IKL

K
a

I:

an xii—DE. t

.. 331:... Z, . 1‘

_ 1%....2?
.33,w.}.1¢.,,
.....ur.., 2:”

1&0.

I‘: ‘k .

‘35.: it.

304‘» clxu In.
fir fili‘v

HE.»

«R.
“flue.-

(Ill:
.. .r VI!

in!

a“ .
. : n.
(A)!

53?.

\l.

.fr n.

 

LCD L _LIBRARY

“big“! State
niversity

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/DmeDuepes-sz

 

AGE DEFICIT S IN THE CONTROL OF PREPOTENT RESPONSES: INHIBITORY
DECLINE OR GOAL NEGLECT

By

Karin M. Butler

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Psychology

2001

ABSTRACT

AGE DEFICIT S IN THE CONTROL OF PREPOTENT RESPONSES: INHIBITORY
DECLINE OR GOAL NEGLECT

By

Karin M. Butler

Older adults have difficulty on behavioral control tasks that require the
suppression of an incorrect response. Two models have been proposed to account for this
finding. The inhibitory deficit hypothesis of aging posits that this impairment is related to
age-related decline in the ability to restrain an incorrect response from being executed.
The goal neglect hypothesis of aging suggests that older adults are as good as young
adults at restraint in these situations, but older adults have difficulty maintaining the
current task goal. When goal maintenance is not optimal a competing task may gain
access to response processes, slowing correct task performance or causing response
errors.

In Experiment 1, the likelihood of goal neglect was manipulated by requiring the
maintenance of the current goal over different task intervals in two paradigms, a spatial
Stroop task and an antisaccade task. Although the response-to-stimulus interval
manipulation led to a pattern of data consistent with goal neglect on the spatial Stroop
task, it did not do so on the antisaccade task. In addition, the goal neglect hypothesis of
aging was not supported because the effects of increasing the likelihood of goal neglect
were equivalent for young and older adults. In Experiment 2, a different manipulation of
the likelihood of goal neglect was tried and, in addition, the difficulty of the inhibitory

task was varied. Again, there was no evidence in support of goal neglect occurring in

either young or older adults. Increasing the difficulty of the inhibitory task disrupted the
performance of older adults more than the performance of younger adults supporting the

inhibitory deficit hypothesis of aging.

Com/right by
KARIN MARIE BUTLER
2001

ACKNOWLEDGMENTS

Afier beating the previously undefeated Greco-Roman wrestler, “Superman”, in
the 2000 Olympics, Rulon Gardner told reporters his strategy for achieving this win was
learned working on his parents’ farm as a kid. When you grow up on a farm you learn
that when you have a hard job to do you put your head down and keep going. My own
parents, Ed and Christine Yager, taught me this lesson as well and I thank them for the
knowledge that I can do anything I put my mind to.

As much as these words served as my mantra during the completion of this
dissertation, there were many people who helped me along the way and without whom
this dissertation might not have been completed. I thank you all.

Rose Zacks was unwavering in her support and skillfiilly guided me through the
process. The members of my dissertation committee, John Henderson, T om Carr, and
Erik Altmann, were always available and enthusiastic to discuss my ideas and offer
constructive suggestions. John Henderson also provided the equipment and space used
for collecting the data.

My husband, Christopher, encouraged me in this pursuit and endured two months
of separation and single-parenthood to make it a reality. My children, Connor and Elora,
were a source of great joy. Beth and Chuck Butler moved across the country to help take
care of my children as I completed this dissertation, giving me the peace of mind that
they were being well cared for in my absence.

Kate Arrington was a generous friend, helping at every step along the way:
listening, offering good advice, providing me with a home, smiling, babysitting, and

bringing me chocolate.

My office and lab mates, Carrick Williams and Doug Davidson, listened to my
ideas, encouraged me, and made me laugh. The research assistants in the Cognitive
Aging Lab, Katie Paquette, Kristina Pecora, and Sarah Burnett, helped enter data and
scheduled participants. Katie Paquette endured the tedium of watching many hours of
eyeballs to help with data collection. Kate An‘ington, Kristina Pecora and Kiel
Christianson recruited older adults to participate.

Several individuals provided technical support. Gary Schrock, the programming
god, was always available to answer my programming questions. David McFarlane
assisted me in the early stages of pilot work. Shari Stockrneyer did typing and copying
and, not least of all, cared.

This research was supported by National Institute of Aging grant AGO 4306 to
Lynn Hasher and Rose T. Zacks and National Science Foundation grant SBR 9617274 to

John Henderson.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... viii
LIST OF FIGURES .............................................................................. ix
1 INTRODUCTION ............................................................................ 1
1.1 Inhibitory Deficit Hypothesis .......................................................... 2
1.2 Goal Neglect Hypothesis .............................................................. 6
1.2.1 Evidence for a goal neglect account of aging ............................... 13
1.2.2 Ideas related to goal neglect .................................................. 16

1.3 Summary ................................................................................ 18
EXPERIMENT 1 ............................................................................ 19
2.1 Method ................................................................................. 21
2.1.1 Participants .................................................................... 21

2.1.2 Design ......................................................................... 22

2.1.3 Apparatus & Stimuli ......................................................... 23
2.1.4 Materials ...................................................................... 25

2. 1 .5 Procedure ..................................................................... 25

2.1.6 Data Analysis ................................................................. 30

2.2 Results .................................................................................. 33
2.2.1 Spatial Stroop Task .......................................................... 33

2.2.2 Antisaccade Task ............................................................ 36

2.2.3 Cumulative Response Distributions ....................................... 42
2.2.4 Relationship between Measures of Interference ......................... 48

2.3 Discussion ............................................................................. 49
EXPERIMENT 2 ............................................................................. 56
3.1 Task Switching ........................................................................ 56
3.2 Inhibitory Difficulty .................................................................. 58
3.3 Predictions .............................................................................. 60
3.4 Method ................................................................................. 62
3 .4. 1 Participants ................................................................... 62

3.4.2 Design ......................................................................... 63

3.4.3 Apparatus & Stimuli ......................................................... 64
3.4.4 Materials ...................................................................... 66

3.4.5 Procedure ..................................................................... 66

3.4.6 Data Analysis ................................................................. 69

3.5 Results .................................................................................. 70
3.5.1 Saccade Direction Accuracy ................................................ 70

3.5.2 Saccade Latencies ............................................................ 72

3.2.3 Residual Switch Costs ...................................................... 75

3.2.4 Reflexive saccades (correct prosaccades and incorrect antisaccades) 76

vii

3.2.5 Target Identification Responses ........................................... 77

3.6 Discussion .............................................................................. 79

4 GENERALDISCUSSION.................................................................87
4.1 Summary of Experimental Results ................................................. 87
4.2 Implications for Theories of Cognitive Aging .................................... 89
4.2.1 Inhibitory Deficit Hypothesis .............................................. 89

4.2.2 Goal Neglect .................................................................. 92

4.3 Implications for Theories of Cognitive Control ................................... 96
4.4 Conclusions ............................................................................ 98
References ....................................................................................... 100

viii

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

LIST OF TABLES

Experiment 1 Participant Characteristics ................................................. 22
Mean reaction times (msec) for correct trials in the Stroop task .............. 34
Experiment 1 Target Identification Responses .................................... 41
Experiment 2 Participant Characteristics . ............................................... 63
Experiment 2 Target Identification Responses ......................................... 79

ix

LIST OF FIGURES

Figure 1. Illustration of a prosaccade trial. In the antisaccade condition, the
double-headed arrow would have replaced the peripheral onset and the single-

headed arrow would have appeared in the opposite box. In the actual displays

stimuli were white presented on a black background ........................................ 29

Figure 2. Mean saccade direction accuracy rate of the initial saccade in the

antisaccade and prosaccade conditions by RSI, short or long, and age group,

young or old.‘Error bars are equal to 1 standard error. Anti = Antisaccade; Pro

= Prosaccade ....................................................................................... 37

Figure 3. Mean latency of the initial eye movement in the correct direction

from the antisaccade and prosaccade conditions and for incorrect eye

movements in the antisaccade condition by RSI, short or long, and age group,

young or old. The mean saccade latency of incorrect antisaccades for young

adults in the short RSI condition excluded 2 young subjects that had fewer

than 7 saccade direction errors on which to base a subject mean. Error bars are

equal to 1 standard error. Anti = Antisaccade, Pro = Prosaccade, Anti_Er =

Incorrect eye movements in the antisaccade condition ....................... _. .............. 38

Figures 4a and 4b. Estimates of the cumulative response distributions reported

by De Jong et al. (1999) for young adults in short (Figure 4a) and long RSI

(Figure 4b) conditions of a spatial Stroop task. Incong. = incongruent, Cong. =
congruent .......................................................................................... 44

Figures Sa-d. Cumulative response distributions for the spatial Stroop task in
Experiment 1 for young adults in the short RSI (Figure 5a) and long RSI

(Figure 5b) conditions and older adults in the short RSI (Figure 5c) and long

RSI (Figure 5d) conditions. Incong. = incongruent, Cong. = congruent .................. 45

Figures 6a-d. Cumulative response distributions for the antisaccade task in

Experiment 1 for young adults in the short RSI (Figure 6a) and long RSI

(Figure 6b) conditions and older adults in the short RSI (Figure 6c) and long

RSI (Figure 6d) conditions. Anti = antisaccade, Pro = prosaccade ........................ 47

Figure 7. Illustration of a prosaccade trial for the central cue condition of

Experiment 2. In the antisaccade condition, the task instruction was

“OPPOSITE” and the target appeared in the box opposite from the cue. In the

peripheral cue condition, a double-headed arrow was presented at fixation

when the peripheral onset occurred. In the actual displays stimuli were white

presented on a black background ............................................................... 68

Figure 8. Mean saccade direction accuracy rate of the initial saccade in the

antisaccade (gray bars) and prosaccade (white bars) conditions by block type,

single task in solids or mixed tasks in stripes, age group, young or older, and

cue type, peripheral or central. Error bars are equal to 1 standard error .................. 71

Figure 9. Mean latency of the initial eye movement in the correct direction

from the antisaccade and prosaccade conditions by block type, single or mixed

task, age group, young or older, and cue type, peripheral or central. Error bars

are equal to 1 standard error. Anti = Antisaccade, Pro = Prosaccade, Sngl. =

Single task, Mix = Mixed tasks ................................................................. 73

Figure 10. Mean latencies of the initial eye movement in the correct direction

for the prosaccade condition and for incorrect eye movements in the

antisaccade condition by cue, peripheral and central, block type, pure and

mixed tasks, and age group, young and old. Error bars are equal to 1 standard

error. Pro = Prosaccade, Anti_Er = Incorrect eye movements in the

antisaccade condition, Sngl = Single task blocks, Mix = mixed task blocks ............. 77

xi

1 Introduction

Older adults have difficulty directing their behavior when the required response is
in competition with a prepotent or highly practiced response. In an eye movement control
task that requires the suppression of a reflexive response toward an onset stimulus in
order to move the eyes in the opposite direction, older adults are more likely to
incorrectly look toward the onset stimulus than young adults are (Butler, Zacks, &
Henderson, 1999; Olincy, Ross, Youngd, & Freedman, 1997). Older adults also have
more difficulty withholding a response in a stop-signal paradigm (Kramer, Humphrey,
Larish, & Logan, 1994; May & Hasher, 1998; Williams, Ponesse, Schachar, & Logan,
1999). In addition, in a Stroop task older adults have more difficulty than young adults
naming the color a word is presented in when the word itself names a color (e. g., West,
1999; but see Verhaeghen & De Meersman, 1998).

