v

. «a

 

"Y "
Va.“

A

or":

 

9023
5H9IH’750 LIBRARY

 

Michigan State
University

 

 

This is to certify that the
dissertation entitled

AGE DIFFERENCES IN VISUAL MEMORY AND THE
RELATION TO EYE MOVEMENTS AND EXECUTIVE
CONTROL PROCESSES IN VISUAL SEARCH

presented by

Carrick C. Williams

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Psychology

 

 

f2“ 7. 7a {a

Major Professor's Signature

37/3/47 3

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

o.---o---o-.-------o-.---.-----o---- .n-c-o--o----o-o-.-----o--u-u--a-----n-o---a-----.-.-o—----.-.-v— —--

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DataDuepGS—p. 1 5

 

AGE DIFFERENCES IN VISUAL MEMORY AND THE RELATION TO EYE
MOVEMENTS AND EXECUTIVE CONTROL PROCESSES IN VISUAL
SEARCH
By

Carrick C. Williams

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology
2003

ABSTRACT
AGE DIFFERENCES IN VISUAL MEMORY AND THE RELATION TO EYE
MOVEMENTS AND EXECUTIVE CONTROL PROCESSES IN VISUAL
SEARCH
By
Carrick C. Williams
The current study examined executive control processes in visual search
and their consequences for visual memory of the searched objects. Younger and
older adults counted the number of targets (identified by a color and category,
e.g., yellow car) present in an array containing 12 real-world objects. In addition
to the targets, the array contained distractors that had differing relationships to
the search target: color distractors, category distractors, and unrelated distractors
(objects that were neither the same category nor color as the search target).
Executive processes were expected to allocate attention differently to the various
classes of objects, leading to varying levels of memory for the different types of
distractors. Participants’ memory was tested using a surprise token
discrimination task after a 10 minute filled interval. Memory for targets was
reliably better than for distractors. Of most interest for this study, however, was
memory for the distractor items. Color and category distractors were
discriminated from foils at similar levels and both were discriminated reliably
better than the unrelated distractors. Additionally, younger adults remembered
the targets better than older adults, but memory for the distractors was equivalent
for the two age groups. Experiment 2 also examined the eye movements that

were made during the search task. These measures indicated that viewing

behavior during search was related to the memory that was maintained for the
objects. There was an additional advantage for target objects with targets being
remembered better than distractors regardless of the number of times they were
looked at or how long they were looked at. Older and younger adults were very

similar in their viewing behavior.

ACKNOWLEDGEMENTS

There have been many people that have assisted me throughout my time
at Michigan State University. First and foremost, I would like to thank Rose
Zacks, who has supported me throughout my time here. She has taught me how
to be a better researcher and writer and how to think like a researcher. I will
always be indebted to her for her constant support. I would also like to thank
John Henderson who allowed me to be a part of his eye movement laboratory
and encouraged me to pursue this line of research. I would not have been able to
get to this point without his help. I would also like to thank the members of my
dissertation committee, Tom Carr and Zach Hambrick for challenging me and
requiring me to think more deeply about the issues.

There have been many people who have walked in the halls of the
Psychology Research Building that have helped me in many ways. The members
of the Cognitive Aging Lab and the Eyelab contributed to this work to this in a
variety of ways. I would especially like to thank Karin Butler, Doug Davidson,
Kate Arrington, Richard Falk, and David Thomas. Each of these people has
contributed in a variety of ways to my development as a researcher and a
person. I also want to thank Sarah Burnett, Janna Whittenberg, and Katie
Paquette for there assistance in running subjects and finding the hundreds of
pictures that were the stimuli for these experiments. Additionally, I need to thank
Gary Schrock and Andrew Hollingworth for their technical help with the eye

tracker. Experiment 2 would not have happened without them.

iv

There are two women who need special thanks. The first is my wife, best
friend, and copy-editor, Camilla. Her love is never-ending and helped get me
through the trials that accompany graduate school. I hope that I can support her
the way that She has supported me. Secondly, I would like to thank Kathleen for
listening and helping me understand how it all fits together. Without these
women, I would not have made it this far.

Finally, I want to thank my parents and my sister. As the son of two
educators, school has always had an important role in my life. I believe that it

was their emphasis on education that helped me to decide to pursue this career.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................... viii
LIST OF FIGURES ............................................................................................... ix
INTRODUCTION ................................................................................................... 1
Visual Search ............................................................................ 4
Executive Control and Visual Search .............................................................. 7
Aging and Visual Search .............................................................................. 17
Visual Memory .............................................................................................. 19
Object Representations and Visual Memory ................................................. 22
Current Study ............................................................................................... 24
EXPERIMENTS 1A AND 18 ............................................................................... 27
Method .......................................................................................................... 32
Participants. ......................................................................................... 32

Design. ................................................................................................. 33
Materials .............................................................................................. 33

Visual Search. ...................................................................................... 35

Visual memory test ............................................................................... 36
Procedure ............................................................................................ 37

Results .......................................................................................................... 39
Visual Search ........................................................................................... 39
Visual Memory Test .................................................................................. 42
Discussion .................................................................................................... 49
EXPERIMENT 2 .................................................................................................. 53
Method .......................................................................................................... 57
Participants. ......................................................................................... 57

Design. ................................................................................................. 57
Materials. ............................................................................................. 58
Equipment ............................................................................................ 58
Procedure ............................................................................................ 59

Eye tracking analyses. ......................................................................... 60

Results .......................................................................................................... 60
Search Results ......................................................................................... 60
Visual Memory Test ............................................................................. 63

Eye Movement Results ............................................................................. 67
Proportion Viewed. ............................................................................... 68

Entry Analysis. ..................................................................................... 70

Total Time Analysis. ............................................................................. 72

Viewing Block and Target Count Analyses .......................................... 73
Regression analyses ................................................................................ 76
Entries .................................................................................................. 76

vi

Total Time ............................................................................................ 79

Discussion .................................................................................................... 82
GENERAL DISCUSSION .................................................................................... 89
Summary and Conclusion ........................................................................ 94
APPENDIX .......................................................................................................... 96
REFERENCES ................................................................................ ' .................... 98

vii

LIST OF TABLES

Table 1. Mean (Standard Error) for Age, Education Level, Vocabulary, and
Momingness/Eveningness Questionnaire (MEQ). .......................................... 33

Table 2. Mean (Standard Error) for (A) Correct Search Times (in ms) and (B)
Search Accuracies (proportion correct) for Experiment 1a and 1b. ................ 40

Table 3. Mean (Standard Error) for (A) Correct Search Times (in ms) and (B)
Search Accuracies (proportion correct) for Experiment 2. .............................. 63

viii

LIST OF FIGURES

Figure 1. Example of a search display. Images in this study were presented
in color. For photocopying purposes, the background presented in this
image is lighter than was true in the experiment. The search targets (1
green drill in panel A and 3 green drills in panel B) are highlighted by the
white boxes. Other objects include a color distractors (green watering can,
upper right), category distractor (yellow drill, upper middle), and unrelated
distractors (yellow-handled axe, right center). ................................................ 31

Figure 2. Example of visual memory test stimuli. Participants had to
determine which green drill token (left or right) had been presented in the
earlier search arrays. Images were presented in color and were the same
size as during the search portion of the experiment. For photocopying
purposes, the background presented in this image is lighter than was true
in the experiment. ........................................................................................... 37

Figure 3. Mean proportion correct for the visual memory test in (A)
Experiment 1a and (B) Experiment 1b. The error bars are standard errors. ...46

Figure 4. Mean proportion correct for the visual memory test in Experiment 2.
The error bars are standard errors .................................................................. 64

Figure 5. A example of the scan pattern for a younger participant looking for
white telephones. The saccades are represented by the lines and fixations
are represented by the circles. The size of the circle is proportional to the
fixation duration. The three white telephones are highlighted in the figure
by the black boxes, which are the size of the scoring region used. Any
fixation that fell within the region was counted as on the object. The color
distractor was the white pipe (lower left corner), the category distractor
was the yellow telephone (middle right), and the unrelated distractor was
the green lady‘s wallet (left center). For photocopying purposes, the
background presented in this image is lighter than was true in the
experiment. ..................................................................................................... 68

Figure 6. Mean proportion of memory-tested objects entered during the
visual search task in Experiment 2. The error bars are standard errors. ........ 70

Figure 7. Mean number of entries into the object regions of the tested objects
during search in Experiment 2. The error bars are standard errors. ............... 71

Figure 8. Mean total fixation time within the object regions of the tested
objects during search in Experiment 2. The error bars are standard errors. ...73

Figure 9. Memory test accuracy as a function of entries. The data points are
the average of each quartile of entries within each condition. The panel A

ix

contains the data for younger adults and panel B contains the data for
older adults. .................................................................................................... 78

Figure 10. Memory test accuracy as a function of total time. The data points
are the average of each quartile of total time within each condition. The
panel A contains the data for younger adults and panel B contains the
data for older adults. ....................................................................................... 81

INTRODUCTION

When looking for a specific book (9.9. a red book) that is in a less than
perfectly organized office, one is likely to encounter books that are not the same
color, things that are the same color but are not books, and things that are
neither books nor the same color. To find a red book in the office, one needs to
ignore the distracting objects in the environment and focus on the combination of
features that define the target book. The current project was interested in the
memory for objects that had been searched earlier, both as targets and
distractors. The different objects (i.e., the targets of visual search and the
distractors) in the search for the red book are expected to receive differential
processing based on their relationship to the search target. It seems reasonable
to predict that the relationship of an object to the search target should influence
the type of processing that is involved. If a person is looking for a red book, he or
she is more likely to glance at a red coffee cup or a blue book (because they
share at least one characteristic that defines the target of the search) than a
yellow pencil (an object that does not share any characteristic of the search
target). This sort of difference in processing could influence the memory that
remains for the different objects that were encountered. Additionally, the person
would have an advantage if he or she avoided, quickly stopped processing, or
quickly forgot the distractor objects so that the search proceeds efficiently. The
process of rejecting distractors also most likely affects the memory that remains

for the objects that were searched.

The current study examined the role of executive control processes in the
selection of target objects in the environment as well as in the ignoring of
distractors and the consequences of the search behaviors on visual memory.
Specifically, a search paradigm involving displays of 12 pictures of real-world
objects was used to investigate how and how much people process different
types of objects in the environment and what they remember. When looking
through an array of objects, the choice of what to look for segregates the
environment into 3 main classes: the object being sought (the target), objects
related to the thing being sought (related distractor objects), and objects
unrelated to what is being sought (unrelated distractor objects). The experiments
described here investigated the effect on memory of the differential processing
that executive control processes apply to these different classes of objects. One
specific executive control process investigated here is inhibition. Inhibition could
play a role by halting the processing of objects that are not the targets of the
search. On the assumption of age-related differences in inhibitory control
(Hasher & Zacks, 1988), the current study included comparisons of younger and
older adults as one means for elucidating the role of inhibitory processing in
visual search.

The current project also addressed a topic of debate in the visual
processing literature: the extent to which people can remember the visual details
of objects once they are out of view. There has been convincing evidence from
transsaccadic studies that people do not build up a veridical representation of the

visual environment (9.9. Irwin, Zacks, & Brown, 1990; Irwin & Zelinsky, 2002).

Some recent theories have gone so far as to claim that no visual information is
maintained (see below). This claim has been questioned by a more moderate
stance that claims that although a veridical picture is not stored, some visual
details are maintained. The current study sought additional evidence to address
whether visual details are maintained for a period of time after attention has been
withdrawn from an object. To examine visual memory, the current project used a
visual search task in which every object presented was a unique token. The
memory test required participants to discriminate between a previously viewed
object token and a nonpresented object token of the same type and color. The
only information that could be used to distinguish the tokens was the visual
details of the tokens because the semantic information was identical. By using
unique tokens, the visual memory for individual targets and distractors from the
previous search could be evaluated. If memory were maintained for either or both
of these types of items, this would argue against the memoryless models of
visual perception. Additionally, because the different objects in the search task
were more or less likely to be processed, the current project examined the effect
of variations in processing on likelihood of visual memory.

Finally, this study examined the eye movements that occur while looking
through an object array for a specified target. The eye movement record
indicated the probability that an object was looked at, how long different objects
were looked at, and whether they were looked at multiple times. These measures
could be informative with regard to the two issues that have been discussed thus

far, executive processing in search and visual memory.

The following sections divide different aspects of this dissertation into the
two main components: visual search and visual memory. For the visual search
component, I provide a brief review of the main points in the literature and then
examine the executive processes that can affect the processing of the objects in
the search, with a special emphasis on inhibition. The visual memory component
focuses on the memory that may or may not be maintained after visual
processing.

Visual Search

The office description above is an example of a visual search task and
demonstrates some complexities that exist when one is looking for a target in a
complex environment. Visual search has been extensively studied in cognitive
psychology. However, the search tasks are usually much simpler than the office
example described above. There are two types of search that are typically
studied: feature search and conjunction search. In both types of search tasks, the
participant typically searches the display for the presence or absence of a
specified target (e.g. Treisman & Gelade, 1980; Treisman & Sato, 1990; Wolfe,
Cave, & Franzel, 1989). In a feature search, the target is defined by a feature
that the distractor objects lack (e.g. a red T in a field of green Ts). In this case,
the target seems to “pop-out” and search proceeds at approximately the same
rate regardless of the number of distractor items that are present (Treisman &
Gelade, 1980). A conjunction search, on the other hand, defines the target item
by the conjunction of two features (usually a shape and color such as red T)

within a field of distractor items that each contains one of the target features, but

not both (e.g. a field of red Os and green Ts). This search task is difficult and
requires that participants search through the items in the display individually. This
is a much less efficient search than the feature search, and the more items there
are in the array, the more difficult the search task is, as indicated by increased
RTS and errors (T reisman & Gelade, 1980; see Wolfe, 1998 for a review).

One of the first comprehensive theories of the processes involved in visual
search, Feature Integration Theory, was provided by Treisman and colleagues.
Treisman and Gelade (1980) claimed that feature search involves a parallel
search process that can examine the entire display and detect the presence of a
target feature. This type of search is very rapid and efficient (e.g. under 10
ms/item in the display). On the other hand, according to these authors,
conjunction search is a serial, attention-dependent process that requires the
viewer to examine, on average, half of the items in the display to find a present
target. In Feature Integration, attention is required during a conjunction search to
bind the various features (e.g., color and identity; red T) of an item together. A
decision about whether or not the item is a target can only proceed after the all of
the features that define the item have been bound together. Because the
distractors in a conjunction search contain one of the target features, the
distractors are likely to receive some activation from the feature it shares with the
target, consequently drawing the searcher to them. The serial conjunction search
is considerably slower than the feature search and the search time is directly
proportional to the number of item in the display (i.e., the search slopes are

positive and significantly greater than 0).

Although Treisman and Gelade’s (1980) theory provided a basic
description of the processes of visual search, there have been demonstrations of
search phenomena that do not conform to their theory. Researchers have
demonstrated instances of conjunction searches that have shallower search
slopes than would be expected (e.g. Treisman & Sato, 1990; Wolfe, et al, 1989).
These shallower search slopes are claimed to be the result of the participant’s
ability to segregate the search display using one of the features that define the
target. Wolfe et al. (1989; Wolfe, 1994) describe this as a guided search, which
allows the participant to perform an efficient search even in a conjunction search
task. That is, given that participants are able to limit a search to a subset of the
displayed items (e.g. only the red items in a display of red Xs and green Os in
which the target was a red 0), they are able to perform a feature search for the
target.

Much has been learned about the processes involved in search using the
techniques described above, but these searches differ from real world searches
in several ways with regard to the types of stimuli that are present. In the typical
experimental search task, the participant is looking through a rather homogenous
set of items (e.g., letters on a blank background). Search arrays typically contain
multiple instances of a small set the distractor items (e.g., several red Xs and
green Os) repeated on all trials. These items tend to be impoverished in
complexity and detail compared to real world objects limiting the extension of the

research to more complex real world scenes. In scenes, objects vary on many

features (shades of color, orientation, etc.), which could influence the ability of
the searcher to find the target object.

The current study took an intermediate step between scenes and typical
search tasks by using pictures of real-world objects in a conjunction search task.
The objects in the search arrays were targets, distractors that were related to the
search target (by either being from the same category or having the same color
as the target), and objects that were unrelated to search target. Unlike most
research on conjunction search, the targets and distractors were single instances
of the particular object (rather than multiple instances of a small set of
distractors). A particular instance of an object (an object token) occurred in only
one array in the search portion of the experiment. With each object being
presented in only one array, the memory for the different objects within each
classification in the search space can be examined.

Executive Control and Visual Search

The application of attention to the items within the search space appears
to play a critical role in visual search (e.g. Treisman 8 Gelade, 1980; Treisman,
1998; Wolfe, 1998). While some processing appears not to require the direct
application of attention (see Treisman, 1998 and Wolfe, 1998 for a discussion of
preattentive processing), there is evidence that more complex searches, like the
ones described in this project, require attention to be focused on the various
items in the search in order to locate the items (e.g. Shih & Sperling, 1996),

conjoin the features of an item (e.g. Treisman, 1988; Treisman & Gelade, 1980),

or, more generally, for further processing (e.g. Wolfe, 1994; Wolfe, Oliva,
Horowitz, Butcher, & Bompas, 2002).

