Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Anisotropy of covert, endogenous orienting of attention across the visual field Roggeveen, Alexa Bleiweis 2007

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

Item Metadata

Download

Media
831-ubc_2007-319178.pdf [ 7.49MB ]
Metadata
JSON: 831-1.0100740.json
JSON-LD: 831-1.0100740-ld.json
RDF/XML (Pretty): 831-1.0100740-rdf.xml
RDF/JSON: 831-1.0100740-rdf.json
Turtle: 831-1.0100740-turtle.txt
N-Triples: 831-1.0100740-rdf-ntriples.txt
Original Record: 831-1.0100740-source.json
Full Text
831-1.0100740-fulltext.txt
Citation
831-1.0100740.ris

Full Text

ANISOTROPY OF COVERT, ENDOGENOUS ORIENTING OF ATTENTION ACROSS THE VISUAL FIELD by A L E X A BLEIWEIS R O G G E V E E N B.A. The Johns Hopkins University, 2001 M . A. The University of British Columbia, 2003 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Psychology) THE UNIVERSITY OF BRITISH C O L U M B I A September 2007 © Alexa Bleiweis Roggeveen, 2007 ABSTRACT Are there advantages to voluntarily attending to some visual field locations over others? The work discussed here explores the question of whether covert, endogenous (voluntary) orienting to various visual field locations has anisotropic effects. Important to this question is understanding how paradigm parameters alter the way that attention is voluntarily distributed across possible target locations, and how that ultimately affects the anisotropy of performance at different visual field locations. When observers are cued to attend with 100% certainty to a visual field location, the effects of endogenous orienting run parallel to previous findings of perceptual and attentional anisotropy (vertical anisotropy; horizontal vs. vertical meridian). This is the first work to demonstrate an attentional benefit to voluntarily attending to the vertical meridian compared to the horizontal meridian. Lowering the cue's predictive value altered the pattern of anisotropic performance, revealing that attention had varying impact with distance from fixation, as well as a greater impact on the lower half of the vertical meridian. This result reflects how attention was distributed across possible target locations in the display, given that the target may not appear at the cued location. Further experiments showed that task parameters have a distinct effect on how attention is distributed across possible target locations. Altering the endogenous distribution of attention by eliminating or reducing an inhibitory gradient at fixation affects performance to the greatest degree close to fixation, and on the horizontal meridian. Both experiments are also the first to reveal an attentional oblique effect, whereby response time to targets presented on the intercardinal axes is consistently and significantly slower than on the cardinal axes. A key role is also demonstrated for the use of placeholders (perceptual objects) which facilitate the objects) which facilitate the voluntary distribution of attention across possible target locations. The effect of placeholders interacts with the cue's predictive value to alter how attention is voluntarily distributed across possible target locations. Overall, the answer to the overarching question is yes: covert, endogenous orienting has anisotropic effects. In light of these findings, four general guidelines are presented to illustrate the impact of covert, endogenous distribution of attention across the visual field. TABLE OF CONTENTS Abstract ii Table of Contents iv List of Tables vii List of Figures viii Acknowledgements x Co-Authorship Statement xi Chapter I: Introduction 1 1.1 References 3 Chapter II: Anisotropy of covert endogenous orienting across the visual field 4 2.1 Introduction 4 2.1.1 Perceptual anisotropy 5 2.1.2 Attentional anisotropy 10 2.1.2.1 Evidence from exogenous orienting ; .11 2.1.2.2 Evidence from endogenous orienting 12 2.1.2.3 Upper vs. lower visual field anisotropy 13 2.1.2.4 Right-left attentional anisotropy 15 2.1.3 Attention distribution. 18 2.2 Experiment 1 24 2.2.1 Methods 25 2.2.1.1 Observers 25 2.2.1.2 Materials 25 2.2.2 Results 27 2.2.2.1 Reaction time 27 2.2.2.2 Errors 29 2.2.3 Discussion 29 2.3 Experiment 2 30 2.3.1 Methods 31 2.3.1.1 Observers 31 2.3.1.2 Materials 31 2.3.2 Results 32 2.3.2.1 Reaction time 32 2.3.2.2 Errors 33 2.3.3 Discussion 33 2.4 Experiment 3 36 2.4.1 Methods 38 2.4.1.1 Observers 38 2.4.1.2 Materials 38 2.4.2 Results 38 2.4.2.1 Reaction time 38 2.4.2.2 Errors 41 2.4.3 Comparison with Experiment 1 41 iv 2.4.4 Discussion 42 2.5 General discussion 44 2.6 Tables and Figures - Chapter 2 50 2.7 References 64 Chapter III: Effects of endogenous attentional distribution across the visual field on performance: another form of attentional anisotropy 74 3.1 Introduction 74 3.1.1 Perceptual and attentional anisotropies 74 3.1.2 Attention distribution effects across visual field 77 3.2 Experiment 1 81 3.2.1 Methods 81 3.2.1.1 Observers 81 3.2.1.2 Materials 82 3.2.2 Results 85 3.2.2.1 Reaction time 85 3.2.2.2 Errors 86 3.2.3 Discussion 87 3.3. Experiment 2 88 3.3.1 Methods 89 3.3.1.1 Observers 89 3.3.1.2' Materials , 89 3.3.2 Results 90 3.3.2.1 Reaction time 90 3.3.2.2 Errors 91 3.3.3 Discussion 91 3.4 Experiment 1 & 2 comparison 93 3.4.1 Reaction time 93 3.4.2 Errors 94 3.4.3 Discussion 94 3.5 General discussion 96 3.5.1 Perceptual and attentional anisotropies 96 3.5.2 Effects of attentional distribution 99 3.6 Tables and Figures - Chapter 3 103 3.7 References 112 Chapter IV: The impact of paradigm parameters on the allocation of endogenous attention: the role of objects and likelihood 120 4.1 Introduction 120 4.1.1 Perceptual and attentional anisotropy 123 4.1.2 Object-based selection: the role of placeholders 127 4.2 Experiment 1 130 4.2.1 Methods 131 4.2.1.1 Observers 131 v 4.2.1.2 Materials 131 4.2.2 Results 133 4.2.2.1 Reaction time 133 4.2.2.2 Errors 134 4.2.3 Discussion 135 4.3 Experiment 2 136 4.3.1 Methods 137 4.3.1.1 Observers 137 4.3.1.2 Materials 137 4.3.2 Results 138 4.3.2.1 Reaction time 138 4.3.2.2 Errors 139 4.3.3 Discussion 139 4.4 Experiment 3 140 4.4.1 Methods 141 4.4.1.1 Observers 141 4.4.1.2 Materials 141 4.4.2 Results 141 4.4.2.1 Reaction time 141 4.4.2.2 Errors 142 4.4.3 Discussion 142 4.5 Experiment 4 143 4.5.1 Methods 143 4.5.1.1 Observers 143 4.5.1.2 Materials 144 4.5.2 Results 144 4.5.2.1 Reaction time 144 4.5.2.2 Errors 145 4.5.3 Discussion 145 4.6 Experiment 5 146 4.6.1 Methods 147 4.6.1.1 Observers 147 4.6.1.2 Materials 147 4.6.2 Results 148 4.6.2.1 Reaction time 148 4.6.2.2 Errors 148 4.6.3 Discussion 149 4.7 General Discussion 149 4.8 Tables and Figures - Chapter 4 153 4.9 References 170 Chapter V: Conclusion 175 5.1 References 185 Appendix A : Supplementary analyses 187 Appendix B: Results summaries 188 V I LIST OF TABLES Table 2.1. Mean cue effects for individual subjects, Experiment 3 50 Table 3.1. Main effect of Validity: Experiment 1 101 Table 3.2. Main effect of Eccentricity, Experiment 1 102 Table 3.3. Main effect of Validity, Experiment 2 103 Table 3.4. Main effect of Eccentricity, Experiment 2 104 Table 4.1 Main effect of eccentricity in error rates, Experiment 1 153 A l l values are in percent error. Table 4.2 Main effect of eccentricity in reaction times, Experiment 2 154 A l l values are in milliseconds. Table 4.3 Main effect of eccentricity in reaction times, Experiment 5 155 A l l values are in milliseconds. Table 4.4 Main effect of Validity - Experiment 5. A l l values are in milliseconds 156 vii LIST OF FIGURES Figure 2.1. Schematic of typical trials in Experiment 1. Upper half depicts trial where. ..51 cue indicates with 100% certainty where the target will appear. The lower half depicts a trial where the cue does not indicate where the target will appear Figure 2.2. Main effect of eccentricity, Experiment 1 52 Figure 2.3. Cue x Location interaction, Experiment 1 53 Figure 2.4. Difference scores (uncued reaction time - cued reaction time) for stimuli...54 presented at all eight radial locations (N, N E , E, SE, S, SW, W, N) , Experiment 1. Figure 2.5. Cue x Eccentricity interaction, Experiment 2 55 Figure 2.6. Difference scores (reaction time to invalidly cued targets - reaction time... .56 to validly cued targets) for stimuli presented at all eight radial locations (N, NE, E, SE, S, SW, W, N), Experiment 2. Figure 2.7. Difference scores (reaction time to invalidly cued targets - reaction time...57 to validly cued targets) for stimuli presented at the four eccentric locations (1°, 4°, 8°, and 12° from fixation), Experiment 2. Figure 2.8. Average reaction times within cuing condition for Experiments 1 and 2 58 Figure 2.9. Cue x Eccentricity interaction for group with consistent cuing 59 effect, Experiment 3. Figure 2.10. Mean reaction times to targets presented at four eccentric locations for 60 subjects who did not show cuing effects, Experiment 3. Figure 2.11. Mean reaction times to targets presented at eight radial locations for 61 subjects who did not show cuing effects, Experiment 3. Figure 2.12. Difference scores (reaction time to invalidly cued targets - reaction 62 time to validly cued targets) for stimuli presented at the four eccentric locations (1°, 4°, 8°, and 12° from fixation), Experiment 3. Calculated only from subjects who showed an overall cuing advantage. Figure 2.13. Difference scores (reaction time to invalidly cued targets - reaction 63 time to validly cued targets) for stimuli presented at all eight radial locations (N, NE, E, SE, S, SW, W, N), Experiment 2. Calculated only from subjects who showed an overall cuing advantage. vin Figure 3.1. Schematic of display sequence in Experiments 1 and 2 105 Figure 3.2. Main effect of Radial location, Experiment 1 106 Figure 3.3. Main effect of Radial location, Experiment 2 107 Figure 3.4. Interaction between Experiment, Radial location, and Eccentricity: 108 comparison between Experiments 1 and 2. Figure 3.5. Interaction between Experiment and Radial location: comparison 109 between Experiments 1 and 2. Figure 4.1. Schematic of sample trial 157 Figure 4.2. Validity x Eccentricity interaction, Experiment 1 158 Figure 4.3. Eccentricity x Radius interaction, Experiment 1 159 Figure 4.4. Main effect of Radial location - Experiment 1 160 Figure 4.5. Radial location x Validity interaction - Experiment 2 161 Figure 4.6. Eccentricity x Radial location interaction - Experiment 3 162 Figure 4.7. Main effect of Radial location, error data - Experiment 3 163 Figure 4.8. Main effect of Radius in reaction time - Experiment 4 164 Figure 4.9. Eccentricity x. Validity interaction - Experiment 4 165 Figure 4.10. Main effect of Radius in error rates - Experiment 4 166 Figure 4.11. Main effect of eccentricity in error rates - Experiment 4 167 Figure 4.12. Main effect of Radius - Experiment 5 168 Figure 4.13. Radius x Eccentricity interaction - Experiment 5 169 ix ACKNOWLEDGEMENTS I would first like to acknowledge my supervisor, Dr. Lawrence Ward, who gave me the opportunity, support, and freedom to pursue this degree. Thank you for having faith in my ability to get to this point, and being the mentor to help me get here. I would also like to thank three of my colleagues in the Psychophysics and Cognitive Neuroscience Lab, each who had their own important, yet different, impact on this dissertation. First, thank you to Dr. David Prime, who taught me what I needed to know, about finding my own answers, and to never think that I'm done learning. Your friendship and mentorship were indispensable in finishing this work. Second, thank you to Stephanie Thai, who tirelessly helped with the data collection for these experiments. Your help allowed me to fulfill two important personal and professional goals: completing this dissertation, and teaching throughout the last year, and I am enormously grateful. Third, thank you to my friend and ally Dr. Michael Baumann. Your unique perspective taught me to question things I may not have otherwise - training an academic should not be without. Finally, I would like to thank my husband, Dr. Paul Hommersen, for his unwavering support throughout the entire process. Thank you for being my best friend, for always knowing what I was capable of, and ultimately for keeping me sane. I can't imagine being able to do this without you. CO-AUTHORSHIP STATEMENT I, Alexa Roggeveen, was principally responsible for the identification and design of the research program detailed within this dissertation. I was personally responsible for all data collection, by being either directly involved in the data collection itself, or for providing direct supervision of the collection of data. I was also solely responsible for the preparation of the experimental setup for data collection, as well as all data analyses. I was also principally responsible for the preparation of the three manuscripts and additional materials included here. I received editorial assistance from my supervisor, Dr. Lawrence Ward in the design of the experiments, as well as preparation of the manuscripts. 1 Introduction When we want to move our attention around the visual field, or the area of the viewable world that we can see when we open our eyes, we typically move our eyes to gaze directly at a particular location in space: we make an overt shift of attention. It is well known, however, that we are also highly capable of shifting our attention to a location in the visual field in a covert way, without moving our eyes (e.g,. Helmholtz, 1867). Both overt and covert shifts of attention are thought to be elicited in two different ways. A shift of attention may be exogenous, or involuntary, caused by an attention-grabbing stimulus out in the world. Attention shifts may also be endogenous (voluntary), where the observer makes the choice to shift their attention to one location or another (see Wright & Ward, 1998). The work described in the following three chapters seeks to answer the question of whether the effects of covert, endogenous orienting are anisotropic (not all identical) at all locations in the visual field. Essentially, are we better at attending to some visual field locations than we are at attending to others? When we attend to a specific location in space, our visual performance improves, usually in the form of faster and more accurate responses to suddenly appearing target stimuli. If the covert, endogenous orienting to different visual field locations has an unequal impact on performance between those locations, the effects of covert, endogenous orienting would be understood to be anisotropic. Does performance on reaction time and accuracy measures change in a uniform way as a result of a covert, endogenous shift of attention, or do we see a larger impact on reaction time and accuracy in some locations more than others? 1 Generally, the answer to this question is yes: the impact of attending voluntarily to different locations in space is not always the same. Moreover, as illustrated in the work to be described in this dissertation, the pattern of anisotropy is not consistent across various task conditions. While there are examples of anisotropic performance in some cases as the result of a cue - the traditional method of asking observers to shift their attention - much of the anisotropy is tied to the way in which observers distribute their attention across possible target locations under conditions where a target will not always appear at the cued location. Overall, the anisotropies of performance observed arise from an interaction between where in the visual field the observer is cued to attend and several important characteristics of the paradigm and stimuli used. Chapter Two discusses the impact of the likelihood that the target will appear at the cued location, and includes a comprehensive review of literature relevant to both perceptual and attentional anisotropies. Chapter Three reveals the role of the placement of possible target locations in shaping the distribution of endogenous attention. Finally, the role of perceptual objects in the selection and distribution of attention, as well as a further manipulation of the cue's predictive validity, are investigated in Chapter Four. Each of the chapters is presented in the format in which the work was submitted to the journals indicated within each chapter, in accordance with the University of British Columbia Faculty of Graduate Studies formatting requirements. The work of the three chapters together provides an initial framework for an understanding of the effects of endogenous attention orienting across the visual field. 2 1.1 References Helmholtz, H. von. (1867; 1924). Treatise on Physiological Optics. Translated by J.P.C. Southall, The Optical Society of America. Wright, R.D., & Ward, L . M . (1998). The control of visual attention. In Visual Attention, R.D. Wright, Ed. Oxford: New York, 132-186. 3 2 Anisotropy of covert endogenous orienting across the visual field1 2.1 Introduction It has been understood for quite some time that visual attention can be oriented covertly, or shifted to a location away from where the eyes are fixated (e.g,. Helmholtz, 1867; Jonides, 1981; Posner, Snyder & Davidson, 1980). What is not yet completely understood is whether attending in a voluntary, covert way to each of the various possible locations in the visual field has equivalent costs and benefits. There are clear anisotropic patterns of both speed and accuracy of response to stimuli presented in different visual field locations. To what degree that anisotropy arises solely from the design of the visual system, however, or rather from anisotropies in the effects of voluntary attention orienting, is unresolved. In the present paper we address two key questions about covert allocation of attention across the visual field. First, is the impact of attending, in a covert, voluntary way the same at all locations in the visual field, or does attending to some locations afford greater benefits than that to other locations? Some answers to this question have come from research concerning involuntary (exogenous) shifts of attention. Little evidence exists in the area of voluntary (endogenous) shifts of attention, however, so we emphasize the latter in the work presented here. Second, i f anisotropic performance across the visual field does occur as a function of endogenous attention orienting, what might this imply for theories of attention, particularly those that deal with the way that attention is distributed among locations and/or objects in the visual field? A great deal of research using covert orienting 1 A version of this chapter has been submitted for publication. Roggeveen, A.B. , & Ward, L . M . (submitted). Anisotropy of covert endogenous orienting across the visual field. 4 paradigms has been conducted to better understand how attention operates. This work has addressed many questions underlying how attention functions in the visual field, including attempts to understand how attention is distributed and divided among a set of attended locations (e.g. Gobell, Tseng, & Sperling, 2004). Reaching an understanding of how well attention can be allocated as a function of location in the visual field contributes to what we know about the mechanism of visual attention more generally. 2.1.1 Perceptual anisotropy Before we consider attentional anisotropy, it is important to understand in detail the ways in which, because of the design of the visual system, perceptual performance is not equivalent at all locations in the visual field, presumably independent of attention. There are four overarching patterns of anisotropic perceptual performance: eccentricity anisotropy, horizontal-vertical anisotropy, vertical anisotropy, and right-left anisotropy. Eccentricity anisotropy refers to the fact that as the distance (eccentricity) of a target from fixation increases, the ability to perceive that object decreases in a linear fashion (e.g. Anstis, 1974). This decrease in acuity is due to the cortical magnification factor: the amount of cortex dedicated to processing at the fovea is significantly greater than the amount of cortex dedicated to processing more peripheral locations (e.g. Cowey & Rolls, 1974). This reflects the fact that there are more ganglion cells allocated to inputs from cones and rods closer to the fovea than in the periphery (Curcio & Allen, 1990; Wassle, Grunert, Rohrenbeck, & Boycott, 1989; Weymouth, 1958). Receptive fields also become larger in the periphery, with the size of the receptive field increasing linearly as eccentricity from fixation increases (e.g. Hubel & Wiesel, 1960). The effects of eccentricity anisotropy on the ability to perceive a stimulus can be compensated for by 5 increasing the size of stimuli in the periphery (e.g. Carrasco & Frieder, 1996; Levi, Klein & Aitsebaomo, 1985). Scaling, however, does not entirely compensate for differences in temporal resolution between the periphery and the fovea. Carrasco, McElree, Denisova, and Giordano (2003) found that stimuli of the same size presented either in the periphery or at fixation are processed more quickly in the periphery. This effect was maintained even when the stimuli in the periphery were magnified to compensate for the cortical magnification factor. As with the decrease in perceptual ability farther from fixation, this temporal advantage is also thought to originate in the neural architecture of the retina, arising from differences between the magnocellular and parvocellular pathways. The magnocellular (M) pathway has greater numbers of ganglion cells in the periphery than the fovea, with increasing numbers as eccentricity increases (Azzopardi, Jones, & Cowey, 1999), and is known to respond more quickly to stimuli than does the parvocellular (P) pathway (Lamme & Roelfsema, 2000). Overall, acuity decreases and processing speed increases as the eccentricity from fixation increases. Horizontal-vertical anisotropy refers to better performance of perceptual tasks on the horizontal than on the vertical meridian of the visual field, and it is a second way in which perception varies according to visual field location (e.g. Cameron, Tai, & Carrasco, 2002; Carrasco, Talgar, & Cameron, 2001; Rovamo & Virsu, 1979). The horizontal and vertical meridians are the cardinal axes which intersect at fixation: two imaginary lines, one parallel and one perpendicular to the horizon. As an example of preferential processing on the horizontal meridian, contrast sensitivity falls off as eccentricity increases along the vertical meridian more rapidly than it does than along the 6 horizontal meridian, with the larger effects occurring at higher spatial frequencies (Rijsdijk, Kroon, & van der Wildt, 1980). More generally, the eccentricity effect is greater on the vertical meridian than the horizontal meridian (Carrasco, Williams, & Yeshurun, 2002). Information accrual is also fastest along the horizontal meridian, with intermediate levels of performance at intercardinal locations, and slowest performance on the vertical meridian (Carrasco, Giordano, & McElree, 2004). Other evidence for the asymmetry of visual perception on the horizontal and vertical meridians is found in work exploring effects of perceptual filling-in, where a peripheral target disappears into the background after a period of fixation. Perceptual filling-in occurs more quickly on the vertical than on the horizontal meridian (Sakaguchi, 2003) when target eccentricity is maintained. As in eccentricity anisotropy, the perceptual advantage of the horizontal meridian over the vertical meridian arises from the design of the visual system. In humans, the density of retinal ganglion cells with receptive fields on the vertical meridian is only 0.7 times that on the horizontal meridian (Curcio & Allen, 1990), allowing for greater neural representation of stimuli on the horizontal meridian. The third perceptual anisotropy is vertical asymmetry, which refers to the fact that performance is better in the lower than in the upper visual field (e.g. Previc, 1990; 1998). Previc (1990; 1998) has argued that the difference in processing between the upper and lower visual fields reflects specialization in dealing with different types of tasks. Principally, the lower visual field is used most extensively for tasks of perception occurring close to the observer, whereas the upper visual field is used most for stimuli that are farther away. The results of many studies support this dichotomy: the preference of the visual system for the lower visual field over the upper visual field has been found 7 in many perceptual tasks that require perceptual processing done close to the observer, viz. illusory contours (Rubin, Nakayama, & Shapley, 1996), achromatic motion processing (Edwards & Badcock, 1993), chromatic motion processing (Bilodeau & Faubert, 1997), and line length estimation (Fukisama & Faubert, 2001). Spatial resolution (Talgar & Carrasco, 2002) and contrast sensitivity (Carrasco et al. 2001) are also better in the lower visual field than in the upper visual field. Vertical asymmetry too can be explained by the underlying neural architecture of the visual system. Increased neural representation of the lower visual field, as with horizontal-vertical anisotropy, starts at the retina, with greater density of ganglion cells and cones in the area of the retina responsible for processing the lower visual field than that for the upper visual field (Perry & Cowey, 1985). And just as there is more cortex dedicated to processing of the fovea, the lower visual field is also represented to a greater degree in both cortical and subcortical structures (e.g. Connolly & Van Essen, 1984; Maunsell & Van Essen, 1987; Tootell, Switkes, Silverman, & Hamilton, 1988; Van Essen, Newsome, & Maunsell, 1984). Neuroimaging studies have also shown that there is a greater neural response to stimuli presented in the lower than the upper visual field, including results from fMRI (Chen et al., 2004) and M E G (Portin, Vanni, Virsu, & Hari, 1999). There also appear to be perceptual advantages for specific areas within the upper and lower visual fields. Several studies have shown that the advantage of performance in the lower visual field over the upper visual field is restricted to the vertical meridian. (vertical meridian asymmetry; Cameron, et al., 2002; Carrasco et al., 2001; Carrasco, Williams, & Yeshurun, 2002; Talgar & Carrasco, 2002). In some cases, the difference 8 between the upper and lower visual fields is only significant on the vertical meridian itself, with faster information accrual in the lower visual field (Carrasco et al., 2004). Liu, Heeger, and Carrasco (2006) showed that this behavioral asymmetry is reflected in a greater neural response to stimuli presented on the lower half of the vertical meridian. This brings us to the fourth perceptual anisotropy: right-left anisotropy. This anisotropy has generally been credited to differential processing of stimuli by the two hemispheres (e.g. Kinsbourne, 1970). The left hemisphere - responsible for processing the right visual field - is better at processing the local structure of stimuli (e.g. Fink et al., 1996) and higher spatial frequencies (e.g., Kitterle, Christman, & Hellige, 1990; Proverbio, Zani, & Avella, 1997). Conversely, the right hemisphere - responsible for processing of the left visual field - is thought to be superior for processing of lower spatial frequencies (e.g., Kitterle, Christman, & Hellige, 1990; Proverbio, Zani, & Avella, 1997) , and analysis of global structure (e.g. Fink et al., 1996). In keeping with preferential processing of lower spatial frequencies, other work has proposed that the right hemisphere is also more efficient at processing blurred and degraded nonverbal stimuli (Michimata & Hellige, 1987). Therefore, certain tasks appear to receive preferential treatment depending upon the visual field responsible for their processing. Fukisama and Faubert (2001), in addition to a lower visual field advantage, found that observers tended to estimate line length to be longer in the right visual field than the left. Words are also processed more quickly in the right visual field than the left (Hagenbeek & Van Strien, 2002). Overall^ these results indicate that preference for the right vs. the left visual field (or vice versa) are heavily 9 task-specific, relying on preferential hemispheric processing of stimuli, rather than anisotropies in neural representation of either the right or left of the visual field. 2.1.2 Attentional Anisotropy Covert attention can be oriented either endogenously, in a manner that is voluntary, or exogenously, where attention is drawn to a location in an involuntary or automatic way (e.g. Jonides, 1981). Endogenous orienting is a process where the observer chooses to shift their attention to a particular object or location. In many experimental paradigms, this shift is prompted by an indirect, usually central, cue, such as an arrow pointing to a to-be-attended location. This type of orienting response is deployed relatively slowly, moving from an initially broad focus to a narrow one, specific to what is being attended, with the narrowest attentional focus achieved at about 500 ms after the onset of the cue (Shepherd & Muller, 1989). Conversely, exogenous orienting is elicited by a direct, usually peripheral, cue, such as a sudden-onset stimulus at the to-be-attended location (e.g. Yantis & Jonides, 1990) . Exogenous orienting is deployed more quickly than endogenous orienting - in many cases, with its greatest effects between 50 and 100 ms after cue onset (e.g. Wright & Ward, 1998). Its impact on performance, however, also decays more quickly than that of endogenous attention, beginning to deteriorate about 100 ms after the presentation of the stimulus that summoned it, unless maintained endogenously (e.g. Cheal & Lyon, 1991) . The two types of orienting are thought to be mediated by separate attentional systems (e.g. Berger, Henik, & Rafal, 2005; Briand & Klein, 1987), although there is probably some sharing of brain regions between the two systems (Wright & Ward, in press). 10 2.1.2.1 Evidence from exogenous orienting. Most of the research conducted thus far that has explicitly explored the anisotropy of attention orienting has focused on involuntary shifts of attention. Given the perceptual anisotropies discussed earlier, it follows that attention might have similar restrictions in its functionality across the visual field. In order to separate the restrictions of perception from those of attention, it is necessary to equate the stimuli to be detected or discriminated for perceptibility at all relevant locations in the visual field. This may be partially achieved by magnifying stimuli in a linear fashion as eccentricity from fixation increases (e.g. Anstis, 1974), or by presenting all stimuli at the same eccentricity from fixation, but in different visual fields. After equating for stimulus discriminability, several studies have found that the allocation of exogenous attention impacts performance differently depending upon where in the visual field the target appears. In many cases, the effects of attention parallel the perceptual anisotropies described above. For instance, exogenous covert attention has been found to increase the level of discriminability to a greater extent as the target eccentricity from fixation increases (Carrasco, Williams, & Yeshurun, 2002), in a sense reversing the perceptual eccentricity anisotropy. Interestingly, the speed of processing is not impacted in the same way, as an exogenous cue has the same effect on the speed of processing of stimuli when it is presented at different eccentricities (Carrasco, Giordano, & McElree, 2006). Exogenous covert orienting also has been shown to speed the accrual of information about a target to a greater degree on the vertical meridian, particularly directly above fixation, than on the horizontal meridian, and to an intermediate degree at the intercardinal locations (Carrasco et al., 2004). This effect on information accrual parallels the horizontal-vertical meridian anisotropy described above. Both of these 11 studies indicate that attention may operate to effectively enhance processing to the greatest degree where it is needed the most: where processing is otherwise impaired compared to other visual field locations because of perceptual anisotropies. However, there are instances where exogenous orienting has the greatest attentional effect in areas of perceptual advantage. Gawryszewski et al. (1987) found that observers were better able to detect a sudden-onset target presented in the lower visual field than in the upper visual field, as well as faster at detecting targets in the right visual field than in the left. As the sudden onset would have elicited an exogenous shift of attention to that location, these differing reaction times probably indicate more efficient processing in the lower visual field as a result of faster exogenous attention orienting to those targets. Evidence is mixed, however, concerning whether the effects of exogenous covert attention orienting are anisotropic. Many experiments exploring exogenous covert attention have found that there are no differences in performance due to attention orienting at different locations in the visual field. Talgar and Carrasco (2002), for instance, failed to find a vertical meridian asymmetry in involuntary attention orienting. More generally across visual field locations, Carrasco and colleagues (e.g. Cameron et al. 2002; Carrasco, 2001) have reported that the impact of exogenously attending to various visual field locations had the same impact, preserving the anisotropic pattern of performance found as a result of perceptual anisotropies. 2.1.2.2 Evidence from endogenous orienting. Might the effects of endogenous orienting map onto the effects found in the exogenous orienting literature? Research concerning the anisotropy of the effects of endogenous orienting to various visual field locations has shown variable results. In the most specific measurement of the 12 effects of endogenous orienting to different visual field locations conducted thus far, Mackeben (1999) asked observers to shift their attention to one of eight possible cued locations, positioned on an imaginary circle with a radius of 7.5° of visual angle and centered at fixation. When a cue was presented, it was always 100% valid, or in the same location as the target; otherwise, no cue was presented. Overall, reflecting previous findings with perceptual differences across the visual field, performance in the absence a cue was highly dependent upon the location of the target: observers performed the letter identification task more accurately on the horizontal meridian than at any other location in the visual field. Mackeben (1999) showed that the impact of shifting attention to a cued location appeared to be nearly equivalent across attended locations, with no particular advantage to one region of the visual field over another. However, orienting attention to the cued location also reduced the amount of variability of response, showing that overall, attention had a very large effect on response accuracy. Altpeter, Mackeben, and Trauzettel-Klosinski (2000) replicated these findings using the same task but with Snellen Es as the targets, showing again that performance was best across the horizontal meridian. Because both Mackeben (1999) and Altpeter et al. (2000) controlled for visual factors, presenting stimuli well above perceptual thresholds, they concluded that any performance anisotropy was due to the effects of sustained attention: therefore, the effect of endogenous orienting is anisotropic across the visual field. 