Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The cognitive processes involved in prospective memory task execution : response switching from an ongoing… Lark, Michelle L. Crease 2019

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2019_november_lark_michelle.pdf [ 1.13MB ]
Metadata
JSON: 24-1.0381005.json
JSON-LD: 24-1.0381005-ld.json
RDF/XML (Pretty): 24-1.0381005-rdf.xml
RDF/JSON: 24-1.0381005-rdf.json
Turtle: 24-1.0381005-turtle.txt
N-Triples: 24-1.0381005-rdf-ntriples.txt
Original Record: 24-1.0381005-source.json
Full Text
24-1.0381005-fulltext.txt
Citation
24-1.0381005.ris

Full Text

     THE COGNITIVE PROCESSES INVOLVED IN PROSPECTIVE MEMORY TASK EXECUTION: RESPONSE SWITCHING FROM AN ONGOING ACTIVITY TO A PLANNED TASK by  Michelle L. Crease Lark  B.A. Hons, The University of Winnipeg, 2008 M.A., The University of British Columbia, 2013  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Psychology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  September 2019  © Michelle L. Crease Lark, 2019 ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: The Cognitive Processes Involved in Prospective Memory Task Execution: Response Switching from an Ongoing Activity to a Planned Task  submitted by Michelle L. Crease Lark in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Psychology  Examining Committee: Prof. Peter Graf Supervisor  Prof Alan Kingstone Supervisory Committee Member  Prof Christiane Hoppmann Supervisory Committee Member Prof Todd Woodward University Examiner Prof Rebecca Todd University Examiner   iii  Abstract Prospective memory is the cognitive function used for carrying out planned tasks upon the occurrence of an appropriate cue. Prospective memory tasks are uniquely challenging because they typically take place while we are otherwise engaged in another ongoing activity. In order to carry out a planned task, three retrieval stages are required: cue noticing, meaning the cue is processed by the sensory system and perceived as a distinct entity; recognition of cue-plan relevance, meaning that the representation of the plan is activated in connection with the cue; and response switching, meaning ongoing activity responding is interrupted and the planned task response is executed. Previous prospective memory research has focused on manipulations and theoretical explanations most relevant to the first two stages. The research presented in this dissertation examined the cognitive processes involved in response switching. In Experiments 1 and 2, a prospective memory task was embedded within the encoding phase (Experiment 1) or retrieval phase (Experiment 2) of a recognition memory test. I examined the pattern of processing for ongoing activity words immediately preceding and following a prospective memory task response. Participants responded more slowly, and recognition test performance was lower, for words immediately following prospective memory task responses. Experiment 3 tested the possibility that the proactive effects observed in Experiments 1 and 2 were a result of processing a distinctive stimulus. For words immediately following distinctive stimuli, participants responded more slowly, but there was little influence on recognition test performance. In Experiments 4a and 4b, the degree of overlap in the processing required for the ongoing activity and the prospective memory task cue was manipulated. Participants showed larger proactive effects when there was a greater degree of overlap. The goal of Experiment 5 was to explore how subliminal primes preceding the prospective memory task cue would influence proactive effects due to response switching. The results showed that category primes produce larger proactive effects compared to repetition primes. Taken together, these findings support the iv  assumption that two mechanisms identified in the task-switching literature, task-set configuration and task-set inertia, are involved response switching in prospective memory task retrieval.    v  Lay Summary Prospective memory tasks, such as remembering to go grocery shopping on our way home from work, usually take place while we are occupied with another activity, such as driving (monitoring your speed, traffic, pedestrians, etc.). In order to carry out the plan to go grocery shopping, we need to switch away from the activity with which we are currently occupied and follow through with the planned activity. The research reported in this dissertation was conducted in order to determine how people are able to make the switch away from an ongoing activity to the task they have planned to carry out, and it examined the consequences of making this switch. Results suggest that it takes extra time and effort to switch back and forth between the activity we are doing and the planned task.     vi  Preface This dissertation is an original intellectual product of the author, Michelle Crease Lark. The experiments reported in Chapters 2- 5 were covered by UBC Ethics Certificate number H03-80566 and U of M Ethics Certificate number HS16703.  A version of Chapter 2 has been prepared for submission. Michelle Crease Lark was primarily responsible for the manuscript composition, with edits from P. Graf and R. K. Jamieson.  Michelle Crease Lark was the primary author and presenter for all conference presentations and posters cited in this document.    vii  Table of Contents Abstract ......................................................................................................................................................... iii Lay Summary ................................................................................................................................................ v Preface ......................................................................................................................................................... vi Table of Contents ......................................................................................................................................... vii List of Tables ................................................................................................................................................ ix List of Figures ................................................................................................................................................ x Acknowledgments ....................................................................................................................................... xiii Dedication .................................................................................................................................................... xiv Chapter 1: Exploring the Unique Characteristics of Prospective Memory Retrieval ....................................... 1 Prospective Memory Task Types ........................................................................................................ 2 Prospective Memory Retrieval: Task Analysis, Stages and Theoretical Implications. ......................... 4 Critical Differences between Prospective and Retrospective Memory Task Retrieval. ............. 5 Stages in Prospective Memory Task Retrieval ......................................................................... 7 Stage 1: Cue Noticing ............................................................................................................. 22 Stage 2: Recognition of Cue-Plan Relevance. ........................................................................ 26 Stage 3: Response Switching. ................................................................................................ 30 Overview of Dissertation ................................................................................................................... 32 Chapter 2: Consequences of Prospective Memory Task Response Switching ............................................ 35 Experiment 1 ..................................................................................................................................... 40 Method .................................................................................................................................... 40 Results .................................................................................................................................... 44 Discussion .............................................................................................................................. 49 Experiment 2 ..................................................................................................................................... 53 Method .................................................................................................................................... 57 Results .................................................................................................................................... 58 Discussion .............................................................................................................................. 62 General Discussion ........................................................................................................................... 67 Chapter 3:  Proactive Influences Due to Perceptually Salient Stimuli .......................................................... 72 Experiment 3 ..................................................................................................................................... 76 Method .................................................................................................................................... 76 Results .................................................................................................................................... 77 Discussion .............................................................................................................................. 82 Chapter 4: High Processing Overlap Increases the Proactive Effects Due to Response Switching ............ 87 Experiment 4a ................................................................................................................................... 92 Method .................................................................................................................................... 92 Results .................................................................................................................................... 94 Discussion ............................................................................................................................ 100 Experiment 4b ................................................................................................................................. 105 Method .................................................................................................................................. 105 Results .................................................................................................................................. 107 Discussion ............................................................................................................................ 112 General Discussion ....................................................................................................................... 1124 Chapter 5: The Influence of Subliminal Primes on the Proactive Effects of Response Switching .............. 117 Experiment 5 ................................................................................................................................... 121 Method .................................................................................................................................. 121 viii  Results .................................................................................................................................. 124 Discussion ............................................................................................................................ 130 Chapter 6: General Discussion .................................................................................................................. 118 Research Contributions .................................................................................................................. 137 Theoretical and Empirical Contributions ............................................................................... 137 Methodological Contributions ................................................................................................ 142 Implications ..................................................................................................................................... 144 Limitations ....................................................................................................................................... 146 Conclusions .................................................................................................................................... 149 References ................................................................................................................................................ 151   ix  List of Tables Table 4.1. ProM task performance, recognition test hit rate for all previously studied words, recognition test false alarm (FA) rate for all unstudied words, as a function of condition (high overlap vs low overlap). ...... 95 Table 4.2. ProM task performance, recognition test hit rate for studied words and recognition test false alarm (FA) rate for unstudied words as a function of cue condition (high overlap vs. low overlap). ........... 109   x  List of Figures Figure 1.1. Stages of prospective memory task execution. ........................................................................... 8 Figure 1.2. Depiction of Cohen et al.’s (2003) spatial displacement of the prospective memory cue .......... 24 Figure 1.3. Depiction of semantic distance of typical and atypical category members as prospective memory cues, inspired by similar models from Collins and Loftus, 1975. .................................................... 28 Figure 2.1 Mean (of participants’ median) response time (in ms) to encoding phase cue words (red bar), cue-surround words (dark blue bars), and non-surround words (light blue bar)........................................... 46 Figure 2.2. Mean recongition test performance (hits) for encoding phase cue words (red bar), encoding phase cue-surround words (dark blue bars), non-surround words (light blue bar) and test phase cue-surround words (yellow bars) ....................................................................................................................... 48 Figure 2.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words, encoding phase non-surround words and test phase cue-surround words. .................................................................................................................. 49 Figure 2.4. Mean recognition test performance for test phase cue words, test phase cue-surround words and non-surround words .............................................................................................................................. 61 Figure 2.5. Mean of participants’ median response time (in ms) required for correct recognition test decisions to ProM task cue words (red bar), test phase cue surround words (dark blue bars), and all non-surround words (light blue bar) .................................................................................................................... 62 Figure 3.1 Mean of participants’ median response time (in ms) to encoding phase von Restorff (VR) words (red bar), von Restorff surround words (dark blue bars), and non-surround words (light blue bar) .............. 79 Figure 3.2. Mean recognition test performance (hits) for von Restorff (VR) words (red bar), encoding phase von Restorff surround words (dark blue bars), non-surround words (light blue bar), and test phase von Restorff-surround words (yellow bars) ......................................................................................................... 81 xi  Figure 3.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions on von Restorff (VR) words (red bar), encoding phase VR-surround words (dark blue bars), non-surround words (light blue bar), and test phase VR-surround words (yellow bars) ...................................... 82 Figure 4.1. Mean of participants’ median response time (in ms) to encoding phase cue words, cue-surround words, and non-surround words for the high (dark blue bars) and low (light blue bars) overlap conditions..96 Figure 4.2. Mean recongition test performance (hits) for encoding phase cue words, encoding phase cue-surround words, and non cue-surround words, for the high overlap (dark blue bars) and low overlap (light blue bars) conditions.................................................................................................................................... 98 Figure 4.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words, non cue-surround words for both the high overlap (dark blue bars) and low overlap (light blue bars) conditions .......................................................... 99 Figure 4.4. Mean of participants’ median response time required for making correct recognition decisions about words following actual cues (Panel A) or potential cues (Panel B) in the test phase only................ 100 Figure 4.5. Mean of participants’ median response time (in ms) to encoding phase cue words, cue-surround words, and non-surround words for the high (dark blue bars) and low (light blue bars) overlap conditions110 Figure 4.6. Mean recongition test performance (hits) for cue words, cue-surround words, and non cue-surround words, for the high overlap (dark blue bars) and low overlap (light blue bars) conditions. .......... 112 Figure 4.7. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words and all non-surround words……………….112  Figure 5.1. Depiction of mask stimulus. ..................................................................................................... 123 Figure 5.2. ProM task performance (Panel A) and time required for making ProM task responses (Panel B) in each prime condition .............................................................................................................................. 126 Figure 5.3. Mean of participants’ median response time (in ms) to cue words, cue-surround words, for the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime xii  condition (grey bars), as well as non-surround words that were not preceded by subliminal primes (black bar).. .......................................................................................................................................................... 127 Figure 5.4. Mean recognition test performance (hits) for cue words and cue-surround words in the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime condition (grey bars), as well as non-surround words that were not preceded by subliminal primes (black bar) ...... 129 Figure 5.5. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words and encoding phase cue-surround words for the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime condition (grey bars), as well as non-surround words that were not preceded by subliminal primes (black bar) .......................................... 130   xiii  Acknowledgments I would like to express my deepest gratitude to my supervisor, Dr. Peter Graf, for his guidance, his wisdom, his suggestions, and his company throughout my PhD program and in preparing this document. I would also like to thank my committee members, Dr. Alan Kingstone and Dr. Christiane Hoppmann for their input and thoughtful questions. Thanks also to my lab mates, Janel Fergusson, Natasha Pestonji, Anna Maslany and Nada Alaifan, for their friendship and encouragement throughout this process.  Special thanks to Dr. Randall Jamieson for collaborating with me on the projects presented in this document, and for all of his insight and support.  Thank you to my husband for his support and patience, my parents and sister for their encouragement and unwavering belief in me, and my daughters for being my inspiration and motivation.    xiv  Dedication      To my family, the reason for everything I do     1. Chapter 1: Exploring the Unique Characteristics of Prospective Memory Retrieval Prospective memory is the cognitive function required for forming, maintaining and executing planned tasks at the appropriate time or upon the occurrence of appropriate cues. This cognitive function is required for a wide range of tasks. In daily life, remembering to stop for groceries while driving home is an example of a prospective memory task. In the lab, participants may be involved in a decision task (e.g., making semantic decisions about words) with the prospective memory task requiring them to make a unique response (i.e., press a different key rather than make the semantic decision) upon the occurrence of a specific word or category member (i.e., the cue). The overall goal of this dissertation was to increase understanding of the cognitive processes unique to the successful execution of a prospective memory task.  Prospective memory tasks differ from their retrospective memory task counterparts in that they typically occur in a dual-task context. If I plan to pick up groceries on my way home from work, this plan needs to be executed while I am driving (monitoring traffic, attending to the speed of my car, etc.). If I am in a lab, and my plan is to press a given key upon the occurrence of a specific word, I must do so while I am occupied making decisions about the familiarity of words. In either case, for my plan to be carried out it is necessary to switch cognitive control, at the right time, from the ongoing activity to the planned task. By contrast, this type of response switching is typically not required for retrospective memory tasks. When the need for retrospective remembering occurs in the context of everyday life, I am focused entirely on the recollection of information and there is no need for switching between tasks. Similarly, when retrospective memory is assessed in the laboratory, the participant is typically not engaged in any concurrent activity and is able to concentrate fully on the assigned memory task. Given the unique dual-task context in which prospective memory tasks are typically carried out, my dissertation focused specifically on the cognitive processes engaged in switching from the ongoing activity response to the prospective memory task response. An additional reason for focusing on response switching is that this aspect of prospective 2  memory tasks has not yet received much attention in the prospective memory literature, and thus, we do not yet have a clear understanding of how the switch from the ongoing activity to the planned task comes about, or of the circumstances under which the response switch succeeds or fails. The first section of this chapter provides a description of different types of prospective memory tasks. Among the long list of available tasks, I identify one that seems most directly related to the recall and recognition tests used for measuring explicit episodic retrospective memory. I continue by highlighting the critical characteristics of prospective memory task retrieval, and briefly outline three distinct stages in prospective memory retrieval. The bulk of this chapter is devoted to a task analysis and literature review, where I describe in detail each of the three retrieval stages, speculate on the theoretical mechanisms that are likely to be required for each stage, and present prospective memory research that is relevant to each stage. I focus in particular on the cognitive processes that mediate switching from the ongoing activity response to the prospective memory task response.  Prospective Memory Task Types Research has focused on a broad range of prospective memory tasks. In this section, I describe the distinctions that exist among types of prospective memory tasks, distinctions that are similar to those that exist among retrospective memory tasks (e.g., short-term/long-term, episodic/semantic). This section also serves to clarify the specific requirements of event-based episodic prospective memory tasks that were targeted by my dissertation research. A number of attributes have been used to differentiate among types of prospective memory tasks. These attributes include the retention interval, the frequency with which a plan is carried out, and the type of cue used to signal plan retrieval.  Differences in retention intervals, the amount of time between making a plan and executing it, have given rise to the distinction between short-term and long-term prospective memory tasks, similar to 3  distinctions between short-term and long-term retrospective memory tasks. Short-term prospective memory tasks (e.g., getting a glass of water from the kitchen) have also been called monitoring or vigilance tasks rather than prospective memory tasks. For these tasks, the plan is actively maintained and/or rehearsed during the retention interval and thus is never outside of conscious awareness (Brandimonte, et al., 2001; Brem, Ran, & Pasual-Ieone, 2013; Graf & Uttl, 2001; Harrison & Einstein, 2010). Monitoring tasks are very similar to tasks associated with working memory (Baddeley & Hitch, 1974). By contrast, for long-term prospective memory tasks (e.g., picking up groceries on the way home from work), the time between when the plan is formed and when it must be executed is filled with other cognitive activities (e.g., writing research reports, meeting with colleagues, monitoring traffic). This ongoing cognitive activity effectively prevents the rehearsal and conscious maintenance of the plan (Burgess et al., 2011; Graf, 2005; Reynolds et al., 2009). Therefore, before a plan can be executed, it first needs to be brought back into conscious awareness. Tulving (1972) used to the label ‘episodic’ for retrospective memory tasks that require the conscious recollection of a specific previous event or experience. Consistent with this usage convention, the label episodic also seems appropriate for prospective memory tasks that require that a previously made plan is brought back into conscious awareness. The term episodic is also used to identify prospective memory tasks that only need to be carried out once or only a few times (e.g., picking up the groceries required for dinner tonight; Graf & Utll, 2001). In contrast, the label habitual prospective memory task is used for plans that must be executed regularly (e.g., taking medication daily).  Just as there are many kinds of cues that signal retrospective memory retrieval (e.g., semantic cues, temporal cues, spatial cues), the kinds of cues that signal plan retrieval in prospective memory tasks can either be a time or an event. Tasks for which the cue is either a specific time (e.g., a meeting at 2 pm) or a specific amount of time that has passed (e.g., removing cookies from the oven in 15 minutes) are 4  called time-based prospective memory tasks (Einstein & McDaniel, 1990). Tasks where the cue is an external/environmental stimulus (e.g., a store where you must stop to purchase groceries) are called event-based tasks (Einstein & McDaniel, 1990). One reason the distinction between these cue types is important is because for time-based tasks, it is possible to anticipate the occurrence of the cue, for example, by regularly checking a watch. In contrast, it is less likely that we are able to anticipate event-based cues, particularly in lab-based prospective memory tasks.  Prospective memory tasks that are episodic and event-based are particularly interesting for a number of reasons. First, they are similar to their episodic retrospective memory counterparts that typically involve a single memory associated with a specific time and place (Tulving, 1972). Second, they require that a plan, which has been out of conscious awareness for some time, be brought back into conscious awareness while we are engaged in another ongoing activity. The processes involved in disengaging from the ongoing activity and engaging in the planned task are not well understood in the context of prospective memory. For this reason, my dissertation focus is on the cognitive processes associated with the execution of these kinds of tasks. For ease of communication, however, in this document, I will use the convenient short-hand label prospective memory tasks or the acronym ProM tasks in place of the much longer and more cumbersome label episodic, event-based prospective memory tasks.  Prospective Memory Retrieval: Task Analysis, Stages and Theoretical Implications.  In this section, I examine what is required for the successful retrieval and execution of a prospective memory task. I begin by identifying the critical differences between prospective memory tasks and their episodic retrospective memory (herein referred to as retrospective memory tasks) counterparts. Next, I identify and describe the stages involved in retrieving and executing a planned task. Following these descriptions, I speculate on the theoretical mechanisms implicated by each stage, augmented by insight from existing general theoretical models of performance on prospective memory tasks.  5  Critical Differences between Prospective and Retrospective Memory Task Retrieval. Prospective memory tasks are similar to their episodic retrospective memory task counterparts in some ways. They each involve an encoding or planning phase, where the to-be-remembered information is learned or the plan is formed. They also each have a retention interval, during which the to-be-remembered information or plan is maintained and consolidated. Finally, each task ends with a retrieval phase, and this phase highlights three critical differences between prospective and retrospective memory tasks.  The first of the differences between prospective and retrospective memory tasks is that in the retrieval phase of retrospective memory tasks, our attention is typically focused entirely on the recollection of the encoded information. For example, if we are attempting to recall a colleague’s phone number, retrieving that information is typically our sole focus until the number is recollected and the phone call is made. A retrospective memory task in a lab might require retrieving a list of words studied earlier. While retrieving this list, we are typically not concerned with any other activity. In contrast, prospective memory tasks typically need to be carried out while we are engaged in another ongoing activity. If the plan is to go grocery shopping on the way home, the other activity in which we are engaged, herein referred to as the ongoing activity, is driving. For an in-lab prospective memory task, the ongoing activity might require making fame judgments about people, and the plan might require making a unique response (e.g., press the q-key) when, for example, the display background includes a potted plant. Because of the dual-task context in which prospective memory tasks are typically embedded, the successful execution of a plan requires noticing the cue (e.g., the grocery store, the potted plant) that is relevant to the planned task.   A second critical difference between retrospective and prospective tasks comes from how retrieval is signalled. In a typical retrospective memory task, retrieval is signalled by the provision of cues and specific instructions on what to do with them. Imagine a typical laboratory retrospective memory task where participants are asked to learn a list of words, and told they must later remember these words on a 6  recognition test. During the recognition test, participants are shown a series of words and instructed to indicate whether or not they were previously studied. By means of instruction, participants are explicitly alerted to each word’s (i.e., each cue’s) relevance to the studied information. In contrast, when cues occur in the retrieval phase of a prospective memory task, we typically are not alerted to their relevance to a previously formed plan (Graf & Uttl, 2001). In a laboratory prospective memory task, participants might be presented with a list of words about which they must make semantic decisions (e.g., familiarity decision), and for the planned task, they are instructed to make a different response to words belonging to a specific category (e.g., types of birds). In this example, participants are not explicitly alerted to the fact that a word is relevant to the plan. Instead, they are required to discover the plan relevance of the cue on their own.  A third difference between retrospective and prospective memory tasks comes from the nature of the cues which are frequently used to signal retrieval. In retrospective memory, cues are typically univalent (i.e., only relevant to the current retrospective memory task). If participants are provided with a list of words and asked to indicate whether or not each word was previously studied, each word is only relevant to the retrospective memory task question (e.g., is the word old or new). In contrast, prospective memory task cues are often bivalent, that is, they are relevant to two tasks: the ongoing activity and the planned task (Meier, Woodward, Rey-Mermet, & Graf, 2009). For example, in a prospective memory task where the ongoing activity requires making semantic decisions and the prospective memory cues are words referring to types of birds, the letter string robin is relevant to both the ongoing activity as well as the prospective memory task. The bivalent nature of prospective memory cues suggests that when such cues occur, there is likely to be some sort of competition between two potential responses (Woodward, Meier, Tipper & Graf, 2003).  The term bivalent used to describe cues that signal prospective memory task retrieval differs from the label bivalent given to stimuli in the task-switching literature. The term bivalent stimuli in task-switching 7  paradigms typically means that a given stimulus has features relevant to two tasks (see Metzak, Meier, Graf & Woodward, 2013). For example, if one task is to identify shapes printed in black (e.g., ▲) and the a second task is to identify coloured numbers (e.g., 444), a coloured shape (e.g., ▲) would be considered bivalent because it has both colour and shape features, and is therefore relevant to both tasks. Using this specific definition, it is possible to consider all the words processed as part of the ongoing activity in a lab-based ProM task would be considered bivalent because they are all processed according their semantic features. It is therefore critical to define the term bivalent in the context of my dissertation. As described in the preceding paragraph, when I describe prospective memory task cues as bivalent I mean that there are two potential responses that could be elicited by the cue, the ongoing activity response and the planned task response. Stages in Prospective Memory Task Retrieval  Consistent with the foregoing description of fundamental differences between retrospective and prospective memory tasks, it is possible to describe three distinct stages that must be completed if a previously made plan is to be successfully executed. For ease of communication, I call these stages: cue noticing, recognition of cue-plan relevance, and response switching. These stages are shown in Figure 1.1.    8   Figure 1.1. Stages of prospective memory task execution. Stage 3, response switching, is the primary focus of this dissertation.  We are typically not alerted to the presence of cues pertaining to prospective memory tasks, so the first stage required for successful plan execution is cue noticing. In retrospective memory tasks, cue noticing is not a concern because our attention is focused on the cue by the nature of the task itself. If we are attempting to retrieve a colleague’s phone number, the phone call we are trying to make is our only concern. In the lab, if we are asked to decide whether a word was or was not previously studied (e.g., for an old-new recognition test), our focus is on the word provided for retrieval (i.e., the cue). By contrast, while driving home from work with the intention of picking up groceries, there is no current information instructing us to focus on the buildings along the way to ensure that the grocery store is seen. Similarly, if a lab task requires that we make fame judgments about people, there is nothing to guarantee that we will notice a potted plant sitting on a nearby desk in the picture background. Or, if the ongoing activity requires that we make recognition decisions about words like basketball, there is nothing to guarantee that the cue basket will be perceived as a distinct word. Given these unique concerns, I use the term cue noticing as a short- Stage 1. Cue is noticed Stage 2.  Cue-plan relevance is recognized Stage 3.  Response switch from ongoing activity response program to planned task response program Plan is executed 9  hand label to mean two things. The first is that the cue is processed by the sensory system (e.g., the grocery store or the potted plant). Second, cue noticing means the cue is perceived as a distinct entity (e.g., the letter string basket). A number of factors may facilitate or hinder the cue noticing stage of prospective memory task retrieval. The physical properties of the cue (e.g., size, brightness, or spatial location of a visual cue) likely influence the probability of sensory system processing. If a cue is very small, dim, or displayed in the periphery of the visual field, it is less likely to be processed by the visual system compared to a very large or bright cue, or a cue that appears in the centre of the visual field. The psychological properties of the cue (e.g., familiarity) also may affect the likelihood that the cue is noticed. For example, an unfamiliar word surrounded by a series of familiar words is more likely to be perceived as a unique entity (e.g., Einstein & McDaniel, 1990). Attention also plays a critical role in cue noticing. Our attention may be directed away from the cue (e.g., towards another stimulus), decreasing the possibility of sensory system processing, or we may not have sufficient attentional resources available to process the cue (e.g., because too many resources are going towards the ongoing activity, as when driving in busy traffic), decreasing the likelihood that the cue will be perceived as a distinct entity. A review of research that has investigated the influence of these factors is included in a later section of this chapter.  The second stage necessary for successful plan execution is recognition of cue-plan relevance. It is not enough for the grocery store to be visually processed and perceived as a distinct entity, we also need to connect this cue with the plan we made to go grocery shopping. Recognition of cue-plan relevance is not an issue in retrospective memory task retrieval for two reasons. First, the nature of the memory test specifically indicates the relevance of the cue to the previously encoded information (e.g., the instructions direct participants to indicate whether each word was previously studied). Second, retrospective memory tasks are typically carried out in a single-task context. However, in prospective memory tasks, the cue 10  associated with the plan is typically bivalent, meaning it is relevant both to the ongoing activity and to the prospective memory task. Therefore, even if a prospective cue is noticed, there is no information that alerts us to its relevance to the planned task, beyond its relevance to the ongoing activity.  A number of variables need to be considered in connection with recognition of cue-plan relevance. One of these variables is the explicit nature of the connection between a cue and the associated plan. Instructions in a lab-based task generally offer an explicit link, which might increase the likelihood of recognition of cue-plan relevance (e.g., when you see a word that refers to a type of bird, press the q-key). In contrast, in the real world example of forming a plan to go grocery shopping, it is more likely that the link between the cue and plan is formed implicitly (i.e., we do not usually create instructions for ourselves such as when I see the grocery store, I will go grocery shopping). A second variable is the novelty of the connection between a cue and plan. The link between the cue and the plan in a lab-based task is typically a brand new association, which may reduce the likelihood of recognition of cue-plan relevance. By contrast, in real-life situations, there is frequently an existing semantic and/or episodic connection between the cue and the plan. For example, if the task is to remember a doctor’s appointment, it is much more likely that a cue such as a stethoscope is connected with that plan compared to a cue such as a hot air balloon (e.g., Cohen et al., 2001). Any of these variables might influence the likelihood that the connection created when the plan was initially formed will be reactivated after the cue is noticed. Research which has considered these and other variables that influence recognition of cue-plan relevance is reviewed in a later section of this chapter.  The third stage necessary for successful prospective memory task retrieval is response switching, by which I mean terminating the ongoing activity response and making the planned task response. Response switching is necessary because prospective memory tasks must be executed while we are otherwise engaged in an ongoing activity. In retrospective memory, there is no need for response switching 11  because the cue that signals retrieval is only relevant to one task, the retrospective memory task. For example, if participants are making old/new decisions about a list of words, the only response possible to each cue word is the retrospective memory task response. In contrast, consider a prospective memory task experiment which requires participants to make lexical decisions about words, using the left and right arrow keys (for yes and no responses, respectively), and for the prospective task, to press the q-key when a word representing a bird (e.g., robin) is displayed. In order to press the q-key, participants need to stop making the same response that they made on the preceding words, and to switch and press a different key on the keyboard. Response switching is particularly difficult in lab-based prospective memory tasks, because cue words are infrequent (a cue may be displayed only about every 20 to 30 trials), thus making the ongoing activity response a strong default. Moreover, a response switch may not occur because on most trials, the participant may be permitted to make only one response on each trial, and the default response may be executed before a decision to switch has been made. The factors that facilitate or hinder response switching in the context of a prospective memory task are likely to be similar to factors identified as influencing task-switching more generally, or factors related to Stroop task interference. For example, in the task-switching literature, when switch trials are predictable, switching is more likely to occur (see Monsell, 2003), while increases in working memory load hinder switching (e.g., Kray & Lindenberger, 2000). Both of these variables are relevant to prospective memory tasks, where the occurrence of a cue is rarely predictable, and where we are typically engaged in another ongoing activity that draws on available working memory resources. In the colour-word Stroop task, switching between potential responses becomes more difficult because one response (i.e., word reading) is automatic or highly practiced (Cohen, Dunbar & McLelland, 1990; Morton & Chambers, 1973; Posner & Snyder, 1975). In a typical prospective memory task, the ongoing activity is executed on the vast majority of 12  trials, making it the dominant, automatic and/or more practiced response. As a result, switching in the context of prospective memory task is uniquely challenging. Theoretical mechanisms involved in each retrieval stage. The first stage required for successful prospective memory task execution, cue noticing, is likely to translate into the activation of the cue’s memory representation. A cue’s memory representation is assumed to include sensory, perceptual and semantic information. Different parts of this representation may be activated by different ongoing activities and different prospective tasks. If the task requires, for example, the identification of letters with enclosed spaces (e.g., Meier & Graf, 2000), sensory and perceptual information will be relevant and thus strongly activated. Alternatively, a task that requires ratings of the pleasantness of words will likely access semantic information from the representation (e.g., Taylor, Marsh, Hicks, & Hancock, 2004). I assume that cue noticing is more likely to occur upon the activation of semantic than perceptual information in its memory representation. Whether or not a cue’s memory representation is successfully activated is likely to depend on its history. For cues which are familiar words (e.g., the letter string robin, a picture of a stethoscope), they are assumed to have a pre-existing representation in semantic memory. The strength of such representations is likely to depend on such factors as the familiarity of a word (Brysbaert & New, 2009; Hulme, Maughan & Brown, 1991), or the category typicality of a word (Nosofsy, 1988b). I assume that stronger representations can be activated more easily, perhaps automatically, or more quickly than weaker representations. However, for cues which are novel items (e.g., an orthographic pattern like gib [Scullin et al., 2010]) or a random abstract shape like  (Simons et al., 2006), there is no pre-existing semantic memory representation. I assume that in order for such items to be noticed and function as cues at the time of retrieval, it is necessary that they become represented in episodic memory in the encoding/planning phase of a prospective memory task. I postulate that the noticing of such novel cues occurs only upon the 13  activation of their newly established episodic memory representations, and that this activation is likely to be relatively slow and may involve attentive and strategic processing. The theoretical assumptions outlined in the preceding paragraphs are consistent with the multiprocess framework proposed by McDaniel and Einstein (2000). The multiprocess framework makes assumptions about the nature of all of the processes required for successful performance on prospective tasks, and is less specific about the processes required for individual stages of retrieval. The framework stipulates that depending on the nature of the prospective memory task and the cues used to signal plan execution, successful execution of a planned task may be mediated by automatic processes, called reflexive-associative processes, or it may be controlled by attention demanding strategic processes, called cue-focused processes (e.g., when a cue is peripheral to the ongoing activity, Maylor, 1996) (McDaniel, Guynn, Einstein & Breneiser, 2004). By contrast to the multiprocess framework, the preparatory attention and memory processes model (PAM) proposed by Smith (2003) appears to be more directly pertinent in connection with cue noticing. According to this model, successful prospective remembering depends to some degree on conscious monitoring of the environment for plan relevant cues, and this monitoring is a process that is demanding of attentional resources (Smith, 2003).  