Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Visual field effects in letter matching : an exploration of hemispheric, attentional, and strategic biases Fecteau, Jillian 2002

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

Item Metadata

Download

Media
831-ubc_2002-750140.pdf [ 20.79MB ]
Metadata
JSON: 831-1.0090743.json
JSON-LD: 831-1.0090743-ld.json
RDF/XML (Pretty): 831-1.0090743-rdf.xml
RDF/JSON: 831-1.0090743-rdf.json
Turtle: 831-1.0090743-turtle.txt
N-Triples: 831-1.0090743-rdf-ntriples.txt
Original Record: 831-1.0090743-source.json
Full Text
831-1.0090743-fulltext.txt
Citation
831-1.0090743.ris

Full Text

VISUAL FIELD EFFECTS IN LETTER MATCHING: AN EXPLORATION OF HEMISPHERIC, ATTENTIONAL, AND STRATEGIC BIASES by J I L L I A N F E C T E A U B.Sc. University of Toronto, 1997 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Graduate Program in Neuroscience) We accept this thesis as conforming to^he'rc^c^ntred standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A February 2002 Copyright by J i l l i a n Fecteau, 2002 In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes maybe granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Graduate Program in Neuroscience The University of British Columbia Vancouver, Canada Date: May 1,2002 Abstract It is easier to determine the identity of a pair of mixed-case letters when they appear on opposite sides of fixation (across-display) than when presented on the same side (within-display). This across-display advantage is taken as evidence by some that the hemispheres can each process letter identity independently (computational complexity theory, Banich, 1998). One unexplored feature of the across-display advantage is that it typically occurs in conjunction with a left-field advantage. If a within- display advantage is obtained instead, it typically occurs in conjunction with a right-field advantage. One reason for studying this relationship further is because both field (left vs. right) and display (across vs. within) advantages may index hemispheric differences in attentional orienting. According to the asymmetrical orienting hypothesis, the left hemisphere orients to right space, generating both right-field and within-display advantages for some tasks. The right hemisphere orients most efficiently to left space, generating a weak left-field advantage, but it can also orient to right space, contributing to an across-display advantage in other tasks. These interpretations were tested in a total of 8 experiments, involving 242 participants. Experiments 1 & 2 confirmed the correlation between left-field and across-display advantages in mixed-case letter rmtching. Experiment 3 revealed similar left-field and across-display advantages in same-case letter matching when a task irrelevant dimension (color) was added. Experiments 4-6 tested matching tasks and response modes favoring the left hemisphere (localization, color reporting, rmtching on rhyme) in an effort to dissociate the field and display effects. This was unsuccessful. Finally, in Experiments 7 and 8, participants were instructed to mentally compare letters in a strict order. This lead to a mimicking of the standard data pattern, or its elimination, depending on which instructions were followed. These findings do not support either theory because strategic effects should not affect parallel processing or hemispheric differences in attentional orienting. The implications are discussed with regard to hemispheric interactions, attentional orienting, and the more general problem of estabfohing priority in visual processing. A qualitative model is presented to account for the data, involving both top-down strategic biases (left-to-right comparison) and bottom-up processing biases (opposite-direction orienting). i i Table of Contents Abstract ii Table of Contents iii Figure Captions vi Table Captions ix Glossary xi Introduction 1 Computational complexity theory 1 Successes and failures of the computational complexity theory 5 Asymmetrical orienting hypothesis 5 Different theories, different predictions 10 Experiment 1: Replicating the basic findings 10 Methods 10 Results 14 Discussion 15 Experiment 2: Does eliminating the possibility of different cognitive sets alter the data pattern? 16 Methods 16 Results 17 Discussion 18 Experiment 3: Is complexity or attention responsible for the differences between same-and mixed-case letter matching? 18 Experiment 3A: Same versus nixed odor displays in same-case letter rmtdoing 20 Methods 20 Results 22 Discussion 23 Experiment 3B: Color vdationship in same-case letter mttdnng 24 Methods 24 Results 24 Discussion 25 Experiment 3C Same versus mixed odor displays in mixed-case letter rmtdoing 25 Methods 26 Results 26 i i i Discussion 27 Experiment 3D: Color relationship in mixed-case letter matohing 27 Methods 28 Results 28 Discussion 29 Oimiewof Experiments 1- 3 29 Experiment 4: Does manipulating field advantage alter the display effects? 31 Methods 31 Results 32 Discussion 35 Experiment 5: Does another task and/or different set-sizes modify field and display effects? 36 Methods 37 Results 39 Discussion 47 Experiment 6: Does changing the material modify field and display effects? 47 Methods 48 Results 50 Discussion 50 Taking toll 51 Experiment 7: Does a right-to-left instructional bias change left-field advantage? 53 Methods 54 Results 54 Discussion 55 Experiment 8: Does a within-to-across bias change the across-display advantage? 57 Methods 58 Results 58 Discussion 59 General Discussion 60 Computational complexity theory and interhemispheric interactions 61 Asymmetrical orienting hypothesis and hemispheric differences in orienting attention 66 Filling in the Void 67 The left-to-right bias in letter comparison. 67 The across-to-within bias in letter comparison. 67 Bringing it all together: A proposed task analysis of mixed-case letter rnatching 69 Conclusions 77 iv References 79 Appendix I: Table summarizing field and display advantages across studies exploring interhemispheric interactions 89 Appendix II: Summary of reaction time and error data 93 Thesis Experiment 1 93 Thesis Experiment 2 94 Thesis Experiment 3 95 Thesis Experiment 4 97 Thesis Experiment 5 98 Thesis Experiment 6 101 Thesis Experiment 7 102 Thesis Experiment 8 103 Appendix III: Ethical Approval 104 V Figure Captions Figure 1. Illustration of the 3-item letter-matching task. On the left is an example of the within-display, same-case letter matching. On the right is an example of the across-display, mixed-case letter matching. Further clarification of these terms can be found in the text. 2 Figure 2. Typical data pattern produced by letter-matching task. Redrawn from Weissman & Banich (2000). 3 Figure 3. Typical findings in mixed-case letter matching illusfrating that the left-field advantage does not interact with the across-display advantage. Redrawn from Weissman & Banich (2000). 4 Figure 4. Hypothetical data of left hemisphere priming in the letter matching task. 8 Figure 5. Examples of the different displays and complexity conditions. In the left panel, the numbers and associated arrows indicate the distance of each item (in degrees of visual angle) from central fixation. See text for further details. 11 Figure 6. Efficiency scores for same-case letter matching in Experiment 1 broken down by display type and probe position. Error bars represent ±1 units of standard error. 14 Figure 7. Efficiency scores for mixed-case letter matching in Experiment 1 broken down by display type and probe position. 15 Figure 8 Efficiency scores for same-case letter matching in Experiment 2 broken down by display type and probe position. 17 Figure 9. Efficiency scores for mixed-case letter matching in Experiment 2 broken down by display type and probe position. 18 Figure 10. Examples of the mixed color displays. The upper two panels illustrate the same color relationship between the target and the matching probe. The lower two panels illustrate the different color relationship between the target and the matching probe. 21 Figure 11. Efficiency scores for the same color displays in same-case letter matching in Experiment 3 broken down by display type and probe position 22 Figure 12. Efficiency scores for the mixed color displays in same-case letter matching in Experiment 3 broken down by display type and probe position 23 Figure 13. Efficiency scores for the mixed color displays in same-case letter matching in Experiment 3 broken down by display type and color relationship between the target and probe. Note that "same" and "different" refer to the color match and mismatch between the target and probe. 25 Figure 14. Efficiency scores for the same color displays in mixed-case letter matching condition in Experiment 3 broken down by display type and probe position 26 Figure 15. Efficiency scores for the mixed color displays in mixed-case letter matching condition in Experiment 3 broken down by display type and probe position. 27 vi Figure 16. Efficiency scores for the mixed color displays in mixed-case letter matching in Experiment 3 broken down by display type, probe position, and color relationship between the target and probe. 28 Figure 17. Efficiency scores for same-case letter matching in Experiment 4 when a detection response was used. Broken down by display type and probe position 33 Figure 18. Efficiency scores for same-case letter matching in Experiment 4 when a localization response was used. Broken down by display type and probe position 33 Figure 19. Efficiency scores for mixed-case letter matching in Experiment 4 when a detection response was used. Broken down by display type and probe position 34 Figure 20. Efficiency scores for mixed-case letter matching in Experiment 4 when a localization response was used. Broken down by display type and probe position 35 Figure 21. Illustration of the 9-item displays. Please note: the background of the displays has been lightened for purposes of reproduction and does reflect the actual color used. 38 Figure 22. Efficiency scores for 3-item same-case letter matching in Experiment 5 when a detection response was used. Broken down by display type and probe position 40 Figure 23. Efficiency scores for 3-item same-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 40 Figure 24. Efficiency scores for 3-item same-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position 41 Figure 25. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when a detection response was used. Broken down by display type and probe position 42 Figure 26. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 42 Figure 27. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position 43 Figure 28. Efficiency scores for 9-item same-case letter matching in Experiment 5 when a detection response was used. Broken down by display type and probe position 43 Figure 29. Efficiency scores for 9-item same-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 44 Figure 30. Efficiency scores for 9-item same-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position 44 Figure 31. Efficiency scores for 9-item mixed-case letter matching in Experiment 5 when a detection response was used. Broken down by display type and probe position 45 Figure 32. Efficiency scores for 9-item mixed-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 46 vii Figure 33. Efficiency scores for 9-item mixed-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position 46 Figure 34. Efficiency scores for the rhyme decision in Experiment 6 broken down by display type and probe position 50 Figure 35. Efficiency scores for the "left first" and the "right first" instructions in Experiment 7 broken down by display and probe position. Note that left and right refers to the field of the matching probe. 55 Figure 36. Efficiency scores for the "across first" and the "within first" instructions in Experiment 8 broken down by display and probe position. Note that left and right refers to the position of the matching probe. 59 Figure 37. Examples of 2-item displays. 64 Figure 38. Selecting the target. Red arrow (origmating from above in middle panel) denotes facilitatory top-down influences involved in selecting the target. The relative size of cylinders denotes the amount of neural activity for each position. The bottom cylinder in the right panel is colored in green (shaded in reproduced copies) to represent that attention is directed to its location. 71 Figure 39. Illustration of top-down left-field bias and bottom-up across-display bias interacting to facilitate performance. Blue arrow (originating from below in middle panel) denotes greater inhibition. Green arrow (in right panel) represents shift of attention. 72 Figure 40. Example of the separate left-field and across-display biases canceling each other out. Attention can be shifted equally well in either direction. 73 Figure 41. Illustration of left-field and across-display biases interacting to impair performance. Large green arrow (thick arrow in right panel) illustrates that attention incorrectly shifted to left probe. Small green arrow (dashed arrow in right panel) illustrates that a second shift of attention shift is necessary to achieve the target in the right position. 73 Figure 42. The different displays used in the Weissman & Banich (2000) study. 75 viii Table Captions Table A2-1. Experiment 1 reaction time and error scores broken down by display and location of matching probe 93 Table A2-2. Source table of statistical analyses for same-case letter matching in experiment 1, all interactions shown. 93 Table A2-3. Source table of statistical analyses for mixed-case letter matching in experiment 1, all interactions shown. 93 Table A2-4. Experiment 2 reaction time and error scores broken down by display and location of matching probe. Note the reaction time data for the target absent condition is an estimate as data from 4 subjects were lost. 94 Table A2-5. Source table of statistical analyses for same-case letter matching in experiment 2, all interactions shown. 94 Table A2-6. Source table of statistical analyses for mixed-case letter matching in experiment 2, all interactions shown. 94 Table A2-7. Experiment 3A and 3C reaction time and error scores broken down by display and location of matching probe for color displays 95 Table A2-8. Source table of statistical analyses involving Color variable in Experiment 3A, all interactions shown. 95 Table A2-9. Source table of statistical analyses involving Color Relationship in Experiment 3B, all interactions shown. 95 Table A2-10. Source table of statistical analyses involving Color variable in Experiment 3C, all interactions shown. 96 Table A2-11. Source table of statistical analyses involving Color variable in Experiment 3D, all interactions shown. 96 Table A2-12. Experiment 4 reaction time and error scores broken down by display and location of matching probe. Detection scores above hashed line, localization scores below hashed line. 97 Table A2-13. Source table of statistical analyses for same-case letter matching in experiment 4. Detection scores above hashed line, localization scores below hashed line 97 Table A2-14. Source table of statistical analyses for mixed-case letter matching in experiment 4. Detection scores above hashed line, localization scores below hashed line 97 Table A2-15. Experiment 5 reaction time.and error scores broken down by display and location of matching probe for 3-item displays. Detection scores above hashed line, localization scores in between hashed lines, and colOr report score below hashed line. 98 ix Table A2-16. Source table of statistical analyses for 3-item displays, same-case letter matching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 98 Table A2-17. Source table of statistical analyses for 3-item displays, mixed-case letter matching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 99 Table A2-18. Experiment 5 reaction time and error scores broken down by display and location of matching probe for 9-item displays. Detection scores above hashed line, localization scores in between hashed lines, and color report score below hashed line. 99 Table A2-19. Source table of statistical analyses for 9-item displays, same-case letter matching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 100 Table A2-20. Source table of statistical analyses for 9-item displays, mixed-case letter matching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 100 Table A2-21. Experiment 6 reaction time and error scores broken down by display and location of matching probe. Note the target absent reaction time and error scores are an estimate of the actual value as data from 3 subjects were lost. 101 Table A2-22. Source table of statistical analyses in experiment 6. 101 Table A2-23. Experiment 7 reaction time and error scores broken down by display and location of matching probe for each instruction set. 102 Table A2-24. Source table of statistical analyses of left-to-right instructions in experiment 7. 102 Table A2-25. Source table of statistical analyses of right-to-left instructions in experiment 7. 102 Table A2-26. Experiment 8 reaction time and error scores broken down by display and location of matching probe for each instruction set. Note the target absent reaction time and error scores are an estimate of the actual value as data from 2 subjects were lost. 103 Table A2-27. Source table of statistical analyses of across-to-within instructions in experiment 8. 103 Table A2-28. Source table of statistical analyses of within-to-across instructions in experiment 8 103 x Glossary Across-display advantage. Superior performance when the target and rmtching probe are presented to opposite sides of fixation. Should be evidenced during computationally complex tasks, such as mixed-case letter matching. See also, Computational complexity, mixed-case letter matching Asymmetrical orienting hypothesis proposes that field and display effects evidenced in the interhemispheric interactions literature index the attentional biases of the primed hemisphere. Priming the left hemisphere causes attention to be biased towards the right field and only that side, explaining the correlation between the right-field advantage and the within-display advantage. Alternatively, priming the right hemisphere causes attention to be more evenly distributed across both fields, explaining the correlation between the weaker left-field advantage and the across-display advantage. Attention must be demanded for such biases to be observed. Adapted from Kinsbourne (1970, 1973). SeeakoHenispbencPriming(Ftmtiond attentional orienting Biased attentional orienting A byproduct of hemispheric priming. In the absence of hemispheric prirning, attention is easily shifted in either direction. However, as one hemisphere becomes primed, attention is shifted more easily in the direction managed by that hemisphere. This facilitation, or bias, to shift attention in one direction enhances the detection and discrimination of objects that are presented in the same direction and a visual field advantage ensues. Computational Complexity Theory proposes that the interactions between the cerebral hemispheres play a fundamental role in the coordination of selective attention. Each hemisphere is believed to possess its own pool of attentional resources and to be able to process the information that it receives directly. The duplication of attentional resources in the two hemispheres, along with the ability of the two hemispheres to work in parallel, doubles the total amount of information processed by the brain at one time. However, the process of integrating the activities of the two hemispheres requires extra cognitive effort. The complexity of the task is proposed to determine whether the benefit of parallel processing by the two hemispheres offsets the cost of integrating the resulting information. See also Computational complexity, Interhemispheric independence, Interhemispheric sharing. Computational complexity. The number of transformations, operations, or computations that must be performed on the input before a decision can be reached (Banich, 1998, p 131). In the letter-matching task, complexity is increased by changing the case of the target letter. See also computational complexity theory, same-case letter matching, and mixed-case letter matching. Hemispheric priiriing (Functional distance hypothesis). Kinsbourne (1970,1973; Kinsbourne & Hicks, 1978) proposed when one hemisphere dorninates performance, it primes other cortical regions with which it is closely interconnected. For non-homologous regions, "close" consists of cortical regions residing within the same hemisphere. On a behavioral level, cortical printing is evidenced when all cognitive and motor abilities associated with one hemisphere are facilitated. xi Interherriispheric independence is proposed for simple tasks, because the cost of integrating the activities of the cerebral hemispheres is greater than the benefit of parallel processing. Performance is best when each hemisphere can solve the task independently. Evidenced by a within-display advantage. See also Computational complexity theory and Within-display advantage Interhemispheric sharing is proposed for complex tasks, because the benefit of parallel processing offsets the cost of integrating information. Performance is best when the hemispheres collaborate and share the workload. Evidenced by an across-display advantage. See also Computational complexity theory and Across-display advantage Mixed-case letter rratching. Trie target letter appears in lower case and the probe letters appear in upper case. Because the identity of the letter must be extracted before a decision can be made, the task is computationally complex. According to the Computational Complexity theory, this task should produce an across-display advantage. See also Target, Probes, Computational complexity, Computational complexity theory, Interhemispheric sharing, Across-display advantage. Probes. In the letter-rmtching task, the probes are upper case letters that are presented above fixation. The participant must decide if one of the probe letters matches the target letter. Within this thesis, "probe" is often used to refer to the matching probe specifically. Same-case letter matching. A l l of the letters in the display (i.e., target and probes) appear in upper case. Because decision-making can be based upon the letters' perceptual characteristics, the task is computationally simple. According to the Computational Complexity theory, this task should produce a within-display advantage. See also Target, Probes, Computational complexity, Computational complexity theory, Interhemispheric independence, Within-display advantage. Target. In the letter- rmtching task, the target is the reference letter to which the probes letters are compared. The target always appears below fixation and can appear in upper or lower case, which manipulates the computational complexity of letter matching. See also Computational complexity, same-case letter matching, and mixed-case letter matching. Within-display advantage. Superior performance when the target and matching probe are presented to the same side of fixation. Should be evidenced during computationally simple tasks, such as same-case letter matching. See also, Computational complexity, same-case letter matching. xii Introduction At every moment we are bombarded by a vast amount of information. For instance, as I sit on my balcony and write this paragraph, I am inundated with competing sensory stimuli. I can hear the thunderous sound of an airplane flying overhead and the melodic songs of birds in a nearby tree. I can feel the heat of the sun on my skin and the pressure of my cat sitting on my lap. I can see the dim screen of my laptop and the swaying branches of a tree in the distance. Despite this wealth of sensory input, I am still able to write this paragraph. Doing so means that I can effectively ignore much of the sensory information that is irrelevant to my task and focus upon that which is relevant. Being able to select some information, while ignoring the rest, is what it means to attend. Selective attention is beneficial in many ways. It allows us to focus on pertinent information and ignore extraneous input. At the same time, there is a price to be paid for this ability because it places limits on the amount of information that can be processed at one time (e.g., Lavie, 1995; Triesman &Kahneman, 1984). Computational complexity theory A recent theory has proposed that interactions between the two cerebral hemispheres play a fundamental role in the coordination of selective attention (Banich, 1998). Each hemisphere is believed to possess its own pool of attentional resources (Friedman & Poison, 1981) and to be able to process the information that it receives directly, in most instances (Banich, 1998). It has even been suggested that the duplication of attentional resources in the two hemispheres, along with the ability of the two hemispheres to work in parallel, effectively doubles the total amount of information that can be processed by the brain at one time (Banich, 1998). However, this doubling of processing capacity also has some associated costs. Namely, the process of integrating the activities of the two hemispheres requires extra cognitive effort (Banich, 1998). Indeed, it is predicted that sometimes the benefit of parallel processing by the two hemispheres is not worth the cost of integrating the resulting information. In this case, each hemisphere is better off working alone, a process called interhemispheric independence. Other times, the benefit of parallel processing offsets the cost of integrating information. In this case, the hemispheres collaborate and share the workload, a process called interhemispheric sharing. It is proposed further that the complexity of the cognitive task detennines whether the hemispheres will be more efficient working alone or together. Computational complexity is defined as the "number and sorts of transformations, operations, or computations that must be performed on the input before a decision can be reached" (Banich, 1998, p 131). Computationally simple tasks elicit interhemispheric independence because the cost of integrating information is greater than the benefit of parallel processing. Computationally complex tasks elicit mterhemispheric sharing because the benefit of parallel processing offsets the cost of integrating information. The primary behavioral support for computational complexity theory has come from the letter-matching task illustrated in Figure 1. Three letters are presented at once to a participant, one letter below fixation (the target) and two letters above fixation (the probes). The task is to decide if one of the two probes matches the target, with the likelihood of a match being 50%. The critical 1 manipulation is the location of the matching probe with respect to the target. The target and its matching probe can appear to the same side of fixation, the within-display (left panel of Figure 1), or can appear to opposite sides, the across-display (right panel of Figure 1). B R + B B R + b Within-Display Same-Case Across-Display Mixed-Case Figure 1. Illustration of the 3-item letter-matching task. On the left is an example of the within-display, same-case letter matching. On the right is an example of the across-display, mixed-case letter niatching. Further clarification of these terms can be found in the text. Computational complexity has been manipulated in this task by changing the case of the target letter. In same-case letter matching, all of the letters appear in upper case (left panel of Figure 1). In this case, decision-making is simple because it can be based upon the letters' perceptual characteristics alone. Only the visual shapes of the letters must be processed. In mixed-case letter matching, the target appears as a lower case letter and the probes appear as upper case letters (right panel of Figure 1). In this case, the decision must be based upon the letter's identity. This is a more abstract relationship, one that depends on a learned mapping between two different shapes and an abstract label1. These letter-matching tasks typically elicit a within-display advantage in same-case letter rmtching and an across-display advantage in mixed-case letter matching, as depicted in Figure 2. This data pattern has a straightforward interpretation within the framework of computational complexity theory. Same-case letter matching is performed most efficiently in within-displays because the cost of integrating the activities of the two hemispheres offsets the benefit of parallel processing (interhemispheric independence). Mixed-case letter matching is performed most efficiently in across-displays because the benefit of parallel processing offsets the cost of integration (interhemispheric sharing) (e.g., Banich & Belger, 1990; Belger & Banich, 1992; Weissman & Banich, 2000). 1 From a historical perspective, the letter-matching task used to assess "computational complexity" was first introduced byPosner and Mitchell (1967) as a way of assessing depth of processing and was later used byPosner et al., (1968) as a way of assessing the manner in which internal representations of same-case and mixed-case letter matching change following a delay. 2 600 .9 500 V, 400 300 • Within-display • Across-display Same-Case Mixed-case Figure 2. Typical data pattern produced by letter-rmtching task. Redrawn from Weissman & Banich (2000). Only when the cognitive task demands resources that reside uniquely in one or the other hemisphere is it predicted that interhemispheric independence will be observed for complex decisions (e.g., Banich, 1998). In many circumstances this restriction is not an issue because both hemispheres can perform the same tasks (e.g., Banich & Nicholas 1998). However, some behaviors do require the participation of a particular hemisphere. For instance, the task of deciding whether letter strings have the same phonological ending (rhyme) does not elicit an across-display advantage, presumably because only the left hemisphere can decide if two words (or letters) share the same sound (Banich & Karol , 1992; Belger & Banich, 1998). These data therefore support that claim that parallel processing is advantageous only when each hemisphere can make the decision independently of the other (e.g., Banich, 1998; Belger & Banich, 1998). Note that relative differences in proficiency of one or the other hemisphere is not of serious concern. Parallel processing may still be advantageous, and will therefore still take place, even if the two hemispheres perform the task with differing degrees of competency (e.g., Banich, 1998). Support for this prediction can be seen in the mixed-case letter-matching task, where even though the right hemisphere is at a general advantage over the left hemisphere, this effect of visual field does not interact statistically with the main effect of the across-display advantage (see Figure 3) (Weissman & Banich, 2000). 3 600 cn —— 550 e • i—i 500 s • 1-H +-> 450 c o +-» 400 CD 350 300 • Within-display • Across-display Left Right Probe Position Figure 3. Typical findings in mixed-case letter matching illustrating that the left-field advantage does not interact with the across-display advantage. Redrawn from Weissman & Banich (2000). The corpus callosum is the neuroanatomical structure of critical importance for interhemispheric interactions. In light of this, the computational complexity theory is supported by reports that clinical disorders involving damage to the corpus callosum (e.g., multiple sclerosis, closed head injuries) are also associated with deficits in attention. For instance, multiple sclerosis is a demyelinating disorder that compromises the functioning of the corpus callosum. Research has shown that as the corpus callosum becomes less myelinated over the course of the illness, patients with multiple sclerosis evidence increasingly serious attentional deficits (Rao, 1995). In summary, according to computational complexity theory, the cerebral hemispheres play a critical role in the coordination of selective attention (Banich, 1998). Each hemisphere has its own pool of resources from which to draw, which doubles the amount of processing that can take place at once. This doubling in processing capacity increases the attentional capacity of the brain. However, the task must be sufficiently complex for parallel processing to be beneficial. If the task is too simple, then the cost of integrating the activities of the two hemispheres offsets the benefit of parallel processing and interhemispheric independence (within-display advantage) will be obtained. If the task is more complex, then the cost of integrating the activities of the hemispheres is offset by the benefit of parallel processing and interhemispheric sharing (across-display advantage) will be obtained. Two prerequisites are necessary for interhemispheric sharing to take place: Both hemispheres must be able to contribute to the cognitive task and the corpus callosum must not be compromised. 