2.1.2.3 Upper vs. lower visual field anisotropy. Neither Mackeben (1999) nor Altpeter et al. (2000), however, found that there was a consistent advantage of attention to the lower visual field over the upper visual field; the lower or upper field advantage varied from observer to observer. He, Cavanagh, and Intriligator (1996), in 13 contrast, found a more general lower field advantage for selecting a target (a tilted grating) that was crowded among distractors, when targets were presented either 20° of visual angle above or below fixation. Intriligator and Cavanagh (2001) also reported a lower visual field advantage for selecting a target location for further processing, indicating that attention may have a finer grain in the lower visual field. Further support for a lower visual field attentional advantage has come from research showing that there are greater effects of attention in the lower visual field in a cuing task (Losier & Klein, 2004), and that shifting covert attention within an object to the lower visual field shortens reaction times to a greater extent than when attention is shifted to the upper visual field (Ashkenazi & Marks, 2004). More recently, however, evidence has favored a lack of lower visual field advantage in attention-demanding tasks which have asked observers to utilize both endogenous and exogenous components of attention. Michael and Ojeda (2005), presenting stimuli randomly from trial to trial in either the upper or lower visual fields, found no significant difference between the visual fields in the ability to identify the orientation of a briefly presented (150 ms) target among distractors. Rather, Michael and Ojeda (2005) found only a general decline in performance with increasing target-distractor similarity. Though the stimulus was presented quickly, an endogenous shift of attention to the visual field in which it was presented was required in order to search the display for the target stimulus among the highly similar distractors. In a similar paradigm where the stimulus display was presented for 280 ms, Levine and McAnany (2005) concluded that better performance in the lower visual field was not likely to arise from attentional factors, as performance did not vary as the task became more demanding. 14 Given the disagreement in these findings, the question of whether there is a lower visual field advantage in endogenous attention orienting deserves further investigation. 2.1.2.4 Right-left attentional anisotropy. As is the case for eccentricity and upper-lower visual field anisotropies, evidence for advantages to attending to the right visual field over the left, the left visual field over the right, or no advantage at all are mixed. The strongest suggestion that there is a right-left anisotropy in attention allocation has come from neuropsychological research. Hemineglect, a neurological condition where one side - typically the left side - of the visual field is systematically not perceived by patients with parietal lobe lesions, has offered some insights into the role of the separate hemispheres in how attention is oriented in the visual field (Heilman, Watson, & Valenstein, 1993). One task routinely used to assess the magnitude of neglect is a line bisection task, where the patient is asked to draw a line at the center point of a long line. Patients with damage to the right posterior parietal cortex, who "neglect" the left side of the visual field, tend to bisect lines to the right of the actual midpoint, as if they simply do not perceive the left half of the line. This pattern of performance, as well as a great number of other cognitive tasks, is thought to reflect a deficit in attentional allocation to the left side of visual space (e.g. Heilman & Valenstein, 1979; Kinsbourne, 1970). This deficit has been credited to either inattention to the left half of space (Heilman & Valenstein, 1979), or to excessive attention to the right side of space (Kinsbourne, 1970). Such findings offer an attentional explanation for the perceptual preferences shown by each hemisphere. Work with hemineglect patients has revealed that there is hemispheric specialization in the level of detail that is attended to on that side of space: the right hemisphere (left visual field) is responsible for attention to global aspects of a 15 stimulus, and the left hemisphere (right visual field) is responsible for attention to the local aspects of a stimulus (Robertson et al., 1988). Further evidence that the right and left hemispheres attend in different ways to the visual field has been demonstrated by Mangun and colleages (1994), who showed that while the left hemisphere allocated attention to the right visual field, the right hemisphere appeared to attend to both halves of the visual field. In adults without neurological damage, preferential processing of words presented to the right visual field could, therefore, be explained by differences in attentional selection of the stimuli within that hemifield. Lindell and Nicholls (2003) showed that cuing the beginning or ending or a word has little effect on the ability to process it when presented in the left visual field; however, when words in the right visual field were cued in the same manner, the location of the cue relative to the word had a significant effect. These results reflect the tendency of the right hemisphere to attend to global structure of the stimulus. The preference of the left hemisphere for attending to local stimulus attributes can also account for the previously mentioned evidence for preferential perceptual processing on the right side of visual space. Greater efficiency of processing blurred or degraded nonverbal stimuli in the left visual field (Michimata & Hellige, 1987) could be considered to be based on greater effects of attention in a task requiring discrimination of a noisy stimulus (Dosher & Lu, 2000) - taking advantage of the global processing of the stimulus afforded by the right hemisphere. Interestingly, research on hemineglect has also shown support for differing attentional processing in the upper and lower visual fields: Ladavas, Carletti and Gori (1994) showed that patients with hemineglect showed greater 16 deficits in performance in the lower visual field in the neglected hemifield compared to the upper half of that hemifield. Normal observers will also reliably showpseudoneglect, or a tendency to bisect lines to the left of the actual midpoint (e.g. Bisiach, Capitani, Colombo et al., 1976). A tendency to bisect a line farther to the left could be credited to lesser allocation of attention to the right side of visual space. Interestingly, McCourt and Garlinghouse (2000) found that pseudoneglect occurred to a greater degree in the upper visual field than the lower visual field, indicating an uneven distribution of attention across the right and left visual fields above and below the horizontal meridian. This result was also interpreted by McCourt and Garlinghouse (2000) to support He et al.'s (1996) finding that attention in the lower visual field has a finer grain than in the upper visual field. There is also evidence for a lack of right-left attentional anisotropy, however. Many authors (e.g. Verfaellie, Bowers, & Heilman, 1988; Kraft et al., 2007) have reported no differences between attention effects in the right and left visual fields. Finding a difference may rely on the use of distracting stimuli: Evert et al. (2003) found that, in a cuing paradigm, visual field differences only appeared when a distracting item that was not a possible target appeared in the uncued location. Reaction times were slower in the right visual field than the left under conditions with a distractor; under conditions without a distractor, performance between the visual fields was equivalent. Similarly, Michael and Ojeda (2005) found a right visual field advantage in sensitivity to the target stimulus when the target was moderately similar to the distracting stimuli; this effect reversed when the target was highly similar to the distractors, showing a slight advantage for the left visual field. However, no advantage to either side was found under 17 conditions of low similarity. Thus, the influence of a potential distractor may occur under both conditions where the target location is cued, and in attentionally demanding situations. From all of these findings, only one point is clear: that further exploration is required to establish the topography of the benefits of voluntarily orienting attention across the visual field. The research presented here is designed to answer the question of whether differences in performance as a result of attention orienting vary according to visual field location, and whether any differences follow a pattern of results similar to previous findings in the attention literature, both endogenous and exogenous, and also how any attentional asymmetries are related to the perceptual anisotropies discussed earlier. 2.1.3 Attention distribution Attentional anisotropies or not, how does attention select specific locations in the visual field for further processing? Consider how attention is used to monitor the visual scene in an everyday situation. There are many degrees to which one can select a location in visual space for additional processing. For instance, while reading a book in a quiet office, attention shifts from word to word (in this case, with eye movements), with little attention allocated to other regions of the visual field. However, if reading the same book at the kitchen table, attention may select what is being read as well as monitor the activity of the pots on the stove. Attention is flexible enough to allow you to read the book, while maintaining vigilance to be ready to jump up to tend to a pot that is ready to boil over. Inherent in this example is the question of how attention may be distributed across the visual field to one or more locations. In experimental paradigms, this question has 18 been explored by manipulating the certainty with which a precue indicates that a target will appear in a certain location. Traditional cuing tasks where an observer is precued to attend to a certain location in the visual field are typically designed with the cue indicating with less than 100% certainty where the target will appear. In other words, the cue will not validly cue the subsequent target location on every trial. Consequently, the manner in which attention must be allocated across possible target locations must change; rather than selecting the cued location to the greatest extent - or, at least, to the greatest extent required by the task - the observer must hedge their bets, and distribute some of their attention to the other possible target locations. Under conditions where the observer can select, with 100% certainty, a particular location in the visual field for further processing, attention does not need to be distributed elsewhere in the visual field in order to successfully perform the task. Returning to the example, if there were two pots on the stove, the pot that was more likely to boil over would be the one where the greatest amount of monitoring would occur. However, i f it was guaranteed that the only activity on the stove would happen in one pot, and not the other, attention could be given entirely to the active pot. Jonides (1983) indeed showed that, under conditions when a precue is invalid, i.e., the subsequent target occurs at an uncued location, attention is then allocated in a diffuse way across all possible uncued locations. As a result of these findings, Jonides (1983) suggested that attention alternates between two states: a focused state, where attention is allocated only to a cued location, and a diffuse state, where attention is allocated across all possible target locations. Precuing the target location allows for attention to be focused at the cued location; were the target not to appear at the cued 19 location, attention would revert to a diffuse state in order to locate and process the target at an uncued location. Similarly, Eriksen and Yeh (1985) showed that under conditions where the validity of the cue is manipulated, performance at both validly and invalidly cued locations varies as a function of the likelihood that the target will appear in the cued location: as cue validity increased, the benefits of attending to validly-cued locations increased, as did the costs of detecting targets that occurred at invalidly-cued locations. Given these results, Eriksen and Yeh (1985) concluded that rather than there being two distinct modes of attentional allocation - focused and diffuse - there were rather "two poles on a continuous range of attentional capacity distribution in the visual field" (p. 595). Attention, rather than a spotlight of unchangeable size, is instead a zoom lens, where the area of the attended region of space could be altered to fit the requirements of the task. Important in this model is the idea that attentional resources are limited - that the larger the space included in the lens, the less efficacious processing will be within it. The zoom lens is also unitary; rather than a split of attention between visual field locations, under the zoom lens model, attention is simply spread as widely as the task demands, capturing whatever stimulus falls within its span. A different approach to understanding how attention is allocated across the visual field is the activity distribution, or gradient model (LaBerge, 1995; LaBerge et al., 1997). Under this model, attention is again allocated in two ways: first, as a state of general readiness to attend to all available target locations in a display, and second, as a mechanism that specifically selects a location in the visual field in response to a cue. However, rather than switching between two modes of attention, as proposed by Jonides 20 (1983) and Eriksen and Yeh (1985), these two modes of attention may co-exist while viewing a display. As in the other theories, targets that appear in the selected location are processed more efficiently than targets that appear in other locations. However, rather than creating a window of attention within which anything is processed, the activity distribution model postulates a gradient of attention that falls off around the cued location. Due to this gradient and the underlying general readiness to attend to all possible target locations, it is possible to respond accurately to a target presented outside of the selected location.2 Varying the likelihood of the target appearing at the cued location can alter the amount of processing resources allocated to the selected location, changing the efficacy of processing of the other locations in the visual field. Two separate areas of research lend support to the activity distribution model. First, there is mounting evidence that attention is flexible enough to be effectively split between locations in the visuaf field (e.g. Awh & Pashler, 2000; Hahn & Kramer, 1998; Kramer & Hahn, 1995), though limited by the type of intervening stimuli (e.g. whether the distracting items onset suddenly; Hahn & Kramer, 1998). Despite these limitations, the division of attention among visual field locations appears to be incredibly flexible. Gobell, Tseng, and Sperling (2004) asked observers to divide their attention across a display of alternating red and green stripes that were of varying spatial frequency. Their results showed that observers were able to divide their attention between the stripes, 2 The idea of an attentional gradient is that attention is allocated preferentially to a cued location and its impact on performance deteriorates with increased distance from the attended location. Many studies (e.g. Downing & Pinker, 1985; Handy, Kingstone, & Mangun, 1996) have specifically explored the spatial extent of the attention gradient and its impact on responses to targets located at various distances from the central attended location. In the work described here, the effect of the attentional gradient as it declines away from the attended location will not be discussed; rather, what is at issue here is how the way attention is allotted (distributed) between potential target locations in the display affects response time and accuracy at various locations in the visual field. Future work could explore the interaction between visual field effects (both perceptual and attentional), such as those presented here, and the shape of the attentional gradient. 21 selecting only those of a particular color, although with decreasing efficacy with higher spatial frequencies. These findings demonstrate that attention can be divided among locations in a fairly complex way in accordance to the task demands - though not without upper limits on its divisibility. The work of Gobell and colleagues (2004) also indicates an interaction between the manner in which attention is distributed across possible target locations, and visual field location itself. Gobell and colleagues (2004) found that observers were better at dividing their attention when the stripes were oriented horizontally, rather than vertically, indicating a preference for allocating attention either on or parallel to the horizontal meridian across the visual field. Similarly, Ashkenazi and Marks (2004) found that, within an object, attention could be guided more effectively when objects were oriented horizontally than when the objects were oriented vertically. These results reflect the above-mentioned attentional advantage on the horizontal meridian, and lend additional support to the hypothesis that attentional selection varies according to visual field location. The second area of research that lends credence to the idea that attention is allocated more according to the activity distribution model than to Jonides' model of attentional mode switching, or the zoom lens model (e.g. Eriksen & Yeh, 1985), is work that shows that i f there is more than one possible target location, it is unlikely that attention is shifted back and forth between those locations. Instead, attention seems to be split in a way to reflect the likelihood that the target will appear in one location over another. Increasing the likelihood that a target will appear in a cued location, without changing anything about the underlying stimulus, will decrease reaction time (RT) to a 22 target appearing at the cued location, and increase RT to targets appearing in an uncued location (e.g. Jonides, 1980). Johnson and Yantis (1995), by manipulating the probability that a target would appear in a cued location (either 100% probability, 50% probability, or a neutral condition), showed that the pattern of performance on the 50% probability condition was unlikely to reflect a shifting between possible target locations. Further, Gottlob, Cheal, and Lyon (1999), using measures of accuracy rather than RT, demonstrated that the attention operating characteristic associated with varying levels of validity and instruction-based allocation of attention implicated a strategy where the observers shared their attention between cued and uncued locations in a manner that reflected the probability that the target would appear in the cued location - not a strategy where attention was shifted between possible locations. Both of these findings suggest that attention may be allocated in a manner where the observer can choose to preferentially attend to one location, while still monitoring another (or several others) for possible target appearance. More interesting, perhaps, is the fact that it appears that the distribution of attention to non-cued locations is specific to visual field locations. Turk-Browne and Pratt (2005) found using an RSVP task that response time to a second target presented in a central location on the vertical meridian in the upper visual field after the first target was presented in the same location was significantly faster than when the second target appeared to the left of the vertical meridian. Interestingly, when the second target was presented to the right of the vertical meridian - an additional case of the target not being presented in the previous target location - the relevant performance difference was not significant. This result indicates that attention is more easily reallocated to the upper right 23 visual field than to the upper left visual field under conditions where the observer has been asked to spread their attention across an array of possible target locations, but shift their attention within that attentional gaze. Although the results just discussed are suggestive that there might be systematic anisotropies in endogenous attention orienting, there have been no systematic explorations of attentional anisotropy of which we are aware. The research presented in this paper thus seeks to address two questions. First, is attention allocated equivalently well to all locations in the visual field? After accounting for visual anisotropies, does attention have different effects at different visual field locations? The work presented here is the first to make a more detailed measurement of the efficacy of endogenous attention orienting across many locations in the visual field, comparing performance on the basis of all previously documented instances of anisotropy: eccentricity, upper vs. lower visual field (including vertical meridian anisotropy), and right vs. left visual field. Second, the work discussed here also explores the effects of manipulations of validity on the way that attention is distributed across the visual field. What do these results indicate about how visual field location and cue validity interact with one another to restrict the way that attention is allocated across the visual field - and distributed between visual field locations? 2.2 Experiment 1 In order to measure the benefits of covert, endogenous attention at different visual field locations, observers were asked to attend either to one of 32 possible cued locations where a target letter would appear with 100% certainty, or to spread their attention across 24 all possible target locations, encompassing 24 degrees of visual angle in diameter across the visual field. 2.2.1 Methods 2.2.1.1 Observers. Twenty observers (6 males and 14 females) participated in the experiment (median age: 21.5, age range 17 - 27). Two observers reported being left handed; all other observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioural Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. 2.2.1.2 Materials. Observers sat in a comfortable chair with their chin in a chinrest located 53 cm in front of a computer screen. The display was shown on a 19.5-inch monitor with a resolution of 1280 by 1024 pixels and a refresh rate of 85 Hz. Observers were given both written and verbal instructions about the procedure of the experiment. After the observer reported feeling comfortable with the task when shown several practice trials, the experiment began. A schematic of an example trial is shown in Figure 2.1. At the beginning of each trial, a fixation cross and 32 placeholder squares appeared at the center of a black screen. The placeholder squares were located at eight radial locations around an imaginary circle - North (0°); Northeast (45°); East (90°); Southeast (135°); South (180°); Southwest (225°); West (270°); and Northwest (315°) -and at 1°, 4°, 8°, and 12° of visual angle from fixation. The placeholder squares were magnified in a linear fashion (cortical magnification factor = 2.3; Anstis, 1974) so that their visibility would be equivalent at each location. 25 The observer was then presented with one of two types of trials: a cued trial, or an uncued trial. On both types of trials, the observer was asked to make a letter discrimination. On cued trials, the observer was informed with 100% certainty where the target letter would occur. The inside of the placeholder square where the target was to appear turned red, stayed red for 550 ms, and then faded back to black over 510 ms. Observers were instructed to shift their attention, without moving their eyes, to the cued location. While a peripheral cue in the sense that it appeared at the target location, the time course and fading of the cue required an endogenous shift of attention to the cued location. On uncued trials, the observer did not know where on the screen the letter would appear. Instead, all of the white placeholder squares turned red, stayed red for 550 ms, and then changed back to white over 510 ms. This interval was chosen to maintain the same timing information about the subsequent appearance of the target. On these trials, observers were asked to attend to all locations in the display equally, as the target letter could appear anywhere. Whether the trial was cued or uncued varied randomly from trial to trial. Two hundred to 400 ms after the cue faded, the target was presented for 150 ms. On cued trials, the letter appeared in the cued location; on uncued trials, the letter appeared randomly in any of the 32 possible locations. The target could be either the letter M or the letter N . The observer's task was to identify the letter presented. Like the placeholder squares, the target letters were magnified, using 2.3 as the cortical magnification factor (Anstis, 1974) so that the targets would be equally visible at all target eccentricities. 26 After the target presentation, the observer made a response, pressing the "/" key on the keyboard with their right hand to respond that they saw an " M " , or the "z" key on the keyboard with their left hand to respond that they saw an " N " . Once the observer had made a response, feedback was presented by either brightening the fixation cross to indicate a correct response, or by replacing the fixation cross with a minus sign. Feedback was presented for 1000 ms, giving ample time for observers to blink between trials. Each observer participated in 16 blocks of 96 trials each; a target appeared in each of the 32 locations in both cuing conditions for a total of 24 trials. Blinks and eye movements were monitored by both the horizontal and the vertical electro-oculograms (EOGs), which were also sampled at 250 Hz. Impedances for EOGs were kept below 20 kQ. Trials on which observers made an eye movement between cue onset and response, those on which they made an error, and trials with reaction times less than 150 ms or greater than 760 ms, were excluded from the analysis. These restrictions left approximately 42 trials per location per observer for the subsequent analyses. 2.2.2 Results 2.2.2.1 Reaction Time. In order to assess whether the effects of endogenously orienting attention to the various visual field locations was anisotropic, a 3-way (Cue condition x Eccentricity x Radius) repeated measures analysis of variance (ANOVA) was performed on the reaction time (RT) data. For instances where the Mauchly's test of sphericity was significant, the Greenhouse-Geisser correction of the degrees of freedom was used to control for any increased possibility of Type 1 error. Effect sizes are reported as partial eta squared (partial r)2) in order to assess the variation that may be attributed to each factor within the A N O V A , while also excluding the other factors from the non-27 2 2 error-related variation (e.g. Cohen, 1973). Partial r\ differs from n in that the values computed within an A N O V A may sum to a number greater than 1 (Cohen, 1973). Response times were shorter for trials were the target location was cued, compared to conditions where it was not cued (Cue condition: F(\, 19) = 62.76; p < 0.001; partial n = 0.77). Speed of response was also affected by the location in the visual field where the target appeared: as distance from fixation increased, overall RT increased as well (Eccentricity: F(3, 57) = 38.0; p < 0.001; partial n 2 = 0.67; see Figure 2.2). The impact of cuing the target location was not equivalent at all target locations. A significant interaction between Cue condition and Radial location (F(7, 133) = 3.20;p < 0.004; partial n 2 = 0.14; See Figure 2.3) indicated that the effects of cuing were greatest for radii extending into the lower visual field (SE, S, SW), showing larger cue effects than those extending into other parts of the visual field. A preference for responding to stimuli presented in the lower visual field was also reflected in a significant main effect of Radial location (F(7, 133) = 7.5; p < 0.001; partial n 2 = 0.28). To further explore where the cuing effects had the greatest - and least - effects in the visual field, five planned comparisons were performed on Cue condition difference scores (uncued - cued) to directly address the known perceptual and attentional anisotropies: horizontal-vertical anisotropy; vertical meridian anisotropy; right-left anisotropy; upper-lower anisotropy, and eccentricity anisotropy. Planned comparisons were performed in conjunction with the omnibus A N O V A in order to directly assess in a more focused way whether the effects of attention mirrored those of previously documented perceptual anisotropies. Figure 2.4 shows the average difference scores at all eight radial locations collapsed across eccentricity. In order to evaluate the differences 28 between performance in the different portions of the visual field, planned comparisons of least squares means were performed. An adjusted significance /rvalue cut-off ofp = 0.01 was used for all planned comparisons to accommodate for possible familywise error due to multiple comparisons. A comparison between the difference scores on the horizontal meridian (E & W) and the vertical meridian (N & S) showed a significant difference (F(l,19) = 12.49;p < 0.01; partial n 2 = 0.36), with the largest differences between cued and uncued trials occurring on the vertical meridian. The difference between the upper visual field (NE, N , NW) and lower visual field (SE, S, SW): (F(l,19) = 6.57;p < 0.02; partial n 2 = 0.26) was marginally significant. However, there was no significant vertical meridian asymmetry (F(l,19) = 3.36;p > 0.08; partial n = 0.15), and no significant difference between the right and left visual fields (F(l,19) = 1.51; jo > 0.23; partial n = 0.07). Additionally, a comparison between eccentricities testing whether a linear relationship existed between eccentricity and attention effects was non-significant (F(l,19) = 1.51; jo > 0.23; partial n 2 = 0.02). 2.2.2.2 Errors. A 3-way (Cue condition x Eccentricity x Radius) repeated measures A N O V A was performed on the error rates. More errors were committed on trials without a cue (Cue condition: F(l,19) = 9.52;p < 0.007; partial r|2 = 0.33). The mean error rate was 4.32%. 2.2.3 Discussion The results of this experiment indicate that the effects of endogenous selection of a target location are anisotropic across the visual field: clearly the effect of endogenously attending to a location in the visual field is mediated by where that location is. The effect of attending to a location on the vertical meridian is significantly different from that on 29 the horizontal meridian, with the greatest impact of attention seen on the vertical meridian. Attention effects were also marginally greater in the lower visual field than in the upper visual field. However, no other significant attentional anisotropies (right vs. left; vertical meridian) were found. Because the size of the stimuli was scaled to equate for perceptual difficulties away from fixation, the occurrence of a significant main effect of eccentricity may be another indication that the effects of attention across the visual field are anisotropic. Because of the magnification we implemented the longer RT at greater eccentricities is unlikely to be accounted for entirely by perceptual factors. Rather, it is likely that attention had a varying impact depending upon the distance from fixation. However, because cuing condition did not interact with eccentricity, these variable effects of attention likely arise from exogenous shifts elicited by the sudden onset of the target letter. The amount to which endogenous orienting affects the performance differences across the different eccentricities will be further explored in Experiment 2. 2.3 Experiment 2 In Experiment 1, the observers either knew with 100% certainty where the target letter would appear, or spread their attention as equally as they could across all possible target locations. However, cuing paradigms have historically used cues with different likelihoods of indicating the target location in order to evaluate both the benefits of attending to a cued location, and the costs of attending away from a target location. In order to more accurately compare performance in Experiment 1 to the previous literature, we used a different validity distribution (70% validly-cued trials) in Experiment 2. Consistent results found between Experiments 1 and 2 should indicate a general 30 preference for attentional selection in some locations in the visual field. Differences between performance in Experiment 1 and Experiment 2, on the other hand, will address two issues. First, should the effects of increasing eccentricity on reaction time be different than those found in Experiment 1, the results would indicate that exogenous orienting alone cannot explain that effect. Rather, some effect of attention distribution dependent on cue validity would be implicated as well. Second, because this is solely a manipulation of the likelihood with which a target will appear at a cued location, and nothing else has been altered about the target display, differing interactions of Cue condition with eccentricity or radius between Experiments 1 and 2 will illustrate how visual field location interacts with the way that attention is allocated across the visual field. 2.3.1 Methods 2.3.1.1 Observers. Twenty observers (8 males and 12 females) participated in the experiment (median age: 20, age range 18 - 23). Four observers reported being left handed; all other observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioural Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. 2.3.1.2 Materials. A l l materials in Experiment 2 were identical to Experiment 1, with the exception of the way in which trials were cued. For all trials, the observer was cued to shift their attention endogenously to a particular location by the same box cue as in the cued condition in Experiment 1. However, this cue indicated with only 70% 31 certainty that the target letter would appear there. On the remaining 30% of trials, the target letter could appear at any of the other 31 locations on the screen. 