with fluent thinking, and that facility of thinking has no part in explanations of the association between originality and ProM. How seriously should this relationship between ProM and original thinking be taken? Could it be the result of extraneous factors, for example subjects' general response to the experimental situation? Original or creative people are often described as intolerant of rules and standards (Guilford, 1975). It is possible that the creative subjects in this experiment had poor compliance with the experimental procedures. However, none of the performance data support this possibility. Subjects with high originality scores performed as well as -or better than- subjects with low originality scores on the simple and choice reaction tasks as well as on the RetM measures employed in Experiment One. ProM is the exception. It is also possible that creative people experienced the ongoing activities for the ProM task as monotonous and paid little attention. This would likely have manifested as lower response speed and/or accuracy on the ongoing task trials. However, subjects 142 scoring high in originality did not differ from subjects low in originality in terms of response speed and response accuracy on any trial block, and most importantly on Block 3, when the ProM cue was presented. Thus the possibility of boredom as a factor appears unlikely. The inverse relationship between ProM and original thinking appears a genuine one. Because the evidence is correlational, no definite conclusions about the direction of the observed relationship can be drawn. In this chapter I develop interpretations that focus on the possible influence of creative thinking abilities on ProM but I am aware that they are only one possible account of the findings. I will return to this point in the General Conclusions, Chapter Seven. In Chapter Five I argued that because original thinking counteracts rote interpretations of events, it might latch onto the more hidden implications of the ProM cue and make ProM more likely. According to this argument, a positive relation should be observed between the two abilities. One explanation for the negative relationship with ProM observed in this investigation is that the ProM cues, and their associated actions, were too obvious. In fact, in both ProM tasks, subjects, as it were, did not have to think far to invoke the planned action. When the picture cue appeared on the screen at increasing sizes, a rather obvious thing to think of was stopping one's activities. When the participation form was handed over at the end of the experiment, a reasonable thing to think of was saying something pleasant about the experiment. Original thinkers might have brought to mind much more unusual and remote connections, and thus missed the rather obvious implications of the ProM cue. In Chapter Five I also discussed an alternative view of creative thinking in terms of low fntering of information and low control over related mental events. According to this view, a negative relation should exist between original thinking and ProM. If this view is taken into account, a different explanation exists for the poor ProM performance of original thinkers in this investigation. It is possible that their thought processes in the ProM test situation, and prior to it, were held captive, as it were, by the ongoing events of the experiment. This might have prevented 143 them from engaging in thoughts about the ProM task, which are known to benefit ProM retrieval (Maylor, 2000). More importantly, it might have prevented them from noticing the ProM cue when it appeared because they were unable to control or stop the processing of ongoing events. In sum, original thinking does not benefit ProM, as hypothesized at the outset of this work. On the contrary, it seems to interfere with the ProM retrieval process. How this interference comes about, remains to be established, and some of the speculations will be followed up in the next investigation. But what kind of thinking helps ProM, then? It could be that it is flexibility of thinking. This possibility will be addressed in the next investigation. It is also tempting to suggest that ProM requires some extent of linearity and convergence. However, this is not directly supported by the findings. Low originality is not necessarily equivalent to high convergent thinking. Second Investigation This investigation sought to replicate and extend the finding of a relationship between ProM and creative tliinking. There were two objectives. The first objective was to make a comprehensive assessment of the relationship between ProM and creative tliinking. For this purpose, in addition to two measures of ProM and creative thinking employed in the first investigation, new measures of both abilities were introduced. Moreover, for the creative tliinking tasks, flexibility was considered in addition to fluency and originality. To address this objective, ProM was assessed by means of the experimental ProM task and by means of a new ProM task, in which the cues were semantic, not pictorial. Moreover, the cues were instances of a category (animals), and only the category was provided at the time of ProM instructions. The reasoning was that in this situation, mvoking the ProM task upon cue presentation required one more tliinking step than in the experimental ProM task -from instance to category, and from category to intended action, rather than directly from picture to intended action. It is possible that in this type of ProM situation, original thinking may be an asset rather than a hindrance as in the previous investigation. 144 Creative thmking was assessed by means of the Uses of Objects test, and by means of a visuo-constructional task requiring people to produce drawings. This latter task was the Picture Completion test, one of the sub-tests in the figural battery of the Torrance Tests of Creative Thirddng (TTCT; Torrance, 1966) 12. The reason for selecting a visually based creativity task was to assess originality in a way that was not verbally mediated. The divergent thinking tests used in the first investigation relied predominantly on verbal processing, and might have measured an individual's skill and aptness at manipulating words. In fact, it has been suggested that original ideas on divergent thinking tests reflect wittiness rather than genuine diversity in thinking (Maddi, 1975). Wittiness, or the tendency to play with words, in turn, might have effects on the verbal encoding of the ProM instructions, possibly complicating or distorting them. The second objective was to examine whether creative thinking affects the way people approach the ProM task. It is possible that highly creative people understand and act upon ProM tasks in ways that are different from those of people who are less creative. For this purpose, subjects were given a questionnaire at the end of the experiment, which probed various aspects of their performance on the experimental ProM task. 1 2 It should be noted that the TTCT battery is not strictly an assessment of divergent thinking, but rather a broad-ranging assessment of multiple skills assumed to be involved in creativity. However the tests are compatible with the understanding of creative thinking in the present work, in that they are open-ended (that is, they do not have a correct answer) and each is scored in terms of the component abilities of Guilford's divergent thinking model. 145 Method Materials The ProM tasks were the experimental and the Story ProM task employed in Experiment One (Chapter Three). The experimental ProM task was identical to the one used in Experiment One except for the minor changes discussed in Experiment Two, Chapter Four. For the Story ProM task, a passage describing the flight of three fugitives from Salt Lake Gty, was adapted from the novel yl Study in Scarlet by Sir Arthur Conan Doyle (see Appendix A). The passage was divided into eight paragraphs of approximately 100 words each. Each paragraph was presented separately on the computer screen, in 24-point Arial font single-spaced against white background. Paragraphs two, four, six and eight contained the word of an animal or arhmals -horses, bear, sheep, dogs, respectively. The animal words were placed approximately midway in each paragraph. Subjects were instructed to read the paragraphs by initiating and ending the presentation of each with a press of the spacebar key on the computer keyboard. The creative tliinking tasks were the Uses of Objects test employed in the first investigation, and the Picture Completion test from the figural battery of the Torrance Tests of Creative Tliinking (Torrance, 1966). For the Uses test, subjects were instructed to write down as many uses as they could think of for four common objects. There were two versions of this test. One version included the objects "rope", "automobile tire", "toothbrush" and "brick" (Form A), the other the objects "shoe", "button", "towel" and "pencil" (Form B). Each version was administered on four lined sheets, at the top of which one object was printed. The first sheet also contained the instructions, which were read to the subjects. The instructions were the following: "On each sheet of this test, you will find the name of a common object. Write down as many different uses as you can tliink of for the object. You can do this in point form. Be sure to describe a use for the object -a purpose one could employ it for- not a characteristic of it." 146 The materials for the Picture Completion task were taken from the Torrance Tests cf Crmtke Thinking Norms and Technical Manual (Torrance, 1966). There were two versions (Forms A and B), each with 10 frames containing incomplete line drawings printed on two sheets (see Figure 6.1 for a sample sheet). The instructions, printed on the first sheet, were read to the subjects. They were: "By adding lines to the incomplete figures on this and the next page, you can sketch some interesting objects or pictures. Try to make each picture tell as complete and as interesting a story as you can. Also, make up an interesting title for each of your drawings and write it at the bottom of each block next to the number of the figure." Participants, Design and Procedure The data for this investigation were collected as part of Experiment Two. There were 80 participants in this experiment, recruited from the University of British Columbia Psychology subject pool as well as from other departments of the university. Fifty-seven were women and 23 men, between 18 and 57 years of age (M age = 23.6). Experiment Two combined manipulations aimed at ProM and manipulations aimed at creativity. The manipulations aimed at ProM involved the distractor and target load of the ongoing target-present / absent decision task, in which the experimental ProM task was embedded. There were four conditions resulting from the factorial combination of the two manipulations. These conditions were described in detail in Chapter Four. In addition a manipulation aimed at probing learning effects on creative test performance was implemented in the experiment. This manipulation, its rationale and its results were not part of the research for this dissertation and will not be discussed in detail. The basic idea was that a subject's performance on a creativity task can facilitate subsequent performance, and that receiving two versions of the same task might facilitate performance to a greater extent than receiving versions of two different tasks. Table 6.4 lists the different combinations of creativity tasks that were employed in the experiment. Twenty subjects received the Uses test twice. Twenty subjects received 147 first the Uses test, then the Picture Completion test. Twenty subjects received first the Picture Completion then the Uses test, and twenty subjects received the Picture Completion test twice. Within each of these four combinations, ten subjects received the order Form A - Form B, the other ten the order Form B - Form A. This design had implications for the present investigation. Measures of each creativity task were available for sixty participants and obtained with two different versions of the task at different times of the experiment. The order of administration of the ProM and divergent thinking tasks is listed in Table 4.1. The general procedures, and those for the experimental ProM task and the post-experiment questionnaire, were described in Experiment Two, Chapter Four. This section outlines only the procedures relating to the administration of the two creativity tasks and the Story ProM task. Subjects received the first creativity task at the beginning of the experiment, the second after 15 to 20 min of mtervening activities. Both tasks were described as investigating ways in which people come up with ideas. For the Uses test, the same procedure as in the first investigation was used, with the exception that subjects were asked to write down their answers. This change was necessary to render the testing conditions for the Uses and Picture Completion tasks as similar as possible. After the practice exercise, subjects were read the instructions printed on the first test sheet, and were then invited to write down as many uses as they could think of for the object printed on the sheet. Subjects were timed unobtrusively. After three rninutes had elapsed, they received the next sheet, and so on, until they had worked on all four sheets. If subjects stated that they had finished prior to the three-min limit, they were encouraged to continue for awhile, after which they received the next sheet, and a note was made of the time taken. Also the Picture Completion test was introduced with a practice exercise. Subjects were given an incomplete drawing, similar to the actual test items. They were told that there were many ways to complete the drawing, and that they should make an interesting picture of it, and give the picture a title. 148 F i g u r e 6.1. P i c t u r e C o m p l e t i o n Te s t , F o r m B, p a g e 1. PICTURE COMPLETION TEST By adding lines to the incomplete figures on this and the next page, you can sketch some interesting objects or pictures. Try to make it tell as complete and as interesting a story as you can. Also, make up an interesting title for each of your drawings and write it at the bottom of each block next to the number of the figure. 