It is possible that the theoretical assumptions outlined by the multiprocess framework (McDaniel & Einstein, 2000) and by PAM (Smith, 2003) are both relevant to cue noticing, contingent upon the type of cue used to signal plan execution. A novel cue, such as a unique orthographic pattern or geometric shape, may require conscious monitoring of the environment in order for the episodic representation associated with that cue to be activated. In contrast, the reflexive-associative processes outlined by the multiprocess framework are likely to be involved in the activation of representations associated with cues that have strong pre-existing semantic memory representations.  14  In order for a planned task to be executed at the right time and place, a cue must be recognized as plan relevant, and I assume that this recognition involves processes similar to spreading activation (see Anderson, 1983; Collins & Loftus, 1975). Consistent with spreading activation models of memory such as Collins and Loftus (1975) and Anderson (1983), I assume that once a cue has been noticed, that is, once a cue’s memory representation has been activated, activation will spread in parallel from this representation to all associated representations. I further assume that included among those associated representations is a representation relevant to the ongoing activity as well as a representation of the plan itself. According to the models, the spreading of activation occurs automatically, is faster along pathways that are more strongly associated with the representation of the cue, and it continues until it reaches at least one associated representation (Anderson, 1983; Collins & Loftus, 1975). I assume that recognition of cue-plan relevance is achieved if activation spreads from the cue’s memory representation to the representation of the planned task. Recognition of cue-plan relevance can be difficult to achieve for two reasons. First, it involves competition in a network of associative paths, and the path from the cue to the planned task representation is likely to be weaker than the path from the cue to the ongoing activity representation. To underscore the nature of this competition, consider a laboratory experiment in which common words (e.g., war, friend, robin) are displayed for an ongoing activity that requires answering a question about each word (e.g., Does this word refer to something positive or negative?), and the prospective task requires pressing a designated key on the computer keyboard (e.g., the q-key) when a word representing a bird is encountered. A property such as positive or negative is likely to be part of the pre-existing semantic memory representation of common words (see Anderson, 1983; Collins & Loftus, 1975; Moscovitch, 1994), whereas there is no reason to assume the prior existence of an association between a word, such as robin, and the representation of pressing the q-key. The latter association exists only if it was created in the instruction 15  phase of the experiment. Therefore, consistent with spreading activation models of memory, when a word such as robin is encountered on an experimental trial, it is likely that the association between its representation and the property positive is activated more quickly than the association between the robin representation and the q-key representation. An additional reason for assuming that the activation competition will be won by the first kind of association stems from the fact that in a typical laboratory prospective memory experiment, prospective memory task cues occur only on a small proportion of trials, such as 1 out of 30 trials. Because the vast majority of experimental trials are ongoing activity trials, the ongoing activity response has a substantial head-start in the activation competition. In light of the foregoing theoretical considerations, what needs to be explained is how the activation competition might be biased in order to bring about the activation of the representation of the plan. One way in which this might occur is if the path between the cue and the plan representation is stronger than the path between the cue and the ongoing activity representation. This type of situation would exist if the ongoing activity required performing an unfamiliar, novel task such as determining how many consonant-vowel-consonant-vowel (cvcv) groupings there are in a letter string, and the prospective memory task built on a strong pre-existing association (e.g., When a real word is displayed, read it aloud). There can be little doubt that counting the number of distinct cvcv groupings in a pseudoword like salutene would be slow and effortful, whereas saying aloud a word like robin would be fast and supported by automatic spreading activation processes. The situation described in the preceding paragraph suggests one way in which the activation competition might be biased so that a prospective memory task response can occur even in experiments where the majority of trials requires making an ongoing activity response. Unfortunately, as the review of the research in the next section of this chapter will show, most experiments on prospective memory do not involve this kind of manipulation. Instead, in nearly all cases, the stimuli used as cues have a strong and 16  pre-existing association with the ongoing activity representations, and the association of the cues and the prospective task representation is newly created in the planning phase of the experiment. Consequently, a different mechanism is required to explain how the activation competition might be biased in favor of making a prospective memory task response. In the following paragraphs, I propose three potential mechanisms.  One such mechanism builds on the Zeigarnik effect (1927), which is the claim that tasks which remain uncompleted are maintained in a state of heightened activation (see Goshke & Kuhl, 1993), and that prospective memory tasks are tied to specific contextual cues (Nowinski & Dismukes, 2005). If the representation of a yet-to-be-completed task is maintained in an elevated state of activation, it is possible that this activation is never substantially lower than the activation of the ongoing activity representation, even in a situation where the latter has been relevant to a long series of trials. If this is the case, only a small amount of additional input may be required to ensure that the activation competition is won by the association between the cue and the planned task. This additional input might be provided by the context in which prospective task cues are expected to occur. In the planning phase of prospective memory experiments, participants may be given specific instructions that link both cues and the context in which they will occur to the representation of the planned task (see Meier, Zimmerman & Perrig, 2006). Consequently, when the expected context is reinstated in the retrieval phase, this gives additional activation to the association between the cue and the planned task, and thereby contributes to cue-plan relevance recognition. This assumption is consistent with the encoding specificity principle (Tulving & Thomson, 1973).  Yet another potential mechanism to explain how the activation competition might be won by the association between the cue and the planned task is suggested by the discrepancy attribution hypothesis (Whittlesea & Williams, 2001). The discrepancy attribution hypothesis has been used to explain a number 17  of phenomena, including the occurrence of false memories (Whittlesea, 2002) and false fame judgments (Jacoby, Woloshyn & Kelley, 1989). According to this hypothesis, when the perceptual or cognitive processing of a stimulus is different from what was expected (e.g., faster or easier than expected), this triggers a discrepancy reaction, something which is akin to a double-take and deeper analysis of the current stimulus or situation. It is possible that in the planning phase of an experiment, when participants receive instructions about a prospective memory task (e.g., whenever in the course of this experiment, you see a word that refers to a bird, press the q-key on the keyboard), this serves to prime the representations of all potential cues. As a consequence of this priming, when such a cue occurs subsequently in the course of the experiment, its processing is more fluent than expected, and thus triggers a discrepancy reaction (see McDaniel & Einstein, 2000; Einstein et al., 2004). In turn, this reaction may interrupt the usual way of responding and delay the default response associated with the ongoing activity, thereby creating an opportunity for recognition of cue-plan relevance.  A final possibility is that the activation competition is won by the representation of the plan because strategic processes are used to guide performance. The PAM theory (Smith, 2003) suggests that strategic processes are engaged for searching for plan relevant cues. To guide this search, it seems likely that the association between the cue and the plan is already primed in some manner. As a consequence, when the search process lands on a possible target, the primed association is fully activated more quickly.   To summarize, it is likely that different mechanisms are involved in recognition of cue-plan relevance for different kinds of cue-plan associations. Strategic processes, like those suggested by PAM (Smith, 2003) might be employed in situations where the association between the cue and the plan is particularly weak. The representation of the plan might remain primed, as suggested by the Zeirgarnik (1927) and Goshke and Kuhl (1993) if the planned task is particularly important (e.g., Kliegel et al., 2001), or when information about the context in which a cue is expected to occur is prominent (e.g., Meier, 18  Zimmerman & Perrig, 2006). A discrepancy reaction, as described Whittlesea and Williams (2001), and applied to prospective memory task execution by Einstein et al. (2004), might be the mechanism involved in recognition of cue-plan relevance in other circumstances. For example, discrepancy reactions might facilitate recognition of cue-plan relevance when specific contextual information about where or when the cue is expected to occur is not available, or when a category of cues (e.g., a type of bird) rather than a specific cue (e.g., robin), is indicated as relevant to the planned task. Regardless of the mechanism, recognition of cue-plan relevance can only occur if activation spreads from the cue’s memory representation to the representation of the planned task. Even if a cue is recognized as plan relevant, a planned task will be carried out only if a switch is made from responding to the ongoing activity to responding to the prospective task. I assume that this type of response switching is mediated by processes that have been used to explain Stroop task interference (Cohen, Dunbar & McLelland, 1990; Dunbar & MacLeod, 1984; MacLeod, 1991; Morton & Chambers, 1973; Posner & Snyder, 1975; Stroop, 1935) and well as task-switching (Mayr & Keele, 2000; Monsell, 2003; Monsell, Sumner, & Waters, 2003; Waszak, Hommel, & Allport, 2002). The Stroop task interference effect is the finding that it takes longer to name the ink colour of a word such as red when it is printed in blue or green, than to name the ink colour of a word such as table which does not refer to a colour (Stroop, 1935). Stroop task interference is assumed to occur because attention demanding executive control processes need to be engaged for inhibiting the faster, more automatic, word reading response so that the more attention demanding colour naming response can be made (see MacLeod, 1991). An alternative view is that the cognitive system has a limited capacity response buffer which can be occupied by only one response at a time. Because word reading is quicker than colour naming, the word reading response gets into the buffer before the colour naming response (Cohen, Dunbar & McLelland, 1990; Morton & Chambers, 1973; Posner & Snyder, 1975). Thus, for the 19  latter kind of response to occur, the word reading response needs to be deactivated and removed from the buffer. It is likely that the same kind of response deactivation or inhibition is required in order to make a prospective task response. Because most trials in a prospective memory experiment require making an ongoing activity response, this response is likely to be generated more quickly, and it needs to be inhibited or deactivated in order for a prospective task response to occur. Importantly, it is assumed that attention demanding executive control processes are required for inhibition or deactivation of default responding, and for switching from the ongoing activity response to the prospective memory task response. The task-switching literature augments our understanding of the processes that might be involved in making the switch from executing the ongoing activity response to the planned task response. In a typical task-switching experiment, participants are trained on two (or more) tasks associated with a set of stimuli that have multiple characteristics, and one (or more) of these characteristics is used to indicate which trained task must be carried out. According to Mayr and Keele (2000), responding in a task-switching situation is guided by a specific task-set or action schema, with each defined by a unique configuration of mental resources. When a switch trial occurs, the task-set from the preceding trial needs to be inhibited (i.e., backward inhibition is required), and a different task-set needs to be instantiated. Switching costs, defined as slower responding on trials that require a switch, occur because of the additional time required for inhibiting the task-set from the preceding trial and for activating a new task-set. Mayr and Keele (2000) and others (e.g., Monsell, Sumner, & Waters, 2003) have explained that switch costs are higher when switching away from highly familiar tasks because their task-sets are more difficult to inhibit. In the context of a prospective memory task, a substantial amount of backward inhibition may be required to switch to the prospective memory task response, because the vast majority of trials require making an ongoing activity response. 20  Allport, Styles and Hsieh (1994) and others (e.g., Allport & Wylie, 1999; Monsell, 2003) have speculated that switching between task may also be associated with task-set inertia, defined as the tendency for a task-set to persist over time and to proactively interfere with creating a new task-set configuration (Wylie & Allport, 2000). Task-set inertia may be particularly strong for less practiced tasks because their existence and maintenance is more dependent on executive control and working memory processes (Waszak, Hommel, & Allport, 2002). Consistent with this assumption, it is likely that in prospective memory experiments, the task-set inertia is greater for switching back to the ongoing activity response after making a prospective task response than vice versa. To summarize, the preceding paragraphs have highlighted three insights that might be characteristic of response switching in prospective memory tasks. First, response switching requires inhibition of the dominant response’s task-set (MacLeod, 1991; Mayr & Keele, 2000). In the case of prospective memory task retrieval, the dominant response requiring a greater degree of inhibition would be the ongoing activity response. Second, the task-switching literature explains that in order to switch from one task’s response to another, a new task-set must be configured, and this process requires executive control (see Monsell, 2003). This process is likely relevant to any trials where we switch from the ongoing activity response to the prospective memory task response, and then back again. Third, task-set inertia, or the tendency for one task-set to proactively interfere with the configuration of a new task-set, is more problematic when switching from less practiced responses to more practices responses (Allport et al., 1994). In other words, we would expect a greater degree of task-set inertia when switching away from the rarely occurring prospective memory task response to the dominant ongoing activity response.  Some caution is warranted, however, in applying the insights described above to response switching in prospective memory task retrieval, because the demands of prospective memory tasks are very different from the demands of the Stroop task or the demands of task-switching situations. Specifically, 21  on each trial of the incongruent condition of the Stroop task, the participant is required to suppress a predominant and highly practiced response in favor of a less practiced one, and there is no requirement to make any kind of response switch. A critical difference between task-switching situations and prospective memory task retrieval is that switches occur much more frequently in task-switching situations, and in many cases occur predictably (see Monsell, 2003). Despite these dissimilarities, the insights provided by the Stroop task and task-switching literature are a good foundation upon which to increase our understanding of response switching in prospective memory retrieval. Review of existing research relevant to each retrieval stage. The vast majority of previous investigations have focused on overall performance on prospective memory tasks, rather than on specific factors that influence cue noticing, recognition of cue-plan relevance, or response switching. Global manipulations that are likely to impact multiple stages of retrieval include task instruction manipulations (e.g., manipulations that vary the relative importance of the prospective memory task) and task difficulty manipulations (e.g., manipulations that change the pool of attentional resources available for the prospective memory task). I briefly outline how these kinds of manipulations might influence multiple stages of retrieval below.   Task instruction manipulations might direct attentional resources towards cue noticing, by emphasizing the relative or absolute importance of executing the prospective memory task (Kliegel, Martin, McDaniel & Einstein, 2001; Brandimonte et al., 2010; Kvavilashvili, 1987; Walter & Meier, 2016). Emphasizing the importance of the prospective memory task encourages participants to focus their attention (i.e., via top-down processes) explicitly on cue noticing by strategically monitoring their environment for cues (Smith, 2003). It is also possible that task instructions that emphasize the importance of the prospective memory task influence recognition of cue-plan relevance. If the plan is important, the 22  representation of the plan might remain in a heightened state of activation, thus ensuring the connection between the cue and the plan is activated quickly and/or automatically after the cue is processed.  The attentional demands of the ongoing activity also might influence both cue noticing and recognition of cue-plan relevance. It is possible that our attention is directed away from the cue (e.g., towards another stimulus [see Hicks, Cook, & Marsh, 2005]) or we do not have sufficient attentional resources available to process the cue as a distinct entity (e.g., because the ongoing activity is too difficult [see Marsh, Hancock, & Hicks, 2002]). Alternatively, a difficult ongoing activity might leave fewer attentional resources available for recognition of cue-plan relevance. I believe attentional resources might be necessary for recognition of cue-plan relevance because in many prospective memory experiments, the associations between the cue and the plan are relatively weak and created only in connection with a specific experiment (e.g., press the q-key when you see a bird word). Such newly created associations are much weaker than a pre-existing association which mediates responding to the ongoing activity. It is possible that the latter are activated automatically, while the former are activated only if sufficient attentional resources are strategically allocated for this purpose. By contrast to these kinds of global manipulations that are likely to affect more than one of the stages involved in the successful execution of a prospective memory task, there is also research which has specifically investigated the unique role of each distinct stage. The section which follows gives a brief review of this type of research. The purpose of this section is not to provide a comprehensive review of previous research, but to illustrate the methods used for investigating each retrieval stage and to summarize critical findings.  Stage 1: Cue Noticing. Cue noticing is the first stage in prospective memory task retrieval, and it is the stage within which the cue must be processed by the sensory system and perceived as a distinct entity. I propose that this is achieved only if the memory representation associated with that cue is 23  activated. There are two categories of manipulations in the prospective memory research that are most clearly relevant to cue noticing: manipulations of perceptual salience and manipulations of psychological salience of the cue.  A cue is more likely to be noticed if it is perceptually salient. Visual salience has been manipulated by increasing the size of the cue relative to the stimuli used for the ongoing activity (Graf, Uttl, & Dixon, 2002). In their experiment, Graf, Uttl, and Dixon (2002) used pictures as the prospective memory task cue. The first time the cue was displayed, its size was 98 x 72 pixels. Each time the cue was displayed again, its size increased in width and height, until a maximum size of 336 x 252 pixels. Results indicated that increasing cue size improved prospective memory task performance. Focusing on a different physical property of a cue, Uttl (2006) had participants complete a card-sorting task during which random sounds were played. Participants were told to ignore the sounds during the ongoing activity, except for one specific sound, the sound of a camera clicking. When they heard the camera click sound, they were asked to stop and tell the experimenter they heard it. The loudness of the cue sound gradually increased across trials until the prospective memory response was made, while the random noise or non-cue stimuli stayed at a constant volume. Results indicated that the louder the volume of the cue, the more successful prospective memory task performance was, particularly for older adults.  Cue noticing is also influenced by the spatial position or location of the cue. Cohen and colleagues (2003) had participants make a prospective memory response to specific letters (e.g., the letter E) that occurred within a visual search task as the ongoing activity. In their manipulation, they spatially displaced the cue from the surrounding letters in the visual search task (see Figure 1.2). This manipulation significantly improved prospective memory task performance (Cohen et al., 2003). Trawley, Law, Brown, Niven, and Logie (2014) studied the effect of the perceptual salience of the cue in the context of a virtual reality scenario. Prospective memory task cues were considered high or low salience based on their 24  locations within virtual rooms (i.e., within or outside the focus of visual attention). As predicted, the prospective memory task was executed more often when high salience cues were used.   Figure 1.2. Depiction of Cohen et al.’s (2003) spatial displacement of the prospective memory cue  Manipulations of psychological salience also are likely to impact cue noticing. One example of a psychological salience manipulation is illustrated in a study where the cue was concealed within a larger stimulus. For example, if the cue word was the letter-string ball, and this string appeared in the word basketball, the cue was less likely to be perceived as a distinct entity, and as a result prospective memory task performance suffered (e.g., Kliegel et al., 2008; Scullin et al., 2010). The cue may not have been noticed because, if perceived as an integral part of the stimulus, the semantic memory representation for the component word ‘ball’ might not have been activated when the word ‘basketball’ was processed.  A related method for manipulating cue noticing has focused on different ways of processing cues. For example, if the prospective memory task requires a unique response to words representing kinds of fruit, I will only make the planned response if I read the word apple as representing a type of fruit. By contrast, if I read apple as a brand of computer, I will not make the planned response. These different ways of processing a stimulus are captured by the distinction between focal and non-focal cues. Focal cues are cues that require processing consistent with an ongoing activity (McDaniel & Einstein, 2000). If the ongoing activity requires semantic processing of words (e.g., making a familiarity decision about each word), and the prospective memory task requires participants to respond to a specific semantic category of words (e.g., types of birds), then the cue is considered focal. In contrast, non-focal cues require a different type of 25  processing, which can be considered optional as it relates to the ongoing activity (McDaniel & Einstein, 2000). If the ongoing activity requires semantic processing of words, but the cue is defined by the physical properties of a word (e.g., if it is written in lower case letters), that cue is considered non-focal. As the ongoing activity in their prospective memory experiment, Meier and Graf (2000) had participants make either semantic judgments (categorizing words as natural or fabricated) or perceptual judgments (categorizing words by the number of enclosed spaces in the letters). Similarly, the prospective memory task cues varied between words that either required semantic processing (they were words that represented types of animals) or perceptual processing (they were words that had three e’s). The results showed that focal cues were facilitative to prospective memory task execution (Meier & Graf, 2000). A number of researchers have replicated this transfer appropriate processing effect (e.g., Einstein et al., 2005; Scullin, McDaniel, Shelton, & Lee, 2010).  Researchers have also manipulated the psychological salience of prospective memory task cues by varying participants’ level of familiarity with them. In a manipulation of the relative familiarity of cues, prospective memory task performance improved markedly when the level of familiarity of words that served as cue words differed from the level of familiarity of words used for the ongoing activity (e.g., frequent words as prospective memory task cues surrounded by infrequent words relevant to the ongoing activity, or vice versa Brandimonte & Passolunghi, 1994). Einstein and McDaniel (1990) varied the level of familiarity of the words for both the prospective memory task and ongoing activity. They had participants study a list of familiar and unfamiliar words as the ongoing activity, to be remembered for a later memory test. The prospective memory task was to press a response key if a specific word appeared (e.g., rake in the familiar condition; monad in the unfamiliar condition). They discovered that using unfamiliar words as cues improved prospective memory performance, even in conditions when the ongoing activity also consisted of unfamiliar words. A potential explanation for the improvement in prospective memory task performance is 26  that rare words, such as monad, attract more attentional resources and require more processing to identify and/or define, thus ensuring the cue is noticed. Stage 2: Recognition of Cue-Plan Relevance. Recognition of cue-plan relevance is the stage in prospective memory task retrieval during which the connection between the cue and the representation of the plan must be activated. In the preceding section, I described a number of theoretical mechanisms that could potentially explain how the connection between the cue and the plan can be activated more quickly and/or to a greater degree than the connection between the cue and the ongoing activity. Consistent with these potential theoretical mechanisms, I describe research that has used two kinds of manipulations likely to influence recognition of cue-plan relevance. The first kind of manipulation varies the strength of the association between the cue and the prospective memory task itself, and the second varies the context in which the cue is expected to occur.  To investigate the role of recognition of cue-plan relevance, a number of studies have focused on the strength of the pre-existing semantic or episodic associations between the cue and the prospective memory task. For example, Cohen, West, and Craik (2001) trained participants on a paired-associates task, where each pair consisted of a picture and an unrelated word. Participants were told to remember the pairs for a later test, and this test would serve as the ongoing activity within which the prospective memory task was embedded. For the prospective memory task, some of the pictures were paired with a cue word and a written intention (e.g., I must go to the doctor, a pair with a picture of a doctor’s waiting room and the word doctor). Participants were told that when the cue pictures came up in the retrieval phase, they were to state aloud the intention they had previously studied. The relatedness of the cue and the intention was varied. For the intention I must go to the doctor, they used either a picture of a doctor’s waiting room or a picture of a hot air balloon as the cue. The cue with a strong semantic association (i.e., the picture of a doctor’s waiting room) significantly improved performance on the prospective memory task compared to the 27  cue that did not did not share a pre-existing semantic association (the hot air balloon). Other researchers have extended Cohen et al.’s (2001) findings, showing participants perform better on prospective memory tasks when the cue-response pairings had high associative connections (e.g., respond with the word ‘paper’ when you see the word ‘document’) compared to low associative connections (e.g., respond with the word ‘wash’ when you see the word ‘document’) (Loft & Yeo, 2007). Pereira, Ellis and Freedman (2012) looked to extend the findings further by encouraging participants to form new episodic connections for previously unrelated cue-plan pairs. In their experiments, participants learned cue-plan pairs that were semantically related (e.g., throw-ball) or semantically unrelated (e.g., throw-lemon). For some pairs, participants were asked to physically act out the cue-action pairs, thus creating an episodic memory representation. They found that performance was superior for the pairs that were higher in relatedness, but the difference was reduced when the action was included (i.e., when an episodic association was formed) (Pereira et al., 2012).  In many prospective memory research paradigms, participants are provided with a category of cues rather than with specific cue words, allowing researchers to manipulate the typicality of cue category exemplars and in turn assess the influence of category typicality on recognition of cue-plan relevance. For example, if the cue category was types of birds, a typical category member might be robin, whereas an atypical member might be penguin. In retrospective memory research, it has been demonstrated that typical category members are retrieved more frequently (Uyeda & Mandler, 1980), more easily (Rothbart et al., 1996), and more quickly (McCloskey & Glucksbery, 1978; Rips et al., 1973). In prospective memory research, similar manipulations result in improved performance for typical category member cues over atypical category member cues (e.g., Cherry et al., 2001; Nowinski & Dismukes, 2005; Penningroth, 2005).  A spreading activation framework (see Anderson, 1983; Collins & Loftus, 1975) is helpful in understanding these results. When the plan is initially formed, its representation is associated with the 28  category of types of birds, and all bird category members might be primed. However, the semantic distance between the plan and typical category memory cue word would be much smaller than the plan and an atypical category member (see Figure 1.3), and thus would be activated more quickly (or, perhaps, to a greater degree). As the cue word and its category are activated, so too is the representation of the plan. When an atypical category member occurs, the degree of activation of the plan could be minimal, very slow, or even non-existent, due to a greater semantic distance between the cue word, its associated category, and the representation of the plan (see Figure 1.3).   Figure 1.3. Depiction of semantic distance of typical and atypical category members as prospective memory cues, inspired by similar models from Collins and Loftus, 1975. Cook, Marsh, Hicks and Martin (2006) used an associative fan manipulation (see Anderson 1974) to investigate the contribution of cue-plan relevance recognition on prospective memory task performance. In Phase One of their experiment, they had participants learn word pairs. Some words were associated with only one other word, and some were paired with four other words. In Phase Two of the experiment, they completed a lexical decision task. It was during Phase Two that the prospective memory task was introduced; it required pressing the I-key in response to any word representing an animal. For this task, three animal words (horse, cow and dog) were each paired with either zero, one or four target words in Phase One. Cook et al. (2006) predicted that increasing the associative fan (i.e., the number of words with which the cue was originally paired) would decrease the likelihood that the representation of the plan would be activated in connection with a cue, given the larger number of alternative representations that could be activated. Their hypothesis was supported by the results. As the number of target words paired with the cue TYPICAL e.g., Robin  ATYPICAL e.g., penguin  PLAN e.g., Press the  q-key  CUE CATEGORY e.g., bird words  29  during the associative-pairs task increased, performance on the prospective memory task itself decreased. The authors explained that the larger associative fan (see Anderson, 1974) of the cue decreased the degree and/or strength of the activation of the representation of the plan associated with that cue.  Implementation intentions, a term coined by Gollwitzer (1990), are if-then statements generated to link a situation with an action, strengthening the association between the cue and the plan (Gollwitzer, 1990, 1993) and helping to ensure recognition of cue-plan relevance. These if-then statements likely help to form a strong episodic memory association between the cue and the planned task and ensure the representation of the plan is more likely to be activated upon the occurrence of the cue, perhaps by maintaining the representation of the plan at a heightened level of activation. According to Gollwitzer, by making an explicit statement such as, I intend to perform goal-directed behavior y when I encounter situation z (Gollwitzer, 1993), a contextual and personal connection between the cue and the plan is formed. McDaniel, Howard and Butler (2008) compared participants’ performance on prospective memory tasks when they were given standard instructions versus instructions that emphasized stating implementation intentions. The implementation intention instructions encouraged participants to imagine themselves executing the plan upon the occurrence of the cue, whereas the standard instructions made no such explicit request. Consistent with Gollwitzer’s (1990, 1993) prediction, McDaniel et al. (2008) found that implementation intentions significantly improved prospective memory task performance. A number of other researchers have replicated this finding (e.g., Meeks & Marsh, 2010; Zimmerman & Meier, 2010).  Recognition of cue-plan relevance may also be manipulated via the context in which a plan needs to be executed. Researchers are able to create contextual expectations about the prospective memory task when the plan is initially formed (i.e., when the prospective memory task instructions are provided). For example, Meier, Zimmerman, and Perrig (2006) had participants complete three tasks during their experiment: a lexical decision task, a concreteness judgment task, a short-term memory task, along with 30  the prospective memory task itself. The order of the three tasks was counterbalanced across conditions. The prospective memory cues were words representing musical instruments, and the task was to alert the experimenter about the words’ occurrence and then briefly describe the instruments. Half of the participants were told that the musical instrument words would occur during the short-term memory task; the other half of participants were not given any information about the context in which they could expect to see cue words. Performance on the prospective memory task was better for participants who had an expectation of the context in which the cue would occur.  Related findings suggest that violating expectations about the context in which cues are expected to occur makes the activation of the representation of the plan less likely in connection with the cue. In Cohen et al.’s (2001) study, described earlier in this chapter, cues were initially presented as words. Participants were told that during the test phase, they must respond to cues by saying out loud the associated intention. However, during the test phase some of the cues were presented as pictures and as a result prospective memory task performance suffered.  Stage 3: Response Switching. The need for response switching is the feature that most clearly distinguishes retrospective from prospective memory tasks, and is assumed to require inhibition of the ongoing activity response, task-set configuration, and task-set inertia. To my knowledge, only one research group has directly investigated the cognitive processes involved in response switching in prospective memory task retrieval.  The costs and consequences of switching from the ongoing activity to the planned task have been investigated in a series of experiments by Meier and Rey-Mermet (2012; 2017). Meier and Rey-Mermet (2012) used a task-switching paradigm as the ongoing activity in which the prospective memory task was embedded. They trained participants on three tasks that served as the ongoing activity: a parity judgment task that used triplicate numbers (e.g., 777), a colour judgment task for symbols displayed in either red or 31  blue (e.g., $$$), and a case decision task for triplicate consonants displayed in black font (e.g., nnn). For the prospective memory task, participants had to press the spacebar if ever they came across any consonant-vowel-consonant triplicates. They found evidence of slowing on the ongoing activity trials immediately after the prospective memory task cue. In a series of follow-up experiments using a similar procedure, Meier and Rey-Mermet (2017) determined that the greater the overlap between the processing required for the ongoing activity and the prospective memory task (e.g., when both tasks required processing of letters), the larger and longer-lasting the costs to the ongoing activity trials immediately following the prospective memory task cue.  The findings from Meier and Rey-Mermet’s (2012; 2017) experiments are consistent with theoretical accounts for the effects observed in task-switching paradigms. First, task-set configuration requires attentional resources (Monsell, 2003), and the configuration of a rarely occurring task-set, like the prospective memory task, requires additional attentional resources (Waszak, Hommel, & Allport, 2002). These demands on attentional resources tend to slow response times following a switch (Allport et al., 1994; Monsell, 2003; Wylie & Allport, 2000). Further, in order to switch from the ongoing activity to the prospective memory, the preceding response (i.e., the ongoing activity response) must be inhibited (Mayr & Keele, 2000), and the more practiced and dominant the response the more inhibition is required (Monsell, Sumner & Waters, 2003). Therefore, switching back to the ongoing activity response following a prospective memory response is particularly difficult. Additionally, Meier and Rey-Mermet (2012; 2017) suggested that increasing the overlap between the ongoing activity and the prospective memory task increased the costs of executing the prospective memory task. One potential explanation for this finding was that the prospective memory task was reactivated during the ongoing activity by stimuli that shared similar features (Meier & Rey-Mermet, 2017). This finding is consistent with task-set inertia, where a task-set persists over time and interferes with the configuration of a new task (Allport et al., 1994).   32  To the best of my knowledge, the research conducted by Meier and Rey-Mermet (2012; 2017) is the only research specifically examining the immediate consequences of executing a prospective memory task. The findings must therefore be viewed with caution. However, the results are encouraging given that they are consistent with the theoretical mechanisms I believe are involved in response switching during prospective memory task retrieval. Additionally, the work of Meier and Rey-Mermet (2012; 2017) highlights the kind of method that can be used to investigate response switching in the context of prospective memory experiments. The experiments included in this dissertation build on this initial research and methodology to address a number of unanswered questions.   