2.3.2 Results 2.3.2.1 Reaction Time. As in Experiment 1, a 3-way (Cuing condition x Eccentricity x Radius) repeated measures A N O V A was performed on the reaction time data. As in Experiment 1, responses to targets presented at the cued location were faster than to targets presented at uncued location (Cue condition: F(l,19) = 123,64;p < 0.001; partial n = 0.86). Responses were also consistently slower to targets presented farther from fixation (Eccentricity: F(3,57) = 45.47; p < 0.001; partial n 2 = 0.76). In a different result than Experiment 1, a significant interaction between Cue condition and Eccentricity was found (F(3,57) = 3.35; p < 0.03; partial n 2 = 0.18; see Figure 2.5), showing that the impact of cuing the target location was greatest both closest (1°) and farthest (12°) from fixation. The varying performance across radial location was also significant (Radius: F(7,133) = 14.5;p < 0.001; partial n 2 = 0.52). As in Experiment 1, planned comparisons were performed on the RT difference scores (invalid - valid) from Experiment 2. Difference scores at the eight radial locations are depicted in Figure 2.6. A comparison of performance on the horizontal and vertical meridians, which was significant in Experiment 1, was not significant here (F(l,19) < 1.0;/? > 0.61; partial n 2 = 0.01). A significant difference between the upper and lower visual fields was also found (F(l,19) = 7.80;p < 0.01; partial i f = 0.29), with a significant difference between performance on the upper and lower halves of the vertical •y meridian (vertical meridian asymmetry; F(l,19) = 9.85;p < 0.005; partial n = 0.34). The 32 difference between performance in the right and left visual fields was not significant (F(l,19) < 1.0;/? > 0.97; partial n 2 = 0.00009). While there was no linear relationship between the difference scores at the four possible target eccentricities (F(l,19) < 1.0;/? > 0.75; partial n 2 = 0.005; see Figure 2.7), a Fisher Least Significant Difference post hoc comparison between the four eccentricities was performed, showing that the effects of attention at four degrees from fixation were significantly less than at 1 degree (p = 0.03) and at 12 degrees (/? = 0.016), and less at eight degrees than 12 degrees (p = 0.02). Interestingly, the eccentricity data significantly fit an a priori quadratic model (F(l,19) = 10.70;p < 0.004; partial n 2 = 0.36), showing that the cuing effects are greatest at 1 degree and 12 degrees. 2.3.2.2 Errors. A 3-way (Cuing condition x Eccentricity x Radius) repeated measures A N O V A was performed on the error rate data. The mean error rate was 4.10%. No significant main effects or interactions were found. Thus, speed-accuracy tradeoffs are unlikely to account for the RT differences. 2.3.3 Discussion In Experiment 2, the pattern of anisotropy appears to be somewhat different than the pattern found in Experiment 1. First, the omnibus A N O V A for all target locations indicated that Cue condition interacted significantly with Eccentricity, rather than Radius. As in Experiment 1, planned comparisons revealed a marginally significant difference between the effects of attention in the upper and lower visual fields, showing that attention had a greater impact in the lower visual field than the upper visual field. The difference between the upper and lower visual fields was significant on the vertical meridian, with a greater advantage to attending to the lower half. 33 More interesting, perhaps, is the relationship between Cuing condition and Eccentricity: as distance from fixation increased, the effects of cuing did not increase in a linear fashion, as would be predicted by research on exogenous orienting showing that the effects of attention increase with increasing distance from fixation. Rather, these results show that the effects of cuing were greatest 1 degree and 12 degrees away from fixation. What is important to keep in mind here is the comparison that is being made between validly and invalidly cued. In this experiment, the effects of attention that are being compared are between trials where the cue validly indicated where the target would occur and trials where the target appeared at a location away from the cued location. In order to effectively perform the task, it would not be possible to allocate all attentional resources to the cued location; instead, attention would have to be preferentially allocated to the cued location, but also distributed among the other possible target locations. The pattern of cuing effects related to eccentricity implies that the distribution of attention to the uncued locations was not isotropic: instead, observers centered any residual attention, or remaining attentional resource leftover after attending preferentially to the cued location, in a ring-shaped configuration to maximize their performance. There is clear evidence that attention is flexible enough to be allocated to a ring-shaped region (e.g. Eimer, 1999); the present results demonstrates that while preferentially attending to a single cued location, this can be done in order to maximize performance on invalidly-cued trials. A direct comparison between Experiment 1, where attention is allocated to the cued location knowing that the target will always appear there, and Experiment 2, where 34 attention is allocated to the cued location with the knowledge that there is the possibility that it will not, begins to create a picture of how attention is more generally allocated across the visual field. Observers were consistently 19 ms slower under conditions where attention was allocated in the manner demanded by Experiment 2. The comparison between the means of the four cuing conditions across both experiments is displayed in Figure 2.8. Performance, not surprisingly, is best when the target is cued with 100% certainty (Experiment 1). As cue validity decreased, performance suffered; when the cue indicated with 70% certainty where the target appeared (Experiment 2), the mean RT across all locations increased by 25 ms. For trials where there was no cue offered at all, requiring observers to distribute their attention as equally as they could across the possible target locations (Experiment 1), RT again increased overall, relative to the 70% valid condition, by 11 ms. Finally, performance on invalidly-cued trials was 25 ms worse than performance on trials with no cue at all. Taken overall, this pattern of results reveals that performance in a task where attention is spread across a wide span of the visual field, with many possible target locations, varies as a function of the likelihood that a cue indicates the subsequent target location. Clearly, performance when the target location is known with 100% certainty is better than when attention is spread across the visual field, as in Experiment 1. As certainty about target location decreases, so does performance. Comparing performance between the no cue (Experiment 1) and invalidly cued (Experiment 2) conditions shows that the effects of spreading attention endogenously across the visual field in order to respond most quickly to the sudden onset target are attenuated by allocating attention to a cued location. The attentional resources available 35 to be spread across the field were lessened by a preferential allocation of attention to the cued location. The most novel finding here, however, is that location in the visual field interacts with the way that the attention not allocated to the cued location is distributed across the visual field. The lesser cuing effect at four degrees under the condition where the observer is cued with only 70% certainty of the target's location, shows that the remaining attention is not allocated throughout the remainder of the visual field in an equivalent way. In fact, what appears to be occurring is that observers are preferentially spreading their attention in a ring-shaped distribution, in order to be closest to most possible target locations. Rather than creating a distribution which is centered at fixation, with greater attentional resources distributed from fixation outwards, attention seems to be distributed from the location most likely to be close to all of the other possible target locations in the display. Before reaching definitive conclusions from these comparisons between Experiments 1 and 2, however, we need to consider results from an experiment that more closely replicates the type of stimulus presentation used in many previous studies (see Wright & Ward, 1998), where the cued and uncued trials are presented alone within blocks, rather than randomly presented from trial to trial. 2.4 Experiment 3 In order to facilitate comparison to some previous work that has investigated whether endogenous orienting of attention is anisotropic, it is necessary to implement a design in which trials with a 100% valid cue and trials without a cue are presented in separate blocks of trials. Mackeben (1999) in particular used such a technique. His results 36 indicated that there were no interactions of visual field location with cuing condition; however, the consistent effects of visual field location on performance were interpreted to represent an attentional advantage for certain visual field locations. In contrast, in our Experiments 1 and 2 we did find interactions between cuing condition and visual field location. Mackeben (1999) also reported a greater degree of between-observer variability than we have seen in our experiments, especially for performance in the upper and lower visual fields. Such between-observer variability, and also the lack of interaction between cue condition and location, could arise from differences in the strategies observers employed in the blocked versus the randomized designs. Thus, we conducted Experiment 3 in order to ascertain, under our stimulus and task conditions, whether a blocked design would alter our results. In the uncued blocks of Experiment 3 we did not use a neutral cue, as we did in Experiment 1, but rather, presented no cue at all. Eriksen and Yeh (1985) reported that there was no behavioral difference between the use of a neutral pre-cue and no pre-cue at all in terms of providing a generalized alerting effect to the observer about the timing of the presentation of the target. Therefore, this methodological difference should have no significant effect on performance; differences between performance in Experiment 1 and Experiment 3 should arise entirely from the effortful nature of switching back and forth between allocating attention specifically to a target location when a single cue is presented, and allocating attention broadly across all target locations when the neutral cue is presented, in Experiment 1, compared to adopting a single strategy for an entire block of trials, in Experiment 3. 37 2.4.1 Methods 2.4.1.1 Observers. Twenty observers (6 males and 14 females) participated in the experiment (median age: 20, age range 18 - 24). One observer reported being left handed; all other observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. 2.4.1.2 Materials. A l l materials in Experiment 3 were the same as those used in Experiment 1, with two exceptions. Rather than presenting the cued and uncued trials randomly from trial to trial, as in Experiment 1, instead for half of the experiment, the target location was cued, using the same cuing technique described in Experiment 1. For the other half of the experiment, the target location was not cued. In the uncued half of the experiment, the target letter would appear 450 - 650 ms after presentation of the feedback from the previous trial to the observer. The order of the cued and uncued halves of the experiment was counterbalanced across observers to account for possible order effects. 2.4.2 Results 2.4.2.1 Reaction Time. A three-way (Cue condition x Eccentricity x Radius) A N O V A was performed for the RT data. There was not a significant difference between response time to targets presented when the location was precued and when it was not. However, response time did increase as distance from fixation increased (Eccentricity: F(3, 57) = 42.00; p < 0.001; partial r\ = 0.69), and performance was, overall, not 38 equivalent at all radial locations (Radius: F(7, 133) = 8.14;p < 0.001; partial n 2 = 0.38). No significant interactions were found in the data. Closer inspection of the RT data, however, revealed that some observers consistently showed a large cuing effect whereas others showed no effect or even a reverse effect, mirroring in this sense the data of Mackeben (1999). Mean cuing effects for each observer are presented in Table 1. We thus divided the observers into two groups using a median split, which resulted in a group of 10 observers who consistently showed cuing effects (average cuing effect: 41 ms), and a separate group of 10 observers who showed no or a reverse cuing effect (average cuing effect: -1 ms). We then analyzed the data of these two groups separately using the same 3-way (Cue condition x Eccentricity x Radius) repeated measures A N O V A . The A N O V A performed on the RT data from the group who did show a consistent cuing effect revealed a marginally significant difference between cued and uncued trials (Cue condition: F(l,9) = 3.91;p < 0.08; partial n = 0.30), with an average difference of 41 ms. The effect of precuing the target location differed with distance from fixation, as reflected in the significant interaction between Cue condition and Eccentricity (F(3,27) = 5.29;p < 0.005; partial n 2 = 0.37; see Figure 2.9). As in the previous experiments, as response time increased with increased distance from fixation (Eccentricity: F(3,27) = 20.28;p < 0.001; partial n 2 = 0.69), and performance differed at the eight possible radial locations (Radial location: (F(7, 63) = 5.48; p < 0.001; partial r\ = 0.38). No other interactions reached significance. In contrast, the A N O V A performed on the RT data from the group who did not show a consistent cuing effect did not show a significant difference between the two 39 cuing conditions, and no significant interactions between any of the factors. Consistent with previous findings, a significant main effect of Eccentricity (F(3,27) = 26.75; p < 0.001; partial n = 0.79; see Figure 2.10) reflected longer RTs to more distant stimuli, and a significant main effect of Radius (F(7,63) = 3.27; p < 0.006; partial n 2 = 0.32; see Figure 2.11) showed that responses were fastest in the lower right portion of the visual field. For the observers who did show a cuing effect, planned comparisons based on previous findings of anisotropy were performed on difference scores (uncued - cued). Paralleling the interaction between Cue condition and Eccentricity, there was a significant linear relationship between eccentricity and cuing effect (F(l,9) = 9.%\;p< 0.02; partial n = 0.52; see Figure 2.12): as eccentricity increased, so did the effect of cuing the target location. However, no significant differences were found between cuing effects between the upper and lower visual fields, the right and left visual fields, the horizontal and vertical meridians, or between the upper and lower halves of the vertical meridian. See Figure 2.13 for difference scores at the eight radial locations. In order to ensure that there were no effects of the order in which the observers participated in the two cuing conditions, a 4-way between-within observers A N O V A (Order x Cue condition x Eccentricity x Radius) was also performed on the RT data. Order interacted significantly with Radius (F(7,l 12) = 2.12;/? < 0.05; partial n 2 = 0.12), showing that observers who performed the half with the cue first performed better at all locations except S, SW, and W. Order did not significantly interact with any of the other factors, nor was the main effect of order significant. Significant main effects were found 40 only for Eccentricity (F(3,57) = 41.50; p < 0.001; partial n 2 = 0.72), and Radius (F(7,133) = 8.67; p < 0.001; partial n 2 = 0.35). 2.4.2.2 Errors. Separate 3-way A N O V A s (Cue condition x Eccentricity x Radius) were performed on the error data for the two groups of observers. For neither observer group were significant main effects or interactions found in the error data. The mean error rate was 3.62%. The group with consistent cuing effects had a mean error rate of 3.05%; the group without consistent cuing effects had a mean error rate of 4.19%. Thus speed-accuracy tradeoffs probably cannot account for the RT effects observed. 2.4.3 Comparison with Experiment 1 In order to evaluate the effect of maintaining a sustained attentional state for longer durations, a 4-way (Experiment x Cue Condition x Eccentricity x Radius) A N O V A was performed on the RT data for the observers who showed a consistent cuing effect in Experiment 3, and the RT data from Experiment 1. Results showed a significant interaction between Experiment, Cue condition, and Eccentricity (F(3,84) = 4.80; p < 0.004; partial n 2 = 0.15), because of an overall RT difference between the two experiments at target locations four and eight degrees from fixation. The interaction between Cue condition and Eccentricity was also significant (F(3,84) = 6.44;p < 0.001; partial n 2 = 0.19), as well as significant main effects of Cue condition (F(l,28) = 23.42; p < 0.001; partial n 2 = 0.46), Eccentricity (F(3,84) = 51.31;/? < 0.001; partial n 2 = 0.65), and Radius (F(7,196) = 10.0; p < 0.001; partial n 2 = 0.26) - all mirroring previous patterns of results. The main effect of Experiment was not significant; the difference between the two experiments' grand averages was only 6 ms. 41 2.4.4 Discussion By presenting the 100% cue trials and no cue trials in blocks, rather than randomly intermixed, interesting patterns emerged. The observers were clearly divided into two groups: observers who showed effects of cuing, and observers who did not. This dichotomy of performance did not occur in either Experiments 1 or 2; in those experiments, all observers showed a significant effect of cuing. When analyzed without separating the groups with such clearly different patterns of performance from one another, the results indicate that there is no attentional anisotropy in the visual field. However, when the groups are separated, those observers who did show a cuing effect also showed anisotropy of performance across the visual field as a function of attention. A clear linear relationship emerged between distance from fixation and the effects of the cue: as distance from fixation increased, so did the impact of the cue. In the group without the cuing effect, there were only effects that differed on the basis of visual field location, not on its interaction with cuing condition. This clear dichotomy between the ways that the two groups performed in this experiment illustrates why some previous work may have found - or not found -attentional anisotropy across the visual field. Organizing the two types of trials into blocks allows observers to develop strategies that they feel facilitate their performance, no matter what they are instructed to do. Interestingly, observers who demonstrated a consistent cuing effect were only 7 ms faster overall than those observers who did not. Due to this small advantage in overall performance, it is understandable why some observers might choose not to allocate their attention according to the cue, which requires effort, and simply ready themselves to respond to the sudden onset of the target letter 42 when it was presented. Presenting each type of trial within blocks also eliminates any immediate reinforcement for shifting attention to the cued location that could arise from a perceived advantage in speed on cued trials, as the two types of trials were not presented intermingled with one another. The observers who did not use the cue, however, were not immune to anisotropy of processing across the visual field: their speed of response was still mediated by target location, with better performance closer to fixation (Figure 2.12), and a small effect of better performance on the right side of the horizontal meridian (Figure 2.13). As in Experiment 1, because the stimuli were magnified to accommodate for the cortical magnification factor, and the stimuli were well above threshold, these results are likely due to anisotropy of exogenous orienting in the visual field. A comparison between Experiment 1 and the cue-effect group of Experiment 3 also revealed two differences. First, while in Experiment 1 cuing effects did not increase as eccentricity from fixation increased, in Experiment 3 this was the case. Looking more closely at the interaction between Cue condition and Eccentricity found in Experiment 3, it is clear that the increased effects of attention arise from the cue reducing the costs of not specifically attending to locations farther from fixation (see Figure 2.11), with performance at cued locations nearly equivalent at the four eccentricities. This may indicate that, in the blocked paradigm, observers adopted a different strategy for spreading their attention across the possible target locations. One possibility is that observers were spreading their attention less equally across all possible target locations, favoring locations closer to fixation. In both Experiment 1 and Experiment 3, however, 43 when the target location was cued with 100% certainty, reaction time was consistent across all eccentricities. Second, in Experiment 1, performance was significantly different on the horizontal and vertical meridians, showing a greater cuing effect on the vertical meridian. In Experiment 3, this pattern of performance did not emerge. This difference may be due to a lack of power, as only 10 observers showed a consistent cuing effect in Experiment 3. 2.5 General discussion Is the allocation of visual attention anisotropic across the visual field? These three experiments clearly show response to a target is dependent upon its location in the visual field - and that location interacts with the way in which attention is allocated according to several factors. Experiment 1 established that, when the observer knows with 100% certainty that the target will appear in a specific location, there is anisotropy of performance as a function of attention. Identifying a target letter is improved by cuing to a greater extent on the vertical meridian than on the horizontal meridian. To our knowledge, this is the first study to specifically demonstrate a horizontal-vertical meridian asymmetry in endogenous attention orienting. Additionally, our finding that there is a marginally significant difference between the effects of attention in the lower visual field and the upper visual field, showing greater impact in the lower visual field, is the first such instance in a sparse display: previous work (e.g. He et al., 1996) has demonstrated a preference for attentional selection in the lower visual field with distracting stimuli in the display. This work is the first to demonstrate that a lower visual field advantage may 44 occur in endogenous orienting with only a single stimulus to process. The results of Experiment 1 did not show any evidence of a right-left anisotropy as a result of endogenous attention orienting. Also, while attention did not interact with eccentricity, performance declined consistently as distance from fixation increased - even when the stimuli were scaled to accommodate for the cortical magnification factor and were well above perceptual thresholds. That this pattern of performance remained consistent across all experiments indicates that there may be a role for exogenous orienting in the results: the sudden onset of the target letter would summon attention to that location in the display. However, it is unclear whether this is entirely due to exogenous orienting -previous research has indicated that exogenous cues serve to balance out the perceptual deficits that arise with greater distance from fixation (e.g. Carrasco et al., 2002). Given that endogenous orienting was found to be anisotropic, Experiment 2 asked the question whether changing the likelihood that a target will appear in the cued location would affect this anisotropy. Changing the predictive value of the cue changed nothing about the underlying stimulus: the exogenous orienting associated with the task would remain the same, as well as any potential perceptual anisotropies, as nothing about the target display changed between Experiments 1 and 2. The only change would be in the way that the observers approached the task: in order to perform the task most effectively, attention should be preferentially allocated to the cued location, but also maintained at the other possible target locations in the display. The benefits of attending to specific visual field locations over others were consistent in the results of Experiment 2: a marginally significant advantage for attending to the lower visual field was driven by vertical meridian asymmetry, where endogenous orienting improved attention more on the lower 45 half than the upper half. As in Experiment 1, there was no advantage to attending to either the right or the left visual field. The results of Experiment 2, however, also demonstrated that the manner in which attention was distributed away from the cued target location was dependent upon the most useful way to allocate attention: between the possible target locations. This was revealed by the fact that cuing had the least effect at four and eight degrees from fixation: when not cued to attend to those locations, what attention remained after attending to the cued location was distributed to those locations in order to be closest to other locations where a target might appear. LaBerge's (1995; LaBerge et al., 1997) gradient model begins to offer an explanation for the pattern of results found in Experiment 2. The dual operation of a selective mechanism at the cued location, as well as a secondary degree of monitoring the other possible target locations, fits the finding that there is a preference for 4 and 8 degrees from fixation on invalidly-cued trials. As an additional explanation for this pattern of results, Dori and Henik (2006) proposed two types of attention gradients established with endogenous orienting: one that gives the cued location priority over other locations in the visual field, and one that inhibits processing in a graded way starting at fixation when the target is cued to appear elsewhere in the visual field. We propose a modification of this model of attention gradients that puts greater emphasis on the properties of the display. Dori and Henik (2006) used displays where the stimuli were presented on the vertical meridian, and on two intercardinal locations on either side of the vertical meridian. Each of these radial locations had two possible target locations at 4 degrees and 9 degrees from fixation. With these types of displays, there were only two 46 possible circular regions in which the target could appear: the outer semicircle, and the inner semicircle. In our experiments, the target could appear in any of four possible circular arrangements which extended below the horizontal meridian - introducing the influence of a greater number of possible target locations, and a wider swath of visual field of which to attend. Were Dori and Henik's (2006) two-gradient model to be applied to our results, we would anticipate that on invalidly-cued trials, the closer the target appeared to fixation the slower response time would be, with a linear decrease in reaction time as eccentricity from fixation increased. The preferential processing of the stimulus locations in Experiment 2 of our work at 4 and 8 degrees from fixation, however, with an increase in RT at 1 and 12 degrees - closest to fixation and farthest from fixation - indicates that inhibition of various visual field locations may be more dependent upon the most advantageous way to distribute the excitation and inhibition gradients in the visual field in order to do the task. It is likely that what is occurring given a display with many possible target locations, such as in our experiments, is an interplay between an inhibitory gradient, as proposed by Dori and Henik (2006), an endogenous manipulation of the state of general readiness, proposed by LaBerge (1995), and selection of the cued location in the visual field. Finally, Experiment 3 was conducted in order to make a direct comparison between our work and previous work on endogenous orienting, which has asked observers in a blocked design to either shift their attention to a cued location on each trial with 100% certainty that the target would appear there, or to respond to targets that appear randomly at any of several locations. Results of Experiment 3 clearly 47 demonstrated that previous instances where no attentional anisotropy was found could have arisen from different strategies adopted by observers for the different conditions they faced. Allowing observers to establish a separate strategy for each condition - which more easily occurs when each trial in a block requires the same approach - will affect how attention is allocated, and, in some cases, can negate the effects of cuing altogether. This finding offers an explanation for why Mackeben (1999) did not find a consistent cuing advantage for any particular location in the visual field: blocking the types of trials allows for a different approach to the task, negating significant anisotropic cuing effects. Interestingly, Mackeben's (1999) finding of a general preference for stimulus processing on the horizontal meridian in his experiment is reflected in the results of our Experiment 1, where cuing explicitly has a greater effect on the horizontal meridian than the vertical meridian. Overall, from our results it appears that attention is not a mechanism that operates independent of task requirements such as the display parameters, likelihood of target appearance at a cued location, and the requirement to maintain an attentional state across long durations. From these considerations, we can derive several suggestions for how the effects of visual field location, attention distribution, and attention state individually play a role in how attention is allocated in the visual field, and how they interact with one another. (1) Under conditions where attention is not allocated with 100% certainty to a target location, attention is preferentially allocated to the cued location, but also distributed among other possible target locations. 48 (2) When attention is fully allocated - at least to the extent required to do the task -to a cued location, the effects of attention are anisotropic at different visual field locations. In this case, attention has the greatest impact on the vertical meridian, relative to the horizontal meridian, with the lower half of the vertical meridian receiving the greatest benefit of attention. This finding is consistent with a lower-visual-field advantage for endogenous attention (e.g. He et al., 1996). (3) The distribution of attention among those possible target locations is based on the most advantageous strategy for orienting attention based on the demands of the display. This means that the effect of location in the visual field on attention is not driven by a preference for one particular visual field location; it is also determined by the demands of the task. (4) Given the findings of Experiment 3, individual strategic differences should be given consideration when evaluating the effects of visual attention at various visual field locations. Anisotropy of visual attention across the visual field, then, arises from the confluence of several factors. The existence of attentional anisotropy - or its non-existence - must be considered in the context of the paradigm in which it was assessed. The pattern of results here indicates a heavy dependence on what the observer is asked to do. Future research should include additional assessment of the stimulus parameters that affect attentional anisotropy across the visual field. 49 2.5 Tables and Figures - Chapter 1 Table 1. Mean cue effects for individual observers, Experiment 3. Observer Mean cuing effect Mean error rate Group 24 -18 5.16% 1 23 -7 2.30% 1 2 -7 4.50% 1 8 -3 3.79% 1 18 -3 5.08% 1 10 -2 1.00% 1 3 1 4.59% 1 17 6 6.11% 1 4 8 4.99% 1 9 10 4.75% 1 20 17 2.62% 2 6 17 2.04% 2 15 26 3.20% 2 21 26 2.83% 2 - 13 31 4.77% 2 7 35 2.11% 2 12 39 8.85% 2 14 42 1.33% 2 19 78 1.68% 2 22 99 1.05% 2 Figure 2.1. Schematic of typical trials in Experiment 1. Upper half depicts trial where cue indicates with 100% certainty where the target will appear. The lower half depicts a trial where the cue does not indicate where the target will appear Target: 150 ms Cue: stays on 550 ms; fades off over 500 ms 100% certainty ° n • rP • • • • + • • • • • • • Cue: stays on 550 ms; fades off over 500 ms No certainty , LBBJ pass^ L j ISI: 2 0 0 - 4 0 0 ms • • • • + • • • • • • • • • • • + • • • • • • • ISI: 2 0 0 - 4 0 0 ms • • • • • • + • • • • • • D • • • • • • + • • • • • 51 Figure 2.2. Main effect of eccentricity, Experiment 1. Experiment 1:. Eccentricity F(3, 57)=?38.01,p-< 0.001 Error bars denote 95% confidence intervals 540 | : i • Figure 2.3. Cue x Location interaction, Experiment 1. Experiment 1: Cue Condition x Radial Location F(7, 133) = 3.2D, p = 0.004 Error bars denote 95% confidence intervals 560 ^ 500 c 490 | . 480 1 470 NE SE S SW Radial location cued uncued 53 Figure 2.4. Difference scores (uncued reaction time - cued reaction time) for stimuli presented at all eight radial locations (N, NE, E, SE, S, SW, W, N), Experiment 1. o C3 QJ L_ QJ Z3 o I •a aj zs ^ U QJ II CO QJ a j o QJ QJ b 55 50 45 40 35 30 25 20 Experiment 1: Radial location F(7, 133) = 3.20; p = 0.004 Error bars denote 95% confidence interval I i I NE SE SW W NW N Radial location 54 Figure 2.5. Cue x Eccentricity interaction, Experiment 2. Experiment 2: Cue condition x Eccentricity F(3, 51) = 3.60, p = 0.02 Error bars denote 95% confidence intervals 600 580 560 •T3 C o o QJ V) | 540 OJ £ 520 *j c o 1 500 CD or 480 460 1 4 8 Eccentricity (degrees of visual angle) Valid Invalid 55 Figure 2.6. Difference scores (reaction time to invalidly cued targets - reaction time to validly cued targets) for stimuli presented at all eight radial locations (N, NE, E, SE, S, SW, W, N), Experiment 2. Experiment 2: Radial location F(7, 133) = 1.66; p = 0.12 Error bars denote 95% confidence interval 52 | • . — • . • a I 48t _ T T o j 44 =3 NE E SE S SW W NW N Radial locations 56 Figure 2.7. Difference scores (reaction time to invalidly cued targets - reaction time to validly cued targets) for stimuli presented at the four eccentric locations (1°, 4°, 8°, and 12° from fixation), Experiment 2. 55 Experiment 2: Eccentricity - Difference scores F(3, 57) = 3.35, p = 0.025 Error bars denote 95% confidence intervals lid RT) 50 45 [Invalid -40 score ( 35 Difference: 30 25 20 1 4 8 12 Eccentricity (degrees of visual angle) 57 Figure 2.8. Average reaction times within cuing condition for Experiments 1 and 2. Experiments 1 & 2: Comparison between RTs Error bars denote 95% confidence intervals 550 , , , , r— cued -100% uncued valid - 70% invalid - 30% 58 Figure 2.9. Cue x Eccentricity interaction for group with consistent cuing effect, Experiment 3. 