4 Successes and failures of the computational complexity theory The computational complexity theory has been developed and supported for the most part by studies using the 3-item matching tasks described earlier. Computationally complex tasks, such as mixed-case letter rmtching, reliably elicit an across-display advantage (Banich & Belger, 1990; Banich et al, 2000; Belger & Banich, 1992,1998; Koivisto, 2000; Reuter-Lorenz et al, 1999; Weissman & Banich, 2000; Weissman, Banich, & Puente, 2000, but see Banich et al., 2000) and computationally simple tasks, such as same-case letter notching, elicit a within-display advantage more than half of the time (Banich & Belger, 1990; Belger & Banich, 1998; Reuter-Lorenz, Stanzak, & Miller 1999; Weissman & Banich, 2000; Weissman, Banich, & Puente, 2000). Therefore, the predictions of the theory with regard to letter rmtching complexity are borne out in the data, even if the across-display advantage is the more robust of the two effects. However, obtaining an across-display advantage for computationally complex tasks becomes less reliable if small modifications are made to the paradigm. For instance, if the matching decision is made between two (as opposed to three) items, an across-display advantage is achieved less often (e.g., Berger & Landholt, 1990; Diamond, Gibson & Gazzaniga, 1972; Leiber, 1982; Liderman, 1986; Liderman, Merola, & Martinez, 1985). In summary, the computational complexity theory cannot always accommodate the reported evidence. Sometimes computationally simple tasks do not elicit a within-display advantage (e.g., Banich et al., 2000; Belger & Banich, 1992; Brown & Jeeves, 1993; Sereno &Kossylyn, 1991; Zhang & Feng, 1999). At other times computationally complex tasks do not elicit an across-display advantage (e.g., Banich et al., 2000; Diamond, Gibson & Gazzaniga, 1972; Leiber, 1982; Liderman, 1986; Liderman, Merola, & Martinez, 1985). Asymmetrical orienting hypothesis An intriguing and unexplored relationship between display type and visual field can be seen in the many studies that have investigated interhemispheric interactions. Visual field advantage could be ascertained in thirty-five studies (containing a total of 63 unique experiments) that examined interhemispheric interactions in right-handed adult participants. When an across-display advantage was obtained using complex cognitive tasks by Banich's definition, 78% (32 / 41) of the experiments reported an advantage for the left visual field. When a v t^hin-display advantage was obtained instead, 72% (16 / 22) of the experiments reported a right visual field advantage. Considering the notable methodological variations2 among the studies under consideration, the tendency of an across-display advantage to appear in conjunction with a left visual field advantage and a within-display advantage to appear in conjunction and right visual field advantage may be an important correlation that points to an underlying common factor. The initial goal of this thesis was to test if hemispheric differences in orienting attention are responsible for this relationship. The rationale for this hypothesis will be developed in several steps. First, visual orienting and hemispheric differences in orienting will be described. Second, an interpretation of field advantages that is different from computational complexity theory will be provided that may explain the relationship between display and field effects that has been frequently reported in the literature. 2 Across studies, different stimuli and different responses were used to explore interhemispheric interactions. In theory, such methodological differences may be responsible for the less than perfect relationship between display and field that has been noted across studies. 5 Visual attention involves the specialized processing of an object or of a particular region of space. Visual orienting is the process of implementing these specialized processes for the chosen object or location. The hemispheres are not equal in their capacity to orient attention. For many cognitive activities, each hemisphere controls actions on the opposite side of the body and each is able to perceptually represent the opposite side of space. This general relationship is not true for orienting of visual attention. The left hemisphere is able to orient to the right side of space, while the right hemisphere is able to orient both to the left and to the right side of space (Heilman & Van Den Abell, 1980; Mesulam, 19.81; 1999). Functional imaging studies have provided much of the support for hemispheric differences in orienting attention. PET, fMRI, and E E G studies have consistently revealed that orienting to the left visual field increases activity in the right posterior parietal cortex and that orienting to the right visual field increases activity in both the left and the right posterior parietal cortices (e.g., Ffeilman & Van Den Abell, 1980; Mesulam, 1999; Nobre et al., 2000). Although well documented in functional imaging studies, this asymmetry has not been documented with behavioral measures in normal participants. Before the right hemisphere's ability to orient attention to both fields was recognized, Kinsbourne (1970,1973) proposed that visual field advantages arise from the attentional bias of the primed hemisphere. Taking a step back for a moment, he proposed that when one hemisphere dominates performance, it primes other cortical regions with which it is closely interconnected. For non-homologous regions, "close" consists of cortical regions residing within the same hemisphere. On a behavioral level, cortical priming is evidenced when all cognitive and motor abilities associated with one hemisphere are facilitated. Biased attentional orienting also happens to be one byproduct of hemispheric priming. In the absence of hemispheric priming, attention is easily shifted in either direction. However, as one hemisphere becomes primed, attention is shifted more easily in the direction managed by that hemisphere. This facilitation, or bias, to shift attention in one direction enhances the detection and discrimination of objects that are presented in the same direction; hence, a field advantage ensues. I will provide evidence supporting Kinsbourne's proposal in just a moment. Before I do, I will provide an example of how Kinsbourne's theory changes the way visual field effects are interpreted. It is easy to see how Kinsbourne's interpretation of visual field effects differs from the more traditional direct access interpretation if we consider the following example. Lexical decision tasks (deciding if a letter string is a word or not) typically elicit a right-field advantage (e.g., Collins, 1999). According to the direct access interpretation, the right-field advantage occurs because words presented to the right-field are directly received and processed by the left hemisphere-the hemisphere dominant for language in most individuals. The disadvantage for the left-field occurs because these words are first received by the right hemisphere, which then must be shuttled off to the left hemisphere before being processed. Whenever information must be transferred from one hemisphere to the other, this indirect information is received later and its quality is degraded compared to information that is received directly. Simply put, field advantages exist because of direct (better) or indirect (worse) access to the hemisphere that is superior at processing a certain type of material (Kimura, 1973). Instead, Kinsbourne proposed that the right-field advantage occurs in lexical decision tasks because the left hemisphere is primed and it's increased activity causes attention to be shifted more efficiently to the right. Simply put, field advantages exist because hemispheric priming causes attention to be biased to one side. 6 Two lines of evidence support Kinsbourne's interpretation of visual field differences. First, participants engaged in a verbal task (dominated by the left hemisphere) are more likely to shift their gaze to the right and participants engaged in a visuospatial task (dominated by the right hemisphere) are more likely to shift their gaze to the left (Kinsbourne, 1972,1974). Or put another way, the spontaneous eye movements of participants engaged in tasks benefiting one hemisphere or another are consistent with the theory of hemispheric prirning. Second, dual task experiments have been used to support this claim. In such experiments, a task is chosen that does not elicit a field advantage (a hemispheric neutral task). This neutral task is performed by itself and alongside a second task that strongly benefits one hemisphere over the other. If prirning does affect attentional orienting, then the neutral task should adopt a significant field advantage when performed alongside the hemispheric advantaged task and should remain neutral when performed alone. For instance, if a neutral task were performed with a task advantaging the left hemisphere (e.g., verbal memory), then the neutral task would adopt a right-field (left hemisphere) advantage in the dual task condition. Such experiments have provided mixed support for Kinsbourne's theory, depending on which hemisphere was primed. When the second task advantages the left hemisphere, Kinsbourne's hypothesis has been supported (e.g., Kinsbourne & Hicks, 1978). But, when the second task advantages the right hemisphere, Kinsbourne's hypothesis has achieved little or no support (e.g., Friedman & Poison, 1981; Hellige, Cox, &Litvac, 1979). An astute reader may realize that a robust left-field advantage would not be expected when the right hemisphere is primed (owing to the right hemisphere's ability to orient to both fields). However, recall that Kinsbourne proposed this theory before the right hemisphere's superior ability to orient attention had been proposed (Heilman & Van Den Abell, 1980; Mesulam, 1981). Consider the ramifications had hemispheric differences in orienting attention been taken into account. A left hemisphere advantage would cause attention to be biased to the right, conferring an advantage for the right-field. A right hemisphere advantage would cause attention to be more evenly distributed across fields, conferring a weak or absent left-field advantage3. Simply put, the failure of dual task experiments to support Kinsbourne's hypothesis when the right hemisphere was primed would be the predicted consequence of an updated version of the same theory. The asymmetrical orienting hypothesis was designed to update Kinsbourne's hypothesis so that it may account for hemispheric differences in orienting attention. In this case, prirning the left hemisphere results in a robust right-field attentional bias because the left hemisphere orients to the right side of space. Prirning the right hemisphere results in a weaker or absent left-field advantage because the right hemisphere orients both to the left and right side of space. This hypothesis may have important consequences for the hemispheric literature more generally. First of all, it can potentially explain the mixed nature of replication successes and failures across studies, in which a right-field advantage is generally much easier to replicate than a left-field advantage (e.g., Sergent, 1983). In addition, if one considers that display advantages, like field advantages, provide attentional signatures of hemispheric advantage, then the asymmetrical orienting hypothesis may also explain 3 Please note, the review of the previous studies provided in the appendix does not compare the magnitude of left-field and right-field advantages. 7 the relationship between these effects that has been observed in the interhemispheric interactions literature. As mentioned, a right-field advantage is often found at the same time as a within-display advantage and a left-field advantage often occurs in conjunction with an across-display advantage. This relationship is consistent with the asymmetrical orienting hypothesis if you consider that display advantages, like field advantages, act as attentional signatures of the primed hemisphere. When the left hemisphere is primed, attention is biased towards the right-field and only the right-field, resulting in a within-display advantage. When the right hemisphere is primed, attention is more evenly distributed across both fields, resulting in an across-display advantage. If this correlation between field and display advantages holds true, then the investigations that were designed to explore interhemispheric interactions may have instead provided evidence for hemispheric differences in orienting attention. In order to make this more concrete, consider the predictions that emerge from the asymmetrical orienting hypothesis for each instance of hemispheric prirning. When the right hemisphere is primed (Figure 3), probes in the left field will be responded to more efficiently than those appearing in the right field. A t the same time, right hemisphere priming will permit probes in both visual fields to be processed efficiently, owing to the unique ability of the right hemisphere to allocate attention to both visual fields. Left-field and across-display advantages therefore arise for the right hemisphere because of data averaging across these two effects: left advantage from direct priming plus an equal attentional allocation to both fields. 600 • Within-display • Across-display Left Right Probe Position Figure 4. Hypothetical data of left hemisphere priming in the letter rmtching task. When the left hemisphere is primed (Figure 4), probes in the right field are generally responded to more efficiently than probes in the left field. Left hemisphere priming will also mean that within-display probes will be processed more efficiently than across-display probes, owing to the inability of the left hemisphere to orient attention to the left field. Right-field and within-display 8 advantages therefore arise for the right visual field because of data averaging over these two effects (see Leiber, 1982 for a similar pattern of results). One important consequence of this theory is that it is restricted to attentionally demanding tasks. Simply put, attention must be engaged so that the attentional bias of the primed hemisphere can be observed. Mixed-case letter rmtching fulfills this requirement because abstracting the identity of the each letter demands that focal attention be directed to each one (e.g., Posner & Dehaene, 1994; Posner & Petersen, 1990; Triesman, 1986). Consider what this task analysis of mixed-case letter matching means for this discussion. Although multiple shifts of attention can be planned concurrently, each individual shift of attention must be carried out in sequence (McPeek, Skavenski, Nakayama, 2001). Put another way, mixed-case letter matching would measure serial processing attributable to multiple serial shifts of attention (see also Woodman & Luck, 1999). One fundamental consequence of this proposal is that the hemispheres will share this attentional bottleneck because only one focus of attention can be implemented in the intact brain (Luck et al., 1989; Sperry, 1982). Therefore, the left-field and across-display advantages produced in mixed-case letter matching will arise from the right hemisphere biasing serial shifts of attention, rather than a right hemisphere advantage and interhemispheric sharing. Alternatively, if minimal demands are placed on attention, then the bias of the primed hemisphere will not be observed. Tasks that place minimal demands on attention often have different behavioral characteristics than do tasks that demand attention. For instance, the amount of time needed to detect a target is independent of the number of distractors if the target can be distinguished on the basis of one feature (e.g. when the target is a red X among green X distractors) because the information may be processed in parallel (Triesman, 1986; Triesman & Gelade, 1980). Same-case letter matching maybe a special case of a minimally demanding task because only the perceptual features of the letters must be compared and only a few letters are presented in the display, allowing the letters to be compared as a group rather than on an individual basis. If a perceptual grouping/odd ball strategy is used in same-case letter matching, then it is more efficient to process the items in parallel (Yeshurun & Carrasco, 2000). If this assessment of same-case letter matching is accurate, then it produces a different pattern of results than mixed-case letter matching because attention does not need to be shifted between the letters before a decision can be reached. As such, the attentional bias of the primed hemisphere will not be observed4. In summary, the asymmetrical orienting hypothesis proposes that field and display effects index the attentional biases of the primed hemisphere. Priming the left hemisphere causes attention to be biased towards the right field and only that side, explaining the correlation between the right-field advantage and the within-display advantage. Alternatively, priming the right hemisphere causes attention to be more evenly distributed across both fields, explaining the correlation between the weaker left-field advantage and the across-display advantage. Attention must be demanded for such biases to be observed. The right hemisphere's bias can be observed in mixed-case letter matching because abstracting the identity of each letter demands that focal attention be directed to each letter (e.g., Posner & Dehaene, 1994; Posner & Petersen, 1990; Triesman, 1986). 4 Saying that attention does not need to be shifted across items to make a decision is not equivalent to saying that precueing attention to a specific location would not enhance the processing of single, simple objects. Detecting a target can be more efficient when attention is shifted to that location before the target appears (e.g., Posner, 1980; Theeuwes, Kramer & Atchley, 1999). Be that as it may, focal attention does not need to be shifted in minimally demanding tasks (Klein, 1988; Triesman, 1986) and can, indeed, negatively influence perceptual grouping in some conditions (Yeshurun & Carrasco, 2000). 9 In contrast, the right hemisphere's bias cannot be observed in same-case letter matching because this task places minimal demands on attention and the letters may be processed in parallel. Different theories, different predictions Display advantages (i.e., within versus across) in letter notching tasks can therefore be interpreted in either of two ways. Consistent with the computational complexity theory, they can be interpreted as the outcome of a cost-benefit analysis involving parallel processing and the coordination of activity in the two hemispheres. Alternatively, they can be interpreted as signatures of hemispheric differences in orienting attention, compatible with the asymmetrical orienting hypothesis. A reasonable way to begin distmgiushing between these theories is to test whether the display and the field effects are really separable or related. The computational complexity theory proposes that field advantage will have no bearing on the parallel processing capabilities of the two hemispheres (Banich &Nicholls, 1997). As a consequence, the field and display effects should be separable. Alternatively, the asymmetric orienting hypothesis predicts that field and display effects will be related because both effects are signatures of the hemisphere whose attentional system has been primed. As a consequence, a right-field advantage will occur in conjunction with a within-display advantage because the left hemisphere is primed, and a left-field (or no field) advantage will be correlated with an across-display advantage because the right hemisphere is primed. Therefore, the experiments to be described began by pursuing the question of whether the field and display effects were dissociable or not. As will be seen, the results clearly showed that they were separable. However, at the same time, the particular pattern of results that were obtained in these experiments could not be accommodated by either the computational complexity theory or the asymmetrical orienting hypothesis. This prompted a second series of experiments in which an entirely new interpretation of these data was tested: an interpretation that was compatible with the unexpected results from the first set of experiments. This thesis began with the preliminary, but necessary, step of replicating the typical findings that have been taken as support for the computational complexity theory. Experiment 1: Replicating the basic findings This experiment attempted to replicate the basic findings associated with the letter-matching task, in which same-case letter matching should elicit a within-display advantage and mixed-case letter matching should elicit combined across-display and left-field advantages (e.g., Banich & Belger, 1990; Weissman & Banich, 2000). Methods Participants. Twenty-four right-handed university students (19 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. 10 Stimuli. The stimuli consisted of 10 upper case and 10 lower case letters (B, b, D, d, F, f, G, g, Ft h, N, n, P, p, Q, q, R, r, T, t) printed in black 48-point Ffelvetica font. From a viewing distance of 57 cm, each letter subtended a maximum of 1.1 degrees horizontally and 1.5 degrees vertically. Three letters were presented at one time. The two probes were presented above fixation, one to each half field, and the target appeared below fixation, either to the left or to the right side. The distances between the letters are depicted in the left panel of Figure 4. One of the probes matched the target on half of the trials. When a match was present, the target and matching probe appeared either to the same field (the within-display) (left side of Figure 5), or to opposite fields (the across-display) (right side of Figure 5). The target and the probes could match in one of two ways. In same-case letter matching, the target and the probes were all upper case letters (left side of Figure 5). In mixed-case letter matching, the target was a lower case letter and the probes were upper case letters (right side of Figure 5). 5.5 B "~ R 2.5 k 2.3 Within-Display Across-Display Same-Case Mixed-Case Figure 5. Examples of the different displays and complexity conditions. In the left panel, the numbers and associated arrows indicate the distance of each item (in degrees of visual angle) from central fixation. See text for further details. In this experiment, same-case and mixed-case letter matching were run in separate blocks of trials. Every other variable, including the presence or absence of a matching probe, the location of the target (left-field or right-field), and the location of matching probe with respect to the target (within-display or across-display), were equally probable and randomly selected. 11 Stimulus Presentation and Data Collection. All stimulus items were displayed on a 15-inch monitor controlled by V-scope™ (Enns & Rensink, 1992) rurining on a Macintosh computer. Procedure. After providing informed consent, the participants were instructed to maintain their gaze on the central fixation marker. They were told that a display consisting of 3-items would appear briefly and that they must decide if one of the probes matched the target (the terms "target" and "probe" were defined to the participants), a match would appear on 50% of the trials, and they were to indicate their decision with a keypress as quickly as possible, without sacrificing accuracy. After having received the instructions, the participants' were comfortably seated in a chair and their heads were stabilized 57 cm from the computer screen with a chin-rest that was self-adjusted in height. Every trial began with a fixation display consisting of a fixation marker (subtending 1.2 by 1.2 degrees of visual angle) for 500 ms, followed by the presentation of the stimulus display for 210 ms to which the subjects responded. This exposure duration is well below the average 280 ms required to execute a saccade in similar situations (Munoz, Broughton, Goldring, & Armstrong, 1998). The participants' responded with their index and middle fingers positioned upon the "g" and "h" keys on the keyboard. The response keys were counterbalanced across conditions. Half of the participants pressed "g" to indicate that a match was present and "h" to indicate that no match was present in same-case letter rmtching and used the opposite mapping in mixed-case letter rrratcliing. This situation was reversed for the remaining participants. The keys were clearly labeled. The hand used to respond remained the same for each participant, but was equally counterbalanced across participants so that half of the subjects responded with their right hand and the rernaining participants responded with their left hand. These counterbalancing procedures followed those described by Weissman & Banich, (2000) precisely. Also in accordance with Weissman & Banich (2000), all participants performed same-case letter matching first and mixed-case letter matching second. Finally, the same number of trials were given: Fourteen observations were made per cell in the match present condition, resulting in 112 match present trials [cell broken down by location of target (left-field or right-field) and location of matching probe (within-display or across-display)]. This number of observations was doubled to grand total of 224 trials for each condition (i.e., 224 trials same-case and 224 mixed-case) to keep the probability of match present and match absent trials identical. For each condition, the trials were broken down into 4 blocks of 56 trials, with an extra block given as practice. During the practice session, the experimenter monitored the performance of the participants and asked them if they had any questions before continuing with the experimental trials. All experiments in this study were conducted in accordance with the ethical standards of the American Psychological Association (4th ed.) and the Behavioral Research Ethics Board at the University of British Columbia (see Appendix IV for documentation of ethical approval). 12 Data handling, A l l trials with a reaction time exceeding 2000 ms. were eliminated from the data set. Less than 0.001 % the trials in this experiment were removed for this reason. Only data from trials containing a matching probe were used in the following analyses. Target absent data is provided in Appendix II. Efficiency scores were used as the primary dependent measure for every experiment in this thesis. To obtain these scores, the mean correct reaction time was divided by the proportion of correct responses for each cell. Thus, if no errors are made, then the efficiency score equals the mean correct reaction time for that cell. Notice that as the proportion of error increases, so does the efficiency score. For instance, if a participant obtained a mean correct reaction time of 700 ms, but was accurate only 80% of the time, an efficiency score of 875 would be attained. Simply put, efficiency scores reflect the participants' reaction time that has been corrected by their accuracy level. Using this score provides a succinct way of describing the results and avoids needless repetition because the reaction time and error analyses yield similar findings (as they should because they are linearly related within most measurable ranges). In appendix II, reaction times, errors and statistical analyses are presented for each experiment to demonstrate that efficiency scores yield similar patterns of data as reaction time and error scores. The efficiency scores were analyzed with repeated-measures analysis of variance ( A N O V A ) including the variables Match (same-case versus mixed-case), Display (within-display versus across-display), and Position of Matching Probe (left-field versus right-field). Note that for convenience of communication, hereafter the term "Probe" will be used to refer to the matching probe, unless otherwise specified. Simple effects were tested with separate A N O V A s or means comparisons (also using the F-distribution) that included the relevant independent and dependent variables. A n alpha value of .05 was adopted for all tests of significance. However, marginally significant effects (p <.10) were also considered if the effect was of theoretical interest. There were several issues to keep in mind when presenting data in graphical format. First, same-case letter matching is performed more efficiently than mixed-case letter matching (3-item displays). Indeed, the differences in efficiency are so large that plotting data from both matching conditions on the same axis obscures any field or display effects obtained for same-case letter matching (see Figure 2 for a noteworthy example). Second, sometimes one group of participants may perform more or less efficiently than another group of participants based on normal population sampling. Finally, different response modes (a topic of relevance in Experiments 4 & 5) are performed more or less efficiently. Such baseline differences in performance are of less theoretical interest than are the relative differences between displays (within vs. across) and fields (left vs. right). A common convention was used when presenting the data in graphical format that took these issues into consideration. First, same-case letter niatching was always plotted on a range of 400 efficiency values and mixed-case letter matching was always plotted on a range on 700 efficiency values to accommodate baseline differences in efficiency across matching conditions. Second, the absolute values of the y-axis were permitted to move within this pre-selected range to accommodate differences in population sampling and response modes. Thus, while the absolute 13 values presented on the y-axis change slightly across experiments and response modes, the relative distance of these values always remains identical (3-item displays). Results Data for same-case and mixed-case letter matching are shown in Figures 6 and 7 , respectively Two findings are immediately evident when comparing these graphs. First, the complexity manipulation was successful: Same-case letter matching was performed more efficiently than mixed-case letter matching. Second, different data patterns were obtained in each version: N o display or probe position advantages were obtained in same-case letter rmtching; whereas, across-display and left-field advantages were obtained in mixed-case letter rnatching. 900 850 800 750 Is700 650 CJ o o <J U u p 600 550 h 500 ° Within • Across Left Right Probe Position Figure 6 . Efficiency scores for same-case letter matching in Experiment 1 broken down by display type and probe position. Error bars represent ±1 units of standard error. 14 1400 ^ 1300 h & S 1200 0 o u u £ ft iioo 1 1000 a g O u 900 h — 800 h 700 Left Right Probe Position Figure 7. Efficiency scores for mixed-case letter matching in Experiment 1 broken down by display type and probe position. These observations were supported by the statistical analyses. The main effect of Match [F (1, 23) = 65.643, p <.05; MSe = 50825.67], indexing an overall advantage for same-case letter matching; was modified by Display[F (1, 23) = 33.769, p <.05; MSe = 6499.469] and by Probe Position [F (1, 23) = 4.870, p <.05; MSe = 22528.352]. Each letter matching condition was analyzed separately to determine the source of these interactions. N o effects were significant in same-case letter matching (all Fs < 1.72, all ps > .1). In mixed-case letter matching, responding was more efficient in the across-display [F (1, 23) = 41.105, p <.05; MSe = 9048.280] and in left-field [F (1, 23) = 10.418, p <.05;MSe = 18875.739]. Consistent with these findings, the overall analysis revealed better performance in the across-display[F (1, 23) = 23.074, p <.05; MSe = 6727.281] and in the left-field [F (1, 23) = 8.924, p < .05; MSe = 9811.843]. N o other effects were significant (all ps > . l ) . Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Discussion The important findings from this experiment were (1) the absence of any display or field effects in same-case letter matching and (2) a clear advantage for the across-display and the left-field in mixed-case letter matching. These results provide mixed support for the computational complexity theory. Obtaining an across-display advantage in mixed-case letter matching supports the notion that mterhemispheric sharing advantages complex decision-making (Banich, 1998). However, a within-display advantage was not found in same-case letter matching. Hence, mterhemispheric independence was not replicated for simple decision-making. There is no obvious reason why a within-display advantage was not obtained because the methodological procedures of Weissman & Banich (2000) were followed precisely. However, sirnilar replication failures have been noted (e.g., Banich, Passarotti, 15 & Janes, 2000; Banich et al., 2000; Belger & Banich, 1992; Brown & Jeeves, 1993; Zhang &Feng, 2000), suggesting that the within-display advantage for same-case matching may not be as reliable an effect as the across-display advantage for mixed-case matching. On the other hand, these findings provide complete support for the asymmetrical orienting hypothesis because an across-field advantage was found in conjunction with a left-field advantage for mixed-case letter matching. As mentioned in the introduction, the asymmetrical orienting hypothesis does not make predictions for same-case letter matching because it presumes that attentional biases cannot be observed when tasks place minimal demands on attention. Yet, the absence of any display or field advantages is consistent with the notion that same-case letter matching can be solved with parallel processing. Experiment 2: Does eliminating the possibility of different cognitive sets alter the data pattern? Consistent with the procedures described by Weissman & Banich (2000), the previous experiment blocked the two conditions, with same-case leiter matching always preceding mixed-case letter matching. Such a design makes the findings vulnerable to other interpretations. Perhaps each version of the letter-rmtching task elicits different patterns of data because the participants adopt different strategies in each case. For instance, same-case letter matching could be solved using an oddball detection strategy, in which the participants press "present" if one odd item appears. Adopting a cognitive set suitable for oddball detection (looking for a different item) is a fundamentally different strategy than the cognitive set assumed to occur in letter matcliing (seeing if two items are the same). The best way to eliminate such concerns is by randomly mixing the two conditions. Methods Participants. Twenty-seven right-handed university students (23 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. Procedure. This experiment was identical to Experiment 1 except that same-case and mixed-case letter imtching trials were randomly mixed together, resulting in one longer experiment consisting of 448 trials equally divided into four blocks of 112 trials. Because of this change, the hand and finger responses for each subject remained the same across the entire experiment, making such counterbalancing a between-subject manipulation. Data handling. Data from three participants, all female, were not included because their reaction time data was not recorded. Of the remaining data, less than 0.001% of all trials were eliminated because the reaction time exceeded 2000 ms. Only data from trials containing a matching probe were used in the following analyses. 