Each of these tasks requires that a prepotent or automatic response be prevented.
The difficulty that older adults have in these task situations may arise from a decline in
the ability to suppress these inappropriate responses. The inhibitory deficit hypothesis
posits that an age-related decline in the ability to suppress prepotent responses, as well as
no longer relevant or distracting information, accounts for age-related changes in
performance on several cognitive tasks (Hasher & Zacks, 1988; Zacks & Hasher, 1994;
Hasher, Zacks, & May, 1999). Although the inhibitory deficit account has received
extensive support, an alternative theory has suggested that interference effects are due to
a failure to access the current goal at the moment of stimulus presentation. In this view, it
is not an impaired inhibitory mechanism that is the cause of interference, but rather the

inhibitory mechanism is not engaged at the proper time because the goal of the task is

momentarily inaccessible. This goal neglect account claims that the inability to access the
current task requirements increases the time needed to produce the required response and,
in the case where the goal is not accessed at all, may result in the execution of a more
practiced or prepotent response (De Jong, Berendsen, & Cools, 1999; Duncan, 1995).
This research attempted to determine the cause of the decreased ability of older
adults to stop prepotent responses. Older adults may show impairments on tasks requiring
suppression of a response because an inhibitory mechanism is impaired. Alternatively,
the source of their deficit may be linked to a failure to maintain or access current task
goals when required, resulting in slowed correct responding or the execution of a default
(prepotent) response. Of course, these two ideas are not mutually exclusive. It may be
that older adults' difficulty on tasks that require suppressing an incorrect response results

from a combination of impairment in both of these processes.
1.1 Inhibitory Deficit Hypothesis

A strength of the inhibitory deficit hypothesis is the breadth of the model. Not
only does it explain deficits on tasks requiring behavioral inhibition, it also accounts for
problems experienced by older adults on tasks requiring attentional selection. Hasher and
lacks (1988; Zacks & Hasher, 1994; Hasher et al., 1999) proposed that inhibitory
processes have three functions: (1) to restrict access to working memory by keeping
irrelevant information fiom entering, (2) to delete information from working memory that
has become irrelevant to the current task, and (3) to restrain responding when a prepotent
response is incorrect. Declines in the effectiveness of inhibition account for aging-related
changes in performance on not only motor control tasks, but also on language

comprehension and memory tasks. On a reading with distraction task, age-related

increases in the interference caused by distracting material embedded in a passage
support the claim that older adults have more difficulty preventing irrelevant information
from accessing working memory than young adults (Connelly, Hasher, & Zacks, 1991).
In addition, when told to explicitly forget information, as in a directed forgetting
paradigm, or when it is clear that a previous interpretation was incorrect, older adults are
more likely to retain this irrelevant information than young adults; they have more
difficulty deleting the information from working memory (Zacks, Radvansky, & Hasher,
1996; Hartman & Hasher, 1994; Hasher & Zacks, 1988; Zacks, Hasher, Doren, Hamm, &
Attig, 1987). Support for an impairment in the restraining function of inhibition, as stated
above, includes findings that older adults have a harder time stopping a response they
have already begun to prepare (Kramer et al., 1994; May & Hasher, 1998) and preventing
the execution of a reflexive response (Butler et al., 1999) than young adults. Older adults
show more disruption on tasks that seem to require inhibition than young adults do.

The performance deficits displayed by older adults on an antisaccade task seem to
provide critical support for the restraint impairment proposed by the inhibitory deficit
model of aging. In an antisaccade task, the viewer is asked to look in the opposite
direction of a peripheral onset stimulus. The antisaccade procedure elicits two competing
response programs: an involuntary eye movement toward the peripheral stimulus and a
voluntary response in the opposite direction. The participant must prevent the automatic
response while programming and executing the voluntary eye movement. This task is
difficult, as indexed by incorrect saccades toward the peripheral onset (Hallet, 1978;
Hallet & Adams, 1980; see Everling & Fischer, 1998, for a review). Older adults have

more difficulty performing antisaccades than young adults; They are more likely to

incorrectly look toward a peripheral onset than young adults in an antisaccade task
(Butler etal., 1999; Olincy et al., 1997; Nieuwenhuis, Ridderinkhof, de Jong, Kok, & van
der Molen, 2000; Klein, Fischer, Hartnegg, Heiss, & Roth,2000) and they need to take
more time to correctly produce antisaccades than young adults when compared to a
control condition (Munoz, Broughton, Goldring, & Armstrong, 1998; Nieuwenhuis et al.,
2000; Fischer, Biscaldi, & Gezeck, 1997).

Saccade direction errors in the antisaccade task have come to be interpreted as the
failure of an inhibitory mechanism. This conclusion follows fi'om evidence suggesting
that on each trial a reflexive and voluntary saccade are being programmed simultaneously
and that errors are the result of the reflexive response program gaining access to response
production. This explanation is supported by error latencies and recovery from errors.
Incorrect responses toward the peripheral onset are as fast, or faster, than saccades toward
the onset in a control condition, indicating that participants are automatically
programming them. When an incorrect response is made, individuals usually will correct
it and look to the correct location. The time needed to initiate the saccade to the correct
location is much shorter than the time needed to program an intentional saccade (e. g.,
Roberts, Hager, & Heron, 1994; Nieuwenhuis et al., 2000) indicating that the correct
response was being programmed prior to the execution of the incorrect saccade, but the
incorrect response gained access to the system and was executed before the completion of
the voluntary saccade program.

Although the inhibitory deficit hypothesis has received support from several
studies using different paradigms (for reviews see Hasher et al., 1999; Stoltzfus, Hasher,

& Zacks, 1996, Zacks & Hasher, 1994), the model has been criticized because it is

lacking in specificity and because attempts to identify a measure of inhibitory fiinctioning
have been unsuccessful (e. g., Burke, 1997; McDowd, 1997; but see Zacks & Hasher,
1997). In addition, the inhibitory deficit hypothesis relies on a model of working memory
that includes an inhibitory mechanism (or mechanisms) that work to control the contents
of working memory. Some researchers have proposed that an inhibitory mechanism is
not necessary to explain how working memory functions during task performance.
Implemented in the ACT-R production system architecture, Kimberg and Farah’s
working memory model does not rely on an inhibitory mechanism to suppress prepotent
responses (Kimberg & Farah, 1993, 2000; Kimberg, D'Esposito, & Farah, 1997). Instead,
it posits that whether the prepotent responses gains control of the system is a function of
two parameters: (1) the strength of the connection between the current task goals (e. g.,
looking away from something that suddenly appears in peripheral vision) and the stimuli
that trigger those goals and (2) the preexisting associations of other task goals to the
stimuli in the current task set (e. g., looking toward a change occurring in peripheral
vision). The relative strengths of these two types of associations determine the extent to
which interference effects will be observed. Interference effects increase when correct
task associations and preexisting associations have similar strengths. Age-relaated
weakening of the strength of the connections between stimuli and new task goals
decreases the difference between the strength of correct task associations and preexisting
associations. Because the difference between association strengths is reduced the
likelihood that the prepotent response (e. g., looking toward the change) will be triggered
by the stimulus and gain control of the system is increased. This model has been used to

account for performance, not only on an antisaccade task, but on several other tasks

thought to require inhibition, including the Stroop task, the Wisconsin Card Sort Test,
and an A-not-B task.

Increasing competition between experimentally-defined and preexisting
associations is difficult to distinguish from increasing difficulty of suppressing a reflexive
or prepotent response and so testing the difference between these two hypotheses is not
the goal of this dissertation. An additional alternative to the inhibitory deficit hypothesis
is the idea that older adults have more difficulty keeping the current task goal active and
it is this goal neglect that leads to increased difficulty on a task that requires the
suppression of a prepotent response. This proposal is similar to the Kimberg and Farah
model of working memory, but its differences allow it to make unique predictions from

those made by the inhibitory deficit hypothesis. These differences are considered below.
1.2 Goal Neglect Hypothesis

Goal neglect has been advanced as an explanation for interference effects. Unlike
the Kimberg and Farah model, goal neglect allows for an inhibitory mechanism, but
proposes that it is not a failure of the inhibitory mechanism that leads interference effects,
but rather that sometimes the inhibitory mechanism is not engaged because of a failure to
access the task goals (De J ong et al., 1999). According to the goal neglect view of
interference effects, different schemas contain the task requirements for the current task,
the previous tasks, and other practiced tasks. In order for the current task to be executed
without interference from other task schemas, the current task goal must be maintained at
a high level of activation. When the current goal is active enough (i.e., when there is
sufficient “goal drive”), the goal-to-schema translation mechanism will operate well,

selecting the correct task schema and executing the correct task. As a result of schema

selection, competing task schemas are suppressed through lateral inhibition. If there is
insufficient “goal drive”, the goal-to-schema translation mechanisms are ineffective and
competing schemas may gain access to response production incorrectly. When the task
goal is not active enough it may not be accessible even though the requirements of the
task are understood. This causes slower correct responding or even incorrect responses
consistent with an automatic or prepotent response (see also, Duncan, 1995).

A critical difference between the goal neglect hypothesis (De Jong et al., 1999)
and the inhibitory deficit hypothesis is the point at which inhibition plays a role. In the
goal neglect hypothesis, inhibition is the result of the selection of the correct task schema
which relies on a high level of goal activation. The task goal activation process is
effortful and is affected by the efficiency of working memory, but it is unrelated to the
strength of the competing response. Only the strength of the current task goal determines
if interference will occur. In the inhibitory deficit hypothesis, inhibitory mechanisms
work in the service of correct task performance. The functioning of the inhibitory
mechanism is not directly related to the preparation for the current task to be performed.
Kimberg and F arah’s working memory model (1993, 2000) also differs from the goal
neglect view in that it claims that interference is the result of the relative strength of
associations of stimuli to correct and incorrect tasks. In this view, strong stimuli-correct
task associations may be sufficient to overcome weak preexisting associations, but as the
preexisting associations become stronger, interference will increase.

Support for a goal neglect hypothesis has come from many different paradigms
that used different methods of manipulating the likelihood that goal neglect would occur.