560 g 440 400 Experiment 3: Cue condition x Eccentricity F(3, 27) = 5.29, p = 0.005 Errors bars denote 95% confidence intervals 1 4 8 12 Eccentricity (degrees of visual angle) Cued Uncued 59 Figure 2.10. Mean reaction times to targets presented at four eccentric locations for observers who did not show cuing effects, Experiment 3. Experiment 1: Eccentricity F(3, 57)=38.01. p < 0.001 Error bars denote 95% confidence intervals 540 | . : • • 530 • to •a 520 • o T ft 480 • or 470 • 460 I 1 1 J 1 4 8 12 Eccentricity (degrees of visual angle) 60 Figure 2.11. Mean reaction times to targets presented at eight radial locations for observers who did not show cuing effects, Experiment 3. Experiment 3: Radial Location F(7, 49) = 3.27; p = 0.006 Error bars denote 95% confidence interval 530 | • • • • . • • 525 • g 520 • 8515f T I Radial Location 61 Figure 2.12. Difference scores (reaction time to invalidly cued targets - reaction time to validly cued targets) for stimuli presented at the four eccentric locations (1°, 4°, 8°, and 12° from fixation), Experiment 3. Calculated only from observers who showed an overall cuing advantage. Experiments: Eccentricity - Difference scores F(3. 27) = 5.29, p = 0.005 Error bars denote 95% confidence intervals 120 | • - • a 100 • -,_ "D aj -40 1 ' •: ' -1 4 8 12 Eccentricity (degrees of visual angle) 62 Figure 2.13. Difference scores (reaction time to invalidly cued targets - reaction time to validly cued targets) for stimuli presented at all eight radial locations (N, NE, E, SE, S, SW, W, N), Experiment 3. Calculated only from observers who showed an overall cuing advantage. 80 a 70 CD ? 60 T3 QJ 3 50 c 3 to £ 40 o o OJ <_> c QJ v_ QJ fc b 30 20 10 Experiments: Radial location - Difference scores F(7, 63) = 0:549; p = 0.79 Error bars denote 95% confidence Interval 1 I 1 1 1 1 1 I NE SE S SW Radial location W NW N 63 2.7 References Altpeter, E., Mackeben, M . , & Trauzettel-Klosinski, S. (2000). The importance of sustained attention for patients with maculopathies. Vision Research, 40, 1539-1547. Anstis, S. (1974). A chart demonstrating variations in acuity with retinal position. Vision Research, 14, 589-592. Ashkenazi, A. , & Marks, L.E. (2004, October). Object-based attention in the visual field. Poster presentation at Fechner Day, Coimbra, Portugal. Awh, E., & Pashler, H. (2000). Evidence for split attentional foci. Journal of Experimental Psychology: Human Perception and Performance, 26(2), 834-846. Azzopardi, P., Jones, K.E. , & Cowey, A . (1999). Uneven mapping of magnocellular and parvocellular projections from the lateral geniculate nucleus to the striate cortex in the macaque monkey. Vision Research, 39(13), 2179 - 2189. Berger, A. , Henik, A. , & Rafal, R. (2005). Competition between endogenous and exogenous orienting of visual attention. Journal of Experimental Psychology: General, 134, 207-221. Bilodeau, L. , & Faubert, J. (1997). The oblique effects and chromatic motion throughout the visual field. Vision Research, 39, 757-763. Bisiach, E., Capitani, E., Colombo, A. , & Spinnler, H. (1976). Halving a horizontal segment: A study on hemisphere-damaged patients with cerebral focal lesions. Archives Suisses de Neurologie, Neurochirurgie et de Psychiatrie, 118, 199-206. Briand, K., & Klein, R . M . (1987). Is Posner's "beam" the same as Triesman's "glue"?: On the relationship between visual orienting and feature integration theory. 64 Journal of Experimental Psychology: Human Perception and Performance, 13, 228-241. Cameron, E.L., Tai, J.C., & Carrasco, M . (2002). Covert attention affects the psychometric function of contrast sensitivity. Vision Research, 42, 949-967. Carrasco, M . & Frieder, K.S. (1996). Cortical magnification neutralized the eccentricity effect in visual search. Vision Research, 37(1), 63 - 82. Carrasco, M . , Giordano, A . M . , & McElree, B. (2004). Temporal performance fields: visual and attentional factors. Vision Research, 44, 1351-1365. Carrasco, M . , Giordano, A . M . , & McElree, B. (2006). Attention speeds processing across eccentricity: Feature and conjunction searches. Vision Research, 46, 2028 - 2040. Carrasco, M . , McElree, B., Denisova, K., & Giordano, A . M . (2003). Speed of visual processing increases with eccentricity. Nature Neuroscience, 6(1), 699 - 700. Carrasco, M . , Talgar, C , & Cameron, E.L. (2001). Characterizing visual performance fields: Effects of transient covert attention, spatial frequency, eccentricity, task, and set size. Spatial Vision, 15, 61-75. Carrasco, M . , Williams, P.E., & Yeshurun, Y . (2002). Covert attention increases spatial resolution with or without masks: Support for signal enhancement. Journal of Vision, 2, 467-479. Cheal, M . , & Lyon, D.R. (1991). Central and peripheral precuing of forced-choice discrimination. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 43A, 859-880. Chen, H. , Yao, D., & Liu, Z. (1994). A study on asymmetry of spatial visual field by analysis of the fMRI BOLD response. Brain Topography, 17(1), 39 - 46. 65 Cohen, J. (1973). Eta-squared and partial eta-squared in fixed factor A N O V A designs. Educational and Psychological Measurement, 33, 107-112. Connolly, M . & Van Essen, D. (1984). The representation of the visual field in parvicellular and magnocellular layers of the lateral geniculate nucleus in the macaque monkey. Journal of Comparative Neurology, 226(A), 544 - 564. Cowey, A. , & Rolls, E.T. (1974). Human cortical magnification factor and its relation to visual acuity. Experimental Brain Research, 21, 447 - 454. Curcio, C.A., & Allen, K . A . (1990). Topography of ganglion cells in the human retina. Journal of Comparative Neurology, 300, 5 -25 . Dori, H. , & Henik, A . (2006). Indications for two attentional gradients in endogenous visual-spatial attention. Visual Cognition, 13(2), 166-201. Dosher, B.A. , & Lu, Z.-L. (2000). Noise exclusion in spatial attention. Psychological Science, 11(2), 139- 146. Downing, C.J., & Pinker, S. (1985). The spatial structure of visual attention. In M.I. Posner & O.S.M. Marin (Eds.), Attention and performance XI, (pp. 171-188). Hillsdale, NJ: Erlbaum. Edgar, G.K., & Smith, A.T. (1990). Hemifield differences in perceived spatial frequency. Perception, 19(6), 759-766. Edwards, M . , & Badcock, D.R. (1993). Asymmetries in the sensitivity to motion in depth: a centripetal bias. Perception, 22, 1013-1023. Eriksen, C.W. & Yeh, Y . Y . (1985). Allocation of attention in the visual field. Journal of Experimental Psychology: Human Perception and Performance, 11(5), 583 -597. 66 Evert, D.L., McGlinchey-Berroth , R., Verfaellie , M . , & Milberg, W.P. (2003). Hemispheric asymmetries for selective attention apparent only with increased task demands in healthy participants. Brain and Cognition, 53, 34 - 41. Fink, G.R., Halligan, P.W., Marshall, J.C., Frith, C.D., Frakowiak, R.S.J., & Dolan, R.J. (1996). Where in the brain does visual attention select the forest and the trees? Nature, 382, 626-628. Fink, G.R., Marshall, J.C., Weiss, P.H., & Zilles, K. (2001). The neural basis of vertical and horizontal line bisection judgments: An fMRI study of normal volunteers, Neurolmage, 14, S59 - S67. Fukisama, S.S., & Faubert, J. (2001). Perceived length in the central visual field: evidence for visual field asymmetries. Vision Research, 41, 2119-2126. Gawryszewski, L. , Riggio, G. de, L., Rizzolatti, G., & Umilta, C. (1987). Movements of attention in the three spatial dimensions and the meaning of "neutral" cues. Neuropsychologia, 25, 19-29. Gobell, J.L., Tseng, C., & Sperling, G. (2004). The spatial distribution of visual attention. Vision Research, 44, 1273 - 1296. Hagenbeek, R.E., & Van Strien, J.W. (2002). Left-right and upper-lower visual field asymmetries for face matching, letter naming, and lexical decision. Brain and Cognition, 49, 34 - 44. Hahn, S. & Kramer, A.F. (1998). Further evidence for the division of attention among non-contiguous locations. Visual Cognition, 5(1/2), 217 - 256. 67 Handy, T.C., Kingstone, A. , & Mangun, G.R. (1996). Spatial distribution of visual attention: Perceptual sensitivity and response latency. Perception & Psychophysics, 55(4), 613 - 627. He, S.H., Cavanagh, P., & Intriligator, J. (1996). Attentional resolution and the locus of visual awareness. Nature, 383, 334-337. Heilman, K . M . , Watson, R.T., & Valenstein, E. (1993). Neglect and related disorders. In K . M . Heilman and E. Valenstein (Eds.), Clinical Neuropsychology (3rd Ed.). New York: Oxford University Press, 1993, 279-336. Heilman, K . M . , & Valenstein, E. (1979). Mechanisms underlying hemispatial neglect. Annals of Neurology, 5, 166-170. Helmholtz, H. von. (1867; 1924). Treatise on Physiological Optics. Translated by J.P.C. Southall, The Optical Society of America. Hubel, D.H., & Weisel, T.N. (1960). Receptive fields of optic nerve fibers in the spider monkey. Journal of Physiology, 154, 572-580. Intriligator, J., & Cavanagh, P. (2001). The spatial resolution of visual attention. Cognitive Psychology, 43, 171-216. Jonides, J. (1981). Voluntary vs. automatic control over the mind's eye's movement. In J. Long & A . Baddeley (Eds). Attention & Performance ZY(pp. 187-203) Hillsdale, NJ: Erlbaum. Jonides, J. (1983). Further toward a model of the mind's eye's movement. Bulletin of the Psychonomic Society, 21, 247 - 250. Kinsbourne, M . (1970). The cerebral basis of lateral asymmetries in attention. Acta Psychologica, 33, 193-201. 68 Kitterle, F.L, Christman, S., & Hellige, J.B. (1990). Hemispheric differences are found in the identification, but not the detection, of low versus high spatial frequencies. Perception and Psychophysics, 48(4), 297 - 306. Kraft, A. , Pape, N . , Hagendorf, H., Schmidt, S., Naito, A., Brandt, S.A. (2007). What determined sustained visual attention? The impact of distracter positions, task difficulty and visual fields compared. Brain Research, 1133, 123 - 135. Kramer, A.F. , & Hahn, A. (1995). Splitting the beam: Distribution of attention over non-contiguous regions of the visual field. Psychological Science, 6(6), 381 - 386. LaBerge, D. (1995). Attentional processing: The brain's art of mindfulness. Cambridge, M A : Harvard University Press. LaBerge, D., Carlson, R.L., Williams, J.K., & Bunney, B.G. (1997). Shifting attention in visual space: Tests of moving-spotlight models versus an activity-distribution model. Journal of Experimental Psychology: Human Perception and Performance, 23(5), 1380- 1392. Ladavas, E., Carletti, M . , & Gori, G. (1994). Automatic and voluntary orienting of attention in patients with visual neglect: Horizontal and vertical dimensions. Neuropsychologia, 32(10), 1195-1208. Lamme, V . A . , & Roelfsema, P.R. (2000). The distinct modes of vision offered by feedforward and recurrent processing. Trends in Neuroscience, 23(11), 571 - 579. Levi, D .M. , Klein, S.A., & Aitsebaomo, A.P. (1985). Vernier acuity, crowding, and cortical magnification. Vision Research, 25(1), 963 - 977. Levine, M.W., & McAnany, J.J. (2005). The relative capabilities of the upper and lower hemifields. Vision Research, 45, 2820 - 2830. 69 Lindell, A . K . , & Nicholls, M.E.R. (2003). Attentional deployment in visual half-field tasks: The effect of cue position on word naming latency. Brain and Cognition, 53, 273 - 277. Liu, T., Heeger, D.J., & Carrasco, M . (2006). Neural correlates of the visual vertical meridian asymmetry. Journal of Vision, 6, 1294 - 1306. Losier, B.J., & Klein, R . M . (2004). Covert orienting within peripersonal and extrapersonal space: young adults. Cognitive Brain Research, 19, 269 - 274. Mackeben, M . (1999). Sustained focal attention and peripheral letter recognition. Spatial Vision, 12, 51-72. Mangun, G. R., Luck, S J., Plager, R., Loftus, W., Hillyard, S. A. , Handy, T. C , Clark, V. P., & Gazzaniga, M . S. (1994). Monitoring the visual world: Hemispheric asymmetries and subcortical processes in attention. Journal of Cognitive Neuroscience, 6, 265-273. Maunsell, J.H., & Van Essen, D.C. (1987). Topographic organization of the middle temporal visual area in the macaque monkey: representational biases and the relationship to callosal connections and myeloarchitectonic boundaries. Journal of Comparative Neurology, 266(A), 535-555. McCourt, M.E. , & Garlinghouse, M . (2000). Asymmetries of visuospatial attention are modulated by viewing distance and visual field elevation: Pseudoneglect in peripersonal and extrapersonal space. Cortex, 36, 713 - 731. Michael, G.A., & Ojeda, N . (2005). Visual field asymmetries in selective attention: Evidence from a modified search paradigm. Neuroscience Letters, 388, 65 - 7 0 . 70 Michimata, C , & Hellige, J. (1987). Effects of blurring and stimulus size on the lateralized processing of nonverbal stimuli. Neuropsychologia, 25, 397-407 Perry, V . H . , & Cowey A. (1985.) The ganglion cell and cone distributions in the monkey's retina: implications for central magnification factors. Vision Research, 25, 1795- 1810. Posner, Snyder & Davidson (1980) Attention and the detection of signals. Journal of Experimental Psychology, 109, 160-74. Portin, K. , Vanni, S., Virsu, V. , & Hari, R. (1999). Stronger occipital cortical activation to lower than upper visual field stimuli. Experimental Brain Research, 124, 287 -294. Previc, F.H. (1990). Functional specialization in the lower and upper visual fields in humans: Its ecological origins and neurophysiological implications. Behavioral and Brain Sciences, 13, 519-575. Previc, F. (1998). The neuropsychology of 3D space. Psychological Bulletin, 124, 123 -164. Proverbio, A . M . , Zani, A. , & Avella, C. (1997). Hemispheric asymmetries for spatial frequency discrimination in a selective attention task. Brain and Cognition, 34, 311 -320. Rijsdijk, J.P., Kroon, J.N., & van der Wildt, G.J. (1980). Contrast sensitivity as a function of position on the retina. Vision Research, 20, 235-241. Robertson, L., & Delis, D.C. (1986). Part-whole processing in unilateral brain damaged patients: Dysfunction of hierarchical organization. Neuropsychology, 24, 363-70 71 Robertson, L., Lamb, M.R., & Knight, R.T. (1988). Effects of lesions of temporalparietal junction on perceptual and attentional processing in humans. Journal of Neuroscience, 5(10), 3757-3769 Rovamo, J., & Virsu, V . (1979). An estimation and application of the human cortical magnification factor. Experimental Brain Research, 37, 495-510. Rubin, N . , Nakayama, K., & Shapley, R. (1996). Enhanced perception of illusory contours in the lower versus upper visual hemifields. Science, 271, 651-653. Sakaguchi, Y . (2003). Visual field anisotropy revealed by perceptual filling-in. Vision Research, 43, 2029-2038. Shepherd, M . , & Muller, H.J. (1989). Movement versus focusing of visual attention. Perception and Psychophysics, 46, 146-154. Talgar, C P . , & Carrasco, M . (2002). Vertical meridian asymmetry in spatial resolution: Visual and attentional factors. Psychonomic Bulletin & Review, 9, 714-722. Tootell, R.B., Switkes, E., Silverman, M.S., & Hamilton, S.L. (1988). Functional anatomy of macaque striate cortex. II. Retinotopic organization. Journal of Neuroscience, 8, 1531 - 1568. Turk-Browne, N.B. , & Pratt, J. (2005). Attending to eye movements and retinal eccentricity: Evidence for the activity distribution model of attention reconsidered. Journal of Experimental Psychology: Human Perception and Performance, 31(5), 1061 - 1066. Van Essen, D.C., Newsome, W.T., & Maunsell, J.H. (1984). The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability. Vision Research, 24, 429 - 448. 72 Verfaellie, M . , Bowers, D., & Heilman, K . M . (1988). Hemispheric asymmetries in mediating intention, but not selective attention. Neuropsychologic, 26, 521 - 531. Weymouth, F.W. (1958). Visual sensory units and the minimal angle of resolution. American Journal of Ophthalmology, 46, 102-113 Wright, R.D., & Ward, L . M . (1998). The control of visual attention. In Visual Attention, R.D. Wright, Ed. Oxford: New York, 132-186. Wright, R.D. & Ward, L . M . (in press). Orienting of attention. New York: Oxford University Press. Yantis, S., & Jonides, J. (1990). Abrupt visual onsets and selective attention: voluntary versus automatic allocation. Journal of Experimental Psychology: Human Perception and Performance, 16, 121 - 134. 73 3 Effects of endogenous attentional distribution across the visual field on performance: another form of attentional anisotropy 3 3.1 Introduction The fact that we can covertly, voluntarily, shift our attentional gaze away from fixation is well-documented (e.g,. Helmholtz, 1867; Jonides, 1981; Posner, Snyder & Davidson, 1980). But are the effects of covertly attending to a particular location in the visual field the same for all locations? We recently provided preliminary evidence that endogenous, or voluntary, covert attending to locations away from fixation is, indeed, anisotropic (Roggeveen & Ward, submitted). That is, the benefit to responses to targets of attending to a particular location in the visual field, compared to not attending to that location, varies with location. But these attentional anisotropies are not well understood, and they interact in complex ways with task demands and perceptual anisotropies. In the present paper we explore this interaction further, focusing in particular on the meridian crossing effect (e.g. Hughes & Zimba, 1985, 1987; Rizzolatti, Riggio, Dascola, & Umilta, 1987). Before describing the present experiments, however, we briefly review what is known about perceptual and attentional anisotropies. Roggeveen and Ward (submitted) contains a more complete discussion of this literature. 3.1.1 Perceptual & attentional anisotropies There is ample evidence that the way in which the visual system is designed dictates several perceptual anisotropies, which are thought to be independent of attention. Generally, four consistent patterns of performance have been found in the way that observers respond to stimuli presented in different locations in the visual field. First, 3 A version of this chapter has been submitted for publication. Roggeveen, A.B. , & Ward, L . M . (submitted). Effects of endogenous attentional distribution across the visual field on performance: another form of attentional anisotropy. 74 eccentricity anisotropy refers to the fact that the farther a stimulus is presented from fixation, the more difficult it is to perceive (e.g. Anstis, 1974). In order to accommodate for this loss of perceptual acuity with greater eccentricity, or distance from fixation, stimuli must be scaled in size (e.g. Anstis, 1974). Even when the perceivability of stimuli is equated, however, the effects of exogenously, or involuntarily, covertly orienting attention to different eccentricities in space are anisotropic. The effects of a non-predictive cue stimulus that suddenly appears in the periphery are greater the farther away from fixation that stimulus is (Carrasco, Williams, & Yeshurun, 2002; Talgar & Carrasco, 2002). The effects of endogenous orienting, however, are less clear-cut: Roggeveen and Ward (submitted) demonstrated that the effects of distance from fixation interact with the manner in which attention is voluntarily distributed across the display, finding that eccentricity effects could be manipulated by changing what the observer was asked to do. Right-left anisotropy is the advantage in processing stimuli when they are presented in either the right or left visual fields. The benefits of perceiving a stimulus in one visual field over the other, however, are highly stimulus-specific, and thought to be driven by specialization of each hemisphere for the processing of certain types of stimuli (e.g. Kinsbourne, 1970). Most relevant here is the preference of the hemispheres for processing individual letter stimuli: the left hemisphere's apparent preference for processing high-frequency, local information (e.g. Fink et al., 1996; Kitterle, Christman, & Hellige, 1990) translates to preferential processing of the identification of letter stimuli. A bias for attending to the right visual field over the left, in contrast, has been clearly demonstrated among both patients suffering from hemineglect, a condition where, 75 due to brain damage, only the right visual field is perceived (e.g. Heilman & Valenstein, 1979; Kinsbourne, 1970), as well as in normal adults (Lindell & Nicholls, 2003). However, many authors (e.g. Kraft et al., 2007; Roggeveen & Ward, submitted; Verfaellie, Bowers, & Heilman, 1988) have failed to find a difference in processing of stimuli in the right and left visual fields due to attentional influences, so the matter is still undecided. Vertical anisotropy is another instance of a consistent perceptual anisotropy. In this case, the visual system seems to be better at processing stimuli presented in the lower half of the visual field, below the horizontal meridian, than above the horizontal meridian (e.g. Previc, 1990; 1998; Talgar & Carrasco, 2002). Many studies have also demonstrated an attentional advantage in the lower visual field: the effects of attending to locations in the lower visual field are greater than when attending to an equivalent stimulus in the upper visual field (e.g. He, Cavanagh, & Intriligator, 1996; Intriligator & Cavanagh, 2001; Losier & Klein, 2004; Roggeveen & Ward, submitted). Examples of inconsistent attentional effects in the lower visual field do exist, however (e.g. Altpeter, Mackeben, & Trauzettel-Klosinski, 2000; Mackeben, 1999). A sub-category of anisotropy under vertical anisotropy is vertical meridian anisotropy: performance is generally better on the lower half of the vertical meridian than the upper half (e.g. Carrasco, Talgar, & Cameron, 2001). Thus far, vertical meridian asymmetry due to either endogenous (Roggeveen & Ward, submitted) or exogenous (e.g. Talgar & Carrasco, 2002) attention orienting has not been found. Finally, horizontal-vertical anisotropy refers to a preference for processing stimuli presented on the horizontal meridian over those presented on the vertical meridian 76 (e.g. Carrasco, et al., 2001; Rovamo & Virsu, 1979). This preference has been demonstrated in many types of perceptual tasks, such as resolution of spatial frequency (Rijsdijk, Kroon, & van der Wildt, 1980). Attentionally, exogenous orienting has greater effects on the speed of processing on the vertical meridian than on the horizontal meridian (Carrasco et a l , 2004). Endogenous orienting, similarly, has greater effects on the vertical meridian than on the horizontal meridian (Roggeveen & Ward, submitted). 3.1.2 Attention distribution effects across visual field Recent work in our laboratory (Roggeveen & Ward, submitted) has begun to build a picture of how attentional anisotropy interacts with the demands of the perceptual task. For instance, uncertainty about whether a target will appear at a cued location alters the pattern of anisotropic performance to reflect the manner in which attention is distributed across possible target locations. Roggeveen and Ward (submitted) asked observers to perform a letter identification task for a target presented at one of 32 possible target locations, presented at eight radial locations (N, NE, E, SE, and so on) and four eccentricities (1°, 4°, 8°, and 12° from fixation). In separate experiments, the target location was cued with either 100% certainty that it would appear in the cued location, (and compared to asking observers to distribute their attention across all possible locations), or with 70% certainty, where on 30% of the trials, the target would appear at one of the other locations on the screen. Under conditions where observers knew with 100% certainty (versus complete uncertainty) where the target would appear, two clear patterns of attentional anisotropy emerged, paralleling the perceptual anisotropies outlined above. Consistent with previous findings (e.g. He et al., 1996), the greatest effect of cuing the target location was seen on the vertical meridian, compared to the horizontal 77 meridian, driven by the largest cuing effect occurring on the lower half of the vertical meridian. However, when the cue predicted the target location with only 70% certainty, with targets appearing at uncued locations 30% of the time, the results revealed different anisotropies. First, cuing effects indicated that attention was preferentially allocated to the cued location - but, in recognition that there was a possibility that the target would not appear there, attention was also distributed among the other possible target locations. Second, the manner in which attention was distributed among uncued locations was strategically advantageous. Rather than attempting to distribute attention equally among all possible locations, observers chose to "hedge their bets" and distribute attention preferentially among locations that would place them halfway between the possible target locations: at 4° and 8°. What this indicates is that the attention distribution mechanism operates flexibly in response to the demands of the task - in this case, the likelihood that the target would appear in a cued location. There also support for the idea that that allocation of attention among possible target locations depends not only on the likelihood that a target will appear in a cued location, but also upon the way in which the display is organized. For instance, Gobell, Tseng, and Sperling (2004) found that when observers were asked to divide their attention among areas delineated by stripes that were oriented either horizontally or vertically across a display, division of attention was better when the stripes were oriented horizontally. Essentially, asking observers to spread their attention to locations along the horizontal meridian, and to locations parallel to the horizontal meridian, was easier than to locations on and parallel to the vertical meridian. Similarly, Ashkenazi and Marks (2004) found that attention could be guided more effectively within an object when the 78 objects were oriented horizontally than when the objects were oriented vertically. This work indicates that patterns of endogenous attentional anisotropy may vary with the manner in which observers have been asked to spread their attention across possible target locations. The general question posed here is this: are we better at attending to particular locations in the visual field than to others? And, more specifically, how is the impact of cuing affected by the way attention is distributed across possible target locations across the visual field? In order to manipulate the manner in which attention was spread across the display without altering essential stimulus characteristics, we took advantage of a theory of attentional distribution proposed by Dori and Henik (2006). Their theory suggests that there are two types of attention gradients established when an endogenous shift of attention is made. One gradient gives the cued location priority over other locations in the visual field, whereas the other inhibits processing in a graded way starting at fixation when the target is cued to appear elsewhere in the visual field. What if a target sometimes appeared at fixation, however, although there was never a cue there? This alteration could disrupt the inhibitory gradient centered at fixation by giving observers a reason to attend there, at least to a certain degree. We performed two experiments to investigate this notion: one with eight possible target locations all away from fixation, and one where the fixation cross was replaced with a possible target location at fixation. Observers were cued to attend to a location on one side of a meridian - either the horizontal, vertical, or one of the two intercardinal meridians - with 70% likelihood that the target would appear at the cued location. On the remaining 30% of trials the target would appear at one of the four locations on the opposite side of fixation. 79 Results of analyses conducted for each experiment, and comparison of the results of the two experiments, demonstrate how anisotropic performance emerges as a result of attentional distribution across the possible target locations. This approach also allows for an assessment of an anisotropic effect of endogenous attention that has not been previously investigated: that of differences in cue effectiveness between the intercardinal, or diagonal, meridians and the cardinal ones. There is some evidence at least that performance on the intercardinal meridians differs from that on the horizontal and vertical meridians. Several studies (e.g. Cameron et al., 2002; Carrasco et al., 2001; Carrasco et al., 2004) found that at intercardinal locations perceptual performance fell somewhere between performance on the horizontal meridian and the vertical meridian. Moreover, Carrasco et al (2004) also showed that exogenously orienting attention to an intercardinal location had effects that fell between those found on the vertical meridian (the greatest effects), and the horizontal meridian (the least effect). Finally, given that some researchers (Cameron, et al., 2002; Carrasco et al., 2001; Carrasco, et al., 2002; Talgar & Carrasco, 2002) have only found a vertical asymmetry on the vertical meridian (vertical meridian asymmetry), and failed to find such effects on the intercardinal locations in the same task, there is reason to believe that processing in these locations differs from processing on the cardinal meridians. An additional reason to expect that the effects of attention will differ along the diagonal axes is the meridian crossing effect (e.g. Hughes & Zimba, 1985, 1987; Rizzolatti, Riggio, Dascola, & Umilta, 1987). Responding to a stimulus that appears across either the horizontal or vertical meridian from where attention is currently oriented appears to be more effortful than responding to a stimulus that appears on the same side 80 or in the same quadrant of visual space where attention is currently oriented. Under conditions where the stimuli are presented on a diagonal axis across the display, observers must reorient across both the horizontal and the vertical meridians, as well as across the other intercardinal meridian, in order to respond to invalidly cued stimuli. If this is so, then the effects of visual attention in the display could be anisotropic, in that cuing would have a larger effect on performance along one of the intercardinal axes than along either the horizontal or vertical axis. In light of the possibility that target processing differs on the intercardinal axes, either perceptually or attentionally, in addition to evaluating the perceptual and attentional anisotropies mentioned earlier, we also directly evaluated performance on the intercardinal meridians as compared to the cardinal meridians. 3.2 Experiment 1 In order to assess the differing effects of endogenous covert attention orienting along the different meridians in the visual field, stimuli were presented within blocks along a single meridian. This design allowed for detailed measurement of the incidence of the perceptual and attentional anisotropies described above, as well as between meridians themselves. In this experiment no targets appeared at fixation, and invalidly cued targets always appeared on the opposite side of fixation from the cued location. 3.2.1 Methods 3.2.1.1 Observers. Twenty observers (10 male; 10 female) participated in the experiment (median age: 20, age range 18 - 24). Two observers reported being left handed; all other observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the 81 purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioural Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. The data collected from one observer were excluded from all subsequent analyses due to error rates greater than 30% in several of the possible target locations (7% errors overall; greater than all other observers). 3.2.1.2 Materials. Observers sat in a comfortable chair with their chin in a chinrest located 53 cm in front of a computer screen. The display was shown on a 19.5-inch monitor with a resolution of 1280 by 1024 pixels and a refresh rate of 85 Hz. Observers were given both written and verbal instructions about the procedure of the experiment. After the observer reported feeling comfortable with the task when shown several practice trials, the experiment began. A schematic of an example trial is shown in Figure 3.1. At the beginning of each trial, a white fixation cross appeared at the center of a black screen, flanked by eight white placeholder squares. The placeholder squares were presented on an imaginary diameter across a circle with a diameter of 24° of visual angle, at 1°, 4°, 8°, and 12° of visual angle on either side of the fixation cross. The squares were magnified in a linear fashion according to Anstis (cortical magnification factor = 2.3; 1974) to equate visibility at all locations. The orientation of the diameter arrangement of the placeholder boxes was randomized across blocks between four possible meridian orientations: North (0°)/ South (180°), East (90°)/ West (270°), Northeast (45°)/ Southwest (225°), and Southeast (135°)/ Northwest (315°). Presenting participants with stimuli in the manner described above allows for specific measurement of the effects of distributing attention along the cardinal 82 and intercardinal axes, rather than across the tested visual field as a whole. To facilitate the allocation of attention across the axis being tested, the placeholder boxes for the locations for each block remained on the screen for the entire block. Overall, there were 32 possible target locations, eight along each of the four meridians. Observers participated in 16 blocks (4 at each meridian in a different overall random order for each observer) of 90 trials each within the two-hour session, completing 1440 trials (45 trials per location). On each trial, the observer was cued to attend to one of the eight locations along the displayed meridian with 70% certainty that the target would appear at the cued location. Therefore, for 70% of trials, the cue correctly indicated the target location: the target was validly cued. On the remaining 30% of trials where the target did not appear at the cued-location (invalidly cued trials), the target always appeared on the opposite side of fixation. Observers were not explicitly informed that the target, if invalidly cued, would always appear on the opposite side of fixation; they were instructed only that the target would appear in one of the other possible target locations on 30%) of trials. In order to reduce confusion about the cued location possibly arising from the presence of eight locations along a meridian, the more typical central cue was not used to cue the target location. Instead, all eight squares turned red, and then all but the cued location faded back to white over 416 ms. The last red square - the cued location - would then fade back to white over the next 416 ms. In this manner, the observer could begin to shift their attention endogenously to the cued location as soon as the non-cue squares began to fade from the screen. Observers were instructed to shift their attention, without moving their eyes, to the cued location. While a peripheral cue in the sense that it appeared at the 83 target location, the time course and fading of the cue required an endogenous shift of attention to the cued location. Whether the cue was validly or invalidly cued varied randomly from trial to trial within the 70/30 ratio of valid to invalid cues. After the cue completely faded to black, and an additional jittered interval of 200 - 400 ms, the target was presented for 150 ms. The target could be either the letter M or the letter N . The observer's task was to identify the letter presented. Like the placeholder squares, the target letters were magnified, using 2.3 as the cortical magnification factor (Anstis, 1974) so that the targets would be equally visible at all target eccentricities. After the target presentation, the observer made a response, pressing the "/" key on the keyboard with their right hand to indicate that the letter M was presented, or the "z" key on the keyboard with their left hand to indicate that the letter N was presented. Once the observer had made their response, feedback was presented by either brightening the fixation cross to indicate a correct response, or by replacing the fixation cross with a minus sign to indicate an incorrect response. Feedback was presented for 1000 ms, giving ample time for observers to blink between trials. Blinks and eye movements were monitored by both the horizontal and the vertical electro-oculograms (EOGs), which were also sampled at 250 Hz. Impedances for EOGs were kept below 20kQ. Trials on which observers made an eye movement between cue onset (beginning of cue fading) and response onset, error trials, and trials with reaction times less than 150 ms or greater than 900 ms were excluded from the analysis. These restrictions left approximately 40 trials per location per observer for the subsequent analyses. 