Because the search task in the current experiment contained objects that
differed in their relationship to the searched-for object, the allocation of attention
was expected to be unequal for the different classes of objects. The concept of
guided search (Wolfe, 1994; Wolfe et al. 1989) indicates that attention is not
evenly spread throughout the search display. The guided search notion implies
that attention is guided to objects that are likely to be the target rather than in a
random fashion.

The selection of which objects to attend to in a visual search task can be
influenced by a variety of processes that may bias selection toward some objects
and away from others. In this dissertation, I refer to processes that control the
deployment of attention to the various objects in the search array as Executive
Control Processes (e.g. Banich, 2004; Gazzaniga, Ivry, & Mangun, 1998).
Executive control processes guide attention to objects that are potential targets
and away from those that are less likely to be targets. Additionally, executive
control processes can limit (stop) the ongoing processing of nontarget objects
that have attracted some initial attention. This definition highlights the importance
of the current goal of the system to the allocation of attention. As was stated in
Kramer, Scialfa, Peterson, and Irwin (2001),“... attentional control relies on an
observer's expectancies about events, and the ability to develop and maintain an

attentional set for particular kinds of environmental events.” (p. 296).

The focus on attention leads to the issue of measuring where attention is
at a particular time. One way to measure where attention is at any one time is
through eye movement data. When viewing a scene, the eyes have periods of
relative stability (called fixations) punctuated by periods of rapid movement
(called saccades). Visual information is effectively acquired only during the
fixation periods. The saccades move the eyes rapidly to the next point that the
observer intends to process (see Rayner, 1998 for a review). In normal vision,
attention tends to precede the eyes to the location of the next fixation and
remains there for part of the following fixation (Deubel & Schneider, 1996;
Henderson, Pollatsek, & Rayner, 1989; Hoffman & Subramaniam, 1995; Rayner,
McConkie, & Ehrlich, 1978). This argues for the use of viewing position as a
marker for the location of attention. In other words, if a person views a particular
object in a scene or array, it is highly probable that attention has been devoted to
the viewed object.

A demonstration of the interrelatedness of the executive control
processes, attention, and eye movements is provided by the attentional capture
paradigm of Kramer, Theeuwes and colleagues (Kramer, Hahn, Irwin, &
Theeuwes, 1999; Theeuwes, Kramer, Hahn, & Irwin, 1998). They demonstrated
that attention could be limited in the objects that are attended by either making
the stimulus salient enough to be detected or by allowing attention to be
adequately focused on the target location. More specifically, participants were
presented with an array of 8 grey circles. On each trial, 7 of the circles would

change color and participants were instructed to move their eyes to the one that

did not change. To examine attention capture, on some trials, concurrent with the
color change, another circle was added to the display. The measure of
attentional control was the probability of fixations on the added circle even
though it was never a target. If attentional control were perfect, the participant
would never look to the added circle. However, these studies found that the eyes
were drawn to the added circle if it was equiluminant with the other distractors or
if there was no indication of the location of the target beforehand. In other words,
attentional control failed if the stimulus did not stand out from the background or
if the executive process involved in targeting the eye movement was not given
adequate focus. On the other hand, if the added circle was salient (i.e. brighter
than the other distractors) or if the target location was predictable, the eyes were
not directed to the new circle. This indicates that the executive control processes
involved can stop the deployment of the eyes to something that would normally
draw attention.

Attention is also critical in the formation of memories of the objects that
have been encountered in search. Treisman and colleagues (T reisman, 1998;
Kahneman, Treisman, 8. Gibbs, 1992) claimed that attention was critical in the
formation of object representations maintained by the visual system (see the
discussion of object files below). This claim was expanded (Hollingworth &
Henderson, 2002; Hollingworth, Williams, & Henderson, 2001) to compare the
relation between fixations and long-term memory of objects. These authors
claimed that as objects are attended/fixated, a long-term memory representation

is built. Hollingworth et al. (2001) state that “. .. over multiple fixations on a scene,

10

visual information from local objects accumulates in memory (p. 762).
However, fixation is not the only influence on memory. Friedman (1979) found
that recognition memory was not as strongly related to total fixation time as it was
to the probability that the object did not belong to a particular scene. Thus,
attention, as indicated by fixations, is a component in the formation of memory
but it appears that fixation pattern is not a perfect predictor of memory.

The executive control processes in search involve the allocation of
attention (as evidenced by memory measures and eye movement data) to the
various objects in the search space. Initially, they can divide the objects into
those that will be searched and those that will not be searched based on a
feature of the target (as described by Treisman & Sato, 1990). Following this
segregation, attention is directed to the potential targets in order to find the
target. The executive control processes potentially play another role here
because if the object is identified as a distractor, the processing of that object
should be halted and attention directed to another object. Following the shift in
attention, the eyes move to the next potential object. The earlier that this occurs
in the processing of the distractor, the more efficient the search can be (i.e., the
faster the searcher can move on and find the target).

In the current study, if the searcher’s gaze falls upon an unrelated
distractor, he or she could quickly reject it as not the target due to the fact that it
shares no features with the target, halting the processing of the unrelated
distractor quickly. The related distractors, on the other hand, may require more

processing to determine that they are not the target before they can be rejected.

11

Finally, when a target was encountered in the current task, the searcher not only
had to identify it as a target, but he or she also had to increment the target count.
This should have required somewhat more extensive processing of the target
objects. Overall, search was expected to proceed quickly through objects not
related to the targets and to dwell on objects that were related to or were the
targets of the search, as evidenced by differences in the memory measures and
the eye movement record for the search.

Inhibition appears to play a critical role in the executive control processes
that underlie efficient search. Specifically, research (e.g., Koshino,
2001 ;Treisman, 1998; Treisman & Sato, 1990) has suggested that inhibition
plays a role in limiting the processing of nontargets in visual search. In the visual
search context, inhibition should work to limit the processing of objects that are
not the target of the search. There are two ways that inhibition could work to limit
the processing of distracting objects. The first is to inhibit the identity of the items
in the search space. Using the features that define the item as a distractor, the
identity of the distractor can be inhibited so that it is either never processed or, if
processed, is more difficult to reaccess at a later time. The second is on the
locations of the distractors in the display. In other words, inhibition could be
applied to the spatial points that were occupied by distractors so that anything
that occurs in these locations will be harder to access.

There are several phenomena within the context of visual search that
appear to result, at least in part, from inhibition operating on the distracting items

in the display or on their locations. These include the segregation of the visual

12

field (i.e. guided search), visual marking, inhibition of return, and negative
priming. Although Wolfe (1994) claimed that guided search was the result of a
process that selected areas of high target probability, Treisman (1998), argued
that some of the ability to segregate the search space using a particular feature
was the result of objects not related to the target being “selectively shut out” (p.
38) On a similar note, Treisman and Sato (1990) stated that the ability to
segregate the search space using features is the result of an inhibitory
mechanism that acts on the distracting features. By inhibiting the nontarget
features, the search space can be effectively reduced to only those items that
contain a specific, selected-for feature.

Visual marking is another phenomenon that appears to involve inhibition
(Watson & Humphreys, 1997; 1998). In a visual marking experiment, a person is
given a modified conjunction search task in which the search display is presented
in two stages. In the first stage, half of the distractors are presented. After a short
period of time, the remaining distractors and the target (if one is present on that
trial) are added to the distractors from the first stage. This modified search
condition, called the gap condition, is compared to search through the same
number of items when they are presented all at one time rather than in two
stages (a standard conjunction search task). Watson and Humphreys found that
search in the gap condition was much more efficient than standard search task,
indicating that the searcher was effectively searching through only those items in
the second array. They claim that the searcher can use the presentation of the

first stage in the gap condition to mark those distractors as not containing the

13

target and thus are only required to search the items presented in the second
stage. In other words, the visual system marks the locations of the distractors
presented in the first stage as irrelevant and thus not needing to be searched
when the second stage items are added. Because the items in the first array no
longer influence search, Watson and Humphreys argue that these nontarget
items have been inhibited.

The third phenomenon described as an inhibitory process is Inhibition of
Return (IOR; Posner & Cohen, 1984). IOR occurs when attention is initially
drawn to a location or object and then proceeds to another location. If attention is
required to return to the initially cued location (in order to make a response),
participants are slower to respond to the target than if it appears at a new
location/object. In other words, because the initially cued location was rejected by
the attentional system, inhibition was applied to prevent its reselection later, and
thus the aftereffect is seen when the participant returns to the location. IOR can
also be found for objects independent of the spatial location the object occupies.
If an object moves after it has been attended, IOR seems to follow the object in a
manner that is consistent with the object-file framework described below. Thus,
although it is primarily thought of as a spatial phenomenon, the inhibitory
processes involved in IOR may also have an object—based aspect. This means
that under certain circumstances if a once-attended object moves, attention will
still be less likely to return to that object (Tipper, Weaver, Jerreat, & Burak,

1 994).

14

The final phenomenon related to the selection of a target and the inhibition
of distractors is negative priming (e.g. Tipper, 1985; see May, Kane, & Hasher,
1995 for a review). In negative priming experiments, a target and a distractor are
presented on each trial. The participant is instructed to respond to the target
while ignoring the distractor. On negative priming trials within the experiment, the
distractor from the previous trial becomes a target on the current trial. This
condition is compared to control trials where there is no relationship between the
current trial and the previous trial. What has typically been found is that
responses on negative priming trials are slower and less accurate than on control
trials. Tipper claimed that these slower responses were the result of an inhibitory
mechanism that acted on the representation of the distractor object making it
more difficult to access. Negative priming has also been shown to act on the
location occupied by a distractor (Tipper, Brehaut, & Driver, 1990). If the
participant attempts to access a location that contained a distractor on the
previous trial, response times are longer than if there was nothing in that location
on the previous trial.

In each of these cases, it can be argued that inhibition plays the role of
limiting the influence of information that has been rejected. In the case of guided
search, the items that can not possibly be the target are ignored (however, see
Wolfe, 1994, for a description of this ability without inhibition). For visual marking,
the items that are marked (rejected as not containing the target) are ignored
when the additional items are added. In IOR, the location that had been attended

to that did not contain the target is more difficult to return to. Finally, in negative

15

priming, the object that was ignored on the previous trial because it was a
distractor is more difficult to respond to when it subsequently becomes the target
(however, see Neill, Valdes, Terry, & Gorfein, 1992, for an alternative explanation
of negative priming). Each of these phenomena contains some aspect to be
ignored because it is not or does not contain the target, and processing of that
information is either halted or actively suppressed. They appear to apply
inhibition to different aspects of a search task and use different functions of
inhibition. Although the display types used in most studies of visual search are
less complex than the arrays of pictures of real-world objects to be used in the
current study, the description of inhibition as a limiting process that can act on
the location or identity of an object is applicable here.

As was mentioned earlier, the current study was also interested in age-
related differences in visual search. Thus, these inhibitory processes will be
considered in relation to the age and inhibition views of Hasher, Zacks and
colleagues (Hasher 8. Zacks, 1988; Hasher, Zacks, & May, 1999). These authors
claim that in the course of normal aging, decreases in the efficiency of inhibition
can lead to processing too much irrelevant or distracting information. This affects
the ability of the person to focus on and process the relevant information. For
example, Connelly, Hasher and Zacks (1993) presented younger and older
adults with passages to read in one font interspersed with distracting information
in another font. They found that older adults were differentially affected by the
distracting text compared to younger adults, especially when the distracting

information was related to the passage. In other words, older adults were less

16

efficient at ignoring related nontarget information than younger adults, which was
indicative of decreased inhibitory efficiency. In the current study, supporting
evidence that inhibition is operating in visual search would be present if the older
adults searched differently than younger adults and had different memory
representations of the objects searched. These differences should be observed
in the distractor object conditions because they are the ones that contain
competing information that would interfere with the search task.
Aging and Visual Search

Visual search has not been a major topic of study in the field of age-
related changes in cognition. Initial studies (Plude 8. Doussard-Roosevelt, 1989;
Zacks & Zacks, 1993) indicate that similar search patterns are present for both
older and younger adults. Both age groups are able to effectively use feature
searches and are slowed significantly in conjunction searches. However, older
adults are slowed more than younger adults by the demands of a conjunction
search task. Additionally, there have been demonstrations of similar processing
between older adults and younger adults in tasks with search components that
appear to rely on inhibition. Studies investigating guided search, visual marking,
IOR and location negative priming have all demonstrated similar processing
between older adults and younger adults (e.g., Connelly 8 Hasher, 1993; Kramer
& Atchley, 2000; Madden, Gottlob, 8. Allen, 1999; McCrae 8. Abrams, 2001). At
first glance, at least, this may seem surprising in the context of the Hasher and
Zacks’ inhibition deficit view and its supporting evidence suggesting that older

adults are more affected by distracting elements of the environment. However, a

17

critical aspect of these inhibitory phenomena is that there are few age differences
when any of them relies upon the inhibition of a location in the search space
rather than the inhibition of the identity or semantic code of an item (Connelly &
Hasher, 1993; Kramer & Atchley, 2000; Hartley & Kieley, 1995). Apparently the
processes involved in inhibiting spatial locations are preserved throughout the
lifespan. This differs from evidence of decreased inhibitory efficiency when the
identity of the item is inhibited. There have been demonstrations of age
differences in negative priming for the identity of the recently rejected distractor
(e.g., Connelly & Hasher, 1993). Additionally, while it is clear that older and
younger adults do not differ in IOR for spatial locations, there have been some
contradictory findings about age differences in IOR that is tied to an object
independently of the spatial location (McCrae & Abrams, 2001). The contrast
between the identity and the location of a distracting item seems to have a large
impact on age differences in inhibition.

In addition to the differences in the properties (location or identity) that are
inhibited, there are other aspects of the stimulus that can affect the likelihood that
age differences occur. One aspect is the salience of the distractors in
comparison to the target items. As mentioned earlier, Kramer et al. (2001) found
that in an attentional capture task, younger adults were able to inhibit their eye
movements to a salient sudden onset distractor (in this case, an added distractor
that was brighter than the other distractors). However, older adults were less able
to prevent eye movements to this salient item. In order for inhibitory processes to

be engaged, it appears to be important that the distractor be a viable competitor

18

for attention (i.e., salient in the environment). If the distractor items are not
salient, there will be little need to inhibit them and thus no age differences in the
responses. In other words, age differences in these tasks will only be found if the
nontarget information is both “strong and wrong” (Radvansky 8 Curiel, 1998, p.
77). The current experiments vary salience by manipulating the relationship
between different objects and the search target. If an object is related to the
target, then it is expected to be more salient and require more processing than
unrelated distractors.
Visual Memory

Earlier studies of visual memory (e.g. Shepard, 1967; Standing, Conezio,
8 Haber, 1970) found that participants were relatively good at remembering
pictures that they had seen even after several days. Counter to these earlier
findings, however, there has been a recent debate as to whether or not the visual
system maintains information about the visual details of objects in the
environment. There have been several demonstrations of insensitivity to
relatively large-scale changes to the visual environment in online viewing tasks
(e.g., Grimes, 1996; McConkie 8 Currie, 1996; Simons 8 Levin, 1998). Change
Blindness, as this has been labeled, is the demonstration that across a visual
disruption (e.g., an eye movement or a briefly presented blank screen), large
changes in a scene can be made without the viewer detecting the change. In a
classic example (Grimes, 1996), participants viewed pictures while their eye
movements were monitored. Changes were made to the picture while the eyes

were in motion. During a saccadic eye movement, no useful visual information is

19

acquired by the visual system; thus changes to the scene can be made during a
period of functional blindness. In a picture of two wedding attendees wearing
differently colored top hats, the hats were shifted between the attendees during a
saccade. Few viewers noticed the changes that occurred during the viewing. In a
more common technique, two scenes differing in one changed element are
repeatedly alternated (separated by blank screens) until a button is pressed or a
time limit is reached. The measure of change blindness is the number of cycles
that the participant views the scenes prior to detecting the change (if s/he detects
it at all). For many changes, it takes several cycles for the change to be noticed.
Such change blindness results led some theorists to claim that no visual
information is maintained from one fixation to the next (O'Regan, 1992; Rensink,
2000). These theorists argue that if participants had memory of the correct visual
details, they should be perfect in change detection. O’Regan (1992) has claimed
that there is no need for visual memory due to the fact that if one needs to
remember what an object in a scene looks like, one simply has to direct the eyes
to that location and resample the information. Rensink (2000; Rensink, O’Regan,
8 Clark, 2000) has claimed that an object coheres only while it is being currently
attended. Once attention is withdrawn, the object representation disintegrates,
losing all visual detail memory. Both of these claims point to an impoverished
visual memory, contradicting the findings of Shepard (1967) and Standing et al.
(1970)

More recent research has called into question models that claim that there

is no memory for the visual details of objects and scenes. Hollingworth and

20

Henderson (2002) found evidence of visual memory in a scene memory
experiment. Participants were able to accurately identify the scene they had seen
earlier from a foil in which one object had changed (a type, token, or orientation
change). The token and rotation changes are of particular interest because in
order to perform accurately in the memory test, one needs to remember the
visual details of the objects. Additionally, Castelhano and Henderson (2001) have
demonstrated that when searching through pictures of scenes, one can
remember the visual details of distracting objects from the search. This was true
even when one was not expecting a memory test. Both of these studies
demonstrate that the visual details of the objects can be remembered.