Overview of Dissertation  A primary purpose of this chapter was to identify and discuss the unique stages involved in prospective memory task retrieval; namely, cue noticing, recognition of cue relevance and response switching. I described my assumptions regarding the theoretical mechanisms involved in each stage of prospective memory task retrieval, and outlined research relevant to the theoretical mechanisms I assume to be involved in each stage. One area in which existing research is minimal is the response switching stage, and a goal of this dissertation is to report new research that improves our understanding of this stage of prospective memory task retrieval. In the chapters that follow, I present a series of experiments conducted to test the assumptions I have made regarding the theoretical mechanisms involved in prospective memory task response switching. Chapter 2 describes two experiments that examined the consequences of response switching during prospective memory task retrieval on the processing of stimuli immediately preceding and following plan execution. Experiments 1 and 2 each had two phases, an encoding phase where participants answered semantic questions about a series of unrelated words, and a recognition test phase, where participants indicated whether or not words occurred during the encoding phase. In Experiment 1, the prospective 33  memory task was embedded within the encoding phase, and in Experiment 2, it was embedded within the recognition test phase. In these experiments, I observed that executing a prospective memory task slowed ongoing activity responses on words following the cue, and it interrupted both encoding and retrieval processes on words that presented immediately after the cue. I attributed these results to two response switching mechanisms: task-set configuration and task-set inertia. The basic method I developed and used in Experiment 1 was carried throughout the rest of the dissertation to further elucidate the processes involved in response switching during prospective memory task retrieval.  In Chapter 3, I describe an experiment that tested an alternative explanation for the findings observed in Experiments 1 and 2. The prospective memory task cues in Experiments 1 and 2 occur rarely, and participants are directing more attentional resources towards them, which in turn makes cues stand out as distinctive. It is therefore possible that any distinctive stimulus is capable of eliciting a similar pattern of results as observed in Experiment 1. In Experiment 3, I presented participants with a perceptually salient stimulus in the place of a prospective memory task cue, and measured the same variables to determine if any unique or distinctive stimulus is sufficient to elicit the pattern of results observed in Experiment 1. I determined that distinctive stimuli are able to slow response time but do not interfere with encoding processes to the same extent as prospective memory task execution.  Chapter 4 reports two experiments that explored the influence of the degree of overlap in the processing required for the ongoing activity and the prospective memory task on prospective memory task response switching. Experiments 4a and 4b each had participants complete a prospective memory task in a high processing overlap condition and a low processing overlap condition. In the high overlap condition, the prospective memory task cue required the same kind of processing as the ongoing activity (e.g., they both required semantic processing). In the low overlap condition, the prospective memory task cue required a different kind of processing than the ongoing activity (e.g., the cue required perceptual processing and 34  the ongoing activity required semantic processing). The results showed that planned task execution is more disruptive to ongoing activity processing in conditions of high overlap between the prospective memory task and the ongoing activity.  In Chapter 5, I present exploratory research motivated by the goal of understanding how primes presented subliminally immediately prior to prospective memory task cues influence response switching processes. Immediately prior to the presentation of the prospective memory task cue words, I subliminally presented repetition, category, and unrelated primes. The results showed that planned task performance did not differ significantly based on prime type, but revealed that ongoing activity processing following a planned task response was disrupted more by category primes, compared to repetition and unrelated primes.  In the final chapter of my dissertation, I summarize the empirical, methodological and theoretical contributions from my dissertation. I describe implications and limitations of the research presented in this dissertation, and suggest avenues for future research.     35  2. Chapter 2: Consequences of Prospective Memory Task Response Switching1 Prospective memory is the cognitive function required for encoding, maintaining and executing planned tasks at the appropriate time or upon the occurrence of a relevant cue. For example, if the plan is to go grocery shopping on the way home, we must retain this plan while at work throughout the day, and then act on it when driving home and when our attention is occupied with the demands of driving. The goal of the research we report in this article was to increase understanding of the cascade of activities that must occur in order for a planned task to be executed on the occurrence of a plan relevant cue. The processes engaged for the execution of a plan are likely to be different for different kinds of prospective memory tasks. Specifically, for an immediate task (e.g., going next door to borrow a stapler from an office colleague), the plan is likely to be held in conscious awareness throughout the retention interval (i.e., the time between making a plan and executing it). For this reason, the sequence of steps required for plan execution is likely to be the same as used for all tasks in which a conscious intention needs to be translated into action (see Baars & Franklin, 2003; Baddeley, 1992; Libet et al., 1983). By contrast, for longer-duration tasks where there is a substantial delay between planning and execution (e.g., getting groceries on route home from work; posting a letter when we encounter a mailbox), the plan is not likely to remain in conscious awareness during the retention interval (Graf & Uttl, 2001). Consequently, when a plan relevant cue occurs, our focus is likely to be on ongoing activities. In order to carry out the planned task, those ongoing activities need to be interrupted and the planned task needs to be brought back into conscious awareness. In this article, we use the label event-based episodic prospective memory for such tasks, herein referred to by the acronym ProM. We also use this label to distinguish such tasks                                                      1 This chapter is a manuscript that has been prepared for submission. As such, some of the content discussed at length in Chapter 1 is summarized again in the introduction of this manuscript. Because this research was submitted as a collaboration between M. Crease Lark, R. Jamieson and P. Graf, the pronoun “we” is used. However M. Crease Lark is the lead author and principle investigator who designed the studies, ran participants through the experiments, analyzed data and wrote this report.  36  from habitual prospective tasks (e.g., taking medication) which tend to be carried out repeatedly, often according to a well-established schedule. A cascade of activities is likely to be required for the successful execution of a ProM task (cf. Mäntylä, 1996). First, when a plan relevant cue occurs, it needs to be noticed (e.g., Mäntylä & Backman, 1992; Maylor, 1990; McDaniel & Einstein, 1993; Smith, 2003). By noticing we mean that the cue has to be processed by the sensory system and perceived as a distinct unit. Second, the cue needs to be recognized as relevant to the planned task (McDaniel & Einstein, 2000; Nowinski & Dismukes, 2005). We assume that at the time of planning a future task (e.g., mailing a letter), we either activate existing associations or forge new associations between this task and potential cues (e.g., a mailbox on the sidewalk) for initiating subsequent retrieval. Consistent with this assumption, we postulate that recognition of cue-plan relevance is mediated by the reactivation of these pre-existing or newly formed associations (see Anderson, 1983; Collins & Loftus, 1975). Third, in order for a planned task to be carried out, it is necessary to switch from the response appropriate for an ongoing activity to the response required for the ProM task (e.g., Meier & Rey-Mermet, 2012; West, Scolaro, & Bailey, 2011). The research reported in this article is concerned specifically with the processes involved in response switching.  Research on ProM has focused primarily on global variables that are likely to influence ProM task performance in a number of ways, and, to a lesser extent, on manipulations of variables relevant specifically to the unique stages of ProM retrieval. Most previous research has been motivated by general theoretical accounts of ProM task performance, such as the preparatory attention and memory processes theory, proposed by Smith (2003), and the multiprocess framework, proposed by McDaniel and Einstein (2000). Consistent with these accounts, researchers have focused on global variables such as task importance (e.g., Brandimonte et al., 2010; Kliegel, Martin, McDaniel & Einstein, 2001; Kvavilashvili, 1987; Walter & Meier, 2016), or on the attentional demands of the ongoing activity (e.g., Hicks, Cook, & Marsh, 37  2005; Marsh, Hancock, & Hicks, 2002) on ProM task performance. These kinds of manipulations may influence performance on a ProM task in a number of possible ways. For example, they might affect the way the ProM task is initially encoded, or in connection with retrieval, they might affect cue noticing, cue-plan relevance recognition, or possibly response switching. By contrast to this type of work, however, there is also research involving the manipulation of variables more directly relevant to the distinct retrieval stages identified in the preceding paragraph. For example, with respect to cue noticing, researchers have manipulated the perceptual salience of the cue, such as its size (Graf, Uttl, & Dixon, 2002), spatial position (Cohen et al., 2003; Trawley et al., 2014), colour (Brandimonte & Passolunghi, 1994) or loudness (Uttl, 2006). Researchers also have varied the psychological salience of the cue, such as its degree of familiarity (e.g., Einstein & McDaniel, 1990), or its position within another stimulus (e.g., Kliegel et al., 2008). In investigations pertinent to cue-plan relevance recognition, researchers have manipulated the degree to which the cue and plan share a pre-existing semantic (e.g., Cohen, West, & Craik, 2001; Loft & Yeo, 2007) or episodic association (Pereira et al., 2012), or they manipulated the availability of plan relevant contextual information (e.g., Cohen et al., 2001; Meier, Zimmerman, & Perrig, 2006).  The processes unique to response switching in ProM tasks have been examined in seminal research by Meier and Rey-Mermet (2012; 2017). Meier and Rey-Mermet (2012; 2017) recognized the connection between task-switching more generally and the kind of response switching required for ProM task retrieval, and they embedded a ProM task into a task-switching paradigm. Participants were required to complete three tasks which collectively served as the ongoing activity, a parity (odd/even) judgment task for which triplicate numbers were displayed (e.g., 777), a colour decision task (red/blue) with symbols displayed in either red or blue (e.g., $$$), and a case decision task (upper/lower case) with triplicate consonants displayed in black (e.g., nnn). The ProM task required participants to press the spacebar if they came across any consonant-vowel-consonant triplicate (e.g., nen). The critical finding from this research 38  was that response times were significantly longer on trials immediately following a ProM task response. In the task-switching literature, the slowing of responses on trials that require response switching is known as a switch cost. Switch costs are thought to occur in part because of the attentional resources and time required for configuring a new task-set (Monsell, 2003). Meier and Rey-Mermet’s (2012, 2017) interpreted their findings consistently with the task-switching literature, suggesting that executive control processes are necessary for the configuration of the ProM task response. Once the ProM task-set has been configured and the response made, the participant must switch back to the ongoing activity. However, reconfiguration and execution of the ongoing activity response is resource demanding, and harnessing the required resources takes time and thus responding is slowed. The research we report here on response switching also was inspired by theoretical insights from the task-switching literature, and it used a variation of the method pioneered by Meier and Rey-Mermet (2012; 2017). We assume that three mechanisms from the task-switching literature are relevant to response switching in ProM tasks. The first mechanism is task-set inhibition. A task-set is defined as a unique configuration of mental resources that is used to guide responding (see Monsell, 2003). Task-switching theorists assume that successful switching from one task to another necessitates that the task-set relevant on the preceding trial is inhibited (Mayr & Keele, 2000). In the context of ProM, successful execution of the ProM task necessitates inhibition of the ongoing activity response. The second mechanism, also identified by Meier and Rey-Mermet (2012), is task-set configuration, defined as the configuration of mental resources, or processing components, required for executing a given task (Monsell, 2003). The third task-switching mechanism relevant to ProM response switching is task-set inertia, defined as the tendency for a task-set to persist over time and to proactively interfere with configuring a new task-set (Wylie & Allport, 2000). Task-set inertia might be particularly relevant to ProM tasks. It is assumed that more executive control processes are required for configuring a less practiced response, like the ProM task 39  response (Waszak, Hommel, & Allport, 2002). Because executive control processes are required for configuring less practiced task-sets, it is more difficult to disengage from them and, as a consequence, they are more likely to interfere with, or delay, the configuration (or reconfiguration) of a different response (see Allport et al., 1994).  In the present article, we report two experiments that were motivated by the goal of investigating the processes involved in ProM task response switching (Crease, Jamieson & Graf, 2014; Crease Lark, Jamieson & Graf, 2015). The specific objectives were to examine whether the three mechanisms highlighted in the task-switching literature, task-set inhibition, task-set configuration, and task-set inertia, are engaged for switching in ProM tasks. Both experiments had two phases. The first phase was an encoding phase, during which a list of words was provided for study, and participants were required to make a semantic decision about each word (e.g., a familiarity decision). The second phase was a recognition a test phase, during which the studied words were presented together with non-studied words, and for each of them, participants’ task was to make an old/new decision. Each experiment also included a ProM task. For the ProM task, participants had to press a unique key (e.g., the 5-key) in response to words representing kinds of birds. This task was inserted into the encoding phase of Experiment 1, whereas it was included in the retrieval phase of Experiment 2.  We asked three questions about the mechanisms involved in response switching. First, can we find evidence of response slowing or memory processing deficits as a consequence of response switching in ProM tasks? Second, if effects such as response slowing or poor recognition memory performance are observed, how long do such effects last? Third, are variables such as the type of processing required for the ongoing activity likely to influence those effects?  Consistent with the three mechanisms described in the preceding paragraphs, and consistent with the findings and theorizing by Meier and Rey-Mermet (2012; 2017), we anticipated the following outcomes. 40  First, on the assumption that the task-set used for guiding ongoing activity responding requires inhibition in order for a ProM task response to occur, we anticipated that the processing of the words presented immediately before the ProM cue might be disrupted, and we herein refer to this disruption in processing as a retroactive effect (Mayr & Keele, 2000). Second, consistent with the assumption that additional time is required for reconfiguring the task-set required for making an ongoing activity response immediately after making a ProM task response (Meier & Rey-Mermet, 2012), we anticipated that the processing of ongoing activity words following each ProM cue would be slowed, and perhaps less elaborately encoded. This prediction is also consistent with the speculations about task-set inertia. Because ProM task responses are rare, the task-set configuration for such responses is heavily dependent on executive control processes, and this kind of task-set is more likely to persist over time and to interfere with the configuration of an alternative task-set (Allport et al., 1994).  Experiment 1 Experiment 1 had an encoding phase, which required participants to make semantic decisions about a long list of unrelated words, followed by a retrieval phase during which participants had to make old/new decisions about words from the encoding phase and an equal number of non-studied words. The ProM task was inserted into the encoding phase of the experiment, and it required participants to press the 5-key on the computer keyboard when they encountered a word that referred to a type of bird.  Method Both Experiments 1 and 2 used the same basic method. For this reason, the method is described in detail here. The method section for Experiment 2 will identify differences between Experiments 1 and 2.  Participants. Given the paucity of research on this topic and the novel paradigm used in these experiments, the total number of subjects required for this (and subsequent) experiments was determined via a power analysis. In order to estimate the effect sizes, we carried out the power analysis after 20 41  participants had completed the experiment. We determined mean differences on encoding phase question response time and recognition memory task performance for words surrounding and not surrounding ProM cues. The power analysis indicated that approximately 80 participants would be required to detect comparable differences. The participants were 86 undergraduate student volunteers. They received partial course credit for their participation in the study, which was approved by the ethics review board (see Preface). Each participant provided informed written consent at the beginning of the experiment. Materials. We required a total of 292 words. Of this total, 280 were high frequency words drawn from the Toronto Word Pool (Friendly et al., 1982), each with between four to eight letters, and being a noun, verb or adjective. An additional 12 words were the names of familiar birds: dove, eagle, falcon, goose, hawk, parrot, pigeon, peacock, robin, seagull, sparrow, and swan. The words were randomly divided to create two sets (A and B), each of which consisted of 140 words from the Toronto Word Pool, plus six bird words. For each participant, the words from one set were presented in the encoding phase of the experiment (140 words) together with six of the bird words which served as ProM task cues. The remaining words were used as recognition test distractors. Procedure. Participants were run in groups of up to eight in a quiet room equipped with eight desktop computers. All parts of the experiment were controlled by means of the E-Prime experimental suite (Schneider, Eschman & Zuccoloto, 2002). The encoding/ProM testing phase had 146 trials. On each trial, participants were shown a word, randomly selected from the encoding list (either set A or B; sets were counterbalanced across participants), together with a question, randomly selected from a set of three questions: Does this word refer to something positive or negative? Does this word refer to something you like or dislike? Does this word refer to something familiar or unfamiliar? The questions were displayed at the top of the screen, with the response options displayed at the bottom of the screen. Each word was displayed centred on the computer 42  monitor, in all upper case letters, and below the encoding questions. The questions, words and response options were displayed against a grey background, in 18-point navy-blue Courier font. Each remained on the monitor for a minimum of two seconds. If a participant responded within the two second minimum, the word remained on the screen until two seconds had elapsed. If a participant required more time, the word remained on the screen until a response was entered. There was a 500 ms inter-stimulus blank display between trials. The encoding phase was arranged into seven blocks, presented in random order, each consisting of 20 trials in which a common word was displayed, and interspersed between the blocks was a trial in which we presented a randomly selected bird-word (i.e., a ProM task cue). The ProM task cue words were displayed in exactly the same manner as the common encoding phase words, except that the cue words were always displayed together with the same question: Does this word refer to something familiar or unfamiliar? Immediately prior to the first encoding phase block, participants received instructions about the ProM task and about the encoding task. For the ProM task, participants were told: If ever in the course of this experiment, you see a word that refers to a bird, ignore all other instruction and immediately press the 5-key on the computer keyboard. To verify understanding of these instructions, we asked participants to tell us what they would do if the word canary appeared on the computer monitor. For the concurrently ongoing encoding task, participants were informed that the computer would show a series of words, one at a time, each together with a simple question, and that their task was to answer each question honestly. They were shown the three encoding questions, and informed to answer each of them using the 1-key for responding positive, like and familiar, respectively, to the relevant encoding question, and using the 0-key for responding negative, dislike and unfamiliar. All task instructions were provided both in spoken and written form, and participants were encouraged to ask questions about them. 43  Immediately after the encoding phase, participants were informed that the ProM task was over; more specifically, they were told that you no longer have to worry about pressing the 5-key in response to bird words. Then they received instructions for the recognition memory test which consisted of 250 trials. On each trial, a word was displayed in the center of the computer monitor, against a grey background, in 18-point green Courier font. Each word remained on display until a response was entered, and was then replaced with a 500 ms inter-stimulus blank display. On each trial, the participants’ task was to make a yes/no decision about the current word based on whether or not that word had been displayed in the encoding phase of the experiment. Participants were instructed to press the y-key on the computer keyboard to indicate yes the word had been presented in the encoding phase, and to press the n-key to indicate no the word did not appear during the encoding phase. Participants were reminded to respond as accuracy as possible.  The recognition memory test included 125 of the words from encoding phase of the experiment: the 6 bird-words used as ProM task cues, the three words immediately preceding and immediately succeeding each cue word (i.e., a total of 36 words) in the study list, plus a random selection of 83 of the remaining common words presented in Phase 1. The test also included 125 new words (i.e., words not presented in Phase 1), 119 of which were randomly selected from the non-studied common words, and 6 were the remaining non-studied bird words. For most trials (i.e., all but 24 trials), one of the words from the 250-word test list was selected randomly and without replacement.  The recognition test included six special blocks, each with four trials; we used these blocks for investigating how recognition test performance is affected by the test phase occurrence of a bird-word (i.e., the type of word used in Phase 1 as ProM task cues). For each of these special blocks, a bird-word was displayed on the first trial. This word was a ProM task cue from Phase 1 for three blocks and a non-presented potential ProM cue for the remaining three blocks (i.e., a bird-word that was not presented during 44  encoding). For each special block, we displayed an old word from Phase 1 of the experiment on Trials 2, 3 and 4; these latter words were selected from words that had not surrounded the cue in Phase 1. By this arrangement of trials, we were able to assess whether the occurrence of bird-words during the encoding phase influenced recognition decisions about studied words even after the ProM task had been terminated. The six special blocks were randomly interspersed among the recognition test trials. Results The main dependent variables were accuracy and speed on the ProM task, speed on the encoding phase task, as well as accuracy and speed on the old/new recognition memory test. All data were screened for outliers and only one was found. On the recognition test, one participant had a median response time of 338 ms, more than three standard deviations below the group mean of 964 ms, most likely because of failing to follow task instructions. None of the data from this participant were included in any of the analyses reported below. We used the 95% confidence interval for making decisions about differences between the means. In a preliminary analysis, we checked ProM and recognition memory task performance for floor and/or ceiling effects. On the ProM task, participants responded correctly to 75.9% (95% CI: 69.5% to 82.3%) of the cues, demonstrating that performance was free of both floor and ceiling effects. An analysis-by-subjects showed that six participants did not make any correct ProM task responses. The mean amount of time required for making correct responses to ProM cues was 1736 ms (95% CI: 1563 ms to 1909 ms), and such responses were made significantly faster than responses to the encoding task questions asked about non-cues, which averaged 2049 ms (95% CI: 1882 ms to 2215 ms). On the old/new recognition test, participants achieved a hit rate of 74.0% (95% CI: 71.4% to 76.7%) and a false alarm rate of 20.8% (95% CI: 16.9% to 24.3%). A signal detection analysis showed this performance to be well above the guessing level, d’ = 1.65 (95% CI: 1.51 to 1.79). 45  Encoding Phase. The present study was designed to find out whether the presentation of, and/or responding to, a ProM cue influences the processing of words displayed nearby to it. To address this issue, a first analysis examined how fast participants responded to the encoding phase questions for the three words displayed immediately after each cue. Figure 2.1 shows the mean of participants’ median response time on these words (herein called the cue surround words); note, the means are only from words following cues that were successfully used to make a ProM response. The figure also shows the mean response time to the actual cue words (Note: For the cues, Figure 2.1 shows only the time for responding to the encoding questions; that is, it does not include the time taken to make a ProM response to the cue), and it shows the average amount of time required for responding to all other (i.e., non-cue surround) words. The means and confidence intervals reveal that roughly the same amount of time was needed for responding to all words; however, a significantly greater amount of time was required for responding to the word which immediately followed each cue (M = 2553 ms, 95% CI: 2266 ms to 2839 ms)2. A separate analysis showed that the time required for responding to the three words preceding each cue was not different from the time required to respond to the non-surround words. Finally, Figure 2.1 also highlights that compared to non-surround words, responding was significantly faster to missed cues (the red bar), that is, to bird words which failed to elicit a ProM task response (M = 1681 ms, 95% CI: 1494 ms to 1868 ms).3                                                      2 Responding to words which immediately followed each cue when it was not correctly recognized as ProM task relevant was not significantly different from responding to non-surround words (2040 ms; 95% CI: 1744 ms to 2336 ms).  3 Fast responding to missed cues may be due to the encoding question that was paired with each cue, as well as due to the concrete nature of the cue words. For cues, the encoding question always was “Does the word refer to something familiar or unfamiliar?” The average time for responding to this question (for all non-surround words) was 1970 ms (95% CI: 1815 ms to 2125 ms). By contrast, the average time for responding to “Does the word refer to something positive or negative” was 2110 ms (95% CI: 1929 ms to 2293 ms) and the average time for responding to “Does the word refer to something you like or dislike” was 2079 ms (95% CI: 1902 ms to 2256 ms). 46   Figure 2.1 Mean of participants’ median response time (in ms) to encoding phase cue words (red bar), cue-surround words (dark blue bars), and non-surround words (light blue bar). The RT’s for cue-surround words are only for words following cues that elicited a planned task response. The RT’s for cue words are only for those trials on which the ongoing activity response was provided for the cue. Error bars represent the 95% confidence intervals.  Recognition Test Phase. We also examined recognition test performance for specific influences due to the occurrence of bird-words that were used as cues in the preceding encoding phase, as well as for influences due to such words that could have been used as cues (i.e., bird-words that had not been presented during the encoding phase). To distinguish between these two types, we herein use the labels actual cues and potential cues, respectively. An initial examination showed that recognition performance was superior for correctly recognized actual cues (M = 81.5%, 95% CI: 77.4% to 86.1%) and potential cues (M = 82.9%, 95% CI: 78.7% to 87.1%) compared to non-surround non-cue old words (M = 74.9%, 95% CI: 72.2% to 77.7%), test phase 500750100012501500175020002250250027503000"Missed" Cues +1 +2 +3 Non-SurroundEncoding Question Response Time (ms)Word Type/Position47  cue surrounds (i.e., the three old words which followed actual cues in the test phase only) (M = 76.1%, 95 CI: 72.6% to 79.7%), and new words that were not potential cues and did not surround an actual or potential cue in the test phase (M = 79.1%, 95% CI: 75.4% to 82.7%).  Recognition was lower on encoding phase cue surrounds (i.e., on the three old words which followed each cue in the encoding phase only) (M = 68.8%, 95% CI: 65.3% to 72.3%) compared to all other types of words. Recognition of the three old words that preceded each cue in the encoding phase was not different from recognition of non-cue surrounds.  A more detailed analysis focused on recognition hits as a function of specific word position relative to the cues. Figure 2.2 shows the recognition hit rate on the actual cue words which had been correctly identified as ProM cues in the encoding phase (red bar), and the hit rate on the three corresponding encoding phase cue-surrounds (dark blue bars). The means and confidence intervals show that correct recognition was significantly higher on the actual cues than on all cue surround words, with the exception of the 3rd word following each cue. However, the hit rate on the 1st word following each actual cue was significantly lower than the hit rate on all other words. Similarly, the hit rate on the 2nd word following each cue also was significantly lower4. Figure 2.2 also shows the recognition hit rate on test phase cue surround words (yellow bars), and it highlights that recognition on these old words was similar to recognition of non-cue surround old words (light-blue bar).                                                       4 Recognition test hit rate for words which immediately followed each cue that did not elicit a planned task response was not significantly different from non-surround words. Mean hit rate for cT+1 words = 68.3% (95% CI: 59.9% to 76.8%) and mean hit rate for cT+2 words = 68.2% (95% CI: 60.6% to 75.58%).  48  Figure 2.2. Mean recongition test performance (hits) for encoding phase cue words (red bar), encoding phase cue-surround words (dark blue bars), non-surround words (light blue bar) and test phase cue-surround words (yellow bars). The figure only includes data for those cue words and their corresponding cue-surround words that successfully elicited the planned task response. Error bars represent the 95% confidence intervals.  A further analysis, not separately shown in Figure 2.2, focused on the three old words that followed each potential cue in the test phase (i.e., bird words that had not been included in the encoding phase list). These words showed a recognition hit rate of 76.4% (95% CI: 72.4% to 80.5%), which is comparable to the hit rate for non-surround words. Figure 2.3 shows the amount of time required for making correct recognition decisions on different word types. The data reveal that roughly the same amount of time was required for correct responses to all 50%55%60%65%70%75%80%85%90%95%100%Recognition Test Performance (Hit Rate)Word Type/Position49  old words, except for the first old word that followed actual cues in the test phase (yellow bars). A similar though less pronounced slowing of responses also occurred for old words that followed studied cues only in the encoding phase (dark blue bars). A supplementary analysis showed the same pattern of slowing for old words which followed potential cues in the test phase, with means of 1317 ms (95% CI: 1084 ms to 1550 ms), 1057 ms (95% CI: 979ms to 1135 ms) and 989 ms (95% CI: 913 ms to 1065 ms), respectively, for the 1st, 2nd and 3rd of these words. Figure 2.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words, encoding phase non-surround words and test phase cue-surround words. Error bars represent the 95% confidence intervals. Discussion The goal of the present study was to understand the mechanisms involved in ProM task response switching, to determine whether task-set inhibition, task-set configuration and task-set inertia are likely to be involved in response switching in ProM. We asked the following questions: First, is there evidence that 500600700800900100011001200130014001500Recognition Test Response Time (ms)Word Type/Position50  the mechanisms listed above are relevant to response switching in ProM? Second, if such evidence does exist, are effects observed before (retroactively) and/or after (proactively) execution of the ProM task, and how long do such effects last? Third, if proactive or retroactive effects occur, are they the same or different if a cue does or does not elicit the planned response? The findings from Experiment 1 give new insights into these and related questions.  The finding that encoding phase responses were made more slowly on words displayed immediately after the ProM task cues is evidence consistent with task-set configuration, as this finding is similar to the switch costs reported in the task-switching literature (see Monsell, 2003). Task-switching experiments tend to use stimuli with multiple characteristics (e.g., vowels or consonants, in upper or lower case, and displayed in red or blue), and participants may be required to make vowel/consonant decisions for letters in red, versus case decisions for letters in blue. Switch costs are assumed to reflect a number of executive control processes, such as task-set configuration for the new response (Cohen et al., 1990; Mayr & Keele, 2000; Monsell et al., 2003; Rogers & Monsell, 1995). A similar set of executive control processes is assumed to be required for making a ProM task response to a cue which occurs in the course of an ongoing activity (Graf, 2005; Meier & Rey-Mermet, 2012).  The cost associated with switching from a ProM task response on a cue trial (cT) to an ongoing activity response on the next trial (cT+1; the first surround word) is highlighted by Figure 2.1. The results show that switching only affected responding to the first word that followed the cue. With the procedure used in Experiment 1, a new trial occurred approximately every 2.5 seconds, and thus, it is possible that this amount of time is sufficient for dissipating the costs arising from the need for task switching. However, it is also possible that time is not a critical factor, that our time estimate is constrained by the procedure of Experiment 1, and more specifically, that switch costs are dissipated by the need for reconfiguring the task-set required for the ongoing activity on cT+1. This latter possibility is consistent with task-switching 51  research showing that even when the response-stimulus interval changed from 1.2 s to .6 s, slowing always occurred only on the first trial of the new task (Rogers & Monsell, 1995). Research by Meier and Rey-Mermet (2012; 2017) also showed that costs related to executing a ProM task are short-lived.  The need for response switching affected not only the time required for making ongoing activity responses, but also the manner of processing words displayed immediately after each ProM cue. This claim is supported by the results in Figure 2.2, by the recognition impairment on the words from trials cT+1 and cT+2. This recognition impairment suggests that the cT+1 and cT+2 words were processed more superficially than the non-surround words, most likely because resources essential for memory encoding were not available. We speculate that the same pool of resources may be used for both task-switching and memory encoding, and that as much as 6 seconds or two ongoing trials may be required for the full reallocation of processing resources from the ProM task to the ongoing activity.  In order to explain the different time courses of effects produced by task-switching, which is highlighted by the results in Figures 2.1 vs 2.2, it may be necessary to postulate two mechanisms. The first mechanism appears to be fast and engaged for shifting attention from one to the next task. This mechanism might be under exogenous control (Rogers & Monsell, 1995), potentially activated by stimulus properties such as visual distinctiveness or novelty (Meier & Rey-Mermet, 2012). A second mechanism may be critical for reconfiguring the task-set required for the ongoing activity response. In light of the recognition data in Figure 2.2, we speculate that this second mechanism is relatively slow and longer lasting, is concerned with allocating processing resources, suggesting that the resources required for encoding were not fully available until the third word that followed each ProM cue.  Encoding phase response times and recognition test decisions were affected only when the planned task was successfully executed in response to a ProM cue. As reported, 24.1% of the bird words were not identified as ProM cues; participants responded to these words in the same manner as to all other 52  ongoing activity stimuli. For these missed ProM task cues, there was no evidence of slower processing or a recognition memory test impairment on the words presented on the trial(s) that followed each cue. In view of this finding, we speculate that the mechanisms postulated in the preceding paragraphs are not engaged automatically by the occurrence of ProM cues; instead, these mechanisms seem to be initiated only when a cue is consciously recognized as relevant to a planned task. The recognition test data support the claim that bird words not identified as ProM task cues were processed in the same manner as all other ongoing task words. The hit rate for missed cues was 77.5% (95% CI: 70.4% to 84.6%), and was not significantly different from the hit rate on non-cue surround words (74.9%). By contrast, the hit rate for successfully identified cues was 81.5% (95% CI: 77.2% to 85.9%). The elevated recognition of successful cues is likely due to their distinctive use in the encoding phase of the experiment. We did not find any evidence that the execution of a planned task retroactively interfered with ongoing activity processing. The theoretical implications of the lack of retroactive effects of ProM task execution are addressed further in the general discussion.  A final outcome of Experiment 1 is the slowing of recognition decisions on the first word displayed immediately after each ProM task cue. This slowing was more pronounced on test phase surrounds (see yellow bars in Figure 2.3) than on surrounds from the encoding phase (see dark blue bars in Figure 2.3). It is possible that this slowing is a consequence of encountering a test word (i.e., a bird word) that stood out in some manner (cf. Whittlesea & Williams, 2001), by virtue of having functioned as a ProM task cue in the preceding encoding phase of the experiment. Previous research has shown that for some time after a ProM task has been completed, previously task-relevant cues continue to trigger ProM task relevant cognitions (see Meier & Rey-Mermet, 2017; Scullin, Bugg, & McDaniel, 2012; Walser, Fischer, & Goschke, 2012, 2014). In the present experiment, participants received explicit instructions that cancelled the ProM task 53  prior to the test phase. The finding of slowed recognition decisions despite these instructions suggests that ProM task relevant cognitions occurred involuntarily (i.e., in the absence of conscious intentions). Experiment 2 The goal of Experiment 2 was to determine if the same pattern of results observed in Experiment 1 would occur when a different combination of processes is required for the ongoing activity and the ProM task. The results of Experiment 1 provided potential answers to the questions related to how response switching affects ongoing activity processing, however, they must be interpreted cautiously. It is possible that the results we found occur only with the specific combination of processes required for the ongoing activity and ProM task used in Experiment 1 (i.e., only when a ProM task uses categorical cues and when this task is embedded in an ongoing activity that requires making semantic decisions about a series of words). Experiment 2 was designed to address this possibility by embedding the same planned task into an ongoing activity that requires a different kind of processing.   There is a wealth of ProM research examining the specific combination of processing required for an ongoing activity and a ProM task, and how this combination influences overall ProM task performance. One example of this research has focused on differences in ProM task performance with what are frequently called focal versus non-focal cues (McDaniel & Einstein, 2000). Focal cues are described as cues which require processing that is consistent with the processing required for the ongoing activity (e.g., semantic processing is required for both cues and the ongoing activity) (e.g., Einstein et al., 2005). Non-focal cues require a different type of processing (e.g., perceptual processing is required for the cue, while semantic processing is required for the ongoing activity). The typical findings are that focal cues facilitate ProM task performance (e.g., Einstein & McDaniel, 2005; Meiser & Shult, 2008; Meier & Graf, 2000; Scullin, McDaniel, Shelton, & Lee, 2010). However, the majority of existing research was not specifically designed to focus on response switching, and the findings might be attributable to other stages that are 54  involved in carrying out a planned task. By way of example, Einstein et al. (2005) reported an experiment where the ProM task required participants to respond differently to a specific syllable (e.g., tor) or a specific word (e.g., tortoise). The ongoing activity required participants to categorize words as belonging to a specific semantic category. In this research, the ProM cues that were full words shared more processing overlap with the ongoing activity, likely facilitating cue noticing. In an example of a processing overlap manipulation likely to influence recognition of cue-plan relevance, McDaniel et al. (1998) used homographs as ProM cues (e.g., chest). Prior to the ProM instructions, the homographs were contextualized with a specific definition (i.e., chest as a body part). During the ongoing activity, the ProM cues were contextualized as consistent or as inconsistent with the earlier definition (i.e., chest as storage equipment). If the representation of the cue word chest contextualized as a body part was associated with the representation of the plan when the ProM task instructions were provided, it is likely that presenting the word chest contextualized in a similar way facilitated recognition of cue-plan relevance.  