84 3.2.2 Results 3.2.2.1 Reaction time. In order to evaluate the effects of visual field location and validity on reaction time, a three-way (Radial location x Eccentricity x Validity) analysis of variance (ANOVA) was performed on the reaction time data. For any instances where the assumption of sphericity was violated, the Greenhouse-Geisser correction was applied. Note that in this and subsequent analyses, locations were divided into eight radii (N, NE, E, SE, S, SW, W, NW) rather than entire meridians to facilitate comparison with previous studies. Pairs of radii were combined into meridians when appropriate. No significant interactions were found between any of the factors. There was a significant main effect of Validity (F(l,18) = 15.97; p < 0.001; partial n 2 = 0.47; see Table 3.1), with longer reaction times to targets presented at invalidly cued locations. A significant main effect of Eccentricity (F(3,54) = 52.58; p < 0.001; partial r\ = 0.74; see Table 3.2) showed that reaction time was not equivalent at all possible target locations, with longer reaction times farther from fixation. The main effect of Radial location was also significant (F(7,126) = 10.28;p < 0.001; partial n 2 = 0.36; see Figure 3.2). In order to assess the pattern of anisotropy in light of previously documented perceptual anisotropies, four planned comparisons were performed: upper (NW, N , NE) vs. lower (SW, S, SE) visual field locations; right (NE, E, SE) vs. left (NW, W, SW) visual field locations; horizontal (E-W) vs. vertical (N-S) meridian locations; and vertical meridian asymmetry (N vs. S). A fifth planned comparison was also performed in order to evaluate the differences in performance between the cardinal (N, S, E, W) and intercardinal (NE, SE, SW, NW) locations. Planned, rather than post-hoc, comparisons were performed in conjunction with the omnibus A N O V A in order to directly assess 85 whether previously reported perceptual anisotropies emerged in the reaction time data. Planned comparisons allowed for a more direct evaluation of the relationship between the reaction times for targets presented at the different visual field locations. In order to accommodate for multiple comparisons, a more conservative significance cutoff of p = 0.01 was chosen to avoid excessive probability of Type I error. The planned comparisons revealed that all but two of the previously reported perceptual anisotropies occurred significantly in the reaction time data. Reaction time was faster in at locations in the lower half of the visual field than at those in the upper half (F(l,18) = 11.03;p = 0.004; partial n 2 = 0.38), and faster at locations on the right side of the visual field than at those on the left (F(l,18) = 20.31; p < 0.001; partial n 2 = 0.53). Despite the significant difference between performance in the upper and lower visual fields, the difference between reaction times on the upper and lower halves of the vertical meridian was not significant (F(l,18) < 1.0;/? = 0.47; partial n = 0.03). Reaction time was also significantly faster at locations on the cardinal axes than at those on the intercardinal axes (F(l,18) = 12.75;p = 0.002; partial n 2 = 0.42). The difference between performance on the horizontal and vertical meridians was marginally significant, given our more conservative significance cutoff (F(l,18) = 5.36;p = 0.03; partial n 2 = 0.23), with faster reaction times on the vertical meridian (by 7 ms). 3.2.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) also was performed on the error rates. Only a significant main effect of Eccentricity was found (F(3,54) = 4.39; p = 0.02; partial n 2 = 0.20), showing that as distance from fixation increased, the number of errors increased. The mean error rate was 86 3.63%. This pattern of results indicates that there was not a speed-accuracy tradeoff in the data, as the number of errors increased with increasing reaction time. 3.2.3 Discussion The results of the RT analysis for this experiment show that validly cuing the target location does not differentially affect performance at the different visual field locations. Rather, the effect of cuing was uniform across the visual field. Taken alone, these findings imply that endogenous covert orienting is isotropic across the visual field. Differences in performance at the various visual field locations did, however, vary in a manner similar to what has been found previously in regard to perceptual anisotropies. First, across both RT and error analyses, eccentricity anisotropy was consistently present, with both RT and errors increasing with increased distance from fixation. This effect persisted despite equating the target for discriminability using size magnification. In addition, there was evidence for upper-lower visual field anisotropy, with faster performance overall at locations in the lower visual field. Despite this finding, however, there was no evidence for vertical meridian anisotropy: better performance on the lower half of the vertical meridian than the upper half. This latter result implies that the upper-lower visual field anisotropy occurred to a greater extent on the intercardinal axes. Finally, we found longer RTs at intercardinal locations than at cardinal locations. This differs from previous findings (e.g. Carrasco et al, 2004) in which performance at intercardinal locations fell between performance on the horizontal meridian - where performance was best - and the vertical meridian - where performance was worst. 87 Interestingly, we also found evidence for right-left visual field anisotropy, showing faster reaction times for target presented on the right half of the display - in this case, on the right arm of the diameter presented within each block, consistent with a preference of the left hemisphere for letter processing. 3.3 Experiment 2 The results of Experiment 1 seem to indicate that endogenous orienting of attention is isotropic, because precuing any of the 32 locations in the visual field in the experiment had similar effects on RT and errors. Voluntarily allocating one's attention to a particular location in the visual field, however, is not the only response to a relevant precue; the manner in which attention is spread across other possible target locations is also an action of endogenous attention. In Experiment 2, we changed a single aspect of the stimulus display but kept the task identical, in order to demonstrate the way in which the distribution of attention across the visual field can be altered by task demands. In Experiment 2, rather than a fixation cross at the center of the screen, a fixation box, in which a target letter occasionally occurred, appeared instead. This alteration changed nothing about the stimulus characteristics of Experiment 1 for any of the stimuli presented away from fixation. The only alteration was in the way that observers had to distribute their attention in order to perform the task on invalid trials. If the results of Experiments 1 and 2 were to be the same, it would indicate that inhomogeneities in reaction time with respect to target location arise solely from perceptual factors, as has been reported in the exogenous orienting literature (e.g. Carrasco et al, 2001). However, should the pattern of performance between the two experiments differ, the differences 88 would have to arise from factors other than perceptual limitations in different visual field locations. 3.3.1 Methods 3.3.1.1 Observers. Twenty observers (4 male; 16 female) participated in the experiment (median age: 20.5, age range 19 - 34). Two observers reported being left handed; all other observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioural Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. Two observers were rejected from the analysis for overall error rates greater than 7%. 3.3.1.2 Materials. The materials used in Experiment 2 were the same as those used in Experiment 1 (see Figure 3.1), with two exceptions. Instead of a fixation cross at the center of the screen, a placeholder box was located there and presented along with the other eight placeholder boxes. Unlike in Experiment 1, in Experiment 2 a cue and a target could appear at fixation. As in Experiment 1, if a location other than fixation was invalidly cued, the target at any of the four locations to the opposite side of fixation from the cue. Observers were not informed that the target would not appear at fixation if cued away from fixation, and were only instructed that the target might appear in one of the other locations on 30% of the trials. If the fixation location was invalidly cued, the target could appear at any of the other eight placeholder locations on either side of fixation. Also, to accommodate for the one additional cue-target location, 96 trials were presented in each of the 16 blocks (1536 trials; 46 trials per location). 89 3.3.2 Results 3.3.2.1 Reaction time. As for Experiment 1, a three-way A N O V A (Radial location x Eccentricity x Validity) was performed on the reaction time data. A significant main effect of Validity was also found, showing that reaction times were longer for invalidly cued targets (F(l,17) = 8.25;p = 0.01; partial n 2 = 0.33, see Table 3.3). The main effect of Eccentricity was significant (F(3,51) = 59.64; p < 0.001; partial r\ = 0.78, see Table 3.4), with reaction time increasing as eccentricity from fixation increased. Finally, the main effect of Radial location was also significant (F(7,l 19) -7.\7;p< 0.001; partial n 2 = 0.30, see Figure 3.3). As for Experiment 1, planned comparisons were performed to evaluate previously found perceptual anisotropies. The pattern of results largely mirrored those found in Experiment 1. Response time to targets presented in the right visual field were significantly faster than that to targets presented on the left (F(l,17) = 8.53; p = 0.009; partial rf2 = 0.33). Reaction time was only marginally significantly faster in the lower than the upper visual field (F(l,17) = 6.49;p = 0.021; partial n 2 = 0.28), which paralleled the marginally significant difference between the upper and lower halves of the vertical meridian (F(l,17) = 4.83;p = 0.04; partial n 2 = 0.22). Unlike Experiment 1, reaction time was faster to targets on the lower half of the vertical meridian than to those on the upper half. The difference between performance on the horizontal and vertical meridians was also marginally significant (F(l,17) = 5.93;p = 0.02; partial r\ = 0.26), and reaction time was significantly faster to targets at cardinal locations than to those at intercardinal locations (F(l,17) = 19.59;p < 0.001; partial n 2 = 0.54). 90 3.3.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) was performed on the error rates for targets presented at the 32 locations away from fixation. No significant main effects or interactions were found. The mean error rate was 3.4%. 3.3.3 Discussion In Experiment 2, as in Experiment 1, there was no significant interaction between cuing condition (valid vs. invalid) and performance at the various target locations. Performance in the two experiments, however, did differ. The sole addition of a potential target location at fixation induced an interaction between the radial locations and eccentricity. Whereas in Experiment 1, increased eccentricity from fixation consistently resulted in longer reaction times, in Experiment 2 on the right half of the horizontal meridian (E), reaction time was nearly equivalent for targets presented both at 1 and four degrees away from fixation. In fact, reaction time was actually faster for targets presented four degrees from fixation. The pattern of perceptual anisotropies was also not identical between the two experiments. The vertical meridian asymmetry not found in Experiment 1 was marginally significant in Experiment 2, showing faster reactions to targets presented on the lower half of the vertical meridian. The horizontal-vertical meridian asymmetry also reversed, compared to Experiment 1: reaction time was marginally significantly faster on the horizontal meridian than the vertical meridian. This pattern of performance is consistent with previous findings (e.g. Carrasco, et a l , 2002). What is conclusive are the consistent findings of faster reaction time to targets presented in the right half of the visual field, and better performance on the cardinal axes 91 than the intercardinal axes. This is likely to have something to do with the organization of the stimuli on the screen: participants are asked to identify a letter that may appear at eight or nine possible locations, organized in a line across a screen. The contrast between cardinal and intercardinal axes may be thought to highlight the possibility that crossing a meridian is more difficult - for intercardinal locations, the observer must cross two cardinal meridians in order to respond on invalidly cued trials. However, the validity effect was not different between cardinal and intercardinal locations (F(l,17) < 1.0;/? = 0.30; partial n 2 = 0.02) - what was different was the overall RT to respond to targets on these meridians. Therefore, what one could conclude is that endogenously spreading attention along an intercardinal axis is more challenging than spreading one's attention along a cardinal axis. In comparison to previous findings (Carrasco et al., 2004) that indicate that the effects of exogenous orienting at the intercardinal locations are smaller than those found on the horizontal axis, and larger than those found on the vertical axis -essentially intermediary between the two - our pattern of results seems to reflect the endogenous distribution of attention across possible target locations. As for the difference between the right and left visual fields, this pattern clearly mirrors that found in previous literature showing a preference for responding to letter stimuli presented in the right visual field - though no interaction with whether or not the location was precued (e.g. Hardyck, et al., 1985). The effects of the location of the objects in the visual field to which an observer is attending are clearly illustrated by the results outlined above. Distribution of attention across locations on the horizontal and vertical meridians is easier than distribution of attention across locations on either of the intercardinal meridians. Therefore, although 92 there are no significantly different effects of cuing a target location in this paradigm, this work is the first to illustrate that the configuration and visual field location of potential target locations has consequences for the ability to attend to those locations. 3.4 Experiment 1 & 2 comparison 3.4.1 Reaction time In order to make a direct comparison between performance in Experiments 1 and 2, a 4-way between-within A N O V A was performed on the RT data from both experiments (Experiment x Radial Location: NE, E, SE, S, SW, W, NW, N x Eccentricity: 1°, 4°, 8°, or 12° x Validity: Valid vs. Invalid). Results revealed three interesting interactions. First, a significant interaction occurred between Radial Location, Eccentricity, and Experiment (F(21, 735) = 1.75;/? = 0.02; partial n 2 = 0.05; see Figure 3.4), showing that the primary difference in performance between the two experiments occurred 1 0 away from fixation. The second interaction gives an additional view of the three-way interaction between Radial Location, Eccentricity, and Experiment: the two-way interaction of Radial Location and Experiment was also significant (F(7,245) = 3.64; p =.00009; partial n 2 = 0.09). As can been seen in Figure 3.5, the effects of having a target at fixation are smallest on the vertical meridian (S and N), and in the lower right visual field (SE). This analysis also showed three significant main effects. The significant main effect of Radial Location (F(7, 245) = 14.16;/? < 0.001; partial n 2 = 0.29) showed that, overall, performance was best on the right half of the horizontal meridian, and the lower half of the vertical meridian. The significant main effect of Eccentricity (FQA05) = 112.50; p < 0.001; partial r\ = 0.76) showed that as distance from fixation increased, so 93 did reaction time. Finally, the main effect of Validity was significant (F(l,35) = 20.03; p < 0.001; partial r\2 = 0.36), with faster reaction time for validly cued targets. Although the main effect of Experiment was not significant, performance in Experiment 2 was consistently faster than that in Experiment 1 at the possible target locations away from fixation (mean difference: 15 ms). 3.4.2 Errors A four-way between-within A N O V A was also performed on the error data from both experiments (Experiment x Radial Location: NE, E, SE, S, SW, W, NW, N x Eccentricity: 1°, 4°, 8°, or 12° x Validity: Valid vs. Invalid). There was only a significant main effect of Eccentricity (F(3,105) = 3.56; p = 0.02; partial n 2 = 0.09), showing that more errors were committed for targets presented farther from fixation. 3.4.3 Discussion A direct comparison between Experiments 1 and 2 illustrates how endogenous attention orienting affects reaction time to stimuli in different locations in the visual field. Here, the results show that the biggest difference between the two experiments lay in RT effects closest to fixation: when a target could appear at fixation, reaction time to nearby targets - whether their location was validly cued or not - was shorter than if there was no possibility that a target would appear at fixation. This indicates that some attention was distributed around fixation to nearby locations but only when it was a possible target location. Interestingly, the interaction between Radial Location and Experiment shows this advantage of having a potential target at fixation on RT at nearby locations was diminished in three of the eight radii: on the vertical meridian (S and N), and in the lower 94 right half of the visual field (SE). In fact, the greatest effects of having a potential target at fixation occured on the horizontal meridian, with a mean difference between the experiments across all four eccentricities of 25 ms on the left half of the horizontal meridian (W), and 12 ms on the right half of the horizontal meridian (E). What these results seem to indicate is an interaction between the well-documented perceptual anisotropies and the voluntary distribution of attention across the display as a function of the characteristics of the stimuli. While we did not find a significant horizontal-vertical anisotropy within each experiment on its own, a comparison between performance on the horizontal and vertical meridians did show a significant difference when the data from both experiments were combined (V7(l,36) = 10.6;p = 0.002; partial n = 0.23), with shorter RTs on average on the horizontal meridian. This is primarily because of longer RTs to targets presented on the upper half of the vertical meridian -results which mirror the marginally significant difference between the upper and lower halves of the vertical meridian found in Experiment 2. Where the effects of a potential target location at fixation are seen most clearly are on the meridian which is known to receive preferential perceptual processing, the horizontal. The greatest impact of a potential target at fixation is seen on the left half of the horizontal meridian where, in this particular task, there is a decline in performance. This implies that placing a potential target location at fixation facilitates the distribution of attention across possible target locations to the greatest degree in a visual field location that is well known to be advantageous for perceptual processing. 95 3.5 General discussion Does voluntary, covert allocation of attention vary in its impact on performance at different visual field locations? What we have seen from the present experiments is that while the effects of a precue somewhere along a diameter of the visual field do not differ in this particular paradigm, the manner in which attention is distributed across the diameter, which is constrained by the display parameters, does interact with visual field location. 3.5.1 Perceptual and attentional anisotropies The results of both of our experiments replicate several known patterns of perceptual anisotropy. First, as distance from fixation increases, reaction time to targets increases, and accuracy declines: a clear replication of eccentricity anisotropy. Interestingly, this result was found despite scaling the stimuli to equate for the decreasing visibility of the target letters with increasing eccentricity. This indicates that the scaling method chosen may not have been entirely effective, leaving the stimuli farther from fixation more difficult to perceive than those closer to fixation. It is also possible, however, that this effect arose from anisotropy of exogenous orienting dependent upon distance from fixation, similar to previous findings (e.g. Carrasco et al., 2002). Whether the scaling was ineffective, or exogenous orienting created the eccentricity anisotropy, could not influence the other perceptual anisotropies we found, as the stimuli were scaled only for eccentricity, and not for radial location. The results of both experiments showed an advantage for responding to stimuli presented in the lower visual field compared to the upper visual field, and the right visual field compared to the left. Additionally, when examined across both experiments, the RT difference between 96 the horizontal meridian and vertical meridian was significant, showing a clear horizontal-vertical asymmetry. An additional anisotropy of performance was seen between the cardinal and intercardinal axes. RTs to targets presented on the intercardinal meridians (NE/SW and SE/NW) were consistently slower than RTs to targets presented on the cardinal meridians (N/S, E/W). However, there is reason to believe this is an attentional effect, not a perceptual one, as there is no evidence, to the authors' knowledge, of differing representation of the intercardinal axes in either the retina or the brain. Therefore, we believe that this work also is the first to demonstrate what appears to be an attentional oblique effect. Similar to the fact that contrast sensitivity declines more rapidly in the periphery for gratings that are oriented away from the horizontal or vertical axes (e.g. Pointer, 1996), our findings demonstrate that when possible stimulus locations are distributed in a linear fashion along a meridian, response time on the intercardinal axes suffers (e.g. NE/SW, SE/NW). Can this result be attributed to the meridian effect (e.g. Hughes & Zimba, 1985, 1987)? In our experiments, observers were asked to shift their attention to a cued location in the display; on the majority of trials, the target appeared there. However, for 30% of the trials, the target appeared on the opposite side of fixation (or at fixation in Experiment 2) in the display. When presented with a block in which possible stimulus locations were distributed along a diagonal meridian, on invalidly cued trials observers were required shift their attention across both the horizontal and vertical meridians in order to respond accurately. As described previously, however, effects of cuing along the intercardinal axes were no greater than those along the cardinal axes. Rather, the RTs to both validly 97 and invalidly cued targets along an intercardinal axis were longer than those to targets presented along a cardinal axis. Also as we noted earlier, under conditions of exogenous orienting RTs to targets presented on the intercardinal axes are between those to targets on the horizontal and vertical axes, and exogenous cuing also has been found to have effects whose magnitude lies between those for the horizontal and vertical axes (Carrasco et al., 2004). If the attentional oblique effect were an effect of exogenous orienting arising from the sudden onset of the target, we would expect to find similar to results to those of Carrasco and colleagues (2004). As we did not find effects similar to those of Carrasco and colleagues (2004), we reason that the attentional oblique effect is likely to arise from the action of an endogenous attentional mechanism that is driven by the manner in which observers must distribute their attention across possible target locations. Essentially, spreading one's attention across a diagonal array seems to be more effortful than doing so across an array along either the horizontal or vertical axis. Because the target was validly cued on only 70% of trials, in order to effectively perform the task, observers had to distribute some attention to other locations along the meridian, in case the target location was invalidly cued (e.g. LaBerge, 1995; LaBerge et al, 1997). Importantly, in light of these findings, we performed an additional analysis of a previous study employing the identical task, target stimuli, cue stimuli, and locations, for which we had initially no theoretical reason to anticipate a difference in performance. We found a marginally significant intercardinal vs. cardinal meridian RT difference in a paradigm where observers had to spread their attention across an entire circular display spanning 24° of visual angle (p = 0.03, using a more conservative significance cut-off to account for potential excessive family-wise 98 error inherent in multiple comparisons; Roggeveen & Ward, submitted). It seems, therefore, that the significant pattern of slower RT for intercardinal locations in the present experiments must be driven by the manner in which the present stimulus display was configured. We conclude that an attentional oblique effect will be found under conditions where observers must spread their attention across an entire meridian of the visual field: performance will be superior along either of the cardinal axes compared to that along either of the intercardinal, or possibly other diagonal, axes. Among all of these anisotropies - perceptual and attentional - we failed to find a simple anisotropic effect of cuing at the various locations in the display. Other work in our laboratory (Roggeveen & Ward, submitted) has demonstrated that attentional anisotropies can be found in a different paradigm, with consistently larger cuing effects in the lower visual field than in the upper visual field, and some evidence for both horizontal-vertical meridian asymmetry and vertical meridian asymmetry interacting with endogenous orienting. We did not find such interactions between cue condition and visual field location in the present experiments. As with the attentional oblique effect, where the parameters of the display seem to determine the manner in which attention is allocated throughout the display, it may be that the attentional anisotropies we found previously appear only under more demanding circumstances. 3.5.2 Effects of attentional distribution Distribution of attention across diagonally arranged locations in the visual field appears to be more difficult than that along one of the cardinal axes. What impact does the manner in which observers distribute attention among meridional locations have on performance, and is that impact anisotropic? The comparison between Experiments 1 and 99 2 - where the only difference was the addition of a possible cue-target location at fixation - reveals that the manner in which observers choose to distribute their attention across possible target locations interacts with the set of relevant visual field locations. The interaction appears in two results. First, when there is the possibility that a target will appear at fixation, RTs on both validly cued trials and on invalidly cued trials are faster for targets presented closer to fixation. Returning to Dori and Henik's (2006) model, which postulates two gradients of attention - one that facilitates responding at the cued location, and one that inhibits responding at uncued locations, with the inhibitory area centered at fixation, it appears that placing a possible cue-target location at fixation disrupts the inhibitory center. Our results indicate that the manner in which attention is distributed - both in terms of its inhibitory and excitatory mechanisms - is dictated by the demands of the stimulus situation with which the observer is presented. Given a task where no stimulus is ever presented at fixation (e.g. Dori & Henik, 2006), or one in which a stimulus is presented at fixation but the observer must ignore it in order to perform the task (e.g. Beck & Lavie, 2005), inhibiting fixation - where it is argued that attentional access is facilitated (Beck & Lavie, 2005) - is clearly beneficial to performing the task. Given a task that requires attending to fixation simultaneously with possible target locations in the periphery, however, maintaining an inhibitory gradient centered at fixation would likely impair performance - and also require greater attentional effort on the part of the observer. Our finding of shorter RTs near fixation when a cue or target is possible there demonstrates the flexibility of the mechanism that creates attentional gradients. 100 The comparison of Experiment 1 and 2 also shows that introducing a possible target at fixation interacts with processing of targets presented in other visual field locations. Although no overall cuing effect was found, this interaction implicates the manner in which attention is endogenously distributed in the display in anisotropic performance. The greatest differences between the experiments were found on the horizontal meridian, driven by a large difference to the left of fixation (25 ms). On the other meridians, the difference between the two experiments averaged across all eccentricities was much smaller, in some cases negligible. Reducing or eliminating the inhibitory gradient at fixation may have larger effects on the horizontal meridian -particularly the left side - than anywhere else because of the well-established advantage of the horizontal meridian for perceptual processing. Removing inhibition improves performance to the greatest extent in the areas where performance is already advantaged. The fact that the performance difference is largest on the left half of the horizontal meridian is likely to be a result of an interaction with faster responses to target letters presented on the right side of the visual field: eliminating inhibition at fixation has little effect at those target locations, as performance is already superior there. Overall, the results of the present experiments demonstrate three interesting effects. First, the demands of the display dictate the manner in which attention is voluntarily distributed within a scene. Placing a potential target at fixation may disrupt any inhibitory gradient that would naturally develop given tasks where it is unnecessary or unhelpful to attend to fixation. Second, the effect of this disruption of the inhibitory gradient on performance is not consistent across visual field locations. Although there is a general facilitation of RT close to fixation, the facilitation is not equivalent on all four 101 meridians investigated in this experiment. The largest effect of removing the inhibitory gradient at fixation - or, without relying on Dori and Henik's (2006) theoretical approach, the largest effect of changing attentional distribution - is found in a location where the perceptual processing is known to be advantageous. Finally, given conditions where observers must voluntarily distribute their attention along locations arranged in a diagonal orientation across a relatively wide area of space, an attentional oblique effect will emerge. Performance will be consistently superior on the cardinal axes due to facilitation of endogenous attentional distribution on the horizontal and vertical axes. 102 3.6 Tables and Figures - Chapter 3 Table 3.1. Main effect of Validity: Experiment 1 Cue condition Reaction time (standard error) Valid 539 ms (14 ms) Invalid 570 ms (16 ms) Table 3.2. Main effect of Eccentricity, Experiment 1. Distance from fixation (degrees of visual angle) Reaction time (standard error) 1° 539 ms (15 ms) 4° 546 ms (14 ms) 8° 560 ms (14 ms) 12° 574 ms (14 ms) 104 Table 3.3. Main effect of Validity, Experiment 2. Cue condition Reaction time (standard error) Valid 519 ms (14 ms) Invalid 560 ms (15 ms) Table 3.4. Main effect of Eccentricity, Experiment 2. Distance from fixation (degrees of visual angle) Reaction time (standard error) 1° 521 ms (13 ms) 4° 529 ms(13 ms) 8° 543 ms(12ms) 12° 563 ms (12 ms) 106 Figure 3.1. Schematic of display sequence in Experiments 1 and 2. Exper iment 1 Cue: all but cued location Cue fades off over 416 ms fades off over 416 ms valid invalid Target: 150 ms time Exper imen t 2 valid Cue: all but cued location Cue fades off over 416 ms fades off over 416 ms ISI: 200-400 ms time 107 Figure 3.2. Main effect of Radial location, Experiment 1. 108 Figure 3.3. Main effect of Radial location, Experiment 2. Main effect of Radial Location: Experiment 2 Error bars indicate 95% confidence interval 570 r 560 • -w 550 -109 Figure 3.4. Interaction between Experiment, Radial location, and Eccentricity: comparison between Experiments 1 and 2. Exper imentxRadia l locationxEccentricity: Experiments 1 & 2 F(21. 