There appear to be two ways to reconcile findings of preserved visual
memory with the change blindness phenomenon. One is that in most change
blindness studies, researchers have not made certain that the objects that were
changed were actually processed before the change occurred. It seems
reasonable to assume that change detection would be nearly impossible if one
had not looked at the object before the change occurred. Additionally, the test of
change blindness (overt change detection) may not be sensitive enough to
determine the memory that is maintained by the visual system. Hollingworth, et
al. (2001) found that in a change detection paradigm (a token change
experiment), when the participants were tested with overt change detection, they
performed better than chance but not exceptionally well. However, they found
evidence that when refixating the critical object, participants examined the object

longer if the object had changed than if it was the same. This was true even

21

though the participant did not overtly respond that anything had changed in the
scene. This indicates that the there may be some level at which the visual details
are maintained. In other words, the visual system appears to have more
information in it than is evident from overt responses.

Object Representations and Visual Memory

In the office example mentioned previously, the search for the red book
took place in a space filled with a variety of objects that could draw attention.
However, the concept of the object and its representation was intentionally left
implicit. In this section, one of the prominent theories of object representation is
discussed along with recent work that discusses the memory maintained for
objects after they have been out of sight.

A great deal of research has explored the way that attending an object
affects the mental representations of that object. Kahneman, Treisman, and
Gibbs (1992) laid out a theory of the mental representation of objects. According
to their object file framework, when an object in the environment is attended to, a
file, metaphorically similar to a file folder, is set up that contains information about
the object (e.g., size, color, etc). This file is indexed by the spatial-temporal
coordinates of the object, which are updated if the object moves. If the file is
accessed at a later time, the information that was stored becomes available to
the executive processes and can be used to facilitate the processing of the
object. This framework describes several abilities of the visual system, including

the ability to track objects as they move within the environment.

22

Although the object file framework was developed using a simple visual
environment, it has been extended to describe the processing of objects in
scenes. Hollingworth and Henderson (2002) claim that during the processing of a
scene, the information about an attended object is stored in a relatively long
enduring format that preserves some of the basic visual details of the object (a
long-term object file). This information is detailed enough to support
discrimination of subtle differences in test stimuli (Hollingworth 8 Henderson,
2002; Hollingworth, et al., 2001). In these experiments, changes were made to a
single object in a scene following a fixation on the object. Participants were able
to recognize the change that had occurred at better than chance levels both
during the viewing epoch and after a long delay. Objects were changed by either
rotating the object 90° in depth or by replacing it with another token of the object.
Because neither of these types of changes alters the basic level category of the
object, the representation constructed during the viewing of the scene must be
detailed enough to detect subtle visual differences in a discrimination task.

Castelhano 8 Henderson (2001) provided further evidence of long-term
object files that is directly relevant to the current study. They demonstrated that
participants Showed memory for the visual details of items that were not the
target of a search in a scene. Participants were able to use long-term
representations of the objects to support both token and rotation discriminations.
Because participants were able to remember visual details of objects that were
not the targets, there is a possibility that the variable processing of the different

classes of objects could lead to differences in the memory representations of

23

those objects. Unlike the current experiments, however, the objects tested in
their experiments were not selected so that they had a specific relationship to the
search target. Thus, the distractor objects that were tested may or may not have
been related to the search that took place, leaving open the question of the effect
of the role that the object played during search on memory.

The critical aspect of the object file framework for the current study is that
attention is vital in the formation of the file (T reisman, 1998). As mentioned
earlier, it was expected that during a search, attention is applied unevenly to the
objects in the search space with some objects (e.g. targets) receiving a great
deal of attention and other objects (e.g., background objects and unrelated
distractor objects) not being attended to much at all. Differences in the
application of attention to objects in a search could lead to variability in the long-
term memory representations of these objects. In other words, because attention
is not directed to targets and distractors equally in a search array, object files that
are created for the various objects will differ in the information contained in them.
These differences could lead to large differences in performance on subsequent
memory tests.

Current Study

The current study modified the typical search task in both stimuli and
dependent measures in order to further study the role played by executive control
of visual attention in search. It used real-worid objects as the basis for search,
and each item appeared in only one array within the search phase of the

experiment. These differences from previous procedures allowed the current

24

study to examine the memory that is formed for each object in a search. By using
individual tokens of the objects, the memory preserved for visual details of
individual objects was examined. Differences in memory for the various classes
of objects would provide evidence of differences in the processing that is
performed during a search task. It was expected that target objects would be the
most likely to be processed in the search, given that they were the goal of the
search. The related distractors should be the next most likely to be processed
due to their relation to the search targets. The unrelated distractors may be
processed little in the search task. Of interest are the comparisons of targets and
related distractors and the related distractors and the unrelated distractors. The
first comparison, between targets and related distractors, gives an indication of
the variable processing that occurs for items that were examined and selected
versus those that were rejected. The second comparison, related and unrelated
distractors, examines any differences that might be present when some items
can be easily rejected and others cannot. The processes applied to these
different classes of objects could yield different memory representations. The
visual memory test examined the long-term object file representations that are
formed during a visual search. The memory test was a two-altemative forced
choice token discrimination test. This test is similar to the one used by
Castelhano and Henderson (2001) in which they tested the memory of distractor
items after a search through a scene.

Additionally, Experiment 2 used eye movement measures as another

dependent measure to examine the processing of objects in the visual search

25

task. Eye movement measures (e.g., the likelihood that an object is examined,
the number of times that a particular object is looked at, and the length of time
that an individual looks at a particular object) were markers for the way that
executive control processes are applied to a search. They can be used to
evaluate the operation of executive processes during search by examining
processing that certain objects in the environment receive based on the way that
they are related to the search goal. Finally, the current study used an age group
comparison to further examine the role played by inhibition. The claim that older
adults have decreased inhibitory efficiency should be reflected in the search
pattern. Specifically, age differences in inhibition should be evident in processing

of the distractor objects.

26

EXPERIMENTS 1A AND 13

Experiments 1a and 1b examined the differences in visual memory for
targets and distractors of a visual search task. Participants were given a modified
conjunction search task in which they had to count the number of targets that fit a
target description. The search was conducted through arrays of 12 real-world
objects (see Figure 1). The target of the search was defined by a color-category
combination (e.g., the drill that is green). In addition to targets, there were objects
that were related to the search targets by either being the same general color as
the search target (green things that are not drills) or the same category as the
search target (drills that are not green). In most of the research that has
examined conjunction search, when the distractors share a feature with the
target, the search proceeds more slowly than when the distractors are unrelated
to the target. This inefficiency indicates that distractors that are related to the
search target are being processed to some extent and subsequently rejected.
Both category and color distractors will be examined because if one of these
features is used to segregate the search space (in a guided search way), then
collapsing across the distractor type would problematic. If, for example, color is
used effectively by the participant to divide the search space into objects to look
at and those not to look at, then only the related color distractors would be viable
competitors for the search target. Thus, in this example the related category
objects would effectively be screened out and thus would not be considered.
Essentially, these category distractors would appear as unrelated category/color

distractors to the participant. The separation of these different types of distractors

27

permits me to examine the possibility that different search strategies could affect
the classification of the distractors. Finally, the search arrays contained objects
that were not related to the target (objects that were neither green nor drills).
Because these objects do not possess a feature that defines the target, it is less
likely they will need to be processed during the search.

The first experiment focused on the memory for objects in visual search.
Participants were given an unexpected memory test following a filled interval of
approximately ten minutes. The presumed difference in the amount of processing
of the visual search target, related distractors, and unrelated distractors could
lead to differences in the memory. Any cognitive processes that played a role in
the selection of targets and the rejection of distractors should influence the
memory maintained after search. The processing of targets as the things that
were searched for could lead to relatively good memory for these items.
Additionally, the executive control processes could include inhibitory processes
that limit the processing of the distractors. As mentioned earlier, inhibition could
potentially serve different functions for the different types of distractors. Related
distractors could be inhibited due to the fact that they are highly likely to draw
attention to them and thus need to be inhibited in order for search to proceed.
Unrelated distractors may need to be inhibited to prevent the search from
processing these items because they are not relevant to the search goal. The
potential use of inhibition in the processing of distractors could lead to poorer

visual memory performance.

28

As was mentioned previously, the memory test was a token discrimination
test in which participants judged which of two pictures fitting the same description
was presented earlier during the search task. This test was designed to measure
memory for the visual details of the search objects, rather than their semantic
categories, and thus evaluates claims that the visual details of perceived objects
are not maintained in memory. Also, because the memory test was unexpected,
any learning that took place during the search was completely incidental.
Therefore, it tested the representations that were constructed without the
intention to remember the visual details.

Because I am interested in the memory for the distractors as well as the
targets, it was important to encourage the participant to examine many of the
objects during a particular search. In order to accomplish this, I altered the
traditional search paradigm in another way: multiple targets. In the object arrays
of these experiments there were as few as 0 or as many as 3 instances of the
target object (e.g., 3 distinct green drills). The use of multiple targets encouraged
participants to continue to search through the object array even after they
encountered an instance of the target object, thereby potentially processing other
distractors. Participants responded with the number of targets that appeared in
the array rather than whether or not the target was present or absent.

Finally, Experiment 1b tested the inhibitory components involved in search
by examining the memory that older adults have for the visual details of the
objects in search. Typically, older adults perform more poorly on memory tests

than younger adults. However, because older adults have decreased inhibitory

29

efficiency, they may process the distractors more than the younger adults. This
additional processing may allow older adults to remember the distractors as well
as, if not relatively better than, the younger adults. In other words, the additional
processing may strengthen the memory for the older adults, and thus they would
not demonstrate the typical pattern of age-related declines in memory. However,
this memory benefit of the older adults should be limited to the distracting objects
in the search.

To preview the results of Experiments 1a and 1b, the visual memory for
the targets of visual search was relatively good even after an intervening period
of at least 10 minutes. Additionally, related distractors (color and category
distractors) were remembered better than chance, although they were
recognized at similar levels. Unrelated distractors were recognized at
approximately chance levels in each of the experiments. Older adults in
Experiment 1b demonstrated relatively well-preserved visual memory for the
distractor objects, but not the search targets. These results support the claims
that visual memory exists for objects that are searched and that this appears to
be tied to the probability that the object was examined. Additionally, there is
evidence that inhibition plays a role in search and that older adults do process

the distractors more than younger adults.

30

 

    

;. \ . W23,

.1;

Figure 1. Example of a search display. Images in this study were presented in
color. For photocopying purposes, the background presented in this image is
lighter than was true in the experiment. The search targets (1 green drill in panel
A and 3 green drills in panel B) are highlighted by the white boxes. Other objects
include a color distractors (green watering can, upper right), category distractor
(yellow drill, upper middle), and unrelated distractors (yellow-handled axe, right
center).

31

Method

The methods were almost identical for experiments 1a and 1b and will be
described together. Any differences will be noted.
Participants.

Forty-eight participants were recruited from general psychology courses at
Michigan State University for Experiment 1a. Experiment 1b had 24 younger
adults (recruited from the same source as Experiment 1a) and 24 older adults
recruited from the general community. The older adults were recruited from a
subject pool maintained by Dr. Rose T. Zacks. The demographics for the
participants in both experiments reported here appear in Table 1. The age
groups differed in their education level and vocabulary scores. In the
Momingness/Eveningness Questionnaire (MEQ; Home 8 Ostberg, 1976) scores,
older adults expressed a stronger preference for the morning (higher values
indicate a preference for morning times; a score of 59+ equals a morning
preference) than younger adults who were neutral (scores 42-58) in their time of
day preference. For this reason, an effort was made to test older adults in the
morning. Younger adults were primarily tested in the later morning to afternoon.

These age differences are typical for experiments run on these samples.

32

 

 

41.8 (1.21)

41.3 (1.40).
62.7 (2.57)

42.4 (2.27)

Age Edfiation Vocabulary MEg

Experiment 1a

Younger Adults 18.8 (0.14) 12.5 (0.12) 27.8 (0.53)
Experiment 1b

Younger Adults 19.8 (0.20) 13.5 (0.21 ). 30.2 (066)

Older Adults 73.5 (1.19) 15.5 (0.61) 35.1 (0.88)
Experiment 2

Younger Adults 20.0 (0.58) 13.0 (0.23). 28.5 (0.82).

Older Adults 76.9 (0.93) 14.7 (0.49) 36.5 (0.57)

61.8 (1 .57)‘

 

'age group comparison p < .05

Table 1. Mean (Standard Error) for Age, Education Level, Vocabulary, and
Momingness/Eveningness Questionnaire (MEQ).

Design.

Experiment 1a had 2 main factors: number of targets and object type. In

the target counting task, the participants had to determine the number of targets

that were present in the search array (ranging from 0-3). The other factor was the

relation that the objects in the search array had to the search targets. Each

search array consisted of four types of items: targets, category distractors, color

distractors, and unrelated distractors. The dependent variables of primary

interest are the measures of visual memory performance for the 4 different types

of array objects. However, measures of search efficiency and accuracy were also

examined.

Experiment 1b was identical to Experiment 1a except that in addition to

number of targets and object type manipulations, there was also an age group

comparison.

Materials

Sixty-four search categories were selected (see appendix). The categories

were divided into two groups of 32 categories each. A participant only searched

33

for targets from one of the two groups of categories. Targets were defined by a
specific category—color combination. For each category, three colors were
selected in which the target could appear (e.g., green, yellow and red drills).
There were 8 possible colors that an object could be (black, brown, white, silver,
red, blue, yellow, and green). Each color was used equally often as a target color
(i.e., for four targets) across categories and participants. For each color within a
category, 5 photographic exemplars were selected, primarily from an electronic
database of pictures of real-world objects. Each object was resized so that its
longest dimension (horizontally or vertically) was 35° in visual angle (from a
viewing distance of 57 cm) while maintaining its original proportions.

Search arrays were constructed for each color—category combination (see
Figure 1). A search array consisted of 0-3 targets (e.g. a green drill), 3-4 category
distractors, 3-4 color distractors and 3-4 unrelated distractors.1 The category
distractors were objects from the other two colors of the search category (i.e. red
and yellow drills). The color and unrelated distractors were selected from the
non-presented set of 32 search arrays for that participant. Thus, the color and
unrelated distractors were from categories that were never searched for by a
participant. Color distractors matched the target on the color that was being
searched for and within an array all came from different categories (e.g. a green
watering can, door, earrings, and tractor). Unrelated distractors were not related

to the search target by either search category or search color. Each unrelated

 

1 The number of color, category, and unrelated distractors was dependent on the number of
targets that were present. For search arrays where there were no targets, 4 color, 4 category, and
4 unrelated distractors were presented. For arrays that had one target present, 1 unrelated
distractor was removed. For arrays that had 2 and 3 targets, one category and one color
distractor, respectively, were also removed.

34

distractor was a different color and from a different category than the other
distractor objects. The objects were counterbalanced in that each object served
as a target, a color distractor, a category distractor and an unrelated distractor
across participants.

Visual Search.

An array of 12 objects was created for each of the factorial combinations
of the number of targets (4 levels), the 3 target colors, and 64 categories. Arrays
were 800 x 600 pixels in size (approximately 31° x 23.5°) with a neutral gray
background (RGB = 120). Each search category was assigned to 1 of 8 object
configurations. The twelve objects were placed in the arrays based on 3.5° x 3.5°
squares (This was done so that the object would be contained within the square
regardless of which was the longest axis). The presentation squares were
arranged with the constraint that objects could not be closer than 35° apart from
the nearest edge of each square. The objects were no nearer the edge of the
display than 40 pixels in the horizontal dimension and 30 pixels in the vertical
dimension.

The arrays were presented in two blocks. Each of the 32 arrays appeared
in each block once. The number of targets and the specific configuration of
objects were changed in each block (e.g., if there were 2 targets in the first
viewing, there could be 1 or 3 targets on the second viewing). The changes were
made to require the participant to perform the search on the second viewing of
the array rather than relying on memory for the answer to the number of targets

or the location of the targets/distractors. In order to ensure that a target was

35

processed for a search array, there was always a target present in the second
block. Participants were not told that in the second block all of the arrays had at
least one target.

All three tested distractors were present both times the array was
searched. If there was no target in the first viewing of the array, then the tested
target was only viewed once. Otherwise, the tested target was viewed on both
presentations of the array.