To our knowledge, Meier and Rey-Mermet (2017) are the only researchers who have directly examined how response switching is affected by manipulations of the overlap in the processing required for an ongoing activity and a ProM task. They embedded a ProM task in a task-switching paradigm, and used a processing overlap manipulation most likely relevant to the cue noticing stage of retrieval. The ongoing activity alternated between three tasks, one that required participants to make odd/even decisions about digits (e.g., 777), another that required participants to make colour decisions about symbols (e.g., %%%), and a final task that required participants to make case decisions about letters (e.g., nnn). The cue signaling ProM task retrieval required processing consistent with one or more of these ongoing activity tasks. For example, in the experiment that included the greatest degree of overlap, ProM cues were triplicate letters, overlapping with the case decision task, and those letters were displayed in either blue or red, overlapping with the colour decision task. Meier and Rey-Mermet (2017) found that when the type of 55  processing required for identifying a ProM cue and for responding to the ongoing activities increased, response time switch costs were larger and longer lasting.  To complement investigations on interactions between the types of processing required for ProM tasks and ongoing activities, Experiment 2 examined how such overlaps affect processing on words immediately following the cues. However, rather than focusing on the overlap in processing required for the cues and ongoing activity stimuli themselves (as in Meier & Rey-Mermet, 2017), we sought to examine the influence of the overlap in memory processing requirements. To examine the effects of this manipulation, Experiment 2 included a ProM task only in the retrieval phase of a recognition memory test. We assumed that by this arrangement the overlap in processing required for the ProM cue and the ongoing activity would be greater than was the case for Experiment 1 because both the ProM task and the recognition memory test were dependent on episodic memory retrieval processes. By inserting the ProM task into the retrieval phase of a memory experiment, Experiment 2 also allowed us to manipulate the episodic status of the cue (i.e., studied or unstudied). To our knowledge, no previous research has examined this kind of cue manipulation. We suspect that this manipulation is likely to influence either the cue noticing stage and/or the recognition of cue-plan relevance stage of ProM retrieval. In previous research, when ProM task cues and ongoing activity stimuli both require semantic memory processing, the ProM cues were considered focal and likely to facilitate cue noticing (see Smith, 2003). It is therefore reasonable to assume that both tasks requiring episodic memory processing would also facilitate cue noticing. Alternatively, ProM task cues that have been recently studied are likely to have a strong association to the encoding phase context that does not exist for new (non-studied) cues. This association to the encoding phase is directly relevant to executing the ongoing activity response (i.e., the recognition memory test response), and therefore might interfere with the association between the cue and the planned task, thereby interfering with recognition of cue-plan relevance.  56  In Experiment 2, we aimed to address three specific questions. First, how does the episodic memory status of the cue influence ProM task performance and proactive effects on the ongoing activity? Based on the theoretical assumptions outlined in the preceding paragraph, we predicted that participants would correctly respond to unstudied ProM task cues more frequently. Second, we asked how an overlap in memory processing requirements between the ongoing activity and the ProM task would influence the proactive effects observed in Experiment 1, that is, the effects we attributed to task-set configuration and task-set inertia. Third, we investigated whether the processing overlap manipulation would influence response time and recognition performance in the same way. Two opposing predictions regarding the influence of processing overlap on the effects of response switching were suggested by previous ProM research. The first prediction is consistent with findings from Meier and Rey-Mermet (2017): that processing overlaps that facilitate cue noticing increases response switching costs, as manifested by longer-lasting and larger increases in response time following a planned task response. This prediction is also consistent with task-switching research, which suggests that when two tasks share overlaps in stimulus-sets (i.e., the stimulus properties relevant to different responses) and the need for a response switch arises, it is more difficult to inhibit the irrelevant response (see Gade & Koch, 2007). The more difficult a response is to inhibit, the more difficult reconfiguration of that task-set becomes once a switch is necessary, which in turn increases switch costs (see Kiesel et al., 2010). The second prediction is consistent with the theoretical assumptions of McDaniel and Einstein’s multiprocess framework (2000). According to this framework, an overlap in the processing required for the ongoing activity and the ProM task facilitates automatic retrieval of the plan (i.e., recognition of cue-plan relevance). Automatic retrieval processes are less likely to be demanding of attentional resources. As such, we might expect automatic retrieval of the ProM task response to reduce switch costs, as more attentional resources would be available for task-set configuration processes.  57  Method We used the same general method as for Experiment 1, except that the ProM task occurred as part of the retrieval phase rather than the encoding phase. In addition, we manipulated the episodic memory status of the ProM cues in Experiment 2. As in Experiment 1, the ProM task cues were bird words. However, half of these cues also were included in the study phase of Experiment 2, while the other half appeared for the first time only in the test phase. Participants. The participants were 78 undergraduate student volunteers. They received partial course credit for their participation in the study, which was approved by the ethics review board. Each participant provided informed written consent at the beginning of the experiment. Materials and Procedure. We used the same words as for Experiment 1; they were again divided into two sets (A and B), each of which consisted of 140 common words from the Toronto Word Pool and six bird words. For each participant, the words from one set were presented in the encoding phase of the experiment together with six of the bird words, with the remaining words (140 common words and six bird words) serving as recognition test distractors. In the encoding phase, each word was again displayed together with one of the three encoding questions from Experiment 1, and for both the encoding phase and the test phase, the words were ordered and displayed in the same manner as for Experiment 1. One critical difference between experiments is that for Experiment 2, we informed participants about the ProM task only after the encoding phase. Upon completing the encoding phase, we instructed participants that they would now complete a recognition memory test. For the prospective memory task, we used the same general instructions as for Experiment 1: If ever in the course of the recognition memory test, you see a word that refers to a type of bird, ignore all other instructions and immediately press the q-key on the computer keyboard. As in Experiment 1, these instructions were given in spoken and written 58  form. To verify understanding of the instructions, we asked participants to tell us what they would do if the word canary appeared on the computer monitor. During the test phase, we displayed 12 ProM cues. Six of these cues were words that had been included in the encoding phase, while six cues were new words displayed for the first time during the test phase. The recognition test procedure was the same as used for Experiment 1, except for the presentation of the ProM task cues and the cue-surround words, that is, the three words that preceded and followed each cue. From the pool of words used for the recognition test, we randomly selected 36 old words (i.e., words that had been in the encoding phase) and 36 new words (i.e., words not in the encoding phase). In the recognition test, three old ProM task cues were surrounded by six words (with 3 words preceding the cue and 3 words following the cue) randomly selected from the set of 36 old words. The other three old ProM task cues were surrounded by six words randomly selected from the set of 36 new words. We used the same method also for displaying and surrounding the new ProM task cues. By this pairing of cues and cue-surrounds, we were able to investigate the influence on ProM task performance due to the oldness/newness of cues, as well as possible interactions between these cues and the oldness/newness of cue-surrounds. During the recognition test, we displayed a ProM task cue after every 20th trial. Results The critical dependent variables were accuracy and speed on the ProM task, as well as accuracy and speed on the old/new recognition memory test. We also examined the speed on the encoding phase task, but given that the ProM task did not occur during the encoding phase, the response times from this phase are provided only for the sake of completeness and for potential comparison with the results from Experiment 1. We used the 95% confidence interval for making decisions about statistically significant differences between and among means. 59  All data were screened for outliers. In the encoding phase, one participant had a median response time of 4972 ms, more than four standard deviations above the group mean of 2011 ms, potentially because of difficulty understanding the task. None of the data from this participant were included in any of the analyses reported below. Encoding Phase. Participants responded equally fast to encoding questions paired with bird words (i.e., words to be used as ProM cues during the recognition test phase) (M = 1930 ms, 95% CI: 1878 ms to 2180 ms), with bird-word surrounds (i.e., the three words following each bird word) (M = 1998 ms, 95% CI: 1855 ms to 2140 ms), and with non-surround words (M = 1965 ms, 95% CI: 1841 ms to 2089 ms). Responding to the three words displayed immediately ahead of each potential ProM cue also was not different from responding to non-surround words. These finding are not surprising, but they validate the assumption that prior to being instructed about the ProM task, participants responded in the same manner to the potential ProM cues as to all other words. Recognition Test Phase. On the old/new recognition test, participants achieved an overall hit rate of 78.4% (95% CI: 75.9% to 81.0%) and a false alarm rate of 21.6% (95% CI: 17.9% to 25.4%). A signal detection analysis showed this performance to be significantly above the guessing level, d’ = 1.76 (95% CI: 1.57 to 1.94). ProM Task Performance. Participants correctly executed the ProM task in response to 71.9% (95% CI: 68.5% to 75.4%) of the cues. They responded correctly to a larger proportion of the new cues, that is, bird words which had not appeared in the encoding phase (M = 75.9%, 95% CI: 71.7% to 80.2%) compared to old cues, that is, bird words which had been included in the encoding phase (M = 67.9%, 95% CI: 63.3% to 72.5%). 60  A separate analysis revealed that participants required significantly less time to make old/new responses to cue words (M = 962 ms, 95% CI: 899 ms to 1024 ms) than ProM responses to cue words (M = 1240 ms, 95% CI: 1144 ms to 1336 ms). Recognition Test Performance. For cues that did not elicit a planned task response, participants had an 83.2% (95% CI: 76.3% to 90.2%) hit rate for previously studied cues, which was higher than the hit rate for all other studied words (M = 78.5%, 95% CI: 76.3% to 80.7%). The false alarm rate for cue words that were not previously studied was 28.3% (95% CI: 20.1% to 36.5%), and significantly higher than the false alarm rate for all other non-studied words (M = 21.1%, 95% CI: 18.1% to 24.0%).  A detailed analysis of recognition test performance examined hits and correct rejections with a focus on the words that surrounded the ProM task cues. Figure 2.4 shows recognition performance on the test phase surround words, which includes the three words displayed immediately after each cue word. The figure shows the hit rate for old surrounds (dark blue bars) and the correct rejection rate for new surrounds (light blue bars). The means show that for the surrounds, word position had no systematic influence on the hit rate, but it had a significant effect on the correct rejection rate. The average hit rate for surround words (M= 79.7%, 95% CI: 76.5% to 82.9%) was similar to the hit rate for non-surround words (M = 79.3%, 95% CI: 77.0% to 81.7%). However, the correct rejection rates for the cue surrounds show a different pattern. The correct rejection rate for the 1st, 2nd and 3rd word that followed each cue was, respectively, M = 68.2% (95% CI: 62.5% to 73.9%), M = 74.8% (95% CI: 69.9% to 79.8%) and M = 80.7% (95% CI: 75.7% to 85.8%). The correct rejection rate for all other non-surround words was 78.9% (95% CI: 75.8% to 81.2 %)5.                                                       5 An additional analysis focused on recognition test performance for those words immediately following cues that did not elicit a planned task response. There were no significant differences in performance for any of the word positions for hits (cT+1: M = 78.4%, 95% CI: 70.9% to 85.9%; cT+2: M = 78.8%, 95% CI: 71.1% to 86.5%,; cT+3: M = 81.3%, 95% CI: 73.4% to 89.3%) or for correct rejections (cT+1: M = 75.8%, 95% CI: 68.2% to 83.4%; cT+2: M = 76.0%, 95% CI: 67.7% to 84.3%,; cT+3: M = 78.8%, 95% CI: 71.5% to 86.1%). 61  We also examined recognition performance on the three words displayed immediately preceding each cue. Performance on these words did not vary across word position and was not significantly different from performance on the non-surround words.  Figure 2.4. Mean recognition test performance for test phase cue words, test phase cue-surround words and non-surround words. The figure highlights the hit rate for studied words (dark blue bars) and correct rejection rate for unstudied words (light blue bars). The cue data are only for cues which did not elicit a planned task response. The data for cue-surround words are only for those words following cues that successfully elicited the planned task response. Error bars represent the 95% confidence intervals. A final analysis focused on the mean of participants’ median response time required for making correct recognition decisions. Overall, making correct old/new responses to surround words (i.e. the three words following each cue) was slower (M =1181 ms, 95% CI: 1129 ms to 1233 ms) than responding to non-surround words (M =1047 ms, 95% CI: 1010 ms to 1083 ms). However, a more detailed analysis showed that this difference is due to just the first surround word which immediately followed each cue (see Figure 50%55%60%65%70%75%80%85%90%95%100%"Missed" Cues +1 +2 +3 Non-SurroundRecognition Test Performance Hit Rate and Correct RejectionsWord/Type PositionStudied WordsUnstudied Words62  2.5), with recognition decisions about this word being significantly slower than decisions about all other words.  In Figure 2.5, response time required for making correct recognition test decisions for hits and correct rejections have been combined. The data were combined because the analysis showed no significant differences in response time for making correct recognition decisions for studied and unstudied words in any of the word positions.    Figure 2.5. Mean of participants’ median response time (in ms) required for correct recognition test decisions to ProM task cue words (red bar), test phase cue surround words (dark blue bars), and all non-surround words (light blue bar). The RT’s for cue-surround words are only for words following cues that elicited a planned task response. The RT’s for cue words are only for those trials on which the ongoing activity response was provided for the cue. Error bars represent the 95% confidence intervals. Discussion The main purpose of Experiment 2 was to investigate whether ProM task responding has the same effects on the processing required for episodic recognition memory retrieval as found in Experiment 1, 400600800100012001400160018002000"Missed" Cues +1 +2 +3 Non-SurroundRecognition TestResponse Time (ms)Word Type/Position63  where the ProM task was embedded in the encoding phase of a retrospective memory task. We made the following predictions: First, we expected that the episodic status of the ProM cue (i.e., whether it was studied or non-studied) would influence performance on the ProM task. Specifically, we anticipated that unstudied cues would be easier to identify as plan relevant. Additionally, we made two predictions regarding the influence of the overlap in the processing required for the ProM task and the ongoing activity. One prediction was that the processing overlap might increase the effects due to response switching, particularly if our processing overlap manipulation is directly relevant to the cue noticing stage of ProM retrieval, consistent with the findings from Meier and Rey-Mermet (2017). Alternatively, if the manipulation is more pertinent to the recognition of cue relevance stage of ProM retrieval, we anticipated a reduction in the effects associated with response switching, consistent with McDaniel and Einstein’s (2000) multiprocess framework.  The finding of higher ProM task performance when the cues were new (non-studied) words rather than old (studied) words is novel, and it shows that the episodic memory status manipulation used in Experiment 2 produces a different effect than, for example, the category typicality manipulations used in previous studies (Cherry et al., 2001; Nowinski & Dismukes, 2005; Penningroth, 2005). When required to make a ProM task response upon encountering a member of a specified category (e.g., a kind of bird), performance is higher when a typical (e.g., sparrow) rather than an atypical (e.g., penguin) member is provided as a cue. This outcome is assumed to occur because the typical cues are more strongly associated with the category and are more likely or quicker to activate the link between the cue to the category (see Collins & Loftus, 1975), and thereby facilitate recognition of cue-plan relevance. By contrast, the episodic memory manipulation used in Experiment 2 was assumed to affect primarily the associations between a cue and its study-phase context, ensuring that these associations existed for old (studied) cues but not for new (non-studied) cues. Therefore, the critical difference between the old and new cues in 64  Experiment 2 is that the former have context-relevant associations that compete directly with the pre-existing associations between the cues and their ProM-task pertinent category membership. We assume that these novel contextual associations speeded the processes essential for responding to the ongoing recognition task. And because only one response is permitted per trial, faster recognition responding might have interfered with recognition of cue-plan relevance. This general interpretation is consistent with our finding that for cues, old/new responses were made substantially faster (M = 962 ms) than ProM task responses (M = 1240 ms).  The recognition test results further highlight the special status of cue versus non-cue words. The hit rate was higher for old cues (83.2%) which did not elicit a ProM task response compared to all other old words (78.5%), and similarly, the false alarm rate was higher for new cues (28.3%) which did not elicit a ProM task response compared to all other new words (21.1%). One explanation for this bias toward making old responses to cue words is that it is due to a category priming effect (Graf, Shimamura, & Squire, 1985). Specifically, we assume that the ProM task instructions to press a unique key whenever a bird word is displayed might have served to prime the semantic memory representations of all instances of the category. As a consequence of this priming, when a cue was displayed in the course of the old/new recognition test, it was more accessible and was perceived as more familiar (relative to non-primed words), and this increased familiarity may have been misinterpreted as evidence of ‘oldness’ (see Whittlesea & Williams, 1998, 2001; Yonelinas et al., 2010). The finding that ProM task execution had a selective effect on the recognition of new but not old words immediately following the cue is also novel. Figure 2.4 highlights this effect, showing that the correct rejection rate was depressed for the first two words with followed the cues (i.e., on cT+1 and cT+2), while the hit rate was not different on cT+1 and cT+2 compared to the hit rate on non-surround old words. The selective effect on correct rejections is difficult to interpret. One possibility is that when a response switch 65  was made from the ongoing activity to the ProM task, this switch also had an influence on the criteria for making recognition decisions. Specifically, it is possible that for a few trials immediately after a successful ProM task response, recognition decisions were made primarily based on a word’s familiarity, rather than on a combination of familiarity and associative information. This change in decision criteria may not have affected performance on studied words because the recent exposure of these words made them highly familiar and decisions about them might have been based on familiarity all along. By contrast, non-studied words are relatively low on familiarity, and decisions about them usually depend more extensively on familiarity as well as associative information. Follow up research is required to explore this account of the selective effect of ProM task responding on hits versus correct rejections.  A detailed analysis of recognition test performance showed that neither hits nor correct rejections were affected on words immediately following those cues which did not elicit a ProM task response. This outcome is consistent with the interpretation outlined in the preceding paragraph, and it suggests that a response switch from the ongoing activity to the ProM task is required for the temporary shift in participants’ recognition decision criterion that produced the post-cue depression in correct rejections.  We also found evidence to suggest that the execution of a planned task interfered with the processing required for recognition memory retrieval. The first piece of evidence comes from the fact that there was a large increase in response time on the word immediately following the cue (a 1.60 fold increase from the time required for responding to non-surround words). A two sample t-test indicated that this slowing was significantly larger than in Experiment 1 (a 1.24 fold increase for the time required to respond to the cT+1 word compared to responding to non-surround words), t(153) = 4.93, p < .001, d = .79. The larger proactive effects on response time may be a result of the increase in the memory processing overlap required for the ProM task and the ongoing activity (i.e., both tasks required episodic memory retrieval processes in Experiment 2). This finding contradicts some existing research, which suggests that focal 66  cues typically result in fewer costs to ongoing activity performance (e.g., Scullin et al., 2010). One explanation for this contradictory finding is that manipulations of the focal nature of the cue are more likely to influence to cue noticing stage of ProM retrieval. The findings are consistent, however, with response competition theories associated with the Stroop task, when increasing the degree of competition between responses results in greater Stroop task interference (e.g., Kane & Engel, 2003), with the task-switching literature on processing overlap (Gade & Koch, 2007; Kiesel et al., 2010), and with recent ProM research (Meier & Rey-Mermet, 2017). These latter manipulations are potentially more relevant to other stages of ProM retrieval, such as recognition of cue-plan relevance and/or response switching. We also assumed that the manipulation used in the present experiment is more likely to affect the recognition of cue-plan relevance stage of ProM retrieval.   Finally, while there was equivalent slowing of recognition response times for both studied and unstudied cT+1 words, recognition accuracy was only reduced for unstudied cT+1 (and to a lesser extent the cT+2) words. Following Experiment 1, we proposed that two mechanisms might be responsible for the proactive effects of response switching on response time and on recognition test accuracy. The former might be the result of an attentional shift mechanism. This attentional shift mechanism would be involved whenever a ProM cue successfully elicits a planned task response, therefore we would not expect differential effects for studied versus unstudied items. In contrast, we proposed that the mechanisms responsible for the effects on recognition test accuracy is due to reconfiguration of the ongoing activity task-set. If studied words were more easily accessible during the recognition test, they would be less likely to be interfered with by the processes involved in reconfiguration of the ongoing activity. However, unstudied words did not have the benefit of a recent memory representation to help counteract the deficits in processing resources, and the resources required for making a correct rejection were not available.  67  General Discussion The goal of Experiments 1 and 2 was to investigate the processes involved in the response switching required for successful ProM task execution. We were specifically interested in determining how response switching interfered with concurrent processing, what variables might influence the effects of response switching, and how long the effects of response switching might last. We were inspired by theoretical mechanisms identified in the task-switching literature, specifically by speculations about task-set inhibition, task-set configuration, and task-set inertia. To determine whether these mechanisms are relevant to response switching in ProM tasks, we looked for both proactive and retroactive effects due to response switching. Specifically, we examined response time and recognition accuracy for words immediately preceding (retroactive) and following (proactive) successful ProM task execution.  In Experiments 1 and 2, there were a number of unique findings associated with switching from an ongoing activity to a planned task and back again. In both Experiments 1 and 2, we found that participants were slower to respond to the first word immediately following ProM cues that elicited a planned task response. Additionally, in Experiment 1, recognition test accuracy was impaired for the two words immediately following ProM cues that elicited a planned task response. In Experiment 2, the influence of executing a planned task on recognition accuracy was limited to correct rejection rate of new words. In both experiments, the proactive effects only occurred when ProM cues elicited a planned task response. We did not find any evidence of retroactive effects of response switching in either experiment. Further, in Experiments 1 and 2, we found that ongoing activity responses to cue words were made more quickly than the planned task response to cue words.  There are two major findings from Experiments 1 and 2 that require further discussion. First, we found no evidence of retroactive effects of ProM task execution. Ongoing activity words immediately preceding the ProM task response were not responded to any more slowly than other words, nor was 68  recognition test performance affected for such words. Second, we did find evidence of proactive effects due to ProM task execution. Response time was significantly slower for the first word immediately following a ProM task response, and recognition memory performance was influenced for the first two words immediately following a ProM task response. The theoretical implications of these findings are discussed below.  The method used in Experiments 1 and 2 constrained our ability to observe all of the influences which are assumed to be required for response switching. Our experiments involve two distinct response switches, first, a switch from an ongoing activity response to a ProM task response, and second, a switch from a ProM task response back to an ongoing activity response. However, the ProM task response required in our experiments is both qualitatively different (pressing the 5-key) from the ongoing activity response (responding to a semantic judgment question), as well as quantitatively different from the ongoing activity response (the former is rare and thus expected to be slower). Because of these differences, the method of our experiments does not permit us to determine whether the switch from the ongoing activity to the ProM task has the same effect as switching from the ProM task to the ongoing activity. Given these methodological constraints, in Experiments 1 and 2 we focused on whether ongoing activity processing was influenced on words immediately preceding and following a successfully executed ProM tasks.  Ongoing activity words immediately preceding a successfully recognized ProM task cue did not show any increase in response time or decrease in recognition test performance in either Experiment 1 or 2. We predicted the possibility of a retroactive effect of response switching because of the task-set inhibition processes we assume to be necessary for making a switch from the ongoing activity to the ProM task (see Mayr & Keele, 2000). If the ongoing activity response must be inhibited, it is possible that this inhibition also interferes with the processing of the stimulus associated with that response which might manifest as poor recognition memory test performance on the words immediately preceding the ProM task 69  response. There was no evidence of an increase in response time or a decrease in recognition memory test performance for words immediately preceding ProM task responses in either Experiment 1 or 2, which suggests that response switching does not have a retroactive effect in the context of ProM task retrieval. This result may, however, be specific to the method used in the present research. The amount of time between words during the encoding phase was approximately three seconds, and it is possible that post-stimulus processing is already completed by the time the ProM task cue was displayed and inhibition of the ongoing activity task-set began. This possibility is consistent with the finding that a stimulus can be recognized with a high degree of accuracy even when displayed for only 1.5 seconds, particularly if it is paired with a semantic question (Craik & Rabinowitz, 1985).  Consistent with the proposal that task-set configuration is required for making an ongoing activity response immediately after making a ProM task response, as well as the notion that a greater degree of task-set inertia is associated with the ProM response task-set, we expected and found significant proactive effects following ProM task responses. In Experiment 1, we found that encoding phase response time increased for the first words immediately following an executed ProM task, and that recognition memory test performance decreased for the words displayed on the first two words which had followed a ProM task in the encoding phase. In Experiment 2, we replicated the finding of increased response times to the first word immediately following a ProM task response. We also found that ProM task execution specifically interfered with participants’ ability to correctly reject new words on the recognition memory test.  When switching from the ProM task back to the ongoing activity, the ongoing activity task-set must be reconfigured, and this process is likely to be both slower and more attentional resource demanding (Mayr & Kliegl, 2003) than on those words where the same response is repeated (Rogers & Monsell, 1995), and particularly difficult following a rarely occurring ProM task, consistent with assumptions about task-set inertia (Wylie & Allport, 2000). We found additional evidence to suggest task-set inertia is relevant to 70  response switching in ProM in Experiment 2. Participants were more likely to correctly execute the ProM task response when cues were not previously studied in the encoding phase. A potential explanation for this finding is that the ongoing activity response task-set was still active during the recognition memory test and interfered with the ability to configure the ProM task response (see Allport et al., 1994).  A possible alternative explanation for the proactive effects observed in Experiments 1 and 2 is related to strategic searching. The preparatory attention and memory processes (PAM) model proposed by Smith (2003) suggests that whenever a ProM task must be executed in the course of an ongoing activity, response time increases are observed for ongoing activity trials (e.g., Smith et al., 2007) . PAM (Smith, 2003) explains that these costs are due to the strategic search for relevant ProM task cues during the course of the ongoing activity, a process that is attentional resource demanding. This theory was not designed to specifically account for response time increases and recognition memory deficits for words immediately following a ProM task response. However, it is possible that the proactive effects from Experiment 1 and 2 arise because the execution of a ProM task briefly reminds participants about the need for strategic searching. It is conceivable that this reminding effect only lasts for one or two trials, as those were the only trials where we found response time increases and recognition test accuracy deficits in Experiment 1. However, it is more difficult to explain why strategic searching would affect the correct rejection rate of unstudied words but not the hit rate of studied words in Experiment 2. A more likely interpretation of the selective effects of ProM task execution on recognition memory test performance in Experiment 2 is described in the preceding paragraph, and consistent with the task-set inertia associated with response switching.  The critical findings from Experiments 1 and 2, namely the proactive effects that result from ProM task execution, support the theoretical assumptions that task-set configuration (and reconfiguration) and task-set inertia are mechanisms involved in response switching during ProM task retrieval. Critically, the 71  proactive effects reported in the preceding paragraphs only occurred when the ProM task response was made, which also supports the idea that these findings are driven by processes like task-set configuration and task-set inertia involved in response switching. These processes are attention-demanding and reduce the pool of attentional resources available for the processing of ongoing activity stimuli. However, it is important to consider an alternative possibility. ProM task responses are rare; participants are directing more attentional resources towards ProM cues, which in turn makes cues stand out as distinctive. It is possible that the same proactive effects could be a consequence of dealing with any stimulus which is different or distinct in some manner. Future research should investigate this possibility by examining the proactive effects (or lack thereof) that follow any kind of distinctive stimulus.   72  3. Chapter 3:  Proactive Influences Due to Perceptually Salient Stimuli Carrying out a prospective memory (ProM) task (e.g., stopping for groceries while driving home from work) upon the occurrence of an appropriate cue requires switching from making an ongoing activity response to making the planned task response. Experiments 1 and 2, and the previous research by Meier and Rey-Mermet (2012; 2017), have shown that this type of response switching has predictable consequences. For at least one trial after making a ProM task response, participants respond more slowly to stimuli presented for the ongoing activity, and they seem to process such stimuli in a more superficial manner. These influences on ongoing activity responses are consistent with two theoretical proposals from the task-switching literature, one focused on task-set configuration and the other on task-set inertia. Task-set configuration is the claim that the cognitive system is a serial processor, and at any one time is able to coordinate only the components required for making one particular response (see Monsell, 2003). Consequently, when required to make an ongoing activity response immediately after making a ProM task response, the cognitive system needs to reconfigure the task-set for the ongoing activity response, and this requirement slows responding for a few trials (Mayr & Keele, 2000). Task-set inertia is the claim that a task-set which is configured only rarely places more demand on executive control processes, and as a consequence, is likely to remain active for longer and to interfere more with the configuration of the task-set required for a new response (Wylie & Allport, 2000). The goal of Experiment 3 was to explore a plausible alternative account for the main findings from Experiments 1 and 2.  A defining property of episodic event-based ProM tasks is that a plan is not rehearsed during the retention interval, and thus needs to be brought back into conscious awareness upon the occurrence of a plan-relevant cue. For these reasons, the research method for ProM tasks ensures that participants’ attention is focused on an ongoing activity, and only a few plan relevant cues are presented in the course of this activity. However, by virtue of these methodological constraints, ProM task responses are relatively 73  rare compared to ongoing activity responses; they are also less well practiced and completed more slowly. It is possible that these factors collectively serve to distinguish ProM task responses from ongoing activity responses, and that the distinctiveness of ProM task responses, rather than the need for response switching, might explain the main findings from Experiments 1 and 2.  The possibility that the infrequency and distinctiveness of a ProM response embedded in a long series of ongoing activity trials might trigger proactive effects on the processing of stimuli presented for ongoing activity trials is consistent with findings from the next-in-line effect (Brenner, 1973). The next-in-line effect occurs in experiments with groups of participants. Each participant in the group reads one word from a given list, most often in a pre-determined order. Then participants are required to recollect the words contributed by all participants (Bond, 1985; Bond, Pitre, & van Leeuwen, 1991; Brenner, 1973). The next-in-line effect is the finding that memory performance suffers for the words immediately preceding and following the participant’s turn to read aloud (Brenner, 1973). Recall performance deficits for words prior to the participant’s turn are typically attributed to attention being paid to performance cues (i.e., because the participant’s turn is coming up), rather than to the stimuli themselves. Recall performance deficits for words following the participant’s turn are typically attributed to the time required to redirect attention towards the encoding task following the participant’s performance (see Brenner, 1973). In summary, the next-in-line effect is generally explained as an encoding problem, due to insufficient attention directed towards the to-be-remembered information (see Bond, 1985). It is possible then that the proactive effects observed in Experiments 1 and 2 can also be attributed to the increased attention directed towards ProM task cues rather than the processes I assume to be involved in response switching.  The explanation from the next-in-line effect must be applied to the current research with caution, because the research paradigm is not a perfect analogy for a ProM task. In the next-in-line paradigm, participants passively receive all of the stimuli except for the words they are required to read aloud (Bond, 74  1985; Brenner, 1973). For most ProM tasks, including those used in Experiments 1 and 2, participants are required to respond in some way to every stimulus. Additionally, in the next-in-line paradigm the distinctive trial occurs more frequently (e.g., 1 out of 11 trials; Brenner, 1973), whereas in Experiments 1 and 2, a ProM cue occurred after 20 ongoing activity trials. Another consideration is that the next-in-line paradigm typically proceeds in numerical order, so participants can anticipate which trial will be their turn. ProM task cues are typically not predictable, and they were not predictable in Experiments 1 and 2. Despite such methodological differences, it is possible that at least some of the proactive effects observed in Experiments 1 and 2 are a result of the same mechanisms that have been used to explain the next-in-line effect.  Research on the von Restorff effect is also potentially relevant to the findings from Experiments 1 and 2. The von Restorff effect, also known as the isolation effect or the distinctiveness effect (Hunt & Lamb, 2001), is the finding that the relative distinctiveness of a stimulus, as compared to the rest of the stimuli in a set, results in enhanced memory for that stimulus (Hunt, 1995; von Restorff, 1933). One way in which the von Restorff effect is studied is by varying the perceptual salience of a stimulus, meaning the stimulus is perceptually (e.g., visually) different relative to the other stimuli in a set (e.g., the salient stimulus is displayed in red and all other stimuli are displayed in black). Other manipulations vary the psychological salience of a stimulus relative to the rest of the stimuli in a set (e.g., an unfamiliar word in a series of familiar words). Hunt (1995) provides two potential explanations for the von Restorff effect. One possibility is that a salient stimulus receives a greater proportion of attentional resources at the time of encoding compared to the other stimuli in the set. An alternative explanation is that the distinctiveness of a stimulus (i.e., its difference from the rest of the items in a list) facilitates discriminative processes at retrieval, meaning that the participant is conscious of the fact that one item in the list was red, and then searches systematically for this item at the time of retrieval (see Hunt, 1995).  75  In the ProM literature, a large proportion of research has focused on manipulating the distinctiveness of the cue, such as its size (Graf, Uttl, & Dixon, 2002), spatial position (Cohen et al., 2003; Trawley et al., 2014), colour (Brandimonte & Passolunghi, 1994) or loudness (Utll, 2006). These manipulations typically result in superior performance on the ProM task. In some cases, these manipulations are thought to influence the likelihood that a cue is noticed (i.e., processed by the sensory system and perceived as a distinctive stimulus) (e.g., Brandimonte & Passolunghi, 1994). In other cases, researchers explain that distinctive ProM cues improve performance by facilitating recognition of cue-plan relevance (i.e., by increasing the likelihood that the plan is activated in connection with the cue) (e.g., McDaniel & Einstein, 2000). However, the existing research has not examined how cue distinctiveness might influence response switching processes. Further, and specifically relevant to the goal of the present experiment, existing research has not attempted to isolate the influence of any kind of distinctive stimulus on the processing of words immediately preceding and following that stimulus. The goal of Experiment 3 was to answer the question of whether exposure to perceptually distinctive stimuli is sufficient to demonstrate the kinds of proactive effects found in Experiments 1 and 2. In Experiment 3, I used the same basic method as for Experiment 1, except that in place of a ProM task cue, I included a word displayed in red font, amid a list of words displayed in dark-blue font, so that it would be perceptually distinctive to participants. Given findings from the next-in-line effect and the von Restorff effect, perceptually salient stimuli are likely to capture or shift participants’ attention. However, what is not clear is whether an attentional shift mechanism is capable of producing the same pattern of proactive effects for ongoing activity trials immediately following the salient stimuli, namely slower response times and poor recognition memory performance.  