1. \ ((• 1-149 Table 6.4. Combinations of creativity tasks and number of subjects who received them. First Creativity Task Second Creativity Task Uses of Objects Test Picture Completion Test Uses of Objects Test Form A -> Form B (n = 10) Form B Form A (n = 10) Form A -> Form B (n = 10) Form B -> Form A (n = 10) Picture Completion Test Form A -> Form B (n = 10) Form B -» Form A (n = 10) Form A H> Form B (n = 10) Form B -> Form A (n = 10) When they had completed the practice drawing, they were read the instructions on the first test sheet, told that they should complete as many of the test drawings as they found possible, in any order they preferred. Also for this test, timing was unobtrusive, and subjects were stopped after 10 min had elapsed. If all ten drawings were completed prior to the ten-min limit, a note was made of the total time taken. Instructions for the Story P r o M task were given after the experimental P r o M test phase. Subjects were told that if in the remainder of the experiment they should come across any animal name, such as cat or squirrel, they should say the word aloud. These instructions were repeated until they were clearly understood. There were two filler activities between these instructions and the test phase of this P r o M task. The first activity was a verbal learning task, in which subjects were presented 18 concrete 150 nouns, one at a time, on the computer screen, and then asked to recall as many as they could, in any order. There were two immediate trials, and a delayed trial, given after the second filler activity. This was a questionnaire asking people to describe their thoughts, motivation, and strategies on the two creativity tasks. The total time required for the two filler activities was from 10 to 15 min. The Story ProM test phase was announced as a reading comprehension task. Subjects were told that they would read a story by the title "The Three Fugitives", presented on the computer screen in eight separate paragraphs. They were instructed to initiate the presentation of each paragraph by pressing the space bar, read the paragraph to themselves once, and move on to the next paragraph by pressing again the space bar. Subjects were told that, during the reading, they were to make sure to understand the main parts of the story because later they would receive a recall test for the story contents. For paragraphs two, four, six, and eight, the experimenter made a note whether subjects did or not remember to say the animal word in each aloud. If a cue was missed, no feedback was given, and at no point were subjects reminded of the ProM instructions for this task. Results This section will first outline the criteria for scoring performance on the ProM and creativity tasks. Next, performance on both types of tasks and the reliability of all instruments will be analyzed. Then the correlations between the ProM and creativity indices will be examined. Finally, the relationship between creativity and various items on the ProM questionnaire will be evaluated. Samng Performance on the experimental ProM task was scored as in the first investigation, by distmguishing between the occurrence of a ProM response and the size of the picture cue at which this response was given. Performance on the Story ProM task was scored as the number of cues successfully identified (there were four cues in total). Three different scores were computed for the Uses and Picture Completion tasks: a fluency score, a flexibility score, and an originality score. For the Uses test, the fluency score was obtained 151 by counting the number of non-repeated responses to each test item. As in the first investigation, this count did not include paraphrases or responses not addressing uses of an item. The item scores were summed and averaged across items, to form a mean fluency score. Inter-rater reliability was not evaluated as the agreement among different raters had been found to be perfect in the Pilot Investigation. To obtain a flexibility score for the Uses test, a group of roughly twenty general categories of uses was identified for each of the eight items. For example, for the item "brick" some categories were "Break things", "Weight", or "Weapon", for the item "button", some were "Counting/Math", "Money/Poker Chips" or "Fasten Qothing". Appendix D lists the entire inventory of categories by item Responses to each item were assigned to one of the categories for the item, and a total count of categories for the item was obtained. The total count of categories for each item was then summed and adjusted for fluency, to indicate the proportion of different categories present in the responses. For example, if a person gave six responses to an item, and five responses fell into different categories, the score for that item would be 5/6 or 0.83. A mean proportion of response categories was computed by averaging across items. Inter-rater reliability for 17 response protocols for the item "brick" was r (15) = 0.92. To score originality for the Uses test, a procedure similar to that used in the first investigation was followed. A lexicon of all subjects' responses to all test items was compiled and the frequency of each response computed. Responses were classified as common if given by more than 5% of the sample, that is, more than 3 subjects; they were classified as unusual if given by 5% or less of the sample, that is, by 3 or fewer subjects 1 3. Sample responses for the two categories are given in Appendix E. Common responses were not further considered, whereas unusual responses were summed for each item, and adjusted for fluency. To illustrate, if a person gave eight responses to an item, and three of these responses were unusual, the score for the item was 3/8 or 0.37. A mean 152 proportion of unusual responses was computed by averaging across items. The scores were based on the statistical infrequency of responses and no rater judgement was involved. For the Picture Completion test, fluency was scored by counting the number of completed drawings. A completed drawing was considered one in which lines were added to the original figure, and a title was given to the drawing. Torrance (1966) reported average inter-rater reliability between Grade 9 teachers to be r = 0.97. Two raters in this investigation separately scored 17 protocols. The correlation between their scores was r (15) = 0.95 for Form A, r (15) = 0.99 for Form B. Flexibility was scored by assigning each drawing to one of 68 categories, published in the TTCT Noms-Techniml Manual (Torrance, 1966). Examples of categories were "Accessories", "Footwear", or "Sea Animals". The number of different categories was summed and adjusted for fluency. To illustrate, if a subject completed eight drawings, and these fell into six categories, the score was 6/8, or 0.75. Torrance (1966) found that inter-rater reliability for flexibility ratings was r = 0.97. Inter-rater reliability in this investigation was r (15) = 0.86 for Form A, r (15) = 0.96 for Form B. The scoring key published in the TTCT Manual was applied to score the originality of the drawings. This key was developed on the basis of drawings submitted by 500 subjects from kindergarten to college age. It distinguishes between three types of responses. Responses that were given by 5% or more of the 500 subjects are awarded 0 points; responses that were given by 2 to 4.99% of the sample are awarded 1 point, and responses that were given by less than 2% of the sample are awarded 2 points. For example, if a subject completes the line drawing 1 in the sample sheet in Figure 6.1 with the picture of a bird, or a heart, he/she receives 0 points. If the line drawing is completed with a cloud, eyebrows, or apples, it is awarded 1 point. If the line drawing is completed with a picture representing other objects or themes than those in the 0- and 1-point group, it is awarded 2 points. The points for all drawings were summed and adjusted for fluency. 1 3 While the total number of subjects participating in the experiment was eighty, responses for each of the Uses items were available only for forty subjects. However, data for another twenty subjects were collected prior to this experiment. 153 For example, if a subject completed five drawings, and obtained a total of eight points, the originality score was 8/5, that is, 1.6 on a scale from 0 to 2. The published inter-rater reliability is r = 0.94 for originality. Inter-rater reliability in this investigation was r (15) = 0.94 for Form A, r (15) = 0.96 for Form B. The Picture Completion test protocols did not only include a picture but also a title, which is normally assessed for originality. The present investigation focused on the visuo-constructional products (that is, the drawings) and did not examine this verbal component of the test. Test Performance and Test Rdiability Table 6.5 shows performance scores on the two ProM tasks achieved by the entire sample, and scores for the creativity tasks achieved by the sixty subjects that received each. For the latter, scores are reported separately for Forms A and B. ProM performance was not at, or close to ceiling. However, it should be noted that the Story ProM task elicited low performance levels, with people on the average responding to a single animal cue. Further examination of the performance on this task revealed that only 36% of subjects responded to any of the animal cues. Performance on the Uses test was almost identical across forms. The mean number of responses per item was roughly seven and a third were unusual. Approximately three responses of four involved a different category. Performance on the Picture Completion test was also very similar across forms. The mean number of drawings produced by subjects ranged from seven to eight. The mean originality scores fell in the intermediate range, with subjects drawing pictures of themes or objects that approximately 5% of the original 500 subjects in the normative study had produced (Torrance, 1966). When performance on Forms A and B of both tests was compared by means of t tests, no significant differences were found, with the exception of the Picture Completion originality score, which was lower for Form A than for Form B. Thus, the response lexicon is based on a total of 60 response protocols for each item. 154 The comparisons between the two forms of each creativity task have important implications for the correlational analyses. They indicate that scores from either form can be used as an index of the abilities measured on the test, with the possible exception of the Picture Completion originality scores. As will be argued later in this section, this provided a basis for disregarding in the correlational analyses the specific version of the creativity task that subjects had received. Test reliability for the ProM measures was evaluated by computing correlations between the experimental and Story task. These were found to be the following: Story task with ProM response, r (78) = 0.23 (p <0.05); Story task with Cue size, r (43) = -0.34 (p <0.05). Compared to the first investigation, this outcome is better as the secondary ProM task correlated with both outcomes of the experimental ProM task. Test reliability for the creativity measures was evaluated where possible in three different ways: By examining the internal consistency of the test, the test-retest correlations, and inter-test correlations. Internal consistency was evaluated for the Uses test, as single item scores were computed. The estimates are based on 40 subjects for each form. The alpha coefficients for fluency were 0.89 (Form A) and 0.85 (Form B). For flexibility they were 0.50 (Form A) and 0.34 (Form B). For originality, they were 0.44 (Form A) and 0.62 (FormB). Clearly internal consistency is adequate for fluency and less than adequate for the other two indices. It is a common finding in the creativity measurement literature that when originality and flexibility scores are adjusted for fluency, their internal consistency drops to relatively low levels (see Runco, 1991). Unfortunately, the scores in this investigation are no exception. Next, test-retest reliability was evaluated for the Uses and the Picture Completion test, for the twenty subjects who received both forms in the same experimental session. For the Uses test, test-retest reliabilities were r (18) = 0.89 for fluency, r (18) = 0.76 for flexibility, and r (18) = 0.24 for originality. For the Picture Completion test, test-retest reliabilities were r (18) = 0.55 for fluency, r 155 (18) = 0.68 for flexibility, and r (18) = 0.17 for originality. Test-retest reliability is good for fluency, acceptable for flexibility, and very low for originality. Published test-retest reliabilities (Torrance, 1966), with college students being tested on both form with a two-week interval, were r = 0.72 for fluency, r = 0.60 for flexibility, and r = 0.63 for originality. These estimates however are based on aggregate scores from three figural tests, and hence, not surprisingly, higher than those found in the present investigation. Finally, correlations between Uses and Picture Completion indices were examined for forty subjects who received both tests. Table 6.6 shows the full correlation matrix. This matrix indicates little overlap between the two tests. Only originality shows some association across tests. Interestingly, within tests, the three scores are more often inversely than positively correlated. That is, highly fluent people, especially on the Picture Completion test, tend to be less flexible, and, to some extent, less original. Also, flexibility and originality scores do not seem to co-vary. In light of Guilford's model, these findings are quite surprising, and some caution is called for in interpreting them. However they suggest that the three abilities implicated in creativity may be more independent of each other than previously thought. Oyrrelations betwen ProM and Creatiuty Scores A significant negative correlation was found between ProM performance and originality scores on both creativity measures. A significant positive correlation was found between ProM performance and flexibility scores on both creativity measures. No correlation was found between ProM performance and fluency scores. Performance data from sixty subjects were available to compute correlations between ProM and creativity. All subjects received both ProM tasks, but not all received the same form of the two creativity tasks. For both the Uses and Picture Completion tests, Form A scores were submitted for forty subjects, Form B scores for the rermining twenty subjects. 156 Performance on the experimental ProM task was assessed under the different ongoing task conditions described in Experiment Two, Chapter Four. One of the ongoing task manipulations (distractor load of the letter displays) had a significant impact on ProM performance, the other (target load of the letter display) did not. To test whether creativity scores might interact with either of the ongoing task manipulations, subjects were assigned to two groups on the basis of their fluency, flexibility, and originality scores on both creativity tasks. Subjects scoring above the median on each index were assigned to the high-scoring group, subjects scoring below the median on each index were assigned to the low-scoring group. ANOVAs with factors distractor load (high versus low), target load (high versus low), and creativity score (high versus low) were conducted for ProM response and for Cue size. No significant interaction between creativity score and the ongoing task variables was found for fluency and originality. However for flexibility, when assessed with the Picture Completion task, not with the Uses test, there were significant interactions both for distractor load, ^ (1,28) =5.5\,MSE = 1.05,p <0.05, and for target load, F( 1,4 8) =A.02,MSE = 0.77, p = 0.05, when ProM Response was considered. The 3-way interaction was not significant. High flexibility scorers performed better than low flexibility scorers when distractor or target load were low. When distractor or target load were high, there was no difference between high and low scorers. It is therefore likely that the relationship between flexibility and ProM depends on the ongoing task demands, and this point will be explored more in depth later on. Table 6.7 lists the correlation coefficients between the three ProM indices and the six creativity indices. There is a clear pattern emerging from the correlation matrix. The ProM index that correlates with creativity indices is ProM response -whether or not the subject responded to the picture cue in the experimental ProM task. Cue size and Story ProM Score do not show reliable correlations with the creativity indices, although for the latter a tendency was noted for r coefficients to go in the same direction as those for ProM response. Table 6.5. Performance on ProM (n = 80) and creativity tasks (n = 40) (second investigation). Task Index Mean and Standard Deviation Experimental ProM Task ProM Response Cue Size Story ProM Task Proportion of responses 0.63 (0.49) Unit from 1 (smallest) to 8.22 (4.72) 19 (largest) Number of cues detected 1.30(1.78) Uses Test Form A Fluency Originality Flexibility Uses Test Form B Fluency Originality Flexibility Picture Compl. Form A Fluency Originality Flexibility Picture Compl. Form B Fluency Originality Flexibility Number of responses 7.41(2.62) Prop, unusual responses 0.29 (0.10) Prop, different categories 0.72 (0.10) Number of responses 7.81(2.45) Prop, unusual responses 0.30(0.14) Prop, different categories 0.72 (0.09) Number of drawings 8.22(2.15) Mean orig. score (0 to 2) 1.10(0.36) Prop, different categories 0.84 (0.17) Number of drawings 7.40 (2.71) Mean orig. score (0 to 2) 1.34 (0.37) Prop, different categories 0.88 (0.13) 158 Table 6.6. Correlations between scores on the Uses and Picture Completion test (n = 40; *= r coefficients with p values < 0.05; ** = r coefficients with p values < 0.001). 