16 The efficiency scores were analyzed with repeated-measures A N O V A , including the variables Match (same-case versus mixed-case), Display (within-display versus across-display), and Probe Position (left-field versus right-field). Results Data from same-case and mixed-case letter matching are shown in Figures 8 and 9, respectively. The important result from this experiment is that mixing the two matching conditions did not affect the data pattern. Same-case letter matching was performed more efficiently than mixed-case letter matching overall. N o display or field effects were obtained in same-case letter rmtching and across-display and left-field advantages were obtained in mixed-case letter rmtching. These observations were supported by the statistical analyses. The main effect of Match [F (1, 23) = 48.050, p <.05; MSe = 90126.229] was modified by Display [F (1, 23) = 26.746, p <.05; MSe = 15348.846] and by Probe Position [F (1,23) = 6.485, p <.05; MSe = 9470.490]. Each condition was analyzed separately to deterrnine the source of these interactions. Nothing achieved significance in same-case letter rmtching (all Fs < 1, all ps > .1). Responding was more efficient for the across-display [F (1, 23) - 23.205, p <.05; MSe = 36964.820] and the left-field [F (1,23) = 5.069, p <.05; MSe = 36882.856] in mixed-case letter matching. 900 — 850 I 800 750 -700 -650 -600 550 500 • Within • Across Left Right Probe Position Figure 8 Efficiency scores for same-case letter matching in Experiment 2 broken down by display type and probe position. 17 I 2 broken down by display type and probe position. Consistent with these patterns of findings, the overall analysis also revealed advantages for the across-display [F (1, 23) = 17.516, p <.05; MSe = 25556.929] and the left-field advantages [F (1, 23) = 3.606, p <.07; MSe = 36667.030]. N o other effects were significant (all ps >.05). Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Finally, Experiments 1 and 2 were compared to see if any statistical differences existed between these experiments. N o factors involving Experiment were significant in this analysis (all Fs <2.35, allps >.l) . Discussion The results of this experiment were identical to those obtained from Experiment 1. N o display or field effects were obtained in same-case letter matching and across-display and left-field advantages were obtained in mixed-case letter matching. Therefore, regardless of whether simple or complex decisions are separated or mixed together, the pattern of results remains the same. The implications for this experiment remain identical to those mentioned in Experiment 1. The computational complexity theory has achieved mixed support because only mterhemispheric sharing was replicated. The asymmetrical orienting hypothesis has achieved full support because an across-display advantage was associated with a left-field advantage. Experiment 3: Is complexity or attention responsible for the differences between same- and mixed-case letter matching? The letter-matching task produces two distinct patterns of data: N o display or field advantages in same-case letter matching and across-display and left-field advantages in mixed-case letter 18 1400 ^ 1300 u cu SH o u cu o u c/5 X ft S g o u • Within n Across 1200 1100 1000 900 800 700 Left Right Probe Position Figure 9. Efficiency scores for mixed-case letter matching in Experiment matching. These distinct patterns of data can be interpreted with the computational complexity theory or the asymmetrical orienting hypothesis that were described in the introduction. These theories not only provide different explanations for the same data patterns, but invoke different assumptions of the letter-matching task itself. The computational complexity theory proposes that the letter-matching task assesses complexity, or the "number and sorts of transformations, operations, or computations that must be performed on the input before a decision can be reached" (Banich, 1998, p 131). Consider the two versions of letter matching. In same-case letter matching, a decision maybe reached on the basis of the letters' form; no transformation is necessary. In mixed-case letter rmtching, a decision can be reached only after the identity of each letter has been abstracted from its form. Abstracting identity requires at least one more computational step, which makes mixed-case letter matching more complex than same-case letter matching. Despite these differences in complexity, both same-case and mixed-case letter nriatching are assumed to measure parallel processing because each hemisphere processes the letter(s) that it receives directly. It is only the relative balance of costs and benefits that determines which pattern of data is obtained: A within-display advantage indicates that the cost of integration was too high and an across-display advantage indicates that the benefit of parallel processing offset the cost of integration. The asymmetrical orienting hypothesis interprets the letter-niatching task in a fundamentally different way. Same-case letter rmtching may place minimal demands on attention because only a few letters are presented and only the visual shapes of the letters need to be compared. In this case, it is misleading to think of same-case letter matching as comparing letters. Instead, participants may adopt a perceptual grouping strategy. When adopting such strategies, it is more efficient to process the objects at the same time because shifting focal attention to individual objects, or parts of a display, actually impairs performance (Yeshurun & Carrasco, 1999). A sirnilar strategy could not be used for mixed-case letter matching because abstracting the identity of each letter demands that focal attention be allocated to each letter (e.g., Posner & Dehaene, 1994; Posner & Petersen, 1990; Triesman, 1986). The act of sequentially directing focal attention to each letter means that mixed-case letter matching measures serial processing attributable to multiple serial shifts of attention. The asymmetrical orienting hypothesis proposes that field and display effects index the attentional biases of the primed hemisphere. A simple consequence of this proposal is that attention must be demanded for such biases to be observed. The right hemisphere's bias cannot be observed in same-case letter matching because this task places minimal demands on attention and the letters maybe processed in parallel. However, the right hemisphere's bias can be observed in mixed-case letter matching because abstracting the identity of the each letter demands that focal attention be directed to each letter in turn (e.g., Posner & Dehaene, 1994; Posner & Petersen, 1990; Triesman, 1986). This experiment was designed to distinguish between these two interpretations of letter matching. To do so, a second irrelevant dimension (color) was added to the letter-matching task Adding color should not affect the complexity of letter notching task because only form is necessary to make a decision. Alternatively, adding a second irrelevant dimension does change a task that places minimal demands on attention into one that demands attention. Previous studies have found that perceptual grouping is hindered when an irrelevant variation in color is present in the 19 display (e.g., Snowden, 1998). If this is the case, then the results of same-case letter notching should be similar to those of mixed-case letter rmtching when color variation is added. The experiment is presented in four steps. Two different analyses were conducted that examined the influence of color, (a) same versus mixed color displays, and (b) same versus different color matches for probe-target pairs in mixed color displays. These two analyses are conducted first for same-case letter rmtching and then for mixed-case letter rmtching. Experiment 3 A: Same versus mixed color displays in same-case letter matching The asymmetrical orienting hypothesis makes no predictions for display or field advantages in same-case letter rmtching because the task may be solved using perceptual grouping strategies. It does not deny that other factors may conspire to produce display and field effects when they appear (see e.g., Theeuwes, Kramer, & Atchley, 1999), but only that they are not reflective of any pattern of asymmetrically orienting attention. Consistent with this view, same-case letter rmtching has been associated with an unstable data pattern across studies, in which within-display (e.g., Banich & Belger, 1990), no-display (e.g., Belger & Banich, 1992), and across-display (e.g., Brown & Jeeves, 1993) advantages have been reported; as have left-field (e.g., Banich & Belger, 1990), no field (Weissman, Banich, & Puente, 2000), and right-field (Banich et al, 1990, left handed subjects only) advantages. If an attentional (i.e., perceptual grouping) interpretation of same-case letter rmtching is accurate, then adding a second, task-irrelevant, dimension to the displays will eliminate the ability of such strategies to be used, necessitating that focal attention be shifted between the letters (see e.g., Snowden, 1998). In this case, same-case leiter rmtching will produce similar left-field and across-display advantages as mixed-case letter matching. Only one change was made to the same-case letter-matching task of previous experiments: Each letter was printed in either black or white against a medium gray background. Note that task complexity remains formally the same because only the shape of the letters is relevant to performance. If this theory is correct, then mixing the color of the letters should not influence performance. Methods Participants. Tliirty right-handed university students (20 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. All of these subjects performed the mixed color version of this task. The data collected from Experiment 2 provided the findings for the same color version of this task. Stimuli. The mixed color displays were identical to the displays of previous experiments, except that the letters were printed in black or white against a medium gray background. The probes always differed in color (i.e., one probe was black and the other probe was white) and the target was either black or white. When a matching probe appeared, half of the time it was the same color as the target and half the time it was a different color, as illustrated in Figure 10. 20 Within-Display Across-Display Same-Case Mixed-Case Figure 10. Examples of the mixed color displays. The upper two panels illustrate the same color relationship between the target and the matching probe. The lower two panels illustrate the different color relationship between the target and the matching probe. Please note: the background of the displays has been lightened for purposes of reproduction and does reflect the actual color used. Procedure. Participants given the mixed color condition also performed another task demanding a different response (localization response described in Experiment 4). The order of the two tasks was counterbalanced across participants. Only the data from the typical letter-imtching task (detection response) will be discussed at this time. The procedures of this task were identical to Experiment 2, except the participants were also informed that the color of the letters was irrelevant to performance. Data handling. Data from one participant was not included because she performed at chance during this task. O f the remaining data, less than 0.001% of all trials were eliminated because the reaction time exceeded 2000 ms. Only data from trials in which a matching probe was presented were used in the following analyses. 21 across-display), and Probe Position (left-field versus right-field). In the interest of brevity, only those effects involving the Color variable will be presented at this time. Results Data for same and mixed color displays are shown in Figures 11 and 12, respectively. The important finding from this experiment is that mixing color changed the display and field effects for same-case letter matching. N o display or field effects were observed when all letters were the same color, whereas, across-display and left-field advantages were observed when the letters were mixed colors. These observations were supported by the statistical analyses. The Color x Display interaction achieved significance [F (1, 51) = 6.766, p <.05; MSe = 2712.161]. To determine the source of this interaction, separate analyses were conducted for same and mixed color displays. N o display or field effects were significant in the same color displays (all Fs < 1), but the mixed color displays yielded a significant left -field advantage [F (1, 28) = 9.592, p < .05; MSe = 9514.381] and a significant Displayx Probe interaction [F (1,28) = 9.431, p <.05; MSe = 1346.890], indicating that the across-display advantage was larger in the right-field [F (1, 29) = 41.971, p <.05] than in the left-field [F (1, 29) = 4.560, p <.05]. Reaction time and error scores produced a similar pattern of results, as shown in appendix II. 900 — 850 800 750 700 650 600 550 500 ° Within Q Across Left Probe Position Right Figure 11. Efficiency scores for the same color displays in same-case letter rmtching in Experiment 3 broken down by display type and probe position 22 Left Probe Position Right Figure 12. Efficiency scores for the mixed color displays in same-case letter matching in Experiment 3 broken down by display type and probe position Discussion The results from this analysis are straightforward. N o field or display effects were obtained in same color displays and left-field and across-display advantages were obtained in mixed color displays. These findings are more consistent with an attentional interpretation of the letter-rmtching task. Perceptual grouping is hindered when a second, irrelevant, dimension is added to the displays (Snowden, 1998), causing focal attention to be shifted across the letters in the display. These findings could not have been anticipated by the complexity interpretation because color wa irrelevant to decision making. However, the definition of "complexity" may not be this clear-cut. Banich (1998) described an unpublished study in which adding an irrelevant dimension was accompanied by data patterns suggestive of "complex" processing. This study was not presented as contradictory evidence, but rather as supporting evidence for the computational complexity theory. Therefore, perhaps the strict definition of computational complexity originally offered by Banich (Belger & Banich, 1992) has now been modfied to include attentional load as a task complexity factor (Banich, 1998). If having to disregard task-irrelevant color variation increases the complexity of the matcliing task enough to make interhemispheric sharing a better strategy, then the immediate relationship between the target and matching probe should change the pattern of results. That is, if discounting color differences between the target and its matching probe increases the complexity of the decision, then the across-display and left-field advantages should be stronger when the target and the rnatching probe do not share the same color. Conversely, a smaller display or field advantage should occur when they are the same color. Remember that the computational 23 complexity theory assumes parallel processing of letters on each side of the display, therefore, variations in color will increase the complexity of matching process, not the initial letter registration process. On the other hand, according to the attentional interpretation, sirnilar across-display and left-field advantages should occur regardless of the immediate color relationship between the target and the matching probe because merely adding the color variation means that the perceptual grouping strategy cannot be used. This comparison was tested in the next analysis. Experiment 3 B : Color relationship in same-case letter matching In mixed color displays, the target is one color (i.e., either black or white) and the two probes are different colors (i.e., one black and one white). Therefore, in each display, the target letter shares the same color as one of the probes and the other probe is unique in color. This arrangement permitted a modified complexity interpretation to be evaluated, which predicts that having to disregard a color difference between the target and the rmtcliing probe increases the complexity of the decision. If this interpretation is accurate, then an across-display advantage will be obtained when the target and rmtching probe are different colors because discounting color increases the complexity of the decision. When they are the same color, no display advantage will be obtained because no color difference has to be disregarded, lessening the complexity of the decision. Put another way, the across-display advantage that was observed in mixed color displays in the previous analysis occurred because more complex and less complex decisions were averaged together. Alternatively the attentional interpretation predicts that the obtaining the across-display advantage was independent of the immediate color relationship between the target and the matching probe because merely adding the second color causes focal attention to be shifted among the items in the display. Methods This experiment is a more detailed analysis of the same-case letter rmtching condition with the mixed color displays. As such, I will only discuss the relevant methodological information at this time. For greater details, see Experiment 3A. Participants Only the 29 participants who contributed data to the mixed color letter-matching task were used in this analysis. Data handling The efficiency scores were analyzed with mixed-design repeated-measures ANOVA including the variables Display (within-display versus across-display), Probe Position (left-field versus right-field), and target-probe Color Relationship (same versus different). Only the significant findings that include the color relationship variable will be described. Results 24 Data for Color Relationship is shown in Figure 13. The important finding from this analysis is that color relationship between the target and probe did not affect the pattern of findings. These observations were supported by the statistical analyses. The color relationship between the target and the matching probe did not affect performance (all Fs <2.5; all ps >. 1). Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. u Ol U u u O O u u CD w ft • i - H 02 u 01 u u o u 800 750 700 650 600 H ~ 550 500 " 450 h 400 ° Within 1 1 Across Same Different Color Relationship Figure 13. Efficiency scores for the mixed color displays in same-case letter niatching in Experiment 3 broken down by display type and color relationship between the target and probe. Note that "same" and "different" refer to the color match and mismatch between the target and probe. Discussion The results from this analysis are straightforward. The immediate color relationship between the target and the matching probe did not influence the across-display or left-field advantages. This finding supports the "attentional" interpretation of the letter matching task, which maintains that adding a second, albeit irrelevant, dimension hinders perceptual grouping strategies (e.g., Snowden, 1998), causing the attention to be shifted among the letters in the display. Had the complexity interpretation been correct, an across-display advantage would have been obtained only when the target and matching probe were different colors. This was not the case. Experiment 3G Same versus mixed color displays in mixed-case letter matching In this analysis, same and mixed color displays were compared to detenriine their influence on mixed-case letter matching. The complexity and attentional interpretations predict the same data pattern, but for different reasons. The complexity interpretation predicts that adding color will not affect the data pattern 25 because color is irrelevant to decision-making. Alternatively, the attentional interpretation predicts that adding color will not significantly change the data pattern because the identity of each letter must already be abstracted before a decision can be reached. Because this step is already required, abstracting the identity across the color dimension would not have any added influence. Methods The methods and analyses described here are identical to those used in Experiment 3-A except that only the data from mixed-case letter-rmtching are considered. Results Data for the same and mixed color displays are shown in Figures 14 and 15, respectively. The important finding from this experiment is that adding color did not affect the magnitude of the display and field effects for mixed-case letter matching. These observations were supported by the statistical analyses. Color relationship was not found to affect any variable in anyway (all Fs <2.1, all ps >.l) . Reaction time and error scores produced a similar pattern of results, except the Color Relationship X Probe interaction was significant for RT, indicating that the disadvantage for the right probe was exaggerated in the color condition, a result that was not predicted by either interpretation. 1400 ^ 1300 u QJ QJ U u fi O O y u QJ " O u 1200 1100 IOOO 900 800 700 • Within 1 1 Across Left Probe Position Right Figure 14. Efficiency scores for the same color displays in mixed-case letter matching condition in Experiment 3 broken down by display type and probe position 26 u C CU u 1400 1300 u cu H 1200 ft 1100 H ft- 1000 -M CU u i i 900 r H O u 800 700 Left Probe Position Right Figure 15. Efficiency scores for the mixed color displays in mixed-case letter matching condition in Experiment 3 broken down by display type and probe position. Discussion The results are straightforward. Adding color did not influence of the pattern of findings for mixed-case letter matching, as both displays were associated with similar across-display and left-field advantages. The findings from the efficiency analysis were predicted by both the complexity and the attentional interpretations of letter rmtching. In the next experiment, I analyzed the mixed color displays for mixed-case letter matching because the color relationship between the target and matching probe will help delineate between these two interpretations. Experiment 3D: Color relationship in mixed-case letter matching In the previous analysis, it was found that adding color did not affect mixed-case letter rmtching. This finding was consistent with both interpretations of the letter-matching task Here the color relationship between the target and probe in mixed color displays was analyzed to help determine which interpretation was better supported. A complexity interpretation predicts that color is irrelevant to decision-making and therefore will not influence the data pattern. However, a modified complexity interpretation, which assumes that adding color increases the complexity of decision-making, may also be forwarded. In this case, a larger across-display advantage is expected when the target and probe are different colors because the magnitude of the across-display advantage increases as the complexity of the decision increases (BelgerSt Banich, 1992). 27 Up to now, an attentional interpretation would have predicted that the color relationship between the target and matching probe would not influence performance because the attentional demands were roughly equivalent for same and mixed color displays in mixed-case letter matching. However, a strong case can be made for another prediction and that is, responding maybe more efficient when the target and the rmtching probe are different colors. Consider for a moment the design of the displays. When the target and rmtching probe are different colors, the matching probe has a unique feature compared to the other letters in the display It is the only letter appearing in that color. Having a unique feature will increase its salience, drawing attention to it (e.g., Egeth & Yantis, 1997). If anything, an advantage of saliency should be most evident in the condition performed most poorly the within-display. Similar saliency effects were not predicted for same-case letter rmtching because performance is typically too efficient, making such saliency effects difficult, if not impossible, to discern. Methods The methods and analyses described here are identical to those used in Experiment 3-B except only the data from mixed-case letter rmtching are considered. Results u CD S o o y u •1-H Q2 •1-H 4—1 W g fi o u 1200 1100 -1000 -900 800 " 700 -600 -500 -400 Same Different Color Relationship Figure 16. Efficiency scores for the mixed color displays in mixed-case letter niatching in Experiment 3 broken down by display type, probe position, and color relationship between the target and probe. Data for Color relationship are shown in Figure 16. The important finding is that a color difference between the target and probe advantaged performance most notably for the within-display. 28 These observations were supported by the statistical analyses. Responding was more efficient when the matching probe was unique in color [F (1, 28) = 4.686, p <.05; MSe = 20463.944]. Although the interaction was not significant (p >.l), planned comparisons of the Displayx Color Relationship interaction revealed that a color difference between the target and probe was advantageous in the within-display (F (1, 28) = 3.096, p < .09), but did not affect performance in the across-display [F (1, 28) < 1, p >.l]. Error scores produced an identical pattern of results as the efficiency scores, although the trend towards an advantage for the color difference between target and probe in the within-display missed significance (F (1, 28) = 2.573, p <.12). RTs produced a similar pattern of data, but the main effect for Color relationship and the planned comparison did not achieve statistical significance (p >.l). Discussion In this analysis, performance was facilitated in the within-display when the target and rmtching probe were different colors. This finding is consistent with a modified attentional interpretation of the letter-rmtching task, in which the color difference between the target and rmtching probe increased the saliency of the rmtching probe compared to the other items in the display. The complexity interpretation was not supported by these findings. If anything, the across-display advantage was smaller when the target and matching probe were different colors, exactly opposite to its prediction. Taken together, the fmdings from this analysis are more consistent with an attentional interpretation of the mixed-case letter-rmtching task. Overview of Experiments 1- 3 In Experiments 1 and 2, the letter-rmtching task was associated with no display or field advantages in same-case letter rmtching and across-display and left-field advantages in mixed-case letter rmtching. This pattern of data is consistent with a complexity interpretation of the letter-rmtching task, which assumes that letter rmtching is always solved with parallel processing. Modifying complexity tips the balance of interhemispheric interactions to favor interhemispheric independence when the decision is simple and interhemispheric sharing when the task is complex (e.g., Banich, 1998). The same task can be interpreted within an attentional framework, which maintains that the two versions of letter rmtching are solved in fundamentally different ways. Same-case letter rmtching places minimal demands on attention, permitting a decision to be reached in parallel. Mixed-case letter rmtching places moderate demands on attention and requires serial processing (Posner & Dehaene, 1994; Posner & Petersen, 1990; Triesman, 1986). To assess which interpretation is better supported, Experiment 3 added a second, irrelevant, dimension to the letter-rmtching task. All told, the findings from this experiment are more compatible with an attentional interpretation and less compatible with a complexity interpretation. 29 Consider first the findings from same-case letter niatching. Same color displays were associated with no display or field advantages, whereas mixed color displays were associated with across-display and left-field advantages (Experiment 3-A). This data pattern was predicted by an attentional interpretation because adding a second irrelevant dimension hinders use of a perceptual grouping strategy when solving this task. The complexity interpretation of same-case letter matching did not predict this outcome because the complexity of the decision remained the same (only the form of the letters was relevant to the task). However, it maybe possible that this strict definition of complexity was not a fair interpretation. Perhaps, having to disregard color tipped the balance of costs and benefits in favor of interhemispheric sharing. To assess the accuracy of this alternative explanation, the mixed color displays were analyzed to see if the immediate color relationship between the target and the probe influenced performance. In this case, the modified complexity interpretation predicts that an across-display advantage will be obtained only when the target and matching probe are different colors because only at that time must color be disregarded. The attentional interpretation predicts that this immediate color relationship will not be of influence because merely adding the second dimension causes attention to be demanded. Again, the findings were more consistent with an attentional view (Experiment 3-B). Consider now the findings from mixed-case letter matching. Comparing the same color and mixed color displays yielded similar results (Experiment 3-C). This finding was consistent with both interpretations of letter matching, but for different reasons. The complexity interpretation predicts a null effect because color was irrelevant to decision making. The attentional interpretation predicts a null effect because serial processing is already demanded, causing a color difference to not have any additional influence. Finally, modified complexity and attentional interpretations were assessed with the mixed color displays to see if the color relationship between the target and the matching probe influenced the pattern of findings. Recall, that a modified complexity view proposes that disregarding color will increase the complexity of the decision. If this is the case, then a color difference between the target and the probe will exaggerate the across-display advantage because as complexity increases, so to should the across-display advantage (Belger & Banich, 1992, 1998). A modified attentional interpretation proposes that saliency effects maybe evident. It is well documented that increasing the saliency of an item increases the efficiency of its detection (e.g., Egeth & Yantis, 1997). Owing to the 3-items and 2 colors of the displays, the matching probe will be more salient when it is a different color than the target because it will be the only letter appearing in that color. Saliency effects should be most notable in the condition that is performed most poorly, in other words, the within-display should demonstrate the greatest benefit. The obtained findings were consistent with an attentional interpretation, as a color difference between the target and the matching probe chiefly advantaged the within-display and had little influence on the across-display (see Figure 14). In general, the analyses from this experiment are more consistent with an attentional interpretation of the letter-matching task This interpretation is not without consequence because it suggests that same-case letter matching is solved in parallel and that mixed-case letter matching demands serial processing. This attentional interpretation of the letter- niatcliing task is also very much at odds with the computational complexity theory, which presumes that both same-case and mixed-case letter matching can be performed using parallel processes. If mixed-case letter rmtching is indeed solved serially, then a parallel processing interpretation (i.e., interhemispheric sharing) cannot be sustained. 