Visual memory test

The explicit memory test was a two-altemative forced-choice task. One
target, one color distractor, one category distractor, and one unrelated distractor
were tested from each search array that a participant viewed. Participants were
asked to determine which of two pictures (the left picture or the right picture; see
Figure 2) was identical to a picture that had been in the earlier search task. The
other picture was a non-presented foil that was selected so that it matched the
condition of the presented item. In other words, if the presented object was a
yellow car, then the foil was also a yellow car that had not been seen by that
participant. The memory foils for half of the participants were the presented
objects for the other half. Objects were presented in the same size (3.5°) as they

were in the search display.

36

 

Figure 2. Example of visual memory test stimuli. Participants had to determine
which green drill token (left or right) had been presented in the earlier search
arrays. Images were presented in color and were the same size as during the
search portion of the experiment. For photocopying purposes, the background
presented in this image is lighter than was true in the experiment.

Participants in Experiment 1a were only tested on objects from 16 of the
32 search categories that they had seen. The objects from the other 16 arrays
were presented in another task, which preceded the explicit memory test.
However, the results from that other task will not be discussed here. The objects
tested were counterbalanced across participants. Participants in Experiment 1b
were tested on objects from all 32 arrays they had seen during the search.
Procedure

Upon arrival, the participant was provided with a brief description of the
experiment and given the informed consent form and a general demographic
questionnaire. After completing these forms, the experimenter positioned the

participant approximately 57 cm from the computer screen and the participant

was presented with the following instructions for the search task:

37

“In this task, you will be given an item to search for in an object array (e.g.,
"the saw that is yellow"). Your job is to count the number of objects in the array
that fit the description given. Input the number of search targets (0, 1, 2, or 3
objects) on the button box in front of you. We want you to be as accurate as
possible in your responding, so take as much time as you need to be certain of
your response.”

The experimenter then gave a sample array to accustom the participant
to the types of judgments that slhe would have to make. After the sample trial,
the experimenter again emphasized that accuracy in the judgment was the
critical component, and answered any questions that arose from the sample trial.
If there were no questions, the experimenter began the experiment. Stimuli were
presented using E prime version 1.0 software.

Following the search task, the participants had a filled interval during
which they completed unrelated tasks (e.g., the Shipley Vocabulary test and the
Momingness/Eveningness questionnaire). After finishing these forms, the visual
memory test was administered. The participants were told that they would see
two pictures and should do their best to decide which one of the pictures was
presented during the visual search task. Additionally, they were told that the test
would be very difficult and to take their best guess if they were uncertain. The
two pictures were presented to the left and the right of the center of the screen.
The participant indicated which of the two pictures was presented earlier by
pressing either the left or right button on the button box. Half of the targets

appeared on the left and half appeared on the right.

38

Results

The results for Experiment 1a and 1b are divided into two sections: visual
search and visual memory test. For all statistical tests, an alpha level of .05 was
used to determine statistical significance.

Visual Search

The two measures of interest for the visual search component are search
time and accuracy in counting the number of targets present. Because each
array was presented twice, the search data are divided into first and second
presentation. However, because there was always a target in the second
presentation, the analyses are restricted to the cases where there was a target in
the array. The results for Experiments 1a and 1b can be found in Table 2.

In Experiment 1a, search times through the search arrays were faster for
the second presentation than the first presentation [F(1, 47) = 112.70, p < .001 ,
MSE = 1284947]. This is not surprising given that people were given no practice
and only one example prior to beginning the first presentation block. By the
second presentation block, the participants had received 32 trials of practice and
thus were more efficient in the task. This could also be influenced by repetition
priming because the pictures had been seen before. There was no main effect of
the number of targets [F(2, 94) = 1.91, p > .10, MSE = 845283]. However, there
was an interaction between the number of targets and the presentation [F(2, 94)
= 3.99, p = .02, MSE = 880041] with longer searches as the number of targets
increased in the first presentation, whereas in the second presentation there was

no difference in the Search time with more targets. As in the search time, the

39

search accuracy also demonstrated improvement between the first and second

presentation [F(1, 47) = 4.41, p = .04, MSE = 0.0073]. There was also a strong

effect of the number of targets on accuracy, with accuracy decreasing as the

number of targets increased in the array [F(2, 94) = 16.14, p < .001, MSE =

0.0178], and there was no interaction between presentation and number of

targets (F < 1).

 

 

 

A
Number of Targets
0 1 2f 3
Experiment 1a
Presentation 1 6043 (251) 5701 (227) 5770 (201) 6269 (343)
Presentation 2 4463 (160) 4611 (176) 4412 (163)
Experiment 1 b
Younger Adults
Presentation 1 6548 (449) 5020 (266) 5335 (331) 5398 (331)
Presentation 2 4377 (256) 4390 (250) 4258 (324)
Older Adults
Presentation 1 9684 (670) 8959 (707) 10408(1 160) 12480(1656)
Presentation 2 8384 (717) 8148 (665) 8800 (698)
B
Number of Targets
0 1 g 3
Experiment 1a
Presentation 1 0.92 (.016) 0.91 (.017) 0.80 (.024) 0.79 (.026)
Presentation 2 0.91 (.013) 0.85 (.018) 0.81 (.019)
Experiment 1b
Younger Adults
Presentation 1 0.94 (.020) 0.95 (.017) 0.85 (.025) 0.81 (.030)
Presentation 2 0.94 (.013) 0.88 (.016) 0.84 (.021)
Older Adults
Presentation 1 0.89 (.027) 0.90 (.017) 0.70 (.038) 0.67 (.040)
Presentation 2 0.87 (.026) 0.73 (.037) 0.75 (.027)

 

Table 2. Mean (Standard Error) for (A) Correct Search Times (in ms) and (B)
Search Accuracies (proportion correct) for Experiment 1a and 1b.

40

The major difference between Experiment 1a and 1b was the testing of
older adults as well as younger adults in Experiment 1b. Overall, older adults
were slower to respond correctly to the search arrays than younger adults [F(1,
46) = 31.97, p < .001, MSE = 50457905]. As in Experiment 1a, correct search
times in Experiment 1b were faster for the second presentation of the search

arrays than the first presentation [F(1, 46) = 23.52, p < .001, MSE = 7264342].

The effect of presentation was modified by an interaction with age group in which

older adults appeared to improve more than younger adults [F(1, 46) = 3.95, p =
.053, MSE = 7264342]. Unlike Experiment 1a, however, there was a main effect
of the number of targets on search times [F(2, 92) = 5.25, p = .007, MSE =
5143581] with increased search times as the number of targets present in the
search array increased. This was also modified by an interaction with age group
[F(2, 92) = 4.30, p = .016, MSE = 5143581], meaning that older adults appeared
to be more affected by the number of targets in the search array than younger
adults. This could indicate greater difficulty for the older adults in the tasks of
updating and maintaining the number of targets that one has encountered. The
target counting task requires that when one encounters another target, the
previous number of targets must be retrieved and then incremented. If either, or
both, of these processes were more difficult for older adults, they would have
more difficulty as the number of targets in the display increased. As in
Experiment 1a, there was an interaction of presentation and target number [F(2,
92) = 5.33, p = .006, MSE = 3666895], indicating that the number of targets had

little effect in the second presentation. Finally, there was a marginal three-way

41

 

interaction of presentation, target, and age group [F(2, 92) = 2.79, p = .067, MSE
= 3666895] with older adults improving more with a greater number of targets.

For search accuracy, younger adults were more accurate than older adults
[F(1, 46) = 25.72, p < .001, MSE = 0.0325]. In contrast to the search times, there
was no main effect of presentation [F(1, 46) = 1.97, p = .17, MSE = 0.0123] or
interaction with age group (F < 1). However, there was an effect of the number of
targets [F(2, 92) = 43.97, p < .001, MSE = 0.0135], with accuracy decreasing as
the number of targets in the search array increased. There was also an
interaction between the number of targets and age group [F(2, 92) = 4.58, p =
.013, MSE = 0.0135] with older adults being more affected by the additional
search targets than younger adults, mirroring the search time results. There was
neither interaction of presentation and number of targets [F(2, 92) = 2.23, p =
.113, MSE = 0.0142] nor a three-way interaction with age group (F < 1).

Visual Memory Test

The more important results for Experiment 1a and 1b are the visual
memory performance results. It is important to note here that because there were
trials in the first presentation block of arrays with no targets present (the 0 target
condition), a subset of the target objects for each participant had only one
opportunity to be viewed in these experiments. All of the analyses reported in this
study, unless otherwise stated, eliminated these singly-viewed target objects
from consideration. This ensured that the target objects and the distractor objects
(which were always present in both presentations of the array regardless of the

number of targets present) had equal opportunities to be viewed within the

42

search portion of the experiment. This amounted to one quarter of the targets
being removed from the analyses for each participant.

Because a two-altemative forced-choice design was used, chance level
for the memory test was 50%. There are two important tests that address
different aspects of this study. The first test is a comparison of the different object
conditions (target, category distractor, color distractor, and unrelated distractor)
to chance level using one-sample t-tests. This test examines the claim that no
visual information is maintained once attention has been withdrawn from an
object. If objects in any of the conditions are recognized at better than chance
levels, then some information about the objects has been maintained after the
visual search task. The second comparison examines the memory for the
different types of objects. Because the objects played different roles in the search
component of the experiment, it is reasonable to assume that the variable
processing that each object received during the search (due to its disparate
relationship to the target) would lead to different memory levels. Additionally, the
executive processes involved in the rejection of the distractors could also lead to
decreased memory performance. It is important to remember that each object in
the memory test served as a target, a color distractor, a category distractor and
an unrelated distractor across participants and that the memory foils for half of
the participants were the presented objects for the other half.

The results of Experiment 1a and 1b are shown in Figure 3. The first
important comparison is to chance levels. As can be seen in the top panel of

Figure 3 (Experiment 1a), when compared to chance level (.50), the target

43

objects [.88; t(47) = 27.1, p < .001], the category distractors [.62; t(47) = 5.39, p <
.001], and the color distractors [.60; t(47) = 6.71, p < .001] were all remembered
better than chance. The unrelated distractors, however, were not remembered
significantly better than chance [.53; t(47) = 1.43, p = .16]. When comparing the
memory results for Experiment 1b, the same pattern can be seen for the younger
adults (bottom panel of Figure 2). Target objects [.86; t(23) = 19.31, p < .001],
category distractors [.61; t(23) = 5.71, p < .001], and color distractors [.60; t(23) =
4.33, p < .001] were all remembered better than chance. The unrelated
distractors were again not remembered significantly better than chance [.52; t(23)
= 1.35, p = .19]. The older adults (also in the bottom panel of Figure 2)
demonstrated a similar pattern with the exception of the unrelated distractors.
Older adults remembered target objects [.80; t(23) = 15.63, p < .001], category
distractors [.60; t(23) = 3.91, p = .001], color distractors [.57; t(23) = 4.45, p <
.001] and unrelated distractors [.56; t(23) = 4.45, p < .001] better than chance
level. The results of Experiment 1a and 1b clearly argue against a visual
perception model that does not maintain at least some visual details from objects
that have been viewed previously.

The second important comparison examined the differences between the
object conditions. In Experiment 1a (top panel of Figure 3) there was a main
effect of object condition [F(3, 141) = 81.73, p < .001, MSE = 0.0140], with

performance declining in the following order". targets, related distractors and

unrelated distractors? The memory for target objects was significantly better than
that of the category distractors [t(47) = 11.89, p < .001] and the color distractors
[t(47) = 13.32, p < .001]. The two related conditions were not significantly
different from each other [t(47) = 0.67, p > .20], and each was significantly better
than the unrelated distractor condition [category distractor, t(47) = 3.53, p = .006;

color distractor, t(47) = 2.85, p = .039]. n

 

 

2 The level of significance for the comparisons of the different conditions was corrected using a

Bonferroni correction for multiple comparisons for each of the experiments described here. The
comparisons of the conditions between the two age groups, however, were not corrected in any
of the experiments.

45

 

1.00 —
0.90 - _I_
0.80 -
0.70 -
0.60 - + —I—
0.50 —
0.40 -

proportion comet

0.30 -
0.20 -
0.10 -

 

 

 

 

 

 

 

0.00 1 . I 1
target category color unrelated

 

 

 

1.00 -
030 . EIYoungerAdullS
IOIder Adults
0.80 -
0.70 -
0.60 -
0.50 -
0.40 -

proportion comet

0.30 -
0.20 -

   

 

 

0.10 T

0.00 I . .
target category color unrelated

 

 

 

 

 

 

 

Figure 3. Mean proportion correct for the visual memory test in (A) Experiment 1a
and (B) Experiment 1b. The error bars are standard errors.

46

The same pattern of data was evident for the younger adults in Experiment 1b
with a main effect of condition [F(3, 69) = 77.56, p < .001, MSE = 0.0067]. Again,
targets were remembered better than the category distractors [t(23) = 14.41, p <
.001] and the color distractors [t(23) = 9.78, p < .001]. As in Experiment 1a,
objects within the two related distractor conditions were remembered at similar
levels [t(23) = 0.27, p > .20], and each was remembered better than the
unrelated distractor condition [category distractor, t(23) = 4.08, p = .003; color
distractor, t(23) = 2.95, p = .043]. The older adults in Experiment 1b, however,
demonstrated a slightly different pattern of visual memory. The older adults did
demonstrate a main effect of condition [F(3, 69) = 31.22, p < .001, MSE =
0.0092]. Similar to the younger adults, older adults remembered targets better
than the related distractors [category distractors, t(23) = 6.77, p < .001; color
distractors, t(23) = 8.47, p < .001]. Additionally, objects in the two related
distractor conditions were remembered at similar levels [t(23) = 0.94, p > .20]. In
contrast to the younger adults, older adults’ memory for the unrelated distractors
was not significantly different from their memory for the related distractors
[category distractor, t(23) = 1.21, p > .20; color distractor, t(23) = 0.31, p > .20].
To examine this difference in visual memory, the two age groups from
Experiment 1b were compared directly. There was an effect of condition [F(3,
138) = 98.09, p < .001, MSE = 0.0079], but there was no overall difference
between the age groups (F < 1). There was, however, a significant interaction
between the two age groups [F(3, 138) = 2.82, p = .041, MSE = 0.0079]. To

explore this interaction, the age groups were compared for each of the object

47

conditions. As expected, younger adults outperformed older adults in their
memory for the visual details of the target objects [t(46) = 2.38, p = .022]. This is
consistent with most comparisons of memory between younger and older adults.
However, there were no significant differences for related distractor conditions
[category distractors, t(46) = 0.18, p > .20, color distractors, t(46) = 0.98, p > .20].
Additionally, older adults marginally outperformed younger adults in the unrelated
distractor condition [t(46) = -1.87, p = .068].

An interesting aspect of the design was that a subset of the tested target
objects were viewed only in the second of the two presentation blocks, whereas
the remainder of the target objects were viewed in both blocks. This was a result
of the fact that during the first presentation block, a quarter of the arrays had no
targets in the array. By comparing the targets that were viewed in both blocks to
those viewed only once, I can provide some indication of the role of additional
processing for visual memory. In Experiment 1a, targets that were viewed on
both trials (.881) were remembered better than targets viewed on only one trial
(.839), but this difference was not significant [F(1, 47) = 2.37, p = .13, MSE =
0.0181]. In Experiment 1b, both age groups had better memory for targets
viewed twice (younger adults: .858; older adults: .796) than for targets that were
viewed only once (younger adults: .775; older adults: .748). In contrast to
Experiment 1a, the difference in the number of viewings was significant [F(1, 46)
= 7.22, p = .01, MSE = 0.0142]. There was neither an overall age difference [F(1,
46) = 2.14, p = .15, MSE = 0.0219] nor an interaction of age group and number of

viewings (F < 1). The lack of an age difference appears to contradict the findings

48

of an age difference in the memory for target objects presented earlier. The
target effect described earlier was an analysis of the twice-viewed targets only.
However, the presentation block analysis added in once-viewed targets and, as
can be seen in the means, the difference between the age groups was smaller in
this condition. As mentioned at the beginning of the results section, the two target
viewing conditions were not equivalent in the number of trials that were present.
The once-viewed targets represent only one quarter of the targets presented (8
targets) and thus reliability of the means in this condition was less robust than
those in the twice-viewed condition (24 targets). In fact, the standard deviations
of the twice-viewed targets (younger adults: 0.086; older adults: 0.093) were
approximately half those in the once-viewed condition (younger adults: 0.180;
older adults: 0.154). However, this analysis weighted the viewing conditions
equally and thus the smaller age difference and increased noise variance from
the once-viewed targets diluted the larger difference in the twice-viewed targets.
The t-test for the twice-viewed targets was presented earlier and was significant,
but when a t-test was computed for the once-viewed targets, the age difference
was not significant [t(46) = 0.56, p > .20]. Thus, it appears that the paradox of the
lack of an age effect in this analysis is an artifact of type of analysis used rather
than a limitation of the experiment.
Discussion

The first experiments demonstrate three important points. The first is that

participants remembered the targets and the related distractors better than

chance. This finding is even more impressive given that the object had been out

49

of sight for at least 10 minutes. The visual memory test used in this experiment
limited the information that could be used to the visual details of the object.
Because the participant was presented with two tokens of the same object, slhe
could not use the knowledge that, for example, a green drill had been presented
to aid in the memory test. The test required that slhe remember which token (i.e.
which green drill) was seen before to respond correctly. The ability of participants
to determine (albeit with some difficulty) the correct token of an object that had
been seen before indicates that the visual details of the object were maintained
after the object was no longer the focus of attention. This finding argues against
the visual theories of O’Regan (1992) and Rensink (2000), which claim that no
visual information is maintained after attention has been withdrawn from an
object. The current finding indicates that at a minimum, some of the visual details
of objects encountered by a viewer were remembered.