735) = 1.75, p = 0.02 Error bars denote 95% confidence intervals 840 6 2 0 6 0 0 "2 5 8 0 o o (V •M 5 6 0 I 540 c o " 520 ac 500 480 460 V 4 :4 I 4 3E Experiment 1 1 4 8 1 2 1 4 8 1 2 1 4 8 1 2 1 4 8 1 2 1 4 8 1 2 1 4 8 1 2 1 4 8 1 2 1 4 812 ~jfZ Experiment 2 N E E S E S S W W NW N 110 Figure 3.5. Interaction between Experiment and Radial location: comparison between Experiments 1 and 2. 580 Experiment x Radial location interaction Error bars denote 95% confidence interval 570 - 5 6 0 E, | 550 540 c o o CO CD * 530 520 510 1 / V\ / / / / / / / / / / / / i l Q-X LLI CM T -Q. Q-X X L U L U LLI QJ LU CM T - CN Q . Q. Q . X X X LU LU LLI i i i LU LU LU CD CO Q . X L U CM t~ Q. Q. X X LU LU CO CO CO' CN T -CL Q. X X LU LU i I CO CM i -Q. Q. X X LU UJ CN T - CM Q . C L Q. X X X LU LU LU £ <: £ 2 z i n 3.7 References Altpeter, E., Mackeben, M . , & Trauzettel-Klosinski, S. (2000). The importance of sustained attention for patients with maculopathies. Vision Research, 40, 1539-1547. Anstis, S. (1974). A chart demonstrating variations in acuity with retinal position. Vision Research, 14, 589-592. Ashkenazi, A. , & Marks, L.E. (2004, October). Object-based attention in the visual field. Poster presentation at Fechner Day, Coimbra, Portugal. Beck, D .M. , & Lavie, N . (2005). Look here but ignore what you see: effects of distractors at fixation. Journal of Experimental Psychology: Human Perception and Performance, 31(3), 592 - 607. Bilodeau, L. , & Faubert, J. (1997). The oblique effects and chromatic motion throughout the visual field. Vision Research, 39, 757-763. Cameron, E X . , Tai, J.C., & Carrasco, M . (2002). Covert attention affects the psychometric function of contrast sensitivity. Vision Research, 42, 949-967. Carrasco, M . & Frieder, K.S. (1996). Cortical magnification neutralized the eccentricity effect in visual search. Vision Research, 37(1), 63 - 82. Carrasco, M . , Giordano, A . M . , & McElree, B. (2004). Temporal performance fields: visual and attentional factors. Vision Research, 44, 1351-1365. Carrasco, M . , McElree, B., Denisova, K. , & Giordano, A . M . (2003). Speed of visual processing increases with eccentricity. Nature Neuroscience, 6(7), 699 - 700. 112 Carrasco, M . , Talgar, C , & Cameron, E.L. (2001). Characterizing visual performance fields: Effects of transient covert attention, spatial frequency, eccentricity, task, and set size. Spatial Vision, 15, 61-75. Carrasco, M . , Williams, P.E., & Yeshurun, Y . (2002). Covert attention increases spatial resolution with or without masks: Support for signal enhancement. Journal of Vision, 2, 467-479. Cheal, M . , & Lyon, D.R. (1991). Central and peripheral precuing of forced-choice discrimination. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 43 A, 859-880. Chen, H. , Yao, D., & Liu, Z. (1994). A study on asymmetry of spatial visual field by analysis of the fMRI B O L D response. Brain Topography, 17(1), 39 - 46. Connolly, M . & Van Essen, D. (1984). The representation of the visual field in parvicellular and magnocellular layers of the lateral geniculate nucleus in the macaque monkey. Journal of Comparative Neurology, 226(4), 544 - 564. Cowey, A. , & Rolls, E.T. (1974). Human cortical magnification factor and its relation to visual acuity. Experimental Brain Research, 21, 447 - 454. Curcio, C.A., & Allen, K . A . (1990). Topography of ganglion cells in the human retina. Journal of Comparative Neurology, 300, 5 -25 . Dori, H. , & Henik, A. (2006). Indications for two attentional gradients in endogenous visual-spatial attention. Visual Cognition, 13(2), 166-201. Edwards, M . , & Badcock, D.R. (1993). Asymmetries in the sensitivity to motion in depth: a centripetal bias. Perception, 22, 1013-1023. 113 Evert, D.L., McGlinchey-Berroth , R., Verfaellie , M . , & Milberg, W.P. (2003). Hemispheric asymmetries for selective attention apparent only with increased task demands in healthy participants. Brain and Cognition, 53, 34 - 41. Fink, G.R., Halligan, P.W., Marshall, J.C., Frith, C.D., Frakowiak, R.S.J., & Dolan, R.J. (1996). Where in the brain does visual attention select the forest and the trees? Nature, 382, 626 - 628. Fukisama, S.S., & Faubert, J. (2001). Perceived length in the central visual field: evidence for visual field asymmetries. Vision Research, 41, 2119-2126. Gawryszewski, L. , Riggio, G. de, L. , Rizzolatti, G., & Umilta, C. (1987). Movements of attention in the three spatial dimensions and the meaning of "neutral" cues. Neuropsychologic, 25, 19- 29. Gobell, J.L., Tseng, C , & Sperling, G. (2004). The spatial distribution of visual attention. Vision Research, 44, 1273 - 1296. Hagenbeek, R.E., & Van Strien, J.W. (2002). Left-right and upper-lower visual field asymmetries for face matching, letter naming, and lexical decision. Brain and Cognition, 49, 34 - 44. Hardyck, C , Chiarello, C., Dronkers, N . , & Simpson, G. (1985). Orienting attention within visual fields: How efficient is interhemispheric transfer? Journal of Experimental Psychology: Human Perception and Performance, 11, 650 - 666. He, S.H., Cavanagh, P., & Intriligator, J. (1996). Attentional resolution and the locus of visual awareness. Nature, 383, 334-337. Heilman, K . M . , & Valenstein, E. (1979). Mechanisms underlying hemispatial neglect. Annals of Neurology, 5, 166-170. 114 Helmholtz, H. von. (1867; 1924). Treatise on Physiological Optics. Translated by J.P.C. Southall, The Optical Society of America. Hughes, H.C., & Zimba, L.D. (1985). Spatial maps of directed attention. Journal of Experimental Psychology: Human Perception and Performance, 11(4), 409 -430. Hughes, H.C., & Zimba, L.D. (1987). Natural boundaries for the spatial spread of directed visual attention. Neuropsychologia, 25(1A), 5-18. Intriligator, J., & Cavanagh, P. (2001). The spatial resolution of visual attention. Cognitive Psychology, 43, 171-216. Jonides, J. (1981). Voluntary vs. automatic control over the mind's eye's movement. In J. Long & A. Baddeley (Eds). Attention & Performance IX (pp. 187-203) Hillsdale, NJ: Erlbaum. Kinsbourne, M . (1970). The cerebral basis of lateral asymmetries in attention. Acta Psychologica, 33, 193-201. Kitterle, F.L, Christman, S., & Hellige, J.B. (1990). Hemispheric differences are found in the identification, but not the detection, of low versus high spatial frequencies. Perception and Psychophysics, 48(4), 297 - 306. Kraft, A. , Pape, N . , Hagendorf, H. , Schmidt, S., Naito, A. , Brandt, S.A. (2007). What determined sustained visual attention? The impact of distracter positions, task difficulty and visual fields compared. Brain Research, 1133, 123 - 135. Levi, D .M. , Klein, S.A., & Aitsebaomo, A.P. (1985). Vernier acuity, crowding, and cortical magnification. Vision Research, 25(7), 963 - 977. 115 Levine, M.W., & McAnany, J.J. (2005). The relative capabilities of the upper and lower hemifields. Vision Research, ¥5 ,2820-2830 . Lindell, A . K . , & Nicholls, M.E.R. (2003). Attentional deployment in visual half-field tasks: The effect of cue position on word naming latency. Brain and Cognition, 55,273-277. Liu, T., Heeger, D.J., & Carrasco, M . (2006). Neural correlates of the visual vertical meridian asymmetry. Journal of Vision, 6, 1294 - 1306. Losier, B.J., & Klein, R . M . (2004). Covert orienting within peripersonal and extrapersonal space: young adults. Cognitive Brain Research, 19, 269 - 274. Mackeben, M . (1999). Sustained focal attention and peripheral letter recognition. Spatial Vision, 12, 51-72. Maunsell, J.H., & Van Essen, D.C. (1987). Topographic organization of the middle temporal visual area in the macaque monkey: representational biases and the relationship to callosal connections and myeloarchitectonic boundaries. Journal of Comparative Neurology, 266(4), 535-555. Michael, G.A., & Ojeda, N . (2005). Visual field asymmetries in selective attention: Evidence from a modified search paradigm. Neuroscience Letters, 388, 65 - 70. Michimata, C , & Hellige, J. (1987). Effects of blurring and stimulus size on the lateralized processing of nonverbal stimuli. Neuropsychologia, 25, 397-407. Mondor, T.A., & Bryden, M.P. (1992). On the relation between visual spatial attention and visual field asymmetries, The Quarterly Journal of Experimental Psychology Section A, 44(3), 529-555. 116 Perry, V . H . , & Cowey A. (1985.) The ganglion cell and cone distributions in the monkey's retina: implications for central magnification factors. Vision Research, 25, 1795- 1810. Pointer, J.S. (1996). Evidence of a global oblique effect in human extrafoveal vision. Perception, 25(5), 523 - 530. Posner, Snyder & Davidson (1980) Attention and the detection of signals. Journal of Experimental Psychology, 109, 160-74. Portin, K., Vanni, S., Virsu, V. , & Hari, R. (1999). Stronger occipital cortical activation to lower than upper visual field stimuli. Experimental Brain Research, 124,2%1 -294. Previc, F.H. (1990). Functional specialization in the lower and upper visual fields in humans: Its ecological origins and neurophysiological implications. Behavioral and Brain Sciences, 13, 519-575. Previc, F. (1998). The neuropsychology of 3D space. Psychological Bulletin, 124, 123 -164. Proverbio, A . M . , Zani, A. , & Avella, C. (1997). Hemispheric asymmetries for spatial frequency discrimination in a selective attention task. Brain and Cognition, 34, 311 -320. Rijsdijk, J.P., Kroon, J.N., & van der Wildt, G.J. (1980). Contrast sensitivity as a function of position on the retina. Vision Research, 20, 235-241. Rizzolatti, G., Riggio, L. , Dascola, I., & Umilta, C. (1987). Reorienting attention across the horizontal and vertical meridians: evidence in favor of a premotor theory of attention. Neuropsychologia, 25(1 A), 31-40. 117 Robertson, L., Lamb, M.R., & Knight, R.T. (1988). Effects of lesions of temporalparietal junction on perceptual and attentional processing in humans. Journal of Neuroscience, 5(10), 3757-3769 Roggeveen, A .B . , & Ward, L . M . (submitted). Anisotropy of covert endogenous orienting across the visual field. Journal of Experimental Psychology: Human Perception and Performance. Rovamo, J., & Virsu, V . (1979). An estimation and application of the human cortical magnification factor. Experimental Brain Research, 37, 495-510. Rubin, N . , Nakayama, K., & Shapley, R. (1996). Enhanced perception of illusory contours in the lower versus upper visual hemifields. Science, 271, 651-653. Sakaguchi, Y . (2003). Visual field anisotropy revealed by perceptual filling-in. Vision Research, 43, 2029-2038. Shepherd, M . , & Muller, H.J. (1989). Movement versus focusing of visual attention. Perception and Psychophysics, 46, 146-154. Talgar, C P . , & Carrasco, M . (2002). Vertical meridian asymmetry in spatial resolution: Visual and attentional factors. Psychonomic Bulletin & Review, 9, 714-722. Tootell, R.B., Switkes, E., Silverman, M.S., & Hamilton, S.L. (1988). Functional anatomy of macaque striate cortex. II. Retinotopic organization. Journal of Neuroscience, 8, 1531 - 1568. Van Essen, D.C., Newsome, W.T., & Maunsell, J.H. (1984). The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability. Vision Research, 24, 429 - 448. 118 Verfaellie, M . , Bowers, D., & Heilman, K . M . (1988). Hemispheric asymmetries in mediating intention, but not selective attention. Neuropsychologic!, 26, 521 - 531. Wright, R.D., & Ward, L . M . (1998). The control of visual attention. In Visual Attention, R.D. Wright, Ed. Oxford: New York, 132-186. Wright, R.D. & Ward, L . M . (in press). Orienting of attention. New York: Oxford University Press. Yantis, S., & Jonides, J. (1990). Abrupt visual onsets and selective attention: voluntary versus automatic allocation. Journal of Experimental Psychology: Human Perception and Performance, 16, 121 - 134. 119 4 The impact of paradigm parameters on the allocation of endogenous attention: the role of objects and likelihood4 4.1 Introduction Voluntary, covert allocation of visual attention has been demonstrated to be more than just a unitary beam of selection. Rather, there is mounting evidence (e.g., Gobell, Tseng, & Sperling, 2004; Roggeveen & Ward, submitted, a) that the manner in which attention is voluntarily allocated in the visual field changes in the face of the demands of the task. Ultimately, task demands alter the impact of attention on performance. Work in our laboratory (Roggeveen & Ward, submitted, a) has shown that there is evidence for anisotropic, or heterogeneous, effects of attention across the visual field - and that the anisotropy is affected by what, exactly, the subject has been asked to do. Several factors have an impact on the manner in which attention is allocated to various visual field locations. One such factor is the likelihood that a target will appear at a cued location, or the cue's predictive value. An additional influence is the presence of a perceptual object or placeholder that may be used to aid in attentional selection of a specific location. Both of these factors may play different, but intersecting, roles in the effects of attention orienting at different locations in the visual field. Likelihood effects and attentional distribution Visual attention is thought to be oriented to locations in space via two processes: exogenous orienting, where attention is allocated involuntarily to a cued location, and endogenous orienting, where attention is moved voluntarily (e.g. Wright & Ward, 1998). Whereas exogenous orienting has a shorter time course, showing the greatest effect 4 A version of this chapter has been submitted for publication. Roggeveen, A.B. , & Ward, L . M . (submitted). The impact of paradigm parameters on the allocation of endogenous attention: the role of objects and likelihood. 120 around 50 - 100 ms after the onset of a cue and quickly deteriorating (e.g. Cheal & Lyon, 1991; Wright & Ward, 1998), endogenous orienting takes longer to reach its full impact on performance, gradually increasing over roughly 500 ms (Shepherd & Muller, 1989). In both cases, attention works to speed responses and improve accuracy for targets that appear at an attended location. It has been shown that the more likely a target is to appear in a cued location, the greater is the amount of attention allocated to the cued location. Gottlob, Cheal, and Lyon (1999) varied the predictive value of a peripherally-presented direct cue, used to elicit exogenous shifts of attention, between 100%, 75%, and 50% validity. Interestingly, though the initial shifts of attention were involuntary given the time course and cue type, discrimination accuracy was better for targets at validly-cued locations when cue validity was higher. Also using an exogenous cue, Johnson and Yantis (1995) demonstrated that cue validity affected performance: reaction times were faster for targets appearing at a location that was cued with 100% validity, as compared to targets appearing at a location that was cued with only 50% validity. Though these tasks induced an involuntary shift of attention, clearly there is an endogenous, or voluntary, component to the manner in which attention was distributed throughout the display - else no matter what the predictive value of the cue was, performance would always be the same if the location was validly cued. Roggeveen and Ward (submitted, a) also showed explicitly that the likelihood that a cue correctly indicates target position has effects in an endogenous cuing task. Comparing target locations that were cued with 100% likelihood and target locations that were validly cued only 70% of the time, response times were universally faster for 100% validly cued 121 targets. These results, compared with the general pattern of performance of the two conditions, indicated that in the experimental condition where the cue's predictive value was less than 100%, some attention was also simultaneously distributed across the uncued possible target locations. This result is consistent with the findings of Johnson and Yantis (1995), who demonstrated that the attention was distributed in parallel to all target locations when an exogenous cue was only 50% predictive. The effects of cue predictive value on the impact of cuing a target location, however, do deserve further consideration. Roggeveen and Ward (submitted, a) found that the manner in which observers distributed their attention among possible target locations given a lower-than-100% predictive cue value was driven to a large degree by the most advantageous approach, given the characteristics of the stimulus array. Observers were presented with a large array of possible target locations (32) which were located at four eccentricities (1°, 4°, 8°, and 12° of visual angle) at eight possible radial locations (NE, E, SE, S, SW, W, NW, N) across the display, only one of which was cued. Distribution of attention across all possible uncued locations would be a demanding task, especially given the relatively low likelihood that the target would not appear at one of them (30%, or slightly less than 1% per uncued location). What observers did, then, was to distribute their residual attention, the attentional resource that wasn't allocated to the cued location, in a way that hedged their bets, by focusing most of it in a ring-like configuration at locations at 4° and 8° from fixation, in the middle of the possible, uncued target locations. The resulting anisotropy of performance varied according to task characteristics, making it clear that the effects of endogenous attention vary according to the task demands. Given the large number of possible target locations and only a single 122 manipulation of cue predictive value (70% likelihood), however, it proved to be impossible in that study to fully characterize how predictive validity impacts the distribution of attention across a display. In the present set of experiments we study this effect further. 4.1.1 Perceptual and attentional anisotropy Task demands also interact with the way in which the visual system preferentially processes various visual field locations-. There are several known perceptual anisotropies, where, due to the design of the visual system, certain areas of the visual field are more useful for the perception of most types of stimuli. In addition to perceptual anisotropies, there is also mixed evidence that these perceptual anisotropies map onto attentional anisotropies: instances where the effects of attention are larger in certain areas of the visual field than in others. These perceptual and attentional anisotropies will be briefly reviewed here; a more comprehensive review of perceptual and attentional anisotropies can be found in Roggeveen and Ward (submitted, a). One of the most obvious and most studied of the perceptual anisotropies is eccentricity anisotropy: as a stimulus is presented farther from fixation, the ability to perceive that stimulus declines. This is largely due to the cortical magnification factor, where the representation of areas of the visual field by the visual system is reduced in an approximately linear manner as distance from fixation increases (e.g. Anstis, 1974). In terms of attention, in the exogenous orienting literature it has been shown that an exogenous cue has a greater impact on the quality of target processing with greater distance from fixation (Carrasco, Williams, & Yeshurun, 2002; Talgar & Carrasco, 2002). With an endogenous cue, however, the pattern of results differs: the effects of 123 cuing seem to be greatest both closest to and farthest from fixation when the cue indicates with 70% accuracy where the target will appear. Therefore, these latter effects appear to be due to the way that attention is voluntarily distributed across possible target locations, rather than to an inherent ability to endogenously attend to one distance from fixation more effectively than to another (Roggeveen & Ward, submitted, a). Another perceptual anisotropy similarly based in the structure of the visual system is vertical anisotropy: perceptual processing is generally better in the lower visual field than in the upper visual field (e.g. Previc, 1990; 1998). Representation of the lower visual field is greater starting in the retina and continuing into cortical areas (e.g. Connolly & Van Essen, 1984; Perry & Cowey, 1985). Greater representation results in better performance on many perceptual tasks, including line length estimation (Fukisama & Faubert, 2001), and contrast sensitivity (Carrasco, Talgar, & Cameron, 2001), to name only two. Data regarding whether attention has greater effects in the lower visual field have been equivocal. While some researchers have not found a consistent lower visual field advantage due to attention (e.g. Altpeter, Mackeben, & Trauzettel-Klosinski, 2000; Mackeben, 1999), there have also been reports that attention has greater effects in the lower visual field under both exogenous (Gawryszewski et al., 1987) and endogenous (He, Cavanagh, & Intriligator, 1996; Intriligator & Cavanagh, 2001; Losier & Klein, 2004; Roggeveen & Ward, submitted, a) cuing conditions. A subset of vertical anisotropy is vertical meridian asymmetry, where performance on the lower half of the vertical meridian is distinctly better than performance on the upper half. This pattern of performance has been reported by 124 researchers who failed to find consistent perceptual advantages at all locations in the lower visual field, but did find advantages only on the vertical meridian (e.g. Cameron, Tai, & Carrasco, 2002; Carrasco et al., 2001. Mirroring the larger representation of the lower visual field generally, it has also been shown that stimuli presented to the lower half of the vertical meridian elicit a larger response in areas V I and V2 than stimuli presented to the upper half (Liu, Heeger, & Carrasco, 2006). This pattern of performance has not, however, been mirrored in exogenous attention cuing experiments (Talgar & Carrasco, 2002), and was inconsistent in endogenous cuing experiments (Roggeveen & Ward, submitted, a). Also clearly tied to the underlying neural architecture is horizontal-vertical anisotropy, where performance on the horizontal meridian is consistently better than performance on the vertical meridian (e.g. Carrasco, et al., 2001; Rovamo & Virsu, 1979). Interestingly, eccentricity anisotropy is more pronounced on the vertical meridian, with a faster fall-off in performance with increased eccentricity (Carrasco, Williams, & Yeshurun, 2002). Anisotropic effects of attention orienting that parallel this perceptual anisotropy can be found in both the exogenous-cuing literature (Carrasco et al., 2004), and the endogenous-cuing literature (Roggeveen & Ward, submitted, a). The effects of both types of orienting are greatest on the vertical meridian, indicating that both processes may serve to compensate for perceptual deficits. Two additional anisotropic patterns of performance appear to be more closely tied to attentional effects, rather than perceptual effects. Right-left asymmetry, where the processing of certain types of stimulus, such as high spatial frequencies (e.g., Kitterle, Christman, & Hellige, 1990) and local structure of stimuli (e.g. Fink et al., 1996) is better 125 in the right visual field than the left, it thought to originate in more effective attention orienting to these types of stimulus characteristics in the right visual field (e.g. Heilman & Valenstein, 1979; Kinsbourne, 1970). However, the effects of cuing a target location have not been found to be better in the right visual field (e.g. Kraft et al., 2007; Roggeveen & Ward, submitted, a; b; Verfaellie, Bowers, & Heilman, 1988). This indicates that there is only a general preference for identifying stimuli presented in the right visual field (Roggeveen & Ward, submitted, b). As Roggeveen and Ward (submitted, a; b) used letter stimuli, this finding may be due to the preferential processing of high spatial frequencies and local structure. Finally, Roggeveen and Ward (submitted, b) reported an attentional oblique effect, where performance was generally poorer on the diagonal meridians crossing fixation than on either the horizontal or vertical meridians. As previous work has shown that the effects of exogenous cuing for location at intercardinal locations fall between those found on the horizontal axis (best performance) and the vertical axis (worst performance; Carrasco et al., 2004), this result was interpreted to be a difficulty in voluntarily distributing attention along diagonal axes. Although many of the perceptual anisotropies are mirrored by attentional anisotropies, there is reason to believe that the effects of endogenous orienting are also greatly affected by the demands of the task, rather than operating in the same manner regardless of what the observer is asked to do. As mentioned above, the effects of an endogenous cue at various eccentricities varied with the cue's predictive validity, rather than always having a consistent effect across eccentricities. An additional case where task demands altered attentional distribution comes from other recent work in our laboratory 126 (Roggeveen & Ward, submitted, b). Observers were asked to perform a letter identification task at one of eight possible locations, presented along one of either the cardinal or intercardinal meridians on either side of fixation. Although the simple effects of cuing were not anisotropic in this experiment, the addition of a potential target location at fixation affected the manner in which attention was distributed along the meridians. Introducing a potential target location at fixation reduced response time to targets presented on the left side of the horizontal meridian by a significant margin. This was thought to occur because of the interruption of an inhibitory process that would, under conditions where an observer has been cued to attend to a location in the periphery, inhibit processing of stimuli presented at fixation (Dori & Henik, 2006). Introducing a potential target at fixation disturbed the customary inhibitory gradient centered at fixation, affecting the manner in which attention could be distributed among potential target locations. Overall, what is clear is that the effects of endogenous orienting are, in many cases, anisotropic. What characteristics of the display and task impact the effects of endogenous attention, however, has only been explored to a limited degree. The work described in the present paper explores this question further: how would a change to the likelihood of a target appearing at a cued location, and the use of placeholders at the possible target locations, affect the impact of endogenous orienting across the visual field? 4.1.2 Object-based selection: the role of placeholders It is clear that changing the predictive value of a cue will change the way that attention is voluntarily distributed across a display - consequently changing the impact of 127 attention. But why might the use of placeholders be important? Many attention cuing experiments, starting with Posner's work (Posner, Snyder & Davidson, 1980), have used placeholders to indicate where targets could possibly occur. The use of placeholders is intuitive; in real-world visual experience, we rarely attend to locations in space without also attending to an object at that location. What could also play a role in the way in which attention is allocated in a display, then, are the effects of object-based visual attention (e.g. Vecera & Farah, 1994; for a review, see Sholl, 2001). While attention may clearly select a location in space for further processing, there is a great deal of evidence that attention may also select an object in space, independent of its location (e.g. Duncan, 1984; Vecera & Farah, 1994). The issue when using placeholders is that measurement of the effects of attending to a cued location in space is confounded with the effects of attending to the placeholders, which offer an object, as well as a location, for selection. Research concerning inhibition of return (IOR), an effect where response to targets presented at a cued location is actually slower than that to targets presented at an uncued location - reversing the typical attention effects - has shown mixed effects of placeholders (Jordan & Tipper, 1998; McAuliffe, Pratt, & O'Donnell, 2001). Whereas in some instances IOR is greater when there are placeholders at the potential target locations (Jordan & Tipper, 1998; McAuliffe et al., 2001, Experiment 1), in others there is an inconsistent effect of placeholders (McAuliffe et al., 2001, Experiments 2 & 3). In light of this inconsistency, as well as the fact that, to our knowledge, no direct evaluation of the effects of placeholders on the effects of endogenous attention orienting in the visual field has been conducted, we decided to explore the latter. 128 Further, although previous work had postulated that endogenous orienting was inherently space-based, rather than object-based (Egly, Driver, & Rafal, 1994), recent work has demonstrated that the mode of attentional distribution - whether focused or distributed - plays a role in whether endogenous orienting may be object-based (Goldsmith & Yeari, 2003). Specifically, when attention is spread in a diffuse manner across possible target locations, endogenous orienting may effectively select objects, rather than locations, for further processing. As discussed above, under conditions where the cue is not 100% informative about the subsequent location of a target, voluntary attention is simultaneously biased to the cued location, and spread across other possible target locations (e.g. Roggeveen & Ward, submitted, a). Therefore, when the likelihood of target appearance at the cued location is manipulated, the diffuse spread of attention will also be manipulated - potentially leading to varying patterns of reliance on the placeholder objects to effectively select the cued locations. Given this potential interaction, the separate influences of cue validity and object-based orienting may interact with one another to affect how endogenous attention impacts performance in the visual field. Overall, two questions are addressed by the five experiments to be described here. First, what impact does the presence of placeholders have on the efficacy of endogenously orienting attention to a cued location? Second, what impact does the manipulation of the likelihood that a target will appear at a cued location have on the efficacy of endogenously orienting attention to a cued location? The work presented here will examine these two effects separately, as well as how their interaction changes the effects of attending to a cued location in the visual field, and the distribution of attention 129 across possible target locations. This is the first paper to directly address the role of these factors on endogenous attention orienting across the visual field. If these types of manipulations, which change little or nothing about the perceptual experience of the letter being identified, alter how quickly observers respond to target stimuli presented in different visual field locations, then the manner in which attention is endogenously distributed must have been altered by those manipulations. This result, in turn, would provide further understanding of how the endogenous distribution of attention interacts with both the requirements of a task and the perceptual anisotropies - and what that means for seeing across the visual field. 4.2 Experiment 1 In the first experiment, in order to reevaluate the occurrence of anisotropy of endogenous orienting in the visual field, we asked observers to identify a target letter presented in one of 32 possible locations on the screen with placeholders present, as in our previous work. In contrast to Roggeveen and Ward's (submitted, a) experiments, however, where there were 32 possible target locations, on each trial in this experiment there were only four possible target locations, presented along only one of eight different radii starting at fixation. This reduced the potential for uncertainty about the target location, and likely required a more even distribution of attention across uncued target locations - with preference for the cued location - than did the paradigm of Roggeveen and Ward (submitted, a). In this first experiment we chose a traditional likelihood for the appearance of a target at the cued location - 70% validity - so that the subject would find a clear performance advantage in attending to the cued location. Placeholders were present at each of the four possible target locations. 130 4.2.1 Methods 4.2.1.1 Observers. Twenty-four observers participated in the experiment (median age: 21, age range 18 - 29). Of these observers, six.were male, and 18 were female. Twenty-two observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were na'ive to the purposes of the study, and each received $20 for two hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. One subject was excluded from analysis because of error rates higher than 10%. 4.2.1.2 Materials. Observers sat in a comfortable chair with their chin in a chinrest located 50 cm in front of a computer screen. The display was shown on a 17-inch monitor with a resolution of 1280 by 1024 pixels and a refresh rate of 85 Hz. Observers were given both written and verbal instructions about the procedure of the experiment. After the subject reported feeling comfortable with the task when shown several practice trials, the experiment began. A schematic of an example trial is shown in Figure 4.1. At the start of each trial, a fixation cross appeared at the center of a black screen. Four white outline squares used as placeholders were presented along either a North (0°); Northeast (45°); East (90°); Southeast (135°); South (180°); Southwest (225°); West (270°); or Northwest (315°) radius originating at fixation. The placeholder squares were presented 1°, 4°, 8°, and 12° of visual angle from fixation. The size of the squares was magnified in a linear fashion with increasing eccentricity (cortical magnification factor = 2.3; Anstis, 1974). Within each block of 96 trials, the locations of the four white placeholder squares remained the same; between blocks, the locations of the squares 131 would change to a different radius. The order in which each radial location was presented to the subject was randomized. Observers performed 16 blocks, with self-timed breaks between each block, and a longer break after the eighth block. This number of trials allowed for 48 trials per location tested. After 1000 ms, the black inside of the placeholder squares would turn red. After the onset of the four red squares, the inside of three of the squares would fade to black over the course of 416 ms. The remaining red-filled square was the cue for the most likely target location; the target appeared there on 70% of the trials. On the remaining 30% of trials, the target could appear randomly at one of the other three locations marked by now-empty placeholder squares (10% per uncued location). Observers were instructed to shift their attention, without moving their eyes, to the cued location. After the three non-cue squares had faded off the screen, the cue also faded, again over the course of 416 ms, leaving a field of four unfilled placeholders. After a further jittered interval of 200 -400 ms, the target was presented inside one of the placeholder squares for 150 ms. The target could be either the letter M or the letter N . The subject's task was to identify the letter presented. Like the placeholder squares, the target letters were magnified, using 2.3 as the cortical magnification factor (Anstis, 1974) so that the targets would be equally visible at all target eccentricities. After the target presentation, the subject made a response, pressing the "/" key on the keyboard with their right hand, or the "z" key on the keyboard with their left hand. Once the subject had made their response, feedback was presented by either brightening the fixation cross to indicate a correct response, or by replacing the fixation cross with a minus sign to indicate an error. 