Because goal maintenance is an effortfirl process, the current activation of the task goal

will vary across task performance. The farther the presentation of the stimulus is from the
instantiation of the task goal (a process thought to occur during task performance) the less
likely the task goal will be at a high level of activation. Therefore, increasing the amount
of time between task performance and the presentation of the next stimulus will increase
the likelihood of goal neglect (De Jong et al., 1999). In a spatial Stroop task, the
response-to-stimulus interval (RSI) was manipulated to vary the likelihood of goal
neglect. It should be harder to keep the goal in a highly active state when there is a long
interval before the next trial begins compared to when there is a short interval. At a short
response-to-stimulus interval (RSI) of 200 ms, when attention was presumably tightly
focused on the task of location naming, no interference between location naming and the
location word was found; by contrast, following a long RSI of 2000 ms, when attention
was presumably less focused and sometimes accessing the current task goal was delayed,
interference was found. It is not clear how lengthening the duration of the RSI would
have made inhibiting the word reading response more difficult and so it appears that goal
neglect is the better explanation for this interference effect.

The larger interference effect observed in the long RSI condition was attributed to
insufficient activation of the task goal, location naming, on some trials. This claim was
supported by an examination of cumulative response distribution functions. A cumulative
response distribution is created by ordering each subjects’ reaction times in each
condition and separating the reaction times into an equal number of bins. Then the mean
reaction times for each bin in each condition are calculated. By comparing the plot of the
congruent and incongruent reaction times, the interference effects when reaction times

are fast can be compared to the interference effects when reaction times are slow (see

Figures 4a and 4b for an estimation of De Jong et al.’s (1999)). The cumulative response
distributions from the spatial Stroop experiment showed that the interference effect found
in the slow paced condition was only apparent at the slow end of the distribution
indicating that when a quick response could be made (i.e., attention was focused) the
interference effect was reduced.

Along slightly different lines, the goal neglect model has been applied to task
switching behavior. In a task switching experiment, participants switch between different
tasks throughout a block of trials. The appropriate task for the current trial is indicated by
a switch cue, a task instruction, the nature of the stimulus, or the ordering of the trials. A
standard finding in the task switching literature is that even when ample preparation time
is given to reconfigure the current task set before a new task is performed, response times
on the first trial following a task switch, the switch trials, are slower than responses on
nonswitch trials (e. g., Rogers & Monsell, 1995). Rogers and Monsell (1995) suggested
that these residual switch costs are due to an inability to completely reconfigure the task
set until the stimulus is presented.

Another idea proposed by De J ong and colleagues (1999; De J ong, 2000) is that
under some conditions, goal neglect, the failure to engage in advanced preparation for the
task switch, can better account for residual switch costs than models that propose a
limitation on the amount of preparation that can occur before a task switch. To evaluate
this hypothesis the cumulative response distributions for switch and nonswitch trials from
conditions with short and long preparation intervals were compared. At short preparation
intervals goal neglect should not play a role in processing because no one has time to

prepare for the upcoming task, but trials in the long preparation conditions should be

separable into those on which preparation occurred and those trials where preparation
was not done. It was found that the fast reaction times on switch trials fi'om the long
preparation condition resembled reaction times on nonswitch trials. It seemed that on
some switch trials participants could completely prepare for the new task and so their
reaction times were the same as for nonswitch trials. In contrast, the reaction times from
the switch trial that were responded to slowly were similar to the reaction times from
trials with short preparation intervals suggesting that on some switch trials with long
preparation intervals participants did not engage in advanced preparation and so their
performance was the same as if there had not been time to prepare. This mixture model
provided a good fit to the observed data (De Jong, 2000; but see Mayr & Keele, 2000, for
an inhibitory account of residual switch costs).

Evidence that antisaccade performance may be disrupted by goal neglect comes
from studies that have examined the relationship between working memory and
antisaccade performance. In one study, individuals with low and high working memory
abilities, as measured by an operation span task, were compared on the antisaccade and
prosaccade tasks (Kane, Bleckley, Conway & Engle, 2001). In the first experiment of this
study eye movements were not directly measured. Instead, performance identifying a
target that appeared for a short time at the location participants were supposed to look at
was used as a measure of success on the eye movement task. It was assumed that if
participants looked in the wrong direction they would not be able to identify the target.
Low span individuals were less likely to correctly identify the target than high span
individuals in the antisaccade condition suggesting that the low span individuals were

more likely to make a saccade direction error, but only when the prosaccade task was

10

performed after the antisaccade task. Kane and colleagues suggested that when the
requirements of the antisaccade condition were made less novel by first performing
prosaccades, the difference in antisaccade performance between low and high span
individuals was eliminated. Reducing the novelty of the antisaccade task should have also
made it easier to maintain that task goal, and so, this finding may indicate that when the
likelihood of goal neglect is reduced, antisaccade performance improves, especially in
low span individuals.

In a second study of working memory's role in antisaccade performance, Roberts,
Hager, and Heron (1994) found that a concurrent working memory load led to increases
in saccade direction errors and saccade latencies, but only in the antisaccade condition.
They suggested that the critical aspect of performance on the antisaccade task was
attentional vigilance. "Successful performance seems dependent on maintaining a high
enough level of activation of the relevant self-instructions to make an eye movement to
the opposite side at the moment the cue is presented (pp. 391, Roberts et al., 1994)."
When the self-instruction activation level declines, e. g., when attention must be devoted
to a concurrent task, goal neglect occurs and the prepotent response is executed.
However, despite this relationship of working memory load to antisaccade performance,
saccade direction errors, antisaccade latency, and target task performance were not
correlated with performance on a version of Daneman and Carpenter's (1980) sentence
span task or a counting span task. This finding is contrary to the differences in
performance based on working memory span found by Kane et al. (2001). This
discrepancy may be accounted for by the smaller variation in working memory span

scores for participants in the Roberts et al. (1994) study. Kane et al. (2001) selected

11

participants with extremely low or high scores on the span measure to compare, whereas
Roberts et al. (1994) did not.

Duncan and colleagues also proposed that goal neglect may be the source of
performance difficulty experienced by individuals with impaired frontal lobe functioning,
including older adults, individuals with frontal lobe damage, and individuals with low
ability to reason and solve problems as measured by fluid intelligence, (Duncan, 1995;
Duncan, Emslie, Williams, Johnson, & Freer, 1996). Although Duncan’s definition of
goal neglect, the inability to adhere to task requirements even when those requirements
are understood, is the same as that of De Jong et al.’s (1999), the conceptualization of the
role of goal activation is somewhat different. Duncan suggests that after very little
practice with the task requirements, or once the task is performed correctly, goal neglect
should no longer occur, suggesting that goal activation is no longer effortful. De Jong et
al. (1999) claimed that goal activation is effortful across the entire period typically
associated with an experimental task and so goal neglect may occur on trial n+1 even if
trial n was performed correctly. This research will consider goal neglect to be an effortful
attentional process that occurs throughout task performance.

Although goal neglect may account, in part, for performance on some interference
tasks it is not clear that it will account for the age-related increase in performance
difficulty seen on tasks that require a prepotent response. The next section will consider
some experimental evidence that does indeed support a goal neglect account of larger

interference effects seen with older adults.

12

1.2.1 Evidence for a goal neglect account of aging

Older adults may be more susceptible to goal neglect than young adults and this
increase in susceptibility may result in the incorrect generation of a prepotent response or
difficulty generating the correct response when a prepotent response is possible. The
results from a Stroop interference task that required participants to switch between color
naming and word reading made an argument similar to the increased susceptibility to
goal neglect of older adults (West, 1999). In this paradigm, the task to be performed was
cued by the stimulus. When a gray word was presented it was supposed to be read and
when a colored word was presented the color was supposed to be named. The proportion
of word reading to color naming trials was varied. When the activation of the competing
goal, word reading, was increased by increasing the proportion of word reading trials, the
presumed result was greater difficulty maintaining the task-relevant color naming goal,
especially for older adults. As a result older adults showed a greater interference effect in
color naming performance relative to young adults as the proportion of word naming
trials increased. This finding suggests that older adults had more difficulty maintaining
the current task goal than young adults. Alternatively, if practicing word reading within
the experimental context makes it more difficult to suppress the word reading response it
could account for the

An interesting parallel to the maintenance of task goals is the maintenance of the
intention to perform a prospective memory task. In a prospective memory task, as
individuals perform a background task they are required to maintain the intention to
perform a second task much in the same way that goals to perform a task must be

maintained in a conflict situation. Generally, older adults have more difficulty on

13

prospective memory tasks than young adults, but these age differences are reduced when
a cue is presented at the moment that the prospective memory task is to be executed (e. g.,
West & Craik, 1999). Age differences on prospective memory tasks support the idea that
older adults are more susceptible to goal neglect. In addition, the elimination of age
differences when prompts are given suggests that external cues that signal the task to be
performed aid the performance of older adults more than the performance of young
adults.

Another piece of evidence in support of a goal neglect account of age-related
prospective memory failure comes from a retrieve-delay prospective memory paradigm
(Einstein, McDaniel, Manzi, Cochran, & Baker, 2000). In this task an event signals that
the required response should be made, but the response must be delayed until a different
phase of the experiment. The intention to respond to the prospective memory cue has to
be kept active either while an unrelated task is performed or during an unfilled interval
when no other task is performed. In both filled and unfilled delay conditions older adults
had more difficulty maintaining a goal across an interval as short as 10 seconds.
Admittedly, the interval over which a prospective memory cue must be maintained is
much short than the difference found between RSIs of 200 and 2000 ms, but note that the
measure in the prospective memory tasks is failure to perform altogether whereas the
measure on the spatial Stroop task that varied RSI was an increase reaction times on
incongruent trials. Perhaps at short prospective memory intervals differences in response
time to prospective memory cues would be observed.

Goal neglect may play a role in older adults’ performance on an antisaccade task

as well. Nieuwenhuis and colleagues (2000) examined the antisaccade performance of

14

older adults under conditions that varied the amount of external information available to
participants about the correct eye movement response. In their study, as in others (e.g.,
Rogers et al., 1994; Butler et al., 1999), a target for a perceptual task was displayed a
short time after the presentation of the peripheral onset at the location that was the goal of
the correct saccade. The requirement to identify the target after completing the eye
movement is used to motivate the participants to try to perform the saccade task correctly
and quickly. However, if the correct antisaccade is not programmed quickly the onset of
the target may visual trigger the correct antisaccade (e. g, Guitton, Buchtel, & Douglas,
1985)

In Experiment 1, Nieuwenhuis and colleagues (2000) varied the cue-to-target
interval and found that for older adults, but not young adults, antisaccade latencies were
slower as the interval increased, suggesting that older adults were relying on the onset of
the target to guide their response. In Experiment 2 this source of environmental support
was eliminated by having distracting stimuli onset at the other potential target locations
simultaneous with the onset of the target. This procedure forced older adults to utilize the
initial cue, the peripheral onset, to generate their correct antisaccades. When the
environmental support was eliminated, the age differences in correct antisaccade
latencies and corrective saccade latencies following saccade direction errors were
reduced at long cue-to-target intervals. The age difference in saccade direction accuracy
was also much larger in Experiment 2 at the long intervals. Nieuwenhuis et al. (2000)
suggested that under antisaccade instructions maintaining the goal is difficult, especially
for older adults, and so they look for an easier strategy for performing the task. In

Experiment 1, older adults were able to utilize the onset of the target to guide their eye

15

movements. At short cue—to-target intervals this led to faster and more accurate
antisaccade performance, but at long cue-to-target intervals this slowed their responses
because they were waiting for the target to occur. When an alternate method for
performing the task was not available, older adults made more saccade direction errors
than they did in Experiment 1 with short cue-to-target intervals. In addition, the
difference in antisaccade latencies compared to prosaccade latencies was greater for older
than younger adults. As with prospective memory tasks, older adults were able to
perform the task better when external prompts signaled the correct response, but unlike
prospective memory tasks, even with external prompts older adults had more difficulty
than younger adults.