30 Experiment 4: Does manipulating field advantage alter the display effects? The findings from Experiment 3 suggest that the leiter-rmtching task gauges the consequences of attention: Same-case letter rmtching measures parallel processing because niinirnal demands are placed on attention and mixed-case letter rmtching measures serial processing because moderate demands are placed on attention. Experiment 4 was designed to assess the asymmetrical orienting hypothesis in another way. Across the studies exploring interhemispheric interactions, a right-field advantage often appears in conjunction with a within-display advantage and a left-field advantage is typically associated with an across-display advantage (see Appendix 1). This relationship is consistent with the asymmetrical orienting hypothesis, which proposes that both field and display effects are attentional signatures of the primed hemisphere. When the left hemisphere is advantaged, attention is biased towards the right-field and only to that side because the left hemisphere orients only to the right. When the right hemisphere is advantaged, attention is more evenly distributed across fields because the right hemisphere orients to both sides of space. Previous work in visual search and in global-local processing has demonstrated that changing the response demands of a task reverses the field advantage, in that detection tasks favor the left-field and localization tasks favor the right-field (Fecteau, Enns, & Kingstone, 2000; Fecteau, Enns, & Kingstone, in preparation). With regards to letter rmtching, partial support for such response effects already exists because left-field advantage is typically reported when the participants detect the presence or absence of a matching probe (e.g., Experiments 1-3; Banich & Belger, 1990; Weissman & Banich, 2000). Changing the letter-rmtching task to demand a localization response is simple: A matching probe will always be present and the participants must indicate the side to which the matching probe appeared (left or right). Similar localization responses have been found to elicit a right-field advantage (Fecteau, Enns, & Kingstone, 2000; Fecteau, Enns, & Kingstone, in preparation). If the field and display effects are related and the detection task elicits joint left-field and across-display advantages and the localization task elicits joint right-field and within-display advantages; then support for the asymmetrical orienting hypothesis will be obtained. On the other hand if the field and the display effects are separable and an across-display advantage is constantly observed, then support for the computational complexity theory will be garnered. Methods Participants. Thirty right-handed university students (20 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. These subjects also provided the data for the mixed color displays in Experiment 3. Stimuli. The stimulus displays were identical to those used in the color version in Experiment 3. 31 Procedure The procedures for the detection response were described in Experiment 3. The localization response differed from the detection response in several important ways. A target was always present and the participants' task was to indicate where the rmtching probe appeared in space, either to the left ("g" key) or to the right ("h") key. The mapping reflected the left or right position of the match and was not counterbalanced across subjects. Because a match was always present, only half of the trials were required to maintain power at the same level. The 224 trials were equally divided into 2 blocks, with 56 extra trials given as practice. Half of the subjects performed the localization response first and the detection response second. For the rermining subjects, this situation was reversed. Otherwise, the procedural details were identical to the previous experiments. Data handling. Data from one participant were eliminated from this analysis because she performed at chance during the detection task. Of the rermining 29 participants, less than 0.001% of all trials were eliminated because the response-latency exceeded 2000 ms. Only data from trials containing a matching probe were used in the following analyses. The efficiency scores were analyzed with repeated-measures ANOVA including the variables Response (detect versus locate), Match (same-case versus mixed-case), Display (witliin-display versus across-display), Probe Position (left-field versus right-field) and Color Relationship (same as versus different than target). In keeping with the main objective of this experiment (i.e., to establish the influence of the different responses on letter rmtching), only the influence of Response on the Display, and the Probe Position variables will be described. Results Same-Case Letter matching Data for same-case letter rmtching are shown in Figure 17 for the detection response and Figure 18 for the localization response. As evident in these graphs, the two response conditions produced similar left-field and across-display advantages. The across-display advantage was larger in the right-field than in the left-field. 32 900 — 850 r u J H J -c u cu o y CO ^ OH <-» - \ .a 2 u 25 H G O u 800 -750 -700 650 600 -550 -500 Left Right Probe Position Figure 17. Efficiency scores for same-case letter matching in Experiment 4 when a detection response was used. Broken down by display type and probe position 800 — 750 700 650 600 £ £ 550 500 450 400 • Within m Across Left Right Probe Position Figure 18. Efficiency scores for same-case letter matching in Experiment 4 when a localization response was used. Broken down by display type and probe position These observations were supported by the statistical analyses. Same-case letter matching was associated with left-field [Detect F (1, 28) = 9.592, p <.05, MSe = 19028.761; Locate F (1, 28) 4.495; p <.05, MSe = 11276.464] and across-display [Detect F (1, 28) = 29.189, p <.05, MSe = 3423.804; Locate F (1,28) 9.380, p <.05, MSe = 2529.183] advantages for both responses. The interaction between these variables was also significant [Detect F (1, 28) = 9.431, p <.05; MSe 33 2693.780; Locate F (1,28) 4.868, p <.05; MSe = 1283.443], indicating that the across-display advantage was large in the right-field (Fs >21, ps <.05) and small, if not absent, in the left -field [Detection, F (1, 28) = 4.560, p <.05; Localization, F (1, 28) <2.2, p >. l ] . Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Mixed-case letter rmtching The data for mixed-case letter rmtching are shown in Figure 19 for the detection response and Figure 20 or the localization response. As evident in these graphs, the two response conditions produced similar left-field and across-display advantages. These observations were supported by the statistical analyses. Mixed-case letter rmtching was associated with left-field (Detect, F (1, 28) = 18.088, p <.05, MSe = 92081.654; Locate F (1,28) = 26.151, p <.05, MSe = 10682.658) and across-display (Detect, F (1, 28) = 24.533, p <.05, MSe = 69591.763; Locate F (1, 28) = 5.982, p <.05, MSe = 18230.869) advantages for both responses. The interaction between these variables was not significant for either response (Detect, F (1, 28) < 2.8, p >. l , M S e =36811.229; Locate F (1,28) = <19, p > . l , M S e = 19665.356). Consistent with these patterns of findings, the overall analyses revealed better performance in the across-display (Detect, F (1, 28) = 32.689, p <.05, MSe = 40278.253; Locate F (1, 28) = 8.890, p < .05, MSe = 13189.651) and in the left-field (Detect, F (1, 28) - 17.884, p <.05, MSe = 82497.043; Locate F (1,28) = 23.097, p <.05, MSe = 12296.883). The interaction of these variables was significant for the detection task[F (1, 28) = 6.222, p <.05, MSe = 18347.035] and approached significance for the localization task [F (1, 28) = 3.353, p <.08; MSe = 10823.636], indicating that the across-display advantage was larger for the right-field. Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. 1400 ~1300 u g g 1200 0 o ^ ft 1100 1 S IOOO 900 n g o " 800 700 • Within l Across Left Right Probe Position Figure 19. Efficiency scores for mixed-case letter matching in Experiment 4 when a detection response was used. Broken down by display type and probe position 34 1100 1000 900 800 700 g 600 o ^ 500 h 400 • Within m Across Left Probe Position Right Figure 20. Efficiency scores for mixed-case letter matching in Experiment 4 when a localization response was used. Broken down by display type and probe position Discussion The most notable finding from this experiment is that the response manipulation did not reverse field advantage. Both the detection and the localization responses evoked similar left-field and across-display advantages. The failure to elicit a right-field advantage with the localization response is puzzling because it does so in visual search tasks (Fecteau, Enns, &t Kingstone, 2000) and in global-local target identification tasks (Fecteau, Enns, & Kingstone, in preparation). A secondary finding was that the across-display advantage was larger, if not present only in the right-field in same-case letter matching. Such an interaction has been noted before (Banich et al., 2000) although its theoretical implications were not addressed in that study. These findings have important implications for both theories under consideration. The computational complexity theory predicts that an across-display advantage will be obtained regardless of how the participants respond. In support of this theory, the localization response also elicited an across-display advantage. Contrary to the computational complexity theory, the across-display advantage was significantly larger in the right-field in same-case letter matching. Such a finding implies that interhemispheric sharing is more beneficial for the left hemisphere— an interpretation that cannot be accommodated by the computational complexity theory in its current form The asymmetrical orienting hypothesis also achieved mixed support. In its favor, an across-display advantage was associated with a left-field advantage, confirming the correlation established in previous experiments. 35 However, the interaction between field and display effects is particularly troublesome for the asymmetrical orienting hypothesis. The across-display advantage was larger in the right-field in same-case letter rmtching. However, notice that the term "across-display advantage" only considers relative differences. If absolute differences are considered instead, then it becomes quite clear that the term "across-display advantage" misrepresents its cause. As evident in Figures 18 and 19, the across-display advantage does not arise from an advantage per se. but from a disadvantage in the right-field, within-display trials. Thus, instead of the right hemisphere sharing attention across fields, this interaction indicates that performance was notably impaired for the right-field. Considering that attention should be evenly distributed across-fields when the right hemisphere is advantaged; fmding that attention is underrepresented in one field is clearly at odds with the predictions of the asymmetrical orienting hypothesis. In summary, this experiment provided imperfect support for both theories. The response manipulation used to alter field advantage did not work and instead both responses elicited left-field and across-display advantages: a data pattern that supports both theories. Be that as it may, the findings from this experiment have negative consequences for both theories. The main problem was the interaction between field and display, indicating that the across-display advantage was larger in the right-field. The computational complexity theory cannot accommodate this fmding because a larger across-display advantage in the right-field indicates that both hemispheres are not equal benefactors of interhemispheric sharing. The asymmetrical orienting hypothesis cannot accommodate this finding because the interaction is better described as a within-display disadvantage for the right-field. Considering that attention should be more evenly distributed across both fields when the right hemisphere is advantaged, finding that attention is less efficiently directed to the right-field contradicts the basic tenets of the asymmetrical orienting hypothesis. Experiment 5: Does another task and/or different set-sizes modify field and display effects? In Experiment 4, the response manipulation used to alter field advantage was unsuccessful. Instead of the detection response eliciting a left-field advantage and the localization response eliciting a right-field advantage, a left-field advantage was obtained in both cases. There are at least four reasons why the localization response did not elicit a right-field advantage. 1. Mixed-case letter matching maybe so dependent upon right hemisphere processing that the localization response could not possibly elicit a right-field advantage. Some support exists for this interpretation, in that mixed-case letter-matching task consistently elicits a left-field advantage (e.g., Banich & Belger, 1990; Weissman & Banich, 2000; Weissman, Banich, & Puente, 2000). However, other evidence challenges this claim because (1) mixed-case letter rmtching also produces right-field advantages in some conditions (e.g., Brown & Jeeves, 1993; Davis & Schmidt, 1973) and (2) mixed-case letter rmtching favors the left hemisphere of split-brain patients (Evitar & Zaidel, 1994). In all likelihood then, the left-field advantage probably does not index a strong right hemisphere advantage. 2. A second explanation arises from methodological issues. Two changes were made to Experiment 4 that may have eliminated the right-field advantage for the localization response. For one, response is usually a between subjects manipulation, in which half of the participants use a detection response and the others use a localization response. Perhaps, having to participate in 36 both conditions diminished the effectiveness of the localization response. Although possible, this explanation seems unlikely because the order in which the different responses were performed did not result in any main effects or interactions (all ps > .05). 3. Another methodological issue concerns set size. Small set sizes (2 items) were performed alongside large display sizes (up to 24 items) in visual search (Fecteau, Enns, & Kingstone, 2000). Perhaps having to adopt a more flexible cognitive set (required by multiple display sizes) is necessary to elicit a right-field advantage in localization responses. Again, this explanation seems unlikely because a global and local processing experiment elicited a right-field advantage using display sizes of 2 (Fecteau, Enns, & Kingstone, in preparation). Taken together, little support exists for the two methodological explanations. Even still, this experiment has corrected for these potential methodological problems. 4. Finally, mixed-case letter-matching may elicit a left-to-right bias in the order that letters are mentally compared. Consider that a left-field advantage is the telltale sign of left-to-right bias when letters are displayed to both sides of fixation (e.g., Heron, 1957; Kimura, 1966; Lubow, et al., 1994; Scheerer, 1972,1973). This letter-matching task is vulnerable to such an explanation because at least one letter is presented to each half field. In this experiment, the two methodological issues for failing to obtain a right-field advantage (points 2 & 3) were explored. Sirnilar to Experiment 4, response manipulations were used to elicit a right-field advantage; however, several important changes were made to increase the likelihood of success. First, a color-report response was included because a right-field advantage is elicited when responses demand an attribute of the target to be reported (e.g., Lamb & Roberston, 1988,1989). Second, each subject performed only one condition, eliminating possible carryover effects from one response to another. Third, 3-item displays were mixed with more demanding 9-item displays to ascertain if adopting a more flexible cognitive-set elicits a right-field advantage. For both theories under consideration, the 9-item displays should yield similar display and field effects as do the 3-item displays. If anything, the computational complexity theory predicts that the magnitude of the across-display advantage may even increase because as the complexity of the task increases, so too should the advantage for interhemispheric sharing (Belger & Banich, 1992, 1998). If these modifications are successful, then the detection response will produce a left-field advantage and the localization and color-report responses will produce right-field advantages. Successfully producing a right-field advantage will help to ascertain if the field and display effects are related or separable and to determine if computational complexity theory or the asymmetrical orienting hypothesis is better supported. If, on the other hand, a left-field advantage is consistently obtained, then no methodological factors were responsible for the failure to evoke a right-field advantage in Experiment 4 and another approach will be used to elicit a right-field advantage. Methods Participants. Ninety right-handed university students (71 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. The participants were equally divided among the three different response conditions. 37 Stimuli. The 3-item displays were identical to those used in Experiment 4. As illustrated in Figure 21, the nine-item displays consisted of eight different upper case probe letters, arranged so that four probes appeared in each visual field, positioned 2.5 or 5.1 degrees above fixation and 3 or 6.1 degrees to the right or left of fixation. The position of the matching and non-matching probes was randomly determined. The target always appeared 2.5 degrees below fixation and either 2.3 degrees to the left or to the right. R T D Within-Display Same-Case Across-Display Mixed-Case Figure 21. Illustration of the 9-item displays. Please note: the background of the displays has been lightened for purposes of reproduction and does reflect the actual color used. Procedures For the detection and localization responses, the procedures were identical to Experiment 4, except that 3-item displays were mixed with 9-item displays. To accommodate the increased number of conditions, the number of observations per cell was decreased to 10 (from 14), which increased the number of trials to 660 for the detection response and 330 for the localization response. A n additional 6 participants performed in each response condition to accommodate for this loss in power. The color-report response was identical to the other responses, except the participants' job was to report the color of the rmtching probe, either being black or white. For half of the subjects, the "g" key was used to indicate black and the "h" key was used to indicate white. For the rermining subjects, this mapping was reversed. Similar to the localization response, a rmtching probe was always presented, resulting in 330 experimental trials. 38 The instructions stressed accuracy over speed. In every other way, the procedural details were identical to the other experiments. Data Handling Data from one participant were eliminated from each of the detection and localization responses because accuracy was too low. Data from two participants were elirninated from the identification task because a computer malfunction caused the reaction times to not be recorded. Less than 0.01% of all trials were elirninated from the remaining data because the response-latency exceeded 2000 ms. The efficiency scores were analyzed with mixed-design, repeated-measures ANOVA including the variables Response (detect versus locate versus color-report), Display Size (3 versus 9 items), Match (same-case versus mixed-case), Display (within-display versus across-display), Probe Position (left-field versus right-field) and Color Relationship (same versus different). In keeping with the main objective of this experiment (i.e., to establish the influence of the different responses on letter rmtching), only the influence of Response and Display Size on the Display, and the Probe Position variables will be described. Results 3-item displays Same-case letter rmtching. Data for 3-item same-case letter matching are shown in Figure 22 for the detection response, in Figure 23 for the localization response, and in Figure 24 for the color-report response. The most important finding was that the typical left-field and across-display advantages were obtained with every response. The statistical analyses were consistent with these observations. For the detection and the color-report responses, performance was more efficient in the left-field [Detect F (1, 28) = 10.165, p < .05, MSe = 60823.007; Color Report F (1, 27) = 15.920, p <.05, MSe = 17733.042] and in the across-display [Detect F (1, 28) = 14.066, p <.05, MSe = 16521.169; Color Report F (1, 27) = 27.006, p <.05, MSe = 12355.422]. For the localization response, the Display X Probe Position was significant [F (1,28) = 7.870, p <.05, MSe = 6252.649], indicating that the across-display advantage was significant in the right-field [F (1, 28) = 25.628, p <.05], but not in the left-field [F (1, 28) < 1.2, p > .1]. Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. 39 1000 — 950 900 850 -800 -750 -700 -650 -600 • Within • Across Left Right Probe Position Figure 22. Efficiency scores for 3-item same-case letter matching in Experiment 5 when detection response was used. Broken down by display type and probe position 800 — 750 h <u g 700 O O £^650 ft C>600 ' tJ 550 §500 u 450 h w 400 • Within • Across Left Right Probe Position Figure 23. Efficiency scores for 3-item same-case letter rmtching in Experiment 5 when localization response was used. Broken down by display type and probe position 40 1300 ^1250 1200 1150 1100 1050 1000 950 900 • Within • Across - T T -T - 1 Hill WPWWVMN Left Right Probe Position Figure 24. Efficiency scores for 3-item same-case letter rmtching in Experiment 5 when a color-report response was used. Broken down by display type and probe position Mixed-case letter rmtching. Data for 3-item mixed-case letter rmtching are shown in Figure 25 for the detection response, in Figure 26 for the localization response, and in Figure 27 for the color-report response. Two important results are immediately evident. First, all responses elicited left-field and across-display advantages. Second, performance was notably poor for the right-field, within- display trials. These observations were supported by the statistical analyses. The Display x Probe position was significant for the detection and localization responses [Detect F (1, 28) = 32.579, p <.05, MSe = 73214.342; Locate F (1, 28) = 12.200, p <.05, MSe = 24253.596] and trended towards significance for the color-report task[F (1,27) = 3.417, p <.08, MSe = 56188.046], all indicatbg that an across-display advantage was obtained in the right-field (Fs > 13, ps <.05) but not in the left-field (Fs < 1.3, ps >.l). Consistent with this pattern, every response yielded left-field [Detect F (1, 28) = 16.649, p <.05, MSe = 154700.547; Locate F (1, 28) = 19.5, p <.05, MSe = 37168.149; Color Report (1,27) = 4.201, p <.06, MSe = 50157.159] and across-display advantages [Detect F (1,28) = 19.411, p <.05, MSe = 106255.337; Locate F (1,28) = 28.247, p <.05, MSe = 22347.251; Color Report (1, 27) = 22.880, p <.05, MSe = 26756.648]. Reaction time and error scores produced a similar pattern of results, as shown in appendix II. 41 1500 ^ 1400 r u % I 1 3 0 0 o o y u V, ft 1200 •8 £ lioo h 1000 h o — 900 800 Left Right Probe Position Figure 25. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when detection response was used. Broken down by display type and probe position 1100 5 IOOO u CD ?H JH fi O O y u CD ^ ft <-> ^ .y s w g 600 h o " 500 900 800 h 700 h 400 • Within 1 1 Across Left Right Probe Position Figure 26. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when localization response was used. Broken down by display type and probe position 42 u cu SH O u 1600 1500 1400 cu u O # X 1300 u ^ j§ g 1100 £ IOOO 1200 900 • Within n Across Left Right Probe Position Figure 27. Efficiency scores for 3-item mixed-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position 9-item displays Same-case letter matching. Data for 9-item same-case letter matching are shown in Figure 28 for the detection response, in Figure 29 for the localization response, and in Figure 30 for the color-report response. Inspection of these graphs reveals that the Display and Probe Position advantages largely disappear when the display consists of 9-items. 2500 — 2250 „ "§ 2000 CU 5H J-H U o o u u ^ ^ 1 5 0 0 u £ cu 1750 E-71250 % 1000 S 750 3, 500 250 0 • Within H Across Left Right Probe Position Figure 28. Efficiency scores for 9-item same-case letter rmtching in Experiment 5 when a detection response was used. Broken down by display type and probe position 43 2500 _ 2250 ^ 2 0 0 0 S o 1750 y u * ^ 1500 g 1 2 5 0 2 2 1000 QJ M o u 750 500 250 0 Left Right Probe Position Figure 29. Efficiency scores for 9-item same-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 3000 ^ 2 7 5 0 OJ I 2 5 0 0 8 8 2 2 5 0 % a 2000 u C QJ H 1750 3 1500 S 1 2 50 8 ,1000 750 500 ° Within n Across :: I:!!!!!;; : Left Right Probe Position Figure 30. Efficiency scores for 9-item same-case letter matching in Experiment 5 when a color-report response was used. Broken down by display type and probe position The statistical analyses were largely consistent with these observations. Because no consistent pattern was observed across responses, each response will be discussed in sequence. 44 Detection response. N o significant display or field effects were obtained (ps >.l) . Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Localization response. A Display x Probe Position interaction was obtained [F (1,28) = 10.001, p <.05, MSe = 734738.649], indicating that an across-advantage was only obtained in the left-field [F (1, 28) = 9.218, p <.05] and not in the right-field [F (1,28) <2.1, p >. l ] . N o other effects involving Display or Probe Position were significant (ps > .1). Reaction time and error scores produced a similar pattern of results, as shown in appendix II. Color-report response. A right-field advantage was obtained overall [F (1, 27) = 4.023, p <.06, MSe = 497138.805]. N o other effects involving Display or Probe Position achieved significance (ps >.l) . Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Mixed-case letter matching. Data for 9-item mixed-case letter matching are shown in Figure 31 for the detection response, Figure 32 for the localization response, and in Figure 33 for the color-report response. Sirnilar to same-case letter matching, all field and display effects were largely eUrninated by increasing the number of items to nine. 2500 ^ 2 2 5 0 cu g 2000 8 8 1750 cn w >p «M500 § 1250 w § R O u 1000 -750 -— 500 h 250 0 • Within • Across Left Right Probe Position Figure 31. Efficiency scores for 9-item mixed-case letter matching in Experiment 5 when a detection response was used. Broken down by display type and probe position 45 2500 2250 -f 2 0 0 0 ; § I 1750 co ~ 1500 | H 1 2 5 0 u ^ 1000 S £ 750 £ 500 250 0 Left Right Probe Position Figure 32. Efficiency scores for 9-item mixed-case letter matching in Experiment 5 when a localization response was used. Broken down by display type and probe position 3000 ^2750 w | 2500 8 8 2 2 5 0 £ a 2000 1^1750 | * 1500 w g 1250 . 8 IOOO 750 500 • Within 1 1 Across Left Right Probe Position Figure 33. Efficiency scores for 9-item mixed-case letter rmtching in Experiment 5 when a color-report response was used. Broken down by display type and probe position The statistical analyses were largely consistent with these observations. Neither the detection nor the identification responses yielded any significant effects involving field or display (all ps > .1). The localization response elicited a significant right-field advantage [F (1, 28) = 6.041, p <.05, 46 MSe = 268070.740]. No other effects involving Display or Probe Position were significant (ps > .1). Reaction time and error scores produced a similar pattern of results, as shown in appendix II. Discussion This experiment produced three important findings. First, the 3-item displays produced similar left-field and across-display advantages for three very different modes of response (target detection, spatial localization, and color-report). Second, sometimes the across-display advantage in mixed-case letter matching was limited to the right-field. Third, increasing the display size to 9-items eliminated the typical left-field and across-display advantages. These findings provide mixed support for the computational complexity theory. In support of this theory, every response produced an across-display advantage (3-item displays) indicating that the across-display advantage is a robust effect that can be obtained irrespective of response demands. However, two findings were less supportive of the computational complexity theory. First, the interaction between display and field was replicated, indicating that the across-display advantage was sometimes limited to the right-field. This finding cannot be accommodated by the computational complexity theory in its current form because interhemispheric sharing should benefit both hemispheres, not just one. Second, elirninating the across-display advantage by increasing the number of letters poses another problem for the computational complexity theory. If anything, the across-display advantage should have been exaggerated in the 9-item displays because the benefit of interhemispheric sharing increases as the complexity of decision-making increases (Belger & Banich, 1992,1998). The asymmetrical orienting hypothesis also received mixed support. Although a left-field advantage was found together with an across-display advantage, replicating the correlation found in previous studies, this support was tempered by the particular form of the interaction between field and display type that was observed. Namely, a right hemisphere advantage should have resulted in matching times that were more similar in the two visual fields, not grossly longer in one. Given this pattern of results, it is clear that methodological factors were not responsible for failing to elicit a right-field advantage using these response manipulations. Two interpretations remain. The letter-matching task maybe so dependent on the right hemisphere that response manipulations cannot curb the right hemisphere's domination. Alternatively, letter matching may elicit a left-to-right bias in the order in which letters are mentally compared. The next experiment assessed if a dominant right hemisphere is responsible for the left-field advantage in letter matching by modifying the task so that it almost certainly requires the left hemisphere's input before the match can be made. Experiment 6: Does changing the material modify field and display effects? The main aim of this thesis is to determine if the correlation between field and display advantages that exists in the interhemispheric interactions literature may be dissociated by manipulating field advantage and observing its influence on display advantage. Experiments 4 and 5 were unsuccessful in achieving this end because a right-field advantage was not elicited by changing the nature of the response. 