2 3 4 5 6 (1) Uses Fluency -.53** .60** -.09 -.10 .18 (2) Uses Flexibility -.24 07 .19 .06 (3) Uses Originality -.26* .06 .31* (4) Pict. Compl. Fluency -.38** -.16 (5) Pict. Compl. Flexibility (6) Pict. Compl. Originality .11 Table 6.7. Correlations between ProM and creativity indices (n = 60; with p values < 0.05). * = r coefficients ProM Response Cue Size Story ProM Uses Fluency -.11 .04 .09 Uses Flexibility .31* .11 .04 Uses Originality -.24 -.02 .06 Pict. Compl. Fluency .00 .28 -.06 Pict. Compl. Flexibility .26* -.21 .21 Pict. Compl. Originality -.35* .13 -.19 159 The overall findings are the following. Fluency, regardless of the task it was assessed with, did not correlate with ProM Response. Flexibility, regardless of the task it was assessed with, was positively correlated with ProM Response. Originality, regardless of the task it was assessed with, was inversely associated with ProM Response. The correlation with Picture Completion originality was significant at the p <0.01 level, the correlation with Uses Originality was marginally significant at the p<0.10 level. To probe the relation between flexibility (assessed by Picture Completion) and ProM further, separate correlations were computed for high and low distractor and target load. The following correlation coefficients were found: r (28) = 0.46 (p <0.01) with low distractor load, r (28) = 0.07 (p >0.05) with high distractor load; r (28) = 0.52 (p <0.01) with lowtarget load, r (28) = 0.16 (p > 0.05) with high target load. These findings consistently indicate that the relationship between flexibility and ProM is conditional on the ongoing task demands. Creativity and Sdf-Reports on Pvospectke Memory Performance The purpose in this section was to determine whether the association between flexibility and originality, and ProM might, at least in part, be based on a different approach to the experimental ProM task. There were differences between highly flexible and less flexible subjects but no differences between highly original and less original subjects, on several task approach parameters. For each of the two creativity indices, a high- and a low-scoring group was formed by a median split on the basis of the Picture Completion scores. Items on the Questionnaire in Appendix B relating to expectations regarding the ProM retrieval context (item 8), the tendency to remind oneself of the ProM task (item 6), and the deployment of a strategy (item 7) were considered. High and low original subjects did not differ in their expectations regarding the ProM retrieval context. Approximately half of each group had no expectation or an incorrect one, the other half the correct expectation that the cue would appear during the decision task trials. They also did not differ on how often they reminded themselves of the ProM task during the experiment: 160 on the average both groups thought about the task now and then. High original subjects were as likely as low original subjects to use a task-specific strategy, which identified the decision task as the likely occasion of cue appearance. Task-specific strategies were implemented by approximately 20% of all subjects. For flexibility the findings are far more revealing. High-flexibility subjects were marginally more likely than low-flexibility subjects to expect the correct ProM retrieval context, chi-squareif) = 2.67, p <010. They reminded themselves significantly more often of the ProM task than low-flexibility subjects, r(52) = 2.17, p <0.05. Finally, they also appeared more likelyto implement a task-specific strategy than a task-general one or none: 31% of subjects from the high-flexibility group but only 15% from the low-flexibility group reported some type of strategy aimed at monitoring the pictures during the decision task. Discussion This investigation aimed to re-examine and extend the observation that a negative association may exist between ProM and creative tliinking. It also aimed to determine whether the relationship between ProM and creative thinking is based on differences in the way people approach the ProM task, for example, their expectations regarding the retrieval context, or their use of a strategy to meet the ProM task demands. It was found that successful ProM was predicted across creativity measures by high levels of flexibility but low levels of originality. There was no association between ProM and fluency It was also found that flexibility had an impact on people's expectations regarding the ProM retrieval context, on their tendency to remind themselves, and possibly, on the use of a monitoring strategy during the ProM test phase. The findings from this investigation confirm and complement the observations made in the first investigation. A negative relationship consistently emerges between ProM and original thinking whereas no systematic relationship seems to exist between ProM and fluent thinking. Moreover, it is likely that a positive relationship exists between ProM and flexible tliinking. It should be noted that 161 in this investigation the originality and flexibility indices had low reliability, and that the observed correlations may not accurately reflect the true relationship between ProM and creative tlimking. When discussing the results from the first investigation, several possibilities were considered to explain the observed relationship between original tliinking and ProM. One was that highly original subjects might be less compliant with the experimental procedures. There was no empirical support for this possibility in the first investigation, and neither is there in this investigation. Highly original subjects performed the tasks assigned to them (for example the letter decision task or the verbal learning task) as well as subjects who had lower originality scores. Another explanation was that highly original people might have become bored with the letter decision task as trials progressed, and simply paid less attention, in this way missing the ProM cue. Once again results from neither investigation bear out this possibility. Highly original subjects performed the critical block of trials in which the ProM cue appeared with the same speed and accuracy as less original subjects Another possibility explored in the present investigation was that the relation between originality and ProM might be specific to verbal abilities. It was suggested that originality, as manifested on the divergent tliinking tasks, might reflect the tendency to play with language, and this tendency might interfere with the creation of a simple verbal rule for the ProM task. To probe this possibility, a visually based creativity task was administered in addition to the verbally based divergent thinking task. Original performance on both tasks was inversely related to ProM, indicating that the relationship between original thinking and ProM is not limited to verbal processes and involves deeper and more general abilities. At the outset of the present investigation, it was reasoned that highly creative people understand and act upon ProM tasks in ways that are fundamentally different from those of people who are less creative. The present investigation revealed no support for this idea in relation to 162 original thinking. Original people were as likely as less original people to have accurate expectations regarding the ProM retrieval situation, and to engage in thoughts about the upcoming task. . * Two explanations for the results from the first investigation focused on a more principled kind of relationship between ProM and original thiriking. One idea was that the ProM cues and their associated actions were too obvious for original thinkers and that a different relationship might be found if more subtle ProM cues, such as the animal names in the Story task, were employed. It was reasoned that these cues required more than one thinking step to be recognized as a signal for an intended action. Correlations between originality scores and ProM as assessed by the Story task were low and failed to reach statistical significance, but, importantly, one (with Picture Completion scores) was of the same sign as those found with the ProM Response measure on the experimental task. However, it must be kept in mind that the Story task did not assess ProM abilities as adequately as the experimental task because on the former people performed close to floor. The drop in correlational strength is more likely the result of reduced variability than of a true change in the relation between original thinking and ProM. In fact this ProM measure did not correlate significantly with any creativity index. The other explanation for the inverse relation between ProM and original thinking was based on the view of creativity in terms of low control over mental processes and the resulting impact of irrelevant external events. It was speculated that this lack of internal control might affect ProM because of the difficulty to stop the processing of ongoing events. Highly original people may have failed to recognize the ProM picture cue because they were unable to ignore the pressing stimulation from the ongoing task -the continuous appearance of the letter displays and the auditory feedback given when responses were slow. If this was the case, then these people should have had comparatively greater difficulty than less original people with the ongoing tasks involving high distractor load. As was noted in the Results section, there was no interaction between originality 163 scores and the experimental ProM manipulations. Also, no reliable inverse relationship emerged between search efficiency (indexed by reaction times to single targets) and originality scores. In sum, while the observed relation between original tliinking and ProM does not appear to be an artifactual result of the experimental setting and is not specific to a particular way of testing originality (i.e., verbally mediated), no compelling argument can be made regarding the possible bases of this relationship. It is possible that the nature of the ProM cue plays a role but it is also possible that the link lies in original people's purported inability to control the processing of mcoming stimulation. Some suggestions will be made in the next section on how to further investigate these possibilities. The link between flexible thinking and ProM is explained more readily. This investigation found a positive relationship between the number of ideational categories invoked while generating uses for an object or completing different line drawings, and the likelihood to respond to the ProM picture cue. There was a tendency for this relationship to be manifest under low but not high ongoing task load. People with high flexibility scores reported more often thinking about the upcoming task, identifying the letter decision task as the likely ProM retrieval context and implementing strategies to monitor this task context for the appearance of the cue than people with low flexibility scores. This pattern of findings suggests that the flexible thinkers in this experiment, achieved ProM by means of deliberate, strategic shifts of attention between the ongoing context and the context of the intended action. These people seemed to periodically bring to mind the ProM task, and to monitor the ongoing activities for a likely retrieval context. Those who identified this context then went on to intensify their watchfulness for the ProM cue. The less flexible thinkers instead tended to approach the ProM task in a more haphazard manner, thinking about it only rarely and considering the ongoing context in a more single-minded manner. 164 It is interesting that the relationship between flexibility scores and ProM was found more reliably when ongoing task demands were low than when they were high. It can be speculated that flexible thinkers were able to bring to bear their strategic skills on the ProM task to a greater extent when the ongoing task demands were comparatively low, that is, when there were few distractors to sort through or a single target letter to look for. When the ongoing demands increased, that is, there were several distractors to sort through or multiple target letters to look for, these strategic skills were probably less employable, and the overall approach to the ProM task was similar for flexible and less flexible thinkers. Conclusions to Part I I The overall goal of Part II was to explore the relationship between ProM and creative thinking. Conceptually and empirically based arguments led to the hypothesis that a link exists between these two domains, and to ask the question whether it is ProM success or ProM failure that is predicted by high levels of creativity. The outcome of the two investigations aimed at answering this question seems itself the product of a creative insight: Creativity can predict both ProM success and ProM failure. This puzzling conclusion must be related to how creativity is understood in the present work. Creativity is seen as a construct made-up of three distinct thinking abilities: fluency, that is, the facility of thought, flexibility, that is, the ease of shifting thought between categories, and originality, that is, the ability to propel thought in a new direction. It was reasoned that all three abilities play a similar role in sustaining ProM retrieval. The evidence clearly contradicts the idea of a common role, and requires making distinctions between the three thinking abilities. In fact, both investigations consistently show no relation between fluent thinking and ProM. Also, they consistently show an inverse relation between original thinking and ProM. Lastly, the second investigation consistently shows a positive relation between flexible thinking and ProM. 165 Flexible thinking is the only of the three creative thinking abilities that truly seems to sustain ProM. The evidence from the second investigation suggests that flexible tliinking is instrumental to ProM in mamtaining a state of readiness for the ProM task. This might be achieved as follows. When people are given ProM instructions, in actuality they are receiving instructions for two tasks. One task is any given activity they are asked to perform in the experimental session. The other is to override that activity in the presence of a designated event whose time of occurrence is not known. To the extent that people think flexibly, they are able to strategically bring to mind the ProM task in the course of ongoing tasks, to secure, as it were, room for ProM-related thoughts in their ongoing thoughts. In more general terms, flexible thinkers seem to do in ProM situations what Graf and Uttl (2001) alluded to when they compared ProM to prospecting for underground mineral deposits. They seem to strategically look ahead, mindful of alternate goals and ready to act upon them when the opportunity arises. Also, to the extent that people think flexibly, they might be more ready to let go of ongoing thoughts or activities and to switch over to new ones; Obviously this kind of ability is fundamental in the context of ProM, where intermpting an ongoing activity and the related mental processes might well be one of the most significant challenges. The inverse relation between original tliinking and ProM has resisted every attempt to explain it on the basis of the available data. It was suggested that original thinkers might in some way overshoot the target. They might represent the retrieval context in some unusual manner, for example imagining the picture cue appear behind them and reveal itself faintly in the reflection on the computer screen. As well, they might produce unusual associations to the cue event, for example thinking of footprints in the snow when receiving the participation credit form. The key question emerging from these speculations is whether there are situations in which original thinking can work in ProM's favor. To test this possibility, the ProM cue event could be manipulated in various ways. For example, it could be manipulated to elicit either a novel or a habitual response within the ongoing context. It might turn out that original people outperform less original people when a 166 habitual response must be counteracted to make a ProM-relevant response. Alternatively, the cue might be strongly or weakly associated with the ProM action. It is possible that when it is weakly associated, original rather than less original people are at an advantage. It is however also possible that original thinking is in principle