Older adults may be more susceptible to goal neglect than young adults. In
addition, difficulty maintain the current task goal may lead to more problems for older
adults when the current task requires the production of a controlled response, but the
stimuli are associated with different automatic or prepotent responses. The specific
question of this research is whether behavioral control difficulties displayed by older
adults are better accounted for by an impaired inhibitory mechanism, reduced ability to
stay on task, or both.

1.2.1 Ideas related to goal neglect

As was described above, Duncan and colleagues’ (Duncan, 1995; Duncan et al.,
1996) view of goal neglect is quite similar to the goal neglect hypothesis put forward by
De J ong and colleagues (De J ong et al., 1999, De J ong, 2000) with one exception being
that goal activation is more transient in De J ong’s model. Consistent with this transient

view of goal activation, research using the prospective memory paradigm has suggested

16

that the activation of the intention to perform a prospective memory task may fluctuate
across the course of an experiment and that older adults may be more prone to these
fluctuations.

Maylor (1996) examined how performance on a prospective memory task may
fluctuate over time by calculating forgetting of the task once it had been successfully
performed and recovery from that forgetting. Older adults were more likely than young
adults to forget to perform the prospective memory task after they had previously
successfully responded, and conversely, older adults were less likely to recover from that
forgetting (Maylor, 1996; West & Craik, 1999). Because the success of prospective
memory performance fluctuated across the course of the experiment the errors were
termed momentary lapses of intention, short periods of time when the activation of the
intention to perform an action falls below the threshold for responding (Craik & Kerr,
1996; West & Craik, 1999). Not only has this idea been applied to failures to respond on
prospective memory tasks, but it has been proposed as a means of accounting for word
reading errors on the color-naming version of the Stroop task (West, 1999). In the Stroop
task, a momentary lapse of intention results in a failure to perform the current task, color-
naming, allowing the more natural or automatic response to control the response system
and resulting in a response error (West, 1999, 2001; West & Alain, 2000). Older adults
appear to be more susceptible to momentary lapses of intention on the Stroop task as well

as on prospective memory tasks (West, 1999).
1 .3 Summary

When the conditions of a situation require the production (or withholding) of a

response in an intentional manner, despite the fact that a different prepotent response is

17

associated with that situation, how does the system select the correct response? One view
posits that an inhibitory mechanism suppresses the reflexive response and that this
inhibitory mechanism may not function as well in older adults compared to young adults.
In contrast, it may be that the inhibitory process functions well, but the control processes
responsible for engaging inhibition are sometimes impaired resulting in a failure to
engage the inhibitory mechanism. Although these two hypotheses make distinct
predictions about how performance should be impaired in older adults, they are not
mutually exclusive. It may be the case that older adults have an impaired inhibitory

mechanism, as well as an increased susceptibility to goal neglect.

l8

2 Experiment 1

The difficulty that older adults have on interference tasks may be due to difficulty
maintaining the activation of the intended goal. If goal neglect is an adequate explanation
of the deficit that older adults experience then, under conditions that increase the
likelihood of goal neglect, older adults should show a greater disruption of performance
than young adults. In this experiment the difficulty of maintaining the goal was
manipulated by varying the interval between the response on trial n and the beginning of
trial n +1, the response-to-stimulus interval (RSI). As was described above, De Jong et al.
(1999) used this manipulation in a spatial Stroop task and found evidence in support of
the goal neglect hypothesis. Because goal maintenance is an effortfirl process the average
activation of the goal following long intervals will be lower than the activation following
short intervals. Higher goal activation will lead to better performance than low goal
activation.

Young and older adults were tested on two interference tasks, the spatial version
of the Stroop task and the antisaccade task. In order to evaluate the goal neglect account
of age-related increases in interference effects, the likelihood of goal neglect was
manipulated between subjects. Participants performed each of these tasks with either long
or short RSIs in a similar manner to that used by De J ong et a1. (1999) in the spatial
Stroop task. In both tasks, the RSI was timed from the button press response made on the
previous trial until the presentation of the next response stimulus. At short RSIs
individuals should be better able to maintain the current task goal and performance on the
interference trials should be less often influenced by the competing task compared to

performance following long R815.

19

In the spatial Stroop task, the interference effect, as measured by the difference in
reaction time between congruent and incongruent trials, should be larger in the long RSI
condition than in the short RSI condition as De J ong et al. (1999) found. In the
antisaccade task, interference will be measured as the frequency of incorrect eye
movements toward the peripheral onset in the antisaccade condition and the difference
between correct saccade latencies in the prosaccade and antisaccade conditions.
Increasing the likelihood of goal neglect should lead to slower reaction times and more
frequent saccade direction errors in the antisaccade condition because of the increased
likelihood that the nongoal task, looking toward the onset, will interfere with
performance. Following a long RSI, if the activation level of the goal is, on average,
lower at the time the onset occurs, then the incorrect response, looking toward the
peripheral onset, will more often interfere with correct performance. The increased
probability of interference in the long RSI condition will result in a larger interference
effect in the spatial Stroop task and more frequent saccade direction errors and greater
latency differences in the antisaccade task.

The question of most interest for this research is the effect of the RSI
manipulation on older adults’ performance. If this group of individuals has more
difficulty maintaining the current goal state than young adults, then the effect of the time
manipulation should be more pronounced for them compared to young adults in both the
spatial Stroop and antisaccade tasks. Participants in this study completed the eye
movement tasks before the Stroop task, but for the exposition purposes the Methods and

results from the Stroop task will be reported first.

20

2.1 Method

2.1.1 Participants

Thirty-four young adults (Age range = 18 - 23 years) were recruited from
undergraduate psychology courses and participated in the experiment for partial course
credit. One young subject (in the long RSI condition) frequently anticipated the
presentation of the peripheral cue and was excluded from the analysis.

Forty-one community-dwelling older adults (Age range = 63 — 80 years old) were
recruited from the Lansing, MI area and paid $10/hour for their participation. The data
from one older adult were excluded from both the Stroop and eye movement analyses
because of frequent anticipations in the eye movement task and a failure (or an inability)
to follow instructions in the Stroop task. Thirty-eight older adults completed the Stroop
task. Two older adults did not complete this portion of the experiment because of time
limitations. Thirty-two of the older adults completed the eye movement tasks
satisfactorily and were included in the eye movement analyses. The data from three
individuals were excluded because of frequent anticipations, eye movements before the
presentation of the peripheral onset, — two in the long and one in the short RSI condition.
In addition, eye movement data were not obtained from five older adults who could not
be reliably tracked during the eye movement tasks because of droopy eyelids and/or
excessive blinking; these individuals did complete the Stroop task, however. Young and
older adults reporting learning disabilities, psychoactive medications, or a history of brain
trauma (including stroke) were excluded from the sample. All participants were able to
discriminate letters that subtended less than 6" of visual angle. Mean ages, vocabulary

test scores, and education levels of the participants in each group in both tasks are

21

presented in Table 1. There was no difference between the RSI conditions in mean age or
education levels, F s < 1. However, despite the random assignment, individuals in the
short RSI condition had lower mean vocabulary scores than individuals in the long RSI
condition, F (l, 68) = 4.8, MSE = 9.3, p = 0.032). This difference is small and unlikely to
have an impact on performance in the experimental tasks. In addition, older adults had
more years of education and higher vocabulary test scores than young adults, Fs (1, 69) =
90.8 and 17.3, MSEs = 9.3 and 3.9, p’s < 0.001, respectively. These differences should

ensure that observed age differences are not due to a low functioning older adult

 

 

population.
Young Old
Short RSI
Age (years) 19.4 72.5
Education 13.2 15.2
Vocabulary (mean 27.9 34.3
n ‘ 18 19
Long RSI
Age (years) 19.7 72.4
Education 13.4 15 .2
Vocabulary (mean 29.0 36.4
n ‘ 16 21

 

Table 1. Experiment 1 Participant Characteristics

2.1.2 Design

The spatial Stroop task had a two age group (young or old) by two RSI (218 ms or
2028 ms) by two item type (congruent or incongruent) design with age group and RS1 as
the between subjects variables and item type as the within subjects variable. The eye
movement tasks had a two age group (young or old) by two RSI (1200 ms or 3200 ms) by

two eye task (prosaccade or antisaccade) design with age group and RSI as between

22

subjects variables and eye task as a within subjects variable. Participants were in the
same RSI condition for both the eye movement and Stroop tasks, i.e., participants in the
short RSI condition of the eye movement tasks also completed the short RSI version of
the Stroop task and vice versa.

2.1.3 Apparatus and Stimuli

Spatial Stroop task. Stimuli for the Stroop task were displayed using E-prime
software rumring on a NEC Multisync XE15 monitor controlled by a Pentium PC-
compatible computer (Psychological Software Tools, Inc., 2000). The experiment
program set the RSI from the button press on trial n to the stimulus display on trial n +1
at 200 and 2000 ms for the short and long RSI conditions, respectively. Because the
computer program took time to prepare the stimuli for presentation after the RSI had
elapsed the actual RSIs were slightly longer. This stimulus preparation time was on
average 18 ms in the short RSI and 28 ms in the long RSI conditions and varied at most 1
ms from trial to trial. This resulted in an actual RSI of 21 8 ms in the short RSI condition
and 2028 ms in the long RSI condition. This small difference in timing fi'om the De J ong
et al. (1999) experiment should have had a minor effect on our results and will not be
discussed further.

The stimulus display contained a row of four asterisks in the center of the screen
that subtended 3.l°. The word ABOVE or BELOW appeared above or below the row of
asterisks in 18 point Sans Serif font. The letters subtended a visual angle of 08° in height
and 05° in width. The entire word subtended a visual angle of 33° and the space
between the asterisks and the word was 02°. The sizes of these stimuli were chosen to

approximate the sizes of those used by De Jong et al. (1999).

23

Participants completed 10 blocks of trials with 100 trials in each block. Each
block contained 50 congruent trials, e. g., the word ABOVE appearing above the row of
plus signs, and 50 incongruent trials, e. g., the word ABOVE appearing below the row of
plus signs. The trials were presented in a random order.

Antisaccade task. Eye movements were recorded with an ISCAN RK-416 high-
speed eyetracker that uses an infrared video-based system to compute and horizontally
track the center of the pupil in the right eye. Signals were generated by the eyetracker at a
frequency of 120 Hz allowing saccade latencies to be calculated with a temporal
resolution of 8.33 ms. The spatial resolution of the apparatus is 02° of visual angle.
Stimuli were displayed at a resolution of 800 by 600 pixels on the computer system
described above.