The second finding of interest was the difference in the visual memory
rates for the different object conditions. The finding that targets were
remembered better than the distractor objects may not appear to be much of a
surprise. It was, after all, the object for which the searcher was looking. The
finding of reliable memory for objects that were not the target of visual search
was more surprising. Additionally, visual memory appears to be dependent on
the relationship of the object to the search target. Objects that were related to the
targets were remembered better than objects that were not related. This supports
the expectation that memory is tied to the likelihood that an object is processed.

The objects that were related to the search targets were more likely to be

50

examined or examined longer than the unrelated objects because they shared a
feature with the target. This seems to have ensured that some of the visual
details of the related objects were remembered. For the younger adults in both
experiments, the unrelated objects were not remembered better than chance.
Finally, the finding that the target objects viewed on both trials were remembered
better than those viewed on only one trial demonstrates that the more an object
is viewed, the better the memory will be. This is similar to the notion that the
higher the probability that an object is viewed the better the memory will be. This
will be more thoroughly examined in Experiment 2.

The age similarities and differences in Experiment 1b were also
informative. The finding of an age difference in the memory for the target objects
was expected. Older adults tend to not perform as well as younger adults in
memory tests. In contrast to this typical pattern of results, older adults and
younger adults remembered the visual details of the distractor objects equally
well. Because this effect was limited to the distracting objects, it appears that the
older adults may have allocated relatively more of their search time to the
distracting objects than younger adults. Given that older adults had longer overall
search times than younger adults, older adults had more opportunities to view
the distracting objects. This possibility was examined in Experiment 2. The
overall finding of equivalent memory for distracting objects for older and younger
adults supports the claims made earlier about the involvement of inhibitory
processes during search. If no inhibitory processes were involved in the search,

there should be an age difference for all of the different types of objects (with the

51

possible exception of the unrelated objects due to the fact that they were at
chance levels for the younger adults). The fact that older adults remember the
distractors, but not targets, as well as younger adults supports the claim made
earlier that older adults somehow process the distractors more than younger
adults. Decreased inhibitory efficiency would lead older adults, in this case, to be
more likely to process the distractors than younger adults and thus have a
greater chance of remembering them. This increased chance could lead to a
relative improvement in memory for older adults in these specific conditions. It is
possible that this improvement could extend to the processing of unrelated
distractors, which would explain the finding that older adults performed slightly,
but not significantly, better than younger adults on these objects. This fits well
with the claim of Treisman and Sato (1990) that inhibition’s role in visual search
is to limit the distribution of attention to those objects that are potential targets.
Experiments 1a and 1b, however, cannot directly measure the processing
that an object received and therefore used memory results to infer processing
during the search. Experiment 2 more directly measured the processing that
occurred during search by examining the eye movements that were made during
the search task. By recording where a participant looked and what they looked
at, Experiment 2 was able to more precisely determine the processing that the

different objects received.

52

EXPERIMENT 2

The accuracy rate measure used in Experiments 1a and 1b gave an
aggregate measure of the processing involved during search. The second
experiment attempted to obtain a more fine-grained analysis of the search
process while replicating the previous results. Experiment 2 used a similar
procedure to Experiment 1 with the exception that during the search task, the
participants’ eye movements were monitored. By measuring how long, and how
many times each of the objects in the array is examined, a clearer picture of the
operation of executive processes during a search could be obtained.

In conjunction search tasks, like the ones presented here, there are many
potential objects to examine that may or may not share a feature of the target.
The second experiment investigated the relation between viewing behavior and
visual memory. There have been several examinations of eye movements and
visual search. L. G. Williams (1967) found that participants preferentially directed
their eyes to stimuli that shared the target color. Other target characteristics,
such as size and shape, were not as preferentially selected. Zelinsky (1996)
found that participants were more likely to examine distractor objects that shared
a feature with the target than those that did not. However, this preference did not
account for all of the variance in the object selection, indicating that eye
movement guidance during search was not perfect.

It was expected that differences in eye movement behavior would partially
account for memory differences. Eye movement measures examine where and

for how long the eye dwells when looking at the stimuli. In the current

53

experiment, it was expected that objects that were related to the search would be
examined more frequently than those that were not. Targets should be viewed on
every trial because the task is to count these objects. Related distractors should
be the next most likely to be viewed and the unrelated distractors should be the
least likely. The same pattern should follow for the number of separate looks to
the objects and the total amount of viewing time on the object.

In addition to the general search aspects, Experiment 2 also examined
search differences between younger and older adults. There has been little
exploration of the differences in the way that older and younger adults look at the
world in general, and specifically in visual search. Scialfa and Joffe (1997)
observed eye movements during a visual search task and found that older adults
make more saccades than younger adults and have a longer average fixation
duration than younger adults. They claimed that this was consistent with a
restricted field of view and a general slowing pattern for older adults. However,
one critical issue makes the comparison between their study and the current
study difficult. Specifically, their particular eye movement measures are not good
indicators of differential processing of the various types of objects. The number of
saccades made and the average fixation duration reflect the processing of the
entire array and do not provide information about the processing of individual
distractors. This limits the usefulness of these data for comparison to the current
study because in the current study the critical comparison is between different
types of objects. Recently, Mayr and Spieler (2002) investigated search

differences between older and younger adults in easy search tasks and relatively

54

more difficult search tasks. Older adults produced more fixations in the search
regardless of difficulty level. However, there was no difference in the likelihood of
refixating an item. These findings are similar to those of Scialfa and Joffe (1997)
and are also limited by the fact that results are based on an aggregate average
over the entire display.

Kramer, Hahn, lnrvin, and Theeuwes (1999) also investigated age
differences in eye movement behavior. Using the search task described in the
introduction, they investigated the effect of a sudden visual onset on eye
movements. They were interested in whether there would be an age difference in
the likelihood that a sudden onset would be fixated. They found that both age
groups were equally likely to execute an eye movement to the sudden onset if
the onset was equiluminant with the other distractor items. This indicated that
older and younger adults were equally likely to suppress eye movements when
the stimulus was not salient. However, if the added distractor was salient (i.e.
brighter than the other objects), younger adults could suppress eye movements
to the added distractor but older adults could not. The results of Kramer et al.
(1999) are important for the current study. If older adults are able to segregate
the search environment based on a feature (as suggested by Plude and
Doussard-Roosevelt, 1989), then the objects that are segregated may not be
salient and thus there may be no difference between the age groups on the
performance on these items. On the other hand, the items that cannot be
segregated from the target based on a feature must be individually processed

and rejected. This additional processing may make these items salient and thus

55

elicit the age differences described by Kramer et al. (1999). Eye movement
measures are required to determine if the search patterns of the older and
younger adults are similar.

The current experiment was intended to explore the role of visual
processing of search arrays on memory. The purpose of examining the viewing
behavior during search was to obtain some insight into the visual memory from
visual search. It was expected that different viewing patterns could influence
visual memory. The experiment also explored the role of inhibition on search
patterns and on eye movement measures. Given the inhibition deficit theory of
Hasher et al. (1999), older adults should demonstrate greater difficulty in
rejecting distractors than younger adults. This greater difficulty should result in
viewing a higher proportion of distracting objects, viewing distracting objects
more frequently, and viewing them for a longer period of time. These measures
would indicate that older adults have decreased efficiency of inhibitory
processing that affects the visual search process. Additionally, the differences
between the search tasks used in the studies discussed earlier and that used in
this study (real-world objects with many different types of distractors) could lead
to a different pattern of results than have been found previously.

The final point to be examined in this experiment is the effect of multiple
views of objects on eye movement measures and whether or not there are any
age differences in this effect. Althoff and colleagues (Althoff 8 Cohen, 1999;
Ryan, Althoff, Whitlow, 8 Cohen, 2000) have demonstrated that on previously

viewed stimuli, there tend to be fewer fixations and a smaller sampling region.

56

Experiment 2 can examine this effect on eye movement measures, but there are
some limitations. The arrays in the current study were different in the two Views
in spatial configuration and number of targets. Thus, any advantage from the
multiple views may be eliminated by the changes in the stimuli. However, if there
is some memory, there may be differences in the eye movement measures used
here. Additionally, any age group interaction with presentation block would speak
to the role played by the inhibition of previously viewed objects on the allocation
of eye movements.

Method

The method used in Experiment 2 closely followed the method of
Experiment 1 with the exception that in Experiment 2 the participants’ eye
movements were monitored during the search task.

Participants.

Twenty-four younger adults and 24 older adults from the same sources as
in Experiment 1b participated in this experiment (see Table 1 for demographic
information).

Design.

As in Experiment 1b, there were three factors: object type, number of
targets and age group. Eye tracking allowed for additional dependent measures
to be collected during the search component of the experiment, including the
proportion of trials on which an object was examined, number of times that an

object was looked at, and the total amount of time spent examining the object.

57

Materials.

The materials were the same as those in Experiment 1b. However, due to
a change in equipment, the visual angle that the display subtended was 12%
larger in this experiment than in Experiment 1b. The display subtended 35°
horizontally and 26° vertically. The objects along their longest dimension
subtended 39° with a minimum of 39° between the edges of objects.
Equipment

Eye movements were monitored using an ISCAN RK-726PCI pupil
tracking system sampling at 240 Hz. This system operates by reflecting an
infrared light into the right eye of the participant and then measuring the reflection
of the infrared light from the eye using a camera sensitive to this wavelength of
light. The ISCAN system calculates two reflections: one from the pupil and one
from the cornea. The two reflections are used to calculate the center of vision,
which is then output to the display computer (to be exact, the pupil is the point of
highest absorption of infrared light on the eye, thus position of the pupil is the
point of the lowest reflection). Prior to data collection, the output from the ISCAN
system and the display computer were calibrated using a 9-point calibration
screen. The accuracy of the calibration was checked periodically throughout the
experiment. The calibrated position is accurate to less than 1° of visual angle
both horizontally and vertically. Because of this limitation, the scoring region of
each object (the area in which a fixation must fall to be counted as on the object)
was slightly larger than the object itself (0.4° larger than the object). The ISCAN

equipment can handle some minor head movements by the participant and thus

58

no bite bar was required. However, a chin rest was used to limit head
movements to some degree.
Procedure

The procedure was the same as Experiment 1b with the stated exceptions
that were required to institute eye tracking. Following the presentation of the
sample display, the participant was informed that while they were looking for the
objects described, their eye movements would be monitored. The participant was
required to place his/her head in a chin and forehead rest to maintain relative
stability during eye tracking. After the participant was comfortable in the chin rest,
the experimenter adjusted the eye-tracking camera to obtain a trackable image.
The participant was then asked to fixate on each of 9 points on an otherwise
blank screen for calibration. After calibration was completed, it was immediately
checked to determine if calibration was sufficient. If not, the 9-point sequence
was repeated. A trial began with a display of the search target description (e.g.,
“the saw that is yellow”). The participant read the description and looked back to
a central box. When the experimenter determined that the participant was looking
at the central box, the search array was presented. The participant viewed the
array and counted the targets while their eye movements were monitored. After
each trial, the calibration screen was presented, which allowed the experimenter
to determine if the system was still tracking adequately. The memory test

proceeded identically to Experiment 1b.

59

Eye tracking analyses.

The eye tracking data consist of XY coordinates of the calibrated eye
position for each sample of the eye tracking system. The ISCAN system used in
this experiment samples at 240 Hz or roughly one sample every 4-5 ms. To
create a more stable eye tracking record, every sample was averaged with the 2
preceding and the 2 following samples. This was done during analysis after eye
tracking had been completed. Fixations were determined by grouping
consecutive samples that were no more than 8 pixels in Euclidian distance from
the previous sample. Any fixation that was less than 90 ms or greater than 4000
ms was discarded. One trial from one older adult was eliminated in the eye
tracking results due to an experimenter error. All trials from all of the other
participants were analyzed.

All of the analyses were performed on object regions. An object region
was defined as the 3.9° x 3.9° box (+0.4° in each dimension) centered on the
object that had been used to place the object in the array. F ixations were
determined to be on a particular object if it fell within this region. An example of
the scoring region can be seen in Figure 5.

Results
Search Results

As can be seen in Table 3, the overall search times were faster than in the
first experiment. Additionally, the accuracy rates were worse than in the first
experiment. While this could be evidence of a speed-accuracy trade-off, I believe

that this it is more likely that the changes in the procedure required for eye

60

tracking led to these differences. The use of an eye tracker may have made
participants focus more on the task at hand (especially the older adults) and thus
sped up their performance. Additionally, because of the eye tracker, participants
were unable to look at the buttons used to input their responses. This likely
decreased the accuracy of the responses.

The search results were analyzed as in Experiment 1b. Overall, older
adults were slower to respond correctly to the search arrays than younger adults
[F(1, 42) = 12.12, p = .001, MSE = 3421588013. Again, correct search times were
faster for the second presentation of the search arrays than the first presentation
[F(1, 42) = 14.68, p < .001, MSE = 3236603]. This difference was modified by a
marginal interaction with age group [F(1, 42) = 3.06, p = .088, MSE = 3236603],
with older adults appearing to improve more in the second presentation. As in
Experiment 1b, there was an increase in the search time as more targets were
introduced into the array [F(2, 84) = 4.64, p = .012, MSE = 3584358], and there
was an marginal interaction with age group [F(2, 84) = 2.56, p = .083, MSE =
3584358] whereby younger adults’ search in the second viewing was practically
unaffected by the number of targets. Older adults appeared to continue to take
more time as the number of targets in the second presentation increased. There
was neither an interaction of presentation and number of targets [F(2, 84) = 1.36,
p > .20, MSE = 3726271] nor a three-way interaction (F <1).

For search accuracy, younger adults were more accurate than older adults

[F(1, 46) = 63.74, p < .001, MSE = 0.0646]. As in the search times, there was an

 

3 Four older adults did not provide any correct responses in the 3-target condition and thus were
excluded from the search time analysis. As mentioned earlier, the change in procedure may have
led to decreases in accuracy.

61

effect of presentation [F(1, 46) = 5.51, p = .023, MSE = 0.0189], with accuracy
improving in the second presentation. There was, however, no interaction with
age group (F < 1). This indicates that both age groups were able to improve their
performance equivalently. There was an effect of the number of targets [F(2, 92)
= 52.48, p < .001, MSE = 0.0241], with accuracy decreasing as the number of
targets the search array increased. Additionally, there was an interaction
between the number of targets and age group [F(2, 92) = 23.93, p < .001, MSE =
0.0241], with older adults being more affected by the additional search targets
than younger adults. In contrast to Experiment 1b, there was an interaction of the
presentation and the number of targets [F(2, 92) = 5.98, p = .004, MSE =
0.0184], with the accuracy improving most in the 3-target condition across the

two presentations. There was no three-way interaction with age group (F < 1).

62

 

 

 

Number of Targets
0 1 g 3
Younger Adults
Presentation 1 4985 (326) 4463 (238) 5005 (363) 4924 (311)
Presentation 2 4287 (276) 4455 (256) 4261 (317)
Older Adults .
Presentation 1 7656 (720) 6500 (581) 7645 (838) 8848 (1521)
Presentation 2 5938 (475) 6543 (557) 6974 (785)
B
Number of Targets
0 1 2 3
Younger Adults
Presentation 1 0.91 (.026) 0.89 (.023) 0.84 (.029) 0.75 (.025)
Presentation 2 0.87 (.021) 0.80 (.025) 0.87 (.028)
Older Adults
Presentation 1 0.84 (.027) 0.80 (.021) 0.55 (.048) 0.37 (.057)
Presentation 2 0.82 (.030) 0.57 (.041) 0.49 (.047)

 

'Four older adults did not have any correct responses in the 3-target conditions in

the first presentation.

Table 3. Mean (Standard Error) for (A) Correct Search Times (in ms) and (B)
Search Accuracies (proportion correct) for Experiment 2.

Visual Memory Test

Again the more important results are the visual memory performance

results. The results of Experiment 2 are shown in Figure 4. As in Experiment 1b,

the visual memory results for each age group were compared against chance

levels. As can be seen in Figure 4, the memory results of Experiment 2

demonstrated a similar pattern to that seen in Experiment 1b. For younger adults,

target objects [.86; t(23) = 19.67, p < .001], category distractors [.60; t(23) = 6.35,

p < .001], and color distractors [.60; t(23) = 5.25 , p < .001] were all remembered

better than chance. In contrast to Experiments 1a and 1b, unrelated distractors

were remembered significantly better than chance as well [.55; t(23) = 3.36, p =

63

.003]. Older adults demonstrated a similar pattern to Experiment 1b with the
exception of the unrelated distractors. Older adults remembered target objects
[.76; t(23) = 11.11, p < .001], category distractors [.55; t(23) = 2.87, p = .009],
and color distractors [.58; t(23) = 4.21, p < .001] better than chance. However,
older adults did not remember the unrelated distractors [.53; t(23) = 1.61, p = .12]

better than chance level.