47 In this experiment, a time-proven method for eliciting a right-field advantage was used. Deciding if two words rhyme produces a right-field advantage (e.g., Banich & Karol, 1992; Rayman & Zaidel, 1991) because it has been proposed that only the left hemisphere is capable of phonological processing (e.g., Belger & Banich, 1998; Levy & Trevarthen, 1977). Functional imaging studies have supported this notion, as the left hemisphere is predominantly, if not uniquely, active during rhyming tasks (Lurito et al, 2000; Poldrack et al., 2001; Rumsey et al., 1997). Moreover, activity found in the right hemisphere (angular gyrus) does not correlate with activity found in other regions of the left hemisphere in normal readers, indicating that activity in the right hemisphere is not functionally connected to classical language centers in the left hemisphere (Horwitz, Rumsey, & Donohue, 1998). Therefore, modifying the letter-matching task to demand a rhyming judgment will ensure that the left hemisphere is consulted before a decision can be reached. The primary motivation for incorporating a rhyming decision in the letter-rmtching task is to see if a right-field advantage can be elicited with this task at all. However, note that if a right-field advantage is obtained in conjunction with a within-display advantage, this finding will not provide support for the asymmetrical orienting hypothesis. The computational complexity theory holds that interhemispheric sharing will only be evidenced when both hemispheres can independently make the decision. Belger & Banich (1998) provided evidence for this claim with a rhyming version of the letter-rmtching task The participants had to decide if one of the two probe letters shared the same sound as a three letter target word (e.g., TEE). Consistent with the computational complexity theory, no display advantage was obtained, presumably because only the left hemisphere can make rhyming decisions (Belger & Banich, 1998). Two findings must be taken into consideration when evaluating the Belger & Banich (1998) study. First, their rhyming task was more difficult than any other task that had been used before. Instead of just matching letters, the participants had to decode and assess the sound of a vertically oriented word (e.g., TEE) before determining if an appropriate match was present by phonologically decoding at least one of the two letters (e.g., B). The difficulty of this decision was evidenced in the high error rates and long reaction times that were obtained. Second, a right-field advantage was also not obtained (ps > .1). Considering that an identical data pattern was elicited with the 9-item displays (a task that both hemispheres could perform), the combined poor performance and absent field and display advantages may indicate that the null effects reported by Belger & Banich (1998) index difficulty and nothing else. To avoid sirnilar interpretational problems, a rhyming task was selected that required one less step. Three-letters were presented and the decision depended on a category judgment of the letters' sound. Two categories were given, letters that rhymed with "ee" (B, D, G P, T) and letters that did not (i.e., F, FL N, Q, R). Therefore, the identical display characteristics were used in this experiment as in the previous experiments. Any differences in performance would therefore have to be attributed to the mental processes engaged in by the participants. Methods Participants. Thirty-three right-handed university students (26 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit. 48 Stimuli, The stimulus displays were identical to Experiment 2, except that a present response now consisted of a category judgment. Two categories were used, letters that rhymed with "ee", (B, D, G, P, T) and letters that did not (F, H, N, Q, R). In half of the trials a match was present and in the rermining trials no match was present. The matching probe was never the same letter as the target. When a match did appear, either one or both of the probes could match the target. These redundant trials were included because pilot work revealed that the participants' could achieve 100% accuracy by simply deciding if the probes were from the same category or not (i.e., without consulting the target). Thus, adding the redundant trials forced the participants to consult the target so that an accurate response could be made. When such redundant trials were presented, the probes were from the same category, but were never the same letter. Procedure The participants' were told that that two categories of letters would be presented to them letters that sound like "ee" (i.e. B, D, G, P, T) and letters that do not (i.e. F, H, N, Q, R) and to determine if either probe matched the category of the target. The same letter was never repeated in one display. To maintain similarity with the other experiments, both upper and lower case letters served as targets. Accuracy was emphasized over speed. The inclusion of the redundant display required the number of trials to be increased. To do so, the number of observations per cell were decreased to about 10 (from 14) and the number of participants were increased to 31 (from 24), resulting in 660 experimental trials, equally divided into 6 blocks of 110 trials. In all other respects this experiment was identical to Experiment 2. Data handling. Data from two subjects were removed from the analysis because of a strong bias to respond "present". Only the trials with one matching probe were included in the analyses because display and field effects can only be assessed when only one of the probes matches the target. The "not ee" category condition proved to be a difficult concept for many subjects to understand, resulting in poor performance for this category (64%) compared to the rhyme with "ee" category (83%) overall. Moreover 64% is an inflated average because 18 of the 28 subjects averaged 53% for the not "ee" condition whereas the rermining 10 subjects averaged 85%. However, scoring poorly on the not "ee" condition did not mean that they were not trying because these same participants averaged 78% for the rhyme with "ee" condition. Because the problem was specific to one category of responses, only the rhyme with "ee" data will be described. The efficiency scores were analyzed with repeated measures ANOVA, including the variables Target Case (upper case versus lower case), Display (within-hemisphere versus between-hemisphere), and Probe Position (left-field versus right-field versus both visual fields). 49 Results The efficiency scores for the rhyme with "ee" decision are shown in Figure 34. The most important finding from this experiment is that performance was more efficient for the left-field and across-display. The statistical analyses confirmed these observations. Performance was superior in the across-display [F (1, 30) = 4.710, p <.05, MSe = 301886.078] and in the left-field (within-display comparison) [F (1, 30) = 3.959, p <.06, MSe = 262141.762]. Nothing else was significant in this analysis (all Fs <2, all ps > .1). The error scores produced a sirnilar pattern of results. However, the reaction time scores yielded a different pattern of results, in which an advantage was found for probes presented to the right-field rather than the left-field. Note that mean correct reaction time data become less reliable as the number of errors increase because there are fewer observations from which to draw. This is one reason to adopt efficiency scores, as they combine both error and reaction times in an evenhanded way (Akhtar & Enns, 1989). 2000 _ 1 9 0 0 ^ 1 8 0 0 g | 1700 ^1600 ft § V l 5 0 0 cu n •a * 1400 • i-l 4-> §j | 1300 £ 1 2 0 0 1100 1000 Left Right Probe Position Figure 34. Efficiency scores for the rhyme decision in Experiment 6 broken down by display type and probe position Discussion Rhyming decisions produce similar left-field and across-display advantages as do mixed-case letter rmtching. The implications of this finding are straightforward. Regarding the letter-rmtching task, finding a left-field advantage for rhyrning decisions indicates that a right hemisphere advantage is not responsible for the left-field advantage obtained with this task The grounds for this conclusion 50 are simple. Rhyming is strongly (e.g., Zaidel & Peters, 1981), if not uniquely (Levy & Trevarthen, 1977; Rayrnan & Zaidel, 1991) lateralized to the left hemisphere. Had the left-field advantage signified the right hemisphere's superiority over letter matching, a right-field advantage would have appeared when the task advantaged the left hemisphere. This was not the case5. Regarding the computational complexity theory, finding joint left-field and across-display advantages provides this theory with no support. For one, field advantages are supposed to reflect hemispheric differences (see e.g., Weissman & Banich, 2000) and these results make a hemispheric interpretation of the left-field advantage unlikely. More problematic was obtaining an across-display advantage for rhyming decisions. Rhyming should not elicit an across-display advantage because the right hemisphere cannot, in theory, contribute to such decision-making (Banich, 1998). Thus, obtaining an across-display advantage implies that (1) both hemispheres could perform the rhyming task or (2) the across-display is not a measure of interhemispheric interactions. These possible explanations will be examined in greater detail in Experiment 8. Regarding the asymmetrical orienting theory, finding joint left-field and across- display advantages also provides this theory with no support. Field advantage must index hemispheric advantage for the asymmetrical orienting hypothesis to have merit. Even if one assumes that the right hemisphere can make rhyming decisions (Zaidel & Peters, 1981), there is no way that its ability can surpass that of the left hemisphere. Therefore, the left-field advantage must be produced by something other than an advantage for the right hemisphere. Taking toll It is time to step back for a moment, to consider the original aim of this thesis, and to evaluate the implications of the data collected so far. A review of the literature on interhemispheric interactions had suggested that there was a tendency for an across-display advantage to be correlated with a left-field advantage and for a within-display advantage to occur in conjunction with a right-field advantage. Two interpretations of this relationship were considered. It may be a spurious relationship and under closer examination the display and the field effects maybe separable. This outcome would support the computational complexity theory because it predicts that hemispheric advantage has no bearing on the hemispheres' ability to process information in parallel (e.g., Banich, 1998; Banich & Nicholas, 1997). Alternatively, if these effects are related, then they might be interpreted as favoring the asymmetric orienting hypothesis that was described in the introduction. So far, several experiments have been conducted to determine if the display and the field effects are separable or related. After first replicating the typical across-display and left-field advantage for mixed-case letter rnatching, I tried to alter the field advantages by using response demands that favor left hemisphere processing (Experiments 4 and 5) and by using a rhyming task that demands left hemisphere processing (Experiment 6). For standard 3-item displays, neither manipulation was successful in producing a right-field advantage. For larger 9-item displays, both the field effects and the display effects vanished. 5 Belger and Banich reported a right-field advantage for rhyming when a similar, but more difficult, task was used. A right-field advantage was not obtained (p >.l). 51 These results indicate that neither theory is well supported. Consider the findings that pose problems for the computational complexity theory. 1. A within-display advantage was not obtained in same-case letter rmtching. Instead, no display effects were found in Experiment 1 and 2 and an across-display advantage was found in Experiments 3 - 5, when black and white letters were used. Although it could be argued that mixing letter color increased the complexity of the cognitive processes involved, tipping it in favor of hemispheric sharing; this does not explain the failure to replicate the within-display advantage in Experiments 1 and 2. It should be reiterated that this is not the first time that a within-display advantage has not been found for a same-case letter rmtching task (see also Banich et al, 2000; Belger & Banich, 1992; Brown & Jeeves, 1993). 2. The letter- rmtching task may not be solved in parallel in mixed-case letter matching. The findings from Experiment 3 suggest that an attentional interpretation may be a more appropriate description of the letter-rmtching task. Same-case letter rmtching maybe processed in parallel because it places mirrirml demands on attention. Mixed-case letter matching may require serial processing because focal attention must be shifted multiple times. Considering that letter rmtching must assess parallel processing for an interhemispheric sharing interpretation to hold, the possibility that the across-display advantage arises from serial processing is very much at odds with the computational complexity theory. 3. An interaction was often seen between the field and display effects, such that the across-display advantage was larger in the right-field. This finding cannot be accommodated by the computational complexity theory because interhemispheric sharing should benefit both hemispheres equally. 4. Increasing the number of items from 3 to 9 elirriinated all display and field effects. This could not be attributed to performance being at floor because accuracy was still well above chance levels (—70%). If anything, the computational complexity theory predicts that this manipulation should have increased the magnitude of the across-display advantage (Belger & Banich, 1992; 1998). The elimination of the across-display advantage with larger display sizes has no explanation within this theory. 5. The across-display advantage was found even when the participants' decided if a probe letter name rhymed with the target letter name. This result contradicts the prediction that interhemispheric sharing will only be evidenced when each hemisphere can make the decision independently of the other (Banich, 1998; Belger 6c Banich, 1998). This is not the case for rhyming decisions, as it is widely accepted that only the left hemisphere can make phonological judgments (e.g., Belger & Banich, 1998; Levy & Trevarthen, 1977). Consider the problems for the asymmetrical orienting hypothesis. 1. The inability to alter the left-field advantage with response manipulations (Experiments 4 & 5) or with rhyming decisions (Experiment 6) suggests that the left-field advantage does not index a right hemisphere advantage. If this is true, then the asymmetrical orienting hypothesis looses credibility as an account of the letter-rmtching task. Field advantages must index hemispheric advantages for this theory to stand. 2. The frequently obtained disadvantage for right-field, within-display trials leaves the asymmetrical orienting hypothesis in poor stead. The basic premise of this theory is that attention 52 will be more evenly distributed across fields whenever the right hemisphere is advantaged. Instead, when the "right hemisphere" is advantaged (indicated by a left-field advantage) performance has, on occasion, been very poor for the condition in which all relevant information was presented to the left hemisphere directly. If anything, this finding is more consistent with the notion that attention was biased to only the left-field, contrary to the basic tenets of the asymmetrical orienting hypothesis. Experiment 7: Does a right-to-left instructional bias change left-field advantage? In the introduction of Experiment 5, four potential causes of the left-field advantage were provided. Now that methodological problems and a strong right hemisphere advantage have been largely ruled out (Experiments 5 & 6, respectively), only one viable explanation for the left-field advantage remains. It is time to consider whether the left-field advantage arises from a left-to-right bias in the order that letters are mentally compared in letter rmtching. Even though reading order biases are not often considered when interpreting field effects, such biases nonetheless elicit robust visual field advantages. A recent study provides strong evidence that the left-field advantage in letter naming arises from just such a reading bias (Lubow, Tsal, Mirkin, &Mazliah, 1994). In this study, 9 letters were presented on an imaginary circle. Participants were bilingual English- Hebrew readers, who were asked to identify as many letters in the display as possible. Two different kinds of displays were presented, one with English letters and the other with Hebrew letters. The rationale of this study was simple. If field advantage depended on hemispheric differences, then the same field advantage would be obtained in both types of displays. However, if field advantages were elicited by reading practices, then a left-field advantage would be obtained for English letters and a right-field advantage would be obtained for Hebrew letters. Their findings were completely consistent with a reading bias interpretation. Many other studies have reached similar conclusions: The left-field advantage for letter rmtching (or naming) is caused by a tendency for English readers to compare letters in a left-to-right order (e.g., Bradshaw, Nettleton, Taylor, 1981; Bryden, 1960; Butler, 1978, 1979, 1981; Heron, 1957; Krueger, 1976; Levine & Banich, 1982; Scheerer, 1972, 1973; Tramer, Butler, &Mewhort, 1985; White, 1976). The 3-item (and similar) letter-matching task is thought to be immune to such biases because the target and probes are arranged on a diagonal axis. Had the letters been distributed along a horizontal axis, bias maybe a factor (e.g., Banich, 1998; Banich & Belger, 1990; Banich & Shenker, 1994). However, this argument is based upon theoretical assumption rather than empirical evidence because presenting letters in a circular array also elicits a left-field advantage (Lubow et al., 1994). These results, along with those of the present study, strongly suggest that it maybe time to reconsider the left-field advantage as signaling a left-to-right bias in the order that letters are mentally compared. One way to test if hemispheric effects or order of comparison biases are responsible for the field and display effects is to ask the subjects to compare the letters in a specific direction. Previous research has reported that biasing attention to one field or to another does not modify hemispheric advantage. For instance, lexical decision tasks consistently elicit a right-field advantage. Hardyck 53 and colleagues (1985) examined if this right-field advantage could be overturned by directing the participants' attention to right-field or to the left-field, respectively. They used several techniques to bias attention including an arrowhead at fixation and asking the participants to pay attention to one field over the other. None of the approaches they used affected the right-field advantage. This experiment adopts a similar strategy to assess the hemispheric nature of the left-field advantage in letter matcliing. After consulting the target, the participants were instructed to compare the probe letters in a left-to-right order or in a right-to-left order. Any hemispheric account would predict that the left field advantage would continue to exist over and above any biasing effects (Hardyck, et al., 1985). Alternatively, if the left-field advantage arises from a left-to-right bias, then requesting subjects to examine the right probe first should significantly alter the pattern of findings. Furthermore, explicitly requesting them to examine the left probe first should produce a pattern of data that niimics the results obtained previously without these instructions. Methods Participants. Twenty-four right-handed university students (13 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit or for $5. Procedure. Every participant performed a blocked version of mixed-case letter rmtching (Experiment 1) two times. Half of the participants began with the "Left First" instructions, in which they were told to maintain their gaze upon central fixation and to decide if the probe in the left field matches the target before deciding if the probe in the right field matches the target. After receiving a block of 56 practice trials and 224 experimental trials, the same participants were then given the "Right First" instructions, which were identical to the "Left First" instructions except they were asked to decide if the probe in the right-field matches the target before deciding if the probe appearing in the left-field matches the target. For the rermining subjects, the order of the instructions was reversed. All other procedural details were identical as Experiment 1. Data handling. Less than 0.001% of trials were eUrninated because the reaction time exceeded 2000 ms. The efficiency scores were analyzed with repeated measures ANOVA, including the variables Instruction ("left first" versus "right first"), Display (within-display versus across-display), and Probe Position (left field versus right field). Only those effects involving the Instruction variable will be discussed. Results Data for the different instructions are shown in Figure 35. Two important findings are immediately apparent. First, instruction successfully modified field advantage: "Left first" instructions produced a left-field advantage and "right first" instructions produced a right-field advantage. Second, these instructions did not affect the across-display advantage, as an across-display advantage was obtained in both cases. 54 The statistical analyses confirmed these observations. The Instruction X Probe Position interaction was significant [F (1, 23) = 15.580, p <.05; MSe = 32161.812]. Each instruction was analyzed separately. For the "left first" instructions, performance was more efficient in the left-field [F (1, 23) 12.956, p <.05; MSe = 58432.267] and in the across-display [F (1, 23) = 29.480, p < .05; MSe = 12009.349]. For the "right first" instructions, performance was more efficient in the across-display [F (1, 23) = 21.006, p <.05; MSe = 15660.508] and the Display x Probe Position interaction [F (1, 23) = 6.572, p <.05; MSe = 6026.056] indicated that the right field was advantaged for the within-display [F (1,23) = 8.635, p <.05], but that no field advantage appeared for the across display (F < 1). Reaction time and error scores produced a sirnilar pattern of results, as shown in appendix II. Finally, the left-first instructions were compared with mixed-case letter rmtching of Experiment 1 to confirm the observation that the left-first instructions yielded an identical pattern of performance as was spontaneously obtained. This observation was confirmed, as there were no significant statistical differences between these experiments (all ps >.05). 1400 1^300 u S I 1 2 0 0 o o >, ftllOO •8 S IOOO h •r-t 4-> m g 900 h o " 800 700 • Within E3 Across Left Right Left Right "Left First' "Right First' Figure 35. Efficiency scores for the "left first" and the "right first" instructions in Experiment 7 broken down by display and probe position. Note that left and right refers to the field of the rmtching probe. Discussion These results showed that field advantages were modified with instructions. A left-field advantage was obtained in the "left first" instructions that was similar, if not indistinguishable, from that normally obtained. However, a right-field advantage was obtained in the "right first" instructions. Importantly, the across-display was not altered with these instructions, indicating that field and display effects are indeed separable. These findings strongly imply (1) that the left-field advantage 55 that has been obtained throughout the course of this study and others (e.g., Banich & Belger, 1990; Banich et al, 2000; Weissman & Banich, 2000) is strongly influenced by the participants' spontaneous tendency to compare letters in a left-to-right order (e.g., Bryden, 1960; Heron, 1957; Kimura, 1966; Lubow et al., 1994) and (2) that mixed-case letter matching requires serial processing because participants can control the order in which letters are mentally compared without altering the time required to make these responses. The implications of these findings are clear. Dissociating the field and display effects demonstrates that the asymmetrical orienting hypothesis is not a viable account of the letter-matching task. Therefore, the combined left-field and across-display advantages are not the attentional consequences of right hemisphere priming. However, dissociating the field and display effects by merely asking the participants to compare the items in a specific direction also implies that the computational complexity theory is not a viable explanation of the letter-matching task. Such instructions demand that participants adopt serial strategies in letter matching. Since these instructions successfully mimicked (left first) and altered (right first) the data pattern implies that mixed-case letter matching may spontaneously elicit similar serial processes. If this synopsis is accurate, the computational complexity theory cannot provide an interpretation of the data patterns elicited with mixed-case letter rmtching. Recall that letter matching must be solved in parallel according to the computational complexity theory. In the next experiment, the across-display advantage was subjected to a similar instructional manipulation. If the across-display advantage indexes the benefits of parallel processing, then asking the participants to compare the items in a certain order will not change the across-display advantage for letter rmtching. The goal of the next, and final, experiment was to establish the accuracy of this prediction. Before moving on, the prospect that systematic eye movements were responsible for the findings described in this experiment should be addressed. This is not a viable scenario because the display was presented for only 210 ms. This means that it would have been erased by the time that most saccades would have been initiated (Munoz et al., 1998). The serial nature of letter rmtching must arise from internal mental processes, not from physical movements of the eye or limited resolution of the fovea. One possibility, however, is that the participants may have spontaneously adopted their own strategies: Perhaps, they fixated on one of the probe's positions at the beginning of each trial rather than at central fixation. Consider however, that changing fixation would have changed the pattern of results. For instance, had the participants fixated upon the left probe's position in the "left-first" instructions, a more exaggerated left-field advantage and a smaller across-display advantage would be expected. The advantage for the left probe would have been exaggerated because probes positioned in the right field would have appeared 11 degrees from the fixated position, making it much more difficult to resolve the identity of the right probe. The across-display advantage would have been smaller because the target would have appeared 5 degrees below and approximately 8 degrees to the right, making it much more difficult to resolve the identity of the target. Instead, the "left first" instructions resulted in a data pattern that was identical to the pattern obtained throughout the course of this investigation and in other published studies (e.g., Banich & Belger, 1990; Weissman & Banich, 2000), indicating that such alternative interpretations are unlikely. 56 Experiment 8: Does a within-to-across bias change the across-display advantage? In the previous experiment, instructions changed the left-field advantage into a right-field advantage. This finding had two consequences. First, it suggests that the left-field advantage in letter matching takes place because participants spontaneously compare letters in left-to-right order. Second, it suggests that the letters are compared in succession because adopting serial strategies mimics the pattern of data found when participants are given no special instructions. The suggestion that mixed-case letter rmtching may require serial processing has grave repercussions for the computational complexity theory. If mixed-case letter matching demands serial processing, then the across-display advantage cannot arise from the benefits of parallel processing. Instead, a mechanism that relies on the serial inspection of letters must be searched for instead. One possible alternative is that a bias in order of processing is responsible for the across display advantage. Several groups of findings make this possibility an important one to consider. First, in rapid saccade tasks, shifts of gaze are faster in the direction opposite to a first saccade. The task is simple: Participants are required to make rapid saccadic eye movements to targets that suddenly appear in the periphery. An "alternation" effect is found, in which a shift of gaze is faster in one direction (e.g., to the right) if it was preceded by a saccade that was made in the opposite direction (e.g., to the left) (Carpenter, 2001; Au et al., in submission). Second, shifts of attention to a target are faster in the direction opposite to a cue. The task is simple: a bright flash (the cue) appears in the periphery and draws attention to one position. After a delay and a reorienting event (another bright flash at fixation), a peripheral target appears. The participants' task is to detect the onset of the target as quickly and as accurately as possible. When a long delay (> 300 ms) is given between the cue and the target, participants are slower to respond if the target appears to the same position as the cue. This slowed responding to a previously attended position has been termed inhibition of return (for a review see Klein, 2000; Posner & Cohen, 1984). In addition, however, the participants respond more efficiently when the target appears in the opposite position as the cue. This facilitation of targets in the opposite position has been termed the "attentional momentum" effect (Pratt et al., 1999). Comparing the descriptions of the rapid eye movement task and the inhibition of return task with the letter-rmtching task used in this study, it may appear that these tasks have little in common. Although many differences do exist, each task assesses the same basic phenomenon— multiple shifts of attention. That is, the eye movement task requires two shifts of gaze (to suddenly appearing targets), the inhibition of return task requires at least two shifts of attention (one to a cue and at least one other to a target), and the letter-matching task requires at least two shifts of attention (one to the target-letter and at least one other to a probe-letter). Although the mechanisms producing this opposite position advantage are currently unknown, it does not depend on stimulating opposite hemispheres because a sirnilar opposition advantage is found above and below fixation (Pratt et al, 1999). It is worth considering whether the across-display advantage can be interpreted in a non-hemispheric manner, namely, as an advantage for exarnining the opposite field position after each 57 attentional shift. Unlike the computational complexity theory, which predicts that parallel processing is responsible for the across-display advantage, this interpretation implies that the need for multiple sequential shifts of attention, in conjunction with biases to inspect items in the opposite visual field, are responsible for the across-display advantage. Evidence that serial letter comparisons can alter the across-display advantage is still absent. This experiment was designed to directly assess an interhemispheric sharing interpretation of the across-display advantage. The logic of this experiment was identical to that of the previous experiment. Participants were instructed to compare the probe letters in a specific order. After consulting the target, the "across first" instructions requested that the participants consult the probe letter appearing in the opposite field as the target first. In the "within first" instructions, participants were requested to consult the probe letter appearing in the same field as the target first. Methods Participants. Twenty-five right-handed university students (20 female) with normal or corrected-to-normal visual acuity participated in this experiment for extra course credit or for $5. Procedure. This experiment was identical to Experiment 7 except for the instructions given to participants. The "Across First" instructions asked the participants, while mamtaining gaze on fixation, to consult the probe appearing in the opposite field as the target before consulting the probe appearing in the same field as the target. In contrast, the "Within First" instructions were identical except the participants were asked to consult the probe appearing in the same field as the target before consulting the probe appearing in the opposite field as the target. Data handling. Data from one participant is not included in this analysis because a computer malfunction caused her reaction time data to not be recorded. Less than 0.001% of the remaining trials were elirninated because the reaction time exceeded 2000 ms. The efficiency scores were analyzed with repeated measures ANOVA, including the variables Instruction ("across first" versus "within first"), Display (within-display versus across-display), and Probe Position (left field versus right field). Only those effects involving the Instruction variable will be discussed. Results Data for the different instructions are shown in Figure 36. Two important findings are immediately apparent. First, instructions successfully altered the display advantage: "Across first" instructions elicited an across-display advantage and "within first" instructions elicited no display advantage. Second, these instructions did not affect field advantage because a left-field advantage was obtained in both cases. 58 The statistical analyses confirmed these observations. The Instruction X Configuration Position interaction was significant [F (1, 23) = 12.390, p <.05; MSe = 36252.544] so each Instruction was analyzed separately. For the "across first" instructions, performance was more efficient in the across-display [F (1, 23) = 40.008, p <.05; MSe = 22395.346] and in left-field [F (1,23) = 8.220, p <.05; MSe = 14277.773]. For the "within first" instructions, no display advantage occurred [F (1, 23) < 1; MSe = 24955.972] and performance was better in the left-field [F (1, 23) = 8.445, p <.05; MSe = 7774.480]. The Display x Probe Position interaction was also significant for the "within first" instructions [F (1, 23) = 5.071, p <.05; MSe = 7174.879]. Breaking down this interaction revealed a trend for a within-display advantage in the left-field [F (1, 23) = 2.569, p <.13] and a trend for an across-display advantage in the right-field [F (1, 23) = 2.502, p <.13]. Reaction time and error scores produced a similar pattern of results, as shown in appendix II. Finally, the across-first instructions were compared with mixed-case letter matching of Experiment 1 to confirm the observation that the across-first instructions yielded an identical pattern of performance as was spontaneously obtained. This observation was confirmed, as there were no significant statistical differences between these experiments (all ps >.05). 1400 ^1300 h 4-> a 11200 o o & Xnoo 5 b I O O O h £ g 900 £ 800 h 700 • Within 1 1 Across Left Right Left Right "Across First' •Within First' Figure 36. Efficiency scores for the "across first" and the "within first" instructions in Experiment 8 broken down by display and probe position. Note that left and right refers to the position of the rmtching probe. Discussion The display advantage was modified by instructions, in that an across-display advantage was obtained in the "across first" instructions and no display advantage was obtained in the "within first" instructions. 