A chin and forehead rest was used to stabilize the participant’s head at a viewing
distance of 49 cm. The fixation display contained a white cross in the center of a black
screen flanked by two white boxes to the left and right. The fixation cross and boxes
subtended visual angles of 08° and 12°, respectively, and the distance fiom fixation to
the inside edge of the box was 105°. The target arrow and nontarget double-headed
arrow each appeared inside one of the boxes on each trial and both subtended 08°. The
onset was defined by the inside of one of the boxes changing fiom black to white.

Participants completed two blocks of prosaccade and four blocks of antisaccade
trials. The second prosaccade block was completed at the end of the experiment. Each
block contained 40 trials, 10 of each peripheral onset location (left or right) by target
arrow direction (up or down) combination. The trials were presented in a different

random order for each subject. The three prosaccade and two antisaccade task practice

24

blocks each contained an equal number of trials fiom each onset location by arrow
direction combination presented in a random order. Two of the prosaccade and one of the
antisaccade practice blocks contained 12 trials and the remaining practice blocks
contained 24 trials.

In order to manipulate the RSI while maintaining an accurate calibration of the
eye movement monitor, strings of trials were performed punctuated by brief periods to
check calibration. The 40 experimental trials in each block were presented as 5 strings of
8 trials each. Within a string of trials, the eye movement monitor was not recalibrated so
that the interval between the response on trial n and the presentation of the peripheral
onset on trial n+1 could be precisely controlled to be either 1200 or 3200 ms.

2.1.4 Materials

Acuig test. Before beginning the eye movement task each participant was asked
to read letters from a photocopied miniature version of the Snellen vision chart taped to
the computer monitor. While sitting with their face in the chin and forehead rest
participants were asked to read as many of the letters as they could starting at the top of
the chart. Participants were said to be able to discriminate letters at a particular visual
angle if they incorrectly identified no more than one letter in the row of that size.

Vocabulm test. At the end of the eye movement tasks participants completed the
Shipley Institute of Living Scale - Vocabulary Test (Shipley, 1940), a 40-item multiple
choice test.

2.1.5 Procedure
Before beginning the experiment, the apparatus was explained to the participant

and the participant signed a consent form. Then, while sitting in the chin and forehead

25

rest, the participant’s acuity to discriminate letters presented at a distance of 49 cm was
tested.

Antisaccade task. Following the acuity test, instructions describing the prosaccade
condition were read by the participant and then the participant was shown a booklet that
illustrated the displays and the order they would be shown, with the experimenter
explaining what the participant should be doing during each display. The instructions
emphasized that it was important to look as quickly and as accurately as possible toward
the peripheral onset when it occurred and to press the button that corresponded to the
direction that the arrow was pointing, either up or down. Participants were instructed to
guess if unsure about the direction of the arrow. One prosaccade block was always
performed before the antisaccade blocks to allow participants to become familiar with the
displays and timing parameters (see Butler et al., 1999).

Before beginning the practice trials, the eye movement monitor was calibrated
horizontally by sampling the center of the pupil while the eyes were looking at a point on
the far right and a point on the far left center of the screen. After the calibration was
checked for accuracy, the participant was reminded of the instructions and then began the
first of three blocks of practice trials. In the first block of 12 trials, an opportunity was
given after each trial to ask questions. Then participants were told that during the actual
experiment several trials would be presented consecutively without pauses between them
and so it was important, after pressing the button, to move the eyes back to the center of
the screen and get ready for the next trial to begin. The second practice block allowed
participants to practice responding to several trials in a row by presenting four strings of

three consecutive trials each for a total of 12 trials. In the final block of practice, the

26

experimental procedure was used. Participants completed three strings of eight trials for a
total of 24 trials in this practice block. Next the first experimental block of 40 prosaccade
trials was completed. Then participants read the antisaccade instructions to look at the
box in the opposite direction from where the peripheral onset was presented. For the
antisaccade task, participants completed two blocks of practice trials, a block of 12 trials
with pauses after each, and then a block of 24 trials presented in three strings of eight
trials each. The antisaccade practice was followed by four blocks of 40 antisaccade
experimental trials and a final block of 40 prosaccade trials. Participants were reminded
of the task instructions following the initial calibration in each block. They were also
given short breaks after each practice and experimental block of trials. In addition, after
the second block of experimental trials, participants filled out a background questionnaire
that asked about medications, mental illness, and learning disabilities and following the
final block of eye movement trials participants completed the vocabulary test. Following
another break participants completed the spatial Stroop task. The entire experiment took
approximately 1.5 hours for young adults and 2 hours for older adults.

The sequence of events for each trial in the eye movement tasks is explained below and
illustrated in Figure 1. At the beginning of a block of trials the fixation display, the
fixation cross flanked by box place markers, was presented. After checking the
calibration the experimenter initiated the string of trials. At the beginning of each string
the fixation display was visible for an additional 1000 or 3000 ms (in the fast and slow
RSI conditions, respectively). The trial began when the fixation cross disappeared, and
then, after 200 ms, the peripheral onset was presented, i.e., the inside of one of the boxes

changed from black to white. After 400 ms the target display was presented for 150 ms.

27

The target display consisted of a single-headed arrow in the box that was the goal of the
correct saccade and a double-headed arrow in the opposite box. The perceptual
identification target was only presented in the box that was the goal of the correct saccade
to motivate the participant to move their eyes as quickly as possible in the correct
direction. In the prosaccade condition, the single-headed arrow replaced the peripheral
onset. In the antisaccade condition, the single-headed arrow appeared in the box opposite
from where the peripheral onset had been. After the target display disappeared, both
boxes appeared empty until a button press response was made. Following the button
press, the fixation cross flashed in the center of the screen for 500 ms at 50 ms intervals
as a cue to move the eyes back to fixation. Then the fixation display remained on the
screen until either 500 or 2500 ms had elapsed (in the fast and slow-paced conditions,
respectively), at which time the next trial began. During each experimental block the
experimenter wrote down the initial looking direction on each trial and coded whether a
trial should be excluded because of poor calibration or failure of the participant to

maintain fixation.

28

 

 

 

Displayed until button
press response

 

 

 

 

 

 

 

 

 

 

 

 

Target Display t l
(150 ms)

 

 

 

 

 

 

 

 

 

 

 

 

Cue Display
(400 ms)

 

 

 

 

Time

 

 

 

 

 

 

Gap Display
(200 ms)

 

 

 

 

 

 

 

 

 

 

 

 

Fixation Display _|_
(1000 or 3000 ms)

 

 

 

 

 

 

 

 

 

Figure 1. Illustration of a prosaccade trial. In the antisaccade condition, the double-
headed arrow would have replaced the peripheral onset and the single-headed arrow
would have appeared in the opposite box. In the actual displays stimuli were white
presented on a black background.

Spatial Stroop Task. The procedures used in the spatial Stroop task were taken

from De J ong et al. (1999). After completing the eye movement tasks and the vocabulary

29

test, participants were given a break. For the spatial Stroop task, the instructions,
presented on the computer screen, told participants to press a button to indicate whether
the word appeared above or below the row of plus signs. The instructions emphasized
that responses should be made as quickly as possible while keeping errors to a minimum.
Participants were asked to respond by pressing the “1” key on the number pad with their
left index finger and the “3” key with their right index finger. The keys were labeled
above with the words “ABOVE” and “BELOW”. The mapping of keys to labels was
counterbalanced for each age group by RSI condition.

For each trial, a row of asterisks was presented in the center of the screen with the
word ABOVE or BELOW above or below it. The word disappeared fiom the screen
when the participant pressed a button. During the RSI nothing was presented on the
screen. After 218 ms in the short RSI condition or 2028 ms in the long RSI condition the
next stimulus was presented. Ten blocks of 100 trials each were presented with the first
block serving as practice. If performance during the practice block was worse than a 90%
accuracy rate, a message appeared on the screen reminding the participant to respond to
the location of the word on the screen as quickly as possible while keeping errors to a
minimum. Participants were given a break after each block and initiated the next block by
pressing the space bar.

2.1.6 Data Analysis

The power to detect a moderate size main effect (f =.25, Cohen, 1977) with 32
participants in each cell was good according to Cohen’s suggested criteria, power = .79.
The power to detect an interaction of age with RSI with 16 participants in each cell was

rather low, power = .55.

30

Spatial Stroop task. For the spatial Stroop task the first block of trials was
considered practice and was excluded from the analyses. In addition the first trial fi'om
each block was excluded because the participant had to press the space bar to begin the
block and may not have been fully prepared to perform the task on the first trial. Also,
trials with reaction time less than 100 ms were excluded as anticipations accounting for
less than 0.1% of the data. Accuracy rates were calculated based on the remaining trials.
Mean reaction times for each participant by item type were based on the reaction times
for correct responses. Mean correct response reaction times and accuracy rates in the
congruent and incongruent conditions were submitted to separate 2 age group (young or
old) by 2 RSI (short or long) by 2 item type (congruent or incongruent) repeated
measures AN OVAs with age group and R81 as the between subjects variables and item
type as the within subjects variable.

Analyses were also run on mean reaction times based on two different sets of
data. The first set of data was created by using the criteria stated above, with the
additional restriction that trials following an error were excluded. The second method
excluded both trials following an error and trials that were more than 3 standard
deviations from the individuals mean reaction time for that item type. The results of
analyses using the two latter methods were the same as those reported here.

Antisaccade task. For the eye movement data, the first step of the data analysis
was to determine the position and duration of fixations. During data acquisition, eye
movement position was sampled at 120 Hz allowing a new sample to be recorded every
8.33 ms. To reduce the size of the data files, a new sample (including the horizontal

position, start time and end time of the sample) was recorded only if the position of the

31

eye changed more than .4° from the initial sampling position. The end time of a sample
was the time at which the eye position changed more than .4° from the initial sampling
position. Any samples of less than 15 ms were removed from the eye tracking record as
representing saccades. Fixations were coded as the time between saccades and the
position of the fixation was calculated by averaging the position of the samples during
that continuous time period weighted for the duration of each sample. The initial saccade
was the first change in fixation position that was greater than 1° from the previous
frxation and that occurred following the presentation of the peripheral onset. The latency
of the initial saccade was the time from the presentation of the peripheral onset until the
end of the fixation prior to the initial saccade. The initial saccade was considered to have
been made in the correct direction if it was toward the location of the peripheral onset in
the prosaccade condition, or if it was toward the box opposite from the peripheral onset in
the antisaccade condition. Trials were excluded from all analyses if the initial fixation
was not within 4° of the center of the screen which could have occurred because of either
poor calibration or an anticipatory eye movement. In addition, any trial that the
experimenter had hand-coded as having poor calibration because of subject movements
was also excluded. The accuracy of button press responses were calculated for trials not
excluded from the eye movement analysis, but including trials on which a correct or
incorrect eye movement response was made. Mean reaction times were calculated from
the correct button press responses.

Four measures from the prosaccade and antisaccade conditions were submitted to

repeated measures ANOVAs with age group and RS1 as between subjects variables.

32

These measures were saccade direction accuracy, mean correct saccade latencies, button

press accuracy, and mean correct button press reaction times.