incompatible with ProM, because of the inability to disregard irrelevant information, possibly associated with creativity. This inability has been indicated by some theorists as the foundation of creative thought (e.g., Eysenck, 1993). In Part I of the present work, the ability to ignore information has been identified as a key resource for ProM. In short, a "see-saw" relationship could exist between original thinking and ProM, in which mechanisms supporting one activity disrupt the other activity. The present conclusions would be incomplete without briefly evaluating the counterintuitive finding that the three creative thinking abilities seem to have very little in common with each other or indeed correlate negatively, such as fluency, and particularly flexibility, with originality. How surprising is this finding and to what extent does it call into question the idea developed until now that these thinking abilities relate differently to ProM? Guilford's idea was that the three creative thinking abilities are highly related, that, in fact, thinking would be original only to the extent that it was fluent and flexible (Guilford, 1959, 1975). But his view has been challenged by empirical findings that fluency, flexibility, and originality scores correlate only moderately or even inversely with each other, particularly when flexibility and originality scores are adjusted for fluency (Baer, 1994; Hocevar, 1979a). There are also theoretical arguments that question the cohesiveness of the creative thinking abilities. Divergent thinking tests like the Uses of Objects test employed in the present investigations, measure rote as well as creative ideation. Associative models of creativity (e.g., Mednick, 1962) predict that individuals engaging in rote ideation are actually likely to produce a larger number of responses than individuals engaging in creative ideation, due to the higher availability of stereotyped responses. On this basis, no relationship or an inverse one would be expected between fluency and originality scores. Also, it is possible that there is more than one way to think creatively. One way might be to exert high control over one's thinking and to detach thinking as much as possible from external events. This way of thmking creatively might be what this work described as flexibility. Another way might be to relinquish control over one's tliinking and let it drift freely, carried by the flow of external events. This way of tliinking creatively might be what this work described as originality. B y this reasoning, flexibility and originality would not necessarily have to have anything in common. In fact they might well be somewhat antithetical, and this is consistent with the indication emerging from the present work, that they play very different roles in the context of P roM. 168 CHAPTER SEVEN: GENERAL CONCLUSIONS The present research investigated prospective memory (ProM), a type of memory that is prominent in everyday life but has been largely ignored in psychological research until very recently. Though much knowledge has been accumulated on ProM in the past two decades (cf. Brandimonte et al., 1996; Ellis & Kvavilashvili, 2000), unifying concepts are only now emerging and theoretical organization is not yet specific enough to generate exact experimental predictions (McDaniel & Einstein, 2000). No definite agreement has yet been reached on such fundamental issues as how ProM should be defined and whether/how it is different from retrospective memory (RetM). The student of ProM still feels somewhat like a pioneer and it is with this frontier spirit that the present research was carried out. ProM plays a critical part in some situations of intention fulfillment. For this research I defined ProM as the ability to discover in some aspect of the present moment the cue to recollect an intention that was formed at an earlier time. This could be an event, for example, walking past a store, or a clock time, for example, 8:00 p.m. This research focused on event-cued ProM. I distinguished ProM from vigilance, that is, actively monitoring the ongoing situation for the opportunity to fulfill an intention. In ProM situations the intention to be fulfilled is not on people's minds and their attention is directed to ongoing activities. I also distinguished ProM from habit, that is, remembering routine actions such as brushing one's teeth in the evening. Thus, ProM in this work is intended as a unique form of long-term episodic remembering. The present research had two separate goals. One goal was to investigate the type of attentional resources that we must bring to bear on the retrieval context in order to discover the ProM cue and regain awareness of an earlier intention. The other goal was to investigate whether we need to have an open mind and see alternatives in order to discover ProM cues, in short, whether the ability to think creatively is related to ProM. 169 Motivation, Goals and Strategies of This Research The goal concerned with the attentional resources required for ProM built on theoretical models of ProM retrieval that view the mental operations underlying this type of remembering as effortful and attention-demanding because of their self-initiated or executive character (Craik, 1986; Shallice & Burgess, 1991). In this respect, ProM tests are similar to RetM tests of free recall, and actually may draw more substantially on the rememberer's resources than the latter (Craik, 1986). But the attentional configuration of ProM tests is very different from that of RetM tests: most or all of the rememberer's resources are directed to ongoing activities (Graf &Uttl, 2001). A resource-demanding reorganization of this configuration must occur for the ProM cue to enter awareness and to initiate a shift in thought and action. Research into the kind of attention or resources that enable ProM can shed light on the mechanisms underlying this type of remembering. There is solid evidence that indeed ProMdepends on the availability of attentional resources. ProM performance has been shown to drop if the ongoing activities the subject is engaged in require high levels of attentional resources than if they require lower levels of resources (e.g., Marsh & Hicks, 1998). ProM performance has also been shown to decline with age (e.g., Uttl et al., 2001), and it is generally accepted that with increasing age there is a depletion of attentional resources. But evidence on the type of attentional resources that are involved in ProM is scarce. Thus, I began this research with no a priori assumptions about the nature of these resources, and as a starting point distinguished two broad types of resources: processing speed and processing capacity. The subsequent focus was on processing capacity, and a new distinction was introduced for the present research that set apart perceptual resources for searching external stimuli from working memory resources for rmtching these stimuli to internal targets. The primary research strategy was experimental and aimed at manipulating the attentional demands of the ongoing task in which the ProM task was embedded, and to evaluate the impact of the manipulations on ProM performance. The manipulations were designed to selectively induce the 170 utilization of distinct types of resources in the ongoing activity, thereby constraining their availability for ProM. A secondary strategy was correlational and aimed at investigating the relationship between individual differences in the availability of attentional resources and ProM. The following paradigm was employed to measure ProM and implement the experimental manipulations. Subjects were shown a picture -the ProM event or cue- on the computer screen and were told that, if they saw this picture again in the experiment, they should stop whatever activity they were engaged in, and carry out a specified action. Subjects were then given a variety of ProM-unrelated tasks with the purpose of directing their thoughts and attention away from the ProM task. At some point, they were presented with letter displays on the screen and asked to respond quickly and accurately in specified ways. The specific tasks subjects received for the letter displays embodied the attentional manipulations. After a number of trials, pictures were shown along with the letter displays. In these pictures was embedded the ProM-cue picture shown repeatedly at increasing sizes until subjects responded to it as instructed or until the letter display task was completed. The ProM cue was a natural part of the situation and did not claim subjects' attention. Thus the uniquely prospective burden of identifying this picture as relevant to the earlier instructions was with the subjects. But this burden did not remain constant across presentations. The picture cue was made more noticeable and intrusive by increasing its size over repeated presentations. It was assumed that the more noticeable the cue, the lower was subjects' own ability to discover it, that is, the lower their own mnemonic contribution. Thus there were two outcome measures. The ProM response indicated whether or not a subject responded to the picture cue, regardless of its size; the cue size indicated how large the picture cue was when the ProM response occurred. It should be noted that size was not a pure index but was confounded with repetition. In Experiment One ProM's demands on processing speed and processing capacity were investigated. For the ongoing task assumed to require processing speed, subjects were given a simple reaction task. They were asked to respond as quickly as possible to the onset of new letter displays. 171 For the ongoing task assumed to require processing capacity, subjects were given an A/B choice reaction task. They were asked to decide as quickly and accurately as possible, which of the two target letters was in the letter displays. Performance on these tasks was taken as an individual difference measure of processing speed and processing capacity. Experiment Two investigated ProM's demands on searching and rmtching-to-memory. The ongoing task required subjects to make target present/absent decisions about the letter displays. To manipulate demands on searching, the displays had either a low or a high load of distractor letters. To manipulate demands on matching-to-memory, the displays had either a single target letter, or one of several potential target letters. The second goal of the present research was concerned with the relationship between ProM and creative thinking, and built on intuition as well as on a body of ideas emerging from the literature that ProM may involve special types of thinking patterns. Among these patterns, the abilities of switching train of thought and rejecting habitual ways of thinking were particularly emphasized (e.g., Ellis, 1996; Schonfield, 1982). These abilities are strongly reminiscent of creative thinking. I reasoned that a relationship ought to exist between ProM and creativity. In a typical ProM situation, the thoughts and events of the present moment are unrelated to ProM and may even be conspiring against ProM when the event designated as the ProM cue elicits highly routinized thoughts and actions rather than thoughts about a non-routine intention. A traffic light is such an event as we are strongly inclined to respond to it in the current context of driving rather than in the context of an intention formed earlier to make a stop and buy a grocery item. For the purpose of the present research I defined creativity in terms of divergent thinking (Guilford, 1959). Divergent thinking is thinking in different directions and seeking alternatives. It is assumed to be made-up of three abilities: fluency, flexibility and originality of thinking. I argued that all three abilities may facilitate ProM and thus a positive relationship should exist between ProM and these abilities. At the same time I raised a note of caution. Some theorists understand creativity as 172 something that "happens" when an individual exerts low control over the processing of mcoming stimuli, such that they become interconnected in unusual ways. If this were the mechanism underlying creative thiriking, then it would likely not be conducive to ProM. Thus I began this part of the present research by postulating that ProM and creative thinking appear in some important way related, but that no basis exists to predict whether the sign of their relationship is positive or negative. The research strategy to address the relation between ProM and creative thinking was correlational. The general approach was to present subjects with different ProM tasks and different tasks assessing creative thinking. In the first investigation subjects received the experimental ProM task and a naturalistic ProM task requiring subjects to make a comment on the experiment upon its completion. They also received two divergent thinking tasks, which required thinking of all the possible uses for some common objects and of all the ways in which pairs of objects were alike. Performance on these tasks was scored for fluency and originality. In the second investigation subjects received again the experimental ProM task and a second ProM task for which they were instructed to say aloud the names of any animals they should come across. The animal names were embedded in a computer-presented story passage. Subjects were given two different creativity tasks. One was the task requiring thinking of possible uses for objects. The other was a picture completion task in which subjects were given incomplete line drawings and asked to complete these to make interesting pictures. Performance on these creativity tasks was scored for fluency, flexibility and originality. Core Findings of This Research With respect to ProM and attention, both experiments consistently showed that ongoing demands on attention do interfere with ProM performance. But not all ongoing demands on attention interfered to the same degree. A solid partem emerged that indicated significant interference between ProM and ongoing tasks requiring searching through distractor items. 