59 The implications of these results are clear. The computational complexity theory cannot explain the across-display advantage in mixed-case letter rmtching because it was eliminated with instructions. Had the across-display advantage indexed the benefits of parallel processing, it would not have been altered with serial letter comparisons. The appearance of similar display advantages in other tasks that demand similar multiple shifts of attention (Carpenter, 2001; Au et al, in submission; Pratt et al, 1999) implies that the across-display advantage may index an important property of how the brain orients attention. Unlike the left-field advantage in Experiment 7, the across-display advantage was not reversed with instructions. Does this differences imply the left-field advantage is more changeable and the across-display is less changeable with instructions? It may. Consider one potential difference between the left-field and across-display advantages. The most straightforward interpretation of the left-field advantage in letter rmtching is that it indexes the tendency of English reading participants to spontaneously compare English letters in a left-to-right order. Such habitual biases may be interpreted as the emergence of a learned, top-down influence on spatial attention. In contrast, the across-display advantage cannot be easily interpreted as a top-down influence because a similar habitual explanation does not exist for an across-to-within serial order comparison. Instead, the across-display advantage may arise from a bottom-up influence on spatial attention arising from how the brain selects the next target of attention (an idea that will be further clarified in the General Discussion). If this distinction between the left-field and across-display advantages is accurate, then the left-field advantage may have been reversed with instructions because top-down influences can be modified more easily, whereas the across-display advantage may have only been eliminated with instructions because bottom-up influences can be modified less easily. General Discussion In the interhemispheric interactions literature, an across-display advantage tends to occur in conjunction with a left-field advantage and a within-display advantage tends to be associated with a right-field advantage for complex cognitive tasks. This thesis was designed to establish if these field and display effects were actually related, as they appeared to be, or whether they were in fact separable when this relationship was examined in greater detail. Arguments were developed to show that each of these outcomes would provide support for a different theoretical framework. Specifically, closely related field and display effects were seen as providing support for the asymmetrical orienting hypothesis. This theory proposes that both field and display effects provide separate signatures of the hemisphere whose attentional system has been primed. The tendency for a right-field advantage to be associated with a within-display advantage is consistent with the ability of the left hemisphere to orient only to the right side of space. The tendency for a weaker left-field advantage to occur in conjunction with an across-display advantage is consistent with the ability of the right hemisphere to orient both to the left and to the right sides of space. On the other hand, it was argued that the finding of separable field and display effects would support the computational complexity theory. This theory proposes that the complexity of the cognitive task determines if parallel processing between the hemispheres is advantageous or not. A within-display advantage illustrates that the hemispheres are better off working independently because the cost of integrating the activities of the two hemispheres is too great. An across-display advantage illustrates that the hemispheres are better off working together because the benefits of 60 parallel processing offset the costs of integration. Hemispheric advantage is not a consideration in this view because it is not believed to affect the manner in which the hemispheres interact (e.g. Banich & Nichols, 1997). The surprising finding of this study was that although field and display effects were shown to be separable in Experiments 7 and 8, the results could still not be interpreted as favorable to the computational complexity theory. Among the evidence most difficult to accommodate to this theory was (1) a failure to replicate the within-display advantage for same-case letter rmtching (Experiment 1); (2) the failure of the signature data pattern predicted for mixed case letter matching to generalize to displays with nine letters (Experiment 5); and (3) the finding of an across-display advantage even when the cognitive task, a rhyming decision, demanded the participation of the left hemisphere (Experiment 6). However, by far the most difficult evidence to accommodate to this theory was the finding that the across-display advantage could be either rnimicked or eliminated by asking observers to compare probe letters with the target letter in a specific order (Experiment 8). In the remainder of this section, the implications of these findings will be discussed for each of the two theories initially under consideration, possible explanations for the left-field and the across-display advantages in letter rmtching will be discussed, and future directions for research will be described. Computational complexity theory and interhemispheric interactions Computational complexity theory The findings of this study did not support the computational complexity theory. First, although every attempt was made to replicate the procedures of previous studies precisely (Experiment 1), same-case letter rmtching did not elicit a within-display advantage. A careful review of previous reports indicated that such a failure to find a display advantage for this condition is not uncommon (e.g., Banich, Passarrotti & Janes, 2000; Banich et al, 2000; Belger & Banich, 1992; Brown & Jeeves, 1993; Zhang &Feng, 2000). I can only conclude that the signature finding taken as evidence for interhemispheric independence is not easily replicated. Second, even though an across-display advantage was consistently obtained in mixed-case letter rmtching, it did not occur for the reasons proposed by computational complexity theory. Most telling in this regard was Experiment 8, where it was shown that the across-display effect could be modified with instructions. When participants were instructed to first compare the probe opposite the target ("across first" instructions), a typical across-display advantage was obtained. No additional cognitive effort or processing time was associated with the following of these instructions. Furthermore, when instructed to first compare the probe appearing to the same side as the target ("within first" instructions), the across-display advantage was eliminated. Such instructions should have not modified the across-display advantage had it indexed the benefits of parallel processing (Weissman & Banich, 2000). The finding that instructions to use a serial leiter-rmtching strategy results in a data pattern that mimics the across-display advantage obtained when no explicit instructions are given implies that mixed-case letter rmtching is solved using serial processes. If this were not the case, then the serial leiter-rmtching strategy should have resulted in longer response times and more errors. The confirmation that participants were actually following instructions to compare letters in a particular order was given in the "within first" condition. Flere following the instructions resulted in a very different data pattern, one that would be expected if participants were successfully overcoming the 61 strategies employed spontaneously in this task. In the next section, I will provide an explanation for the across-display advantage that does not depend on the notions of computational complexity and hemispheric interactions. However, even before the instructional experiments were conducted in Experiment 7 and 8, aspects of the data in previous experiments had indicated that the across-display advantage was not reflective of interhemispheric sharing. For one, the across-display advantage was sometimes larger for the right-field than for the left-field (as in Experiments 4 & 5). Interhemispheric sharing should benefit both hemispheres equally, not the left hemisphere more. Also, the across-display advantage was not observed in the very conditions that should have optimized this effect: letter matching in 9-item displays (Experiment 5). As the complexity of the decision increases, so too should the advantage of parallel processing (Belger & Banich, 1992; 1998). Yet it was under these conditions that the across-display advantage largely disappeared, a finding that could not be attributed to floor effects on performance because participants were still performing well above chance in accuracy (—70%). Third, an across-display advantage was obtained when it should not have been. Rhyming requires phonological analyses that are thought to depend exclusively upon the left hemisphere (e.g., Banich &Karol, 1992; Rayman & Zaidel, 1991). An across-display advantage should occur only when each hemisphere can independently make the decision (e.g., Banich, 1998; Belger & Banich, 1998). Yet an across-display advantage was obtained in a rhyming version of letter matching (Experiment 6) that was indistinguishable from other versions of mixed-case letter matching. Moreover, other evidence that has been cited in favor of the computational complexity theory in the past is of questionable merit. For instance, evidence from neuropsychological populations with damage to the corpus callosum and associated attentional deficits has been taken as support (Banich, 1998). The problem with this line of reasoning is that every clinical population that has been forwarded as supporting evidence of this kind has brain damage to other brain structures as well. If attentional functions are carried out by a widely distributed network of brain regions, as many believe (e.g., Mesulam, 1999; Posner & Dehaene, 1994, Posner & Petersen, 1991), then damage to any part of the network may result in attentional disorders. For instance, multiple sclerosis (MS) is proposed to be accompanied by attentional deficits because the corpus callosum is damaged during the course of this disease (e.g., Banich, 1998). In support of this notion, one study has reported that as the corpus callosum becomes more atrophied, the patients' attentional deficits become more exaggerated (e.g., Rao, 1995). However, MS does not only damage the corpus callosum, it destroys white matter throughout the brain. As a result, greater demyelination of the corpus callosum could well be associated with greater attentional deficits. However, it must be understood that the same evidence would also indicate that other myelinated tracts directly involved in attentional functioning had received progressive damage. It is unreasonable to expect the corpus callosum to be specifically responsible for the attentional deficit observed in MS patients when other brain regions are also affected. Following a similar line of questionable reasoning, studies of children with phenylketonuria (PKU) have been offered as support for the computational complexity theory (Banich et al., 2000). It is true that the disorder known as PKU is accompanied by decreased size of the corpus callosum. Children with PKU also have a reduced across-display advantage in letter rmtching compared to normal children (Banich et al, 2000). However, the conclusion that this implicates the corpus callosum in the control of the letter-rmtching task is not required by the data. Instead the children 62 with PKUin this study demonstrated a reduced across-display advantage because they performed better overall than did the normal children. Unlike the normal children, the children with PKU evidenced no disadvantage for the right-field, within-display trials. This is not convincing support for the claim that damage to the corpus callosum decreases the parallel processing capacity of the brain. Other evidence cited as support is equally problematic. Fmdings from an fMRI study showing unilateral brain activity in a computationally simple task and bilateral brain activity in a computationally complex task (Klingberg, O'Sullivan, & Roland, 1997) has been cited as evidence in support of computational complexity theory (Weissman & Banich, 2000). However, other studies have reported bilateral activation in other simple tasks (e.g., Just et al., 1996). In summary, there is very little evidence supporting the claim that computationally simple tasks are performed better by one hemisphere working at a time and that complex tasks are performed better when the hemispheres work in parallel. First, the present study was unable to confirm a within-display advantage for a computationally simple task. Second, the results of the present study show that the across-display advantage appears to be reflective of the serial order with which letters are compared rather than of interhemispheric sharing. Third, evidence cited previously as supporting this theory taken from neuropsychological populations and from functional imaging studies are open to alternative explanations. Interhemispheric interactions In recent years, the computational complexity theory has provided the main theoretical framework for the study of interhemispheric interactions. Accordingly, the findings from the present study, which suggest that the across-display advantage should not be taken as evidence for interhemispheric sharing, seriously undermine this theoretical framework. The findings from this study raise serious questions about the use of the 3-item leiter-rmtching task as a method of probing interhemispheric interactions. However, many other studies have used 2- item paradigms, which can elicit similar across-display advantages (e.g., Brown & Jeeves, 1993; Diamond & Beaumont, 1971; Serano &Kossyln, 1991). Like the computational complexity account, the typical interpretation of the across-display advantage is that dividing information between the hemispheres confers an advantage because the hemispheres can process the information in parallel (e.g., Diamond & Beaumont, 1971). But the across-display advantages reported with these 2-item displays are also subject to alternative interpretations. In particular, several methodological artifacts maybe responsible for the across-display advantage obtained with the 2-item displays. One possible confound is that the stimuli are often presented in a horizontal configuration in the across-display (right panel in Figure 36) and in a vertical configuration in the within-display (left panel in Figure 36). This difference raises the question as to whether the across-display advantage indexes parallel processing between the hemispheres or an advantage for comparing horizontally versus vertically oriented items (e.g., Banich StShenker, 1994; Coney, 1985; Ludwig et al., 1993; Schmitz-Gielsdorf, et al., 1988). Even if a diagonal orientation is used for the across-display instead, as shown in the middle panel in Figure 37, the relative distance between the items is different in the across- and the within-displays. Confounded distance between items is a problem because the size of the across-display advantage has been shown to increase along with the distance between items (Coney, 1985; Ludwig et al., 1993; Miller, 1981). Furthermore, studies that have equated the distance between the items 63 have produced mixed results. Some have found that the across-display advantage disappears (e.g., Schmitz-Gielsdorf, et al, 1988); others have reported that an across-display advantage remains, albeit reduced in magnitude (e.g., Coney, 1985); and others have reported that the across-display advantage still holds (e.g., Liberman, 1986). Vertical Within Diagonal Across Horizontal Across Figure 37. Examples of 2-item displays. However, even if an across-display advantage can still be found after these other factors have been ruled out, one alternative explanation remains. The planning of horizontal, diagonal, and vertical shifts of attention have been shown not to be the same. Shifts of attention made along a horizontal or a diagonal plane are programmed unilaterally meaning that only one hemisphere programs and initiates the shift. In contrast, shifts of attention made along a vertical plane are programmed bilaterally meaning that both hemispheres must work together to initiate the shift (e.g., Goldberg, Eggers, & Gouras, 1992). This difference has been supported by electrical stimulation studies in the frontal eye fields. Applying mild electrical pulses to this fore brain region causes eye movements to be produced. Unilateral stimulation elicits either horizontal or diagonal eye movements to the opposite visual field. Vertical eye movements require bilateral stimulation to be evoked (e.g., Bruce et al, 1985). Quite ironically then, when the hemispheres must work together to compare items, as in the vertical arrangement of letters, performance is hindered compared to when one hemisphere can make the attention shifts on its own, as in the horizontal and diagonally arranged letters. But note this explanation assumes that shifts of attention between items are required to perform the letter-matching task Two lines of evidence suggest that even these 2-item displays require serial shifts of attention. First, if the hemispheres were processing information independently and in parallel, then performance should be the same if one item was presented to one hemisphere or if two items were presented, with one item to each hemisphere. Comparisons of this kind indicate that performance is more efficient for displays containing only one item, suggesting that the hemispheres do not process information independently and in parallel (e.g., Miller, 1981). Second, consider the findings from a recent ERP investigation (Woodman & Luck, 1999). In this study, two items were presented to participants who had to decide if one of the objects was oriented in a certain direction. The participants were biased to compare the items in a certain order 64 because one was more likely to be the target. The study was designed to establish if visual search was better described by parallel versus serial models of processing. The primary measure of serial processing was an ERP correlate of focal attention, the N2pc wave. The authors reasoned that if the items were processed in parallel, the N2pc wave would not alternate across the hemispheres. However, if the items were processed serially, the N2pc wave would alternate across the hemispheres, corresponding to the serial inspection of the items. The results unambiguously supported a serial model of visual search. A secondary aspect of the data from this study is even more directly related to the present studies. Similar to other studies involving 2-item displays, an across-display advantage was obtained when both items were examined (i.e., when the second item was the target) (Woodman, personal communication, August 24th, 2001). Finding an association between an across-display advantage and a neural correlate of serial processing gives further plausibility to one of the main findings of the present study, namely, that a data pattern previously thought to be an index of parallel processing can be either mimicked or altered simply by instructions to participants to engage in letter comparisons in a particular order. In summary, although the findings from this study do not directly address the across-display advantage that can be found with 2-item displays, there is good reason to believe that the across-display advantage does not index interhemispheric sharing. First, the across-display advantage may arise from an advantage for horizontal comparisons (Banich & Shenker, 1994) or from an advantage for items separated by a larger distance (Coney, 1985; Schmitz-Gielsdorf, et al., 1988). Second, it may arise from a disadvantage for shifts of attention in a vertical direction (Goldman et al, 1992). However, for this alternative explanation to hold, the 2-item displays must be solved with serial processing, a notion that has its own base of support in the data (Miller, 1981; Woodman & Luck, 1999). Future directions for the computational complexity theory and interhemispheric interactions The results of this study suggest that the across-display advantage in letter matching does not index the advantages of parallel processing between the hemispheres. If this is the case, it is important to consider the question of how future research in this area should proceed. How indeed should the important question of how the hemispheres interact be studied in the future? If limitations on spatial shifts of attention form the bottleneck of the across-display advantage, then similar "across" advantages might be expected whenever "across" trials are compared to "within" trials regardless of which sensory modality is examined. Consider for a moment that other sensory modalities map space in topographically organized structures that share a close connection with the structures that map visual space (e.g., Andersen et al., 1998). If, for instance, an across advantage is found in the auditory modality, this advantage may demonstrate that that spatial attention shares similar characteristics across sensory modalities (e.g., Driver & Spence, 1998), rather than demonstrating an advantage of parallel processing. Being unable to interpret the across advantage within the framework of interhemispheric interactions poses a serious problem for the computational complexity theory because it will be difficult, if not impossible, to obtain empirical support for this theory. This situation does not, however, pose a problem for the field of interhemispheric interactions because there are many other ways that interhemispheric interactions could be investigated. In fact, so few alternative methodologies have been employed to explore interhemispheric interactions that any technique will be a novel contribution. 6 5 In my view, an interesting and important direction that research should take is to examine the neurophysiological basis of interhemispheric interactions in a simpler system. The motor system would be an ideal candidate because the interactions between homologous regions in each hemisphere could be assessed by combining behavior with neurophysiology. For instance, it is well documented that each hand can perform independent actions if the actions are relatively simple, but that the actions of each hand become destabilized as the task becomes more demanding (e.g., Carson et al., 1997). The behavioral shift between independent and synchronized activity should be associated with a distinct neurophysiological correlate involving independent activity and then synchronized activity between the hemispheres. Detailed examination of the border region between independent and synchronous activity may provide important insights into the physiological mechanisms governing interhemispheric interactions in this system that may extend to other systems as well. Although this is little more than a thought experiment at this time, it is an example of a possible avenue for further research that would be less plagued by the problems identified in this thesis. Asymmetrical orienting hypothesis and hemispheric differences in orienting attention The findings from this study did not support the asymmetrical orienting hypothesis. Although this theory proposed that the left-field and the across-display advantages in letter matching were indicative of the same underlying mechanisms, the results showed that these effects were clearly separable (Experiments 7 & 8). However, this evidence should not be taken to imply that there are no differences between the hemispheres in the spatial orienting attention (e.g., Heilman & Van Den Abell, 1980; Mesulam, 1981,1999). Instead, the lesson should be that visual field and display effects in letter matching are not necessarily an index of this difference. Rather than trying to establish if other visual field effects can be reinterpreted as signatures of the hemisphere whose attentional system has been primed, more direct ways of assessing hemispheric differences in orienting attention should be developed. In doing so, it is important to bear in mind that only the posterior parietal cortex shows an asymmetry in orienting (e.g., Mesulam, 1999; Nobre et al., 2000). Therefore, further studies of this asymmetry should take clues from our current understanding of the role of the posterior parietal cortex in attentional orienting. For example, some recent research suggests that the posterior parietal cortex is more involved in directing reflexive shifts of attention, whereas the frontal cortex is more involved in directing intentional shifts of attention (e.g., Gaymard et al., 1998). It is likely then that comparing differences in reflexive and intended shifts of attention may help to dissociate between the bilateral and unilateral control of orienting attention. If bilateral control confers some kind of advantage, then reflexive shifts to the right of space should be superior in some way than those made to the left. Importantly, a similar advantage should not exist for endogenous orienting because shifts in either direction are controlled unilaterally at the level of the frontal cortex. Some evidence already suggests that this line of research may be fruitful because (1) orienting both to the left and to the right is disrupted after lesions to the right posterior parietal cortex (e.g., Mesulam, 1999) and (2) saccadic eye movements are faster if elicited to abruptly appearing targets on the right side of space (Munoz et al., 1998). 66 Filling in the Void Mixed-case leiter rmtching consistently elicits a left-field and an across-display advantage. Finding that the left-field advantage may not arise from a right hemisphere advantage and the across-display advantages may not arise from interhemispheric sharing leaves these otherwise replicable visual field effects without theoretical interpretation. In this section, alternative explanations for the left-field and across-display advantages will be provided. The left-to-right bias in letter comparison. It is unlikely that the left-field advantage in letter matching indexes an advantage for the right hemisphere because a right-field advantage could not be elicited with response manipulations (Experiments 4 & 5) or with rhyming decisions (Experiment 6) that have been shown to benefit the left hemisphere in other studies. Instead, a right-field advantage was elicited when participants were asked to compare the letters in a right-to-left order (Experiment 7), a manipulation that should have not modified left-field advantage had it reflected an advantage for the right hemisphere (Hardycket al., 1985). One interpretation for the left-field advantage is that it reflects the spontaneous tendency of participants to compare letters in a left-to-right order. English readers have learned over many years to process text in a particular order and so this left-to-right bias reveals itself in letter rmtching tasks when letters must be examined individually to identify their name (Experiment 1-5 & 8) or sound (Experiment 6) (see also Bradshaw, Nettleton, Taylor, 1981; Bryden, 1960; Butler, 1978, 1979, 1981; Heron, 1957; Krueger, 1976; Levine & Banich, 1982; Lubow et al, 1994; Scheerer, 1972, 1973; Tramer, Butler, &Mewhort, 1985; White, 1976). One important consequence of this interpretation is that a right-field advantage in letter rmtcliing would be expected had the participants been skilled readers in a language in which the text is read from right-to-left (see Lubow et al, 1994). Although this finding is the expected outcome, establishing if reading practices do in fact alter the left-field advantage in letter matching can only be decided with further empirical investigation. In sum, the simplest and best-supported interpretation for the left-field advantage in letter matching is that it indexes the tendency of English reading participants to spontaneously compare English letters in a left-to-right direction. Therefore, the left-field advantage appears to be a consequence of learning to read in a specific direction, which is a product of culture, not of a right hemisphere advantage. Using terms from cognitive psychology, such habitual biases maybe interpreted in another way, as the emergence of a learned, top-down influence in the direction of spatial attention. The across-to-within bias in letter comparison. It is unlikely that the across-display advantage indexes interhemispheric sharing because the across-display advantage was eliminated when participants were asked to compare the probe letter appearing to the same side as the target first. Such serial strategies should not have altered the across-display advantage had it indexed interhemispheric sharing (Weissman & Banich, 2000). An alternative explanation for the across-display advantage is that it represents an advantage for shifting attention in the direction opposite to the preceding shift. Note that similar opposite 67 direction advantages have been documented in tasks that demand multiple shifts of attention. In the rapid saccade task, a saccade is made to one location and then to another. The speed of the second saccade is faster when it made in the direction opposite the first (Carpenter, 2001; Au et al, in submission). In the inhibition of return task, attention is drawn to two locations in sequence, the location of the cue and the ultimate location of the target. Detecting the target is more efficient when it appears in the position opposite the cue (Pratt et al., 1999). Finally, in mixed case letter matching, after the target is attended, attention is immediately diverted to one of the probe positions. Detecting, locating, or reporting the color of the rmtching probe is more efficient when the probe is in the field opposite to the target (Experiments 1-7). As such, I will adopt the neutral term opposite position advantage to refer to this general phenomenon. Obtaining an opposite position advantage in different tasks that demand multiple shifts of attention does not mean that this advantage occurs for the same reasons: Indeed, it could be a spurious correlation. However, if sirnilar experimental manipulations affect the opposite position advantage in comparable ways, then more support for the generality of this phenomenon will be provided. Comparing these literatures, three experimental manipulations have been used in more than two tasks, all suggesting that the opposite position advantage occurs for sirnilar reasons. Moreover, the influence of such manipulations may also provide some important insights into the mechanisms responsible for the distant location advantage. First, the opposite position advantage is eliminated with instructions in the letter-rmtching task . and in the rapid saccade task (Au et al., in submission). In mixed-case letter rmtching in the present study, the opposite position advantage was eliminated when participants were instructed to compare the probe appearing to the same side as the target first (Experiment 8). In a rapid saccade task, the opposite position advantage was elirninated when subjects were instructed to direct their gaze with an arrow at fixation (Au et al, in submission). The ability of instructions to eliminate the opposite position advantage suggests that it can be overridden with the intention of the participants. Second, presenting items sequentially may reduce the reliability of the opposite position advantage. Obtaining an across-display advantage was often hit or miss in the early studies exploring interhemispheric interactions (e.g., Dimond & Beaumont, 1971; Dimond, Gibson, &Gazzaniga, 1972). Subsequent research has indicated that the unreliability of the across-display advantage may have occurred because the to be compared visual objects were presented in sequence in these studies. If the objects are presented at the same time instead, then the across-display advantage becomes quite reliable (Banich & Belger, 1990; Berger, 1988; Serano &Kosslyn, 1991; Weissman, Banich, & Puente, 2000). A similar effect may exist in inhibition of return. Although this statement may seem odd because the inhibition of return task demands sequential presentation, consider that one study has challenged the opposite position advantage because it was found to be weak and often absent in many participants (Snyder, Schmidt, & Kingstone, in press). Albeit little more than speculation at this time, perhaps achieving the full opposite position advantage requires the simultaneous presentation of the relevant objects to be reliably observed. Third, the opposite position advantage maybe temporary in appearance. In the interhemispheric interactions literature, the opposite position advantage is eliminated and a repetition effect ensues after multiple sessions in a word narning task (Liberman, Merola, & Martinez, 1985). Similarly, a repetition effect is observed after multiple sessions in the rapid saccade task (Dorris, Pare, & Munoz, 2000). Since different tasks show that practice eventually reverses the distant location advantage suggests that its appearance is limited to a certain window in time. The reason why the 6 8 distant location advantage is overturned with practice is unknown at this time and should be addressed in future studies. Taken together, the opposite position advantage may be a general feature of orienting attention between multiple items. This effect maybe eliminated with instructions, weakened by when the items are sequentially presented, and overturned after much practice. However, beyond illustrating that similar behavioral consequences result from tasks that demand multiple shifts of attention, no explanation has been provided for this phenomenon. Here, I describe how the neurophysiological properties of target selection may be responsible for the opposite position effect. The frontal eye fields (FEF) is a brain region that is mtimately involved in planning saccadic eye movements, and in voluntary