2.2 Results

2.2.1 Spatial Stroop Task

Reaction Times. The mean reaction times for each age group and RS1 condition
are presented in Table 2 by item type. Responses to congruent trials were faster than
responses to incongruent trials (Ms = 533 and 555 ms, respectively), F (l, 68) = 54.2,
MSE = 333.4, p < 0.001, and young adults were faster than older adults (Ms = 464 and
623 ms, respectively), F (1, 68) = 42.3, MSE = 21329, p < 0.001. Consistent with the
findings of De J ong et al. (1999), individuals were faster in the short RSI condition than
in the long RSI condition (Ms = 508 and 580 ms, respectively), F (1, 68) = 8.6, MSE =
21329, p < 0.01, and the effect of RSI did interact with item type, F (1, 68) = 14.2, MSE
= 333.4, p < 0.001. The effect of item type was reduced in the short RSI condition
compared to the long RSI condition (differences of 11 and 38 ms, respectively). The
interference effect in the short RSI condition is the same size as the difference found by
De J ong et al. (1999) which they reported to be nonsigrrificant. In contrast, when the
mean reaction times for only the short RSI conditions were submitted to an ANOVA in
this experiment, the effect of item type was still significant, F (l, 34) = 10.5, MSE =

205.5, p < 0.01.

33

Young Old

 

 

Mean Acc Mean Acc
Short RSI
Congruent 431 (20) 96.5% 574 (22) 98.5%
Incongruent 443 (24) 96.1% 584 (25) 98.2%
Difference 12 10
Long RSI
Congruent 478 (24) 97.6% 647 (24) 99.0%
Incongruent 505 (29) 96.5% 688 (27) 98.1%
Difference 27 41

 

Table 2. Mean reaction times (msec) for correct triab in Stroop task

Surprisingly, age did not interact with any variables, Fs < 1.6. Although the
difference in reaction time between congruent and incongruent trials was slightly larger
for the older adults than for the young adults (interference effects of 26 and 19 ms,
respectively), the difference was not statistically significant, F < 1. Because the goal
neglect hypothesis predicted that the age difference in the interference effect would be
more pronounced in the long RSI condition than in the short RSI condition, mean
reaction times in the long RSI condition were analyzed separately. Again the interaction
of age group and item type was not significant, F = 1.8. Under conditions that
presumably increased the likelihood of goal neglect (long RSI condition) older adults
were not more adversely affected than young adults.

It is possible that the large amount of practice in this experiment eliminated an
age difference in the interference effect. To evaluate this hypothesis, performance in the
first block of experimental trials was analyzed separately. The results of the first block
and whole experiment analyses were the sarrre indicating that, even early in task

performance, the interference effects were equivalent for older and young adults.

34

In the long RSI condition, the size of the interference effect was larger for the
older adults than for the young adults, but not significantly so. If the slower responding of
older adults in the congruent condition is taken into account, this difference may be
eliminated. The logic for this suggestion is that this general slowing exhibited by older
adults is the result of the slowing of each stage of information processing, and that stages
of processing that take more time will be slowed to a greater extent. Thus, when
comparing conditions that take different amounts of time to complete(like congruent and
incongruent conditions) the condition that takes more processing time(incongruent) will
be slowed to a greater extent resulting in a larger interference effect. The slower response
times of older adults was controlled by considering the interference effect as a proportion
of the speed of responding. This proportion was calculated by dividing each participant's
interference effect by their congruent trial reaction time. Young adults were 5% slower
on incongruent trials in the long RSI condition, older adults were 6% slower, F < 1,
suggesting that older and younger adults were slowed to the same extent in the long RSI
condition. Admittedly, if RSI had a moderate effect on age differences this experiment
would have detected it only 55% of the time.

Accuracy Rates. In general accuracy rates were quite high, 97.3%. The results of
the ANOVA were similar to the results fiom the reaction time analysis. Accuracy was
higher in the congruent condition than in the incongruent condition, F (l, 68) = 22.6,
MSE = 0.000069, p < 0.001, although the difference was less than 1% (98.0 vs. 97.3%,
respectively). RSI did modulate this effect though, with the difference in accuracy rates
between the congruent and incongruent conditions being greater for the long RSI

condition than for the short RSI condition, 1.2% vs. 0.3%, respectively, P (1, 68) = 6.6,

35

MSE = 0.000069, p < 0.02. In addition, older adults were more accurate than young
adults (98.4 and 96.5%, respectively), F (1, 68) = 14.2, MSE = 0.0008, p < 0.001, but age
did not interact with any other variable, F s < 1. The increased accuracy of older adults
coupled with their slower reaction times compared to young adults suggests that the age
difference in overall performance on the spatial Stroop task may be attributed to a
speed/accuracy trade-off, at least in part.
2.2.2 Antisaccade Task

Trials were excluded from the eye movement analyses if the calibration was
misaligned for that trial, if the participant was not looking at the center of the screen
when the onset occurred, or if a saccade was not made within 1000 ms of the beginning
of the trial.I These criteria excluded 6.3% of the young adults’ data and 15.8% of the
older adults’ data, a statistically significant difference, F ( 1, 61) = 14.0, MSE = 0.02078,
p < 0.001. In addition, a greater percentage prosaccade trials were excluded than
antisaccade trials, probably because the prosaccade task was performed first and
participants were more likely to move (misaligrring the calibration) early in the
experiment (Ms = 12.0 and 9.9 %, respectively), F (l, 61) = 4.6, MSE = 0.00326, p =
0.036. There was no difference in percent of trials excluded based on RSI and this
variable did not interact with any other variables, F s < l.

Sagde Direction Accuracy Rm Mean saccade direction accuracies are
presented by age group, eye movement task, and RSI condition in Figure 2. Antisaccade
responses were less accurate than prosaccade responses (Ms = 80.9 and 96.7%,

respectively), F ( 1, 61) = 106.4, MSE = 0.00765, p < 0.001, and older adults were less

36

accurate than young adults (Ms = 85.6 and 92.0%, respectively), P (1, 61) = 11.8, MSE =
0.01103, p = 0.001. Age and eye movement task interacted to indicate that older adults
had more difficulty producing correct antisaccades than young adults, F (1, 61) = 19.0,
MSE = 0.00765, p < 0.001. Accuracy was better in the short RSI condition than in the
long RSI condition (Ms = 90.6 and 87.0%, respectively), F (1, 61) = 4.0, MSE = 0.01103,
p = 0.05, and this difference was larger in the antisaccade condition than in the
prosaccade condition, F (l, 61) = 6.7, MSE = 0.00765, p = 0.012. Although RSI did have
a larger effect on saccade direction accuracy in the antisaccade condition, it did not enter

into any interactions with age, F s < 1.1.

 

 

' Short RSI Long RSI
1.00

+-
0.95 7 7P7 , , . —I— , 7 r1— - - 7 ,
0.90 77 7 TI— 7 7 7 7 7 7 7 7 7 7
0.85 7 7 7 7 777777 7 7
0.80 7 7 7
0.75 J 7 7 7 ,
0.70 7 - Em"
0.65 77 7 7 7 7 7
0.60 7 7 7 7 7
0.55 7 7 7 7 7 7 7
0.50 + . 7
Young Older Young Older

Age Group

DEC

Mean Accuracy Rate

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2. Mean saccade direction accuracy rate of the initial saccade in the antisaccade
and prosaccade conditions by RSI, short or long, and age group, young or old. Error bars
are equal to 1 standard error. Anti = Antisaccade; Pro = Prosaccade.

 

‘ Anticipation errors and misalignment errors were indicated by the same eye movement
pattern (the eyes more than 4 degrees from fixation when the peripheral onset was
presented). Therefore these types of errors will not be considered separately.

37

Correct Saccade Latencies. The mean latencies by age group, RSI, and eye movement
task are presented in Figure 3. Antisaccades were initiated more slowly than prosaccades
(Ms = 345 and 255 ms, respectively), F (1, 61) = 289, MSE = 915, p < 0.001, and older
adults were slower than young adults (Ms = 338 and 261 ms, respectively), F (1, 61) =
58.6, MSE = 3252, p < 0.001. Age interacted with eye movement task as well, F (1, 61) =
5.4, MSE = 915, p =0.023. Older adults’ eye movements slowed down more from the
prosaccade to the antisaccade conditions (Ms = 287 and 390 ms, respectively) than the
eye movements of young adults (Ms = 223 and 301 ms, respectively) indicating that they

had more difficulty than young adults producing antisaccades.

500
4507777
400-7 7 77
350—7 77 777
300777
25077

 

,SI‘E’rl BS} - , , , , -. , W $0093.31 , , ,1

1:] Pro —i
[:1 Anti i
I Anti _ErJ

 

Latency (msec)

 

 

 

 

 

 

 

Young Older Young Older
Age Group

 

 

 

Figure 3. Mean latency of the initial eye movement in the correct direction from the
antisaccade and prosaccade conditions and for incorrect eye movements in the
antisaccade condition by RSI, short or long, and age group, young or old. The mean
saccade latency of incorrect antisaccades for young adults in the short RSI condition
excluded 2 young subjects that had fewer than 7 saccade direction errors on which to
base a subject mean. Error bars are equal to 1 standard error. Anti— - Antisaccade, Pro =
Prosaccade, Anti_ Er= Incorrect eye movements in the antisaccade condition.

A possible explanation of the larger difference between antisaccade and

prosaccade latencies for the older adults is that it is the result of overall slower responses

38

by the older adults. To address this possibility, the difference between each individuals’
antisaccade and prosaccade latencies was converted to a proportion by dividing the
difference by the prosaccade latency. Indeed, when saccade latency differences were
considered as a proportion of saccade latency, the age difference was eliminated, F < 1.
Both young and older adults were 36% slower to initiate antisaccades than prosaccades.

Surprisingly, saccade latencies were slower in the short RSI condition than in the
long RSI condition, F (1, 61) = 19.3, MSE = 3252, p < 0.001. As was found in the
accuracy data, there was a marginally significant interaction of eye movement task and
RS1 indicating that the difference between the short and long RSI conditions was a bit
larger in the antisaccade condition (Ms = 372 and 319 ms, respectively) than in the
prosaccade condition (Ms = 272 and 238 ms, respectively), F (1, 61) = 3.4, MSE = 915, p
= 0.070. The effect of RSI on saccade latencies was in the opposite direction of the effect
found in the spatial Stroop task where reaction times were faster in the short than long i
RSI condition. In fact, these latency results taken together with the accuracy data suggest
that the difference between the short and long RSI conditions may be due to a
speed/accuracy tradeoff; in the short RSI condition saccades were slower, but more
accurate, than saccades in the long RSI condition.

Reflexive saccades (correct prosaflades and incorrectfiantisaccades). Saccade
direction errors in the antisaccade condition occurred when the participant looked toward
the peripheral onset before looking at the opposite box. If these error saccades were
instances when the reflexive response was made instead of a more controlled response,
then error saccades should have been faster than antisaccades. In fact, the error saccade

latencies should have been as fast as prosaccade latencies because prosaccades were also

39

saccades toward a peripheral onset. The mean saccade latencies for incorrect responses in
the antisaccade condition were compared to the saccade latencies of prosaccades. As can
be seen in Figure 3, incorrect antisaccades were faster than prosaccades, F (1, 61) = 58.5,
MSE = 782, p < 0.001. The eye movement condition did not interact with any other
variable.