76  Experiment 3 Experiment 3 was modelled after Experiment 1. It had an encoding phase, during which participants were required to answer semantic questions about a list of unrelated words, and a retrieval phase during which participants had to make old/new judgments about a list of unrelated words, half of which were presented in the encoding phase and half of which were new, unstudied words. However, instead of including a ProM task, the encoding phase included words that were visually distinctive. The majority of words during the encoding phase were displayed in dark-blue font, with the visually distinctive words displayed in red font.  Method The method of Experiment 3 was nearly identical to the method used in Experiment 1, save for the inclusion of a ProM task. Below, I highlight the differences between the method for Experiment 3 and the method for Experiment 1.  Participants. The participants were 93 undergraduate student volunteers. They received partial course credit for their participation in the study, which was approved by the ethics review board. Each participant provided written informed consent at the beginning of the experiment.  Materials and Procedure. All materials were exactly the same as for Experiment 1. As in Experiment 1, I divided the words into two sets (A and B), each of which consisted of 140 common words from the Toronto Word Pool, plus six bird words. For each participant, the common words from one set were presented in the encoding phase of the experiment (140 words) together with six of the bird words, with the remaining words (140 common words and six bird words) serving as recognition test distractors.  In the encoding phase, each word was again displayed together with one of three binary questions (Does this word refer to something familiar or unfamiliar? Does this word refer to something positive or negative? or Does this word refer to something you like or dislike?). For both the encoding and the 77  recognition memory test phases, the words were ordered and displayed as for Experiment 1, with one critical exception. The bird words that served as ProM task cues in Experiment 1 now served as what will herein be referred to as von Restorff words. In order to make these words perceptually salient in the encoding phase of the experiment, I displayed them in 18-point red Courier font, whereas all other words were displayed in 18-point navy blue Courier font. In the test phase, all words were displayed in 18-point green Courier font. As in Experiment 1, each word was displayed for a minimum of two seconds. If a participant responded within the two second minimum, the word remained on the screen until two seconds had elapsed. If a participant required more time, the word remained on the screen until a response was entered. There was a 500 ms inter-stimulus blank display between trials. The instructions were also the same as for Experiment 1, except that there were no ProM task instructions. Instead, participants were informed that I wished to find out whether a word’s appearance would affect how that word is processed. Participants were given the following instructions regarding von Restorff words: If ever in the course of the experiment, you see a word displayed in a different colour, simply respond to the questions as you would for any other word. The recognition test instructions and procedure were identical to those used in Experiment 1. Following the recognition memory test, participants were given a brief post-experiment questionnaire, which asked them to identify if they noticed anything about the words that were presented in red font during the encoding phase. I asked this question to determine whether participants were aware that all of the words displayed in red font represented kinds of birds, as this might have an implication on their recognition memory test performance for both studied and unstudied bird words.  Results The main dependent variables were the time required for responding to encoding phase questions, as well as the performance and speed on the old/new recognition memory test. The data were screened for 78  outliers and none were found. The results showed that participants were significantly slower in responding to encoding questions paired with von Restorff words (M = 2323 ms, 95% CI: 2112 ms to 2534 ms) compared to all other non-surround words (M = 2018 ms, 95% CI: 1858 ms to 2179 ms). On the old/new recognition test, participants achieved an overall hit rate of 74.2% (95% CI: 71.5% to 77.0%) and a false alarm rate of 20.7% (95% CI: 17.6% to 23.8%). A signal detection analysis showed this performance to be significantly above the guessing level, d’ = 1.67, (95% CI: 1.53 to 1.81). Encoding Phase. To find out whether the display of, and/or responding to, a von Restorff word influenced the processing of words that preceded or followed each such word in the encoding phase, I examined how quickly participants responded to the encoding phase questions. Figure 3.1 shows the mean of participants’ median response time on the three words displayed immediately after each von Restorff word (herein defined as von Restorff surrounds), the mean response time to the actual von Restorff words, and the mean response time for all other non-surround words. The means and confidence intervals reveal that responding to von Restorff words was significantly slower (M = 2323 ms, 95% CI: 2112 ms to 2534 ms) than responding all non-surround words (M = 2018 ms, 95% CI: 1858 ms to 2179 ms). Responding to the first trial immediately following the von Restorff words (vrT+1) was not significantly slower (M = 2573 ms, 95% CI: 2346 ms to 2800 ms) than responding to the von Restorff words themselves. Responding to the vrT+1 was significantly slower than responding to the vrT+2 (M = 2095 ms, 95% CI: 1918 ms to 2272 ms) and vrT+3 words (M = 2109 ms, 95% CI: 1919 ms to 2299 ms), and all non-surround words. There were no significant differences in response time for the three words preceding each von Restorff word compared all non-surround words (vrT-1: M = 1981 ms, 95% CI: 1803 ms to 2159 ms; vrT-2: M = 2077 ms, 95% CI: 1864 ms to 2291 ms; vrT-3: M = 1985 ms, 95% CI: 1810 ms to 2160 ms). 79   Figure 3.1 Mean of participants’ median response time (in ms) to encoding phase von Restorff (VR) words (red bar), von Restorff surround words (dark blue bars), and non-surround words (light blue bar). Error bars represent the 95% confidence intervals.  Recognition Test Phase. I examined recognition test performance for specific influences due to the occurrence of bird words which had been displayed in red font in the encoding phase (i.e., von Restorff words), as well as for influences due to words which belong to the same category and therefore could have served as von Restorff words (i.e., bird words that had not been presented during the encoding phase). To distinguish between them, I herein use the labels ‘actual von Restorff words’ and ‘potential von Restorff words’, respectively. An initial examination showed that recognition performance was significantly higher on actual von Restorff words (M = 83.7%, 95% CI: 80.0% to 87.3%), compared to potential von Restorff words (M = 78.8%, 95% CI: 74.4% to 83.2%). The recognition hit rate for actual von Restorff words also was significantly higher than recognition of surrounds that followed the von Restorff words (M = 70.5%, 95% CI: 500750100012501500175020002250250027503000VR +1 +2 +3 non-surroundEncdoing Question Response time (ms)Word Type/Position80  66.2% to 72.4%), of surrounds that preceded the von Restorff words (M = 75.2%; 95%CI: 71.1% to 79.3%), of words that followed von Restorff words only in the test phase (M = 77.4%, 95 CI: 73.8% to 81.0%), and of all non-surround old words (M = 74.2%, 95% CI: 71.3% to 77.1%). Recognition of the latter three word types did not differ significantly from one another.  A more detailed analysis focused on recognition hits for the von Restorff words, the three words displayed immediately after the von Restorff words in the encoding phase only, and on the three words that followed the von Restorff words in the test phase only. The results are shown in Figure 3.2. The means and confidence intervals show that recognition was significantly higher on the actual von Restorff words (red bar in Figure 3.2) (M = 83.7%, 95% CI: 80.0% to 87.3%) than on all surround words (dark blue bars in Figure 3.2) (M = 70.1%, 95% CI: 66.9% to 73.4%) and non-surround words (light blue bar) (M = 74.2%, 95% CI: 71.3% to 77.1%). The hit rate did not differ among the three study-phase surrounds; (vrT+1: M = 70.5%, 95% CI: 66.2% to 74.8%; vrT+2: M = 68.8%, 95% CI: 64.5% to 73.1%; vrT+3: M = 71.8%, 95% CI: 67.1% to 76.6%). However, the hit rate was lower for the vrT+1 and vrT+2 words compared to all other non-surround words. Finally, the yellow bars in Figure 3.2 show the hit rate for words that followed von Restorff words in the test phase only; the first two of these words were recognized at the same rate as the non-surround words, but recognition increased across the three words and was significantly higher on the 3rd word than on the 1st word. Compared to non-surround words, there was no difference in recognition test accuracy for words that immediately followed potential von Restorff words during the recognition test phase only (M = 72.7%, 95% CI: 67.7% to 77.7%). A more detailed analysis demonstrated that the second word following potential von Restorff words was recognized more accurately (M = 75.7%, 95% CI: 69.8% to 81.2%) than the first word following a potential von Restorff word (M = 68.9%, 95% CI: 62.4% to 75.4%). Recognition test 81  accuracy for the third word following a potential von Restorff word did not differ from either of the other test phase surrounds (M = 72.3%, 95% CI: 67.4% to 77.3%).   Figure 3.2. Mean recognition test performance (hits) for von Restorff (VR) words (red bar), encoding phase von Restorff surround words (dark blue bars), non-surround words (light blue bar), and test phase von Restorff-surround words (yellow bars). Error bars represent the 95% confidence intervals. Figure 3.3 shows the time required for making correct recognition decisions for each of the word types displayed in Figure 3.2. The data reveal that recognition was roughly equally fast for the von Restorff words and for the non-surround words. Correct rejection decisions were also roughly equally fast for potential von Restorff words (M = 936 ms, 95% CI: 888 ms to 984 ms), though these data were not represented in Figure 3.3. By contrast, participants were slower to make decisions about the three words that had followed the von Restorff words in the encoding phase (dark blue bars) and on the three words 50%55%60%65%70%75%80%85%90%95%100%Recognition Test Performance(Hit Rate)Word Type/Position82  that followed von Restorff words in the test phase (yellow bars). For both of these two word types, responding was significantly slower on the first than third word that followed the von Restorff words.    Figure 3.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions on von Restorff (VR) words (red bar), encoding phase VR-surround words (dark blue bars), non-surround words (light blue bar), and test phase VR-surround words (yellow bars). Error bars represent the 95% confidence intervals.    A supplementary analysis showed the same pattern of slowing for old words that followed each potential von Restorff word (i.e., bird words that did not appear in the encoding phase) in the test phase, with means of 1156 ms (95% CI: 1008 ms to 1303 ms), 1067 ms (95% CI: 973 ms to 1161 ms) and 980 ms (95% CI: 897 ms to 1063 ms), respectively, for the 1st, 2nd and 3rd of these words. Discussion Experiment 3 asked whether the processing of a perceptually distinctive stimulus would have the same proactive effects or different effects on ongoing activity processing as response switching during ProM task retrieval. Three findings from the current experiment are directly relevant to this question. First, during the 500600700800900100011001200130014001500Recognition TestResponse Time (ms)Word Type/Position83  encoding phase participants were slower to respond to the von Restorff words themselves (vrT), as well as for the first word immediately following a von Restroff word (vrT+1). Second, during the recognition test, performance was lower for the three words immediately following von Restorff words in the study phase only. Third, participants required significantly more time for making correct decisions to recognition test words that followed von Restorff words in the test phase.  In the encoding phase, von Restorff words were responded to more slowly than all other non-surround words. This finding stands in contrast to the results of Experiment 1, where responding to bird words that were not recognized as ProM cues was faster compared to all non-surround words. This difference in encoding phase response time for von Restorff words suggests that the manipulation was effective, as von Restorff words successfully shifted participants’ attention. It also indicates that ProM task cues, that are distinctive due to the unique response they require, are processed differently or at the very least more quickly, than stimuli that are distinctive due to their perceptual properties.   In addition to requiring extra time to respond during the encoding phase, von Restorff words were more likely to be accurately recognized during the recognition test phase compared to all other words. This finding has two implications. First, it is consistent with the classic von Restorff findings (see Hunt, 1995), and provides additional evidence that the manipulation of perceptual distinctiveness in Experiment 3 was effective. Second, ProM cue words in Experiment 1 were also recognized more accurately compared to all other words. The comparable performance between ProM cue words and von Restorff words suggests that ProM cue might attract a similar amount of attention at encoding as perpetually distinctive stimuli.  Participants were slower to respond to the first word immediately following von Restorff words during the encoding phase of the experiment. This finding is similar to the results for the first words immediately following planned task responses in Experiments 1 and 2. The similarity in the results among the three experiments suggests that an attentional shift mechanism, exogenously driven by the distinctive 84  nature of the cue itself (Meier & Rey-Mermet, 2012; Rogers & Monsell, 1995), might be responsible for at least some of the proactive effects following ProM task response switching that were observed in Experiments 1 and 2. ProM task cues attract special attention, and this additional attention might delay participants’ ability to return to the ongoing activity, irrespective of the fact that a response switch has been made. This attentional shift mechanism would also apply to perceptually distinctive stimuli like the von Restorff words.  Recognition test accuracy was impaired for the first words that immediately followed von Restorff words compared to all non-surround words. This finding suggests that the attentional shift caused by a perceptually distinctive stimulus does have an impact on participants’ ability to fully engage in encoding processes. However, the impairment observed in Experiment 3 was markedly smaller than the recognition test performance impairment for the first two words following planned task responses observed in Experiment 1 (see Figure 2.2 in Chapter 2). A two sample t-test indicated that the impairment to recognition test performance for the cT+1 words compared to non-surround words was significantly larger in Experiment 1 (an average of performance drop of 16.3%) than Experiment 3 (an average performance drop of 3.7%), t(176) = 3.98, p < .001, d = .59. In Experiment 1, I proposed that recognition test accuracy following a planned task response was impaired due to the executive control processes that must be engaged for the configuration of the task-set relevant to the ProM response, as well as the reconfiguration of the ongoing activity task-set. Given the appropriation of executive control processes for task-set configuration, fewer resources were available for encoding of ongoing activity words, which in turn negatively impacted recognition test performance. In Experiment 3, no response switch was made and therefore there was no need for executive control processes to configure different task-sets. As such, processing resources were more readily available for encoding of the words immediately following the von Restorff words.  85  Participants required more time to make correct recognition decisions for words immediately following actual and potential von Restorff words in the recognition test phase (labelled test phase surround words). This finding was comparable to the results observed for test phase surround words in Experiment 1. In Experiment 1, I attributed response slowing to test phase surround words to the fact that previously task-relevant cues continue to trigger ProM task relevant cognitions (see Meier & Rey-Mermet, 2017; Scullin, Bugg, & McDaniel, 2012; Walser, Fischer, & Goschke, 2012, 2014), and this appears to occur involuntarily even when the ProM task has been explicitly cancelled. In Experiment 3, bird words also may have involuntarily shifted attention, given that they were the category of words displayed in a perceptually distinctive manner in the encoding phase. While the connection between the category birds and the perceptual distinctiveness of the von Restorff words was never made explicitly, I included a post-experiment questionnaire that asked participants if they were aware of anything unique regarding the words displayed in red font. Twenty participants noticed that the von Restorff words were all birds; an additional seven participants said they noticed the words were all animals, and it is possible that other participants were aware of the distinction but did not indicate as much in the post-experiment questionnaire. Even for those participants who were not consciously aware of this association, there may have been a category priming effect (Ferrand & New, 2003), that influenced the way in which bird words were processed, and, in turn, the response time on the recognition test trials that followed bird words. The same effect may not have occurred in Experiment 1 because the special distinction of bird words was made explicit, as was the cancellation of that special distinction, which was not the case in Experiment 3. The results of Experiment 3 provide additional insight into the processes that are likely involved in ProM task response switching. First, the word immediately following perceptually distinctive stimuli that did not require a response switch was responded to significantly more slowly than all other non-surround words. These results support my interpretation that an attentional shift mechanism is at least partially 86  responsible for the response time increases on the first word immediately following a successfully recognized ProM cue. Second, recognition test accuracy for words immediately following von Restorff words did not differ significantly from all other non-surround words. This finding supports my interpretation from Experiment 1 that recognition test accuracy declined due to attentional resources being diverted to task-set configuration processes and therefore unavailable for encoding processes. In sum, the goal of Experiment 3 was to determine whether perceptually distinctive stimuli are sufficient to account for the proactive effects attributed to response switching processes in Experiment 1. The results summarized above indicate that the proactive effects on recognition test accuracy observed in Experiment 1 can be uniquely attributed to the processes required for successful response switching, namely task-set configuration and task-set inertia.   87  4. Chapter 4: High Processing Overlap Increases the Proactive Effects Due to Response Switching A strong predictor for the successful completion of a prospective memory (ProM) task is the degree of overlap in the processing required for the ongoing activity and for the ProM task. Research has shown, for example, that when semantic processing is required for both an ongoing activity and a ProM task, the latter task is more likely to be carried out, compared to when these two tasks require different kinds of processing (e.g., semantic processing and perceptual processing; Meier & Graf, 2000). The finding that the overlap of processing affects completion of a planned task is consistent with the multiprocess framework (McDaniel & Einstein, 2000). This framework suggests that completion of a planned task is guided by automatic versus controlled processes, depending on whether the degree of overlap in the processing required for the ongoing activity and the planned task is high versus low, respectively (see Einstein & McDaniel, 2005; Kelly et al., 2013; Kliegel, Jäger & Phillips, 2008; Loft & Remington, 2013; Scullin et al., 2010). However, more pertinent to the goal of my dissertation research is that most previous investigations were not designed to focus on whether the degree of overlap in processing between an ongoing activity and a ProM task exerts its influence via cue noticing, recognition of cue-plan relevance, or response switching. The goal of the experiments reported in this chapter was to address this information lacuna. More specifically, I wanted to find out whether processing overlap manipulations modulate the proactive effects due to response switching attributed in Chapter 2 to task-set reconfiguration and task-set inertia. The type of processing required for completing the ongoing activity may overlap selectively with the processing required for cue noticing, for recognition of cue-plan relevance, or for response switching. An overlap with cue noticing could be achieved, for example, by using a list of words composed of both simple and compound words (e.g., tree, flower, basketball, crosswalk). For an ongoing activity, participants could be required either to make a semantic decision about each stimulus (e.g., is this word familiar or unfamiliar?) or they might have to decide whether the current stimulus is a simple word or a compound 88  word. For the ProM task, they might have to press the q-key when the word basket occurs anywhere in the experiment. Using this kind of method, the degree of processing overlap with the ProM task is likely to be greater when the ongoing activity requires a simple/compound word decision than when it requires a familiarity decision, and this increase in overlap is likely to facilitate ProM task performance primarily via the cue noticing stage of plan retrieval. ProM researchers have used this kind of manipulation in the past. For example, in an experiment by Einstein et al. (2005), the ProM task instructions were to respond either to a specific syllable (e.g., tor) or to a specific word (e.g., tortoise) during the course of an ongoing activity that required participants to categorize words as belonging to a specific semantic category.  Previous research also has used processing overlap manipulations that likely exerted their influence primarily via recognition of cue-plan relevance. To illustrate, McDaniel et al. (1998) used homographs as ProM cues (e.g., bat, chest, pool), in an experiment with two parts. In Part 1, participants made truth (i.e., true or false) decisions about a series of statements, some of which included words later used as ProM cues. The statement in which each word initially appeared was designed to bias one of its interpretations (e.g., A bat is active at night and sleeps during the day). In Part 2, the ProM task required participants to press the F10-key if they ever saw the word bat (or chest or pool), and the ongoing activity was to make truth decisions about statements. The ProM cues were embedded in statements that either were consistent with the meaning biased in Part 1 (e.g., A dark attic is one possible place to see a bat flying) or inconsistent with the meaning biased in Part 1 (e.g., Equipment used in a baseball game includes a bat and a glove). Performance on the ProM task was significantly better in the consistent condition. One interpretation of this finding is that at the time the plan is formed, it is associated with the cue’s memory representation that is consistent with a specific homographic definition (e.g., bat as a flying mammal). At retrieval, when a cue word is presented in the context of the same homographic definition, the 89  representation of the plan is more likely to be activated (McDaniel et al., 1998), thereby implicating recognition of cue-plan relevance. Processing overlap manipulations that might facilitate or hinder the response switching stage are more difficult to conceptualize, and to my knowledge, the effects of such manipulations remain to be investigated. However, it might be possible to manipulate the overlap between an ongoing activity and response switching, for example, by using an ongoing activity that either requires the same response on each trial or a variety of responses across trials. The need for switching among several responses for the purpose of carrying out the ongoing activity might facilitate the response switching required for executing a planned task.  In the preceding paragraphs, I have described various processing overlap manipulations that could be used to alter the overlap between the processing required for the ongoing activity and the processing required for each of the unique stages of ProM task retrieval: cue noticing, recognition of cue-plan relevance, and response switching. There is research that has manipulated some of these processing overlaps (e.g., Einstein et al., 2005; McDaniel et al., 1998). However, the goal of previous research that manipulated processing overlaps was to examine how such overlaps affect overall ProM task performance; the focus was not on investigating the unique influence of such manipulations on the individual stages involved in ProM task retrieval. Moreover, to my knowledge, none of the research I described in the preceding paragraphs has examined how processing overlap manipulations influence the proactive effects produced by executing a ProM task, or whether all kinds of processing overlap manipulations influence such proactive effects in the same way. Only one research group has directly examined the way in which the proactive effects I have attributed to response switching during ProM retrieval are influenced by a processing overlap manipulation. Meier and Rey-Mermet (2017) used a processing overlap manipulation which most likely overlapped 90  selectively with the processing required for cue noticing. They embedded a ProM task into an ongoing activity that always alternated between three tasks. One task required participants to make odd/even decisions about digits (e.g., 777), another required participants to make red/blue colour decisions about symbols (e.g., %%%), and a final task required participants to make upper/lower case decisions about letters (e.g., nnn). In a series of experiments, they varied the shared properties between the ProM cue and the stimuli used for the three ongoing activity tasks. For example, in one experiment, ProM cues were triplicate letters displayed in either blue or red, thus sharing properties with stimuli used for both the case and colour decision tasks. Meier and Rey-Mermet (2017) reported that when the processing overlap between the cue and the ongoing activity increased, response time switch costs were larger and longer lasting.  Given the paucity of research specifically examining the influence of processing overlap manipulations on the proactive effects associated with planned task execution, the goal of the experiments reported in this chapter was to identify how one specific processing overlap manipulation influenced the proactive effects I have attributed to the response switching mechanisms, task-set inertia and task-set configuration. The experiments reported in this chapter were modelled after Experiment 1. In Experiment 4a, participants began with an encoding phase, where they were required to answer semantic questions about words. The encoding phase also included a ProM task, but the instructions for the ProM task differed between experimental conditions. The high overlap condition was a direct replication of Experiment 1. For the ProM task, participants were instructed to make a unique response to words from the category birds. In the low overlap condition, ProM cues were words displayed in lower case letters, and participants were instructed to make a unique response to such words. Following the encoding phase, all participants completed a recognition memory test. 91  The processing overlap manipulation used in Experiment 4a differs substantially from those used by previous investigations (e.g., Einstein & McDaniel, 2005; Kelly et al., 2013; Kliegel, Jäger & Phillips, 2008; Loft & Remington, 2013; Meier & Graf, 2000; Scullin et al., 2010). In previous investigations, the critical overlap was operationalized by the type of processing required for the ongoing activity (e.g., semantic versus perceptual processing) and by the type of processing required for identifying a stimulus as a ProM task cue (e.g., cues defined by their semantic or perceptual properties). By contrast, for the present experiment, I also varied the display properties of cues between conditions, showing them in upper versus lower case letters, respectively, in the high versus low overlap conditions. I varied the cue display properties on the assumption that visually distinctive cues would facilitate overall ProM task performance (see Cohen et al., 2003; Graf et al., 2002; McDaniel & Einstein, 1993; Trawley et al., 2014), and thus help to reduce or eliminate the performance difference between the high and low overlap conditions that has been reported by previous investigations (see Kelly et al., 2013; Kleigel et al., 2008; Scullin et al., 2010). This effort to reduce or eliminate the typical difference in ProM task performance in the high versus low overlap condition was motivated by the possibility that the proactive influences due to response switching might be different when ProM task responses are frequent versus infrequent.   Previous investigations that used processing overlap manipulations provide a basis for predicting overall ProM task performance, as well as guidance about how such manipulations might influence effects attributable to task-set inertia and task-set configuration. Consistent with the multiprocess framework (McDaniel & Einstein, 2000), it has been argued that ProM task performance is elevated in conditions with high versus low processing overlaps because responding to ProM cues is mediated by automatic processes in the former condition versus controlled processes in the latter condition (McDaniel & Einstein, 2007). Although the assumption about the involvement of automatic versus controlled processes may still hold for the present study, I expected no difference in overall ProM task performance because of the 92  visually distinctive cues used in the low overlap condition. More importantly, based on the assumption that ProM task performance depends more on controlled processes in the low than high overlap condition, fewer resources would be available for configuring the ongoing activity task-set, thereby increasing the proactive effects associated with this mechanism in the low overlap condition compared to the high overlap condition. An alternative prediction is that the higher overlap between the processing required for ProM task and the ongoing activity will increase the proactive effects of response switching (see Gade & Koch, 2007). When there is a greater degree of processing overlap between two tasks, the ProM task-set might be more likely to remain active during subsequent ongoing activity trials, consistent with the assumptions about task-set inertia (Allport et al., 1994). If task-set inertia is more demanding of executive control processes, I expect that this process would leave fewer attentional resources available for processing of the stimuli associated with the ongoing activity, and in turn result in larger proactive effects. This result would be consistent with the typical findings from the task-switching research (e.g., Gade & Koch, 2007; Keisel et al, 2010), as well as with the findings from recent ProM research by Meier and Rey-Mermet (2017).  Experiment 4a The basic method of this experiment was the same as for Experiment 1. Exactly as for Experiment 1, for the high processing overlap condition the ongoing activity and the ProM task cue both required semantic processing. For the low processing overlap condition, the ongoing activity required semantic processing and the ProM task cue required perceptual processing. Processing overlap was manipulated between-subjects.  Method Participants. The participants were 183 undergraduate student volunteers, 94 of them randomly assigned to the high overlap condition and 89 to the low overlap condition. Each participant received partial 93  course credit for participating in the study and provided informed written consent at the beginning of the experiment. The study was approved by the ethics review board.  Materials and Procedure. A new list of 268 words randomly selected from the Toronto Word Pool (Friendly et al. 1982), plus 12 bird words, were split into two sets, A and B. Each set consisted of 134 common words plus 6 bird words. As in previous experiments, for each participant, the common words from one set were presented in the study phase of the experiment together with six of the bird words (140 words total), with the remaining 140 words (common words and six bird words) serving as recognition test distractors. The set of words that were presented during the study phase and the set that served as recognition test distractors were counterbalanced across participants.  The study procedure for the high overlap condition was a replication of the procedure used in Experiment 1, but this procedure was modified for the low overlap condition. In both conditions, the ongoing activity required participants to answer semantic questions about words, exactly as described in the Method of Experiment 1. For the ProM task in the high overlap condition, participants were instructed to press the q-key in response to words referring to types of birds (e.g., EAGLE). For the ProM task in the low overlap condition, participants were instructed to press the q-key in response to words that were displayed in lower case letters (e.g., eagle). All other words were displayed in upper case letters. The same bird words were used as cues in the low overlap condition, however participants were not provided any information about the category membership of the words during the instructions. Instead participants in the low overlap condition were required to respond to the perceptual properties of the cues. This manipulation is consistent with the kinds of high/low overlap manipulations used by others (e.g., Einstein et al., 2005; Meier & Graf, 2000; Scullin et al., 2010). As in Experiment 1, each word was displayed for a minimum of two seconds to ensure the duration of participants’ exposure to each word was roughly equivalent irrespective of how quickly they answered the encoding question displayed with each. 94  The procedure for the recognition memory test was modelled after Experiment 1. The test phase consisted of all 140 words from Phase 1 (either set A or B), and all 140 words from the other set (the words not presented in Phase 1). The recognition test also included six special blocks for investigating how performance is affected by the test phase occurrence of ProM cue words. Participants were asked to respond with the left arrow key to indicate that a word was old, meaning they recognized it as having occurred in the first phase of the experiment, and the right arrow key if the word was new, meaning they had not seen the word in Phase 1. In the low overlap condition only, all bird words were displayed in lower case letters, and all other words were displayed in upper case letters. In the high overlap condition, all words were displayed in upper case letters. Participants were instructed to respond as quickly and accurately as possible.  Results As in Experiment 1, the main dependent variables of interest were accuracy and speed on the ProM task, response time on the encoding phase task, as well as accuracy and decision time on the old/new recognition memory test. I used the 95% confidence interval for making decisions about differences among means. All data were screened for outliers. On the recognition test, four participants had median response times that were more than two standard deviations below the group mean of 1030 ms. These same participants also showed a bias in recognition memory test responding, by responding to over 90% of words with an old response, most likely because of failing to follow task instructions. None of the data from these participants, three from the high overlap condition and one from the low overlap condition, were included in any of the analyses reported below. I checked prospective and retrospective memory task performance for floor and/or ceiling effects, and compared performance between the high and low overlap conditions. Table 4.1 presents descriptive 95  statistics on overall performance on the prospective and retrospective memory task. The means show that there was no evidence of floor or ceiling effects on the ProM task in either condition. Participants executed the planned task response to a significantly smaller proportion of cues in the high overlap condition compared to the low overlap condition. Performance on the ProM task in the high overlap condition was comparable to ProM task performance in Experiment 1. The mean amount of time required for making correct responses to ProM cues in the high overlap condition was 1831 ms (95% CI: 1634 ms to 2027 ms) which was faster than ongoing activity responses to non-cue words at 2080 ms (95% CI: 1922 ms to 2239 ms). In the low overlap conditions, the mean of participants’ median response time for correct responses to ProM cues was 1343 ms (95% CI: 1254 ms to 1432 ms), significantly faster than the same kind of response in the high overlap condition, as well as faster than ongoing activity responses to non-cue words, which averaged 2137 ms (95% CI: 1974 ms to 2301 ms). For the recognition memory test, performance was similar across the overlap conditions (see Table 4.1). A signal detection analysis indicated that recognition performance was significantly above chance in both the high overlap condition, (d’ = 1.67, 95% CI: 1.54 to 1.80) and the low overlap condition (d’ = 1.62, 95% CI: 1.52 to 1.72).  Condition ProM Task Performance Recognition Test Hit Rate Recognition Test FA Rate High Overlap M = 70.3%  95% CI: 63.4% to 77.1% M = 76.6%  95% CI: 74.3% to 78.9% M = 21.9%  95% CI: 19.0% to 24.8% Low Overlap  M = 89.0% 95% CI: 83.9% to 94.0% M = 76.8% 95% CI: 74.6% to 79.0% M = 21.8% 95% CI: 19.2% to 24.4% Table 4.1. ProM task performance, recognition test hit rate for all previously studied words, recognition test false alarm (FA) rate for all unstudied words, as a function of condition (high overlap vs low overlap).    Encoding Phase. The first analysis was used to determine how quickly participants responded to encoding phase questions about cue words, cue-surround words, and all other non-surround words. The means and confidence intervals are represented in Figure 4.1. They show that the mean of participants’ median response times were roughly equal for all word types except for the first word immediately following 96  the ProM cue (cT+1), in both the high (M = 2754 ms, 95% CI: 2390 ms to 3119 ms) and low (M = 2686 ms, 95% CI: 2433 ms to 2938 ms) overlap conditions. There was no significant difference in response time on the cT+1 words between the two conditions. A separate analysis showed that the time required for responding to the three words which preceded each high overlap condition cue and each low overlap condition cue was not different from the time required to respond to the non-surround words.  There was a significant difference in response time for missed high overlap and low overlap cues. When high overlap cues were missed, they were responded to significantly faster (M = 1749 ms, 95% CI: 1548 ms to 1949 ms) than low overlap cues (M = 2266 ms, 95% CI: 1940 ms to 2592 ms).   Figure 4.1. Mean of participants’ median response time (in ms) to encoding phase cue words, cue-surround words, and non-surround words for the high (dark blue bars) and low (light blue bars) overlap conditions. The RT’s for cue-surround words are only for words following cues that elicited a planned task response. The RT’s for cue words are only for those trials on which the ongoing activity response was provided for the cue. Error bars represent the 95% confidence intervals. 500750100012501500175020002250250027503000"Missed" Cues +1 +2 +3 Non-SurroundEncoding Question Response Time (ms)Word Type/PositionHigh OverlapLow Overlap97  Recognition Test Phase. In the high overlap condition, the results replicated the findings from Experiment 1. Recognition performance was superior for correctly recognized actual cues (M = 81.8%, 95% CI: 76.9% to 76.8%) and potential cues (M = 82.7%, 95% CI: 78.4% to 86.9%) compared to non-surround old words (M = 76.5%, 95% CI: 73.7% to 79.3%) and test phase cue-surrounds (i.e., the three old words which followed actual cues in the test phase only) (M = 74.7%, 95% CI: 70.8% to 78.6%). For the low overlap condition, there were no significant differences in recognition test performance for any word types: correctly recognized cues (M = 73.6%, 95% CI: 68.7% to 78.5%), potential cues (M = 74.9%, 95% CI: 69.9% to 80.0%), non-surround non-cue old words (M = 76.7%, 95% CI: 74.4% to 79.1 %) and test phase surrounds (M = 76.4%, 95% CI: 73.0% to 79.9%).  For the surround words, overall recognition performance was lower in the high overlap condition (M = 73.3%, 95% CI: 68.0% to 78.6%) than the low overlap condition (M = 77.5%, 95% CI: 73.0% to 81.9%), though this difference was not statistically significant (see Figure 4.2). However, a detailed analysis that focused on recognition hits as a function of word position indicated that this difference was driven by low a hit rate on the cT+1 (M = 68.7%,  95% CI: 63.2% to 74.3%) and cT+2 (M = 71.3%,  95% CI: 65.8% to 76.9%) words in the high overlap condition. The same analysis of recognition test performance for individual surround words in the low overlap condition indicated that only the cT+1 (M = 73.5%, 95% CI: 69.4% to 77.5%) and not the cT+2 (M = 78.1%, 95% CI: 73.7% to 82.5%) showed a drop in hit rate compared to non-surround words.  98   Figure 4.2. Mean recongition test performance (hits) for cue words, cue-surround words, and non cue-surround words, for the high overlap (dark blue bars) and low overlap (light blue bars) conditions. The figure only includes data for those cue words and corresponding cue-surround words which had been successfully recognized as ProM task relevant. Error bars represent the 95% confidence intervals. Figure 4.3 shows the mean of participants’ median response times for making correct recognition decisions for the same words as represented in Figure 4.2. Participants responded more slowly to the first two words that followed cues in the encoding phase compared to the third word and all other non-surround words, in both the high overlap and low overlap conditions. For most word types/positions, there were no significant differences in response time between the two conditions. The only notable difference was in the response time for low overlap cue words (M = 1356 ms, 95% CI: 1233 ms to 1477 ms), which was significantly slower than response time for high overlap cues (M = 1050 ms, 95% CI: 988 ms to 1112 ms). It was also significantly slower than response time for all other word types (see Figure 4.3).  50%55%60%65%70%75%80%85%90%95%100%Cues +1 +2 +3 Non-SurroundRecognition Test Performance(Hit Rate)Word Type/PositionHigh OverlapLow Overlap99   Figure 4.3. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words, non cue-surround words for both the high overlap (dark blue bars) and low overlap (light blue bars) conditions. Error bars represent the 95% confidence intervals.  I also analyzed the response times and recognition hit rate for test-phase surround words, which were previously studied words that followed actual and potential cue words in the test phase only. There was no significant difference in recognition for test phase surround words that followed actual high overlap cues (M = 74.7%, 95% CI: 70.8% to 78.6%), actual low overlap cues (M = 76.4%, 95% CI: 73.0% to 79.9%), potential high overlap cues (M = 74.0%, 95% CI: 70.5% to 77.6%), potential low overlap cues (M = 75.6%, 95% CI: 71.8% to 79.3%). There were significant differences in response times for making correct recognition decisions to the test phase surround words in each of these conditions (see Figure 4.4). These differences were driven primarily by the first test phase surround word immediately following each actual cue (high overlap test phase +1 M = 1602 ms, 95%CI: 1391 ms to 1821 ms; low overlap test phase +1 M = 1632 ms, 95% CI: 1336 ms to 1929 ms) and potential cue (high overlap test phase +1 M = 1322 ms, 95%CI: 1168 ms to 1476 ms; low overlap test phase +1 M = 1530 ms, 95% CI: 1350 ms to 1771 ms) in the 500600700800900100011001200130014001500Cues +1 +2 +3 Non-SurroundRecognition Test Response Time (ms)Word Type/PositionHigh OverlapLow Overlap 100  test phase only (see Figure 4.4, panels A and B). Additionally, correct recognition decisions were made more quickly for the first test phase surround words following a high overlap potential cue compared to a low overlap potential cue (see Figure 4.4, Panel B).     