173 Experiment One showed that subjects performing the A / B choice reaction task were less likely to discover the embedded ProM-cue picture and required larger cue sizes when they did discover it, than subjects performing the simple reaction task. Also, subjects perforating the A/B choice reaction task were more likely to respond to the cue picture when the concomitant letter display had no distractors, than when it had four or eight distractors. Experiment Two showed that subjects making target present/absent decisions about letter displays with a high number of distractors, were less likely to discover the ProM cue than subjects who made the same decisions about letter displays with a low number of distractors. The former also required larger cue sizes than the latter when they did discover the ProM cue. Experiment Two also showed that subjects who made present/absent decisions for multiple potential target letters were as likely to discover the ProM cue as subjects who made decisions for a single target letter, and the former required the same cue size as the latter. The effect of distractor letters on ProM was more evident when decisions were made about a single target letter rather than multiple potential target letters. These findings implicate the processing capacity resource in ProM. Processing capacity is understood as the total amount of items the cognitive system can process at any given time. At the same time they point to a specific processing locus that cannot be accommodated within the processing capacity construct. To review, when the perceptual demands of the ongoing task are increased, there is a cost for ProM but when the working memory or executive control demands of the ongoing task are increased, there is none. Thus, the resource critical for ProM is searching, that is, scanning mcoming information for relevant items. The correlations between individual difference measures of resource availability and ProM support the experimental findings. In both experiments, measures of search speed were correlated with ProM when identification/response selection speed was held constant. Specifically, in Experiment One, an inverse correlation was found between search slope and ProM Response but 174 no correlation was found between search intercept and ProM. In Experiment Two, reaction times to single targets were found to be negatively related to ProM Response and positively related to Cue Size; reaction times to multiple potential targets were not found to be related to either ProM measure. The balance of these findings is consistent and clear. In these experiments the discovery of the ProM cue hinged on the presence of distractor items in the ongoing task. If there were distractor items to process, the discovery of the ProM cue picture was more unlikely, and at the very least required a large and noticeable size. It was concluded that ProM requires attentional input early on in the processing of the cue. With respect to ProM and creativity, the findings were consistent across investigations for performance on the experimental ProM tasks. The following pattern was found. Original thinking, operationalized as the number of unusual ideas or drawings, that only few people in the overall sample produced, was inversely related to ProM. This relationship emerged in the first investigation for the verbally based divergent thinking measures, and in the second investigation also for the visually based creativity measure. Flexible thinking, operationalized as the number of different categories invoked in the responses, was positively related to ProM. This relationship was addressed only in the second investigation, but it emerged both for the verbal and the visuo-constructional creativity tasks. People with high flexibility differed from people with low flexibility in adopting a more strategic approach to the ProM task, for example, by reminding themselves of the task and by fonning expectations regarding the retrieval context. Fluent thinking, operationalized as the total number of ideas or drawings produced, was not found to have a relationship with ProM. The conclusion emerging from these findings is that patterns or abilities of thinking are important in relation to ProM. The general label creative thinking is not useful but the distinction between flexibility and originality is, as it indicates that thinking, depending on its nature, can either facilitate, or interfere with, ProM. 175 Theoretical Implications The present research has provided evidence that ProM retrieval is mediated by attentional processes. This evidence is relevant in the context of current theorizing about ProM and attention. One theoretical view is that ProM retrieval is in principle an effortful process that draws on the system's attentional resources. This view has been most clearly articulated by Craik (1986) and by Shallice and Burgess (1991). Craik focused on the comparison of ProM tests with other memory tests and concluded that ProM tests rely heavily on effortful processes that are self-initiated by the rememberer. Shallice and Burgess linked their view of ProM to the Norman-Shallice model of action control (cf. Norman & Shallice, 1986) and indicated the intervention of the supervisory attentional system in ProM retrieval. A second theoretical view is that ProM retrieval may take advantage of two different types of processes. According to this multiprocess framework (McDaniel & Einstein, 2000), ProM retrieval could occur by means of a resource-demanding process that strategically monitors the environment, and in this way eventually identifies the opportunity to act on an intention. Alternatively, ProM retrieval could occur by means of a resource-saving automatic process that brings to mind the intention to act upon encountering a triggering event. The system will use one or the other in an opportunistic fashion, but the reliance on resource-demanding strategic processes is most likely when there is little or no processing of the ProM cue in the context of the ongoing task. At a general level, this research supports the view that ProM requires attention. It cannot contribute to the issue whether attentional demands characterize all ProM situations or only some. It cannot rule out that under task conditions different from the ones in the present experiments, in which for example the ProM cue is more fully processed or attended as part of the ongoing task, attentional manipulations of the type implemented in this research might not have effects. It is not yet possible to evaluate this possibility because the variables defining extent of cue processing have yet to be articulated clearly. 176 The important contribution of this research lies at a more specific level. This research has provided the first empirical demonstration that ProM requires attentional input for the perceptual processing of the event designated to cue the retrieval of an intention. This evidence suggests that perception is an important stage in the chain of mental events that eventually lead to the fulfillment of an intention to act. It was suggested that processing of the ProM cue competes for resources with processing of ongoing stimuli. It was also speculated that the infusion of resources into cue processing enables a process of figural emphasis. In this way the cue event is selected over other ongoing events whose processing is attenuated or interrupted. With respect to the existing theoretical views, this evidence raises the question where exactly the effortful processes involved in ProM begin. Up until now, theoretical views have assumed implicitly (Craik, 1986) or explicitly (Burgess & Shallice, 1997; McDaniel & Einstein, 2000; Shallice & Burgess, 1991) that attention is required in ProM retrieval to sustain the voluntary, strategic processes of the central executive. These processes might be reminding oneself that a task needs to be carried out at a future time (Craik), taking control over one's thoughts or actions (Shallice and Burgess), or monitoring the environment for cue events (McDaniel and Einstein). Thus it can be said that according to these views ProM is a conscious, willful act of reorganizing one's thoughts and actions. This claim is not easily reconcilable with definitions of ProM such as the one adopted in the present work, that emphasize the engagement of conscious awareness in ongoing activities and thoughts, not in the to-be-remembered intention (cf. Graf &Uttl, 2001), at the time of retrieval. If by definition our thoughts are absorbed in other things it seems unlikely that we willfully bring ourselves to remember. The proposal that ProM retrieval begins with the attentional selection of the cue event, may at least in part overcome this problem. This proposal strengthens the idea that discovery or recognition of "telltale signs" in the environment (Graf & Uttl, 2001) is the defining process of ProM, its unique attribute. It postulates a plausible psychological process for the term discovery. We 177 do not deliberately search for the cue event, we do not automatically notice it, but we emphasize it perceptually in a way that it acquires mental prominence or attensity. At this point it is possible that the postulated executive control processes take over to evaluate the circumstances and shift course of action. The focus on the perceptual events of ProM advocated here has some important implications. First, it broadens the scope of the speculations on the cognitive mechanisms underlying ProM retrieval. At the very least, there seem to be perceptual as well as more central mechanisms involved in this type of memory. Second, it effectively counters the argument examined in Chapter One that event-cues provide more support for retrieval than time cues, and that therefore event-cued tasks are less resource demanding and easier to perform than time-cued task (Einstein & McDaniel, 1996). The main idea emerging from this research is that processing an event in a way that leads to identifying it as a ProM cue, demands resources. Whether the mechanisms underlying the identification of event and time cues are similar, remains to be established and will be more closely examined in Limitations of this Research. Third, the focus on perceptual events in ProM leads to conceptualizing the retrieval context in terms of an entire scene rather than an isolated event. It draws attention to the other items present in the scene that might have a considerable impact on ProM. But the proposal that ProM retrieval is mediated by selective attention mechanisms raises its own set of problems. If ProM by definition does not involve a consciously controlled search for the cue event, how does the system "know" when to select certain perceptual features, for example the color or contour of the butterfly picture? Could it be that in the time interval between setting up an intention and encountering the retrieval context some kind of preparedness is maintained for the cue features? Several authors have considered this possibility. For example, West et al. (2000) have suggested in their ERP study that the process of noticing the cue maybe achieved through a state of attentional preparedness of the neural systems that support processing of the defining features of 178 that cue. In other words, if the defiriing feature of the cue is a color, this preparedness will involve neural systems for color processing, whereas if the defiriing feature is a letter case, it will involve neural systems for letter processing. Marsh and Hicks (1998), in their study on working memory components and ProM, speculate that ProM may involve the formation of a production rule, for example "if fruit then respond". This production rule lies in wait until its condition is matched by the ongoing events. Burgess et al. (2001), in their PET study of brain regions involved in ProM, have considered the possibility that the retention period in ProM may not be "cognitively silent" but in fact may involve some form of anticipatory processing in which certain search parameters are specified. There is currently no known mechanism for such long-term non-conscious state of preparedness (see for example Plude et al., 1994, for a review of expecting mechanisms). However, a reasonable starting point would be to link research and theorizing on ProM retrieval to investigations of how intentions are represented in memory. There have been empirical demonstrations that intentions are represented in long-term memory with a heightened level of resting activation (Goschke &Kuhl, 1993; Maylor, 2000). It is possible that these representations include not only the content of a to-be-performed intention, but also a specification of the characteristics of the retrieval context. This research has also produced evidence that the type of thinldng one engages in when finding oneself in a ProM situation, may play an important role in whether memory will be successful or not. Prior research has shed light on the important influence the content of ProM-concurrent thoughts, or more accurately, concurrent processing has on ProM. Meier and Graf (2000) demonstrated concurrent transfer-appropriate processing (TAP) for ProM performance. Subjects were presented with words for two types of ongoing activity a semantic decision task (does the word represent something natural or man-made?) and perceptual decision task (does the word have two or fewer enclosed spaces?). Half of the subjects performing each task were given a 179 semantic ProM cue (the name of an animal), the other half were given a perceptual ProM cue (words containing three e's). ProM performance was higher when ongoing processing and processing required for the cue matched (e.g., semantic-semantic) than when they did not match (semantic-perceptual). These findings suggest that the closer the content of one's thinking to the ProM cue content, the more likely is the discovery of the latter. For example, if one reflects on whether a word represents a natural object or not, one is more ready to identify an animal name as such than if one examines a word for enclosed spaces. This research suggests that not only the contents of ongoing thoughts but also their structure play a role in ProM. In light of this possibility it appears necessary to develop a comprehensive framework that incorporates the various ideas on ProM and thinking reviewed in Chapter Five, and the key notions from concurrent TAP. Methodological Implications The ProM paradigm in this experiment yielded two measures. One was a binary measure, indicating whether the subject had or not responded to the cue. The other was a continuous measure, indicating the size of the cue at which the response occurred, on a scale of arbitrary units. An important observation emerging from the present research is that all effects produced by the experimental manipulations showed up on both measures. This lends substance to the findings and at the same time suggests that the cue size measure indexes in a more fine-grained manner the same abilities as the more conventional binary measure of success or failure. ProM performance is first evaluated in terms of whether the subject was successful at responding to the ProM cue. The response is then subjected to an additional measurement in terms of how much stimulation from the cue -number of repetitions and/or sheer size- is needed to build up the ProM response. The use of two performance measures is recommended in future studies on ProM as it generally permits stronger confidence in the experimental fmdings and may possibly reveal more subtle effects by means of the cue size measure. 180 This pattern was not observed for the correlational findings. Most of the significant correlations, whether with measures of search efficiency or creative tliiriking, were with the ProM response measure, and very few with the cue size measure. A corrected cue size in which values one unit above the largest actual value were assigned to people who failed to respond to the cue, did not yield any valuable information beyond the ProM response measure. In future correlational studies, researchers might increase the usefulness of the cue size measure by ensuring that all or most subjects obtain an actual score on it. This might be achieved by adding larger cue presentations such at virtually no subject can fail the ProM task. Of course in such studies one would have to relinquish the binary performance measure. There is also another source of difficulty for correlational research on ProM. This research found very modest inter-test correlations between different ProM tasks, ranging from r = 0.08 to 0.34. The most likely reason was that performance scores were based on single observations that are inherently unreliable. But the possibility that different ProM tasks do not reliably measure the same abilities must also be considered. One important goal for researchers in the field should be to develop procedures for assessing the reliability of ProM tests. Because ProM is by definition based on single observations, even when they are administered in the same task (cf. Einstein & McDaniel, 1990), the only way to approach this issue is by means of test-retest assessments. Massive learning effects must be expected. In fact ProM can be considered similar to problem-solving: once the "solution" is found, the re-administration of the task does not measure the same ability. In my understanding, the only viable strategy is to construct two versions of a ProM task that capitalizes on a continuous measure such as cue size. These versions should be equated on all variables known to be important, but should not appear the same to the subjects that receive them. This research stressed the importance of identifying and putting into practice procedural requirements to elicit ProM rather than vigilance or monitoring in response to instructions to do something at a later time. These requirements were that some time had to elapse between the ProM 181 instructions and the test phase, and that subjects needed to be actively engaged in ProM-unrelated activities during this interval and at the time the ProM cue was presented. They were incorporated into the experimental paradigm by administering two ProM-unrelated tasks for a total duration of 10 to 15 min between the instructions and the test phase. This research also employed a post-hoc strategy for testing whether subjects did or not divide their attention between performing the