 

1.00 7

0.80 _ IOIder Adults
0.70 -
0.60 -

0.50 ~
0.40 -

ropertlon correct

p

.0

(A)

O
l

0.20 -
0.10 -

 

 

 

 

 

0.00

 

target category color unrelated

 

 

 

Figure 4. Mean proportion correct for the visual memory test in Experiment 2.
The error bars are standard errors.

In the comparison of differences between the object conditions for the
younger adults in Experiment 2, there was a main effect of condition [F(3, 69) =
80.43, p < .001, MSE = 0.0057]. Again, targets were remembered better than

category distractors [t(23) = 9.98, p < .001] and color distractors [t(23) = 11.70, p

< .001]. As in Experiment 1a and 1b, objects in the two related conditions were
remembered at similar levels [t(23) = 0.14, p > .20]. However, unlike in
Experiment 1a and 1b, the related distractors were not remembered significantly
better than the unrelated distractors [category distractor, t(23) = 2.28, p = .20;
color distractor, t(23) = 2.02, p = .33]. These comparisons were not significant
with the Bonferroni correction. However, using a less stringent criterion for
comparisons, the unrelated distractor condition was different the category
distractor condition [t(23) = 2.28, p = .032] and marginally different from the color
distractor condition [t(23) = 2.02, p = .055]. This pattern is consistent with
Experiments 1a and 1b.

Older adults, however, demonstrated a slightly different pattern of visual
memory in Experiment 2 than in Experiment 1b. The older adults did
demonstrate a main effect of condition [F(3, 69) = 41.40 p < .001, MSE =
0.0066]. Similar to the younger adults, targets were remembered better than the
related distractors [category distractors, t(23) = 8.11, p < .001; color distractors,
t(23) = 7.79, p < .001], and related distractors were remembered similarly [t(23) =
1.20, p > .20]. Older adults’ memory for the unrelated distractors was not
significantly different from their memory for the related distractors [category
distractor, t(23) = 1.24, p > .20; color distractor, t(23) = 2.06, p > .201‘.

As in Experiment 1b, the two age groups were compared directly. There

was an effect of condition [F(3, 138) = 116.69, p < .001, MSE = 0.0062].

 

‘ For the older adults, the difference in memory between the color distractors and the unrelated
distractors was significant (t(23) = 2.06, p = .051) using less stringent comparisons than the
Bonferroni correction. However, the difference between the category distractors and the unrelated
distractors was not significant even by less stringent measures.

65

However, in contrast to Experiment 1b, there was a difference between the age
groups [F(1, 46) = 6.75, p = .013, MSE = 0.0034]. There was a marginally
significant interaction of age group and object condition, [F(3, 138) = 2.35, p =
.075, MSE = 0.0062]. To explore the interaction, the age groups were compared
for each of the object conditions. As expected, younger adults outperformed
older adults in their memory for the visual details of the target objects [t(46) =
3.07, p = .004]. Additionally, as in Experiment 1b, there were no significant age
differences for the color distractor condition [t(46) = 0.60, p > .20] or the
unrelated distractor condition [t(46) = 0.88, p > .20]. The category distractor did,
however, yield a marginally significant age difference with younger adults
performing better than older adults [t(46) = 1.95, p = .057]. This last difference
contrasts with the results of Experiment 1b.

As in the Experiment 1a and 1b, the targets viewed on only one trial were
compared to targets viewed on both trials. Both age groups had better memory
for targets viewed twice (younger adults: .855; older adults: .764) than for those
viewed once (younger adults: .784; older adults: .696). Consistent with
Experiment 1b, the difference in the number of viewings was significant [F(1, 46)
= 8.92, p = .005, MSE = 0.0130]. However, in contrast to Experiment 1b, there
was the expected age difference in this experiment [F(1, 46) = 8.46, p = .006,
MSE = 0.0115]. Finally, there was no interaction of age group and number of
viewings (F < 1). Again, the number of viewings possible seems to influence

visual memory. This will be further examined using the eye movement data.

66

Eye Movement Results

There are several ways to interpret the eye tracking data. These data will
be discussed in two sections: age comparisons of eye movement behavior during
the search portion of the experiment and the relationship of viewing behavior to
visual memory. Both sections examined the viewing behavior for the objects that
were tested from each array. This means that eye tracking results for one target,
one category distractor, one color distractor, and one unrelated distractor from
each array were examined. However, because the participants did not know
which object was going to be tested (or for that matter that there was a test at all)
while they were viewing the array, the tested objects are equivalent to the other
objects in each condition. The three measures of specific interest were the
proportion of arrays in which an object was viewed (proportion viewed), the
number of separate looks to an object (number of entries), and the total time that
an object was examined during the search. The comparison of visual memory
and viewing behavior was only preformed for the number of entries and the total

time analyses. An example scan pattern can be seen in Figure 5.

67

 

Figure 5. A example of the scan pattern for a younger participant looking for
white telephones. The saccades are represented by the lines and fixations are
represented by the circles. The size of the circle is proportional to the fixation
duration. The three white telephones are highlighted in the figure by the black
boxes, which are the size of the scoring region used. Any fixation that fell within
the region was counted as on the object. The color distractor was the white pipe
(lower left comer), the category distractor was the yellow telephone (middle
right), and the unrelated distractor was the green lady's wallet (left center). For
photocopying purposes, the background presented in this image is lighter than
was true in the experiment.

Proportion Viewed.

The proportion viewed measure determines the proportion of arrays in
which a particular object was looked at in either (or both) of the trials that an
array was seen. “Looked at” in this case means that at least one fixation landed
in the object region. This fixation could have been as short as 90 ms, but as long

as the fixation was contained in the object region, the object was counted as

68

viewed. The tested targets were viewed in almost every array for targets that
appeared in each viewing block. Out of a possible 576 targets that could have
been viewed for each age group, only 4 tested targets were not viewed by
younger participants and only 6 tested targets were not viewed by older adults.
Because almost every target was viewed, causing a ceiling effect, only the
distractor conditions were analyzed in the proportion entered analysis. However,
all conditions are represented in Figure 6. Most of the objects tested in the
memory test phase were viewed during the search phase. There was an effect of
the object condition [F(2, 92) = 54.01, p < .001, MSE = 0.0073], with the two
types of related distractors being viewed equally often [t(47) = 0.78, p > .20], and
both were viewed more frequently than the unrelated distractors [category
distractors, t(47) = 8.36, p < .001; color distractors, t(47) = 8.56, p < .001]. This
pattern is similar to that of the memory results. This similarity supports the
importance of viewing the object for finding memory for the visual details. There
was neither an effect of age group (F < 1) nor an interaction of age group and

object condition [F(2,92) = 1.24, p >20, MSE = .0013).

69

 

1.00 -
Cl Younger adults
0.90 ‘ I Older adults
0.80 -
0.70 ~
0.60 -

0.50 -

proportion entered

0.10 -

 

 

 

 

 

 

0.00
target category color unrelated

 

 

 

Figure 6. Mean proportion of memory-tested objects entered during the visual
search task in Experiment 2. The error bars are standard errors.

Entry Analysis.

The number-of-entries analysis is a more fine-grained analysis than the
proportion entered analysis because a participant could look at a particular object
several times on one trial. Entries can be described as the number of separate
looks to an object. For example, in keeping track of the number of entries to a
green drill, a participant could fixate at the green drill (Entry 1), and then fixate a
yellow drill or other objects, eventually fixating the green drill (Entry 2) once
again. This pattern would yield an entry count of 2 for that trial. If during the
second viewing of the array the green drill were viewed 3 times, the green drill
would have an entry count of 5. If visual memory is dependent on the amount of
processing an object receives during search, then the number of separate looks

to an object could serve as a predictor of the memory that is retained.

70

The different types of objects were entered at different rates [F(3, 138) =
140.94, p < .001, MSE = 0.164]. As can be seen in Figure 7, targets were
entered the most frequently, followed by category distractors [t(47) = 9.97, p <
.001], then color distractors [t(47) = 2.93, p = .034], and finally unrelated
distractors [t(47) = 9.91, p < .001]. There was neither an age difference nor an
interaction with age group (Fs < 1). The lack of an age difference or an age group
interaction replicates the proportion viewed analysis. However, this is a bit of a
surprise given that older adults looked at the search arrays for a longer period of
time than younger adults. The additional time apparently was not taken up by

searching the objects multiple times.

 

3.50 .

3.00 — I:I Younger adults
I Older adults
2.50 —

2.00 -

entrles

1.50 ~

1.00 -

0.50 -

 

 

 

 

 

 

0.00 .
target category color unrelated

 

 

 

Figure 7. Mean number of entries into the object regions of the tested objects
during search in Experiment 2. The error bars are standard errors.

71

Total Time Analysis.

The final eye movement analysis to be discussed here is the total time
analysis. The total time is simply the total number of milliseconds that a
participant viewed an object combined across the two viewings. This is most fine-
grained analysis discussed here as a participant could enter an object
infrequently but still spend a great amount of time examining the object during
those infrequent entries. Only those objects that were viewed at least one time
were included in the total time analyses. The total time could be more informative
than entries or proportion entered as it can be thought of as a direct measure of
processing time.

The total time analysis yielded similar results to the entry analysis with a
main effect of object condition [F(3, 138) = 162.47, p < .001, MSE = 63562; see
Figure 8]. Target objects were looked at longer than category distractors [t(47) =
12.15, p < .001]. Category distractors were in turn viewed longer than color
distractors [t(47) = 3.37, p = .01], followed by unrelated distractors [t(47) = 10.10,
p < .001]. In contrast to the entry analysis, however, there was an age effect, with
older adults looking at objects longer than younger adults [F(1, 46) = 8.82, p =
.005, MSE = 67917]. This is not surprising given that older adults had longer
search times. It appears that the extra search time for the older adults is spent
dwelling on the individual objects rather then more extensively searching the
array. There was also an interaction between the type of object and age group

[F(1, 46) = 5.03, p = .002, MSE = 63561]. The interaction was caused by the fact

72

that older adults spent much more time examining the target objects than the

younger adults in comparison to the other conditions.

 

2000 —
1800 ~
El younger adult
1600 " lolder adult
1400 -
1200 1

1000 ~

total time (ms)

800 -

600 -

400 —

200 -

 

 

0

 

 

target category color unrelated

 

 

Figure 8. Mean total fixation time within the object regions of the tested objects
during search in Experiment 2. The error bars are standard errors.

Viewing Trial and Target Count Analyses

Because the arrays were presented in two blocks, differences in the two
viewings for each of the eye movement measures could be examined.
Additionally, the effect of the number of targets in the array for each trial (target

count) was examined. For all of the analyses in this section, four factors were

73

analyzed: object condition, age group, viewing trial, and target count of the
array.5

Viewing trial

For the proportion entered and the number of entries measures, the only
significant effect was a main effect of viewing trial in the number of entries
analysis [F(1, 46) = 5.70, p = .021, MSE = .179]. In presentation block 1, the
tested objects were entered on average 1.01 times per object whereas in
presentation block 2, they were entered on average 0.95 times per object. No
other significant effects or interactions with viewing trial were found (all Fs <
1.78). This indicates that, although the same proportion of objects were entered,
participants returned to previously viewed objects (on a particular trial) less
frequently in the second presentation block.

As in the number of entries, there was a main effect of viewing trial on the
total time measure [F(1, 46) = 19.21, p < .001, MSE = 62744] with objects being
looked at for 494 ms on average in the first presentation block and 429 ms in the
second block. However, in contrast to the number of entries analysis, there was
an interaction between the presentation block and the object condition [F(3, 138)

= 9.46, p < .001, MSE = 34804]. This appeared to be the result of the greater

 

5These analyses were performed in a different manner than the other eye movement analyses.
Rather than combining the two viewing trials, these analyses examined the viewing behavior for
each trial separately. Only those objects (targets and distractors) that came from arrays with at
least one target present on both viewing trials were included, eliminating objects from 8 of the 32
arrays viewed for each participant. The differences in the data used here make direct
comparisons to previous analyses, especially in the proportion entered analyses, difficult. Any
missing cell in the analysis for the total time measure was filled with the participant's condition
average. All of these missing cells (5 for younger adults and 1 for older adults) were in the
unrelated distractor condition.

74

drop in total time for objects in the target condition than objects in the distractor
conditions. No other interaction was significant (all Fs < 1.26).

Target Count
There were differences in the proportion entered [F(2, 92) = 26.56, p <

.001, MSE = 0.0242] and the number of entries [F(2, 92) = 9.75, p < .001, MSE =
0.113] as the number of targets increased in an array. The average proportion
entered for all objects fell as the number of targets increased (1 target = 0.73; 2
targets = 0.70; 3 targets = 0.65) as did the number of entries (1 target = 1.04
entries; 2 targets = 0.99 entries; 3 targets = 0.93 entries). There was also an
interaction between object condition and target count [proportion entered, F(6,
276) = 3.00, p = .007, MSE = 0.0232; number of entries, F(6, 276) = 4.43, p <
.001, MSE = 0.0841]. The interaction was the result of both measures decreasing
for the distractor conditions only. The target condition was examined on an equal
proportion of trials and with an equal number of entries regardless of number of
targets present. A possible reason for this interaction is that participants were
spending more time viewing the targets when there were more targets present in
order to keep track of them. There was a numerical (though not significant)
increase in the total time spent on target objects as the number of targets
increased (1 target = 768 ms; 2 targets = 787 ms; 3 targets = 802 ms). Though
this difference is not significant, it may be an indicator of the reason for the
interaction in the other measures. No interactions were significant for the total

time measure (all Fs < 1.27).

75

Regression analyses

The next section examines the relationship of the different eye movement
measures to the visual memory test accuracy. It was expected that the more an
object was processed, the greater the memory test accuracy would be.
Regression analyses were computed for both the number of entries and the total
time measures.
Entries

The analysis regressed the number of entries onto accuracy from the
memory test for each object condition, resulting in a point-biserial correlation. As
was done in Hollingworth and Henderson (2002), a procedure was used for the
regression analyses that accounted for the fact that there were 128 observations
from each of the 24 different subjects in each age group. This procedure involves
first removing the variance attributable to the individual participant’s means
(through the use of dummy variables for each participant) and then determining if
there was an effect of the predictor variable (entries) on the dependent variable
of interest (memory test accuracy). In other words, the regression analyses used
here were hierarchical regression analyses that first removed the variance
attributable to the participant’s overall mean, and then examined the effect of the
predictor variable. The data are presented in Figure 9. For the figure, the entries
were divided into quartile bins for each object condition for each participant. The
mean accuracy for each quartile is presented in the figure.

The regressions were performed separately for each condition. The

statistics provided here are the F values for the change in the R2 of the

76

regression model above the individual participant variance. For younger adults
(top panel of Figure 9), there was a significant relationship between the number
of entries and memory test accuracy for the category distractors [rpb = .16,
Panama, 743) = 18.39, p < .001], and color distractors [rpb = .10, Fmangeu, 743)
= 7.55, p = .006]. There was not a significant relationship between memory test
accuracy and the entries in the target condition [rpb = .04, Fmangen, 551) = 10.2,
p > .20] or unrelated distractor condition [rpb = .05, Fchangen, 743) = 1.51, p >
.20]. For older adults (bottom panel of Figure 9), there was a significant
relationship between entries and accuracy for the category distractors [rpb = .09,
Fmangeu, 742) = 6.47, p = .01]. Unlike younger adults, however, older adults did
not demonstrate a relationship between accuracy and entries in the color
distractor condition (rpb = -.02, Fchange < 1). Older adults also did not demonstrate
a relationship between the number entries in the target condition [er = .07,
Fmangeu, 551) = 2.50, p = .12]6 or the unrelated condition and accuracy [rpb =

.06, Penance“, 742) = 2.49, p = .11].

 

° As In most of the analyses described in this study, the target condition regressions reported
above were performed on the data from targets that could be viewed twice. However, it is
possible to examine the relationship of memory and the entries for all targets. If number of entries
is related to visual memory, than the number of opportunities that one has to view the target is not
as critical as the number of separate looks to the object. Therefore, an additional regression
between the number of entries and visual memory was computed with all targets included. For
both age groups there was a significant correlation of memory and entries [younger adults, r9], =
.08, meu, 743) = 4.69, p = .03; older adults, rpb = .08, chmu, 742) = 4.71, p = .03].