Figure 4.4. Mean of participants’ median response time required for making correct recognition decisions about words following actual cues (Panel A) or potential cues (Panel B) in the test phase only. Dark blue bars represent test phase surround words in the high overlap condition and light blue bars represent test phase surround words in the low overlap condition. Error bars represent 95% confidence intervals Discussion The goal of Experiment 4a was to determine the influence of a specific manipulation of processing overlap between the ongoing activity and the ProM task on the processes I assume to be required for response switching, namely, task-set configuration and task-set inertia. I manipulated the kind of processing required for the ProM task cue (semantic versus perceptual) while participants were engaged in an ongoing activity that required semantic processing. I also included an intentional confound in the 50075010001250150017502000TestPhase +1TestPhase +2TestPhase +3Recognition Test Response Time (in ms)Word Type/PositionPotential CuesHigh OverlapLow Overlap50075010001250150017502000TestPhase +1TestPhase +2TestPhase +3Recognition Test Response Time (in ms)Word Type/PositionActual CuesHigh OverlapLow OverlapB) A) 101  experiment, by making cues in the low processing overlap condition perceptually salient. This confound was included to eliminate the ProM task performance differences that typically occur between the high and low processing overlap conditions (see Kleigel et al., 2008). Four findings require further discussion. First, in the encoding phase, the first word following planned task execution (cT+1) was responded to significantly more slowly than all other words, and the time required for responding to this word was similar in the high and low overlap conditions. Second, recognition test performance was impaired for the cT+1 and cT+2 words, and this impairment was greater in the high overlap condition compared to the low overlap condition. Third, participants required more time for making correct recognition decisions for the cT+1 and cT+2 words in both processing overlap conditions. Finally, ProM task performance was significantly better in the low overlap condition compared to the high overlap condition.  The finding that participants performed better on the ProM task in the low overlap condition than the high overlap condition was expected, but it is different from what is usually reported by investigations that have focused on the effects of processing overlaps on ProM performance. In previous investigations of the effects of processing overlaps, ProM task performance was typically superior in conditions of high processing overlap (e.g., Einstein et al., 2005; Meier & Graf, 2000). The multiprocess framework accounts for this type of performance difference by proposing that when the degree of processing overlap is high versus low, execution of the planned task is guided by automatic versus controlled processes, respectively (see Einstein & McDaniel, 2005; Kelly et al., 2013; Kliegel, Jäger & Phillips, 2008; Loft & Remington, 2013; Scullin et al., 2010). Despite the pattern of ProM task performance observed in Experiment 4a, this explanation might still hold. The findings from Experiment 4a do, however, call into question the general claim that a greater degree of processing overlap always leads to better performance on ProM tasks, and suggest that the generalization about the effects of processing overlap on ProM task performance need to be made more cautiously. It is possible, consistent with what is proposed by McDaniel and Einstein’s 102  (2000) multiprocess framework, that conditions of high processing overlap facilitate the automatic activation of the representation of the plan, thereby influencing the recognition of cue-plan relevance stage of retrieval. However, other characteristics of cues, such as their perceptual salience, can influence overall ProM performance by affecting a different stage of ProM retrieval, such as cue noticing. These findings, and their implications, point to the importance of a detailed understanding of the processes involved in all the stages of ProM retrieval.  The finding that encoding phase responding was slower on the first word immediately following a planned task response (the cT+1 word), and response time was roughly equivalent in the low overlap and high overlap conditions (see Figure 4.1), suggests that the processing overlap manipulation used in Experiment 4a had no influence on the time required for reconfiguration of the ongoing activity task-set. Switch costs, such as slower responding to the cT+1 word, are assumed to reflect the executive control processes required for task-set configuration (Mayr & Keele, 2000; Monsell et al., 2003; Rogers & Monsell, 1995; Wylie & Allport, 2000). It is also assumed, under the multiprocess framework (McDaniel & Einstein, 2000), that executive control processes are required for the planned task to be executed under conditions of low processing overlap. However, the lack of difference between overlap conditions suggests that the resources required for plan execution do not influence reconfiguration of the ongoing activity. A potential explanation for this lack of difference is that the influence of the processing overlap manipulation was limited to the processing resources required for cue noticing and recognition of cue-plan relevance stages of ProM retrieval. An alternative explanation for the lack of difference between the experimental conditions on encoding phase response time is that I used a procedure which equated overall ProM task performance across processing overlap conditions and the time required for task-set reconfiguration on the ongoing activity could be linked to the overall level of ProM task performance. A more typical processing overlap 103  manipulation would test the possibility that reconfiguration of the ongoing activity task-set is not influenced by manipulations of processing overlap. Experiment 4b explored this possibility.  The finding that impairment on the recognition test was larger and longer-lasting in the high overlap condition compared to the low overlap condition suggests that the processing overlap manipulation did have an influence task-set inertia processes, and potentially task-set reconfiguration. In the high overlap condition, recognition test performance was comparable to the results of Experiment 1. Impairment on the recognition memory test in Experiment 1 was explained by the fact that the rarely occurring ProM task is likely to remain active and interfere with the reconfiguration of the ongoing activity task-set, consistent with the assumptions about task-set inertia (Wylie & Allport, 2000). The persistent activation of the ProM task-set effectively reduced the attentional resources essential for memory encoding (Allport et al., 1994). This explanation is also relevant to the impairment of recognition memory test performance observed in the high overlap condition of Experiment 4a. In Meier and Rey-Mermet’s (2017) series of experiments, they provided a similar explanation for the finding that a greater degree of processing overlap increased the size and duration of switch costs. Specifically, they posited that ongoing activity trials which shared greater overlap with the ProM cue may have reactivated the planned task, and slowed performance for the ongoing activity. Task-set inertia was less likely to be involved in the low overlap condition because the ongoing activity and ProM task required different kinds of processing. The ongoing activity trials following planned task execution were therefore less likely to keep the ProM task active. As such, in this condition more attentional resources were available for encoding processes. This explanation is also consistent with the task-switching literature. Under conditions of low processing overlap, researchers observed smaller switch costs, and speculated that fewer processing resources were required for differentiating between the two responses (Gade & Koch, 2007; Kiesel et al., 2010; Monsell et al., 2003). 104  The time required for making correct recognition decisions for the words that followed a planned task response during encoding was uninfluenced by the processing overlap manipulation, suggesting that a separate mechanism must be involved in recognition test decision time effects versus recognition test decision accuracy. For each trial during the recognition memory test, participants had to configure an old or a new response, and the response configured was likely dependent on a combination of familiarity and contextual information that was available for each word. Given that the words that immediately followed a ProM task response in the encoding phase were processed less extensively (as described in the preceding paragraph), the contextual information required for making correct old decisions to these words was less readily accessible. Participants therefore required more time to retrieve this contextual information and subsequently configure the correct old response for those words.  Taken together, the findings from Experiment 4a are similar to those from Experiment 1, showing the same effects on encoding phase response speeds in both processing overlap conditions, thus suggesting that the processing overlap manipulation used in Experiment 4a had no influence on task-set configuration. However, the results from Experiment 4a showed an effect due to the processing overlap manipulation on recognition test performance. In Experiment 1, and in the preceding paragraphs, the reduced recognition of words displayed on trials immediately after each successful ProM task response was explained with respect to task-set inertia. The different degrees of impairment on the recognition memory test in the high and low processing overlap condition suggests that only this construct is influenced by the processing overlap manipulation used in Experiment 4a. The conclusion that processing overlap manipulations only affect task-set inertia and not task-set configuration must be accepted with caution, however, for two main reasons. The first reason is that Experiment 4a incorporated a confound that was deliberately included in order to equate performance between the high and low overlap conditions. This confound cannot explain the difference between the 105  overlap conditions in the recognition performance data. However, it is possible that the confound acted to eliminate differences between the overlap conditions in encoding phase response times and in recognition decision speeds. A second reason for caution with accepting conclusions based on the finding from Experiment 4a comes from the limited conditions that were explored in this experiment. It is possible, for example, that the findings from Experiment 4a hold only under conditions where a ProM task needs to be carried out in the context of a semantic ongoing activity. Processing overlap manipulations might have different effects in experiment where the ongoing activity requires perceptual processing. Experiment 4b Experiment 4b was designed to explore the generality of the findings from Experiment 4a. More specifically, the goal of Experiment 4b was to find out if the effects found in Experiment 4a would occur also when the ongoing activity requires a form of perceptual processing. Additionally, Experiment 4a included an intentional confound, where the low processing overlap cues were also perceptually salient. In Experiment 4b, this confound was removed in order to determine the effects observed in Experiment 4a were a result of the processing overlap manipulation, or a result of the confound that changed overall ProM performance.  The processing resources required for completing an ongoing activity is a predictor of the successful execution of ProM tasks (e.g., Einstein, Smith, McDaniel & Shaw, 1997; Marsh, Hancock & Hicks, 2002; Rendell et al., 2007). Consistent with Levels of Processing (Craik & Lockhart, 1972), it might be argued that the semantic ongoing activity used in Experiment 4a is more dependent on processing resources, than for example, an ongoing activity which requires perceptual processing. On the assumption that task-set configuration and task-set inertia are associated with the availability of processing resources, it is possible that reconfiguration of the ongoing activity task-set occurs more slowly and persistence of the ProM task-set lasts longer after making a ProM response in the context of a semantic rather than a perceptual ongoing activity. This possibility leads to the prediction that executing a ProM task in the context 106  of an ongoing activity less demanding of attentional resources (i.e., one that requires perceptual rather than semantic decisions about words) will result in a smaller proactive effects overall. Further, I anticipate that by removing the confound from Experiment 4a I will observe better ProM task performance in the high overlap condition compared to the low overlap condition, consistent with previous research (e.g., Einstein & McDaniel, 2005; Kliegel et al., 2008).  Method  In order to explore the generality of the findings from Experiment 4a, I modified the method for Experiment 4b. For the ongoing activity, participants were asked to complete a task that required perceptual processing of a list of words. Specifically, they had to classify words based on the proportion of letters within each word that were displayed in lower case. In the high overlap condition, the ProM task was to identify words that were displayed all in lower case letters (i.e., cues were defined by perceptual properties). In the low overlap condition, the ProM task required participants to identify words representing kinds of birds (i.e., cues were defined by semantic properties). High overlap cues and low overlap cues were displayed in exactly the same way, that is, in all lower case letters.  Participants. The participants were 173 undergraduate student volunteers, 88 randomly assigned to the high overlap condition and 85 randomly assigned to the low overlap condition. They received partial course credit for their participation in the study, which was approved by the ethics review board. Each participant provided informed written consent at the beginning of the experiment. Materials and Procedure. In this experiment, a new set of 288 words were selected from the Toronto Word Pool (Friendly et al., 1982), plus an additional 12 bird words. This set was selected to include four-, five-, six-, and seven- letter words, as these word lengths corresponded to the length of the bird words selected to serve as ProM task cues. There were a total of 28 four-letter words, 30 five-letter words, 60 six-letter words, and 40 seven-letter words. These proportions were selected based on the available 107  words in the Toronto Word Pool (Friendly et al., 1982). As in previous experiments, the words were arranged into two sets (Sets A and B), in a pseudo-random manner which ensured an equal number of four-, five-, six-, and seven- letter words in each set.  The words used for the ongoing activity were displayed so that some of the letters of the word were in lower case and some of the letters were in upper case (e.g., tABle). For each word length, the number of lower case letters was between one and one less than the total number of letters in the word. An equal proportion of words of each length (i.e., 4- 5- 6- and 7- letter words) were presented with fewer than half of the letters in lower case (e.g., WatCH) and more than half of the words in lower case (e.g., tABle). None of the words had exactly half of the letters in lower case. The ProM task cue words were the only words that were displayed in all lower case letters. The position of the letters that were in lower case varied randomly from word to word.  The ongoing activity required participants to classify words based on the number of letters within each word that were displayed in lower case. The same question was paired with each word: How many letters are in lower case? Participants were asked to respond by selecting either the left arrow key, indicating fewer than half, or the right arrow key, indicating more than half.  Processing overlap was manipulated between-subjects. For the high overlap condition, the ProM task instructions were to ignore the case question and press the q-key if ever in the course of the ongoing activity the participant came across a word displayed in all lower case letters. For the low overlap condition, participants were instructed that if ever they came across a word referring to a type of bird, they should press the q-key. In the low overlap condition, these words were also displayed in all lower case letters, however no reference was made to this difference in display properties. As in Experiment 1, each word was displayed for a minimum of two seconds to ensure the duration of participants’ exposure to each word was roughly equivalent irrespective of how quickly they answered the encoding question. For the recognition 108  memory test, in both conditions all words were displayed in all upper case letters. All other procedures were identical to Experiment 4a.  Results The main dependent variables were the accuracy and speed of responding on the ProM task, the accuracy and speed of responding on the encoding phase task, as well as the accuracy and speed of responding on the recognition memory test. I used 95% confidence intervals for making decisions about the differences among means.  All data were screened for outliers. A signal detection analysis for performance on the recognition memory test indicated that four participants from the high overlap condition and one participant from the low overlap condition had d’ scores less than or equal to zero. An additional two participants from the low overlap condition showed biased responding to recognition test words; one participant responding almost exclusively (91.5%) to all words as old and another responding almost exclusively (90.0%) to all words as new. These participants were most likely not following task instructions, and consequently, none of their data were included in the analyses reported below.  I checked prospective and retrospective memory performance for floor and/or ceiling effects, and also compared performance between the overlap conditions. On the ProM task, participants executed the planned task significantly more often to cues in the high overlap condition than in the low overlap condition (see Table 4.2). Performance was not at floor in either condition, but was near ceiling in the high overlap condition. There was also much more variability in performance for the low overlap condition compared to the high overlap condition, as evidenced by the 95% confidence interval (see Table 4.2). The mean amount of time required for making the planned task response to ProM cues was not statistically different in the high overlap condition (M = 1335 ms, 95% CI: 1254 ms to 1416 ms) and the low overlap condition (M = 1511 ms, 95% CI: 1269 ms to 1753 ms). Planned task response times to cue words were slower than 109  ongoing activity responses to missed cue words (M = 1005 ms, 95% CI: 862 ms to 1147 ms) in the high overlap condition, but ongoing activity responses to missed cues in the low overlap condition took roughly the same amount of time (M = 1325 ms, 95% CI: 1187 ms to 1464 ms) as planned task responses to cues. For the recognition memory test, the hit rate was not significantly different from chance in the high overlap condition. In the low overlap condition, the lower limit of the 95% confidence interval was less than 1% different from chance performance. False alarm rate was significantly better than chance in both conditions (see Table 4.2). Taken together, overall performance on the recognition test was above chance as indicated by d’ scores, for the high overlap condition, d’ = .51 (95% CI: .46 to .56) and for the low overlap condition, d’ = .64 (95% CI: .58 to .71).  Condition ProM Task Performance Recognition Test Hit Rate Recognition Test FA Rate High Overlap  M = 92.6%  95% CI: 88.9% to 96.4% M = 51.4%  95% CI: 48.2% to 54.5% M = 33.0%  95% CI: 30.1% to 35.8% Low Overlap  M = 69.9%  95% CI: 62.7% to 77.1% M = 54.3%  95% CI: 50.9% to 57.6% M = 31.9%  95% CI: 28.5% to 35.2% Table 4.2. ProM task performance, recognition test hit rate for studied words and recognition test false alarm (FA) rate for unstudied words as a function of cue condition (high overlap vs. low overlap). Encoding Phase. Participants’ mean response time to encoding phase questions for cue words, cue-surround words, and all other non-surround words replicated the general pattern of results from Experiment 4a. The means and confidence intervals in Figure 4.5 show that response times were roughly equal for most word types except for the first word (cT+1) immediately following the ProM cue in the high overlap (M = 2530 ms, 95% CI: 2321 ms to 2739 ms) and low overlap (M = 2754 ms, 95% CI: 2479 ms to 3029 ms) conditions. Between the high overlap and low overlap conditions, there was a significant difference in response time for the first word following the cues that elicited a planned task response. Ongoing activity responses to high overlap cues were significantly faster (M = 1005 ms, 95% CI: 861 ms to 1148 ms) than ongoing activity responses to low overlap cues (M = 1511 ms, 95% CI: 1269 ms to 1753 110  ms). In both cases, response time was faster for ongoing activity responses to cues than for all other word positions (see Figure 4.5).  Figure 4.5. Mean of participants’ median response time (in ms) to encoding phase cue words, cue-surround words, and non-surround words for the high (dark blue bars) and low (light blue bars) overlap conditions. The RT’s for cue-surround words are only for words following cues that elicited a planned task response. The RT’s for cue words are only for those trials on which the ongoing activity response was provided for the cue. Error bars represent the 95% confidence intervals. In addition to encoding phase response time, I analyzed overall performance accuracy for the ongoing activity. Performance was not statistically different in the high overlap (M = 85.3%, 95% CI: 82.6% to 88.1%) and low overlap (M = 86.4%, 95% CI: 83.8% to 88.9%) conditions. In the high overlap condition, mean performance on the ongoing activity for cue words that did not elicit the ProM task response was 70.0% (95% CI: 52.0% to 88.0%), which was lower than performance on the same words in the low overlap condition (M = 91.2%, 95% CI: 84.6% to 97.7%). Recognition Test Phase. The analysis of the recognition hit rate by word type/position showed that for most word types/positions, hit rate was not significantly different from chance (see Figure 4.6). The 500750100012501500175020002250250027503000"Missed" Cues +1 +2 +3 Non-SurroundEncoding QuestionResponse Time (ms)Word Type/PositionHigh OverlapLow Overlap111  only words that were correctly recognized at better than chance levels were cue words that successfully elicited a planned task response in the low overlap condition (M = 76.8%, 95% CI: 71.4% to 82.4%) and non-surround words in the low overlap condition (M = 55.4%, 95% CI: 51.9% to 58.8%). For cues in the high overlap condition that failed to elicit a ProM task response, performance (M = 39.4%, 95% CI: 19.5% to 59.2%) was comparable to low overlap cues that failed to elicit the ProM task response (M = 58.6%, CI: 48.3% to 68.9%). Correct rejection rate for potential cue words (i.e., bird words) that did not occur in the encoding phase was significantly better in the low overlap condition (M = 72.9%, 95% CI: 67.6% to 78.2%) and the high overlap condition (M = 78.5%, 95% CI: 73.7% to 83.2%). Further, performance was better for cues in both processing overlap conditions when compared to correct rejection rate for all other new words in both the low overlap (M = 67.7%, 95% CI: 64.1% to 71.2%) and high overlap (M = 67.0%, 95% CI: 64.2% to 69.9%) conditions.   Figure 4.6. Mean recongition test performance (hits) for cue words, cue-surround words, and non cue-surround words, for the high overlap (dark blue bars) and low overlap (light blue bars) conditions. The figure only includes data for those cue words and their corresponding cue-surround words that successfully elicited the planned task response. Error bars represent the 95% confidence intervals. 0%10%20%30%40%50%60%70%80%90%100%Cues +1 +2 +3 Non-SurroundRecognition Test Accuracy (Hit Rate)Word Type/PositionHigh OverlapLow Overlap112  Figure 4.7 shows the mean of participants’ median response times for making correct recognition decisions for cue words, cue-surround words, and all other non-surround words. Responding was significantly slower for the three words that followed cues in the study phase compared to all other non-surround words, and more so in the low overlap condition. Additionally, response time for low overlap cue words (M = 1266 ms, 95% CI: 1145 ms to 1387 ms) was significantly slower than for high overlap cue words (M = 1068 ms, 95% CI: 986 ms to 1150 ms) and all other non-surround words (high overlap condition: M = 1021 ms, 95% CI: 965 ms to 1076 ms; low overlap condition: M = 1006 ms, 95%CI: 958 ms to 1054 ms).   Figure 4.7. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words, encoding phase cue-surround words and all non-surround words. Dark bars repesent the data for the high overlap condition and light bars represent the data for the low overlap condition. Error bars represent the 95% confidence intervals. Discussion The goal of Experiment 4b was to determine whether the proactive effects observed in Experiment 4a could be replicated in the context of an ongoing activity that required perceptual processing. The 500600700800900100011001200130014001500Cues +1 +2 +3 Non-SurroundRecognition Test Response Time (ms)Word Type/PositionHigh OverlapLow Overlap113  confound that was introduced in Experiment 4a to equate performance across overlap conditions was removed in Experiment 4b. Two findings from Experiment 4a were replicated. Participants required more time to respond to encoding phase questions and to make correct recognition test decisions for the words that immediately followed a planned task response in both processing overlap conditions. However, two findings from Experiment 4a were not replicated in this Experiment 4b. First, recognition memory test performance did not differ between processing overlap conditions. However, overall recognition performance was very poor; the hit rate for previously studied words was at or near chance levels in both processing overlap conditions. Second, ProM task performance was significantly better in the high overlap condition compared to the low overlap condition. The theoretical implications of these findings are discussed in the following paragraphs.  The finding that ProM task performance was better in the high overlap condition than the low overlap condition is the opposite pattern of ProM task performance observed in Experiment 4a. In Experiment 4a, a confound was included to equate performance across conditions, and a potential explanation for finding no difference in encoding phase response time and recognition memory test decision time was due to the inclusion of this confound. In Experiment 4b, the confound was removed, and ProM performance was better in the high overlap condition compared to the low overlap condition consistent with the typical findings from research using similar processing overlap manipulations (e.g., Meier & Graf, 2000). Any replication of the findings from Experiment 4a therefore cannot be attributed to the inclusion of the confound.  The finding that encoding phase responding was slower for the first word immediately following a ProM task response in both the high and low overlap conditions suggests that processing overlap manipulations do not influence task-set configuration processes. In the ProM literature, plan execution under conditions of low processing overlap is assumed to involve controlled processes (see Einstein & 114  McDaniel, 2005). Given that task-set configuration is also thought to require executive control processes (see Monsell, 2003), the lack of difference between processing overlap conditions was unexpected, however it replicates the outcome observed in Experiment 4a. The combined findings strengthen the view that reconfiguration of the task-set relevant to the ongoing activity is not influenced by manipulations of processing overlap.  Recognition test hit rate was at or near chance in both the high and low overlap conditions, which limited my ability to draw conclusions about the influence of processing overlap manipulations on task-set inertia, and potentially task-set reconfiguration. Consistent with levels of processing, the low hit rate on the recognition memory test is likely accounted for by the fact that ongoing activity words were processed primarily on a perceptual level (Craik & Lockhart, 1972). Also consistent with this explanation is the finding that ProM cues that elicited a planned task response in the low overlap condition, which must have been processed semantically in order to be identified as birds, were recognized significantly more accurately than all other word types or positions. In Experiment 4a, I explained the greater degree of impairment on the recognition test in the high overlap condition by the fact that task-set inertia was particularly relevant under conditions of high processing overlap. However, with hit rate at chance levels for the words that immediately followed a planned task response in both processing overlap conditions, I cannot make inferences about whether processing overlap manipulations influence task-set inertia in the context of a perceptual ongoing activity.  The floor effects observed for the recognition memory test performance in Experiment 4b make the decision speed results difficult to interpret. Consistent with what was observed in Experiment 4a, participants required more time to make correct recognition test decisions for the words that had followed a planned task response during the encoding phase in both conditions, however in Experiment 4b this effect was longer lasting in the low overlap condition. The explanation offered for this finding in Experiment 4a 115  was that the time required for making correct recognition decisions was a result of the additional time needed to retrieve contextual information about each word. However, with performance at floor levels, participants were potentially guessing for most responses, so it is unclear whether correct decisions were based the retrieval of contextual information from the encoding phase. The findings from Experiment 4b provide support for the idea that the processing overlap manipulation used does not influence task-set configuration processes in the context of a perceptual ongoing activity. However, given the floor effects on the recognition test, the influence of this manipulation on task-set inertia in the context of a perceptual ongoing activity remains unclear. A challenge for future research would be to address this problem of floor effects when processing is perceptual, perhaps by having participants study geometric shapes rather than words.  General Discussion The goal of Experiments 4a and 4b was to explore variables that might influence the proactive effects I assume to be the result of response switching processes. The first variable I explored was processing overlap. In both experiments, for the high processing overlap condition ProM task cues required the same kind of processing as ongoing activity stimuli, and for the low overlap condition ProM task cues required a different kind of processing from the ongoing activity stimuli. This processing overlap manipulation was used, in part, because it is the most commonly used processing overlap manipulation in the ProM literature (cf. McDaniel & Einstein, 2007). Given the exploratory nature of the research conducted in Experiments 4a and 4b, a logical starting point was to use a well-established processing overlap manipulation. While this processing overlap manipulation is most relevant to the cue noticing and/or recognition of cue-plan relevance stages of ProM retrieval, I was interested in determining whether it influenced task-set configuration and task-set inertia likely involved in response switching. Consistent with what I observed in Experiment 1, I found that encoding phase responses were made more slowly, correct 116  recognition test decisions were made more slowly, and recognition test performance was impaired for ongoing activity words that immediately followed successful execution of the planned task. Critically, the latter effect was influenced by the processing overlap manipulation, while the former effects were not. The theoretical implications of these findings are discussed below.  The finding that the manipulation of processing overlap influenced the proactive effects associated with task-set inertia but not the effects associated with task-set configuration processes underscores the complexity of response switching during ProM retrieval. The results of Experiment 4a and 4b suggest that we cannot assume the consequences of plan execution are consistent across different contexts. Specifically, the findings suggest that after a response switch, it is more difficult to reengage in an ongoing activity that shares a greater degree of processing overlap with the ongoing activity likely because the ProM task-set is more likely to remain active under conditions of high processing overlap.  An important consideration to come out of Experiments 4a and 4b is the benefit of having two measures of the proactive effects associated with response switching, response time and recognition memory test performance. To measure the cost of switching from one response to another, research in both task-switching (see Monsell, 2003 for a review) and the limited research in ProM (Meier & Rey-Mermet, 2012; 2017) have focused almost exclusively on response time, and occasionally on decision errors (e.g., in a lexical decision task; see Gade & Koch, 2007). Using the method established in Experiment 1, and carried through to Experiments 4a and 4b, I was able to disentangle response time effects that I assume are due primarily to task-set configuration processes, and encoding difficulties that I assume are a result of the construct task-set inertia. The pattern of proactive effects for the high overlap and low overlap conditions highlights the sensitivity of the method I have used throughout this dissertation research to disentangle these different mechanisms.  117  To the best of my knowledge, no other research has directly examined the influence of processing overlap manipulations on the mechanisms involved in response switching during ProM retrieval. As such, the conclusions drawn from Experiments 4a and 4b must be accepted with caution. This initial exploratory research used only one type of processing overlap, and the kind of processing overlap used in the experiments reported in this chapter is likely most pertinent to cue noticing or recognition of cue-plan relevance. What remains unclear is how processing overlap manipulations more directly relevant to response switching would influence task-set configuration and task-set inertia. For example, if the ongoing activity required regular switching between tasks, including the ProM task, it is conceivable that fewer proactive effects might be observed, given that task-set configuration processes are regularly engaged throughout the ongoing activity. Additionally, a confound was introduced in Experiment 4a in order to equate performance across processing overlap conditions. Future research might attempt be to equate the performance across processing overlap conditions without including a confound in the procedure. This goal might be achieved by making the ProM task in the high processing overlap condition as challenging as it is in the low overlap condition. For example, in the context of an ongoing activity requiring semantic processing, participants could be asked to categorize words based on less familiar rules (e.g., identify words that fall within a grammatical category such as adverbs) or using less typical category members (e.g., emu) in the high overlap condition. A final limitation of the research presented in this chapter is that the perceptual ongoing activity in Experiment 4b unintentionally resulted in floor effects on the recognition memory test, and limited my ability to draw conclusions based on recognition test performance. Future research might focus different kinds of ongoing activities, such as one involving orthographic processing that would be less likely to result in recognition test floor effects.    118  5. Chapter 5: The Influence of Subliminal Primes on the Proactive Effects of Response Switching Research has shown that successful completion of a prospective memory (ProM) task is augmented if the display of a ProM task cue is preceded by a related stimulus that primes the processing of the cue (Ellis, Burkes & Milne, 1997; Meier, Zimmermann & Perrig, 2005; Taylor, Marsh, Hicks & Hancock, 2004). Researchers have used several different kinds of primes, such as words that are orthographically similar to the cue (e.g., words beginning with the same letters as the cue) or words that are semantically related to the cue (e.g., words from the same semantic category). These primes are often presented as part of the ongoing activity. For example, a word that is semantically related to the cue might be presented during the course of the ongoing activity where participants are rating the pleasantness of words (see Taylor et al., 2004). In other experiments, primes have been presented subliminally immediately prior to the presentation of the cue (e.g., Graf, 2005). The goal of Experiment 5 was similar to that of Experiments 4a and 4b, that is, to investigate variables that might affect response switching processes. In Experiment 5, I explored the influence of different kinds of subliminal primes, repetition primes, category primes and unrelated primes, on the proactive effects associated with response switching. The priming manipulations used in Experiment 5 were another way of gaining a better understanding of variables that affect the task-set configuration (Monsell, 2003) and task-set inertia (Wylie & Allport, 2000) processes that were used to explain the proactive effects reported in the preceding chapters.  ProM researchers have explained that primes facilitate ProM task execution in two ways. Consistent with the multiprocess framework (McDaniel & Einstein, 2000), many researchers have attributed the benefits of priming the ProM cue to the ease with which the representation of the plan is activated (e.g., Taylor et al., 2004), suggesting that primes increase the likelihood of spontaneous or automatic retrieval of the plan’s representation (e.g., Meier et al., 2005). Alternatively, Graf (2005) suggested that after a prime is presented, the processing of the cue is likely faster or easier than expected, which triggers a discrepancy 119  reaction (cf. Whittlesea & Williams, 2001a). This reaction encourages a deeper analysis of the current stimulus, which stalls the default ongoing activity response and allows the opportunity for the ProM task response to be executed (Graf, 2005).  It is possible that different kinds of primes, such as repetition primes or category primes, have different influences on ProM retrieval stages. However, to the best of my knowledge, none of the existing research has used primes that were intended to modulate the distinct stages of ProM task retrieval. As a consequence, it is not clear whether all kinds of primes have the same or different effects on cue noticing, recognition of cue-plan relevance, and response switching. It is also unclear whether the explanations that have been offered to explain priming effects on overall ProM task performance (see Graf, 2005; Meier et al., 2005; Taylor et al., 2004) can be generalized to all priming manipulations. Finally, and most relevant to the goal of the Experiment 5, we do not yet know whether all kinds of primes have the same effects on the task-set configuration and task-set inertia processes that are assumed to be involved in response switching from the planned task to the ongoing activity. I propose that a repetition or identity prime, i.e., a prime that is identical to the cue word itself, is most likely to facilitate the cue noticing stage of ProM retrieval and to a lesser extent recognition of cue-plan relevance. Repetition primes are assumed to make a word’s representation more readily accessible (Marcel, 1983), and I have assumed that activation of the representation of the cue word is necessarily for cue noticing. Consistent with the assumptions of spreading activation (Anderson, 1983; Collins & Loftus, 1975), activation of a cue’s memory representation will also spread to the category within which the cue belongs (i.e., birds), and eventually to the representation of the plan associated with that category. In this way, repetition primes also might facilitate recognition of cue plan relevance, albeit to a much lesser degree. Further, given the bivalent nature of ProM task cues, the representation of the cue is also relevant 120  to the ongoing activity response. It is therefore possible that repetition primes could facilitate the speed or accuracy of the ongoing activity response instead of the ProM task response.  In contrast to repetition primes, I propose that a prime that activates the category to which a cue belongs likely influences the recognition of cue-plan relevance stage of ProM retrieval, and to a lesser extent the cue noticing stage. ProM research concerned with priming supports this possibility. Meier, Zimmerman and Perrig (2006) presented semantically related primes as part of the ongoing activity (e.g., music stand when cues belonged to the category musical instrument). They found that presenting semantically related words prior to the cue word itself increased the phenomenological experience of the plan being spontaneously retrieved. Other researchers have found similar effects on overall ProM task performance when presenting semantically related primes (e.g., Ellis et al., 1997; Taylor et al., 2005). ProM researchers have attributed these findings to the likelihood that when a prime successfully activates the representation of the category to which the cue belongs, that activation presumably spreads to all associated representations (see Anderson, 1983) including the representation of the plan, which facilitates recognition of cue-plan relevance, and eventually to representation of the cue word itself. In this way, cue noticing might also be facilitated by category primes, but to a lesser degree than recognition of cue-plan relevance.  What remains unclear is how different types of primes influence the processes required for response switching during ProM retrieval, and Experiment 5 was motivated by the goal of exploring how priming manipulations influence these processes. Consistent with this goal, Experiment 5 included subliminal primes that were presented immediately prior to the ProM task cue. As in previous experiments, in Experiment 5, participants began with an encoding phase, where they were required to answer semantic questions about words. The encoding phase also included the ProM task, and the ProM task cues were immediately preceded by one of three different kinds of primes. I subliminally presented repetition primes to 121  primarily facilitate cue noticing, category primes to primarily facilitate recognition of cue-plan relevance, and unrelated primes to serve as a ProM control condition. I used this priming manipulation in part because this research is exploratory in nature, and I wanted to test the possibility that different primes influence different stages of ProM retrieval and potentially have different effects on the processes involved in response switching. Given that repetition and category primes are assumed to modulate different stages of ProM retrieval, it is possible that they also have different influences on the proactive effects attributed to task-set configuration and task-set inertia. Repetition primes activate the representation of the cue, which might facilitate the configuration of the ongoing activity task-set and reduce the duration and size of the proactive effects associated with response switching. In contrast, category primes activate the representation of the plan, which might facilitate the persistent activation of the ProM task-set and increase the proactive effects attributed to task-set inertia. I also used the priming manipulation to challenge the general assumption that primes always facilitate overall ProM task performance.  