132 Feedback was presented for 1 second (1000 ms), giving ample time for observers to blink between trials. Trials on which observers made an error and trials with reaction times less than 200 ms and greater than 900 ms were excluded from the analysis. Trials on which blinking and eye movement artifacts occurred also were removed from the analysis by applying automated artifact detection routines. Blinks and eye movements were monitored by using both horizontal and vertical electro-oculograms (EOGs), which were sampled at 250 Hz. Impedances for EOGs were kept below 20kQ. An average of 5% of the total number of trials was removed from the analysis because of one or more of these criteria. 4.2.2 Results 4.2.2.1 Reaction time. In order to evaluate the effects of visual field location and validity on reaction time, a three-way (Radial location x Eccentricity x Validity) Repeated Measures Analysis of Variance (ANOVA) was performed on the reaction time data. In cases where the sphericity assumption was violated, the Greenhouse-Geisser correction was applied. There was a significant main effect of Validity (F(l,20) = 19.63; p < 0.001; partial r f = 0.50): RT was longer for invalidly cued trials by 17 ms. There was a significant interaction between Eccentricity and Validity (F(3,60) = 3.15;/? = 0.03; partial r\ =0.14), indicating that the validity effect varied with eccentricity: as can been seen in Figure 4.2, in most radial locations the difference in RT between validly-cued and invalidly-cued trials was largest 1 ° from fixation. A significant main effect of Eccentricity (F(3,60) = 48.21; p < 0.001; partial n 2 = 0.71) showed that reaction times generally increased as distance from fixation increased, 133 and a significant interaction between Radial location and Eccentricity (F(21,420) = 2.71; p = 0.005; partial r| =0.12, Figure 4.3) showed that the effect of eccentricity differed with radial location. There was also a significant main effect of Radial location (F(7,140) = 3.10; p = 0.005; partial r|2 =0.13, Figure 4.4), showing that across all eccentricities, performance along certain radii was faster than that along others. In order to test directly the perceptual anisotropies discussed in the introduction, five planned comparisons were performed. Planned comparisons (rather than post-hoc) were performed in conjunction with the omnibus A N O V A in order to make focused comparisons between average reaction times at the specific visual field locations of interest. In order to compensate for potential family-wise error, a more conservative p-value significance cutoff of 0.01 was chosen. RT was significantly shorter in the lower visual field than in the upper visual field (F(l,20) = 13.96;p = 0.001; partial n 2 =0.41) and in the right visual field than in the left (^(1,20) = 8.56; p = 0.008; partial n 2 = 0.30). RT was also shorter on the lower half of the vertical meridian (F(\,20) = 3.0\;p = 0.09; partial n = 0.143). There were no differences in RT between the horizontal and vertical meridians (F(l,20) = 1.76;p = 0.062; partial n 2 = 0.08) nor between the cardinal and intercardinal locations (^(1,20) < 1.0; p = 0.94; partial n 2 = 0.0003). 4.2.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) was also performed on the error rates. Only a significant main effect of Eccentricity was found (F(3,60) = 5.26;p = 0.003; partial i f = 0.21), showing that the number of errors committed was larger 1° from fixation than at 4°, and then increased with eccentricity, indicating that there was a possible speed-accuracy tradeoff for the 1 ° location. This tradeoff cannot account for the overall eccentricity effect. Moreover, it can 134 account only partially for the larger validity effect at 1 ° because, although RT on validly-cued trials was shorter and less accurate there, consistent with such a tradeoff, RT was longer and accuracy was lower for invalidly-cued trials, inconsistent with such a tradeoff. The mean error rate was 3.67%. 4.2.3 Discussion First, the results of Experiment 1 provide a clear example of anisotropic attention effects in the visual field. The effects of cuing the target location were largest at 1 ° from fixation, because of a larger impairment of performance on invalid trials. What this result indicates is that when observers need to shift attention back towards fixation in responding to a target there when originally cued to attend more peripherally, they are impaired to a greater extent than when asked to shift attention to the other three possible eccentricities. This finding provides support for the presence of an inhibitory gradient centered at fixation (Dori & Henik, 2006). It also lends credence to the idea that the effects of endogenous orienting across the visual field are not homogeneous, and that they are strongly affected by the manner in which attention is voluntarily distributed across possible target locations. Experiment 1 also provided strong evidence for a well-documented perceptual anisotropy: preferential processing in the lower visual field, both directly in shorter RTs for targets in the lower than in the upper visual field, and in the presence of vertical meridian anisotropy, with faster responding on the lower part of the vertical meridian. A clear preference for responding to targets presented in the right visual field over the left was also found. This effect did not change, however, depending on whether or not the subject was cued to attend to the right visual field; faster responses to targets on the right 135 were consistent throughout. This finding replicates that of Roggeveen and Ward (submitted, a; b), who also did not find a greater cuing advantage of the right visual field over the left in a letter identification task. Interestingly, we found no evidence for the other reported perceptual anisotropies, neither faster or more accurate responses to targets on the horizontal meridian compared to the vertical meridian (e.g. Carrasco et al., 2001), nor better performance on the cardinal axes than on the intercardinal axes (Roggeveen & Ward, submitted, b). We assessed the occurrence of all of these perceptual anisotropies in all of the following experiments in order to evaluate their consistency across variations in this paradigm. 4.3 Experiment 2 Having established how both perceptual and attentional anisotropies emerge under conditions where the subject is fairly certain that the target will appear at the cued location, we conducted Experiment 2 in order to evaluate the effects of changing the likelihood that a target will appear at a cued location. Roggeveen and Ward (submitted, a) found that changing the likelihood that a target would appear at a cued location from 100% to 70% induced observers to distribute some residual attention across the uncued locations in the display while simultaneously giving preferential attention to the cued location. Due to the large number of locations in the display, however, observers were not able to distribute their residual attention equally well to all uncued locations, and instead appeared to allocate residual attention to uncued locations in a way that would maximize performance. In Experiment 1, the likelihood that the target would appear in the cued location was 70%, making the likelihood 30% that the target would appear in one of the other three locations (10% each). In Experiment 2, the likelihood that the 136 target would appear at the cued location was only 40%, making the likelihood that the target would appear at each of the uncued locations individually 20%. This means that the likelihood of an invalidly-cued trial was larger than that of a validly-cued trial. In this way, it is possible to see, under conditions with a smaller number of locations, the effect of cue validity on how observers distribute their attention across possible target locations - and how that effect interacts with both perceptual and attentional anisotropies. 4.3.1 Methods 4.3.1.1 Observers. Seventeen observers participated in the experiment (median age: 20, age range 18 - 35). Of these observers, 11 were male, and six were female. Two observers reported that they were left handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $40 for four hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. One subject (right handed female) was excluded from the analysis for error rates greater than 10%. 4.3.1.2 Materials. A l l materials were the same as in Experiment 1, with the following exceptions. Rather than having the cue indicate the target location on 70% of the total trials, the cue now indicated the target location with only 40% validity. Invalid trials occurred for 60% of the trials, with a 20% chance of the target occurring at each of the other, uncued locations. Observers were instructed to shift their attention to the cued location, as it was more likely that the target would appear at the cued location than at any one of the other locations. Observers also ran in two sessions of the experiment, 137 rather than one, resulting in a total of 32 blocks of 64 trials, or a total of 2048 trials (64 trials per location). 4.3.2 Results 4.3.2.1 Reaction time As for Experiment 1, a three-way A N O V A (Radial location x Eccentricity x Validity) was performed on the RT data remaining after errors and eye movement trials (3%) were discarded. A l l three main effects were significant. The significant main effect of Eccentricity (F(3,48) = 5.08; p < 0.001; partial n 2 = 0.24) showed that as distance from fixation increased, so did RT. Responses to validly-cued targets were significantly faster than responses to invalidly-cued targets (F(l,16) = 15.88; p = 0.001; partial n = 0.50), and RT differed significantly for the different radii (F(7,\ 12) = 5.08;/? < 0.001; partial n 2 = 0.24). Only the interaction between Radial location and Validity was marginally significant (F(7,l 12) = 1.87; p = 0.08; partial n 2 = 0.11; see Figure 4.5), again indicating attentional anisotropy but differently from that found in Experiment 1. Here the smallest difference between RT to targets presented at validly- and invalidly-cued locations occurred in the upper right visual field (NW). Because the main effect of Radius was significant, the same five planned comparisons were performed as for Experiment 1. Again, because of the large number of comparisons, a more conservative p-value significant cutoff of 0.01 was chosen to control for family-wise error. As in Experiment 1, responses were significantly faster to targets presented in the lower visual field (F(l,16) = 18.0;p = 0.0006; partial n 2 = 0.73) compared to the upper visual field, as well as to targets presented in the right visual field CF(1,16) = 10.84; p = 0.005; partial n 2 = 0.40) compared to the left. The comparison between the upper and lower halves of the vertical meridian showed only a marginally 138 significant difference, however, with faster RTs to targets on the lower half (F(l,16) = 3.85;p = 0.06; partial r| =0.19). The differences between performance on the horizontal and vertical meridians (F(\A6) = 0.02;p = 0.89; partial i f = 0.001) and the cardinal and intercardinal meridians (F(l,16) = 2.36;p = 0.14; partial n 2 = 0.12) were not significant. 4.3.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) was also performed on the error rates. No significant interactions or main effects were found. The mean error rate was 3.88%. 4.3.3 Discussion The pattern of anisotropy differed between Experiments 1 and 2. Where in Experiment 1 the effects of attention were roughly equivalent across all radial locations, and differed across distances from fixation, in Experiment 2 the effects of attention were different across radial locations (although only marginally), and were the same across all eccentricities. Roggeveen and Ward (submitted, b) found that an inhibitory gradient centered at fixation can be interrupted by the presence of a potential target location there. What also appears to be a critical characteristic of the presence of this gradient is a high likelihood that the target will appear at the cued location, away from fixation. The presence of an interaction between Eccentricity and Validity when cue validity is high (Experiment 1), and its absence when cue validity is low (Experiment 2), implies that locations closest to fixation are inhibited under conditions where it is highly unlikely that the target will appear there (high cue validity), but not when it is fairly likely that a target will appear there (low cue validity). Under the conditions of Experiment 2, a subject would be well-advised to put less emphasis on the cued location than they may have under the 139 conditions in Experiment 1, and to distribute a greater amount of attention across the uncued locations. Because of the greater attentional weight placed at each of the uncued locations, the decrement in performance at those locations, arising from attentionally emphasizing the cued location, would be reduced. In fact, this was the case in this experiment: the RT disadvantage for responding to a target presented at an uncued location averaged only 7 ms, as compared to 18 ms in Experiment 1. The pattern of perceptual anisotropies - both what was significant, and what was not significant - was nearly identical between Experiments 1 and 2. In both experiments, there were faster responses to stimuli presented in the lower visual field than the upper visual field, and faster responses to stimuli presented on the right than on the left. In a slight departure from the results of Experiment 1, the vertical meridian asymmetry in Experiment 2 was only marginally significant. There was still no significant evidence for a horizontal-vertical meridian asymmetry, or preferential processing on the cardinal axes when compared to the intercardinal axes. 4.4 Experiment 3 In addition to the likelihood of target presentation at the cued location, the presence of placeholders at the possible target locations may also affect the manner in which attention is distributed across them. In particular, the presence of placeholders would be expected to increase the precision of allocation of attention to potential target locations (e.g. Jordan & Tipper, 1998). By eliminating the placeholders in Experiment 3, we were able to evaluate the role of placeholders in the attentional anisotropy of found in Experiment 1, and in the pattern of perceptual anisotropies found across both Experiments 1 and 2. 140 4.3.1 Methods 4.3.1.1 Observers Twenty observers participated in the experiment (median age: 21, age range 18 - 38). Of these observers, two were male, and 18 were female. A l l observers reported that they were right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $40 for four hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. Two observers were excluded from the analysis due to error rates higher than 10%. 4.3.2.1 Materials The methods of Experiment 3 differed from those of Experiment 1 only in that the placeholder squares were removed from the screen. The timing and presentation of the red cuing squares remained the same. 4.3.2 Results 4.3.2.1 Reaction time. As in the previous experiments, a three-way A N O V A (Radial location x Eccentricity x Validity) was performed on the RT data remaining after errors and eye movement trials were discarded (4%). A significant main effect of Eccentricity (F(3,57) = 11.16; p < 0.001; partial n 2 = 0.39), showed that RT increased as distance from fixation increased. Responses to validly-cued targets were also significantly faster than responses to invalidly-cued targets (F(l,19) = 22.85;p < 0.001; partial n 2 = 0.57). A significant interaction between Radial Location and Eccentricity was also found (F(21,357) = 1.97;p = 0.04; partial n 2 = .10; see Figure 4.6). 141 Although the main effect of Radial Location was not significant, in order to parallel the analyses of Experiments 1 and 2, the same five planned comparisons were carried out. No significant differences were found between the upper and lower visual fields, the right and left visual fields, the upper and lower halves of the vertical meridian, the horizontal and vertical meridians, or the cardinal and intercardinal locations. Thus, in Experiment 3, there was no evidence of perceptual anisotropies. 4.4.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) was also performed on the error rates. A significant main effect of Radial location was found (F(7,\ 19) = 2.25; p = 0.04; partial n 2 = 0.12). As can been seen in Figure 4.7, the highest number of errors occurred in the upper visual field. As responding was no faster in the upper visual field, however, this is unlikely to indicate a speed-accuracy tradeoff. The mean error rate was 2.89%. 4.4.3 Discussion Interestingly, the removal of placeholders at the four possible target locations in the display eliminated the well-documented perceptual anisotropies found in Experiments 1 and 2. The disappearance of previously found advantages in reaction time when responding to targets presented in certain areas of the visual field, despite consistency between the target stimuli used in the experiments discussed here, indicates a role for the endogenous distribution of attention in perceptual anisotropies. Endogenous orienting to a cued location, however, did not have anisotropic effects: while cuing did have a significant impact on performance, that effect was the same across all 32 of the possible target locations. Also, similar to Experiment 2, as distance from fixation increased, so did reaction time. 142 4.5 Experiment 4 Placeholders clearly have a powerful effect on both attentional and perceptual anisotropies. Experiment 4 was performed in order to evaluate whether altering the likelihood of the target appearing at the cued location - thus changing the manner in which attention was distributed across possible target locations - would reduce the negative impact of eliminating the placeholders. In Experiment 2, it was seen that observers distributed their attention more evenly across possible target locations when cue validity was low. With this more even distribution of attention across locations, will the perceptual anisotropies clearly present in the results of Experiment 2 also be seen in the absence of placeholders? In order to answer this question, we replicated Experiment 2, but without placeholders. 4.5.1 Methods 4.5.1.1 Observers. Nineteen observers participated in the experiment (median age: 21, age range 18-31). Of these observers, 10 were male, and nine were female. Two observers reported that they were left handed; the rest reported being right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $40 for four hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. Two observers were excluded from the analysis for error rates greater than 10%. 143 4.5.1.2 Materials. Experiment 4 was identical to Experiment 2, with the following exception: the white boxes that served as placeholders in Experiment 2 were absent. The timing, size, and presentation of the red cue squares remained the same. 4.5.2 Results 4.5.2.1 Reaction time. As in the previous experiments, a three-way A N O V A (Radial location x Eccentricity x Validity) was performed on the RT data remaining after errors and eye movement trials (4%) were eliminated. Significant main effects of Eccentricity (F(3,48) = 24.27;p < 0.001; partial n 2 = 0.60) and Validity (F(l,16) = 10.79; p - 0.005; partial n = 0.40) showed that as distance from fixation increased, so did RT, and that RT was faster to validly-cued targets. The main effect of Radius was marginally significant (F(1A 12) = 2.39; p = 0.05; partial n 2 = 0.13; see Figure 4.8). A significant interaction between Eccentricity and Validity (F(3,48) = 5.28; p = 0.003; partial n = 0.25; see Figure 4.9) showed that the difference between the reaction time to validly-cued and invalidly-cued targets was greatest 1° from fixation, and declined as eccentricity increased. In light of this interaction and the previous comparisons made, five planned comparisons were performed in order to evaluate the perceptual anisotropies. With the more conservative p-\a\ue significance cut-off, none of the five comparisons were significant. However, three were marginally significant. Responses to targets in the lower visual field were faster than to those in the upper visual field (F(\,\6) = 6.07;p = 0.03; partial n 2 = 0.27), and responses to targets in the lower half of the vertical meridian were faster than those to targets in the upper half (F(l,16) = 6.04;p = 0.03; partial n 2 = 0.27),. Although the difference between the cardinal and intercardinal locations was also 144 2 marginally significant (F(l,16) = 4.74;p = 0.05; partial r| = 0.22), the actual average RT difference was only 3 ms. Neither the difference between the right and left visual fields, or that between the horizontal and vertical meridians, was significant. 4.5.2.2 Errors. A three-way A N O V A (Radial location x Eccentricity x Validity) was also performed on the error rates. Significant main effects of Radius (F(7,l 12) = 2.6;p = 0.001; partial n 2 =0.18; see Figure 4.10) and Eccentricity (F(3,48) = 6.40; p = 0.01; partial i f = 0.28) were found. The mean error rate was 3.34%. None of these effects, however, indicated the presence of a speed-accuracy tradeoff: more errors were made closer to fixation (see Figure 4.11); however, this parallels the longer RTs found on invalidly-cued trials when the target was presented 1° from fixation. 4.5.3 Discussion Two interesting patterns emerged when both the likelihood of the target appearing at the cued location was reduced and placeholders were not used. First, rather than finding consistent cuing effects at all eccentricities, as in Experiment 1, the effects of validly precuing the target location were greatest when the target was presented 1 ° from fixation. Second, the perceptual anisotropies found clearly in Experiment 2 disappeared or were much weaker - mirroring the disappearance of the same anisotropies in Experiment 3. The emergence of slower RTs to invalidly-cued targets presented close to fixation brings the discussion back to the possible role of an inhibitory gradient centered at fixation. In Experiment 1, under conditions of high cue validity and with placeholders present at all possible target locations, an inhibitory gradient appeared to interfere with performance for targets close to fixation. In Experiment 4, an inhibitory gradient appears 145 to interfere with responses to targets close to fixation under conditions of low cue validity and with no placeholders present. Why might this pattern of performance have emerged? What appear to be interacting here are the emphasis that the subject gives to each of the locations - driven by the likelihood that the target will appear there - and the effect of the presence of perceptual objects on the ease of distributing attention among those locations. In Experiment 1, distributing attention among the uncued locations would have been relatively easy, because of the presence of the placeholders, but there was little residual attention to distribute, because of high cue validity. In Experiment 4, the absence of placeholders made it more difficult to distribute attention among the uncued (or cued, for that matter) locations, but cue validity was low, requiring observers to maintain a more even distribution of attention, allocating more attention to uncued locations that were unmarked after the red squares faded. In both experiments, the cost of allocating attention to locations close to fixation would be prohibitive, but for different reasons: in Experiment 1 because there was a low probability of the target appearing there, and in Experiment 4 because the effort required was too great when the near-fixation location was not marked by a perceptual object, in spite of a higher probability of a target appearing there. Therefore, in both situations, we see that the inhibitory gradient centered at fixation was not overridden by attention: in one case, where the result was not worth the effort (Experiment 1), in the other, where without perceptual objects, the task became too difficult (Experiment 4). 4.6 Experiment 5 Because the larger cuing effect at 1 ° from fixation arose from a decrement in performance on invalidly-cued trials, we conducted Experiment 5 in order to evaluate 146 more precisely the role of placeholders at the uncued locations. In this experiment a placeholder only appeared, with the onset of the cue, at the cued location. In this experiment also cue validity was high (70%), replicating the cue validity of Experiments 1 and 3. 4.6.1 Methods 4.6.1.1 Observers. Seventeen observers participated in the experiment (median age: 21, age range 19 - 22). Of these observers, seven were male, and 10 were female. Two observers reported that they were left handed; the rest reported being right handed. A l l reported normal or corrected-to-normal vision, and no neurological defect. A l l observers were naive to the purposes of the study, and each received $20 for 2 hours of participation. The methods of the study were approved by the Behavioral Research Ethics Board of the University of British Columbia and all observers gave informed consent prior to beginning the study. Two observers were excluded from the analysis for error rates greater than 10%. 4.6.1.2 Materials. The materials used in Experiment 5 were identical to those in Experiment 1, with a single change: rather than having the four possible target locations within a block marked with a white placeholder square, only the cued location within each trial had a placeholder square. The white placeholder square appeared at the to-be-cued location with the four red squares, and then remained on throughout the trial, disappearing only after the response was made to the target. In this manner, only validly-cued targets appeared inside a placeholder square. Invalidly-cued targets appeared at locations where a red square had briefly appeared and then had faded, leaving only a blank screen. 147 4.6.2 Results 4.6.2.1 Reaction time A three way A N O V A (Radial location x Eccentricity x Validity) was performed on the RT data remaining after errors and eye movement trials (5%) were discarded. A l l three main effects were significant: RT to validly-cued targets was faster than to invalidly-cued targets (F(l,14) = 53.80;p < 0.001; partial n 2 = 0.79), response speed slowed significantly as target eccentricity increased (F(3,42) = 40.80;p < 0.001; partial n 2 = 0.74), and RT varied with Radial Location (F(7,98) = 3.83; p = 0.001; partial n =0.21, see Figure 4.12). There was only a marginally significant interaction was between Radial location and Eccentricity (F(21,294) = 1.85; p = 0.07; partial r\ = 0.12; see Figure 4.13). As the main effect of radius was significant, as for the other experiments five planned comparisons were carried out. Responses to targets presented in the lower visual field were significantly faster than to those presented in the upper visual field (F(l,14) = 13.63;/? = 0.002; partial n 2 = 0.49). The advantage of performance in the right visual field over the left was also significant (F(l,14) = 9.55;p = 0.007; partial n 2 = 0.41). The difference between performance on the upper and lower halves of the vertical meridian, with faster RTs on the lower half, was marginally significant using the more conservative significance cut-off (F(l,14) = 5.80;p - 0.03; partial n 2 = 0.29). However, the differences between the horizontal and vertical meridians, and cardinal and intercardinal meridians, were not significant. 4.6.2.2 Errors A three-way A N O V A (Radial location x Eccentricity x Validity) was also performed on the error rates. No significant interactions or main effects were found. The mean error rate was 4.18%. 148 4.6.3 Discussion The introduction of a placeholder only at the cued location in Experiment 5 did not change the result of Experiment 3 with respect to attentional isotropy: attention was isotropic in both experiments. There were differences between the two experiments in the perceptual anisotropies, however, without any placeholders there was no evidence for a processing advantage for targets in the lower visual field over the upper visual field, or the right versus the left, reintroducing a placeholder at the cued location restored these effects. 4.7 General Discussion Returning to the two general questions we posed earlier, how do cue validity and the presence of placeholders affect endogenous orienting to various locations in the visual field? Precuing locations in the visual field only had anisotropic effects in Experiment 1, where cue validity was 70% and placeholders were present, and in Experiment 4, where cue validity was only 40% and placeholders were absent. In both experiments, the cost of allocating attention to the location 10 from fixation must have been greater than the benefit derived from such allocation, because responses to invalidly-cued trials at that location were slowed there more than at other invalidly-cued locations, presumably by an inhibitory gradient centered at fixation. That this pattern of performance emerged both with high cue validity and placeholders present, and with low cue validity and placeholders absent, is initially confusing. The same effect occurs under seemingly opposite conditions. As stated earlier, without a large incentive to maintain equivalent distribution of attention across the uncued target locations (as in Experiment 1), it is easier to allow the "natural" occurrence 149 of an inhibitory gradient centered at fixation (Dori & Henik, 2006). As the target is most likely to appear at the cued location - a location away from fixation - ignoring locations at or near fixation is strategically smart. Expending the effort to attend equally to all uncued locations would be unlikely to have a palpable effect on performance, given that targets only appeared at the uncued location 1° from fixation on 10% of trials. However, in Experiment 4, while the likelihood that the target will appear 1 ° from fixation when that location has not been cued was doubled (20%), without placeholders, the ability to selectively attend to any of the four possible target locations was hampered. This, again, allowed for an inhibitory gradient centered at fixation to affect performance. The idea that attention orienting is facilitated by the presence of a perceptual object, versus no object at all, is supported by evidence showing that IOR is enhanced when a placeholder is used (Jordan & Tipper, 1998; McAuliffe, Pratt, & O'Donnell, 2001). The work presented here is the first in the endogenous orienting literature, to our knowledge, to evaluate explicitly the effect of placeholders on performance in a cuing task. Another important finding is the variation across the five experiments in what have been considered to be stable perceptual anisotropies. Removing placeholders, for instance, negated the preferential processing of the right visual field over the left. Similarly, when placeholders were removed at all possible target locations, the preference for responding to targets in the lower visual field was eliminated. However, the lower visual field advantage was maintained so long as a placeholder was present at the cued location. Both of these patterns perhaps indicate a potential role of endogenous attention orienting to a perceptual object in a perceptual task. Therefore, the influence of the presence of perceptual objects on perceptual task performance should be considered. 150 Additionally, across all studies, there was no evidence for faster responses on the horizontal meridian over the vertical meridian, nor faster responses on the lower half of the vertical meridian than on the upper half (vertical meridian asymmetry). Why neither of these well-documented effects (e.g. Carrasco et al., 2001) occurred may be a result of the type of task used here. Compared to a simple perceptual discrimination (e.g. a Gabor patch orientation discrimination), a letter discrimination may require levels of processing not as sensitive to differences in quality of sensory information between the horizontal and vertical meridians, or the upper and lower half of the vertical meridian. These findings also place an interesting limitation on the finding of an attentional oblique effect, as reported by Roggeveen and Ward (submitted, b). It appears that such an effect is absent when stimuli are presented within a single quadrant, rather than along a meridian crossing the center of the visual field. The difficulty of attending to locations along a meridian may have been attenuated by reducing either the distance or number of locations across which attention had to be distributed: the task used in the present experiments may not have been effortful enough to draw out this effect. Future research could attempt to disentangle the effects of the area across which the observer must spread attention, and the number of individual locations to which the observer must attend. What is critical here is that these five experiments demonstrate an important role for two sometimes-ignored attention orienting task characteristics: the likelihood that a target will appear at a cued location, and the use of placeholders at target locations. Both appear to have different effects on the impact of endogenous orienting at various visual field locations - and their interaction with one another illustrates how the distribution of attention across a display can be altered by simple task characteristics. The manner of 151 attention distribution affects the impact of cuing a target location differently at different visual field locations - in this particular case, with respect to distance from fixation. Future research should explicitly explore further the variables that affect the manner in which attention is distributed across potential target locations. 