ongoing task and monitoring for the cue. This strategy consisted of comparing reaction times to the ongoing trials prior to and following the occurrence of a ProM response. The reasoning was that if subjects showed slowing prior to the ProM response, it was likely that they were indeed allocating some of their attention to spotting the ProM cue. There was no evidence of slowing in Experiment One or Experiment Two. However, when subjects were asked in a post-experiment questionnaire whether they had employed a particular strategy to spot the ProM-cue picture, 21% of subjects indicated that they had in some way divided their attention between performing the ongoing task and spotting the ProM-cue picture. Evidently, a filled delay was not sufficient to prevent these subjects from adopting a monitoring strategy and from bringing to mind the ProM task well before the cue was actually presented. But their strategy of dividing attention was not evident in their ongoing performance: it was only revealed by the questionnaire. The implication is that the filled delay in itself, widely employed in studies on ProM, is not sufficient to completely divert subjects' thoughts from the ProM task. Most experiments on ProM, including the ones conducted for this research, have a feature that may contribute to this difficulty. When subjects are given ProM instructions to respond to a certain item, whether a picture or a word, should they come across it, they might well create an additional rule for themselves, such as "pay attention when a task in the experiment involves words (pictures)". In this way they may ask themselves at the beginning of each task whether it presents a likely retrieval context or not. It is possible that the characteristics of the tasks administered prior to the ProM test activity determine 182 the particular stance the subject adopts during the test activity. In the present experiments, no task prior to the ProM test activity presented pictures, and this may have induced some subjects to switch to a state of vigilance when the ProM test activity was administered. There are two options. It is possible to define as a ProM cue something that is present throughout the experiment, for example the background pattern of the computer screen (cf. Park et al, 1997). Alternatively, it is possible to administer a variety of "lure" tasks prior to the actual ProM test activity, that present items like the ProM cue but not the ProM cue itself. Limitations of This Research The ProM paradigm used in this research presented letter displays with one to nine items in them, centered on the computer screen. These displays were the stimuli for ongoing tasks requiring a high degree of attention. In addition, quadruplets of distractor pictures were presented laterally at the top or bottom of the computer screen. Among these pictures, in different locations, was presented the ProM-cue picture, whose size across presentations increased to three times the size of the largest distractor pictures. Clearly, this task makes substantial perceptual demands on subjects in order to satisfy the ongoing activity and at the same time spot the ProM-cue picture. There are two important questions that arise with respect to the features of this paradigm To what extent did it measure memory abilities and to what extent also the ability to simply see the pictures? And consequently, how task-specific are the findings on ProM's resource demands from this research? Would they generalize to different types of ProM situations? Several arguments can be made to support the claim that indeed the paradigm measured ProM. In both experiments, a substantial number of subjects (~20%) discovered the cue when it was either smaller than, or equal to, the largest distractor picture size (3x4 cm). The questionnaire administered in Experiment Two revealed that the majority of subjects who did respond to the cue, saw it appear two or more times before they interrupted the task. The reasons they reported for not responding immediately were not remembering the relevance of the picture and having difficulty to 183 interrupt responding to the letter displays. These difficulties appear strongly related to ProM not to problems in seeing the picture. Furthermore, in both experiments performance on the experimental ProM task was correlated with performance on the secondary ProM task, and though correlations were modest, they were no smaller than correlations found between ProM measures in other studies (for example, Jacova et al, 2001). But one finding challenges these arguments. Most subjects who failed to respond to the picture cue, when asked whether it had appeared in the course of the experiment, indicated not having seen the picture. There are two possible interpretations. Either subjects really did not see the picture and this would weaken the claim that the paradigm always measured ProM. Or, as suggested in Chapter Four, they failed to render it perceptually distinctive and hence to remember it when asked later, which is exactly what this work has indicated as the mechanism critical for ProM. For a related phenomenon -inattentional blindness- both accounts, seeing but not remembering, and not seeing in the first place have been proposed (Mack & Rock, 1998; Moore & Egeth, 1997). While it is not possible to settle this issue with definite data, it is my intuition that people who claimed not having seen the ProM-cue picture, failed to notice it as something special and hence were unable to recall its appearance at the end of the experiment. Would distractor items in the ongoing task also interfere with ProM if the cues were not pictures or other perceptually defined items (i.e., a certain color, texture, etc.)? If for example the ProM cue were a clock time, would there be similar interference? There are good reasons to suspect that this might happen. In experimental settings as well as everyday life, ProM time cues are represented by perceptual tokens, for example a digital clock on a computer screen or children rushing into the house on school day afternoons (see Rabbitt, 1996, for an extensive discussion of the environmental cues that embody temporal information). There is no reason in principle why the mechanism of perceptual selection should not be important in such time-cued ProM situations. 184 But there are ProM situations where the mechanism of perceptual selection might play a minor or no role. Suppose the cue is a category, such as the category of animals used in Meier and Graf's study (2000) or fruit used in Marsh and Hicks' study (1998). Would perceptual mechanisms be prominent in identifying this type of cue? Or suppose that subjects in the present research had been told, without displaying the rdeiant picture, that if they saw a butterfly in the course of the experiment, they were to stop ongoing activities? It is difficult to imagine that in these situations perceptual selection of the cue event would be a factor that is critical to enabling ProM. It can be assumed that cue discovery would rely massively on conceptual processes. Whether there would be also a perceptual component in such situations (after all, these cues are external events), remains to be established. Another type of limitation of this research regards the findings on ProM and creative thinking. These findings are based on correlations. Flexible and original thinking tended to co-vary with ProM abilities, but this kind of evidence does not perse say anything about the nature of the relationship. On logical grounds the interpretation in this work assumed a specific direction, that is, thinking abilities facilitate, or interfere with, ProM. On statistical grounds alternative possibilities must be considered. ProM variables (for example encoding or retrieval abilities) may affect tliinking or a third variable not identified in this research mediates the relationship between thinking and ProM. The first of these possibilities received some support in Experiment One: FEghly original subjects were much more likely than less original ones to produce semantic intrusions on RetM r tests. Only experimental manipulations can shed light on this issue, and some are suggested in the next section. One final, very important limitation of this research is that a critical component of ProM, that is, the planning phase was not examined. Several theorists have drawn attention to the primary role planning processes play in ProM. Dobbs and Reeves (1996) state that planning is one of the component processes of a variety of tasks referred to as prospective remembering (recall that these 185 authors do not consider ProM a form of memory). Rabbitt (1996) urges to study ProM in the context of plan structuring and plan coordination, and in fact asks "why are studies of 'prospective memory' planless?". Shallice and Burgess (1991) advance compelling theoretical arguments for investigating planning, with their notion that a marker flagging future events is created when an intention is formed. Empirical evidence also suggests that planning may determine whether ProM is successful. For example, Einstein et al. (1997) found that young people's ProM performance was affected by increased attentional demands when these demands were introduced in the encoding stage of the ProM instructions. Jacova, Woodward, Schafer and Graf (2002) found that cognitive interventions to improve prospective memory in patients with schizophrenia were only effective when they supported both plarming and retrieval abilities. These findings suggests that planning a future action carries its own resource demands, and that some variables related to the structure and organization of plans are probably a primary factor in ProM retrieval. Suggestions for Future Research The research on ProM and attention suggests several avenues for future research on ProM. The findings implicate inhibitory processes in ProM but do not provide definite evidence for their involvement. One way to firmly establish their role in ProM would be to investigate their contribution to age-related declines in ProM. Uttl et al. (2001), in their investigation of ProM- and RetM declines and their underlying mechanisms in older adults, noted that existing operationalizations of processing resources did not account for as great a portion of age-related variance in ProM as would be theoretically expected. It may be that if resources are operationalized in terms of inhibitory processes, for example, in terms of negative priming measures, they might provide a more compelling explanation of age declines in ProM. The role of inhibitory processes might also be investigated in young people by designing ProM tasks that build on the Stroop interference effect (Stroop, 1935). As was observed earlier, one direct implication of this research is that the attributes of the non-cue elements in a ProM situation 186 impact cue discovery just as much as the cue attributes. It would follow that if non-cue elements are presented that are difficult to suppress because they involve prepotent processing, such as the printed words in the Stroop task, ProM would be much more difficult than if these elements were more neutral. To investigate whether a form of long-term preparedness for specific perceptual features might underlie ProM, one strategy might be to establish perceptual priming effects in ProM. If they do, then it would follow that some mechanism exists that could mediate a form of long-term non-conscious readiness to process certain perceptual features. There is already some evidence indicating the existence of semantic priming effects on ProM. Mantyla (1996) asked subjects to perform an action whenever they came across instances of certain categories, for example vehicles or body parts. Prior to the test phase, he gave subjects semantic fluency tasks for some but not all categories, for example, to name all vehicles they could tliink of. Subjects were more likely to remember the ProM task when the cue was an instance of a category that had been primed with the earlier fluency task than when the cue belonged to a category that had not been primed. This research might also contribute to the search for neuropsychological correlates of ProM. As noted in Chapter Two, frontopolar and prefrontal cortical structures have already been linked to ProM. If it is true that selective attention mechanisms mediate ProM, then regions that have recently been implicated in selection and filtering of information such as the thalamus, in particular its largest nucleus, the pulvinar (e.g., LaBerge & Buchsbaum, 1990) should receive particular attention in future investigations. One PET study of brain regions involved in ProM discussed in Chapter Two (Burgess et al., 2001) did in fact reveal increased activity in the thalamus when a condition in which the ProM cue was presented, was compared with one in which ProM instructions were given but no ProM cue was presented. In light of the findings emerging from this research, the increased thalamic activity observed in this study could reflect the activation of selection processes rather than, as was argued in Chapter Two, the initiation of a novel response. 187 In Experiment Two high potential target load in the letter decision task did not produce ProM performance decrements. In fact there was a nominal difference in favor of ProM performance with high target load. However, the slower reaction times obtained with this manipulation showed that this load was attentionally demanding. It is possible that ProM does not require the type of attention needed to mentally access a string of potential target letters and match them to mcoming stimuli. But another possibility was also discussed. Maybe the attention that subjects must mobilize to conduct an internal search carried in some way over to ProM. Possibly subjects became more watchful and cautious, possibly mamtaining the system in a retrieval mode may have facilitated ProM retrieval. This raises the very interesting question whether some attentional manipulations directed to ongoing activities might have effects on ProM that fall outside the dual-task model. That is, they might stimulate ways of utilizing attention that facilitate ProM. To make bold suggestion, there might be concurrent transfers of attention that might better be accounted for within a TAP than a dual-task framework. The surprising negative relationship between original thinking and ProM needs further investigation. As argued in Chapter Six, the question really is whether this is a principled relationship reflecting some incompatibility between thinking in unusual ways and ProM, or whether it might be specific to a certain type of ProM tasks. If the latter is true, then at least in some ProM situations, thinking originally, defying rote and predictable responses, may turn out to be beneficial for ProM. This might be the case for example if ProM cues are inserted into highly habitual sequences of action. Main Contribution of This Research This research has revealed that ProM requires attention for the perceptual discovery of the cue event. A plausible mechanism by which ProM cue discovery occurs is selection or figural emphasis. If the rememberer succeeds in suppressing ProM-unrelated items in the ongoing scene, the ProM cue is rendered prominent and is noticed. In this way, the uniquely prospective 188 requirement of becoming aware of aspects in the environment which signal the need to recollect an intention is satisfied. At a theoretical level this research prompts current models to understand ProM not only as a deep executive process but one extending outward toward perception and the senses. This research has also provided evidence that certain thinking abilities are systematically related to ProM. The ability to think flexibly predicts ProM success whereas the ability to think in unusual ways predicts ProM failure. It is known that thought content plays a role in ProM but this research draws attention to the style and structure of thinking. ProM seems to benefit from the ability to shift thoughts and be quick to abandon them when they cannot satisfy the job at hand. Can the findings on attention and thinking be linked in some way, that is, do they hold a common message for ProM? Maybe they show from a bottom-up and top-down perspective that the key cognitive challenge of ProM is to interrupt and reorganize all ongoing processing. Phrased simply, the system must hit all the brakes it has available to make possible the extraordinary feat of prospective remembering. 