_T_arget Identification Responsg The mean accuracy rates and reaction times for
target arrow identification responses are presented in Table 3. The identification accuracy
rate analysis mirrored the saccade latency analysis. In conditions that resulted in slower
eye movement latencies, target identification accuracy rates were lower. This is not
surprising because if the eyes moved slowly the participant would have had less time to
view the target stimulus, the arrow. Older adults did worse than young adults (Ms = 87.1
and 96.4%, respectively), F (l, 61) = 17.8, MSE = 0.01554, p < 0.001, and antisaccade
performance was worse than prosaccade performance (Ms = 89.1 and 94.4%,
respectively), F( 1, 61) = 21.2, MSE = 0.00421, p < 0.001. In addition, older adults
showed a greater decline in identification accuracy from the prosaccade to the
antisaccade condition compared to young adults, F (l, 61) = 4.9, MSE = 0.00421, p =
0.031. A main effect of RSI indicated that in the short RSI condition (the slower eye
movement condition) participants were more likely to incorrectly identify the direction of
the arrow than in the long RSI condition (Ms = 89.0 and 94.5%, respectively), F (1, 61)
= 6.3, MSE = 0.0155, p = 0.015. This variable did not interact with any other variable,

however, Fs < 1.8.

40

Accuracy Reaction Time

 

Pro Anti Pro Anti
Young Adults
Short RSI 97.2% 93.0% 1061 1053
Long RSI 98.3% 97.0% 1033 1023
Older Adults
Short RSI 87.3% 78.5% 1264 1311
Long RSI 94.7% 88.0% 1243 1252

Table 3. Experiment 1 Target Identification Responses

Mean button press reaction times were also examined to deterrrrine if they were
affected by RSI. There was no effect of RSI and this variable did not interact with age or
eye task, Fs < 2.3. Older adults responded more slowly than young adults (Ms = 1268 and
1042 ms, respectively), F (1, 61) = 64.9, MSE = 25403, p < 0.001. Age interacted with
eye task as well indicating that although there was little difference in the button press
reaction times of young adults between the prosaccade and antisaccade conditions, the
older adults were slower to respond correctly in the antisaccade condition than in the
prosaccade condition, F (1, 61) = 8.0, MSE = 1401, p = 0.006. When performing
optimally young and older adults had their eyes at the location of the target arrow when it
onset and their button press accuracy rates and reaction times were not affected by the
previous eye movement task. Only on trials where the participant took longer to look at
the target location, and perhaps did not view the target long enough for discrimination,
were the button press reaction times and accuracy rates disrupted. This situation was
more likely for older adults in the antisaccade condition.

To explore this suggestions two additional analysis were conducted. In the first,
each subj ects' mean antisaccade latency and mean button press reaction time were

correlated. Because of differences between the age groups and RS1 conditions in mean

41

saccade latencies the correlations were calculated separately for each of the four groups.
The correlations ranged from .43 to .66 confirming the suggestion that performance on
the target identification task was related to how quickly an individual could move their
eyes to look at the target. The second analysis was made of target identification accuracy
when the initial eye movement was a saccade direction error because in these conditions
participants should have been less likely to view the target and so target identification
accuracy would be reduced compared to trials on which a correct eye movement had been
made. Only the antisaccade conditions was examined because enough saccade direction
errors did not occur in the prosaccade conditions to compute a reliable accuracy measure.
Target identification accuracy was lower and target identification reaction times were
slower following an initial saccade direction error than following a correct initial saccade
supporting the idea that target identification was affected by the time needed by the
participant to move their eyes to the correct location (Ms = 83.9% and 1178 ms for young
adults in the short RSI condition, 91.7% and 1083 ms for young adults in the long RSI
condition, 65.4% and 1463 ms for older adults in the short RSI condition, and 76.1% and
1455 ms for older adults in the long RSI condition).
2.2.3 Cumulative Response Distributions

In support of their view that Stroop interference results from a failure to maintain
the current goal, De J ong et al. (1999) presented cumulative response distributions of
reaction times by item type and RS1 conditions. The cumulative response distribution is
constructed by ordering each subject’s reaction times in a particular condition and then
dividing the ordered reaction times into reaction time bins containing equal proportions

of responses (e. g., 10%). Then each subjects' average reaction time for each bin by item

42

type is averaged and plotted. The plot illustrates where the differences between the item
types is localized (or not) on the distribution. Figures 4a and 4b are an estimate of De

J ong et al.’s (1999) data from young adults in the short and long RSI conditions,
respectively. In the short RSI condition, across the entire distribution there is little
difference between the congruent and incongruent conditions. However, in the long RSI
condition the difference between congruent and incongruent reaction times becomes
larger as reaction times become slower. De J ong and colleagues took this pattern as
support for a goal neglect account of the interference effect. On trials with sufficient goal
activation for immediate task schema selection (like those in the short RSI condition and
the fast end of the long RSI condition distribution) reaction times were fast for both
congruent and incongruent trials and an interference effect was not indicated. On trials
with reduced goal activation (the slow end of the long RSI condition distribution),
incongruent trials were slowed to a greater extent than congruent trials and a large
interference effect was observed. Cumulative response distributions were calculated and
graphed for both the spatial Stroop task and the eye movement tasks.

Spatial Stroop task. The cumulative response distributions from this experiment
for the young and older adults are presented in Figures 5a-5d. The cumulative response
distributions for the young adults are similar to those reported by De J ong et al. (1999). In
the short RSI condition, the difference between the congruent and incongruent reaction
times was small across the entire distribution whereas in the long RSI condition the
difference between the two curves increased through the slower end of the distribution.
Given that the interference effects for young and old in the long and short RSI conditions

were not statistically different, it was not surprising that the cumulative response

43

 

 

 

 

 

4a _ Short RSI 4b Long R31
9 0.85 _ 20.85 ‘
3 3 “
_I=_ _ E _
g: 0.65 - 3.0.65
5 g YA-lnc 5 1
D 7 YA-Con U
2? 0 45 - O E045 -
a _ if 7
5, 0.25 $0.25
8' ‘ 8' ‘
:r _ 3 _
0 05 0.05
200 500 800 1100 200 500 800 1100
Reaction Time Reaction Time (ms)

 

 

 

 

Figures 4a and 4b. Estimates of the cumulative response distributions reported by De
Jong et al. (1999) for young adults in short (Figure 4a) and long RSI (Figure 4b)
conditions of a spatial Stroop task. Incong. = incongruent, Cong. = congruent.
distributions for the older adults displayed the same pattern. These qualitative
assessments were confirmed by a mixed AN OVA with item type and bin as within
subjects variables and age group and RSI as between subjects variables. As De J ong et al.
(1999) found, there was a significant bin by RSI by item type interaction, F (9, 612) =
3.1, MSE = 821, p= 0.001 , that indicated that the difference between reaction times to
congruent and incongruent stimuli was greater at the slow end of the distributions in the
long RSI condition than at both the fast end of the long RSI distributions and for the all
bins of the short RSI distributions. This finding suggested that the incorrect word reading
response only interfered when responses were slow. When responses were fast,
presumably because attention was tightly focused on the task of location naming, the

interference effect was eliminated. In addition, bin interacted with age group, F (9, 612) =

5.74, MSE = 821, p < 0.001, indicating that the difference in mean reaction time between

 

 

 

 

 

 

 

 

 

 

55‘ Short RSI - Young Adults 5b Long RSI- Young Adults
“i 7;.“- ‘
9 0785 j €0.85
E _ s. .
g 0.65 30.65
5» 5 A
o g ‘
Z"? 0 45 - OCong. §045 .
3' @ Incong. ' r
E 0.25 €0.25 ‘
2' g
0.05 i __ 0.05
200 500 300 1100 200 500 800 1100
Reaction Time Reaction Time (ms)
50 Short RSI- Older Adults 5d Long R31- Older Adults

9
on
or

.0
or
or

  
 

4 DCong.
[:1 Incong.

.0
N
or

.O
N
0'1
uonnqgrrsra eArrelnurnQ
O
A
01

uorrnqursrd emeinwno
O
A
01

 

.0
o
01

J

 

 

0.05

 

T_v

200 500 800 110
Reaction Time (ms)

71—~——_—;—.?..—». ——».__q *7 ..—

200 500 800 1100
Reaction Time

 

 

 

 

Figures 5a-d. Cumulative response distributions for the spatial Stroop task in Experiment
1 for young adults in the short RSI (Figure 5a) and long RSI (Figure 5b) conditions and
older adults in the short RSI (Figure 5c) and long RSI (Figure 5d) conditions. Incong. =
incongruent, Cong. = congruent.

young and older adults perfomrance was larger at the slow end of the distributions for
both the short and the long RSI conditions, but age did not enter into any other

interactions with bin, Fs <1 . l.

45

Antisaccade task. The cumulative response distributions for correct saccade
latencies can be seen in Figures 6a—6d. These distributions divided the data into only 5
bins because fewer trials were completed in the eye movement tasks. In addition, the data
from 3 young and 3 older adults in the long RSI condition and 2 older adults in the short
RSI condition were excluded fiom this analysis because they had fewer than 50 correct
prosaccades or antisaccades leaving less than 10 observations to go into each bin. The
shapes of these curves are markedly different from those for the Stroop task. Not
surprisingly, in both the short and the long RSI conditions there are large differences
between antisaccade and prosaccade latencies reflecting the difference between
generating a voluntary and a reflexive saccade. Unlike the Stroop data, however, these
differences are larger in the short RSI condition than in the long RSI condition suggesting
that the interference effect was larger in the short than in the long RSI condition, although
this difference is difficult to interpret because of the speed/accuracy trade-off associated
with the RSI manipulation. In addition, the curves in the long RSI condition appear to be
parallel suggesting that the in this condition interference was as likely to occur when
attention was tightly focused (fast end of the distribution) as when it was not (slow end of
the distribution). In contrast, in the short RSI condition the antisaccade and prosaccade
conditions diverge at the slow end of the distribution as they did in the long RSI
condition of the spatial Stroop task suggesting, if anything, that goal neglect was
occurring in the short RSI condition of the antisaccade task. These observations were
confirmed by the mixed ANOVA. Bin interacted with RSI, F (4, 212) = 4.8, MSE = 497,
p = 0.001, and bin interacted with eye movement task, F (4, 212) = 3.6, MSE = 497, p =

0.007. Bin also interacted with age group, F (4, 212) = 17.8, MSE = 497, p < 0.001, to

46

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

63 Short RSI - Young Adults 5b Long RSI - Young Adults

0 0.9 00.9

g 7 s -

a 0.7 §0.7

2’. 1:.