Experiment 5 explores these possibilities.  Experiment 5 The basic method of Experiment 5 was the same as the method used for Experiment 1, but this method was modified to include subliminal primes that were displayed immediately prior to the presentation of the ProM cues. Primes were either repetition primes, category primes or unrelated primes. I included repetition and category primes to test the possibility that different prime types influence different stages of retrieval, and therefore have potentially different effects on response switching and overall ProM performance. The unrelated prime condition served as a control condition for the ProM task.  Method  In order to explore the influence of different kinds of subliminal primes on the proactive effects attributed to response switching, Experiment 5 included a within-subjects priming manipulation. Each participant was exposed to two ProM cues preceded by repetition primes (e.g., eagle for the cue word 122  eagle), two ProM cues preceded by the category label (e.g., bird for the cue eagle), and two ProM cues preceded by unrelated primes (e.g., mirror for the cue eagle). There was also a control condition in which six ongoing activity words from one category were also preceded by primes. An additional six ongoing activity words from another category were displayed without primes to serve as non-primed control words. This condition was included to verify the effectiveness of the priming manipulation.  Participants. The participants were 93 undergraduate student volunteers. They received partial course credit for their participation in the study, which was approved by the ethics review board. Each participant provided informed written consent at the beginning of the experiment. Materials. I required a total of 316 words for this experiment. Of this total, 280 words were randomly selected from the Toronto Word Pool (Friendly et al., 1982). Twelve words representing kinds of birds served as ProM cues. Unrelated primes were nouns unrelated to the cue category, also selected from the Toronto Word Pool (Friendly et al., 1982). An additional 24 words served as encoding phase control words, half of which were also primed. The control words came from the categories fruits and musical instruments, and were selected based on prototypicality norms by Uyeda and Mandler (1980). The words representing birds, fruits and musical instruments were matched for length and frequency (Brysbaert & New, 2009). The 316 words were divided into two sets, Set A and Set B. Each set consisted of 140 randomly selected words from the Toronto Word Pool (Friendly et al., 1982), six words representing birds, six words representing fruits and six words representing musical instruments. In Set A, fruit words served as primed controls, and musical instrument words served as control words that were not preceded by a prime. In Set B, musical instrument words served as primed controls and fruit words served as control words that were not preceded by a prime.  The primes in the experiment were preceded and followed by a mask, which was a letter string displayed with the mirror image of those letters superimposed on top of the original letter string (see Figure 123  5.1). The mask was generated with Microsoft Paint and remained on the screen for 250 ms before and after the prime.  Figure 5.1. Depiction of mask stimulus. Procedure. The basic procedure was similar to Experiment 1, except for the inclusion of the priming manipulation. The experiment was arranged into two phases: An encoding/ProM task phase and the recognition memory testing phase. Each participant saw two repetition primes, two category primes and two unrelated primes immediately preceding cue and control words. The repetition primes were identical to the ProM cue/control words, the category primes were the label for the category to which the cue/control words belonged (e.g., bird, fruit, instrument), and the unrelated primes were unrelated nouns selected at random from the Toronto Word Pool (Friendly et al., 1982). I created six different arrangements of the materials, so that each bird word, each fruit word, and each musical instrument word was paired with each kind of prime at least once.   Prior to Phase 1, participants were provided instructions for the ongoing activity and the ProM task both orally and in written form via the experiment program. These instructions were similar to those provided in Experiment 1, except that participants were also informed that they would see a series of letters flashed quickly prior to the display of each word, and that these letters were meant to prepare them for the next trial. They were given practice on both the ongoing activity and the ProM task. The instructions for the ProM task were the same as in Experiment 1: If ever in the course of [the ongoing activity] you see a word that refers to a kind of bird, press the q-key. Phase 1 consisted of 158 trials. Each trial began with a fixation cross displayed in the centre of the screen for 500 ms. The fixation cross was followed by a prime mask, which remained on the screen for  124  250 ms. For 146 of the trials, no prime followed this mask. For each of the six ProM cues and six control trials, a prime was displayed for 40 ms between mask presentations as this duration has been shown to be optimal for subliminal priming effects (Meier, Morger & Graf, 2003). The prime mask was displayed again for 250 ms. Following the post-prime mask, participants were immediately provided with the ongoing activity encoding question (e.g., Does this word refer to something familiar or unfamiliar?) for 500 ms on its own, then along with the word to which they were responding. The question and word remained on the screen for a minimum of two seconds. In Phase 1, all primed words (ProM cues and control words) as well as all cue-surround words were shown with the question Does this word refer to something familiar or unfamiliar?  The procedure for the recognition memory test was modelled after the procedure in Experiment 1. All words from Phase 1 (either Set A or B) and all remaining words participants did not see (either Set A or B) were presented. Participants were asked to respond old (left arrow) if they thought words were shown in the encoding phase or new (right arrow) if they believed they did not see the words earlier in the experiment. Words were presented one at a time, randomly and without replacement.  Results  The main dependent variables of interest were accuracy and speed on the ProM task, speed on the encoding phase task, as well as accuracy and speed on the old/new recognition memory test for ProM task cues, primed non-cue words, and all other ongoing activity words. As in previous experiments, I used 95% confidence intervals for making decisions about differences among the means. All data were screened for outliers. On the recognition test, three participants had median response times that were more than two standard deviations faster than the group mean of 954 ms. In a signal detection analysis of recognition memory test performance, those same participants had d’ scores that did not differ from zero, and/or showed bias in responding, i.e., they responded almost exclusively with old or 125  new decisions to all words. These three participants were likely not following the task instructions and their data were not included in any subsequent analyses.  I began by checking ProM task and recognition memory test performance for floor and/or ceiling effects. On the ProM task, participants correctly identified 76.4% (95% CI: 70.4% to 82.5%) of ProM cues, indicating a lack of ceiling and floor effects. For the recognition test, participants correctly identified 70.6% (95% CI: 68.1% to 73.1%) of old words, and had a false alarm rate of 16.5% (95% CI: 13.9% to 19.2%) on new words. A signal detection analysis indicated that performance was significantly better than chance (d’ = 1.68, 95% CI: 1.57 to 1.79). The mean of participants’ median amount of time for making correct responses to ProM cues was 1873 ms (95% CI: 1688 ms to 2058 ms), which was not significantly different from the time required for making all other ongoing activity responses, (1876 ms, 95% CI: 1753 ms to 1998 ms).  I examined ProM task performance in the different prime conditions. ProM cues preceded by category primes elicited the planned task response more frequently (M = 81.1%, 95% CI: 74.4% to 87.8%) than ProM cues preceded by both repetition (M = 75.0%, 95% CI: 67.4% to 82.6%) and unrelated primes (M = 76.1%, 95% CI: 68.5% to 83.7%), however none of the differences among prime conditions were statistically significant (see Figure 5.2, Panel A). An analysis of the time required for making ProM responses in each prime condition revealed that cues preceded by unrelated primes elicited a planned task response more quickly (M = 1920 ms, 95% CI: 1771 ms to 2068 ms) than cues preceded by repetition primes (M = 2185 ms, 95% CI: 1877 ms to 2494 ms) or category primes (M = 2001 ms, 95% CI: 1768 ms to 2234 ms). The difference between the category and repetition prime conditions was not statistically significant (see Figure 5.2, Panel B).  126   Figure 5.2. ProM task performance (Panel A) and time required for making correct ProM task responses (Panel B) for each prime condition. Error bars represent 95% confidence intervals.   Encoding Phase. I began my analysis by examining the overall pattern of encoding phase response time collapsing across prime conditions. Following a correctly recognized ProM cue (cT), the word on the first trial (cT+1) was responded to significantly more slowly (M = 2423 ms, 95% CI: 2234 ms to 2613 ms) than the cT+2 word (M = 1813 ms, 95%CI: 1642 ms to 1984 ms), cT+3 word (M = 1620 ms, 95% CI: 1482 ms to 1758 ms) and all other non-surround words (M = 1873 ms, 95% CI: 1688 ms to 2058 ms). The findings replicate the pattern of proactive effects on encoding phase response time also observed in previous experiments. Figure 5.3 shows the mean of participants’ median response time to encoding task questions for the cue words, the three words immediately following a correctly recognized cue for each of the prime conditions, as well as the response time data for all other non-surround words. Results show no significant differences in response time among the prime conditions for the cT+1 words (see Figure 5.3). For “Missed” Cues (i.e., cues that elicited an ongoing activity response and not a ProM task response), the response time was significantly slower for cues preceded by unrelated primes (M = 2164 ms, 95% CI: 1886 ms to 50%55%60%65%70%75%80%85%90%95%100%Repetition Category UnrelatedProM Task PerformancePrime Type500750100012501500175020002250250027503000Repetition Category UnrelatedProM TaskResponse Time (ms)Prime TypeA) B) 127  2442 ms) than cues preceded by category primes (M = 1900 ms, 95% CI: 1663 ms to 2138 ms), but not significantly different than cues preceded by repetition primes (M = 1933 ms, 95% CI: 1614 ms to 2251 ms).   Figure 5.3. Mean of participants’ median response time (in ms) to encoding phase cue words and cue-surround words for the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime condition (grey bars), as well as non-surround words that had not been preceded by subliminal primes (black bar). The RT’s for cue-surround words are only for words following cues that elicited a planned task response. The RT’s for cue words are only for those trials on which the ongoing activity response was provided for the cue. Error bars represent the 95% confidence intervals.  I also examined the speed of responding to prime control words. Consistent with previous research (see Marcel, 1983), I expected that words preceded by repetition and category primes would be responded to more quickly than non-primed matched control words. The results from the experiment showed that words preceded by repetition primes (M = 1783 ms, 95% CI: 1659 ms to 1907 ms) were responded to significantly more quickly than non-primed matched controls (M = 1920 ms, 95% CI: 1802 ms to 2037 ms). However, the time required for responding to control words preceded by category primes (M = 2026 ms, 500750100012501500175020002250250027503000"Missed" Cue +1 +2 +3 Non-SurroundEncdoing QuestionResponse Time (ms)Word Type/PositionRepetitionCategoryUnrelated128  95% CI: 1840 ms to 2212 ms) and unrelated primes (2138 ms, 95% CI: 1914 ms to 2362 ms) did not differ significantly from non-primed matched controls.  Recognition Test Phase. The overall pattern of recognition test results for cue words, cue-surround words, and non-surround words demonstrated that actual cue words (M = 81.7%, 95% CI: 77.1% to 86.2%) were recognized significantly better than surround words (M = 66.8%, 95% CI: 60.8% to 72.7%) and non-surround words (M = 71.0%, 95%CI: 68.3% to 73.6%). I also found that recognition test performance of the cT+1 words (M = 63.7%, 95% CI: 57.6% to 69.9%) was significantly lower compared to all non-surround words. The difference in recognition test performance between all non-surround words and the cT+2 words (M = 66.8%, 95% CI: 61.0% to 72.7%) was not statistically significant.  A more detailed analysis examined the specific influence of the different prime conditions on recognition memory test performance. Figure 5.4 shows recognition test performance accuracy for same word types/positions depicted in Figure 5.3. Here I highlight two significant differences in recognition test performance. First, for ProM cues, performance was significantly better for cues preceded by repetition primes (M = 84.4%, 95% CI: 78.1% to 90.7%) as compared to unrelated primes (M = 76.6%, 95% CI: 69.2% to 84.1%). Recognition performance for the cues preceded by category primes did not differ from recognition performance for the cues preceded by repetition or unrelated primes (M = 82.9%, 95% CI: 76.4% to 89.4%). Second, recognition test performance in the category prime condition was impaired for the cT+1 words (M = 59.8%, 95% CI: 50.7% to 68.8%) and cT+2 words (M = 61.6%, 95% CI: 53.5% to 69.7%) compared to all other non-surround words (M = 71.0%, 95%CI: 68.3% to 73.6%). Performance in the repetition prime condition was not significantly different from all other non-surround words for the cT+1 words (M = 69.5%, 95% CI: 60.8% to 78.1%) and cT+2 words (M = 73.4%, 95% CI: 60.8% to 78.1%).  129   Figure 5.4. Mean recognition test performance (hits) for cue words and  cue-surround words in the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime condition (grey bars), as well as non-surround words that were not preceded by primes (black bar). The figure only includes data for those cue words and their corresponding cue-surround words that successfully elicited the planned task response in each of the prime conditions. Error bars represent 95% confidence intervals. Consistent with the pattern of recognition memory tests results for primed cue words, an analysis of recognition memory test performance for the control primed words revealed better performance in the repetition and category prime conditions. Specifically, control words preceded by repetition primes (M = 75.0%, 95% CI: 68.4% to 81.6%) and category primes (M = 73.3%, 95% CI: 66.9% to 79.8%) were correctly recognized more frequently that control words following unrelated primes (M = 68.9%, 95% CI: 62.0% to 75.7%). Recognition memory test performance on repetition and category primed control words was also superior to performance for non-primed matched controls (M =70.4%, 95% CI: 65.7% to 75.0%), however these differences were not statistically significant.  50%55%60%65%70%75%80%85%90%95%100%Cue +1 +2 +3 Non-SurroundRecongition Test Performance (Hit Rate)Word Type/PositionRepetitionCategoryUnrelated130  Figure 5.5 shows the mean of participants’ median response time required for making correct recognition test responses for the same words represented in Figure 5.4. The data demonstrate that the time required for making correct recognition test decisions was significantly longer for all word positions in all three prime conditions as compared to time required for making correct recognition test decisions for non-surround words.    Figure 5.5. Mean of participants’ median response time (in ms) required for making correct recognition decisions for cue words and encoding phase cue-surround words for the repetition prime condition (dark blue bars), category prime condition (light blue bars) and unrelated prime condition (grey bars), as well as non-surround words that were not preceded by subliminal primes (black bar). Error bars represent the 95% confidence intervals.  Discussion The goal of Experiment 5 was to investigate variables that influence the processes assumed to be involved in response switching, namely task-set configuration and task-set inertia. In line with this goal, Experiment 5 investigated the influence of different types of subliminal primes on the proactive effects of plan execution. The experiment compared performance among three conditions, which were defined by the 500600700800900100011001200130014001500Cue +1 +2 +3 Non-SurroundRecognition TestResponse Time (ms)Word Type/PositionRepetitionCategoryUnrelated131  use of repetition primes, category primes, and unrelated primes (i.e. a prime control condition). Prime type was manipulated within-subjects, and for this reason, Experiment 5 included only two ProM trials in each prime type condition. As a consequence, encoding phase response time, recognition memory test performance and recognition test decision speed showed substantially more variability than occurred in the preceding experiments, and only a few findings achieved significance. Nevertheless, the findings did show a pattern of effects that is generally consistent with predictions. First, performance on the ProM task was higher in the category prime condition compared to the other prime conditions. Second, encoding phase responses were made more slowly on the ongoing activity trial following each successful ProM task response, and there was no difference in the amount of slowing across the different prime type conditions. Third, findings also showed an impairment on recognition test performance for the ongoing activity words which occurred immediately after each successful ProM task response, and this impairment on the recognition test was greater in the category prime condition compared to the other priming conditions.  The finding that overall ProM task performance was higher and ProM task responses were faster in the category prime condition compared to the repetition prime condition is consistent with the idea that the two prime types facilitate different stages of ProM retrieval. Category primes are likely to facilitate both cue noticing and recognition of cue-plan relevance. When the category bird is primed, activation is assume to spread from this category to all associated representations (Anderson, 1983; Collins & Loftus, 1975), including the cue word and the representation of the plan. The activation of the representation of the plan is likely critical for the configuration of the ProM response task-set (see Monsell, 2003), and therefore the category prime will more frequently result in execution of the planned task rather than the ongoing activity response. In contrast, repetition primes are likely to activate the representation of the cue word (Marcel, 1983), and this representation is also relevant to the ongoing activity. It is for this reason that repetition primes might at times facilitate the configuration of the ongoing activity response.   132  The finding that encoding phase response time was significantly slower for the first word (cT+1), and to a lesser extent the second word (cT+2), immediately following a planned task response replicates the results from Experiments 1 through 4b. In preceding experiments, I attributed slowing on the cT+1 trial to the need for task-set reconfiguration of the ongoing activity response. The finding that the same pattern of results for encoding phase response time occurred across the three prime type conditions (i.e., repetition, category and unrelated) suggests that subliminally presented primes have no effect on this reconfiguration process.  The finding that recognition was impaired on the cT+1 and cT+2 words in the category prime condition but not the repetition prime condition suggests that the priming manipulation influenced the construct of task-set inertia. The category prime likely activated the representation of the plan more quickly than other prime conditions (Meier et al., 2006; Taylor et al., 2004), facilitating recognition of cue-plan relevance. The activation of the representation of the plan may have resulted in stronger and longer lasting activation of the task-set relevant to the ProM response. This persistent activation of the ProM task-set, i.e., task-set inertia (Allport et al., 1994) increased the recognition test impairment due to the limited resources available for encoding processes. The repetition prime, on the other hand, likely activated both the task-set relevant to the ProM response and the task-set relevant to the ongoing activity, and task-set inertia was therefore less impactful in this condition.   In all prime conditions, participants required more time to make correct recognition decisions for the words that followed a planned task response during encoding, and this finding is consistent with results from previous experiments. This finding suggests that the priming manipulation does not influence task-set configuration processes. This finding further supports the explanation provided in Experiment 4a and 4b that the configuration of an old response for words that had followed a planned task response during 133  encoding required more time because those words were processed more superficially. As a result, the contextual information required to configure correct old responses was not readily accessible.  The findings from the exploratory research conducted in Experiment 5 provide encouraging support for the idea that different kinds of primes likely facilitate different stages of ProM retrieval and challenge the general assumption that primes always facilitate ProM task performance. The results also suggest that different types of primes may have different influences on the mechanisms I assume to be involved in response switching. However only two ProM trials were included in each prime condition as compared to six ProM trials in the preceding experiments. The limited number of ProM trials per prime condition led to issues with statistical power and limited my ability to draw strong conclusions from the results. The findings from Experiment 5 must be replicated under conditions of greater statistical power. Future research might address power issues by manipulating the priming conditions between-subjects, increasing the number of participants threefold, or by including more ProM trials in each condition. With regard to this latter suggestion, care must be taken to ensure ProM cue trials occur very rarely to allow the plan to be outside of conscious awareness throughout the course of the ongoing activity (Graf & Utll, 2001).    134  6. Chapter 6: General Discussion Prospective memory (ProM) is the cognitive function required for forming, maintaining, and executing a planned task at the appropriate time or upon the occurrence of an appropriate cue. This cognitive function is required for a number of daily activities, such as remembering to stop and purchase groceries while driving home from work. In laboratory ProM tasks, participants might be engaged in an activity that requires making semantic judgments about words (e.g., familiarity judgments), and the planned task might be to make a unique response (e.g., press the q-key) to words belonging to a specific category (e.g., types of birds). These examples highlight one specific kind of ProM task, an episodic event-based ProM task. This kind of task must be executed once or only very few times upon the occurrence of a specific environmental cue (e.g., the grocery store, the word eagle). In contrast to episodic retrospective memory tasks, very little is known about the processes involved in episodic, event-based ProM tasks. The overarching goal of this dissertation was to increase our understanding of the cognitive processes involved in ProM retrieval.  ProM tasks are uniquely challenging, in part, because they typically take place in a dual-task context. The dual-task context is demonstrated in the examples provided above. While engaged in the activity of driving (monitoring traffic, driving speed, cyclists, etc.), we must remember and execute the plan to go shopping when we pass the grocery store. While evaluating the familiarity of a list of words displayed as part of a laboratory task, we must remember and execute the plan to press the q-key whenever we see a word representing a kind of bird. The dual-task context within which ProM tasks typically occur means that ProM retrieval involves a number of distinct stages. The first stage is cue noticing, by which I mean it must be processed by the sensory system and perceived as a distinct entity. If the plan is to go grocery shopping while driving home from work, we must process the grocery store with our visual system and we must identify the store as a separate unit from other buildings. The second stage is recognition of cue-plan 135  relevance, meaning the representation of the plan must be activated in connection with that cue. The third and final stage is response switching, where we must interrupt the ongoing activity within which we are currently engaged and switch to the planned task response. Exactly how the switch between these two tasks occurs is still unclear. The research presented in this dissertation was guided by the goal of gaining a better understanding of the cognitive processes involved in response switching during ProM retrieval.  The processes involved in response switching are still poorly understood, in part, because most of the previous research on ProM has been guided primarily by two theoretical frameworks that were intended to account for overall performance on ProM tasks. The preparatory attention and memory processes (PAM) framework, proposed by Smith (2003), states that ProM performance is guided primarily by strategic monitoring processes, and that successful execution of planned task is dependent on an attentional resource demanding search for plan-relevant cues. The multiprocess framework presented by McDaniel and Einstein (2000) agrees that under some circumstances, ProM performance is dependent on strategic monitoring for cues (e.g., when the cue is peripheral to the ongoing activity), but in other circumstances, ProM performance is guided by automatic processes (e.g., when the cue and the planned task are highly associated) (McDaniel et al., 2004). Motivated by these theoretical frameworks, the majority of existing research has focused on overall ProM task performance, for example, by manipulating the relative importance of the ProM task (e.g.,, Kliegel, Martin, McDaniel & Einstein, 2001; Brandimonte et al., 2010; Kvavilashvili, 1987; Walter & Meier, 2016) or the difficulty of the ongoing activity (Hicks et al., 2005; Marsh et al., 2002). A relatively small proportion of the existing research has focused on cue noticing; such studies manipulated the perceptual salience of the cue (e.g., Cohen et al., 2003; Graf et al., 2002; Uttl, 2006). A small number of studies investigated recognition of cue-plan relevance, for example, by manipulating the pre-existing connection between the cue and the planned task (e.g., Cohen et al., 2001; Loft & Yeo, 2007; Pereira et al., 2012). To my knowledge, however, only one research group has looked specifically at the 136  processes involved in response switching (Meier & Rey-Mermet, 2012; 2017). The research reported in this dissertation builds on the seminal work of Meier and Rey-Mermet (2012; 2017) in order to gain a deeper understanding of the cognitive processes involved in response switching.  Guided by this goal of illuminating the cognitive processes involved in response switching, the research reported in this dissertation was inspired by theoretical proposals from the task-switching literature. A number of mechanisms are thought to be involved in task-switching, and I have identified three mechanisms that are potentially relevant to response switching during ProM retrieval. The first is task-set inhibition, which is the process by which the response set required for trials preceding a switch must be inhibited in order for the switch to be successful (Mayr & Keele, 2000). The second mechanism is used for harnessing and organizing the mental resources required for each response set, and is known as task-set configuration (see Monsell, 2003; Rogers & Monsell, 1995). The third mechanism is task-set inertia, which is the tendency for a task-set to persist over time and to proactively interfere with the configuration of a new task-set (see Allport et al., 1994; Wylie & Allport, 2000). One of the most consistent findings attributed to these mechanisms is the cost associated with switching between tasks. This switch cost is typically demonstrated by slower responses on trials that require a switch, as well as on trials that follow a switch compared to trials where the same response is repeated (see Monsell, 2003).  In ProM task retrieval, there are two response switches that must occur. The first switch required for planned task execution is the switch from the ongoing activity response to the ProM task response, and the second is the switch from the ProM task response back to the ongoing activity response. The research presented in this dissertation focused on costs associated with the switch from the ProM task back to the ongoing activity. This choice was motivated by two critical issues. First, ProM task responding in the experiments reported in this dissertation was qualitatively different from ongoing activity responding. The ProM task required responding to an item from a specific category, and the ongoing activity required 137  making semantic judgments about words. Additionally, ProM task responding was quantitatively different from ongoing activity responding. ProM task responses were required very rarely (after every 20 ongoing activity trials) and ongoing activity responding occurred on almost every trial (except for six ProM task trials). The difference in the quantity of responses has two implications. First, rare responses tend to be slower, and second, the small number of ProM responses makes responding more demanding of attentional resources, and less likely to be automatic as compared to ongoing activity responding. Because of the difference in the type of question that is relevant to each kind of response as well as the relative frequency of the two responses, a direct comparison between switching to the ProM task and switching to the ongoing activity is not possible. For these reasons, my research focused entirely on the costs and consequence of response switching on ongoing activity responding. The choice to focus entirely on the consequences of planned task execution on ongoing activity responding is consistent with the seminal research of Meier and Rey-Mermet (2012; 2017).   Research Contributions My dissertation makes three kinds of contributions: theoretical, empirical and methodological. I will briefly summarize those contributions in this section. The theoretical and empirical contributions are described together, followed by a description of the methodological contributions.  Theoretical and Empirical Contributions  This dissertation provides a theoretical framework that allows both an investigation of and explanation for the cognition processes involved in response switching during ProM task retrieval. This framework builds on the seminal work of Meier and Rey-Mermet (2012; 2017) in the ProM field, and on insights from the task-switching literature. Taking inspiration from these sources, I speculated that three mechanisms are likely to be involved in response switching during ProM retrieval: task-set inhibition, task-138  set configuration and task-set inertia. In this section, I describe each mechanism and what the research reported in this dissertation tells us about the mechanism’s relevance to response switching.   Task-set inhibition. My dissertation research investigated whether response switching in event-based episodic ProM tasks involves task-set inhibition, one of the mechanisms known to be critical for response switching in task-switching experiments (see Mayr & Keele, 2000) as well as in the colour word Stroop task (Cohen, Dunbar & McLelland, 1990; Morton & Chambers, 1973; Posner & Snyder, 1975). As described by Mary and Keele (2000), task-set inhibition is the process by which the response set on trials preceding a switch must be inhibited in order for the switch to be successful. Switch costs are thought to occur on switch trials, in part, because more time and attentional resources are required for inhibiting the task-set from a preceding trial (see Monsell, 2003). In the colour word Stroop task, word reading is described as quicker, more automatic, and/or easier than colour naming (see MacLeod, 1991), and therefore requires inhibition in order for the colour naming response to be executed (Cohen, Dunbar & McLelland, 1990; Morton & Chambers, 1973; Posner & Snyder, 1975). My dissertation research was designed to find out whether task-set inhibition occurs in event-based ProM tasks by examining changes in the processing and/or responding to words presented on ongoing activity trials that immediately preceded the switch to the planned task response.  Across the five experiments, I found no evidence that planned task execution influenced the processing of ongoing activity words preceding the planned task response. When the ProM task was inserted into the encoding phase (Experiment 1) or the retrieval phase (Experiment 2) of a memory experiment, when the degree of processing overlap between the ongoing activity and the planned task was high or low (Experiments 4a and 4b), and when ProM cues were preceded by subliminal category and repetition primes (Experiment 5), there was no evidence implicating task-set inhibition. Taken together, the 139  findings from these experiments suggest that task-set inhibition was not required for switching from the ongoing activity to the planned task in the context of the ProM tasks that I used in my research. There are two possible explanations for the combined findings from my dissertation research. One explanation is that the absence of task-set inhibition effects in my research is due to specific aspects of the research method. In my experiments, and characteristic of all research on event-based episodic ProM, the planned task is executed very rarely. Similarly, the cues that signal plan retrieval for event-based episodic ProM tasks occur unpredictably throughout the course of an ongoing activity. Moreover, these two characteristics of event-based ProM tasks highlight the difficulty of ProM retrieval. Planned tasks are outside of conscious awareness for some time while we are otherwise engaged in an ongoing activity, and we are typically not aware of exactly when the planned task must be retrieved and executed. Task-set inhibition might be more relevant in traditional task-switching paradigms, where switches occur much more frequently than switching in ProM tasks (see Monsell, 2003) and are often predictable (see Monsell et al., 2003; Rogers & Monsell, 1995).  A second explanation for the absence of task-set inhibition effects in my research is that time constraints for responding, the kind of stimuli to which participants were responding, and the actual responses required, were similar across all experiments. Because there was a minimum of two seconds between ongoing activity trials, the processing of the ongoing activity words may have been completed before the response set was inhibited. It is possible that the inhibition of the ongoing activity response may have occurred but might not have impacted response time or encoding processes given the additional time available for post-event processing. Additionally, all my experiments involved responding to words, and the ongoing activity was almost always a semantic judgment task. Task-switching research typically uses a combination of stimuli (words, images, orthographic shapes, letters, etc.) and often requires response associated with the perceptual properties of the stimulus (e.g., its location on the computer screen) (see 140  Monsell, 2003). It is possible that if I used combinations of stimuli and/or ongoing activities similar to typical task-switching paradigms, I would find evidence of task-set inhibition during planned task execution.  Task-set configuration and task-set inertia. My dissertation also investigated whether response switching during retrieval in event-based ProM tasks involves two other mechanisms that have been postulated in the task-switching literature: task-set configuration and task-set inertia. Task-set configuration is an attentional resource demanding process whereby mental resources are organized into a specific task-set or action schema in order to guide responding (Monsell, 2003). Task-set configuration is thought to be particularly demanding of attentional resources for a rarely occurring task-set (Waszak, Hommel, & Allport, 2002). Task-set inertia is the tendency for a task-set to persist over time and to proactively interfere with the configuration of a new task-set (Allport et al., 1994; Allport & Wylie, 1999; Monsell, 2003; Wylie & Allport, 2000), and is assumed to be particularly strong for less practiced task-sets because their existence and maintenance is more dependent on executive control and working memory processes (Waszak, Hommel, & Allport, 2002). Given that the ProM task is rarely occurring, I assumed that these mechanisms would be particularly relevant to response switching during ProM retrieval.  The results from the experiments reported in this dissertation provide a substantial amount of support that task-set configuration and task-set inertia are involved in response switching during ProM retrieval. The pattern of results was similar across experiments. In general, response times were slower and recognition memory test performance was reduced for ongoing activity words that immediately followed planned task execution. This outcome occurred when the ProM task was inserted into the encoding phase (Experiment 1) or the retrieval phase (Experiment 2) of a memory experiment, when the degree of processing overlap between the ongoing activity and the planned task was high or low (Experiments 4a and 4b), and when ProM cues were preceded by subliminal category and repetition primes (Experiment 5).  141  The effects I observed across the experiments reported in this dissertation are consistent with the kinds of effects typically observed in task-switching research. Response time is typically slower for trials that require a switch versus trials that do not require a switch (see Monsell, 2003). These switch costs are typically attributed to mechanisms such as task-set configuration, which is slow and demanding of attentional resources (Monsell et al., 2003). In the research reported in this dissertation, response time was slower when switching from the ProM task back to the ongoing activity. This finding can be attributed, in part, to fewer resources being available to process the word’s semantic content, leading to slower response time to the ongoing activity question. Additionally, the attentional resources available for proper encoding of the ongoing activity wordswas also limited, leading to poor recognition memory test performance. The evidence also supports the idea that task-set inertia is likely involved in the switch from the ProM task back to the ongoing activity. Under conditions where there is a high degree of processing overlap between the ongoing activity and the ProM task, the ProM task is more likely to proactively interfere with the configuration of the ongoing activity task-set. Significantly lower performance on the recognition test in the category prime condition compared to the repetition prime condition in Experiment 5 can be attributed to the possibility that category primes keep the ProM task-set active for longer, interfering with the reconfiguration of the ongoing activity task-set and leaving fewer resources available for memory encoding processes.  The effects attributed to task-set configuration and task-set inertia were fairly similar across all experiments. However, with the exception of Experiment 3, my experiments were not designed to look specifically for differences in these effects. One explanation for the consistency of findings is that I used the same basic method for all of my experiments. It is possible that memory encoding processes are particularly vulnerable to the diversion of attentional resources for task-set configuration. Other ongoing activities, such those that require perceptual processing, might not show the same kind of performance 142  deficits as I observed. In fact, the results from Experiment 4b suggests that for a perceptual ongoing activity performance, accuracy is not disrupted by planned task execution in the same way as encoding processes are. More research focused on the consequences associated with planned task execution in a variety of different contexts would address these possibilities.  My dissertation research also showed that some of the effects attributed to task-set configuration and task-set inertia might be incidental to response switching. Instead, these effects might be due to the fact that ProM task execution involved only a small number of distinctive trials. In Experiment 3, I included a perceptually salient stimulus (a word displayed in a different colour font) in place of a ProM task cue. Participants were slower to respond to words immediately following a perceptually distinctive stimulus. However, compared to the effects observed following planned task execution, recognition test performance was less dramatically impacted by the processing of a perceptually distinctive stimulus. These results suggest that an attentional shift mechanism, one that is faster and short-lived compared to task-switching mechanisms, may be engaged for shifting participants’ attention from one task to the next (Rogers & Monsell, 1995). This same mechanism is also activated by stimulus properties such as perceptual distinctiveness. Additional support for the possibility that a faster, short-lived attentional shift mechanism impacts response time but not recognition memory test performance comes from the different time course of the proactive effects following planned task execution. In experiments where a switch between tasks was required, slower ongoing activity response time lasted only one trial, whereas recognition memory test deficits lasted for two trials. This latter effect is more likely due to slower, longer-lasting response switching mechanisms, namely task-set configuration and task-set inertia.   Methodological Contributions  My dissertation focuses attention on the three distinct stages that are required for the successful execution of a planned task, and it showcases a proven method that is available for the detailed 143  investigation of the cognitive processes that are involved in one of these stages, response switching. This method was inspired by the limited existing ProM research that has looked specifically at response switching during ProM retrieval (Meier & Rey-Mermet, 2012; 2017) as well as the task-switching literature (see Monsell, 2003), both of which focused primarily on the costs associated with switching between two responses. Most of the existing ProM research on the costs associated with planned task execution has focused on global costs, such as overall mean response time for ongoing activity trials that contained a ProM task compared to mean response time for the same number of ongoing activity trials that did not contain a ProM task (e.g., Smith, 2003). The method used throughout this dissertation provided a more fine-grained analysis of the specific costs associated with planned task execution.  