189 R E F E R E N C E S Abelson, R P. (1981). Psychological status of the script concept. A mericanPsychologist, 36, 715-729. Amabile, T. M. (1996). Creatiiity in context. Boulder, CO: Westview Press. Andreasen, N . J., & Powers, P. S. (1975). Creativity and psychosis: An examination of conceptual style. British J amid of Psychiatry, 125, 452-456. Baddeley, A (1990). Humanmemory Theory and practice. London, UK: Lawrence Erlbaum Baer, J. (1993). Creativity and divergent thinking A task-specific approach. Flillsdale, NJ: Lawrence Erlbaum Barchillon, J. (1961). Creativity and its inhibition in child prodigies. In Personality Dimensions cf Creatiiity. New York: Lincoln Institute of Psychotherapy. Bartlett, F. G (1932). Remembering. Cambridge, UK: Cambridge University Press. Biebach, H., Gordjin, M., Krebs, J.R (1989). Time-and-place learning by garden warblers. Syhia bonn A nimd Behavior, 37, 353-360. Bisiacchi, P. (1996). The neuropsychological approach in the study of prospective memory. In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), ProspectiiE Memory Theory and Applications . Mahwah, NJ: Lawrence Erlbaum. Blair, J. R, & Spreen, O. (1989). Predicting premorbid IQ: A revision of the National Adult Reading Test. Clinical Neuropsychologist, 3(2), 129-136. Brandimonte, M. A , Passolunghi, M C . (1994). The effect of cue familiarity, cue distinctiveness, and retention interval on prospective remembering. Quartedy J amid cf Experimented Psychdogy, 47, 565-587. Brandimonte, M., Einstein, G. O , & McDaniel, M. A (Eds.). (1996). ProspectiiE Memory Theory and applications. Mahwah, NJ: Erlbaum. Burgess, P. W., & Shallice, T. (1997). The relationship between prospective and retrospective memory: Neuropsychological evidence. In M. A Conway (Ed.), Cognitiie models of memory (pp. 247-272). Cambridge, M A The MIT Press. Burgess, P. W., Veitch, E., de Lacy Costello, A , & Shallice, T. (2000). The cognitive and neuroanatomical correlates of multitasking. Neurcpsydodogia, 38, 848-863. Burgess, P. W., Quayle, A , & Frith, C. D. (2001). Brain regions involved in prospective memory as determined by positron emission tomography. Neimjpsycbdogia, 29, 545-555. Campbell, D. (1960). Blind variation and selective retention in creative thought as in other knowledge processes. PsychologicalReview, 67, 380-400. Carlsson, I., Wendt, P. E., &Risberg, J. (2000). On the neurobiology of creativity. Differences in frontal activity between high and low creative subjects. Neunpsydodogia, 38, 873-885. 190 Cattell, R B., & Butcher, H. J. (1969). 7fe/mfoiwzc/^zez07^^^C7mft'af)'. Indianapolis, IN: Bobbs-Merrill. Cockburn, J. (1995). Task interruption in prospective memory: A frontal lobe function? Cortex, 31, 87-97. Cockburn, J., Keene, J., &Hope, T. (2000). Are prospective memory measures of the RBMTmore vulnerable to cognitive impairment than retrospective memory measures? Paper presented at the First International Conference m Prospective Memory, Hatfield, UK. Cook, R G., Brown, M.F., Riley, D.A. (1985). Flexible memory processing by rats: Use of prospective and retrospective information in the radial maze. Journal cf Experimental Psychology: A nimdBehavxrrProcesses, HQ), 453-469. Craik, F. I. M., & Lockhart, R S. (1972). Levels of processing: A framework for memory research. Journd of Verbal Learnir^ and Verbal Behavior, 11, 671-684. Craik, F. I. M. (1986). A functional account of age differences in memory. In F. Klix & FL . Hagendorf (Eds.), Human rmnory and cognitive capabilities: Mechanisms and performances (pp. 409-422). New York, NY: Elsevier. Qikszentmihalyi, M. (1996). Cneativity Flaw, and the psychology ofdiscoveryand invention. New York, NY: Harper Collins. Dobbs, A. R, & Reeves, M. B. (1996). Prospective memory. More than memory. In M. Brandimonte, G. O. Einstein, & M. A. McDaniel (Eds.), Prospective memory: Theory and Applications (pp. 199-225). Mahwah, NJ: Erlbaum. Einstein, G. O., McDaniel, M A (1990). Normal aging and prospective memory. Journal of Experimental Psychology: Learning Memory, and Cognition, 16, 717-726. Einstein, G. O., & McDaniel, M. A (1996). Retrieval processes in prospective memory Theoretical approaches and some new empirical findings. In M. Brandimonte, G. O. Einstein, & M. A. McDaniel (Eds.), Prospective Memory: Theory and Applications (pp. 115-141). Mahwah, NJ: Erlbaum. Einstein, G. O., McDaniel, M. A , Smith, R E . , & Shaw, P. (1997). Aging and prospective memory The influence of increased task demands at encoding and retrieval. Psychology and Aging 12(3), 479-485. Einstein, G. O., McDaniel, M. A , Smith, R E., & Shaw, P. (1998). Habitual prospective memory and aging: Remembering intentions and forgetting actions. Psychological Science, 9(4), 284-288. Ellis, J. (1996). Prospective memory or the realization of delayed intentions: A conceptual framework. In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Prospective memory-Theory and Applications (pp. 1-22). Mahwah, NJ: Erlbaum. Ellis, J., & Kvavilashvili, L. (2000). Prospective memory in 2000: Past, present, and future directions. Applied Cognitive Psychology, 14(7), 1-9. Eysenck, H. J. (1993). Creativity and personality: Suggestions for a theory. Hillsdale, NJ: Lawrence Erlbaum. 191 Gallistel, G R. (1990). Theorganizatimcflmrning. Cambridge, MA: The MIT Press. Ghiselin, B. (1963). Automaticity, intention, and autonomy in the novelist's production. Daedalus, 92(2), 297-311. Gill, F. B. (1988). Trapline foraging by hermit hunimingbirds. Competition for an undefended renewable resource. Ecology, 69, 1933-1942. Glisky, E. L. (1996). Prospective memory and the frontal lobes. In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Pmspective Memory: Theory and Applications. Mahwah, NJ: Lawrence Erlbaum. Goldberg, L. R (1999). The Curious Experiences Survey: A revised version of the Dissociative Experiences Scale: Factor structure, reliability, and relations to demographic and personality variables. Psychcic^ad Assessment, 11(2), 134-145. Goldstein, K. (1944). The mental changes due to frontal lobe damage. Journal cf Psychology, 17,187-208. Graf, P., & Schacter, D. L. (1985). Implicit and explicit memory for new associations in normal subjects and amnesic patients. Journal cf Experimental Psychology Learning Memory, and Cognition, 11, 501-518. Graf, P., & Uttl, B. (1995). Component processes of memory: Changes across the adult lifespan. Swiss Journal cf Psychology, 54(2), 113-130. Graf, P., & Uttl, B. (2001). Prospective memory: A new focus for research. Consciousness and Cognition, 10(4), 437-450. Green, M. J., & Williams, L. M. (1999). Schizotypy and creativity as effects of reduced cognitive inhibition. Personality and Individual Differences, 27, 263-276. Grice, G. R, Nullmeyer, R & Schnizlein, J. M. (1979). Variable criterion analysis of brightness effects in simple reaction time. Journal cf Experimental Psychology Human Performance and Perception, 5, 303-314. Gruber, F i E., & Davis, S. N . (1988). Inching our way up Mount Olympus: the evolving systems. In R J. Sternberg (Ed.), The Nature cf Creativity Contemporary Psychological Perspectives . New York: Cambridge University Press. Guilford, J. P. (1959). Three faces of intellect. The A merican Psychologist, 469-479. Guilford, J. P. (1968). Intelligence, Creativity, and Their Educational Implications. San Diego, C A Knapp. Guilford, J. P. (1975). Creativity: A quarter century of progress. In I. A Taylor & J. W. Getzels (Eds.), Perspectives in creativity. Chicago, IL: Aldine Publishing Company. Flasher, L., & Zacks, R T. (1979). Automatic and effortful processes in memory. Journal cf Experimental Psychology: General, 108(3), 356-388. 192 Honig, W. K., & Wasserrnan, E.A. (1981). Performance of pigeons on delayed simple and conditional discriminations under equivalent training procedures. LearnmgandMotivation, 12, 149-170. Honig, W. K., &Dodd, P.W.D. (1986). Anticipation and intention in working memory. In D. F. Kendrick, Rilling, M.E., Denny, M.R. (Ed.), Theories cf animal memory. Hillsdale, NJ: Lawrence Erlbaum Associates. Huppert, F. A , Johnson, T., & Nickson, J. (2000). High prevalence of prospective memory impairment in the elderly and in early-state dementia: Findings from a population-based study. Applied Cognitive Psychology, 14(7), 63-81. Jacova, C , Meier, B., & Graf, P. (2001, July). Prospective memory in later adulthood Paper presented at the 17th Congress cf the Internaticml Association of Gerontology, Vancouver, Canada. Jacova, G, Woodward, T. S., Schafer, M., & Graf, P. (2002). Component skills in prospective memory Evidence from rehabilitation with schizophrenic patients. Poster presented at the 12th A mrual Meeting of the Canadian Societyfor Brain, Behaviour, and Cognitive Science, Vancouver. Johnston, J., McCann, R S., & Remington, R. W. (1995). Chronometric evidence for two types of attention. Psychological Science, 6(365-3 69). Kahneman, D. (1973). A ttention and effort. Englewood Cliffs, NJ: Prentice Hall. Kasof, J. (1997). Qeativityand breadth of attention. Cneativity Research Journal, 10(4), 303-315. Kausler, D. H. (1994). Learrimgandrrmxjry in normal aging. New York: Academic Press. Kerns, K. (2000). The CyberCruiser: An investigation of development of prospective memory in children. Journal of the International Neuropsychological Society, 6, 62-70. Koriat, A , Ben-Zur, H., &Nussbaum, A (1990). Encoding information for future actions: Memory for to-be-performed tasks versus memory for to-be-recalled tasks. Memory and Cognition, 18, 568-578. Kuhn, T. S. (1963). The essential tension: Tradition and motivation in scientific research. In F. Barron & C. W. Taylor (Eds.), Scientific Cneativity: Its ReccgnitimandDevdcpment New York: Wiley. Kvavilashvili, L. (1987). Remembering intention as a distinct form of memory. British Journal of Psychology, 78, 507-518. Kvavilashvili, L., & Ellis, J. (1996). Varieties of intentions: Some distinctions and classifications. In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Prospective rnemory: Theory and Applications (pp. 23-51). Mahwah, NJ: Erlbaum Kvavilashvili, L., Kombrot, D., Mash, V., Cockburn, J., & Milne, A (2000). Remembering event, time-, and activity-based tasks in young, young-old and old-old. Paper presented at the First International Conference on Prospective Memory, University of Heilfordshire, Hatfield, UK. LaBerge, D., StBuchsbaum, M. (1990). Positron emission tomographic measurements of pulvinar activity during an attention task. Journal qfNeurvscienoe, 10, 613-619. 193 Lavie, N. , & Tsal, Y. (1994). Perceptual load as a major deternrinant of the locus of selection in visual attention. PerceptimandPsydocpbysics, 56(2), 183-197. Mack, A &Rock, I. (1998). Imttentional blindness. Cambridge, MA: MIT Press. Maddi, S.R. (1975). The strenuousness of the creative life. In IA. Taylor &J.W. Getzels (Eds), Perspectives in Creativity. Chicago, IL: Aldine PubHshing Company. Maier, N . R. E. (1929). Delayed reaction and memory in rats. Comparative Psychology Monographs, 36, 538-549. Mantyla, T. (1994). Remembering to remember: Adult age differences in prospective memory. Journal cf Gerontology: Psychological Sciences, 49, 276-282. Marsh, R L., & Hicks, J. L. (1998). Event-based prospective memory and executive control of working memory. Journal cfExpenmentalPsychology Learning Memory, and Cognition, 24(2), 336-349. Martin, M., Kliegel, M., McDaniel, M.A, & Einstein, G.O. (2000). The neuropsychology of prospective memory: The role of central executive functions. Paper presented at the First International Conference m Prospective Memory, Hatfield, UK. Maylor, E. (1996). Does prospective memory decline with age? In M. A Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Pmspextiw mernory Theory and Applications (pp. 173-191). Mahwah, NJ: Erlbaum. Maylor, E. A (2000). Prospective memory across the lifespan Paper presented at doe First International Conference onPmspective Memory, Hatfield, UK. Maylor, E. A , Darby, R J., Smith, G., Delia Sala, S., & Logie, K H. (in press). Prospective memory across the lifespan. In P. Graf&N. Ohta (Eds.), Lifespan memory deidcpment. Cambridge, M A The MIT Press. McCann, R S., & Johnston, J. G (1992). Locus of the single-channel bottleneck in dual-task interference. Journal cfExperimental Psychology, Human Perception and Performance, 18, 471-484. McCrae, R R, Arenberg, D., & Costa, P. T. (1987). Declines in divergent thinking with age: Cross-sectional, longitudinal, and cross-sequential analyses. Psychology and Aging, 2(2), 130-137. McDaniel, M. A , Robinson-Riegler, B., & Einstein, G. O. (1998). Prospective remembering: Perceptually driven or conceptually driven processes? Memory and Cognition, 26(1), 121-134. McDaniel, M. A , Glisky, E. L., Rubin, S. R, Guynn, M. J., & Routhieaux, B. G (1999). Prospective memory: A neuropsychological study. Neuropsychology, 13(1), 103-110. McDaniel, M. A , & Einstein, G. O. (2000). Strategic and automatic processes in prospective memory retrieval: A multiprocess framework. Applied Cognitive Psychology, 14(7), 127-144. McDowd, J. M., & Shaw, R J. (2000). Attention and aging: A functional perspective. In F. I. M. Craik & T. Salthouse (Eds.), The handbook cf aging and cognition. Mahwah, NJ: Lawrence Erlbaum. 194 Mednick, S. A. (1962). The associative basis of the creative process. PsychologicalReziew, 69(3), 220-232. Mehrabian, A. (1977). A questionnaire measure of individual differences in stimulus screening and associated differences in arousability. Enuronmentd Psychdcgy andNo^ .7(2), 89-103. Meier, B., & Graf, P. (2000). Transfer appropriate processing for prospective memory tests. Applied CognitizePsychology, 14, 511-527. Mendelsohn, G. A (1976). Asociative and attentional processes in creative performance. Journal cf Personality, 44, 341-369. Miller, G. A (1956). The magical number seven plus or minus two: some limits on our capacity for processing information. Psychological Renew, 63, 81-97. Moore, C. M. &Egeth, FL (1997). Perception without attention: Evidence from grouping under conditions of inattention. Journal of Expenrrental Psychdcgy Human Perxeptim and Perforrmna^ 11, 56-60. Moscovitch, M. (1982). A neuropsychological approach to memory and perception in normal and pathological aging. InF. I. M. Craik&S. Trehub (Eds.), Adiances inthestudyof communication and affect- Vol. 8. Aging and Cognitize Processes. New York, NY: Plenum Press. Moscovitch, M. (1994). Memory and working with memory Evaluation of a component process model and comparisons with other models. In D. L. Schacter & E. Tulving (Eds.), Memory Systems 1994 . Cambridge, M A The MIT Press. Navon, D. (1984). Resources: A theoretical soupstone. PsychologicalReziew, 91, 216-234. Neisser, U. (1967). Cognitize Psychology. New York- Appleton-Century-Crofts. Neumann, O. (1996). Theories of attention. In O. Neumann & A F. Sanders (Eds.), Handbook of perceptionandaction. Vol 3 (pp. 389-446). New York, NY: Academic Press. Newell, A , & Simon, FL A (1972). Human problem sdzing. New York Prentice Flail. Norman, D. A , & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior. In R J. Davidson, G. E. Schwartz, & D . Shapiro (Eds.), Consciousness and Self-Regulation A dzances in Research and Theory, Volume 4 . New York and London: Plenum Press. Okuda, J., Toshikatsu, F., Atsushi, Y., Ryuta, K., Takashi, T., Reiko, F., Suzuki, K., Ito, M., & Fukuda, FL (1998). Participation of the prefrontal cortices in prospective memory Evidence from a PET study in humans. Neuroscience Letters, 253, 127-130. Otani, PL, Landau, J. D., Libleman, T. M., St. Louis, J. P., Kazen, J. K., & Throne, G. W. (1997). Prospective memory and divided attention. Memory, 5, 343-360. Park, D. G , Hertzog, G, Kidder, D. P., Morrell, R W., &Mayhorn, G B. (1997). Effect of age on event-based and time-based prospective memory. Psychology and Aging 12, 314-327. 