77

 

1.0

.0
(D
1

.0
on
1

Memory Test Accuracy
9 o
a1 «1

 

. + Targets

0.5 - --0— Category Distractors
C + Color Distractors

I —v— Unrelated Distractors

 

 

 

 

 

 

 

 

0.4 I L ‘ l ' I ‘ I 1 L I A ‘ A 1 I I ' l l I l 1 l ' I
0 1 2 3 4 5 6
Entries
1 .0 F
: + Targets

- —0— Category Distractors
" + Color Distractors
—v— Unrelated Distractors

9
ID
1

 

 

 

 

I

Memory Test Accuracy
9 o
O) ‘1

0.5 I

 

 

 

)-
0 4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
. I I I I 1 1

Entries

Figure 9. Memory test accuracy as a function of entries. The data points are the
average of each quartile of entries within each condition. The panel A contains
the data for younger adults and panel B contains the data for older adults.

78

One interesting aspect of the regression analyses is that it appears that
the target objects were remembered better that the distractors regardless of the
number of separate looks to the objects. This is evident from Figure 9, which
shows that even in the quartile with the most entries for each of the distractor
conditions (all of which are have more entries than the than the smallest number
of entries quartile for the target objects), memory performance does not reach
the level of the worst quartile for the target objects.

Total Time

The same type of regression analysis related the total time and memory
test accuracy within each object condition. As‘in the entry analysis, younger
adults (top panel of Figure 10) demonstrated a significant relationship between
total time and memory test accuracy for category distractors [rpb = .19, Fmangeu,
656) = 25.50, p < .001], and color distractors [rpb = .13, Fmangefl, 656) = 11.58, p
= .001]. The relationship between visual memory and total time spent on targets
was marginally significant [rpb = .08, Fmangem, 548) = 3.45, p = .06]. Replicating
the entry analysis, there was no significant relationship between total time spent
on an object in the unrelated distractor condition and memory test accuracy (rpb =
.03, Fmange < 1). For older adults (bottom panel of Figure 10) there was a
significant relationship between total time and accuracy for category distractors
[rmJ = .12, Fmangeu, 631) = 8.71, p = .003]. Total time spent viewing color
distractors demonstrated a marginal relationship with accuracy [rm> = .07,

Fchangeh, 644) = 3.53, p = .061]. Older adults did not demonstrate a relationship

79

between total time and accuracy for targets [rpb = .07, Fmangeu, 545) = 2.42, p =
.12]7 or unrelated distractors (r...D = .04, Penna, < 1).

The total time regression analysis yielded a similar pattern to the number
of entries analysis, with target objects being remembered better than distractor
objects regardless of the total time spent examining the objects. It is apparent
from Figure 10 that for both age groups target objects were remembered better
than distractors at all levels of total time and that memory performance in the
distractor conditions does not reach the level of the lowest quartile for the target

objects.

 

7 Again, an additional regression was computed for target objects. In this regression of total time
and visual memory accuracy, all targets viewed at least once were included. Again, both age
groups demonstrated a significant relationship between total time spent viewing the target objects
and memory test accuracy in this analysis [younger adults, rm, = .10, FWWU, 734) = 7.74, p =
.006; older adults, rpb = .08, me.(1, 717) = 4.05, p = .045]

80

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 /\
i
0194
a?
g 0.8 1
< b
g; OJ'4b
'—
E;
g .
c) Oti~.
2
Z + Target
0.5 - -0- Category Distractors
- + Color Distractors
: —v— Unrelated Distractors
0.4”1414]1111111111111L'1111'1111' ..
0 500 1000 1500 2000 2500 3000 3500
1 0 Total Time
0.9 «I
>~
o .
g 0.8 a
< I
3 0.7 4
'- C
E: .
E
d) 0.6 a
:5
_ + Targets
0.5 — —O- Category Distractors
- + Color Distractors
—v— Unrelated Distractors
0.41.1.11E11‘1.1.'.114T44,,I,,,,',1l.

 

 

0 500 1000 1500 2000 2500 3000 3500
Total Time

Figure 10. Memory test accuracy as a function of total time. The data points are
the average of each quartile of total time within each condition. The panel A
contains the data for younger adults and panel B contains the data for older
adufis.

81

Discussion

The primary goal of the second experiment was to examine the visual
memory from search as a function of the viewing behavior that occurred during
search. Visual memory should be tied to viewing behavior, which in turn should
be tied to the application of executive control processes. Certain objects could be
viewed more frequently or for more time than others based on their relationship
to the search target. Differences in the viewing behavior of the two age groups
could indicate the contribution of inhibition. The data from this experiment clearly
support the role of viewing behavior in visual memory for different objects, but
evidence for the contribution of inhibition is less clear.

The visual memory results of Experiment 2 were similar to those of
Experiment 1b. As in Experiment 1b, younger adults remembered target objects
better than related distractors, which were remembered better than unrelated
distractors. Older adults again showed better memory for target objects than
distractors, and the distractors were remembered at similar rates, though the
category distractors were remembered slightly better than the unrelated
distractors. The differences in the memory pattern were evident in the
comparisons to chance because younger adults remembered unrelated
distractors better than chance, whereas older adults did not. This is the opposite
pattern to that seen in Experiment 1b. This different pattern of results in
comparison to chance seems to indicate a small memory effect for the unrelated
distractors in both age groups; however, it may be easy to miss. The replication

of the finding that older adults could remember distractors as well as younger

82

adults (with the possible exception of category distractors in this experiment)
supports the claims made in Experiment 1b. The advantage for older adults in
memory for distractor objects indicates that only these objects were differentially
available to them in comparison to the younger adults.

The replication of the memory pattern from Experiment 1b was important
as it supports the claims of Experiment 1a and 1b that visual memory can exist
even after a period of time in which the object is out of sight. The finding that
visual memory is detailed enough to support token discrimination supports the
claims of Hollingworth and Henderson (2002) that there are long-term object files
that contain some visual information about objects that were processed
previously.

Turning now to the critical eye movement data, Experiment 2
demonstrated that the eye movement data generally parallel the memory data.
Overall, most of the tested objects were viewed at least once during the search
task. Even in the unrelated condition, objects were viewed at least once in
approximately 70% of the arrays. This seems to indicate that simply having
viewed the object (viewed here means that at least one fixation fell in the object
region during the search) is not sufficient for remembering the object. Othenrvise,
the participants would have been considerably more accurate overall on the
memory test. However, from another perspective, the proportion-entered analysis
showed a pattern that mirrored the visual memory results, with the target being
viewed in almost every array and showing the highest memory rates. Related

distractors were viewed in fewer arrays than the target, but category and color

83

distractors were viewed on an equal proportion of the arrays. This similarity in the
proportion viewed mirrors the similarity in the visual memory performance.
Unrelated distractors were viewed on the smallest proportion of trials and had the
worst memory performance for the younger adults. Older adults also showed
similar patterns between the proportion entered and the visual memory results
with the exception that unrelated distractors were not necessarily remembered
more poorly than related distractors.

The number of entries analysis and the total time analysis showed a
pattern similar to the memory results, and they each demonstrated a relationship
to accuracy on the visual memory test. These two eye movement measures are
related in that an increase in entries for an object increases the likelihood that the
object will be viewed for a longer period of time, but, as was mentioned earlier,
they are not necessarily perfectly correlated. In the end, the two measures
yielded similar patterns of results. Target objects were entered the most
frequently and were looked at for the longest amount of time, and unrelated
distractors were viewed the least frequently and for the shortest time. Category
distractors received the second most frequent entries and the second longest
fixation time. However, in contrast to the proportion analysis and the visual
memory results, color distractors were entered significantly fewer times than
category distractors.

The difference between the two types of related distractors in the entry
and total time analyses indicates that while the objects were viewed on an equal

proportion of trials, they were not viewed with equal frequency or for an equal

84

length of time. One potential reason for this difference is that it may be easier to
determine the category membership of an object in this experiment than the
predominant color of the object. Because the stimuli were photographs of real-
world objects, lighting differences and other factors made the hues of a particular
color look different. Take, for example, the green drill. The green used could vary
from a yellowish-lime green to a more bluish-green while still maintaining the
property of green. The decision about the color of an object may be relatively
more difficult than the identification of the photographs as members of a
semantic category. This is a possibility because the objects were chosen so that
their semantic labels would be easily identified. Thus, even when a participant
landed on a color distractor, it would be much easier to determine that it was not
a member of the target category than to label its exact color. If this were the
case, it would explain the findings of increased looks to and time spent on the
category distractors. Additionally, real-world objects are rarely just one color. A
yellow car, for example, could contain several other colors. The tires, the grille,
and the bumpers could all be (and most likely were) a color other than yellow.
However, because the prominent color of the object was yellow, it was still
referred to in this experiment as a yellow car. It seems reasonable to suggest
that if the prominent color of some objects was more difficult to determine, then a
participant would require more time or more looks to the object to accurately
judge the color. This would lead participants to look at the category distractors
less than the color distractors. Thus, given the pictures used here, it appears that

it is easier to say that a drill is a drill than to determine the color of the drill.

85

The presentation block analyses provided evidence that some aspects of
viewing behavior were affected by repeated viewings of the same objects. This
was surprising given the number of changes between the two arrays and
supports some of the claims of Althoff and colleagues (Althoff and Cohen, 1999).
The lack of an age interaction with the viewing factor indicates that older adults
were as capable as younger adults of using memory to aid in the deployment of
the eyes to the various objects in an array.

The regression analyses for the entry and total time data indicate that for
the related distractor conditions, the more an object is viewed, the better visual
memory tends to be. The target objects and unrelated distractors did not
demonstrate this relationship. Although the results were mixed, there does
appear to be some relationship between the viewing behavior and memory test
accuracy. By demonstrating this relationship, this study provides evidence of the
role of viewing behavior in the formation of memory. The lack of a relationship in
the unrelated distractor conditions is likely the result of the fact that memory is
barely better than chance in that condition. The lack of systematicity in the
memory data in this condition would make trying to find a relationship with
another variable difficult. A restriction of range does not appear to be an issue in
the target condition which makes the lack of a correlation puzzling. Another issue
is that the significant correlations were relatively small and thus accounted for
only small portions of the variance in memory performance. However, it is
important to remember that the search task was separated from the memory test

by a filled interval of at least 10 minutes, which could have increased the noise in

86

the memory test performance. In general, the regression analyses support the
notion that visual memory is at least partially dependent on how the object was
viewed during the search task.

Although the regression analyses demonstrate some relationship between
viewing behavior and the visual memory results, target objects were remembered
better than distractor objects regardless of the viewing that a distractor object
received. Target objects are looked at more frequently and for a longer time, but
by examining Figures 9 and 10, it is apparent that memory for target objects is
privileged beyond what would be expected based on viewing time. The privileged
status of target memory indicates that there is something more to the role played
by these objects in search. One possible explanation is that because the target is
the goal of search, its representation in memory is tagged to be remembered.
Another possibility is that rather than a benefit for target objects, distractor
objects are impaired because they are not the goal of the search. In other words,
distractor objects (regardless of relationship to the target) are tagged as such
and forgotten due to either decay or active suppression. Unfortunately, it is
impossible for this experiment to separate these two explanations for the finding
of better overall memory for target objects. Regardless, it is important that target
objects are remembered better overall because it demonstrates an additional
aspect to memory performance that is not dependent on viewing behavior.

The lack of an age difference in the proportion entered and the number of
entries analyses indicates that older adults were not searching through more

objects than younger adults. Additionally, the total time analysis demonstrated an

87

interaction in which older adults were processing the distractor objects as much
as younger adults. The interaction of age group and object condition was
primarily caused by a large age difference in the target condition and no age
difference in the category distractor condition. A test of the application of
inhibition in this study was a difference between younger and older adults on
distractor objects. It was critical for the inhibition argument that the older adults
process the distractor objects more than younger adults. However, the results of
the eye movement analyses did not demonstrate any age interactions that would
result from the application of inhibition. The two age groups were equally likely to
enter distractor objects and older adults did not spend differentially more time
examining the distractor objects. There are two possible reasons for this finding:
1) the search task used here does not require the application of inhibition, or 2)
the type of inhibition used in the search task is age invariant. Either of these is a
possibility given the results. The search task itself may not require the participant
to inhibit the distractor objects because they were not sufficient competitors for
search. Thus, no age differences would be expected on eye movement
measures. The age differences in memory would result from other (potentially
inhibitory) mechanisms that operate outside of the search itself. The other
possibility is that the type of inhibition that is applied during search is age-
invariant. This would place the current task in a class of supposedly inhibitory
paradigms such as visual marking, IOR, and spatial negative priming, that all

demonstrate age-invariance in the application of inhibition. Unfortunately, the

88

current experiment cannot differentiate between these two possibilities. Further
research will be required to distinguish between these two explanations.
GENERAL DISCUSSION

The current study had three main goals. The first was to examine the
visual memory maintained from visual search. The second was to examine the
role of executive control processes in search and the subsequent effect on
memory for objects that were not the target of search. The final goal was to
explore potential age differences in the search of and subsequent memory for
targets and distractors of a visual search.

With respect to the first goal, the current project found that people could
remember the visual details of objects that had been both the targets and
distractors in a visual search task. The issue of what remains after attention has
been withdrawn has led some researchers (O'Regan, 1992; Rensink, 2000) to
rather extreme views of the paucity of visual memory, with some researchers
claiming that there is nothing at all maintained by the visual memory system. The
current study found that these claims greatly underestimate the amount of
information that is maintained. Each of the experiments in the study
demonstrated above-chance memory for the targets of search and for distractors
that were related to the search targets replicating findings of preserved memory
of the visual details from previous studies (Castelhano 8 Henderson, 2001;
Hollingworth 8 Henderson, 2002). These studies had found relatively long term
memory for the details of objects after instructions to either memorize the scene

or search the scene. The current study was a stringent test of the maintenance of

89

visual details in memory. Firstly, participants did not expect the memory test,
meaning that they had no reason to attempt to memorize the objects in the array.
Additionally, the objects were learned in an array that did not form a coherent
scene, limiting the role of organization strategies in memory. Finally, the visual
memory test used in this experiment required that participants decide which of
two tokens of an object was presented. The test was devoid of context from the
search phase (see Figure 2), and thus the decision could be based on only the
visual details attached to the object. Given the demanding nature of the memory
test, the finding of preserved visual memory in these experiments is even more
impressive.

The second goal was to examine the role played by executive control
processes in search, as evidenced by the effect of an object’s relationship to the
search target on the memory maintained after viewing the array. The
experiments presented here clearly demonstrate that the memory that is
maintained is influenced by the role played by the object in the search. Objects
that were the targets of search were remembered relatively well. Target objects
were entered the most frequently and were viewed for the longest time. Of
greater interest was the memory for distractor objects. The finding that related
distractors were remembered better than unrelated distractors (at least by
younger adults) indicates that executive processes are involved in the selection
of objects to be examined. The eye movement data support this claim because
related distractors (both category and color distractors) were viewed more

frequently than unrelated distractors. Thus, the processes involved are able to

90

select objects related to the target and can limit the processing of unrelated
distractors. The processes were not perfect in their ability to prevent entries to
the unrelated distractors, as evidenced by the fact that the approximately 70% of
the tested unrelated distractors were entered on at least one viewing of the
arrays. This proportion was significantly less than distractors that were related to
the search target, but it still demonstrates that executive processes could not
perfectly guide the search. The inability of search processes to ignore the
unrelated distractors indicates that in this experiment, the inhibitory processes
described by Treisman and Sato (1990) could not completely eliminate
consideration of the objects that did not share a feature with the target. This
could be the result of the differences in the use of real-world objects, which vary
in many aspects, in this experiment as opposed to traditional laboratory stimuli,
which tend to be rather homogenous. However, given that the number of entries
and the total time were significantly less than for the related distractors, it
appears that the executive processes involved in search can limit the processing
of objects unrelated to the search once they have been so identified.

These experiments demonstrate the importance of the selective
processing that occurs in visual search not only for the efficiency of the search
itself, but also for the memory that remains afterwards. The memory patterns
generally mirrored viewing behavior during the search. The more an object is
examined in the search, the better memory tends to be as evidenced by the
regression analyses for the related distractors. The more frequently an object is

examined or the more time spent viewing an object, the more accurate the visual

91

memory tends to be. Unfortunately, this was not tme for targets or the unrelated
distractors. The finding that targets were recalled better than would be expected
from viewing behavior indicates that while there is some relationship between
viewing behavior and memory, other processes appear to play a role as well.
Further research will be needed to explore the contribution of nonsearch aspects
to the visual memory of targets.

The third goal of this project was to examine differences between younger
and older adults in their memory of objects encountered during search and their
viewing behavior during search. The hope was to examine the role played by
inhibition in the memory that was built during the search task. With regard to
memory, older and younger adults were relatively similar in their memory for
objects encountered during search. The only condition that consistently yielded
an age difference was the target condition, with younger adults outperforming
older adults. This is the typical finding in experiments that examine age
differences in memory. In contrast, younger and older adults performed equally
well in their memory for distractor objects (with the exception of a marginal
difference in category distractors in Experiment 2). This demonstration of
equivalent memory for distractor objects supports a claim that older adults
process these objects more than younger adults, which in turn could be evidence
of inhibition operating during search. Because this was limited to the conditions
that the searcher was trying to avoid, the age equivalence could be the result of
older adults processing the distractor objects to a greater extent due to an

inability to effectively filter them out. However, in the eye movement analyses,

92

there were few age differences, and those present did not support the additional
processing argument. Older adults entered the distractor objects as frequently as
younger adults and, although they looked at the distractor objects longer than the
younger adults, this was not specific to the distracting conditions. The lack of an
age effect in the distracting conditions indicates that differences in the viewing
behavior for distracting objects were not the cause of the age equivalence in
memory for these items. In fact, older adults spent significantly more time
processing target objects and still demonstrated comparably worse memory. It is
therefore reasonable to assume that the age difference that allowed older adults
to remember the distracting objects as well as younger adults can be found in
processes that continue after the search has been completed.