152 4.8 Tables and Figures - Chapter 4 Table 4.1. Main effect of eccentricity in error rates, Experiment 1. All values are in percent error. Eccentricity (degrees of visual angle) Error rate (standard error) 1° 3.18% (0.37%) 4° 2.23% (0.28%) 8° 2.81% (0.38%) 12° 3.68% (0.42%) 153 Table 4.2. Main effect of eccentricity in reaction times, Experiment 2. All values in milliseconds. Eccentricity (degrees of visual angle) Reaction time (standard error) 1° 525 ms (14 ms) 4° 521 ms (13 ms) 8° 529 ms(13 ms) 12° 540 ms (13 ms) Table 4.3. Main effect of eccentricity in reaction times, Experiment 5. All values in milliseconds. Eccentricity (degrees of visual angle) Reaction time (standard error) 1° 508 ms(lOms) 4° 508 m s ( l l ms) 8° 523 ms (11 ms) 12° 534 ms (12 ms) Table 4.4. Main effect of Validity - Experiment 5. All values are in milliseconds. Cuing condition Reaction time (standard error) Valid 507 m s ( l l ms) Invalid 530 ms (11 ms) 156 Figure 4.1. Schematic of sample trial • • • • • Cue: all but cued location fades off over 416 ms Cue fades off over 416 ms ISI:200-400 ms Target: 150 ms time 157 Figure 4.2. Validity x Eccentricity interaction, Experiment 1 590 580 570 560 550 540 530 I 520 .1 510 500 <r> TJ C o CJ. OJ 0) QJ 490 480 470 Experiment 1: Eccentricity x Validity interaction F(3, 60) = 3.15, p = 0.031 Error bars denote 95% confidence intervals 1 4 8 Eccentricity (degrees of visual angle) 12 Valid Invalid 158 Figure 4.3. Eccentricity x Radius interaction, Experiment 1 O o QJ QJ Experiment 1: Eccentricity x Radius interaction F (21,420) = 2.71, p = .005 Error bars denote 95% confidence intervals 600 580 560 540 1 5 2 0 4-1 £= O o 5 0 0 03 QJ QL 4 8 0 460 NE SE S SW W NW Radial location 1 degree 4 degrees 8 degrees 12 degrees 159 Figure 4.4. Main effect of Radial location - Experiment 1 Radial location: Experiment 1 Error bars denote 95% confidence intervals 550 NE S E S S W W Radial location 160 Figure 4.5. Radial location x Validity interaction - Experiment 2 m X3 c o u CD £ c 2 +-» o C3 580 570 560 550 540 530 520 510 500 490 480 470 Experiment 2: Validity x Radial location interaction F(7, 112) = 1.87, p = 0.08 Error bars denote 95% confidence intervals NE E SE S SW W NW Radial location N -3E Valid 3 E Invalid 161 Figure 4.6. Eccentricity x Radial location interaction - Experiment 3 Experiment 3: Eccentricity x Radial location interaction F(21, 357) = 1.97, p = 0 04 Error bars denote 95% confidence intervals 640 to X3 C o (_> CD CO CD E 4-) c o O CO CD 460 NE E SE S SW W Radial location NW N 1 degree 4 degrees 8 degrees 12 degrees 162 Figure 4.7. Main effect of Radial location, error data - Experiment 3 Radial location - Errors: Experiment 3 Error bars denote 95% confidence intervals 5.5 i ' • 1 1 ' 1 ' 5.0 _ 4.5 • NE E S E S S W W NW N Radial location 163 Figure 4.8. Main effect of Radius in reaction time - Experiment 4 Radial location: reaction time - Experiment 4 Error bars denote 95% confidence intervals 570 565 -w 560 T3 C 8 555 CD GO 1 550 CD E 545 c B 540 o CD CD OH 535 530 525 I 21 I NE S E S S W W Radial locations NW N 164 Figure 4.9. Eccentricity x Validity interaction - Experiment 4 Experiment 4: valdity x Eccentricity interaction F(3, 48) = 5.28. p = 0.003 Error bars denote 95% confidence, intervals 610 | • ' , 600 ~ 590 580 Eccentricity (degrees of visual angle) 165 Figure 4.10. Main effect of Radius in error rates - Experiment 4 166 Figure 4.11. Main effect of eccentricity in error rates - Experiment 4 Experiment 1: Eccentricity - error rates F(3,48) = 6.40,-p = .001 Error bars denote 95% confidence intervals 7.5 p • • • 7.0 • cn 5.0 OJ 2 4.5 S 4,0 3.5 1.5 • 1.0 I • • • -1 4 8 12 Eccentricity (degrees of visual angle) 167 Figure 4.12. Main effect of Radius - Experiment 5 168 Figure 4.13. Radius x Eccentricity interaction - Experiment 5 Experiment 5: Eccentricity x Radial location interaction F(21,294) = 1.85, p = 0 014 Error bars denote 95% confidence intervals 600 3 4 5 6 Radial location 3 E 1 degree W- 4 degrees 3 1 8 degrees 12 degrees 169 4.9 References Altpeter, E., Mackeben, M . , & Trauzettel-Klosinski, S. (2000). The importance of sustained attention for patients with maculopathies. Vision Research, 40, 1539-1547. Anstis, S. (1974). A chart demonstrating variations in acuity with retinal position. Vision Research, 14, 589-592. Cameron, E.L., Tai, J.C., & Carrasco, M . (2002). Covert attention affects the psychometric function of contrast sensitivity. Vision Research, 42, 949-967. Carrasco, M . , Giordano, A . M . , & McElree, B. (2004). Temporal performance fields: visual and attentional factors. Vision Research, 44, 1351-1365. Carrasco, M . , Talgar, C , & Cameron, E.L. (2001). Characterizing visual performance fields: Effects of transient covert attention, spatial frequency, eccentricity, task, and set size. Spatial Vision, 15, 61-75. Carrasco, M . , Williams, P.E., & Yeshurun, Y . (2002). Covert attention increases spatial resolution with or without masks: Support for signal enhancement. Journal of Vision, 2, 467-479. Cheal, M . , & Lyon, D.R. (1991). Central and peripheral precuing of forced-choice discrimination. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 43A, 859-880. Connolly, M . & Van Essen, D. (1984). The representation of the visual field in parvicellular and magnocellular layers of the lateral geniculate nucleus in the macaque monkey. Journal of Comparative Neurology, 226(4), 544 - 564. 170 Dori, H. , & Henik, A. (2006). Indications for two attentional gradients in endogenous visual-spatial attention. Visual Cognition, J3(2), 166-201. Duncan, J. (1984). Selective attention and the organization of visual information. Journal of Experimental Psychology: General, 775(4), 501-517. Egly, R., Driver, J., & Rafal, R.D. (1994). Shifting visual attention between objects and locations: evidence from normal and parietal lesion observers. Journal of Experimental Psychology: General, 123(2), 161 - 177. Fink, G.R., Halligan, P.W., Marshall, J.C., Frith, C D . , Frakowiak, R.S.J., & Dolan, R.J. (1996). Where in the brain does visual attention select the forest and the trees? Nature, 382, 626 - 628. Fukisama, S.S., & Faubert, J. (2001)..Perceived length in the central visual field: evidence for visual field asymmetries. Vision Research, 41, 2119-2126. Gawryszewski, L. , Riggio, G. de, L., Rizzolatti, G., & Umilta, C. (1987). Movements of attention in the three spatial dimensions and the meaning of "neutral" cues. Neuropsychologia, 25, 19-29. Goldsmith, M . , & Yeari, M . (2003). Modulation of object-based attention by spatial focus under endogenous and exogenous orienting. Journal of Experimental Psychology: Human Perception and Performance, 29(5), 897 - 918. Gottlob, L.R., Cheal, M . , & Lyon, D.R. (1999). Attention operating characteristics in a location-cuing task. The Journal of General Psychology, 126(3), 271 - 287. He, S.H., Cavanagh, P., & Intriligator, J. (1996). Attentional resolution and the locus of visual awareness. Nature, 383, 334-337. 171 Heilman, K . M . , & Valenstein, E. (1979). Mechanisms underlying hemispatial neglect. Annals of Neurology, 5, 166-170. Intriligator, J., & Cavanagh, P. (2001). The spatial resolution of visual attention. Cognitive Psychology, 43, 171-216. Johnson, D.N. , & Yantis, S. (1995). Allocating visual attention: Tests of a two-process model. Journal of Experimental Psychology: Human Perception and Performance, 21(6), 1376- 1390. Jordan, H. , & Tipper, S.P. (1998). Object-based inhibition of return in static displays. Psychonomic Bulletin and Review, 5(3), 504-509. Kinsbourne, M . (1970). The cerebral basis of lateral asymmetries in attention. Acta Psychologica, 33, 193-201. Kitterle, F.L, Christman, S., & Hellige, J.B. (1990). Hemispheric differences are found in the identification, but not the detection, of low versus high spatial frequencies. Perception and Psychophysics, 48(A), 297 - 306. Kraft, A. , Pape, N . , Hagendorf, H. , Schmidt, S., Naito, A. , Brandt, S.A. (2007). What determined sustained visual attention? The impact of distracter positions, task difficulty and visual fields compared. Brain Research, 1133, 123 - 135. Liu, T., Heeger, D.J., & Carrasco, M . (2006). Neural correlates of the visual vertical meridian asymmetry. Journal of Vision, 6, 1294 - 1306. Losier, B.J., & Klein, R . M . (2004). Covert orienting within peripersonal and extrapersonal space: young adults. Cognitive Brain Research, 19, 269 - 274. Mackeben, M . (1999). Sustained focal attention and peripheral letter recognition. Spatial Vision, 12, 51-72. 172 McAuliffe, J., Pratt, J., & O'Donnell, C. (2001). Examining location-based and object based components of inhibition of return in static displays. Perception and Psychophysics, 63(6), 1072- 1082. Perry, V . H . , & Cowey A. (1985.) The ganglion cell and cone distributions in the monkey's retina: implications for central magnification factors. Vision Research, 25, 1795- 1810. Posner, Snyder & Davidson (1980) Attention and the detection of signals. Journal of Experimental Psychology, 109, 160-74. Previc, F.H. (1990). Functional specialization in the lower and upper visual fields in humans: Its ecological origins and neurophysiological implications. Behavioral and Brain Sciences, 13, 519-575. Previc, F. (1998). The neuropsychology of 3D space. Psychological Bulletin, 124, 123 -164. Roggeveen, A .B . , & Ward, L . M . (submitted, a). Anisotropy of covert endogenous orienting across the visual field. Journal of Experimental Psychology: Human Perception and Performance. Roggeveen, A .B . , and Ward, L . M . (submitted, b). Endogenous attentional distribution and attentional anisotropy across the visual field. Vision Research. Rovamo, J., & Virsu, V . (1979). An estimation and application of the human cortical magnification factor. Experimental Brain Research, 37, 495-510. Shepherd, M . , & Muller, H.J. (1989). Movement versus focusing of visual attention. Perception and Psychophysics, 46, 146-154. Scholl, B.J. (2001). Objects and attention: The state of the art. Cognition, 80(1-2), 1 - 46. 173 Talgar, C P . , & Carrasco, M . (2002). Vertical meridian asymmetry in spatial resolution: Visual and attentional factors. Psychonomic Bulletin & Review, 9, 714-722. Vecera, S.P., & Farah, M.J. (1994). Does attention select objects or locations? Journal of Experimental Psychology: Human Perception and Performance, 123(2), 146-160. Verfaellie, M . , Bowers, D., & Heilman, K . M . (1988). Hemispheric asymmetries in mediating intention, but not selective attention. Neuropsychologia, 26, 521 - 531. Wright, R.D., & Ward, L . M . (1998). The control of visual attention. In Visual Attention, R.D. Wright, Ed. Oxford: New York, 132-186. 174 5 Conclusion Returning to the underlying question that all of these experiments addressed, are the effects of covert, endogenous attention anisotropic at different visual field locations? As mentioned in Chapter 1, the general answer to this question is yes - there are instances where the effect of attending to certain locations in a covert, voluntary way has varying levels of effect on reaction time. However, what is clear from the ten experiments described herein is that the pattern of anisotropy is not always the same: there is no universal preference for voluntarily attending to certain locations in the visual field over others. (See Appendix B for a summary of the findings from all experiments.) Rather, the impact of endogenous, covert orienting interacts with the way in which an observer approaches the task at hand. There are several ways to manipulate the task demands for the observer; across all of these experiments, while keeping the basic task (letter identification) the same, other characteristics of the experiment were changed to assess their impact on attention orienting. First, the likelihood of a target appearing at a cued location was manipulated (Chapters 2 and 4). When attending to a specific location as indicated by a cue, if the cue specifies with 100% certainty where the target will appear, there is no reason to attend to other locations in a display. However, if the cue only tells the observer with 70% certainty where the target will appear, in order to do the task when the target does not appear at the cued location, the observer also attends somewhat to other potential locations in the display. The degree to which those other locations are attended is not the same, however, as when no location in particular has been cued, and observers spread their attention as equally as they can across all locations. Rather, there is a decline in the 175 ability to distribute attention across those locations. In light of this restriction, the manner in which attention is distributed among the possible target locations is strategic. Finding this pattern of results supports theoretical accounts of attention that stipulate that attention may be both preferentially allocated to a cued location in the visual field, and simultaneously spread in a general fashion among all possible target locations (e.g. LaBerge, 1995; LaBerge et al, 1997). Additionally, Treisman (2006) has proposed several guidelines for how distributing attention across a broad swath of the visual field to encompass many objects at once - for instance, across a natural scene - would affect the impact of attention. Essentially, these guidelines stipulate that the information available to the observer remains on a global level, though allowing for guidance of attention in a more focused way to a more local level. Treisman (2006) argues that the distribution of attention across a display with many objects is a "separate mode of processing" (p. 10) from that which is used under conditions of focused attention, which is at the opposite end of a single continuum. Distributed attention is used to extract global, statistical information from a display, whereas focused attention is required for binding features together and separating objects from one another. In response to the change in the way in which attention was voluntarily distributed across the display, the pattern of anisotropy was altered. With 100% certainty where the target would appear (compared to a condition where the observer was asked to spread their attention across all possible locations), the impact of cuing the lower visual field was marginally significantly larger than that in the upper visual field. A larger effect of endogenous orienting to the lower visual field replicates other work (e.g. He, et al., 1996). This result was accompanied by a significant difference between the effects of 176 attention on the horizontal and vertical meridians: the effect of attention orienting was greatest on the vertical meridian, particularly the lower half. These findings provide insight into the fact that the impact of attention is affected by the manner in which it is distributed. One theoretical approach to understanding how visual attention impacts perceptual experience is feature integration theory (e.g. Treisman & Gelade, 1980). This theory, originally used to explain findings in a visual search task, put forth that attention serves to integrate information about the presence of features detected by separate feature maps. For instance, separate feature maps would be responsible for the processing of color and shape; attention would put this information together, integrating the features into a single object presented at a particular location. The location information for the features would be present in the feature maps as a function of how the information was encoded, but without attention the fact that different features appeared at spatially coincident locations would not be perceived correctly. This is particularly relevant under conditions of visual search, where locating the conjunction of certain stimulus features, such as red and circle, would be important: attention would put those features together at the same location. This theory is also relevant under conditions where the location of the target stimulus is precued. In such situations (especially where the location is precued with 100% validity), the location of the target is not in question. In an identification task, the observer is asked to attend to a particular location and make an identification of the target that appears there. The features of the target must be integrated in order to make the identification. Attention, then, would serve to put together the features that appear at the attended location. When attention is distributed across space, the efficacy of integrating 177 the features of a stimulus in order to quickly and accurately identify it is reduced. The work described here further illustrates the impact of distribution of attention across a broader display in the face of less than 100% likelihood that the target will appear at the cued location - as well as how visual field location also interacts with the effects of attention. Changing the certainty with which the cue indicated the target location revealed an effect not found under conditions of 100% certainty. As seen in Experiment 2 of Chapter 2, in spreading some of their attention across the non-cued target locations in anticipation of the target not appearing at the cued location, observers found the greatest benefit of attention both close to (1°) and far away (12°) from fixation, with intermediary effects in-between (4° and 8°). What this indicates is that observers would "hedge their bets" about where the target would appear, and distribute their residual attention preferentially to the most felicitous group of uncued locations. Given the design of the display, that was a ring-like configuration in the middle of the potential target locations. A n additional change to the paradigm - presenting conditions where the cue indicated with 100% certainty where the target would appear versus no cue at all in a blocked, rather than mixed, design - illustrated an additional effect on anisotropy of performance. The effects found when the two types of trials were mixed disappeared, revealing two distinct groups of observers: one that oriented to the cue as instructed, and one that did not. This finding highlighted the importance of the way in which observers are asked to carry out the task: in studies where conditions had been presented in a blocked fashion (e.g. Mackeben, 1999), inconsistent effects of attention were found 178 between observers, particularly in differences in performance between the upper and lower visual fields. Under conditions where the number of possible locations to which the observer would need to distribute their attention was reduced, cue validity was further manipulated (Chapter 4). Varied between experiments, the cue's predictive value was either 70%, leaving a 10% chance that the target would appear at each of the other, uncued locations, or 40%, with a 20% chance of target appearance at each of the uncued locations. On its own, changing cue validity revealed little about changes in anisotropy of orienting as a result of altered attentional distribution. These studies did reveal, however, an impact of the presence of placeholders at the potential target locations - and the way that providing the observer with a perceptual object to which to attach their attention interacted with the cue's predictive validity. That is, when the cue predicted the target location with relatively high certainty - 70% - an inhibitory gradient centered at fixation (Dori & Henik, 2006) impaired performance on invalidly-cued trials when the target appeared 1 ° from fixation. Such a gradient, which is advantageous when a cue has instructed an observer to attend to a location in the periphery because it suppresses input that may come from fixation, clearly interferes with performance close to fixation when only a small amount of, or no, attention has been distributed to possible, but uncued, target locations. The decrement in performance on invalidly-cued trials at 1° of eccentricity from fixation also occurred under conditions of low cue validity (40%), and no placeholders. As indicated in Chapter 4, this finding seems contradictory to the previously described results: high probability and placeholders, and low probability and no placeholders result 179 in similar effects. It is plausible, however, that this similar effect was brought about by the same mechanism working under different conditions. Without a perceptual object to which to attach attention, selecting the four possible target locations likely became more difficult. Therefore, despite the higher likelihood that the target would not appear at the cued location - inducing the subject to work to override the inhibitory gradient close to fixation when there is a cue to attend elsewhere - without the aid of the placeholders the distribution of attention to the uncued locations was impeded. In this way, the same pattern of results could emerge from two entirely separate conditions, revealing how the likelihood that a target will appear at a cued location and the presence of placeholders interact. The finding that a placeholder presented only at the cued location eliminates this effect, mirroring conditions where the cue's validity was 70% and no placeholders were used, implicates the placeholders at the uncued locations as having an important role in effective distribution of attention across possible target locations. This effect, interestingly, also emerged under the conditions of Chapter 2, where the stimuli could appear at 32 possible target locations distributed across the display. A larger cuing advantage at 10 from fixation, upon re-examination of the results of Experiment 2, Chapter 2 (70% cue validity), originates in a decrement in performance 1° from fixation on invalidly-cued trials. The parallel findings between these two experiments with such different stimulus configurations implicates the inhibitory gradient in the selection of the central two eccentricities (4° and 8°) by the distribution of attention. The large impact of attention at 12°, however, still supports the idea that the pattern of distribution of attention across the four eccentricities revealed by the cuing effect was, in part, chosen for its proximity to all uncued locations. 180 Finally, the results of Chapter 3 reveal how altering the way that attention is distributed across the display impacts the inhibitory gradient. By introducing a possible target location at fixation and changing nothing else about the task, distributing attention across the horizontal meridian was facilitated, speeding responses to stimuli particularly on its left half. More interesting in Chapter 3, perhaps, is the emergence of a previously undocumented anisotropy in response time to targets presented on the intercardinal meridians as compared to the cardinal meridians. Responses to targets on the intercardinal meridians was consistently slower than that on the cardinal meridians - an effect not explained by either the greater attentional demands of crossing the horizontal or vertical meridians (e.g. Hughes & Zimba, 1985, 1987; Rizzolatti, Riggio, Dascola, & Umilta, 1987), or by previous findings that perceptual processing and the effects of exogenous cues at intercardinal locations fell between performance on the horizontal (best) and vertical (worst) meridians (e.g. Cameron et al., 2002; Carrasco et al., 2001; Carrasco et al., 2004). Faster response to targets on the intercardinal axes, interestingly, only emerges to a marginally significant degree in the results of the experiments in Chapter 2 (see Appendix A), which was not analyzed in that context as looking for this particular anisotropic pattern fell outside of the scope of that paper. However, faster response to stimuli presented on the diagonal radii did not emerge in any of the five experiments in Chapter 4. Why might this have occurred? The experiments in Chapter 4 required observers to respond to targets presented at one of four possible locations, all on the same side of the visual field. In no case was there any demand that the observer distribute their attention among the four locations, 181 while simultaneously crossing the one of the cardinal or intercardinal meridians. The absence of a meridian-crossing effect in the experiments of Chapter 4 would be an explanation consistent with previously documented difficulties with crossing the meridian, although in those cases, the difficulties arose when comparing cued trials to uncued trials, not on general performance overall (e.g. Hughes & Zimba, 1985, 1987; Rizzolatti, Riggio, Dascola, & Umilta, 1987). This is not the only possible explanation, however. What also differs between the experiments of Chapters 3 and 4 - as those are the two sets of experiments most similar to one another - are both the number of possible locations where a target might appear (8 or 9, vs. 4) and the degrees of visual angle across which the possible target locations are distributed (12° vs. 24°). Future work should attempt to prise apart which of these three possible factors induce slower performance overall on the intercardinal meridians. What is clear, however, is that this effect is most likely to arise from the vicissitudes of distributing attention across those locations; which stimulus parameters affect this process under these conditions remains to be investigated. Several commonalities in how visual field location affected performance occurred across experiments and paradigm manipulations. A l l of the experiments showed several important commonalities in performance that reflected consistent perceptual anisotropies. For instance, all experiments showed a right visual field advantage except under conditions without placeholders at all possible target locations (Experiments 3 - 5 , Chapter 4). Similarly, all experiments showed an advantage of responding to targets in the lower visual field. This effect interacted with attention in the experiments in Chapter 2. As with the right-left anisotropy, however, the significant RT advantage in the lower 182 visual field also disappeared with the absence of placeholders at the possible target locations for invalid targets (Experiments 3 & 4, Chapter 4). In light of all of these findings, future research should explore how what are currently understood to be perceptual anisotropies are affected by properties of the configuration of stimuli used. While outside the scope of the work discussed here, the fact that properties of the stimuli used could impact well-documented perceptual anisotropies indicates a further role of the endogenous spread of attention, rather than simply perceptual processing. From all of these experiments, the following guidelines for how endogenous attention has an impact across the visual field were compiled: (1) When attention is fully allocated, to the extent required to do the task, to a cued location, the effects of attention are anisotropic at different visual field locations. In this case, attention has the greatest impact on the vertical meridian, relative to the horizontal meridian, with the lower half of the vertical meridian receiving the greatest benefit of attention. This finding is consistent with a lower-visual-field advantage for endogenous attention (e.g. He et al., 1996). (2) Under conditions where the cue does not indicate with 100% certainty where a target will appear, attention is preferentially allocated to the cued location in a manner that reflects cue validity, but residual attention is also simultaneously distributed among other possible, uncued, target locations. (3) The distribution of attention among uncued target locations is determined by at least three factors: the most advantageous strategy for orienting attention based on the demands of the display, the influence of a gradient of inhibition centered at fixation (Dori & Henik, 2006), and the use of perceptual objects to aid in the 183 selection of the attended locations in the display (placeholders). These factors interact with one another to create an anisotropic pattern of performance in responding to targets at various locations in the visual field. That this anisotropy varies with the manipulation of these factors indicates that the anisotropic effects are not simply perceptual: they are a function of the manner in which attention is voluntarily allocated across a display as a result of task demands. (4) Individual strategic differences should be given consideration when evaluating the effects of visual attention at various visual field locations, as the way in which the stimuli and conditions are presented to the observer have an influence over the way cues are used to facilitate performance. As a whole, these guidelines could be used to inform future theoretical understanding of endogenous, covert orienting. The manipulations used in the ten experiments detailed here are highly simplified versions of the ways that the natural visual environment can vary from scene to scene. Therefore, in order to achieve a fuller understanding of the way that attention is voluntarily allocated under normal viewing conditions, it is essential to understand the role of factors like the ones explored here. Furthermore, the interaction of these simple manipulations with each other implies that the way in which endogenous attention is allocated is highly complex in the natural world, and is far from being entirely understood. The work here represents a starting point from which to approach the impact of stimulus characteristics on attention distribution and consequent performance. 184 5.1 References Cameron, E.L., Tai, J.C., & Carrasco, M . (2002). Covert attention affects the psychometric function of contrast sensitivity. Vision Research, 42, 949-967. Carrasco, M . , Giordano, A . M . , & McElree, B. (2004). Temporal performance fields: visual and attentional factors. Vision Research, 44, 1351-1365. Carrasco, M . , Talgar, C , & Cameron, E.L. (2001). Characterizing visual performance fields: Effects of transient covert attention, spatial frequency, eccentricity, task, and set size. Spatial Vision, 15, 61-75. Dori, H. , & Henik, A. (2006). Indications for two attentional gradients in endogenous visual-spatial attention. Visual Cognition, 13(2), 166-201. He, S.H., Cavanagh, P., & Intriligator, J. (1996). Attentional resolution and the locus of visual awareness. Nature, 383, 334-337. Hughes, H.C., & Zimba, L.D. (1985). Spatial maps of directed attention. Journal of Experimental Psychology: Human Perception and Performance, 11(A), 409 -430. Hughes, H.C., & Zimba, L.D. (1987). Natural boundaries for the spatial spread of directed visual attention. Neuropsychologia, 25(1 A), 5-18. LaBerge, D. (1995). Attentional processing: The brain's art of mindfulness. Cambridge, M A : Harvard University Press. LaBerge, D., Carlson, R.L., Williams, J.K., & Bunney, B.G. (1997). Shifting attention in visual space: Tests of moving-spotlight models versus an activity-distribution model. Journal of Experimental Psychology: Human Perception and Performance, 23(5), 1380- 1392. 185 Mackeben, M . (1999). Sustained focal attention and peripheral letter recognition. Spatial Vision, 12, 51-72. Rizzolatti, G., Riggio, L., Dascola, I., & Umilta, C. (1987). Reorienting attention across the horizontal and vertical meridians: evidence in favor of a premotor theory of attention. Neuropsychologia, 25(1 A), 31-40. Treisman, A. (2006). How the deployment of visual attention determines what we see. Visual Cognition, 14(4-%), 411 - 443. Treisman, A. , & Gelade, G. (1980). A feature integration theory of attention. Cognitive Psychology, 72 ,97 - 136. 186 Appendix A: Supplementary analyses Comparison of performance at cardinal and intercardinal locations for Experiments 1 & 2, Chapter 1 SS df MS F P Partial n2 Experiment 1 1967.04 1 1967.04 5.34 0.03 0.2193 error 7002.17 19 368.54 SS df MS F P Partial n Experiment 2 3268.43 1 3268.43 5.68 0.03 0.25 error 9776.33 17 575.08 187 Appendix B: Results summaries Table B.l . Summary of results from Experiments 1,2, and 3 in Chapter 2. Significance cutoff for difference scores: p = 0.01 Significant results: overall ANOVA Planned comparisons: Difference scores (invalid - valid) Reaction Time Errors Horizontal vs. vertical meridian Upper vs. lower visual field Vertical meridian asymmetry Right vs. left visual field Eccentricity comparison Experiment 1: 100% valid cues vs. no cues - Cue condition x Radius - Main effects radius, eccentricity, cue condition - Main effect of cue condition p = 0.004 p = 0.019 p = om p = 0.23 No significant linear relationship (p = 0.54) Experiment 2: 70% valid cues vs. 30% invalid cues - Validity x Eccentricity - main effects radius, eccentricity, cue condition No interactions or main effects p = 0.62 /? = 0.011 p = 0.005 p = 0.96 Significant quadratic relationship (p = 0.004) Experiment 3: 100% valid cues vs. no cues - blocked design - A l l observers: main effects of Radius and Eccentricity - Only observers with cuing effects: Cue condition x Eccentricity interaction For all observers and observers with cuing effects, no main effects or interactions in error data p = 0A5 p = 0.24 p = 0.44 p = 0.S3 Only observers with cuing effects: linear relationship (p = 0.012) Table B.2. Summary of results from Experiments 1 and 2 in Chapter 3. Significance cutoff for planned comparisons: p = 0.01 Significant results: overall ANOVA Planned comparisons Reaction time Errors Horizontal vs. vertical meridian Upper vs. lower visual field Vertical meridian asymmetry Right vs. left visual field Cardinal vs. intercardinal Experiment 1: No potential target at fixation - Main effects of Radial location, Eccentricity, and Validity Main effect of Eccentricity p = 0M p = 0.004 p= 0.47 /? = 0.0003 p = 0.002 Experiment 2: potential target at fixation - Main effects of Radial location, Eccentricity, and Validity No significant main effects or interactions p = 0.03 p = 0.02 p = 0.04 p = 0.0095 p = 0.0004 Table B.3. Summary of results from Experiments 1 - 5 in Chapter 4. Significance cutoff for planned comparisons: p = 0.01 Significant results: overall ANOVA Planned comparisons Reaction time Errors Horizontal vs. vertical meridian Upper vs. lower visual field Vertical meridian asymmetry Right vs. left visual field Cardinal vs. intercardinal Experiment 1: 70% valid; placeholders Eccentricity x Radius Eccentricity x Validity Main effects of Radial location, Eccentricity, and Validity Main effect of eccentricity p = 0.20 p = 0.001 p= 0.10 /? = 0.008 p = 0.94 Experiment 2: 40% valid; placeholders Main effects of Radial location, Eccentricity, and Validity None p = 0.89 /? = 0.0006 p = 0.07 p = 0.005 p = 0A4 Experiment 3: 70% valid; no placeholders Eccentricity x Radial location Main effects of Eccentricity, Validity Main effect of Radial location p = 0.66 /7 = 0.10 p = 0.57 p = 0.08 p = 0.76 Experiment 4: 40% valid; no placeholders Eccentricity x Radius Eccentricity x Validity Main effects of Radial location, Eccentricity, and Validity Main effects of Radial location and Eccentricity /? = 0.39 p = 0.03 /? = 0.03 p = 0.24 p = 0.05 Experiment 5: 70% valid; placeholder only where 1 cue appears Eccentricity x Radius Main effects of Radial location, Eccentricity, and Validity None p = 0.64 p = 0.002 p = 0.03 p = 0.008 p = 0.49 

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0100740/manifest

Comment

Related Items