195 Pashler, PL, & Johnston, J. (1989). Interference between temporally overlapping tasks: Chronometric evidence for central postponement with or without response grouping. Quarterly Journal cf Experimental Psychology, 41A, 19-45. Perrig, W. J., Hofer, D., Kling, V., Meier, B., Perrig, P., Ruch, M., & Serafin, D. (1994). Computeruntmtutzter Gedfcht^ Manual. Basel: Forschungberichte des Instituts fiir Psychologic Plude, D. J., Enns, J. T., & Brodeur, D. (1994). The development of selective attention: A life-span overview. Acta Psychologka, 86, 227-272. Prentky, R A (1980). Creativity andpsychopathology A neurocognitvwperspective. New York, NY: Praeger Publishers. Rabbitt, P. M. (1965). An age decrement in the ability to ignore irrelevant information. Journal cf Gerontolcgy, 20, 233-238. Rabbitt, P. (1996). Commentary: Why are studies of "prospective memory" planless? In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Prospective Memory Theory and Applications . Mahwah, NJ: Lawrence Erlbaum. Reason, J. T. (1984). Lapses of attention in everyday life. In R Parasuraman & D . R Davies (Eds.), Varieties cf attention. Orlando, FL: Academic Press. Rendell, P. G., & Craik, F. I. M. (2000). Virtual week and actual week. Applied Cognitive Psychology, 14(7), 43-62. Roberts, W. A (1998). Principles ofammd cognition, Boston, M A McGraw-Hill. Roediger, H. L., &McDermott, K. B. (1993). Creating false memories: Remembering words not presented in lists. Journal cf Experimental Psychology Learning Memory, and Cognition, 21, 803-814. Runco, M. A (1991). Divergent thinking. Norwood, New Jersey Ablex Publishing Corporation. Salthouse, T. A (1991). Theoretical perspectives on cognitive aging. FEllsdale, NJ: Lawrence Erlbaum. Salthouse, T. A (1996). The processing-speed theory of adult age differences in cognition. Psychological Bulletin, 103(3), 403-428. Schacter, D. L. (1996). Searchingfor memory. New York: Basic Books. Schonfield, D. (1982). Attention switching in higher mental processes. In F. I. M. Craik & S. Trehub (Eds.), Advances inthestudycfcomnm^timardafkt: VoL 8. Aging and cognitive processes (pp. 212-223). New York, NY: Plenum Press. Shallice, T. (1988). FmmNeuropsychology to Mental Structure. Cambridge, UK: Cambridge University Press. Shallice, T., Burgess, P.W. (1991). Deficits in strategy application following frontal lobe damage in man. Brain, 114, 727-741. 196 Shallice, T. (1996). Commentary: The neuropsychology of prospective memory. In M. Brandimonte, G. O. Einstein, & M . A McDaniel (Eds.), Prospective Memory Theory and Applications . Mahwah, NJ: Lawrence Erlbaum. Sherrington, C. S. (1906). The integrative action cf the nervous system New Haven, CT: Yale University Press. Sibly, R. M., &McCleery, KIT. (1983). The distribution between feeding sites of herring gulls breeding at Walney Island , U K . Journal of A nimd Ecology, 52, 51-68. Smith, S. M., Ward, T. B., &Finke, R, .A, (Eds.). (1995). The creative cognition approach. Cambridge, MA: The MIT Press. Stavridou, A , &Fumharn, A (1996). The relationship between psychoticism, trait-creativity and the attentional mechanism of cognitive inhibition. Personality and Individual Differences, 21(1), 143-153. Sternberg, S. (1969). Memory-scarming: Mental processes revealed by reaction time experiments. American Scientist, 57, 421-457. Stroop, J. R (1936). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643-662. Torrance, E. P. (1966). Torrance Tests of Creative Thinking. Princeton, NJ: Personnel Press. Torrance, E. P. (1988). The nature of creativity as manifest in its testing. In R J. Sternberg (Ed.), The nature cf creativity. Cambridge, UK: Cambridge University Press. Tulving, E. (1983). Elements ofepisodk memory. New York: Oxford University Press. Uttl, B., & Graf, P. (2000, July). Event-cued ProM proper differences in old age. Paper presented at the First International Conference on Prospective Memory, Hatfield, UK. Uttl, B., Graf, P., Miller, J., & Tuokko, H. (2001). Pro- and retrospective memory in late adulthood. Consciousness and Cognition, 10(4). Vurpillot, E. (1968). The development of scanning strategies and their relation to visual differentiation. Journal of Experimental Child Psychology, 6, 632-650. Wallach, M. A , &Kogan, N . (1965). Modes cfhinking in young children. New York: Holt, Rinehart, and Winston, Inc. Wallas, G. (1926). The art cf thought. New York Hartcourt, Brace. Welford, A T. (1958). A ging and human skill. London: Oxford University Press. West, R L. (1988). Prospective memory and aging. In R N . Sykes & P. E. Morris (Ed.), Practical Aspects of Memory Current Research andlssues. Volume 2 . Chichester, UK: Wiley West, R, Herndo, R W., & Ross-Munroe, K. (2000). Event-related neural activity associated with prospective remembering. Applied Cognitive Psyhology, 14(7), 115-126. 197 Wickens, C D. (1984). Processing resources in "attention. In R. Parasuraman & R Davies (Eds.), Varieties cf attention (pp. 63-102). New York, NY: Academic Press. Wilkie, D. M., &Willson, RJ. (1990). Time-place learrring bypigeons, Columbia livia. Journal cfthe Experimental Analysis cfBehazior, 57,145-158. Wilkie, D. M., Carr, J .AR, Siegenthaler, A , Lenger, B., Liu, M., & Kwok, M. (1996). Field observations of time-place behavior in scavenging birds. Behavioural Processes, 38, 77-88. Willons, A J., & Baddeley, A (1978). Remembering to recall in everyday life: An approach to absent-mindedness. In M. M. Gruneberg, P. E. Morris, & R N . Sykes (Eds.), Practical aspects cf memory. LondonJJK: Academic Press. Williams, L. (1982). Cognitive load and the functional field of view. HumtnFactors, 24, 683-692. Wilson, B., Cockburn, J., Baddeley, A , & Morns, R (1989). The development and validation of a test battery for detecting and monitoring everyday memory problems. Journal cf Clinical and Experimental Neuropsychology, 11(6), 855-870. Winograd, E. (1988). Some observations on prospective remembering. In M. M. Gruneberg, E. P, Morris, & R N . Sykes (Eds.), Practical aspects cf memory Current research and issues. Vol I: Memory in everyday life. Toronto: Wiley. Woodman, G. F., Vogel, E. K., &Luck, S. J. (2001). Visual search remains efficient when visual working memory is full. Psychological Science, 12(3), 219-234.. Zangwill, O. L. (1966). Psychological deficits associated with frontal lobe lesions. International Journal cfNeumlcgy, 5, 395-402. 198 APPENDICES Appendix A: The Story ProM Task 199 (Note that cue words were not underlined in the presented version) -1 -All night the three fugitives' course led through intricate gorges and over irregular and rock-strewn paths. More than once they lost their way, but Jefferson Hope's intimate knowledge of the mountain enabled them to regain the track once more. When morning broke, a scene of marvelous though savage beauty lay before them. In every direction the great snow-capped peaks hemmed them in. So steep were the rocky banks on either side of them that the larch and the pine seemed to be suspended over their heads, and to need only a gust of wind to come hurtling down upon them. - 2 -The fear was not entirely an illusion, for the barren valley was thickly strewn with trees and boulders, which had fallen in a similar manner. Even as they passed, a great rock came thundering down with a hoarse rattle, which woke the echoes in the silent gorges and startled the weary horses into a gallop. As the sun rose above the eastern horizon, the caps of the great mountains lit up one after the other, like lamps at a festival, until they were all ruddy and glowing. The magnificent spectacle cheered the hearts of the three fugitives and gave them fresh energy. - 3 -At a wild torrent, which swept out of a ravine, they called a halt and partook of a hasty breakfast. Lucy and her father would have rested longer, but Jefferson Hope was inexorable. "They will be upon our track by this time," he said. "Everything depends upon our speed. Once safe in Carson, we may rest for the remainder of our lives." So they packed up their gear, and struggled on through the gorges. At nighttime they calculated that they were more than thirty miles from their enemies. They enjoyed a few hours' sleep at the base of a beetling crag. -4 -On the second day of their flight their scanty store of provisions began to run out. Jefferson chose a sheltered nook, and piled together a few dried branches to make a blazing fire for his companions. He bid them adieu and set out with his rifle. He walked for a couple of miles through one ravine after another. From the marks on the trees, he judged that there were bears in the vicinity. Walking on, he found himself in a valley divided and sub-divided into many gorges, which were so like each other that it was impossible to distinguish them. - 5 -After hours and hours of fruitless search, he was thinking of turning back in despair. He followed one gorge for a mile. Convinced he had taken the wrong turn, he tried another and another, until he came to a mountain torrent, which he was sure that he had seen before. Even then it was no easy matter to keep to the right track, for the high cliffs on either side prevented him from getting his bearings. Weary from his exertions, he stumbled along, keeping up his heart by the reflection that every step brought him nearer to Lucy. 200 -6-A l l o f a s u d d e n , h e s a w a s ight, w h i c h s en t a thrill of p l e a s u r e t h r ough h i s heart. O n the e d g e o f a jutt ing p i n n a c l e , t h r e e o r f ou r h u n d r e d f ee t a b o v e h i m , t h e r e s t o o d a c r e a t u r e s o m e w h a t r e s e m b l i n g a s h e e p in a p p e a r a n c e , but a r m e d wi th a pa i r of g i gan t i c horns . It w a s p r o b a b l y a c t i n g a s a g u a r d i a n o v e r a f lock , w h i c h w e r e i nv i s i b le to t he hunter; but f o r tuna te l y it h a d not p e r c e i v e d h im. L y i n g o n h i s f a c e , he r e s t ed h i s rifle u p o n a rock, a n d t o o k a l ong a n d s t e a d y a i m . H e d r e w t he t r igger a n d h i s t a rge t c a m e c r a s h i n g d o w n into the va l l ey . - 7 -Night w a s c o m i n g o n rapid ly, a n d it w a s d a r k be f o r e h e c a m e to t h e m o u t h of the ve r y g o r g e in w h i c h h e h a d left t he o ld m a n a n d Lucy . T h e y must , h e r e f l e c ted , be awa i t i ng h im anx i ou s l y . In t he g l a d n e s s of h i s hear t he put h i s h a n d s to h i s m o u t h a n d s h o u t e d a s i g na l into t he d a r k n e s s . H e p a u s e d a n d l i s t ened for a n a n s w e r . N o n e c a m e s a v e his o w n , e c h o i n g cry, w h i c h c l a t t e red up the d rea ry , s i l ent r a v i ne s , a n d w a s b o r n e b a c k to h i s e a r s in c o u n t l e s s repet i t ions . A v a g u e , n a m e l e s s d r e a d c a m e o v e r h im, a n d he hur r ied o n w a r d f rant ica l l y . - 8 -W h e n h e t u r ned t he co rne r , h e c a m e full in s ight of the s po t w h e r e t he f i re h a d b e e n lit. T h e s a m e d e a d s i l e n c e still r e i g ned al l r ound . W i t h h i s f e a r s a l l c h a n g e d to c onv i c t i on s h e hur r ied o n . T h e r e w a s no l iv ing c r e a t u r e n e a r t he r e m a i n s o f t he f i re: t he o ld m a n a n d L u c y h a d g o n e . T h e g r o u n d w a s al l s t a m p e d d o w n by t he f ee t o f m e n a n d dog s , s h o w i n g that a l a r ge par ty h a d t a k e n the two fug i t i ves . It w a s o n l y t oo c l e a r that a s u d d e n a n d ter r ib le d i s a s t e r h a d o c c u r r e d du r i n g h i s a b s e n c e - a d i s a s t e r w h i c h h a d e m b r a c e d t h e m a l l . 201 Appendix B: The Prospective Memory Questionnaire 1. At some point I showed you a picture against black background and gave you instructions? What was the picture and what were the instructions? 2. Did you see this picture again in the course of the experiment? Yes No (If no go to 5) 3. When you saw the picture again, did you stop what you were doing at the time as instructed? Yes No 4. If not, why not? 5. How many times do you think the picture appeared? 6. During the experiment, you reminded yourself of stopping when the picture appeared: (a) not at all (b) rarely (c) every now and then (d) frequently (e) continually 7. Did you have a particular strategy? 8. Did you expect the picture to appear during a specific task in the experiment? If yes, which task? Appendix C: Common, Unusual, and Unique Responses to the Item "Brick" Common Responses Unusual Responses Unique Responses To build a well As an anchor As giant chopsticks As a doorstop To sharpen knives To jack up a car As a hammer To roof a house To bang together To line flowerbeds As a computer stand To cook until it is soft As a weapon As a footrest To fix a wall As a paperweight To displace water As Lego pieces For bookshelves To confine an area As a hat To sit on For martial arts To make concrete To break a window To trace a rectangle To practice ping pong To build a fireplace To make yourself taller As a pot holder To draw or write on it To hold down the gas pedal As a ruler As an exercise tool To make red color To rhyme with it To pave a driveway As insulation As a talking stick To weigh down a tarpaulin To fill a hole As a wrench To stop a car on a hill As a gift To make a universe To make pebbles As a diving toy For a pendulum As a heating device To mark a spot As a baseball bat To make a sculpture To build a dog house As domino pieces To balance something To sell it for profit As a pillow 203 Appendix D: Flexibility Categories BRICK 1) Arts and Crafts 2) Break Things 3) Build 4) Close/Stop Holes 5) Container 6) Decoration 7) Furniture/Household Items 8) Heat Source 9) Holding/Preventing Movement 10) Increase Height 11) Make Noise 12) Marker/Barrier 13) Recreation 14) Recycle 15) Tool 16) Weight 17) Weight for Exercise 18) Weapon 19) Other PENCIL 1) Accessories 2) Arts and Crafts 3) Build 4) Burn 5) Collection 6) Decoration 7) Eraser 8) Fiddle/Chew 9) Hold/Connect 10) Make Noise 11) Marker/Barrier 12) Poke/Probe/Extension 13) Recreation/Sports 14) Tool 15) Weapon 16) Write/Draw 17) Other SHOE 1) Burn BUTTON 1) Arts and Crafts 2) Build 3) Collection 4) Counting/Math 5) Decorating Clothing/Accessories 6) Decoration/Ornament 7) Fasten Clothing or Accessories 8) Furniture/Household Accessories 9) Money/Poker Chips 10) On/Off Button 11) Recreation 12) Tool 13) Weapon/Harm People 14) Weight 15) Other ROPE 1) Arts and Crafts 2) Burn 3) Capturing Person/Animal 4) Climb 5) Clothing and Accessories 6) Decoration 7) Hanging Something 8) Harming Person/Animal 9) Hold Things Open or Closed 10) Leash 11) Lift/Pull/Drag 12) Marker 13) Measure/Counting 14) Recreation 15) Safety/Rescue 16) Tie things together/Tie Knots 17) Tie things up/Secure things 18) Tool TIRE 1) Absorb impact/buffer 2) Arts and Crafts 3) Build with/stack 4) Burn 5) Commercial use 2) Collection 3) Commercial Use 4) Container 5) Creating something physical 6) Decoration 7) Extra weight/height 8) Fashion statement 9) Gift 10) Laces for use 11) Make noise 12) Marker 13) Recreation 14) Smell 15) Tool 16) Weapon 17) Wear 18) Weight 19) Other TOOTHBRUSH 1) Arts and Crafts 2) Build/Create 3) Clean/Scrub 4) Comb/Brush 5) Decoration/Advertising 6) Geometry/Measuring 7) Gift 8) Make Noise 9) Probe/Poke 10) Recreation 11) Scratching/Tickling 12) Stick 13) Tool 14) Weapon 15) Other TOWEL 1) Arts and Crafts/Creating 2) Burn 3) Carry things inside 4) Clean/Scrub 5) Clothing/Accessories 6) Cover Below 7) Cover On Top 8) Decoration 9) Drying/Soaking Up 10) Furniture/Household Accessories 11) Health Purposes 12) Protection (wrap with) 13) Recreation 14) Rope/String 15) Signal/Flag 16) Shelter 17) Weapon 18) Other 205 Appendix E: Common and Unusual Responses to the Item "Rope" Common Responses Unusual Responses To swing from For a boxing ring As a belt As a bracelet To hang a person To build a bridge As a clothes line As dental floss As material for a fire To fly a kite To hold up a tent As a harness As an exercise tool To make a basket As a lasso To measure things To move a car To hoist things To make a trap For bunjee jumping To rescue a person To imitate a snake To tie a boat As a tight rope To tie a prisoner To catapult things To play tug-of-war For a fishing rod As a whip As a reminding device To wrap things For a strength test As a hair band To tickle somebody For rock climbing To trip someone 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items