and reflexive shifts of attention (e.g., Nobre et al., 2000; Rosen et al, 1999; Schall & Thompson, 1999). Recent investigations have revealed important properties about the representation of targets and distractors in the FEF during visual search. An array of objects is presented to a monkey who must decide if one of the objects is different than the others (an oddball detection task). About 80 milliseconds after the onset of a display, the target and distractors are represented in the FEF by an increase in the firing rate of the neurons encoding those positions. Shortly after, the target of attention is selected by suppressing the responses evoked by the distractor items and by mamtaining or enhancing the responses elicited by the target (e.g., Schall & Hanes, 1993; Thompson, Bischot, & Schall, 1997). One important feature of these neural interactions is that the distractor items are differentially inhibited depending on their relative position to the target. To be specific, distractors that flank the target most closely receive greater inhibition than do distractors that are located farther away (Schall & Hanes, 1993). Another important quality of the FEF is its responsiveness to bias, or "intention". Thompson & Schall (2000) reported that sometimes the baseline activity of a neuron encoding one position was elevated before the array of objects appeared. When this occurred, monkeys were faster to respond to the item appearing in that position. This baseline shift in activity may index the neural correlate of strategic bias to a particular location (Thompson & Schall, 2000). In summary, the across-display advantage may arise in part from how the nervous system plans and execute shifts of attention. Inhibitory interactions that permit target selection cause the neural representation of objects that are closer to a target of attention to receive greater inhibition than other objects that are positioned farther away (e.g., Schall & Hanes, 1993). According to this interpretation then, the opposite position advantage may index a bottom-up influence on multiple shifts of attention. Bringing it all together: A proposed task analysis of mixed-case letter matching Mixed-case letter matching consistently produces left-field and across-display advantages. The evidence presented in this study indicates that a hemispheric interpretation of such effects (i.e., right hemisphere advantage or interhemispheric sharing) is not supported. Instead, mixed-case letter matching seems to index two separate attentional biases: A top-down bias to compare letters in a left-to-right order and a bottom-up influence advantaging the probe letter that is farther away. These behavioral biases may arise from, or at least be correlated with, known neurophysiological properties of the frontal eye fields (e.g., Schall & Hanes, 1993; Thompson & Schall, 2000). The top-down bias for the left-field maybe evidenced by an increase the baseline firing rate of neurons 69 encoding leftward shifts of attention (Thompson & Schall, 2000). The bottom-up bias for the opposite position may occur because this position experiences less inhibition than does the closer item (Schall &Hanes, 1993). Here a model is described that shows how these two interacting biases may produce the left-field and the across-display advantages associated with mixed-case letter rmtching. The layout of the model is simple. The middle rectangle denotes the "map of locations" in which object locations are represented; the bottom rectangle represents the bottom-up interactions involved in producing the across-display bias; and the top rectangle represents the top-down influences involved in first selecting the target letter and then facilitating the left probe's position. The scope of this model considers only shifts of attention among the target and probe letters. The identification of the letters and the subsequent decisions that are made with regard to response selection and execution are not considered because they are common to all the tasks considered. Consider same-case letter rmtching. Even though all of the display and response parameters are the same as mixed-case letter matching, same-case letter rmtching elicits no field or display effects. It has been argued that this occurs because a match can be made without shifting attention between the individual letters. There are two caveats to keep in mind when considering this model. First, although known properties of the FEF were used in its development, other brain regions that are involved in shifts of attention have demonstrated sirnilar target selection properties (e.g., superior colliculus, Basso & Wurtz, 1997, 1998). Second, this model is only intended to provide a simple description of how facilitory and inhibitory interactions may account for the data pattern obtained with the mixed-case letter-rmtching task To keep the model as simple as possible, the illustrations are neuroanatomically inaccurate (i.e., the mapping of objects does not reverse to opposite hemispheres) and neurophysiologically inaccurate (i.e., the absolute spatial mapping of the objects is not updated with every saccadic eye movement). These simplifications do not undermine the model; instead, they make it easier to present. There are two major assumptions in this model. The first is that mixed-case letter rmtching demands more than one shift of attention before a correct response can be made. This is a reasonable assumption because serial strategies mimic and eliminate the field and display effects (Experiments 7 & 8). The second assumption is that participants solve this task in a systematic manner, in that they inspect the target before inspecting either of the two probes. This assumption is reasonable because participants must consult each probe in order to make a correct response. On the other hand, an exhaustive inspection of two or more probe letters is not a requirement on each trial. When a match has been found the participant's task is formally over. After the appearance of the stimulus display (left panel, Figure 38), each letter is represented by an increase in the firing rate of the neurons representing its position (middle panel, Figure 38). Soon after, the neurons representing the target's position experience maintained or enhanced activity and the neurons representing the probe letter positions are differentially inhibited depending on their approximate location to the target. This differential activity allows attention to be shifted to the target's position (right panel, Figure 38). 70 Objects Target Registered Selected Figure 38. Selecting the target. Red arrow (originating from above in middle panel) denotes facilitatory top-down influences involved in selecting the target. The relative size of cylinders denotes the amount of neural activity for each position. The bottom cylinder in the right panel is colored in green (shaded in reproduced copies) to represent that attention is directed to its location. After the target has been identified, attention is then shifted to one of the two probe locations. Two interacting factors determine which probe will be selected as the next target of attention. First, owing to the bias to compare letters in a left-to-right order, the left probe position experiences a positive baseline shift in activity, providing a buffering effect against the inhibition that it otherwise receives. Second, owing to the design of the display and the differential inhibition among the object positions in the FEF (Schall & Hanes, 1993), the probe letter that is closer to the target letter (same visual field) experiences more inhibition than does the probe that is farther from the target (opposite visual field). For the displays in which matches are achieved most rapidly (left-field, across-display), the combined inhibitory and biasing effects interact in such a way to maximally advantage performance. The facilitatory baseline shift for the probe appearing in left-field combined with reduced inhibition of this position permits the representation of the left probe to quickly achieve high levels of activation, allowing attention to be efficiently shifted to this position, as depicted in Figure 39. 73 Display Activity pattern of Probe preceding shift Selected Figure 39. Illustration of top-down left-field bias and bottom-up across-display bias interacting to facilitate performance. Blue arrow (originating from below in middle panel) denotes greater inhibition. Green arrow (in right panel) represents shift of attention. For the conditions leading to intermediate matching times, the left-field bias and the inhibitory interactions counteract each other. Specifically, for the left-field, within-display trials, a bias towards the left-field facilitates the probe's selection, whereas the inhibitory interactions work against its selection because the target and probe appear in the same field. In contrast, for the right-field across-display trials, the right position experiences less inhibition because the target is in the opposite field, but the bias to the left-field works against its selection (illustrated in Figure 40). On a behavioral level, it is noteworthy these opposing forces often have similar clout, as performance is equally efficient for these display types (see Figure 7). Finally, for the displays in which matching takes the most time (right-field, within-display), the left-field bias and the inhibitory interactions interact in such a way that maximally disadvantage performance because both the baseline shift and inhibitory interactions work against this probe's location. These biases make it very likely that a second shift of attention is required so that this probe can be examined (illustrated in Figure 41). Necessitating a third shift of attention may explain why performance can sometimes be atypically poor for these trials. Greater activation may be required for attention to be shifted to this position because it was inhibited twice before. 72 Display Activity pattern of preceding shift Probe Selected Figure 40. Example of the separate left-field and across-display biases canceling each other out. Attention can be shifted equally well in either direction. Display Activity pattern of preceding shift Probe Selected Figure 41. Illustration of left-field and across-display biases interacting to impair performance. Large green arrow (thick arrow in right panel) illustrates that attention incorrectly shifted to left probe. Small green arrow (dashed arrow in right panel) illustrates that a second shift of attention shift is necessary to achieve the target in the right position. The accuracy of this model can be evaluated in several ways. Serial processing in mixed-case letter matching. This model assumes that serial processing, necessitated by multiple shifts of attention, allows the left-field and opposite position advantages to occur. If this is true, then mixed-case letter matching should behave like any other serial task when 73 subjected to similar circumstances. One way to delineate between tasks demanding serial or parallel processing is to compare performance between sequentially and simultaneously presented stimulus displays. If sequentially presenting items facilitates performance, then serial processing is supported. If simultaneously presenting items facilitates performance, then parallel processing is supported (e.g., Hoffman, 1998). This study has already been conducted. Weissman and Banich (2000) compared the across-display advantage for sequentially presented, partially overlapping, and simultaneously presented displays. The baseline reaction times between these different conditions revealed that performance was fastest when the items were presented in sequence, was intermediate when the items partially overlapped in time, and was slowest when the items were presented simultaneously. This finding provides much support for a serial processing interpretation of mixed-case letter rmtching. Top-down left-to-right bias. The left-field advantage is proposed to be a consequence of a top-down bias in comparing letters in a left-to-right order. Considering that this left-to-right bias is absent for geometric shapes (e.g., Bryden, 1960), a straightforward prediction is that the left-field advantage should be absent when the stimuli are used that do not elicit such top-down biases. This study has already been conducted. Koivisto (2000) used similar display parameters as the letter-matching task, except pictures were presented instead of letters. Although an across-display advantage was found for complex comparisons (category matching), a left-field advantage was not elicited. This finding provides support that a top-down bias to compare letters in a left-to-right order may be responsible for the left-field advantage. Bottom-up across-display bias. According to this model, the critical factor involved in producing the across-display advantage is the unequal distance between the target and the probe letters. Changing the target's position so that it is positioned directly in between the two probe's positions results in a straightforward prediction: The across-display advantage should be eliminated. This experiment has already been conducted. Weissman & Banich (2000) presented participants with a three-item letter matching task that was identical to the one used in this study except the target could also appear at the midline (see Figure 42). Comparing the "midline-displays" to the typical across- and within-displays revealed a noteworthy pattern: Performance for the midline-displays was intermediate to that for the across and within trials (Experiment 1). This finding provides support that the unequal distance between the letters may be responsible for the advantage for the across-display and the disadvantage for the within-display (see also Schmitz-Gielsdorf, et al., 1988). However, there are two other factors that may be involved in producing the opposite position effect. The locations of the objects are important because objects that flank the target receive greater inhibition than do objects that are farther away from the target (Schall & Hanes, 1993). The trajectory, or direction, of attention is also important because an object appearing along the path of an attention shift may receive less inhibition than do objects that are on a different path (Pratt etal., 1999). 74 B R + b B R + b Within-Display Across-Display Midline-Display Figure 42. The different displays used in the Weissman & Banich (2000) study. Notice that up to now, I have limited the discussion of the opposite position effect to the distance principle (i.e., that farther objects may experience less inhibition than do closer objects) for good reason. Distance is only relevant for displays that consist of 3-items because objects must flank or not flank the initial target for such effects to occur. However, when more than 3-items are presented, the relative positions of the objects may be more important than the absolute distances between the objects. Some evidence exists to support this view because objects that flank an initial target experience more inhibition than do objects that do not flank the target, even when the distance between the items has been equated in an inhibition of return task (Pratt et al., 1999). If the relative positioning of the items is important (after distance is controlled), then this position effect may index how the brain is reconfigured to accommodate changes in workspace. Workspace refers to the region of space that is relevant for a task When a workspace is created for observers and its size is varied, participants are better able to detect objects that are presented in a small workspace and are less sensitive to identify targets that are presented in a larger workspace (e.g., LaBerge & Brown, 1986; Robertson et al., 1993). Such changes in sensitivity imply that the brain is reconfigured dynamically to accommodate changes in workspace. The inhibition of return paradigm creates a very stable workspace for the observer, in which ghost boxes (empty squares) indicate where all of the relevant information will be presented. Perhaps creating this stable visual environment creates the boundaries for the workspace of attention. In this workspace, even if the opposite position is the same distance to the target as the other items, it is still not adjacent to the target. Perhaps, in such environments greater inhibition may exist for flariking targets, rather than for absolute distances. In addition to creating a 4-item version of the letter-rmtching task that specifically manipulates the relative distancing and the relative positions between the letters, this alternative explanation maybe examined by creating an unstable environment, which may eKminate such position effects. The direction principle only holds for sequentially presented displays because attention must be heading in a circumscribed direction. This is the case in an inhibition of return experiment. Namely, a cue appears to the periphery, which is followed by a "reorienting" event (essentially another cue) at center fixation, and then a target appears. Each of these events drive attention in different directions: The initial cue drives attention out to a position in the periphery, the reorienting event drives attention back towards fixation (and in the opposite direction of the initial cue), and then the target appears, which draws attention to its position. So, if the target appears in the opposite position as the initial cue, then attention can maintain its trajectory and its momentum 75 remains the same. If the target appears in another position, then attention must be swerved from its trajectory to reach the target, which slows its momentum. Simply put, keeping attention moving in the same direction facilitates performance. However, the direction that attention is heading in cannot be determined in the letter-matching task because all letters appear simultaneously. Without a reorienting event, there is no reason to assume, or to even expect, that attention would shift back to the fixation marker before shifting to one of the probe locations. Moreover, if attention was shifted from the target to either of the probe positions directly, its trajectory would be straight and therefore the attentional momentum would be the same in both conditions. In sum, even though only distance was discussed as the cause of the opposite position effect up to now (because only distance was relevant for the 3-item letter-matching task), this does not mean that other factors are uriimportant to our ultimate understanding of the inhibitory interactions that may occur during target selection. On the contrary, the separate and perhaps interactive influences of distance, flanking, and direction need to be systematically evaluated because each factor would have its own unique affect on brain activity. Role of the vertical midline. This model holds that the vertical midline (or different hemispheres) does not contribute to the across-display advantage because the vertical midline does not have a special influence on target selection in the FEF (Schall & Hanes, 1993). Some evidence supports this statement because the vertical midline has no special influence on the opposite position advantage found in studies of the inhibition of return (Pratt et al., 1999; Snyder, Schmidt, & Kingstone, in submission). A future study that rotated the displays, so that the within- and across- display differences are now mapped to the horizontal midline, would directly assess this possibility. If this model is correct, then a similar across-display advantage should be noted for the upper and lower visual fields even though there would be no reason to expect parallel processing mediated by the hemispheres under these conditions. Top-down influences eliminating biases. Finally, this model predicts that the intentions of the participant can counteract the top-down (i.e., left-to-right) or bottom-up (i.e., across-to-within) attentional bias. Experiments 7 and 8 dealt with this issue most directly, in which the left-field advantage was reversed and the across-display advantage was eliminated with instructions. Boundary conditions of the model It is time to consider the data from the same-case letter matching conditions and the 9-item displays. Same-case letter rmtching. Same-case letter matching (3-item displays) provides a unique way of assessing the boundaries of the proposed model because it demonstrates the importance of serial processing in eliciting the left-field and across-display advantages in mixed-case letter rmtching. Same-case letter matching is assumed to be solved in parallel and for this reason no field or display advantages are found. The most straightforward way to assess the accuracy of a parallel processing interpretation of 3-item, same-case, letter rmtching is to determine if presenting the letters simultaneously facilitates performance over sequential presentation (e.g., Hoffman, 1998). This study has not yet been conducted. 76 9-item displays. The 9-item displays provide another boundary condition for the model because these displays did not elicit consistent field or display advantages. The most straightforward explanation for this effect is that increasing the number of items (and the difficulty of the task) changed both the left-to-right strategy that participants used as well as the bottom-up opposite direction effect. Whether it is task difficulty of the spatial array that is the more critical variable has yet to be determined. However, we already know that some very difficult tasks, even when they only involve 3 item arrays, work to eliminate the field and display effects. For instance, a difficult rhyming decision resulted in a sirnilar absence of display and field effects in a previous study (Belger & Banich, 1998). Guiding the participants' attention during letter matching may be one way to return strategy to the 9-item displays (e.g., Woodman & Luck, 1999). In this case, the number of items would remain the same, but the participants would select among only a subset of those letters on the basis of a single feature; for instance, color. Even though the number of items remains identical, the participants could adopt a systematic scanning order for the small number of distinctively located items in the array. Tests following a sirnilar logic, but examining spatial variables, could be used to see which spatial variables are required to return the across-display advantage to the results. In summary, the left-field and across-display advantages associated with mixed-case letter matching arise from two separate, but interacting, biases. This model describes how these biases work together to facilitate performance (positive bias and less suppression), work against one another and seemingly nullify such influences (e.g., positive bias and greater suppression), or work together to hinder performance (no bias and more suppression). Focal attention must be demanded and attention must be shifted among multiple items for such biases to be observed. If a task can be completed without requiring serial shifts of attention across space, as maybe true for same-case letter matching, then no field br display advantages are obtained. Furthermore, if the difficulty of the task alters the strategies that participants adopt in shifting attention among letters, as may have been the case in the 9-item displays, then the field and display advantages are also elirninated. Conclusions This study has made important contributions to two different fields of study. First, the study tested alternative explanations for a robust pattern of data in letter-matching tasks that can be found in the literature on interhemispheric interactions. These tests showed, in a variety of ways, that it is unlikely that the letter-matching task produces data that is an index of interhemispheric interactions. As such, the strongest line of support for an influential theory of hemispheric interactions, the computational complexity theory (Banich, 1998; Banich & Belger, 1990; Weissman & Banich, 2000), has been underrnined. Other methods will need to be developed to study the important questions of interhemispheric interactions. Second, the findings of this study inadvertently pointed to the mixed-case letter matching task as a potentially useful tool in the study of how attention is shifted among multiple items in the visual field. The tests that were conducted in this study, and the subsequent model that was developed to account for the data, suggest that this task may yet become an important one in the study of the interactions between top-down and bottom-up influences on attentional orienting. Among the useful features of this behavioral task is that it is simple in its basic design (a niinimum of three items can be used), it lends itself very naturally to the separation of various influences on attentional priority (bottom-up influences of biased object competition, top-down influences of a 77 priori location biases), and it can be easily modified to accommodate other stimuli (letters, images, words), cognitive analyses (feature rmtching, abstract identity, rhyming, categorization) and response requirements (identity, location). 78 References Andersen, RA, Snyder, LH, Batista, AP, Buneo, CA, Cohen, & YE (1998). Posterior parietal areas specialized for eye movements (LIP) and reach (PRR) using a common coordinate frame. Novartis Foundation Symposium, 218.109-122. Au, C.S., Fecteau, JH, Armstrong, IT, &Munoz, D. P. (submitted). Sensations have consequences: Reinterpreting the alternation advantage as inhibition of return. Psychological Science Banich, M.T. (1998). Integration of information between the cerebral hemispheres. Current Directions in Psychological Science 7, 32-37. Banich, M.T. (1998). The missing link: The role of interhemispheric interaction in attentional processing. Brain and Cognition. 36. 128-157. Banich MT, Goering S, Stolar N, & Belger A. (1990) Interhemispheric processing in left- and right-handers. International Journal of Neuroscience. 54. 197-208 Banich MT, Passarotti AM, & Janes D. (2000) Interhemispheric interaction during childhood: I. Neurologically intact children. Developmental Neuropsychology. 18, 33-51. Banich MT, Passarotti AM, White DA, Nortz MJ, & Steiner RD. (2000) Interhemispheric interaction during childhood: II. Children with early-treated phenylketonuria. Developmental Neuropsychology, 18, 53-71. Banich, MT & Belger, A (1990). Interhemispheric interaction: How do the hemispheres divide and conquer a task? Cortex. 26, 77-94. Banich, MT & Karol D (1992). The sum of the parts does not equal the whole: Evidence from Bihemispheric processing. Journal of Experimental Psychology: Human Perception and Performance. 18, 763-784. Banich, MT & Nicholas, CD (1998). Integration of processing between the hemispheres in word recognition. In M Beeman & C Chiarello (Eds) Right hemisphere language comprehension: Perspectives from cognitive neuroscience. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc., (pp. 51-77). Banich, MT & Shenker, JI (1994). Dissociations in memory for item identity and item frequency Evidence from hemispheric interactions. Neuropsychologia. 32. 1179-1194 Beaumont, JG, Dimond, SJ (1973). Transfer between the cerebral hemispheres in human learning. Acta Psychologica, 37, 87 - 91 Beaumont, G & Dimond, S (1975) Interhemispheric transfer of figural information in right and non right handed subjects. Acta Psychologica, 39, 97-104 Belger & Banich (1992) Interhemispheric interaction affected by computational complexity. Neuropsychologia. 30. 923-929. 79 Belger, A & Banich, MT (1998). Costs and benefits of integrating information between the cerebral hemispheres: A computational perspective. Neuropsychology. 12, 380-98. Berger, JM (1988) Interhemispheric cooperation and activation in integration of verbal information. Behavioral Brain Research. 29,193-200. Berger, JM & Landolt, M (1990) Bilateral processing of numbers: Parallel processing versus functional unity. Cortex, 26, 555-556. Berger, JM, Perret, E (1986) Interhemispheric integration of information in a surface estimation task. Neuropsychologia, 24, 743-746 Berger, JM, Perret, E, Zirnmermann, A (1988). Role of interstimulus and stimulus-hemisphere compatibility in the process of interhemispheric integration. Perceptual and Motor Skills. 6Z, 503-512. Bradshaw, JL, Nettleton, NC, & Taylor, MJ (1981). The use of laterally presented words in research into cerebral asymmetry Is direction scanning likely to be a source of artifact? Brain and Language. 14.1-14. Brown, V (1998). Comparing parallel and sequential multinomial models of letter identification. In CE Dowling & FS Roberts (Eds) Recent progress in mathematical psychology: Psychophysics. knowledge, representation, cognition, and measurement, pp. 253-284. Mahwah, NJ, US: Lawrence Erlbaum Associates, Inc., Publishers Brown, WS & Jeeves, MA (1993) Bilateral visual field processing and evoked potential interhemispheric transmission time. Neuropsychologia, 31,1267-1281. Brown, WS, Larson, EB, & Jeeves MA (1994). Directional asymmetries in interhemispheric transmission time: Evidence from visual evoked potentials. Neuropsychologia. 32. 439-448 Bruce, CJ, Goldberg, ME, Bushnell, MQ & Stanton, GB (1985). Primate frontal eye fields. II. Physiological and anatomical correlates of electrically evoked eye movements. Journal of Neurophysiology. 54, 713-34. Bryden, MP (1967). A model for the sequential organization of behavior. Canadian Journal of Psychology. 21. 37 - 56. Bryden, MP (1970). Left-right differences in tachistoscopic recognition as a function of familiarity and pattern recognition,. Journal of Experimental Psychology, 84, 120 - 122. Bryden, MP. (1978) Strategic effects in the assessment of hemispheric asymmetry. In G. Underwood (Ed.). Strategies of Information Processing. London: Academic Press. Butler, BE (1978) Sequential and parallel operations in tachistoscopic recognition. Canadian Journal of Psychology, 32, 241-256. Butler, BE (1979). A note of familiarity and scanning with tachistoscopic recognition. Canadian Journal of Psychology, 33. 65 - 70. 80 Butler, BE (1981). Canadian studies of visual information-processing: 1970-1980. Canadian Psychology. 22. 113-128. Callaghan, T.C (1989). Interaction and dominance in texture segregation: Hue and geometric form. Perception & Psychophysics. 46. 299-311. Callaghan, T.C., Lasaga, M.I. & Garner, W.R. (1986). Visual field segregation based on orientation and hue. Perception &C Psychophysics. 39. 32-38. Carpenter, RH (2001) Express saccades: is bimodality a result of the order of stimulus presentation? Vision Research. 41.1145 - 51. Carson, RG, Summers, TJ, Walters, MR, & Semjen, A (1997). The dynamics of bimanual circle drawing. Quarterly Journal of Experimental Psychology A, 50. 664-83. Chelazzi, L & Corbetta, M (2000). Cortical mechanisms of visuospatial attention in the primate brain. In MS Gazzaniga (Ed.). New Cognitive Neurosciences USA; MIT Press, (pp. 667-686) Chellazzi, L, Duncan, J, Miller, EK, & Desimone, R. (1998). Responses of neurons in inferior temporal cortex during memory-guided visual search. Journal of Neurophysiology. 80. 2918 - 40 Collins, M (1999) Differences in semantic category priming in the left and right cerebral hemispheres under automatic and controlled processing conditions. Neuropsychologia. 37. 1071 - 85. Coney, J (1985) Dual hemispheric processing in a letter rmtching task. Perception and Psychophysics. 38. 331-42. Corbetta, M (1998) Frontoparietal cortical networks for directing attention and the eye to visual locations: identical, independent, or overlapping neural systems? Proceedings of the National Academy of Science. USA. 95. 831-838. Cowan, N (2001). The Magical Number 4 in Short-term Memory A Reconsideration of Mental Storage Capacity. Behavioral and Brain Sciences 24 (1) preprint, no page numbers giim Davis, R & Schmit, V (1973). Visual and verbal coding in the interhemispheric transfer of information. Acta Psychologica. 37. 229-240. diGirolamo, GJ, & Posner MI (2000). Introduction (to attention). In MS Gazzaniga (Ed.). New Cognitive Neurosciences USA: MIT Press, (pp. 623 - 632) Dimond S. & Beaumont G (1971). Use of two cerebral hemispheres to increase brain capacity. Nature. 232. 270 - 271 Dimond, SJ & Beaumont, JG (1972). On the nature of the interhemispheric effects of fatigue. Acta Psychologica. 36. 443-449 81 Diamond, S. & Beaumont, G (1974). Hemisphere function and paired-associate learning. British Journal of Psychology. 65. 275-278 Diamond, SJ, Gibson, AR, Gazzaniga, MS (1972). Corss field and within field integration of visual inforamtion. Neuropsychologia. 379-381 Dorris, MC, Pare, M, &Munoz, DP (2000). Immediate neural plasticity shapes motor performance. Journal of Neuroscience. 20. RC52. Egeth, H, & Yantis, S (1997). Visual attention: Control, representation, and time course. Annual Review of Psychology. 48. 269-97. Enns, J. T. & Kingstone, A (1997). Hemispheric coordination of spatial attention. In S. Christman (Ed.) Cerebral asymmetries in sensory and perceptual processing (pp. 197-231). New York: Elsevier. Enns, J. T., &Rensink, R A (1992). VScope: General purpose tachistiscope for the Macintosh. Vancouver, Canada: Micropsych Software. Eviatar, Z & Zaidel, E. (1994). Letter matching within and between the disconnected hemispheres. Brain and Cognition. 25.128 - 37. Fecteau, JH, Enns, JT, & Kingstone, A (2000) Competition-induced visual field differences in search. Psychological Science. 11. 386 - 93. Friedman, A & Poison, MC. (1981). Hemispheres as independent resource systems: limited-capacity processing and cerebral specialization. Journal of experimental psychology Human perception and performance. 7. 