With respect to the inhibition deficit claim of Hasher and Zacks (1988), this
experiment was unable to find evidence of age-related inhibition differences in
the search through an array of real-world objects. It appears that inhibition could
be operating in restricting the objects that are examined. The lack of an age
difference in the search processes seems to indicate that, at the least, this type
of inhibition does not diminish as people age. However, the finding that older
adults were able to remember the distracting objects as well as younger adults
does suggest that older adults are less able than younger adults to effectively
inhibit distractors after the search has been completed.

Given the lack of an age difference in the eye movement measures during
search, it appears likely that the age difference that allowed the distracting

objects specifically to be remembered as well for older as for younger adults

93

occurs following the search. This argument is similar to that of Hartman and
Hasher (1991 ). In their experiment, participants were required to generate '
endings to sentences. They were then given an ending to remember that either
matched or did not match the ending that was generally produced. Of interest
was the access that individuals had to the ending that they had generated after
being told that they had to remember another ending (called the disconfirmed
ending). The authors found that older adults had better access to the
disconfirmed endings they had produced (using an implicit memory test)
compared to younger adults. They claim that older adults were not as efficient at
suppressing these generated endings, and thus the disconfirmed endings could
more readily accessed (see also May, Zacks, Hasher, 8 Multhaup, 1999).

In the current study, the distractor objects could be more readily available
after the search for older adults. In other words, once the long-term object file of
a distractor is constructed during the search, it is possible that inhibition operates
on it to prevent its reactivation because the object has been identified as a
distractor. Older adults are not as effective as younger adults at suppressing
these long-term object files allowing those objects to be accessible. This is
analogous to the argument Hartman and Hasher (1991) made because the older
adults have more ready access after the search has been completed to objects
that were previously rejected.

Summary and Conclusion
Overall, the current study demonstrates that even when people are not

intending to remember the visual details of objects during search, they

94

demonstrate relatively good visual memory. People are also able to remember
the details of objects that they were not intending to find in a search. These
findings argue against any model of visual perception that does not maintain
some visual details after an object has been viewed. The total time examined
and number of viewings also appear to contribute to the visual memory for at
least some of the objects. It was further evident that the role of an object in the
search was critical for the finding of memory, with memory being found primarily
for objects that were related to the search target. There was evidence of age
differences in visual memory, but this does not appear to be the result of age
differences during the search itself. Thus, there was mixed evidence for age-
related inhibition deficits in the task. Returning to the office search example
presented in the introduction, it appears that people can remember a red coffee
mug or a blue book when they look for a red book. However, people tend to not

remember the yellow pencil on the desk.

95

APPENDIX

 

 

Target Category Color 1 Color 2 Color 3

Set 1
Apple Red Green Yellow
Axe Handle Red Brown Yellow
Backpack Blue Yellow Red
Basket Brown White Red
Binoculars Black Green Silver
Bird Yellow White Green
Boat White Blue Brown
Briefcase Black Brown Silver
Bucket Brown Red Silver
Camera Silver Yellow Black
Car Yellow Blue White
Cat Black Brown White
Chair Blue Yellow Green
Cow Brown White Black
Dogs White Black Brown
Door Green Blue Brown
Earrings Green Silver White
Garbage Can Blue White Green
Guitar Brown Red Black
Hand Gun Silver Black Brown
Lamp Blue Silver Black
Lock Yellow Green Silver
Microscope White Black Red
Pan Silver Black Green
Pen Yellow Red White
Telephone White Yellow Blue
Pot Silver Red Yellow
Suitcase Red Brown Blue
Telescope Black Silver Blue
Tractor Green Blue Red
Truck Red Green Yellow
Watering Can Green Silver Blue

Set 2
Beetle Green Brown Red
Bicycle Red Green Yellow
Boots Brown Black Yellow
Drill Green Yellow Red
Fish Silver Blue Brown
Glasses Blue Black Brown

96

Hairdryer
Hammer

Hat
Iron

Knife Handle

Leaf
Lighter

Microphone
Motorcycle

Mug
Pipe
Pitcher

Airplane
Pliers Handle

Purse

Scissors Handle

Silver
Silver
Yellow
White
Silver
Yellow
Green
Black
Red
Blue
White
Yellow
White
Red
Brown
Black

Screwdriver Handle Blue

Shoes
Stapler
Teapot

Toolbox
Train Car
Typewriter
Umbrella
Lady’s Wallet
Wheelchair

Yellow
Red
White
Black
Brown
Green
Blue
Brown
Black

97

Blue
Red
Green
Silver
Brown
Red
Brown
Silver
Yellow
Yellow
Brown
White
Silver
Blue
Green
Red
Black
White
Silver
Yellow
Red
Green
White
White
Black
Blue

White
Black
White
Blue
Black
Green
Silver
Blue
Blue
White
Green
Silver
Green
Silver
White
Blue
Red
Silver
Black
Red
Yellow
Black
Brown
Yellow
Green
Brown

REFERENCES

Althoff, R. R. 8 Cohen, N. J. (1999). Eye-movement-based memory effect: A
reprocessing effect in face perception. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 25, 997-1010.

Banich, M. T. (2004). Cognitive neuroscience and neuropsychology (2nd ed.).
Boston, MA: Houghton Mifflin Company.

Castelhano, M. S., 8 Henderson, J. M. (2001). Detailed visual memory in scenes
following intentional and incidental learning tasks. Paper presented at the
42"Cl annual meeting of the Psychonomic Society, Orlando, FA.

Connelly, S. L., 8 Hasher, L. (1993). Aging and the inhibition of spatial location.
Journal of Experimental Psychology: Human Perception and Performance,
19, 1238-1250.

Deubel, H. 8 Schneider, W. X. (1996). Saccade target selection and object
recognition: Evidence for a common attentional mechanism. Vision
Research, 36, 1827-1837.

Friedman, A. (1979). Framing pictures: The role of knowledge in automatized
encoding and memory for gist. Journal of Experimental Psychology:
General, 108, 316-355.

Gazzaniga, M. S., Ivry, R. B., 8 Mangun, G. R. (1998). Cognitive neuroscience:
The biology of the mind. New York: W. W. Norton 8 Company.

Grimes, J. (1996). On the failure to detect changes in scenes across saccades.
In K. Akins (Ed.), Perception: Vancouver studies in cognitive science (Vol.
5. pp. 89-110). Oxford, England: Oxford University Press.

Hartley, A. A., 8 Kieley, J. M. (1995). Adult age differences in the inhibition of
return of visual attention. Psychology and Aging, 10, 670-683.

Hartman, M., 8 Hasher, L. (1991). Aging and suppression: Memory for previously
relevant information. Psychology and Aging, 6, 587-594.

Hasher, L. 8 Zacks, R. T. (1988). Working memory, comprehension, and aging:
A review and a new view. In G. H. Bower (Ed.), The psychology of
Ieaming and motivation (Vol. 22. pp. 193-225). San Diego, CA: Academic
Press.

98

Hasher, L., Zacks, R. T., 8 May, C. P. (1999). Inhibitory control, circadian
arousal, and age. In Gopher, D., 8 Koriat, A. (Eds), Attention and _
performance XVII: Cognitive regulation of performance: Interaction of
theory and application. (pp. 653-675). Cambridge, MA: MIT Press.

Henderson, J. M., Pollatsek, A., 8 Rayner, K. (1989). Covert visual attention and
extrafoveal information use during object identification. Perception 8
Psychophysics, 45, 196-208.

Hoffman, J. E., 8 Subramaniam, B. (1995). The role of visual attention in
saccadic eye movements. Perception and Psychophysics, 57, 787-795.

Hollingworth, A., 8 Henderson, J. M. (2002). Accurate visual memory for
previously attended objects in natural scenes. Journal of Experimental
Psychology: Human Perception and Performance, 28, 113-136.

Hollingworth, A., Williams, C. C., 8 Henderson, J. M. (2001). To see and
remember: Visually specific information is retained in memory from
previously attended objects in natural scenes. Psychonomic Bulletin 8
Review, 8, 761-768.

Home, J. A., 8 Ostberg, O. (1976). A self-assessment questionnaire to determine
momingness-eveningness in human circadian rhythms. International
Joumal of Chronobiology, 4, 97-110.

Irwin, D. E., Zacks, J. L. 8 Brown, J. S. (1990). Visual memory and the
perception of a stable visual environment. Perception 8 Psychophysics,
47, 35-46.

Irwin, D. E., 8 Zelinsky, G. J. (2002). Eye movements and scene perception:
Memory for things observed. Perception and Psychophysics, 64, 882-895.

Kahneman, D., Treisman, A., 8 Gibbs, B. J. (1992). The reviewing of object files:
Object-specific integration of information. Cognitive Psychology, 24, 175-
21 9.

Koshino, H. (2001). Activation and inhibition of stimulus features in conjunction
search. Psychonomic Bulletin and Review, 8, 294-300.

Kramer, A. F., 8 Atchley, P. (2000). Age-related effects in the marking of old
objects in visual search. Psychology and Aging, 15, 286-296.

Kramer, A. F., Hahn, S., Irwin, D. E., 8 Theeuwes, J. (1999). Attentional capture

and aging: Implications for visual search performance and oculomotor
control. Psychology and Aging, 14, 135-154.

99

Kramer, A. F., Scialfa, C. T., Peterson, M. S., 8 Irwin, D. E. (2001). Attentional
capture, attentional control, and aging. In Folk, C. 8 Gibson, B. (Eds.)
Advances in Psychology, Vol. 133: Attraction, distraction, and action:
Multiple perspectives on attentional capture. Amsterdam: Elsevier
Sciences.

Madden, D. J., Gottlob, L. R., 8 Allen, P. A. (1999). Adult age differences in
visual search accuracy: Attentional guidance and target detectability.
Psychology and Aging, 14, 683-694

May, C. P., Kane, M. J., 8 Hasher, L. (1995). Determinants of negative priming.
Psychological Bulletin, 118, 35-54.

May, C. P., Zacks, R. T., Hasher, L., 8 Multhaup, K. S. (1999). Inhibition in the
processing of garden-path sentences. Psychology and Aging, 14, 304-
313.

Mayr, U., 8 Spieler, D. H., (2002). Eye movements during visual search. Poster
presented at the biannual meeting of the Cognitive Aging Conference,
Atlanta, GA.

McCrae, C. 8., 8 Abrams, R. A. (2001). Age-related differences in object- and
location-based inhibition of return of attention. Psychology and Aging, 16,
437-449.

McConkie, G. W., 8 Currie, C. B. (1996) Visual stability across saccades while
viewing complex pictures. Journal of Experimental Psychology: Human
Perception and Performance, 22, 563-581.

Neill, W. T., Valdes, L. A., Terry, K. M., 8 Gorfein, D. S. (1992). Persistence of
negative priming: ll. Evidence for episodic trace retrieval. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 18, 993-
1000.

O’Regan, J. K. (1992). Solving the “real” mysteries of visual perception: The
world as an outside memory. Canadian Journal of Psychology. 46, 461-
488.

Plude, D. J., 8 Doussard-Roosevelt, J. A. (1989). Aging, selective attention, and
feature integration. Psychology and Aging, 4, 98-105.

Posner, M. l., 8 Cohen, Y. (1984). Components of visual orienting. In H. Bouma

8 D. G. Bouwhuis (Eds.) Attention and Performance X: Control of
Language processes. (pp. 531-554). Hillsdale, NJ: Erlbaum.

100

Radvansky, G. A., 8 Curiel, J. M. (1998). Narrative comprehension and aging:
The fate of completed goal information. Psychology and Aging, 13, 69-79.

Rayner, K. (1998). Eye movements in reading and information processing: 20
years of research. Psychological Bulletin, 124, 372-422.

Rayner, K., McConkie, G. W., 8 Ehrlich, S. (1978). Eye movements and
integrating information across fixations. Journal of Experimental
Psychology: Human Perception and Performance, 4, 529-544.

Rensink, R. A. (2000). The dynamic representation of scenes. Visual Cognition, ’-
7, 17-42. '

Rensink, R. A., O'Regan, J. K., 8 Clark, J. J. (2000). On the failure to detect
changes in scenes across brief interruptions. Visual Cognition, 7, 127-145.

Ryan, J. D., Althoff, R. R., Whitlow, S. 8 Cohen, N. J. (2000). Amnesia is a deficit I

 

in relational memory. Psychological Science, 11 , 454-461.

Scialfa, C. T., 8 Joffe, K. M. (1997). Age differences in feature and conjunction
search: Implications for theories of visual search and generalized slowing.
Aging, Neuropsychology, and Cognition, 4, 227-246.

Shepard, R. N. (1967). Recognition memory for words, sentences, and pictures.
Journal of Verbal Learning and Verbal Behavior, 6, 156-163.

Shih, S. l., 8 Sperling, G. (1996) Is there feature-based attentional selection in
visual search? Journal of Experimental Psychology: Human Perception
and Performance, 22, 758-779.

Simons,-DanieI-J; Levin,-Daniel-T. (1998). Failure to detect changes to people
during a real-world interaction. Psychonomic Bulletin and Review, 5, 644-
649.

Standing, L., Conezio, J, 8 Haber, R. N. (1970). Perception and memory for
pictures: Single-trial Ieaming of 2500 visual stimuli. Psychonomic Science,
19, 73-74.

Theeuwes, J., Kramer, A. F., Hahn, S., 8 Irwin, D. E. (1998). Our eyes do not
always go where we want them to go: Capture of the eyes by new objects.
Psychological Science, 9, 379-385.

Tipper, S. P. (1985). The negative priming effect: Inhibitory priming by ignored

objects. Quarterly Journal of Experimental Psychology: Human
Experimental Psychology. 37A, 571 -590.

101

Tipper, S. P., Brehaut, J. C., 8 Driver, J. (1990). Selection of moving and static
objects for the control of spatially directed action. Journal of Experimental
Psychology: Human Perception and Performance, 16, 492-504.

Tipper, S. R, Weaver, 3., Jerreat, L. M., 8 Burak, A. L. (1994). Object-based and
environment-based inhibition of return of visual attention. Journal of
Experimental Psychology: Human Perception and Performance, 20, 478-
499.

Treisman, A. (1988). Features and objects: The fourteenth Bartlett memorial
lecture. Quarterly Journal of Experimental Psychology, 40A, 201 -237.

Treisman, A. (1998). The perception of features and objects. In Wright, R. D.
(Ed.) Visual attention (pp. 26-54). New York: Oxford University Press.

Treisman, A. M., 8 Gelade, G. (1980). A feature-integration theory of attention.
Cognitive Psychology, 12, 97-136.

Treisman, A., 8 Sato, S. (1990). Conjunction search revisited. Journal of
Experimental Psychology: Human Perception and Performance, 16, 459-
478.

Watson, D. G., 8 Humphreys, G. W. (1998). Visual marking of moving objects: A
role for top-down feature-based inhibition in selection. Journal of
Experimental Psychology: Human Perception and Performance, 24, 946-
962.

Watson, D. G., 8 Humphreys, G. W. (1997). Visual marking: Prioritizing selection
for new objects by top-down attentional inhibition of old objects.
Psychological Review, 104, 90-122.

Williams, L. G. (1967). The effects of target specification on objects fixated during
visual search. Acta Psychologica, 27, 355-360.

Wolfe, J. M. (1994). Guided Search 2.0: A revised model of visual search.
Psychonomic Bulletin and Review, 1, 202-238.

Wolfe, J. M. (1998). Visual Search. In Pashler, H. (Ed.), Attention (pp. 13-73).
Hove, East Sussex, UK: Psychology Press.

Wolfe, J. M., Cave, K. R., 8 Franzel, S. L. (1989). Guided Search: An alternative

to the feature integration model of visual search. Joumal of Experimental
Psychology: Human Perception and Performance, 15, 419-433.

102

Wolfe, J. M., Oliva, A., Horowitz, T. S., Butcher, S. J., 8 Bompas, A. (2002).
Segmentation of objects from backgrounds in visual search tasks. Vision
Research, 42, 2985-3004.

Zacks, J. L., 8 Zacks, R. T. (1993). Visual search times assessed without
reaction times: A new method and an application to aging. Journal of
Experimental Psychology: Human Perception and Performance, 19, 798-
81 3.

Zelinsky, G. J. (1996). Using eye saccades to assess the selectivity of search
movements. Vision Research, 36, 2177-2187.

103

 

‘51

_n1(

 

     

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

Illlll III III III I

31293 02461 5019

in