In the reported experiments, I was able to manipulate and determine the influence of a number of variables on the costs associated with planned task execution. I varied memory processing requirements, perceptual distinctiveness and processing overlap, and I included subliminal primes. I was able to examine three mechanisms that are assumed to be involved in task-switching more broadly and potentially involved in response switching during ProM retrieval: task-set inhibition, task-set configuration, and task-set inertia. Moreover, I was able to dissociate effects that can be partially attributed to an attentional shift mechanism and task-set configuration (i.e., response time effects), versus those more likely to be a result of task-set inertia (i.e., recognition test performance).  In addition to the questions I was able to answer in my dissertation research, the method I developed is versatile, and could be modified to address any number of additional research questions. For example, it could be used to ask whether the predictability of a ProM task influences the retroactive or proactive effects associated with response switching. This question might be addressed by signalling the upcoming occurrence of a cue during the ongoing activity (e.g., including a special symbol or colour for a specific number of trials prior to the presentation of the ProM cue). An alternative question that could be 144  answered is whether time-based ProM tasks result in the same pattern of retroactive and proactive effects as event-based ProM tasks.  Implications A deeper understanding of the processes involved in successful execution of a planned task is likely to require a careful description of the stages involved and the development of methods for investigating each stage. My dissertation is a step toward this goal. I identified the three stages required for planned task execution, cue noticing, recognition of cue-plan relevance, and response switching. I have also speculated about the theoretical mechanisms involved in each stage, and focused in particular on a theoretical framework outlining the mechanisms involved in response switching. Additionally, my review of the ProM literature shows that there is research on some of the stages, particularly cue noticing and recognition of cue-plan relevance. However, there is very little research relevant to response switching and more work is required in order to understand this stage. My dissertation advances this goal as well, by presenting a series of experiments designed to investigate the cognitive processes involved in the response switching stage.  The theoretical framework that guided my research complements the existing general frameworks that have been proposed to account for ProM task performance more generally. To the best of my knowledge, my description of the processes involved in response switching during ProM retrieval is the first attempt to explain in detail the processes involved in this stage. A theoretical framework based on an individual stage in ProM retrieval is a useful complement to theoretical explanations such as the multiprocess framework (McDaniel & Einstein, 2000) and the preparatory attention and memory processes framework (Smith, 2003). These global theoretical frameworks provide invaluable information regarding the processes that are involved in ProM performance more generally, but fail to provide significant detail 145  regarding the processes involved in the more specific and unique components of ProM retrieval (i.e., response switching).  The identification of the unique stages of ProM retrieval, as well as the speculations and assumptions regarding the theoretical mechanisms involved in each stage, provides an opportunity for more focused research questions. For example, under what experimental conditions is a discrepancy attribution mechanism (see Whittlesea & Williams, 2001a) likely to be involved in recognition of cue-plan relevance? Are there variables that uniquely influence response switching, without influencing cue noticing or recognition of cue-plan relevance? These questions are difficult to answer. However, identification of the unique stages associated with ProM retrieval is critical in attempting to tackle them.  A deeper understanding of the processes involved in specific stages of ProM retrieval inspires new avenues for future research in ProM. There are a number of variables that existing research has determined as influences on ProM performance more broadly. These variables include individual difference variables (e.g., participants’ age), task instruction variables (e.g., task importance manipulations), or task-specific variables (e.g., manipulations of task difficulty). Future research might investigate how such variables influence the individual stages of ProM retrieval, including response switching. For example, older adults tend to perform more poorly on laboratory ProM tasks than younger adults (see Rendell et al., 2007). What remains unclear is whether older adults’ poor performance is a result of difficulty with cue noticing, recognition of cue-plan relevance, or response switching. The method developed in this dissertation could address questions regarding age differences (and a variety of other individual differences) specific to response switching during ProM retrieval.  The findings from the research conducted in this dissertation also have practical implications, particularly in terms of the phenomenological experiences associated with executing planned tasks. In the everyday experience of remembering to execute a plan, people might identify strongly with the finding that 146  it takes longer to re-engage in the activity they were engaged in prior to the plan execution. For example, if we remember our plan to return a phone call while in the middle of drafting an email, we might find it difficult to pick up where we left off in the email after having returned the phone call. Additional practical applications of the present research are slightly more challenging. Everyday ProM tasks tend to be much more complex than the types of tasks used in the present experiments. Additionally, people are likely to use coping mechanisms to help them remember to carry out planned tasks (e.g., reminders, calendars, alarms, etc.). This additional support for plan execution has implications for the processes involved in response switching. For example, it is possible that a reminder to execute a planned task reduces some of the burden on attentional resources associated with planned task execution, and might therefore minimize the negative impact on ongoing activity processing. Another interesting difference with planned tasks in daily life is that we typically assign ProM tasks to ourselves, identify our own cues and highlight the importance of the tasks for ourselves. While these variables are not addressed within the research presented in this dissertation, a framework and method for investigating how such variables (e.g., selecting one’s own ProM cue, the relative importance of plan execution) might influence response switching.  Limitations The research presented in this dissertation provides a specific way of examining the processes involved in response switching during ProM retrieval. To my knowledge, only one other research group has looked specifically at the immediate consequences of executing a planned task on ongoing activity processing (Meier & Rey-Mermet, 2012; 2017). It is therefore premature to make definitive statements about the processes that are involved, and processes that are not involved, in response switching during ProM retrieval. Additionally, a large part of my dissertation research was exploratory. More research must be conducted before broad conclusions can be reached regarding the validity of the assumptions made in this dissertation.  147  I used the same basic method for all experiments reported in this dissertation, and this method poses limitations on the conclusions that can be drawn from the findings. One of the methodological constraints throughout the majority of the experiments was that the amount of time before a new trial began throughout the ongoing activity was a minimum of two seconds. Additionally, response speed was not emphasized. It is conceivable that allowing trials to change whenever participants made a response or telling participants to respond as quickly as possible would change the trajectory of the response time slowing effects or of recognition memory test deficits. That said, in Experiment 2 participants completed the ProM task during a recognition test, where speed and accuracy of responding was emphasized and not constrained to a minimum of 2 seconds between trials, and I found a similar pattern of proactive effects. It is also possible that the effects I observed throughout this dissertation research are limited to the specific ongoing activity used. Replication of the findings in the context of different ongoing activities would address this concern. A final challenge with the method used throughout this dissertation, and with any experiment interested in episodic ProM, is that the number of ProM task execution trials is necessarily limited to ensure that the planned task is outside of conscious awareness for a period of time during the ongoing activity. The limited number of ProM task trials reduces the amount of data available on ProM task execution and the statistical power to detect processing changes. This limitation was a particular challenge in Experiment 5, where I may have lacked the statistical power necessary to find significant effects among the experimental conditions.  In the series of experiments presented in this dissertation, I found no evidence to suggest that task-set inhibition is involved in response switching during ProM retrieval. However, effects associated with task-set inhibition have been regularly observed in the task-switching literature (see Mayr & Keele, 2000) as well as in the next-in-line effect (see Brenner, 1973). One explanation for why I failed to find such effects is the time constraint imposed on ongoing activity trial responding, specifically each stimulus remained on the 148  screen for a minimum of two seconds. It is possible that inhibition of the ongoing activity task-set did occur on trials preceding planned task execution, but by the time the ProM cue was presented processing of the ongoing activity words had already been completed. Inhibition of the task-set relevant to the ongoing activity would therefore not interfere with responding to or encoding of the words presented on those trials preceding planned task execution. If ongoing activity trials occurred in more rapid succession, or if presentation was not time constrained, I might find evidence of task-set inhibition. It is also possible that manipulations to make planned task execution more similar to task-switching paradigms or to the next-in-line paradigm would allow me to find evidence of task-set inhibition. For example, participants are able to anticipate their turn in the next-in-line paradigm (Brenner, 1973). If participants were able to anticipate ProM task cues in the context of the experiments presented in this dissertation (e.g., if the ProM task was time-based rather than event-based), processing and/or response changes might be more likely for ongoing activity trials immediately prior to the planned task response.  The experiments presented in this dissertation were all conducted in a laboratory setting, and the tasks are artificial compared to real-world ProM tasks, limiting the generalizability of the findings. In real-world ProM tasks, plan execution typically has higher stakes, and participants would likely be more motivated to successfully complete them. Manipulating the relative importance of the ProM task within the context of the methods used in the present experiment could address this limitation. Additionally, in real-world ProM tasks, participants have the benefit of additional external cues to help signal plan retrieval, such as reminders, day planners, calendars or alarms. Using the same basic method from the experiments presented in this dissertation, participants might be allowed to select their own ProM task cue, which could address this limitation. Finally, all participants were undergraduate university students, the majority of who were between the ages of 18-22 years old. Therefore, the results may not generalize to a broader population. It is unclear to what extent this age constraint might influence response switching processes.  149  The research reported in this dissertation shows basic mechanisms that seem to be involved in response switching during ProM task retrieval, namely task-set configuration and task-set inertia. I was also able to identify an attentional shift mechanism that may account for some of the effects observed across the experiments. More research is required to determine whether the findings replicate across different kinds of ProM tasks (e.g., time-based tasks), different ongoing activities (e.g., lexical decision tasks). Additionally, much of the research reported in this dissertation was exploratory in nature. For example, Experiments 4a and 4b examined the influence of one specific processing overlap manipulation on the consequences of planned task execution. The manipulation I used is most likely to impact the cue noticing stage of ProM retrieval, and future research might look at a processing overlap more directly pertinent to the response switching stage. The research reported in this dissertation was able to demonstrate a valid method to examine these response switching mechanisms, and the variables that might influence them. Conclusions The overall goal of my dissertation research was to identify the cognitive mechanisms that underlie our ability to switch from an ongoing activity to a ProM task and back again. We complete many tasks every day that require us to execute a plan while we are otherwise engaged. Previously, very little work focused on the response switching stage of ProM retrieval. Overall, my dissertation shows that there are significant costs associated with switching back to an ongoing activity after having executed a planned task.   The results of the experiments reported in the preceding chapters indicate that task-set configuration and task-set inertia are likely relevant mechanisms for response switching in ProM retrieval. When switching back to an ongoing activity from a planned task, participants must reconfigure the ongoing activity response, and this reconfiguration is particularly difficult because of interference from the ProM task-set. As a result, fewer attentional resources are available for ongoing activity processing. The experimental method developed in this dissertation leaves substantial opportunities to further investigate 150  variables that might influence the identified mechanisms. Further research is required in order to determine whether the consequences associated with lab-based ProM task execution are similar to consequences of executing a planned task in the context of everyday life, where tasks are more complex, we have more experience with them, more coping mechanisms available, and may be more motivated to successfully execute our plans.    151  References Allport, D. A., Styles, E. A., & Hsieh, S. (1994). Shifting intentional set: Exploring the dynamic control of tasks. In C. Umiltà & M. Moscovitch (Eds.), Attention and performance series. Attention and performance 15: Conscious and nonconscious information processing (pp. 421-452). Cambridge, MA, US: The MIT Press. Allport, A., & Wylie, G. (1999). Task-switching: Positive and negative priming of task-set. In G. W. Humphreys, J. Duncan, & A. Treisman (Eds.), Attention, space, and action: Studies in cognitive neuroscience (pp. 273-296). New York, NY, US: Oxford University Press. Anderson, J. R. (1974). Retrieval of propositional information from long-term memory. Cognitive Psychology, 6(4), 451–474. doi: 10.1016/0010-0285(74)90021-8  Anderson, J. R. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press.  Baars, B. J., & Franklin, S. (2003). How conscious experience and working memory interact. Trends in Cognitive Sciences, 7(4), 166-172. doi: 10.1016/S1364-6613(03)00056-1 Baddeley, A. (1992). Working memory. Science, 255(5044), 556-559. doi: 10.1126/science.1736359 Ballhausen, N., Rendell, P. G., Henry, J. D., Joeffry, S., & Kliegel, M. (2015). Emotional valence differentially affects encoding and retrieval of prospective memory in older adults. Aging, Neuropsychology, and Cognition, 22(5), 544-559. doi: 10.1080/13825585.2014.1001316 Bisiacchi, P. S., Schiff, S., Ciccola, A., & Kliegel, M. (2009). The role of dual-task and task-switch in prospective memory: Behavioural data and neural correlates. Neuropsychologia, 47(5), 1362-1373. doi: 10.1016/j.neuropsychologia.2009.01.034 Bond, C. F. (1985). The next-in-line effect: Encoding or retrieval deficit? Journal of Personality and Social Psychology, 48(4), 853-862. doi: 10.1037/0022-3514.48.4.853 152  Bond, C. F., Pitre, U., & van Leeuwen, M. D. (1991). Encoding operations and the next-in-line effect. Personality and Social Psychology Bulletin, 17(4), 435-441. doi:  10.1177/0146167291174012 Brandimonte, M. A., Ferrante, D., Bianco, C., & Grazia Villani, M. (2010). Memory for pro-social intentions: When competing motives collide. Cognition, 114(3), 436-441. doi: 10.1016/j.cognition.2009.10.011 Brandimonte, M. A., & Passolunghi, M. C. (1994). The effect of cue-familiarity, cue-distinctiveness, and retention interval on prospective remembering. The Quarterly Journal of Experimental Psychology, 47A(3), 565-587. doi: 10.1080/14640749408401128 Brem, A. K., Ran, K., & Pascual-Leone, A. (2013). Learning and memory. Handbook of Clinical Neurology, 116, 693-737. doi:10.1016/B978-0-444-53497-2.00055-3 Brenner, M. (1973). The next-in-line effect. Journal of Verbal Learning and Verbal Behavior, 12(3), 320-323. doi: 10.1016/S0022-5371(73)80076-3 Brysbaert, M. & New, B. (2009). Moving beyond Kučera and Francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. Behavior Research Methods, 41(4), 977-990. doi: 10.3758/BRM.41.4.977  Burgess, P. W., Gonen-Yaacovi, G., & Volle, E. (2011). Functional neuroimaging studies of prospective memory: What have we learnt so far? Neuropsychologia, 49(8), 2246-2257.  doi: 10.1016/j.neuropsychologia.2011.02.014 Busemeyer, J. R., & Townsend, J. T. (1993). Decision field theory: A dynamic-cognitive approach to decision making in an uncertain environment. Psychological Review, 100(3), 432-459. Recovered from: http://psycnet.apa.org/buy/1993-40080-001  Cherry, K. E., Martin, R. C., Simmons-D’Gerolamo, S. S., Pinkston, J. B., Griffing, A., & Gouvier, Wm. D. (2001). Prospective remembering in younger and older adults: Role of the prospective cue. Memory, 9(3), 177-193. doi:  10.1080/09658210143000092 153  Cohen, A., Dixon, R. A., Lindsay, D. S., & Masson, E. J. (2003). The effect of perceptual distinctiveness on the prospective and retrospective components of prospective memory in young and old adults. Canadian Journal of Experimental Psychology, 57(4), 274-289. doi: 10.1037/h0087431 Cohen, J. D., Dunbar, K., & McClelland, J.L. (1990). On the control of automatic processes: A parallel distributed processing account of the Stroop Effect. Psychological Review, 97(3), R332-361. doi: 10.1037/0033-295X.97.3.332 Cohen, A., West, R., & Craik, F. I. M. (2001). Modulation of the prospective and retrospective components of memory for intentions in younger and older adults. Aging, Neuropsychology, and Cognition, 8(1), 1-13. doi: 10.1076/anec.8.1.1.845 Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82(6), 407-428. doi: 10.1037/0033-295X.82.6.407 Cooks, G. I., Marsh, R. L., Hicks, J. L., & Martin, B. A. (2006). Fan effects in event-based prospective memory. Memory, 14(7), 890-900. doi: 10.1080/09658210600816079 Craik, F. I., Govoni, R., Naveh-Benjamin, M., & Anderson, N. D. (1996). The effects of divided attention on encoding and retrieval processes in human memory. Journal of Experimental Psychology: General, 125(2), 159-180. doi: 10.1037/0096-3445.125.2.159 Craik, F. I., & Rabinowitz, J. C. (1985). The effects of presentation rate and encoding task on age-related memory deficits. Journal of Gerontology, 40(3), 309-315. doi: 10.1093/geronj/40.3.309 Craik, F. I. M. (1986). A functional account of age differences in memory. In F. Klix & H. Hagendorf (Eds.), Human memory and cognitive capabilities (pp. 409-422). Amsterdam: North-Holland. Crease Lark, M. L., Jamieson, R. K., & Graf, P. Cue contamination: Consequences of prospective memory task response switching. Manuscript in preparation.   154  Crease, M. L., Graf, P. & Jamieson, R. K. (2014). Bird is the word: The next-in-line effect and prospective memory. Poster presented at the Canadian Psychological Association's 75th Annual Convention, Vancouver, BC, June 5-7. Crease Lark, M. L., Graf, P. & Jamieson, R. (2015). Cue contamination: Prospective memory cues disrupt retrospective memory retrieval. Poster presented at the 25th annual meeting of the Canadian Society for Brain, Behavioural and Cognitive Science, Ottawa, ON, June 5-7. Crease Lark, M. L., Graf, P. & Jamieson, R. (2015). Prospective memory cues contaminate memory for surrounding stimuli. Poster presented at the 11th biennial meeting of the Society for Applied Research in Memory and Cognition, Victoria, BC, June 24-27.  Crease Lark, M. L., Jamieson, R., & Graf, P. (2016). Differential effects of focal and non-focal cues on ongoing task performance in prospective memory. Poster presented at the 77th CPA Annual Convention, Victoria, BC, June 9-11.   Crease Lark, M. L., Jamieson, R., & Graf, P. (2018). Killing two birds with one Prime. Poster presented at the 28th annual meeting of the Canadian Society for Brain, Behavioural and Cognitive Science, St John’s, NL, July 4-8.   Crease Lark, M. L., Jamieson, R., & Graf, P. (2019). Repetition primes slow prospective memory task performance. Poster to be presented at the 80th Annual National Convention, Halifax, NS, May 31-June 2.   Dunbar, K., & MacLeod, C. M., (1984). A horse race of a different colour: Stroop interference patterns with transformed words. Journal of Experimental Psychology: Human Perception and Performance, 10(5), 622-639. doi: 10.1037/0096-1523.10.5.622 Eich, E. (1995). Searching for mood dependent memory. Psychological Science, 6(2), 67-75. doi: 10.1111/j.1467-9280.1995.tb00309.x 155  Einstein, G. O., & McDaniel, M. A., (1990). Normal aging and prospective memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 717-726. doi: 10.1037/0278-7393.16.4.717 Einstein, G. O., & McDaniel, M. A., (2000). Strategic and automatic processing in prospective memory retrieval: a multiprocess framework. Applied Cognitive Psychology, 14(7), 127-144. doi:  10.1002/acp.775 Einstein, G. O., & McDaniel, M. A. (2005). Prospective memory multiple retrieval processes. Current Directions in Psychological Science, 14(6), 286-290. doi:10.1111/j.0963-7214.2005.00382.x Einstein, G. O., McDaniel, M. A., Richardson, S. L., Guynn, M. J., & Cunfer, A. R. (1995). Aging and Prospective Memory: Examining the Influences of Self-Initiated Retrieval Processes. Journal of Experimental Psychology: Learning, Memory and Cognition, 21(4), 996-1007. doi: 10.1037/0278-7393.21.4.996 Einstein G. O., McDaniel M. A., Manzi M., Cochran B., Baker M. (2000). Prospective memory and aging: Forgetting intentions over short delays. Psychology and Aging, 15(4), 671-683. doi: 10.1037/0882-7974.15.4.671 Einstein, G. O., McDaniel, M. A., Thomas, R., Mayfield, S., Shank, H., Morrisette, N., & Breneiser, J. (2005). Multiple processes in prospective memory retrieval: factors determining monitoring versus spontaneous retrieval. Journal of Experimental Psychology: General, 134(3), 327-342. doi: 10.1037/0096-3445.134.3.327 Ellis, J. & Milne, A. (1996). Retrieval cue specificity and the realization of delayed intentions. Quarterly Journal of Experimental Psychology, 49(4), 862-887. doi: 10.1080/713755662 Ellis, J., Kvavilashvili, L., & Milne, A. (1999). Experimental tests of prospective remembering: The influence of cue-event frequency on performance. British Journal of Psychology, 90(1), 9-23. doi: 10.1348/000712699161233 156  Fisher, A. V. (2011). Automatic shifts of attention in the dimensional change card sort task: Subtle changes in task materials lead to flexible switching. Journal of Experimental Child Psychology, 108(1), 211-219. doi: 10.1016/j.jecp.2010.07.001 Friendly, M., Franklin, P. E., Hoffman, D., & Rubin, D. C. (1982). The Toronto Word Pool: Norms for imagery, concreteness, orthographic variables, and grammatical usage for 1080 words. Behavior Research Methods & Instrumentation, 14(4), 375-399. doi: 10.3758/BF03203275 Gade, M., & Koch, I. (2007). The influence of overlapping response sets on task inhibition. Memory & Cognition, 35(4), 603-609. doi: 10.3758/BF03193298 Godden, D. R., & Baddeley, A. D. (1975). Context-dependent memory in two natural environments: On land and underwater. British Journal of Psychology, 66(3). 325-331. doi: 10.1111/j.2044-8295.1975.tb01468.x Gollwitzer, P. M. (1990). Action phases and mind-sets. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2, pp. 53-92). New York: Guilford Press. Recovered from http://psycnet.apa.org/record/1990-98254-002 Gollwitzer, P. M. (1993). Goal achievement: The role of intentions. European Review of Social Psychology, 4(1), 141-185. doi: 10.1080/14792779343000059 Goshke, T., & Kuhl, J. (1993). Representation of intentions: Persisting activation in memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19(5), 1211-1226. doi: 10.1037/0278-7393.19.5.1211 Graf, P., & Uttl, B. (2001). Prospective memory: A new focus for research. Consciousness and Cognition, 10(4), 437-450. doi: 10.1006/ccog.2001.0504 Graf, P., Uttl, B., & Dixon, R. A. (2002). Prospective and retrospective memory in adulthood. In P. Graf & N. Ohta (Eds.), Lifespan Development of Human Memory (pp. 257-282). Cambridge, MA: MIT Press.  157  Graf, P. (2005). Prospective memory retrieval revisited. In: N. Ohta, C. M. MacLeod, & B. Uttl (Eds.), Dynamic cognitive processes (pp. 305–332). Tokyo: Springer. Recovered from https://link.springer.com/chapter/10.1007/4-431-27431-6_13  Harrison, T. L., & Einstein, G. O. (2010). Prospective memory: Are preparatory attentional processes necessary for a single focal cue? Memory & Cognition, 38(7), 860-867. doi: 10.3758/MC.38.7.860 Heathcote, A., Loft, S., & Remington, R. W. (2015). Slow down and remember to remember! A delay theory of prospective memory costs. Psychological Review, 122(2), 376-410. doi: 10.1037/a0038952.supp Hering, A., Wild-Wall, N., Gajewski, P. D., Falkenstein, M., Kliegel, M., & Zinke, K. (2016). The role of cue detection for prospective memory development across the lifespan. Neuropsychologia, 93, 289-300. doi: 10.1016/j.neuropsychologia.2016.11.008 Hostler, T. J., Wood, C., & Armitage, C. J. (2018). The influence of emotional cues on prospective memory: a systematic review with meta-analyses. Cognition and Emotion, 32(1), 1-19. doi: 10.1080/02699931.2017.1423280 Hulme, C., Maughan, S., & Brown, G. D. A. (1991). Memory for familiar and unfamiliar words: evidence for a long-term memory contribution to short-term memory span. Journal of Memory and Language, 30(6), 685-701. doi: 10.1016/0749-596X(91)90032-F Hunt, R. R. (1995). The subtlety of distinctiveness: What von Restorff really did. Psychonomic Bulletin & Review, 2(1), 105-112. doi: 10.3758/BF03214414 Hunt, R. R., & Lamb, C. A. (2001). What causes the isolation effect? Journal of Experimental Psychology: Learning, Memory, and Cognition, 27(6), 1359-1366. doi: 10.1037/0278-7393.27.6.1359 Jacoby, L. L., Woloshyn, V., & Kelley, C. (1989). Becoming famous without being recognized: Unconscious influences of memory produced by dividing attention. Journal of Experimental Psychology: General, 118(2), 115-125. doi: 10.1037/0096-3445.118.2.115 158  Kane, M. J., & Engle, R. W. (2003). Working-memory capacity and the control of attention: the contributions of goal neglect, response competition, and task-set to Stroop interference. Journal of Experimental Psychology: General, 132(1), 47-70. doi: 10.1037/0096-3445.132.1.47 Kelly, A. J., Hertzog, C., Hayes, M. G., & Smith, A. D. (2013). The effects of age and focality on delay-execute prospective memory. Aging, Neuropsychology, & Cognition, 20(1), 101-124. doi: 10.1080/13825585.2012.691152 Kiesel, A., Steinhauser, M., Wendt, M., Jost, K., Philipp, A. M., & Koch, I. (2010). Control and interference in task switching – a review. Psychological Bulletin, 136(5), 849-874. doi: 10.1037/a0019842 Kliegel, M., Jäger, T., & Phillips, L. H. (2008). Adult age differences in event-based prospective memory: A meta-analysis on the role of focal versus nonfocal cues. Psychology and Aging, 23(1), 203-208. doi: 10.1037/0882-7974.23.1.203 Kliegel, M., Martin, M., McDaniel, M. A., & Einstein, G. O. (2001). Varying the importance of a prospective memory task: Differential effects across time- and event-based prospective memory. Memory, 9, 1-11. doi: 10.1080/09658210042000003 Koch, I., Gade, M., Schuch, S., & Philipp, A. M. (2010). The role of inhibition in task switching: A review. Psychonomic Bulletin & Review, 17(1), 1-14. doi: 10.3758/PBR.17.1.1 Kray, J., & Lindenberger, U. (2000). Adult age differences in task switching. Psychology and Aging, 15(1), 126-147. doi: 10.1037/0882-7974.15.1.126 Lavie, N., Hirst, A., de Fockert, J. W., & Viding, E. (2004). Load theory of selective attention and cognitive control. Journal of Experimental Psychology: General, 133(3), 339-354. doi: 10.1037-0096-3445-133.3.339 159  Libet, B., Gleason, C. A., Wright, E. W., & Pearl, D. K. (1983). Time of conscious intention to act in relation to onset of cerebral activity (readiness-potential). The unconscious initiation of a freely voluntary act. Brain, 106(3), 623-42. Retrieved from: https://www.ncbi.nlm.nih.gov/pubmed/6640273  Loft, S., & Remington, R. W. (2013). Wait a second: Brief delays in responding reduce focality effects in event-based prospective memory. The Quarterly Journal of Experimental Psychology, 66(7), 1432-1447. doi: 10.1080/17470218.2012.750677 Loft, S. & Yeo, H. (2007). An investigation into the resource requirements of event-based prospective memory. Memory & Cognition, 35(2), 263-274. doi: 10.3758/BF03193447 MacLeod, C. M. (1991). Half a century of research on the Stroop Effect: An integrative review. Psychological Bulletin, 109(2), 163-203. doi: 10.1037/0033-2909.109.2.163  Mäntylä, T. (1996). Activating actions and interrupting intentions: Mechanisms of retrieval sensitization in prospective memory, In: M. A. Brandimonte, G. O. Einstein, & M. A. McDaniel (Eds.), Prospective Memory: Theory and Applications (pp. 93-113). New Jersey: Psychology Press. doi: 10.4324/9781315806488 Mäntylä, T., & Bäckman, L. (1992). Aging and memory for expected and unexpected objects in real-world settings. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18(6), 1298-1309. doi: 10.1037/0278-7393.18.6.1298 Mäntylä, T. (1994). Components of recollective experience in cued recall. Scandinavian Journal of Psychology, 35(3), 263-270. doi: 10.1111/j.1467-9450.1994.tb00950.x Marcel, A. J. (1983). Conscious and unconscious perception: Experiments on visual masking and word recognition. Cognitive Psychology, 15(2), 197-237. doi: 10.1016/0010-0285(83)90009-9 160  Marsh, R. L., Hancock, T. W., & Hicks, J. L. (2002). The demands of an ongoing activity influence the success of event-based prospective memory. Psychonomic Bulletin and Review, 9(3), 604-610. doi: 10.3758/BF03196319  Marsh, R. L., Hicks, J. L., & Cook, G. I. (2005). On the relationship between effort toward an ongoing task and cue detection in event-based prospective memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31(1), 68-75. doi: 10.1037/0278-7393.31.1.68 Maylor, E. A. (1990). Age and prospective memory, Quarterly Journal of Experimental Psychology, 42(3), 471-493. doi: 10.1080/14640749008401233 Maylor, E. A. (1996). Age-related impairment in an event-based prospective memory task. Psychology and Aging, 11(1), 420-428. doi: 10.1037/0882-7974.11.1.74 Mayr, U., & Keele, S. W. (2000). Changing internal constraints on action: The role of backward inhibition. Journal of Experimental Psychology: General, 129(1), 4-26. doi: 10.1037/0096-3445.129.1.4 Mayr, U., & Kliegl, R. (2003). Differential effects of cue changes and task changes on task-set selection costs. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29(3), 362-373. doi: 10.1037/0278-7393.29.3.362 McCloskey, M. E., & Glucksberg, S. (1978). Natural categories: Well defined or fuzzy sets? Memory & Cognition, 6(4), 462-472. doi: 10.3758/BF03197480 McDaniel, M. A., & Einstein, G. O. (1993). The importance of cue familiarity and cue distinctiveness in prospective memory. Memory, 1(1), 23-41. doi: 10.1080/09658219308258223 McDaniel, M. A., & Einstein, G. O. (2000). Strategic and automatic processes in prospective memory retrieval: A multiprocess framework. Applied Cognitive Psychology, 14, S127-S144.  161  McDaniel, M. A., Glisky, E. L., Rubin, S. R., Guynn, M. J., & Routhieaux, B. C. (1999). Prospective memory: A neuropsychological study. Neuropsychology, 13(1), 103-110. Retrieved from: https://www.ncbi.nlm.nih.gov/pubmed/10067781 McDaniel, M. A., Guynn, M. J., Einstein, G. O., & Breneiser, J. E. (2004). Cue-focused and reflexive-associative processes in prospective memory retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30(3), 605-614. doi: 10.1037/0278-7393.30.3.605 McDaniel, M. A., Howard, D. C., & Butler, K. M. (2008). Implementation intentions facilitate prospective memory under high attention demands. Memory & Cognition, 36(4), 716-724. doi: 10.3758/MC.36.4.716 McDaniel, M. A., Robinson-Riegler, B., & Einstein, G. O. (1998). Prospective remembering: Perceptually driven or conceptually driven processes? Memory & Cognition, 26(1), 121-134. doi: 10.3758/BF03211375 Meeks, J. T., & Marsh, R. L. (2010). Implementation intentions about nonfocal event-based prospective memory tasks. Psychological Research, 74, 82-89. doi: 10.1007/s00426-008-0223-x  Meier, B., & Graf, P. (2000). Transfer appropriate processing for prospective memory tests. Applied Cognitive Psychology, 14, S11-S27.  Meier, B., Morger, V., & Graf, P. (2003). Competition between automatic and controlled processes. Consciousness and Cognition, 12(), 309-319. doi: 10.1016/S1053-8100(02)00069-7 Meier, B. & Rey-Mermet, A. (2012). Beyond monitoring: After-effects of responding to prospective memory targets. Consciousness and Cognition, 21(4), 1644-1653. doi: 10.1016/j.concog.2012.09.003 Meier, B. & Rey-Mermet, A. (2017). After-effects without monitoring costs: The impact of prospective memory instructions on task switching performance. Acta Psychologica, 184, 85-99. doi: 10.1016/j.actpsy.2017.04.010 162  Meier, B., Woodward, T. S., Rey-Mermet, A., & Graf, P. (2009). The bivalency effect in task-switching: General and enduring. Canadian Journal of Experimental Psychology, 63(3), 201-210. doi: 10.1037/a0014311 Meier, B., Zimmerman, T. D., & Perrig, W. J. (2006). Retrieval experience in prospective memory: Strategic monitoring and spontaneous retrieval. Memory, 14(7), 872-889.  Meiser, T., & Schult, J. C. (2008). On the automatic nature of the task-appropriate processing effect in event-based prospective memory. European Journal of Cognitive Psychology, 20(2), 290-311. doi: 10.1080/09541440701319068 Metzak, P. D., Meier, B., Graf, P., & Woodward, T. S. (2013). More than a surprise: The bivalency effect in task switching. Journal of Cognitive Psychology, 25(7), 833-842. doi: 10.1080/20445911.2013.832196 Monsell, S. (2003). Task switching, Trends in Cognitive Sciences, 7(3), 134-140. doi: 10.1016/S1364-6613(03)00025-7  Monsell, S., Sumner, P., & Waters, H. (2003). Task-set reconfiguration with predictable and unpredictable task switches. Memory & Cognition, 31(3), 327-342. doi: 10.3758/BF03194391 Monsell, S., Taylor, T. J., & Murphy, K. (2001). Naming the colour of a word: Is it responses or tasks sets that compete? Memory & Cognition, 29(1), 137-151. doi: 10.3758/BF03195748 Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16(5), 519-533. doi: 10.1016/S0022-5371(77)80016-9 Morton, J., & Chambers, S. M. (1973). Selective attention to words and colours. Quarterly Journal of Experimental Psychology, 25(3), 387-397. doi: 10.1080/14640747308400360 163  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 (pp. 269-310). Cambridge, MA, US: The MIT Press. Neely, J. H. (2012). Semantic priming effects in visual word recognition: A selective review of current findings and theories. In Basic processes in reading (pp. 272-344). Routledge. Nosofsky, R. M., & Palmeri, T. J. (1997). An exemplar-based random walk model of speeded classification. Psychological Review, 104(2), 266-300. doi: 10.1037/0033-295X.104.2.266 Nosofsy, R. M. (1988b). Exemplar-based accounts of relations between classification, recognition, and typicality. Journal of Experimental Psychology, 14(4), 700-708. doi: 10.1037/0278-7393.14.4.700 Nowinski, J. L., & Dismukes, K. (2005). Effects of ongoing task context and target typicality on prospective memory performance: The importance of associative cueing. Memory, 13(6), 649-657. Pereira, A., Ellis, J., & Freeman, J. (2012). Is prospective memory enhanced by cue-action semantic relatedness and enactment at encoding? Consciousness and Cognition, 21(3), 1257-1266.  Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In: R. L. Solso (Ed.), Information Processing and Cognition: The Loyola Symposium (pp. 55–85). Hillsdale, NJ: Erlbaum.  Ratcliff, R., & Smith, P. L. (2004). A comparison of sequential sampling models for two-choice reaction time. Psychological Review, 111(2), 333-367. Rensink, R. A., O’Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceived changes in scenes. Psychological Science, 8(5), 368-373.  Reynolds, J. R., West, R., & Braver, T. (2009). Distinct neural circuits support transient and sustained processes in prospective memory and working memory. Cerebral Cortex, 19(5), 1208-1221. doi: 10.1093/cercor/bhn164 164  Rips, L. J., Shoben, E. J., & Smith, E. E. (1973). Semantic distance and the verification of semantic relations. Journal of Verbal Learning and Verbal Behavior, 12(1), 1-20. doi: 10.1016/S0022-5371(73)80056-8 Rogers, R. D., & Monsell, S. (1995). Costs of predictable switch between simple cognitive tasks. Journal of Experimental Psychology, 124(2), 207-231. Recovered from http://psycnet.apa.org/buy/1995-31890-001  Rothbart, M., Siriam, N., & Davis-Stitt, C. (1996).The retrieval of typical and atypical category members. Journal of Experimental Social Psychology, 32(4), 309-336. doi: 10.1006/jesp.1996.0015 Rubinstein, J. S., Meyer, D. E., & Evans, J. E. (2001). Executive control of cognitive processes in task switching. Journal of Experimental Psychology: Human Perception and Performance, 27(4), 763-797. doi: 10.103/0096-1523.27.4.763  Schacherer, J., & Hazeltine, E. (2017). How conceptual overlap and modality pairings affect task-switching and mixing costs. Psychological Research, 1-13. doi: 10.1007/s00426-017-0932-0 Schneider, W., Eschman, A., & Zuccolotto, A. (2002) E-Prime Reference Guide. Pittsburgh: Psychology Software Tools Inc.  Scullin, M. K., Bugg, J. M., & McDaniel, M. A. (2012). Whoops, I did it again: commission errors in prospective memory. Psychology and Aging, 27(1), 46-53.  Scullin, M. K., McDaniel, M. A., Shelton, J. T., & Lee, J. H. (2010). Focal/nonfocal cue effects in prospective memory: Monitoring difficulty or different retrieval processes? Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(3), 736-749. doi: 10.1037/a0018971 Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: Sustained inattentional blindness for dynamic events. Perception, 28(9), 1059-1074. doi: 10.1068/p281059 165  Smith, R. E. (2003). The cost of remembering in event-based prospective memory: Investigating the capacity demands of delayed intention performance. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29(3), 347-361. doi: 10.1037/0278-7393.29.3.347 · Smith, R. E., Hunt, R. R., McVay, J. S., & McConnell, M. D. (2007). The cost of event-based prospective memory: Salient target events. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(4), 734-746.  Smith, R. E., Hunt, R. R., & Murray, A. E. (2017). Prospective memory in context: Moving through a familiar space. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43(2), 189-204. doi: 10.1037/xlm0000303 Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology: General, 121(1), 15-23. doi: 10.1037/0096-3445.121.1.15 Taylor, R. S., Marsh, R. L., Hicks, J. L., & Hancock, T. W. (2004). The influence of partial-match cues on event-based prospective memory. Memory, 12(2), 203-213. doi: 10.1080/09658210244000559 Trawley, S. L., Law, A. S., Brown, L. A., Niven, E. H., & Logie, R. H. (2014). Prospective memory in a virtual environment: Beneficial effects of cue saliency. Journal of Cognitive Psychology, 26(1), 39-47. doi: 10.1080/20445911.2013.852199 Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Donaldson (Eds.) Organization of memory (pp. 381-403). Oxford, England: Academic Press.  Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80(5), 352-373. doi: 10.1037/h0020071 Uttl, B. (2006). Age-related changes in event-cued visual and auditory prospective memory proper. Aging, Neuropsychology, and Cognition, 13(2), 141-172. doi: 10.1080/138255890969618 Uttl, B. (2008). Transparent meta-analysis of prospective memory and aging. PLoS one, 3(2), e1568. 166  Uyeda, K. M., & Mandler, G. (1980). Prototypicality norms for 28 semantic categories. Behaviour Research Methods & Instrumentation, 12(6), 587-595. doi: 10.3758/BF03201848 von Restorff, H. (1933). About the effect of area formations in the trace field. Psychological Research, 18(1), 299.342. doi: 10.1007/BF02409636 Walser, M., Fischer, R., & Goschke, T. (2012). The failure of deactivating intentions: aftereffects of completed intentions in the repeated prospective memory cue paradigm. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38(4), 1030–1044. doi: 10.1037/a0027000. Walser, M., Goschke, T., & Fischer, R. (2014). The difficulty of letting go: moderators of the deactivation of completed intentions. Psychological Research, 78(4), 574-583. doi: 10.1007/s00426-016-0795-9 Walser, M., Goschke, T., Möschl, M., & Fischer, R. (2017). Intention deactivation: Effects of prospective memory task similarity on aftereffects of completed intentions. Psychological Research, 81(5), 961-981. doi: 10.1007/s00426-016-0795-9 Walter, S., & Meier, B. (2015). The impact of absolute importance and processing overlaps on prospective memory performance. Applied Cognitive Psychology, 30, 170-177.  Waszak, F., Hommel, B., & Allport, A. (2003). Task-switching and long-term priming: Role of episodic stimulus-task bindings in task-shift costs. Cognitive Psychology, 46(4), 361-413. doi: 10.1016/S0010-0285(02)00520-0 Waszak, F., Hommel, B., & Allport, A. (2005). Interaction of task readiness and automatic retrieval in task switching: Negative priming and competitor priming. Memory & Cognition, 33(4), 595-610. doi: 10.3758/BF03195327 West, R., Scolaro, A. J., & Bailey, K. (2011). When goals collide: The interaction between prospective memory and task-switching, Canadian Journal of Experimental Psychology, 65(1), 38-4. doi: 10.1037/a0022810 167  Woodward, T. S., Meier, B., Tipper, C. & Graf, P. (2003). Bivalency is costly: bivalent stimuli elicit cautious responding. Experimental Psychology, 50(4), 233-238. doi: 10.1026//1618-3169.50.4.233 Whittlesea, B. W. A., & Williams, L. D., (1998). Why do strangers feel familiar, but friends don’t? A discrepancy-attribution account of feelings of familiarity. Acta Psychologia, 98(2-3), 141-165. doi: 10.1016/S0001-6918(97)00040-1  Whittlesea, B. W. A., & Williams, L. D. (2001a). The discrepancy-attribution hypothesis: I. The heuristic basis of feelings of familiarity. Journal of Experimental Psychology Learning, Memory, and Cognition, 27(1), 3-13. doi: 10.1037//0278-7393.27.1.3 Whittlesea, B. W. A., & Williams, L. D. (2001b). The discrepancy-attribution hypothesis: II. Expectation, uncertainty, surprise, and feelings of familiarity. Journal of Experimental Psychology Learning, Memory, and Cognition, 27(1), 14-33. doi: 10.1037//0278-7393.27.1.14 Whittlesea, B. W. A. (2002). False memory and the discrepancy-attribution hypothesis: The prototype-familiarity illusion. Journal of Experimental Psychology: General, 131(1), 96-115. doi: 10.1037/0096-3445.131.1.96 Wylie, G., & Allport, D. A. (2000). Task-switching and the measurement of “switch costs”. Psychological Research, 63(3-4), 212-233. doi: 10.1007/s004269900003 Yonelinas, A. P. (2002). The nature of recollection and familiarity: A review of 30 years of research. Journal of Memory and Language, 46(3), 441-517. doi: 10.1006/jmla.2002.2864 Zeigarnik, B. (1927). On finished and unfinished tasks. Psychologische Forschung, 9, 1-15. Retrieved from: https://pdfs.semanticscholar.org/edd8/f1d0f79106c80b0b856b46d0d01168c76f50.pdf  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0381005/manifest

Comment

Related Items