1031-1058. Gaymard B & Pierrot-Deseilligny C. (1999). Neurology of saccades and smooth pursuit. Current Opinion in Neurology. 12.13 - 9. Gaymard B, Ploner CJ, Rivaud S, Vermersch AI, & Pierrot-Deseilligny C. (1998) Cortical control of saccades. Experimental Brain Research. 123.159 - 63. Goldberg, ME, Eggers, FDVI, & Gouras, P (1992). The ocular motor system. In ER Kandel, JH Schwartz, &TMJessell (Eds). Principles of Neuroscience. 3rd Edition. USA Appleton & Lange Inc. Hardyck, C, Chiarello, C, Dronkers, NF, & Simpson GV (1985) Orienting attention within visual fields: How efficient in interhemispheric transfer? Journal of Experimental Psychology Human Perception and Performance. 11. 650-66 Hatta, T & Tuji, S (1993) Interhemispheric integration of number stimuli: Comparisons of Arabic with Kanji numerals Cortex. 29. 359-64. Heilrnan, K M & Van Den Abell, T (1980) Right hemisphere dominance for attention: the mechanism underlying hemispheric asymmetries of inattention (neglect). Neurology. 30. 327 - 330. 82 Heilrnan, KM, & Gonzalez-Rothi, LJ, (1997). Limb apraxia: A look back. In LJ Gonzalez Rothi & K M Heilrnan (Eds) Apraxia: The neuropsychology of action. Brain damage, behaviour and cognition series, (pp. 7-18). Hove, England: Psychology Press/Erlbaum (UK) Taylor & Francis. Hellige, JB, Cox, PJ, & Litvac, L (1979). Information processing in the cerebral hemispheres: Selective activation and capacity limitations. Journal of Experimental Psychology General. 108,251- 279. Hellige, JB, Cox, PJ, & Litvac, L (1979). Information processing in the cerebral hemispheres: Selective hemispheric activation and capacity limitations. Journal of Experimental Psychology: General. 108. 251-279 Heron, W. (1957) Perception as a function of retinal locus and attention. American Journal of Psychology. 70. 38-48. Jeeves, MA & Lamb, A (1988). Cerebral asymmetries and interhemispheric processes. Behavioural Brain Research. 29.211-233. Kahneman, D & Triesman, A. (1984). Changing views of attention and automaticity. In R. Parasuraman & D.R. Davies (Eds.), Varieties of Attention (pp. 29 - 61). New York: Academic Press. Kimura, D. (1966). Dual functional asymmetry of the brail in visual perception. Neuropsychologia. 4. 275 - 285. Kimura, D (1973). The asymmetry of the human brain. Scientific American. 228. 70-78. Kinsbourne, M. (1970). The cerebral basis of lateral asymmetries in attention. Acta Psychologica. 33, 193-201. Kinsbourne, M (1972) Eye and head turning indicates cerebral lateralization. Science. 176. 539 -541. Kinsbourne, M. (1973). The control of attention by interaction between the cerebral hemispheres. In S. Kornblum (Ed.), Attention and Performance IV (pp. 239-256). New York Academic Press. Kinsbourne, M (1974). Direction of gaze and distribution of cerebral thought processes. Neuropsychologia. 12. 279 - 281. Kinsbourne, M (1977). Hemi-neglect and Hemisphere rivalry. Advances in Neurology. 18. 41 -49. Kinsbourne, M & Hicks, RE (1978). Functional cerebral space: A model for overflow, transfer, and interference effects in human performance: A tutorial review. In J. Requin (Ed.) Attention and Performance VII Hillsdale, NJ: Erblaum, pp. 345 - 362. 83 Kinsbourne, M & Hiscock, M (1983) Asymmetries of dual task performance. In JB Hellige (Ed.), Cerebral hemisphere asymmetry Method, theory and application (pp. 255-334). New York Praeger. Klein, RM (2000) Inhibition of return. Trends in Cognitive Science. 4.138-147. Klein, RM &MacInnes, WJ (1999). Inhibition of return is a foraging facilitator in visual search. Psychological Science. 10. 346-352. Klingberg T, O'Sullivan BT, Roland PE. (1997) Bilateral activation of fronto-parietal networks by incrementing demand in a working memory task Cerebral Cortex. 7. 465-71. Koivisto, M (2000) Interhemispheric interaction in semantic categorization of pictures. Cognitive Brain Research. 9.45 -51. Krueger, LE (1976). Evidence for directional scanning with the order-of-report factor excluded. Canadian Journal of Psychology. 30. 9 - 14. LaBerge, D, & Brown, V (1986). Variations in size of visual field in which targets are presented: An attentional range effect. Perception and Psychophysics. 40. 188 - 200. Lamb MR, & Robertson LC. (1989). Do response time advantage and interference reflect the order of processing of global- and local-level information? Perception and Psychophysics. 46, 254 - 8. Lamb, M, & Robertson, LC (1988). The processing of hierarchical stimuli: effects of retinal locus, locational uncertainty, and stimulus identity. Perception and Psychophysics, 44. 172- 81. Larson, EB & Brown WS (1997). Bilateral field interactions, hemispheric specialization, and evoked potential interhemispheric transmission time. Neuropsychologia. 35, 573-581. Lavie, N (1995). Perceptual load as a necessary condition for selective attention. Journal of experimental psychology Human perception and performance, 21. 451-468. Leiber, L (1982). Interhemispheric effects in short-term recognition for single words. Cortex. 18. 113-124 Levine, SC & Banich, MT (1982). Lateral asymmetries in the naming of words and corresponding line drawings. Brain and Language. 17. 34 - 43. Levy, J & Trevarthen, C (1977) Perceptual, semantic, and phonetic aspects of elementary language processes in split-brain patients. Brain. 100.105-18. Liderman, J. (1986) Interhemispheric interference during word naming. International Journal or Neuroscience, 30. 43-56. Liderman, J &Meehan, P (1986) When is between hemisphere division of labor advantageous? Neuropsychologia. 24. 863-874 84 Liderman, J., Merola, J., Martinez, S. (1985). Interhemispheric collaboration in response to simultaneous bilateral input. Neuropsychologia. 23, 673-683. Lubow, RE, Tsal, Y, Mirkin, A, Mazliah, G (1994). English and Hebrew letter report by English-and Hebrew-reading subjects: evidence for stimulus control, not hemispheric asymmetry. Brain and Cognition. 25.34 - 51. Luck SJ, Hillyard SA, Mangun GR, & Gazzaniga MS. (1989). Independent hemispheric attentional systems mediate visual search in split-brain patients. Nature. 342. 543-5. Ludwig, TE, Jeeves, MA, Norman, WD, & DeWitt, R (1993). The bilateral field advantage on a letter matching task Cortex. 29. 691-713 McPeek RM, Skavenski AA, Nakayama K. (2000). Concurrent processing of saccades in visual search. Vision Research. 40. 2499 - 2516. Merikle, PM, Lowe, DG, & Coltheart, M. (1971). Familiarity and method of report as determinants of tachistoscopic recognition. Canadian Journal of Psychology. 21. 167-174. Merola, JL & Liderman, J (1987). Developmental versus individual differences in the ability of the hemispheres to operate independently. International Journal of Neuroscience. 35.195-204. Mesulam, MM (1981) A cortical network for directed attention and unilateral neglect. Annals of Neurology. 10. 309 - 325. Mesulam, MM (1999) Spatial attention and neglect: parietal, frontal and cingulate contributions to the mental representation and attentional targeting of salient extrapersonal events. Philosophical transactions of the Royal Society of London B: Biological Sciences. 354. 1325 - 1346. Mewhort, DJ, Merikle, PM, & Bryden, MP. (1969) On the transfer from iconic to short-term memory. Journal of Experimental Psychology. 81. 89-94. Miller, EK & Cohen, JD (2001) An integrative theory of prefrontal cortex function. Annual Review of Neuroscience. 24. 167 - 202. Miller, LK (1981) Perceptual independence of the hemifields in children and adults. Journal of Experimental Child Psychology. 32. 298-312. Munoz DP, Broughton JR, Goldring JE, & Armstrong IT (1998). Age-related performance of human subjects on saccadic eye movement tasks. Experimental Brain Research. 121. 391 -400. Nobre, AC, Gitelman, DR, Dias, E Q & Mesulam MM (2000) Covert visual spatial orienting and saccades: overlapping neural systems. Neuroimage. 11. 210 - 216. Norman, WD, Jeeves, MA, Milne, A, & Ludwig, T (1992). Hemispheric interactions: The bilateral advantage and task difficulty. Cortex. 28. 623-642. 85 Olson, CR (2001) Object-based vision and attention in primates. Current Opinion in Neurobiology. 11. 171-9. Pratt, J. Spalek, T.M., & Bradshaw, F (1999). The time to detect targets at inhibited and non-inhibited locations: Preliminary evidence for attentional momentum Journal of Experimental Psychology: Human Perception and Performance. 25. 730-746 Posner, MI, Boies, SJ, Eichelman, WH, & Taylor, RL (1968). Retention of visual and name codes of single letters. Journal of Experimental Psychology. 79.1 - 16 Posner, MI, & Cohen, Y (1984). Components of visual orienting. In H Bouma & D. Bouwhuis Eds. Attentionaand Performance. Volume X. Hillsdale, NJ: Erlbaum. Pp. 531-556. Posner MI &Dehaene S. (1994). Attentional networks. Trends in Neuroscience. 17. 75 - 9. Posner MI & Mitchell RF (1967) Chronometric analysis of classification. Psychological Review. 74, 392-409. Posner, MI & Petersen, SE (1990). The attention system of the human brain. Annual Review in Neuroscience. 13. 25- 42. Pylyshyn, ZW (2001). Situating vision in the world. Trends in Cognitive Science. 4.197-207. Rao, SM (1995) Neuropsychology of multiple sclerosis. Current opinion in neurology. 8. 216 - 20. Rasmussen P. (1993) Persistent mirror movements: a clinical study of 17 children, adolescents and young adults. Developmental Medicine and Child Neurology. 35. 699-707. Rayman J, &Zaidel E. (1991). Rhyming and the right hemisphere. Brain &c Language. 40. 89-105. Reuter-Lorenz, PA, Stanczak, L, & Miller, AC (1999). Neural recruitment and cognitive aging: Two hemispheres are better than one, especially as you age. Psychological Science. 10. 494-500. Rosen AC, Rao SM, Caffarra P, Scaglioni A, Bobholz JA, Woodley SJ, Hammeke TA, Cunningham JM, Prieto TE, Binder JR. (1999) Neural basis of endogenous and exogenous spatial orienting. A functional MRI study. Journal of Cognitive Neuroscience. 11.135 — 152. Schall, JD & Hanes DP (1993). Neural basis of saccade target selection in frontal eye field during visual search. Nature. 366. 467 - 9. Schall, JD & Thompson, K G (1999) Neural selection and control of visually guided eye movements. Annual Review of Neuroscience. 22. 241-259. Scheerer, E (1972). Order of report and order of scanning in tachistoscopic recognition. Canadian Journal of Psychology. 26. 382-390. Scheerer, E (1973) A further test of the scanning hypothesis in a tachistoscopic recognition. Canadian Journal of Psychology. 27. 95-102. 86 Scheerer, E (1974) Task requirement and hemifield asymmetry in tachistoscopic partial report performance. Acta Psychologica. 38. 131-147. Schmitz-Gielsdorf, J., Willmes, K., Vondenhoff, G, Hartje, W. (1988). Effects of spatial arrangement of letter pairs in a name-matching task with unilateral and bilateral hemifield stimulation. Neuropsychologia. 26. 591-602 Sereno, AB & Kosslyn SM (1991). Discrimination within and between hemifields: A new constraint on theories of attention. Neuropsychologia. 29. 659-675 Sergent, J (1983). Role of input in visual hemispheric asymmetries. Psychological Bulletin. 93. 481 - 512. Snowden, RJ. (1998). Texture segregation and visual search: a comparison of the effects of random variations along irrelevant dimensions. Journal of Experimental Psychology: Human Perception and Performance. 24.1354-1367. Snyder, JJ, Schmidt, WC, & Kingstone, A (in submission). Attentional momentum does not underlie the inhibition of return effect. Sperry, R (1982). Some effects of disconnecting the cerebral hemispheres. Science. 217.1223 - 6. Stuphorn, V, Taylor, TL, & Schall, JD (2000). Performance monitoring by the supplementary eye field. Nature. 408. 857-60. Tanji, J & Hoshi, I (2001). Behavioral planning in the prefrontal cortex. Current Opinion in Neurobiology. 11.164 - 170. Theeuwes, J, Kramer, AP, & Atchley, P (1999). Attentional effects on preattentive vision: Spatial precues affect the detection of simple features. Journal of Experimental Psychology. 25. 341-347. Thompson, KG, Bichot, NP, & Schall JD (1997). Dissociation of visual dkcrirnination from saccade prograrrirning in macaque frontal eye field. Journal of Neurophysiology. 77.1046 -50. Thompson, K G & Schall, JD (2000). Antecedents and correlates of visual detection and awareness in macaque prefrontal cortex.Vision Research. 40.1523 - 38. Tramer, O, Butler, BE, & Mewhort, DJ (1985) Evidence for scarming with unilateral visual presentation of letters. Brain and Language. 25.1-18. Trick, LM & Pyfyshyn, ZW (1993). What enumeration studies can show us about spatial attention: evidence for limited capacity preattentive processing. Journal of Experimental Psychology Human Perception and Performance. 19. 331- 51. Triesman AM (1986) Features and objects in visual processing. Scientific American. 255. 114 -125. 87 Triesman, AM & Gelade, G (1980) A feature-integration theory of attention. Cognitive Psychology. 12. 97 - 136. Weissman, D H & Banich, MT (2000). The cerebral hemispheres cooperate to perform complex but not simple tasks. Neuropsychology. 14. 41-59. Weissman, DH, Banich, MT, & Puente, EI (2000). An unbalanced distribution of inputs across the hemispheres facilitates interhemispheric interactions. Journal of the International Neuropsychological Society. 6. 313-321. White, MJ (1976). Order of processing in visual perception. Canadian Journal of Psychology. 30. 140 - 156. Woodman, GF & Luck, SJ (1999). Electrophysiological measurement of rapid shifts of attention during visual search. Nature. 400. 867-869. Yeshurun, Y & Carrasco, M (2000) The locus of attentional effects in texture segmentation. Nature Neuroscience. 3. 622-7. Yohsizaki, K & Tsuji, Y (2000). Benefits of interhemispheric integration on the Japanese Kana script-matching tasks. Perceptual and Motor Skills. 90. 153-165. Young, AW & Ellis AW (1983) Directional scanning or reporting preferences are not the cause of visual hemifield * report interactions. Cortex. 19. 475 - 480. Zaidel E & Peters AM (1981) Phonological encoding and ideographic reading by the disconnected right hemisphere: two case studies. Brain and Language. 14. 205 - 34. Zhang, W &Feng, L (1999). Interhemispheric interaction affected by identification of Chinese characters. Brain and Cognition. 39. 93-99. 88 Appendix I: Table summarizing field and display advantages across studies exploring interhemispheric interactions The table below lists all behavioral studies of letter rmtching in which both visual field effects and display effects were reported (adult, right-handed participants only). The last two columns summarize whether the data provided positive (V), mixed (~), or negative (X) support for the computational complexity and asymmetrical orienting theory. Positive support for computational complexity theory included (1) a within-display advantage for simple decisions and/or (2) an across-display advantage for complex decisions. Negative evidence included (1) an across-display advantage for simple decisions or (2) a within-display advantage for complex decisions. Note that the table does not include all studied relevant to the computational complexity theory. Some studies have provided positive (e.g., Reuter-Lorenz, Stanzak, & Miller, 1999), mixed (e.g., Sereno &Kosslyn, 1990) and negative (e.g., Schmidt et al., 1988) support for this theory but they were not included because field advantages were not reported. Positive support for asymmetrical orienting theory was counted as (1) an across-display advantage occurring in conjunction with a left field, or no display advantage and (2) a within-display advantage occurring in conjunction with a right-field advantage. Negative evidence was counted as (1) a right-field advantage occurring in conjunction with an across-display advantage or (2) a left-field advantage occurring in conjunction with a within-display advantage or no display differences. Many of the studies listed included multiple experiments that are not mentioned individually in this table. Only different tasks or similar tasks that yielded different results were given a separate entry. Left or right field advantage was determined on within-displays or on unilateral displays. "No display advantage" indicates that performance for one (or both) of the unilateral displays equaled or surpassed performance of the bilateral displays when broken down into left and right field unilateral displays. Similar to previous studies (Banich & Karol, 1992; Belger & Banich, 1998), the absence of an across-display (or bilateral) advantage was taken as a within-display (or unilateral) advantage because there was no benefit for parallel processing between the hemispheres. Reference Task Field Advantage Display Advantage CC Theory AO Theory 3- item or similar displays Belger & Banich (1992) Letter matching LVF Across display advantage V V Weissman, Banich, & Puente (2000) Letter matching LVF Across display advantage V Weissman & Banich (2000) Letter matching Categorv-matching LVF Across display advantage 89 Banich & Belger (1990) Letter matching Summation-matching Ordinal- matching LVF Across display advantage Belger & Banich (1998) Letter matching Rhyme- matching LVF RVF trend Across display advantage No advantage < Banich, Goering, Stolar, & Belger (1990) Word and letter matching LVF trend Across-display advantage X Zhang & Feng (1999) Homonym-matching Synonym-matching. LVF trend RVF trend Across display advantage ~ Koivisto (2000) Picture- matching None Across display advantage 2- item or similar displays Koivisto (2000) Picture matching None Across display advantage < Davis & Schmidt (1973) Letter- matching None Across display advantage — X Norman et al., (1992) Letter- matching LVF Across display advantage Coney (1985) Letter- matching None Across display advantage V Yoshiazaki & Tsuji (2000) Letter- matching LVF Across display advantage V Larson & Brown (1997) Pattern- matching Letter matching LVF None Across display advantage ~ V Banich & Shenker (1994) Identity memory-Frequency estimates LVF Not consistent Within display advantage X X Banich & Belger (1991) Line judgment RVF Within display advantage X Banich & Karol (1992) Rhyme- matching RVF Within display advantage V Jeeves & Lamb (1988) Dot-matching LVF Across display advantage Brown & Jeeves (1993) Letter matching RVF Across display advantage X 90 Ludwig et al., (1993) Letter matching Experiments: 1-3 4 LVF RVF Across display advantage ~ ~ Liderman (1986) Word-naming Word-recognition RVF None No advantage No advantage X Liderman & Meehan (1986) Chunked orientation letter-naming Unilateral Mixed orientation letter-naming None None Across display advantage No advantage Liderman, Merola, & Martinez (1985) Word categorization RVF Within display advantage (with practice) X Hatta & Tuji (1993) Digit summation English digits Kanji digits RVF LVF Across display advantage Berger, Perrett, & Zimmermann (1988) Picture-word naming: Most combinations Bilateral: Picture (LVF) Word (RVF) RVF Cannot determine No advantage Across display advantage ~ Berger & Landolt (1990) Digit summation Words, dots Bar graphs, Mixed displays RVF trend LVF trend RVF trend No advantage X Berger (1988) Semantic decision Simultaneous Sequential None RVF Across display advantage No advantage V V Berger & Perrett (1986) Size-matching LVF Within display advantage X X 91 Miller (1981) Letter- matching Word- matching None RVF Across display advantage Leiber (1982) Word memory RVF Within display advantage X Dimond, Gibson, & Gazzaniga (1972) Word repetition RVF trend Wthin display advantage X V Dimond & Beaumont (1974) Paired associate learning RVF Within display advantage X V Beaumont & Dimond (1975) Figure- matching Long exposure Short exposure LVF None No advantage No advantage X X Dimond (1975) Word-naming Bilateral RVF Across display advantage X Dimond & Beaumont (1971) Digit-naming RVF Across display advantage V X Dimond & Beaumont (1972) Picture-matching LVF Across display advantage 92 Appendix II: Summary of reaction time and error data Thesis Experiment 1 Display X Field of matching probe Same-case letter matching Mixed-case letter matching RT Error RT Error Within-display, left probe 642 .052 828 .114 Across-display, left probe 635 .055 771 .074 Within-display, right probe 634 .035 860 .170 Across-display, right probe 637 .063 799 .108 Target Absent 675 .022 875 .058 Table A2-1. Experiment 1 reaction time and error scores broken down by display and location of rmtching probe Source Same-case letter matching RT Error D f F P MSe df F P MSe Display i, 23 <1 >.l 62.89 1, 23 <3.6 >.05 .002 Probe Location 1, 23 <1 >.l 228.595 1, 23 <1 >.l .007 Display x Probe location 1, 23 <1 >.l 669.313 1, 23 <1.2 >.l .003 Table A2-2. Source table of statistical analyses for same-case letter rmtching in experiment 1, all interactions shown. Source Mixed-case letter matching RT Error df F P MSe df F P MS e Display i, 23 33.30 6 <05 2501.398 1, 23 16.387 <05 .004 Probe Location 1, 23 58.08 <05 3791.888 1, 23 6.487 <05 .007 Display x Probe location 1, 23 <1 >.l 2166.626 1, 23 1.129 >.l .002 Table A2-3. Source table of statistical analyses for mixed-case letter rmtching in experiment 1, all interactions shown. 93 Thesis Experiment 2 Display X Field of matching probe Same-case letter matching Mixed-case letter matching RT Error RT Error Within-display, left probe 635 .037 803 .142 Across-display, left probe 628 .038 732 .125 Within-display, right probe 645 .042 834 .217 Across-display, right probe 637 .046 766 .116 Target Absent 746 .046 818 .067 Table A2-4. Experiment 2 reaction time and error scores broken down by display and location of matching probe. Note the reaction time data for the target absent condition is an estimate as data from 4 subjects were lost. Source Same-case letter matching RT E rror df F P MSe df F P MSe Display 1, <1.2 >.l 1067.955 1, <1 >.l .0002 23 23 Probe Location 1, <1 >.l 3757.485 1, <1 >.l .003 23 23 Display x Probe 1, <1 >.l 799.269 1, <1 >.l .002 location 23 23 Table A2-5. Source table of statistical analyses for same-case letter matching in experiment 2, all interactions shown. Source Mixed-case letter matching RT E rror df F P MSe df F P MSe Display i, 23 66.743 <05 1715.470 1, 23 12.486 <05 .007 Probe Location 1, 23 5.256 <05 4856.398 1, 23 4.705 <05 .006 Display x Probe location 1, 23 <1 >.l 972.689 1, 23 13.043 <05 .003 Table A2-6. Source table of statistical analyses for mixed-case letter matching in experiment 2, all interactions shown. 94 Thesis Experiment 3 No color display: Mean correct RTs and mean error, see Table A2-4 Display X Field of matching probe Same-case letter matching Mixed-case letter matching RT Error RT Error Within-display, left probe 647 .025 786 .118 Across-display, left probe 625 .028 732 .071 Within-display, right probe 685 .069 871 .195 Across-display, right probe 654 .036 787 .125 Table A2-7. Experiment 3 A and 3C reaction time and error scores broken down by display and location of matching probe for color displays Source Same-case letter matching RT Error df F P MSe df F P MSe Color i, <l >.l 27435.54 1, < >.l .005 51 6 51 Color x Display 1, 4.790 <05 1102.543 1, 3.178 <.09 .001 51 51 Color x Probe 1, <2.6 >.l 3109.391 1, <1.3 >.l .004 51 51 Color x Display x Probe 1, <1 >.l 781.834 1, 3.316 <08 .001 51 51 Table A2-8. Source table of statistical analyses involving Color variable in Experiment 3 A, all interactions shown. Source Same-case letter matching RT Error df F P MSe df F P MSe Color Relationship l , 28 <1.8 >.l 4593.832 1, 28 <l >.l .003 Display x Color Relationship 1, 28 <1.5 >.l 1611.237 1, 28 <i >.l .003 Probe x Color Relationship 1, 28 <1 >.l 1399.345 1, 28 <i >.l .003 Display x Probe x Color Relationship 1, 28 <1 >.l 2224.996 1, 28 <l >.l .003 Table A2-9. Source table of statistical analyses involving Color Relationship in Experiment 3B, all interactions shown. 9 5 Source Mixed-case letter matching RT Error df F P MSe df F P MSe Color l, 51 <1 >.l 53766.17 5 1, 51 <l >.l .043 Color x Display 1, 51 <l >.l 1102.543 1, 51 <l >.l .006 Color x Probe 1, 51 4.318 <05 4319.038 1, 51 <2.1 >.l .007 Color x Display x Probe 1, 51 <1.8 >.l 2015.143 1, 51 <2.8 >.l .005 Table A2-10. Source table of statistical analyses involving Color variable in Experiment 3C, all interactions shown. Source Mixed-case letter matching RT Error df F P MSe df F P MSe Color Relationship l, 28 <l >.l 1932.037 1, 28 4.661 <04 .004 Display x Color Relationship 1, 28 <l >.l 4004.315 1, 28 <1 >.l .008 Probe x Color Relationship 1, 28 <1 >.l 6053.903 1, 28 <1 >.l .003 Display x Probe x Color Relationship 1, 28 <i >.i 3219.064 1, 28 <1 >.l .009 Table A2-11. Source table of statistical analyses involving Color variable in Experiment 3D, all interactions shown. 96 Thesis Experiment 4 Display X Field of matching probe Same-case letter matching Mixed-case letter matching RT Error RT Error Within-display, left probe 647 .025 786 .118 559 .024 671 .060 Across-display, left probe 625 .028 732 .071 554 .016 644 .068 Within-display, right probe 686 .069 871 .195 591 .030 726 .093 Across-display, right probe 654 .036 787 .125 566 .023 695 .083 Target Absent (Detection only) 787 .024 844 .057 Table A2-12. Experiment 4 reaction time and error scores broken down by display and location of matching probe. Detection scores above hashed line, localization scores below hashed line. Source Same-case letter matching RT Error df F P MSe df F P MSe Display i, 28 18.992 <05 2261.911 1, 28 10.344 <05 .001 6.964 <05 1811,487 2.031 >.l .001 Probe Location 1, 28 13.011 <05 5154.057 1, 28 3.932 <06 .010 6.682 ^ <05 4328.879 <1 >.l .006 Display x Probe location 1, 28 <1 >.l 1535.024 1, 28 8.907 <05 .002 4.677 <05 1186.463 <1 >.l .001 Table A2-13. Source table of statistical analyses for same-case letter niatching in experiment 4. Detection scores above hashed line, localization scores below hashed line Source Mixed- c as e letter mate hing RT Error df F P MSe df F P MSe Display i, 28 30.832 <05 8997.355 1, 28 16.786 <.05 .012 6.517 <05 7545.194 2.748 >.l .005 Probe Location 1, 28 36.949 <05 7755.269 1, 28 16.286 <.05 .015 43.443 <05 3724.632 <l".8 1 >.l .003 Display x Probe location 1, 28 <2.2 >.l 5742.888 1, 28 <1 >.l .011 <1 >.l 4508.526 4.163 <06 .007 Table A2-14. Source table of statistical analyses for mixed-case letter niatching in experiment 4. Detection scores above hashed line, localization scores below hashed line 97 Thesis Experiment 5 3-item Displays Display X Same-case letter Mixed-case letter Field of matching probe matching mate hing RT Error RT Error Within-display, left probe 718 .048 829 .108 651 .038 739 .055 1025 .043 1163 .072 Across-display, left probe 686 .028 830 .106 645 .027 719 .043 976 .027 1121 .074 Within-display, right probe 793 .079 972 .234 689 .033 833 .120 1101 .042 1254 .087 Across-display, right probe 738 .056 837 .108 633 .013 739 .069 1024 .038 1131 .068 Target Absent 840 .035 909 .063 (Detection Only) Table A2-15. Experiment 5 reaction time and error scores broken down by display and location of matching probe for 3-item displays. Detection scores above hashed line, localization scores in between hashed lines, and color report score below hashed line. Source 3-item display, same-case letter matching RT Error df F P MSe df F P MSe Display l, 28 27.321 <05 3972.03 7 1, 28 6.047 <05 .005 16.812 <05 3309.95 1 5.268 <05 .003 1, 27 30.345 <05 7394.28 9 1, 27 <L.6 >.l .003 Probe Location • 1, 28 15.5 <05 15109.1 45 1, 28 3.494 <08 .010 <2.1 >.l 4499.26 1 3.039 <.l .002 1, 27 21.349 >.l 10189.2 10 1, 27 <1 >.l .003 Display x Probe location 1, 28 <2.1 >.l 3734.96 8 1, 28 <1 >.l .002 6.213 <05 5692.62 0 <1 >.l .001 1, 27 <L.4 >.l 8509.96 5 1, 27 <1 >.l .003 Table A2-16. Source table of statistical analyses for 3-item displays, same-case letter matching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 98 Source 3-item display, mixed-case letter matching RT Error df F P MSe df F P MSe Display 28 23.677 <05 10940.814 1, 28 11.946 <05 .020 53.019 <05 3538.289 4.816 <05 .012 1, 27 30.040 <05 12475.856 1, 27 <1 >.l .006 Probe Location 1, 28 20.355 <05 16014.839 1, 28 9.767 <05 .024 17.761 <05 10661.272 12.136 <05 .010 1, 27 7.629 <05 19721.941 1, 27 <1 >.l .007 Display x Probe location 1, 28 31.739 <05 8523.404 1, 28 2.690 >.l .009 12.107 <05 6727.898 <1 >.l .001 1, 27 8.418 <05 10912.173 1, 27 <1 >.l .011 Table A2-17. Source table of statistical analyses for 3-item displays, mixed-case letter rmtching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 9-item displays Display X Same-case letter Mixed-case letter Field of matching probe mate hing mate hing RT Error RT Error Within-display, left probe 911 .271 960 .273 933 .334 977 .353 1232 .330 1259 .366 Across-display, left probe 876 .292 932 .319 882 .219 960 .068 1165 .344 1220 .312 Within-display, right probe 880 .263 946 .258 883 .229 924 .093 1199 .308 1228 .355 Across-display, right probe 902 .210 968 .253 887 .318 953 .342 1185 .274 1244 .342 Target Absent 994 .26 1020 .29 (Detection Only) Table A2-18. Experiment 5 reaction time and error scores broken down by display and location of matching probe for 9-item displays. Detection scores above hashed line, localization scores in between hashed lines, and color report score below hashed line. 99 Source 9-item display, same-case letter matching RT Error df F P MSe df F P MSe Display i, 28 <l >.l 5696.927 1, 28 <1.1 >.l .014 4.84 <05 6707.448 <1 >.l .035 1, 27 10.062 <05 9224.192 1, 27 <l >.l .017 Probe Location 1, 28 <1 >.l 5533.963 1, 28 6.351 <05 .019 <3.1 >.05 9533.165 <1 1 >.l .027 1, 27 <1 >.l 12981.67 2 1, 27 <2.5 >.l .048 Display x Probe location 1, 28 <3.4 >.05 14079.01 5 1, 28 3.303 <08 .025 4.013 <06 11132.39 4 12.512 <05 .048 1, 27 6.164 <05 16306.60 8 1, 27 <1.4 >r 1 .025 Table A2-19. Source table of statistical analyses for 9-item displays, same-case letter rmtching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. Source 9-item display, mixed-case letter matching RT Error df F P MSe df F P MSe Display 1, 28 <1 >.l 7077.134 1, 28 <1 >.l .010 <1 >.l 5215.310 <1.1 >.l .035 1, 27 <1 >.l 10707.653 1, 27 <2.7 >.l .024 Probe Location 1, 28 <1 >.l 8973.411 1, 28 <1 >.l .015 7.549 <05 6923.140 5.346 <05 .014 1, 27 <1 >.l 11038.259 1, 27 <1 >.l .032 Display x Probe location 1, 28 <3.1 >.05 11905.856 1, 28 <1 >.l .028 <i.7 1 >.l 17687.189 2.491 >.l .081 1, 27 3.677 >06 11596.761 1, 27 <1.1 >.l .023 Table A2-20. Source table of statistical analyses for 9-item displays, mixed-case letter rmtching in experiment 5. Detection scores above hashed line, localization scores in between hashed lines, and color report below hashed line. 100 Thesis Experiment 6 Display X Field of matching probe Rhyme with EE RT Error Within-display, left probe 1236 .138 Across-display, left probe 1201 .150 Within-display, right probe 1178 .197 Across-display, right probe 1161 .141 Target Absent Data 1415 .096 Table A2-21. Experiment 6 reaction time and error scores broken down by display and location of rmtching probe. Note the target absent reaction time and error scores are an estimate of the actual value as data from 3 subjects were lost. Source Rhyme with EE RT Error df F P MSe df F P MSe Display i , 30 1.444 >.l 28930.45 2 i, 30 7.051 < .05 .012 Probe Location (Within-display comparison) 1, 30 5.589 <05 18772.05 4 1, 30 2.377 <15 .012 Display x Probe location 1, 30 <1 > . l 18772.05 4 1, 30 <1.7 > . l .010 Table A2-22. Source table of statistical analyses in experiment 6. 101 Thesis Experiment 7 D i s p l a y X F i e l d o f m a t c h i n g probe Left- to-r ight Right-to- lef t RT Error RT Error Witliin-display, left probe 807 .106 854 .154 Across-display, left probe 749 .069 783 .084 Within-display, right probe 873 .194 818 .148 Across-display, right probe 816 .126 758 .110 Target Absent 879 .058 876 .062 Table A2-23. Experiment 7 reaction time and error scores broken down by display and location of matching probe for each instruction set. Source Left- to-r ight R T E r r o r df F P MSe df F P MSe Display 19.457 <05 3995.568 i, 22.703 < .003 23 23 .05 Probe Location 1, 15.135 <05 7039.096 1, 13.512 <05 .009 23 23 Display x Probe 1, <1 >.l 2040.068 1, <1.9 >.l .003 location 23 23 Table A2-24. Source table of statistical analyses of left-to-right instructions in experiment 7. Source Right-to- lef t R T E r r o r df F P MSe df F P MSe Display i, 23 34.154 <05 2980.133 i, 23 11.877 < .05 .006 Probe Location 1, 23 4.102 <06 5323.622 1, 23 <1 >.l .008 Display x Probe location 1, 23 <1 >.l 1188.413 1, 23 <2 >.l .003 Table A2-25. Source table of statistical analyses of right-to-left instructions in experiment 7. 102 Thesis Experiment 8 Display X Field of matching probe Ac ros s - to-within Within-to-across R T Error R T Error Within-display, left probe 898 .115 862 .088 Across-display, left probe 802 .069 892 .085 Within-display, right probe 946 .143 905 .120 Across-display, right probe 850 .061 895 .097 Target Absent 861 .049 868 .040 Table A2-26. Experiment 8 reaction time and error scores broken down by display and location of matching probe for each instruction set. Note the target absent reaction time and error scores are an estimate of the actual value as data from 2 subjects were lost. Source Ac ross-to-within RT Error df F P MSe df F P MSe Display i, 23 82.393 <05 2677.747 i, 23 21.846 <.05 .004 Probe Location 1, 23 13.304 <05 4081.318 1, 23 <1 >.l .002 Display x Probe location 1, 23 <1 >.l 2511.116 1, 23 3.375 <.08 .002 Table A2-27. Source table of statistical analyses of across-to-within instructions in experiment 8. Source Within-to-across RT Error df F P MSe df F P MSe Display i, 23 <1 >.l 2516.296 i, 23 < 2.2 >.l .002 Probe Location 1, 23 4.537 <05 2694.238 1, 23 4.58 6 <.05 .003 Display x Probe location 1, 23 4.257 <06 2343.520 1, 23 <1 >.l .003 Table A2-28. Source table of statistical analyses of within-to-across instructions in experiment 8 103 

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}]}"
                            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.831.1-0090743/manifest

Comment

Related Items