2' “ < '

g 0.5 30.5

g: ‘i 0 Pro. 2. 4

:35 0.3 0 Anti. 30.3

5. Eu _.

6' ‘ o

3 0.1 3 0.1
O 300 600 0 300 600

Latency (ms) Latency (ms)
5C Short RSI - Older Adults 6d Long RSI - Older Adults

'1 1i

O 0.9 00.9

c _ s .4

307 507

2’. i 2’. '

2. _, 2' u

t'D t0

:. _ a —

§= 0.3 D Pro; gas

5- E Anti. a, d

8' ‘l o

3 0.1 3 0.1
0 300 500 0 300 600

Latency (ms) Latency (ms)

 

Figures 6a-d. Cumulative response distributions for the antisaccade task in Experiment 1
for young adults in the short RSI (Figure 6a) and long RSI (Figure 6b) conditions and
older adults in the short RSI (Figure 6c) and long RSI (Figure 6d) conditions. Anti =

antisaccade, Pro = prosaccade.

indicate that older adults’ distributions were more spread out than young adults’

distributions. Bin did not enter into any hi glrer order interactions, Fs < 1.3.

 

 

2.2.4 Relationship between Measures of Interference

In this experiment young and older adults were compared on two interference
tasks. Antisaccade error rates and the difference between antisaccade and prosaccade
latencies are two measures of the interference between a prepotent and a voluntary
response. In the Stroop task, the interference effect was the difference between generating
a response when the meaning of the word indicates the alternate response compared to
when the correct response and the word meaning indicate the same response. (Note: This
interference measure contains a facilitation component as well, but it is assumed that the
facilitation component of Stroop performance would be unaffected by this manipulation.)
Are individuals who are more likely to have difficulty generating an antisaccade also
more likely to have trouble responding to incongruent trials compared to congruent trials
in the Stroop task?

To answer this question, two sets of correlations were done: one comparing the
interference effect (incongruent trial reaction times — congruent trial reaction times) with
antisaccade saccade direction accuracy rates and another set comparing them with the
difference between correct antisaccade and prosaccade latencies. Because antisaccade
accuracy rates varied by age group and RSI, correlating accuracy with Stroop
interference across these groups may lead to significant correlations caused by group
differences, rather than a true relationship between the two measures. To avoid this
potential problem, the correlations were done separately for each age group by RSI. Only
one of the correlations was significant at the p < 0.05 level with a one-tailed test, for the
young adults in the long RSI condition, the Stroop interference effect was correlated with

antisaccade accuracy rate, r = -0.47. On examination of the scatter plots, one outlier

48

 

seemed to be driving this correlation. With this person’s data removed the correlation was
eliminated, r = 0.05. The interference effect in the Stroop task was unrelated to
performance in the antisaccade task suggesting that these two measures reflect different
types of processing. In addition, this finding is consistent with the observation that older
adults had more difficulty in the antisaccade task compared to young adults, but that the

interference effect in the Stroop task was not affected by age.
2.3 Discussion

In this experiment the delay between the end of trial It and the beginning of trial
n+1 was manipulated between subjects in an antisaccade and a spatial Stroop task. It was
expected that in both tasks goal neglect would be more apparent in the long RSI
condition than in the short RSI condition. In addition, if older adults are more susceptible
to goal neglect than young adults, there should be a larger age difference in performance
under conditions that increase goal neglect, i.e., in the long RSI conditions.

The RSI manipulation had the predicted effect in the spatial Stroop task, but not
in the antisaccade task. Replicating the findings of De J ong et al. (1999) in the spatial
Stroop task, a larger interference effect was found in the long compared to the short RSI
condition. In the antisaccade task, although error rates were higher in the long than in the
short RSI condition, this effect was tempered by a speed/accuracy trade-off. Unlike the
spatial Stroop task, saccade latencies were faster in the long RSI condition than in the
short RSI condition making it difficult to determine whether the increase in error rates in
the long RSI condition was due to an increased likelihood of forgetting the current task
goal or, alternately, to a lower threshold for initiating a response that led to faster reaction

times and an increase in error rates.

49

Were age differences in performance modulated by the RSI manipulation, as a
goal neglect account of interference effects would suggest? In the Stroop task, where the
RSI manipulation had the intended effect, there was not a significant difference between
the interference effects of older and young adults when difference scores were analyzed
and when those scores were considered as a proportion of congruent reaction times.
Counter to the prediction, older adults were not slowed to a greater extent when the
likelihood of goal neglect was higher in the long RSI condition.

In the eye movement task, older adults made more saccade direction errors in the
antisaccade condition than young adults for both the short and long RSI conditions. This
finding confirms previous research indicating that older adults have more difficulty
correctly performing antisaccades than young adults (Butler et al., 1999, Olincy et al.,
1998, Nieuwenhuis et al., 2000). In addition, the difference between antisaccade and
prosaccade latencies was greater for older adults than young adults, but not when that
difference was considered as a proportion of prosaccade latency implying that a larger
antisaccade-prosaccade interference effect for older adults may be due to general slowing
rather than an inhibitory deficit or goal neglect.

Although the proportion analysis did not indicate an age difference in the
antisaccade-prosaccade interference effect when age-related slowing was controlled, the
larger difference between latencies represents a failure to replicate the saccade latency
findings from Butler et al. (1999) that indicated the size of the antisaccade-prosaccade
interference effect was the same for young and older adults. This discrepancy may be due
to a difference between the two studies in target presentation. Nieuwenhuis et al. (2000)

found that older adults will use target presentation as a means of visually triggering their

50

correct antisaccades when that information indicates the correct response, but when the
target location is not the only visual change occurring (i.e., the changes on the screen can
not trigger the correct antisaccade) older adults' antisaccade-prosaccade interference is
greater than that of young adults. Butler et al. (1999) used conditions that allowed older
adults to rely on visual triggers to generate antisaccades, but this experiment did not.
Because older adults could not rely on target presentation to visually trigger their
antisaccades they had to program the correct antisaccade voluntarily and this slower
process was affected to a greater extent by the slower responding of older adults.

Neither of these age differences in performance was moderated by the RSI. Older
and younger adults were affected in the same way by the RSI manipulation, responding
more quickly and less accurately in the long RSI compared to the short RSI condition.

In neither the spatial Stroop nor in the antisaccade task did older adults show
more interference in performance as a result of a long compared to a short RSI. This
finding suggests that goal neglect may not account for older adults’ impaired
performance on interference tasks. However, this conclusion hinges on the assumption
that maintaining a goal across a long interval is more difficult than maintaining a goal
across a short interval. This issue is considered below.

It is puzzling that the RSI manipulation had different effects in the spatial Stroop
and eye movement tasks. In the spatial Stroop task both young and older adults had more
difficulty responding under interference conditions in the long RSI condition than in the
short RSI condition, but for the eye movement task the manipulation of RSI did not lead

to a pattern of data that suggested goal neglect in either young or older adults.

51

One potential explanation is that the task structures of the spatial Stroop and
antisaccade tasks differ in an important way. This idea is supported by the lack of
correlation between antisaccade error rates and Stroop interference. In the spatial Stroop
task, optimal performance depends on the ability to ignore the meaning of the word and
prevent the response associated with the word meaning fiom interfering with
performance. The meaning of the word is irrelevant to the correct response. The situation
is different in the antisaccade task. Although participants are told not to look at the
peripheral onset, they must code the location of that onset in order to know the correct
direction to move their eyes. The location of the peripheral onset must be explicitly
attended on every trial for the correct response to be programmed. Attending to the
location that is in the direction of the incorrect response, may prime the incorrect task
schema leading to a need for a controlled inhibitory mechanism to be instantiated for the
correct response to be generated. In essence, attending to a location is presumed to be
equivalent to activating the task schema to respond to that location. The premotor theory
of attention posits that, indeed, the processes associated with attentional shifts are the
same as the processes required for the programming of a response to a particular location
(Rizzolatti, Riggio, & Sheliga, 1994; Craighero, F adiga, Rizzolatti, & Umilta, 1999).

According to the description above, interference can arise from two points in
processing. If the incorrect task schema is accessed because it overlaps with the correct
task schema the incorrect task schema must be inhibited in a controlled manner. If the
early stages of the incorrect and correct task schemas do not overlap, as in the spatial
Stroop task, controlled inhibition of the incorrect task schema is not required. Automatic

processing of the identity of the word was unnecessary for the determination of the

52

location of the word and thus in early stages of stimulus processing there was no overlap
between the two task schemas associated with the stimuli. Therefore, in the spatial Stroop
task a controlled attentional mechanism is not required. Interference occurs when the
current task goal is not active enough to allow the task to be executed and more
interference occurs when the task goal is harder to maintain, e.g., across a long RSI.

Another potential explanation for the difference in the effect of the RSI
manipulation on performance in the Stroop and antisaccade tasks comes from studies of
foreperiod effects. Although the RSI manipulation was used as a means of varying the
likelihood of goal neglect it could also be seen as a manipulation of the time provided to
prepare for the stimulus to occur, the preparation interval or foreperiod. In foreperiod
studies, a warning signal is presented sometime before the imperative stimulus and serves
to alert the participant to the impending presentation of the stimulus. Because the warning
signal does not provide any information that is necessary for performing the task,
systematic variation in reaction times following changes in the length of the foreperiod
are said to result from nonspecific preparation occruring dining the interval. In studies
that vary foreperiod across blocks, short preparation intervals result in faster reaction
times to the imperative stimulus than longer preparation intervals (N ierrri & Naatanen,
1981)

An RSI manipulation can be seen as a variation of the preparation interval if the
response on the previous trial (or the onset of the delay screen) is considered to be the
warning signal for the presentation of the stimulus on the current trial. This mapping
makes the prediction that when the RSI (or preparation interval) is short, reaction times

will be faster compared to when the RSI is long. In this spatial Stroop task and in that of

53

De J ong et al. (1999), the RSI and foreperiod could be viewed as the same interval and, in
fact, the difference in reaction times between the RSI conditions is consistent with the
foreperiod literature.

In the eye movement tasks, however, the foreperiod and RS1 were not confounded
though. In both the short and the long RSI conditions a second warning stimulus was
presented before the onset occurred. This warning stimulus was the offset of the fixation
point 200 ms before the peripheral onset, a time interval that is about 75 ms longer than
the interval that leads to optimal levels of preparation in choice reaction time tasks
(Bertelson, 1967). In this task, RSI and foreperiod are dissociated by the presentation of
the second warning stimulus, thus the failure to find the predicted effect of RSI may be
due to an unconfounding of foreperiod and RS1.

The distinction between foreperiod and RS1 may not be an important one for a
theory of goal neglect. Both the historically more accepted view, that foreperiod effects
are the result of expectancy and time uncertainty (N iemi & Naatanen, 1981), and the
more recent argument, that foreperiod effects are controlled by the parameters of trace
conditioning (Los & Van Den Heuvel, 2001), could accommodate goal activation as part
of the nonspecific preparation associated with foreperiod processing. If manipulating the
RSI is equivalent to manipulating the foreperiod, with goal activation being part of the
nonspecific preparation processes, then in the eye movement task the warning signal 200
ms before the trial began may have acted to reactivate the task goal eliminating any
differences between the RSI conditions in goal activation. If this is indeed the case, it also
suggests that reactivation is not more difficult for older adults. The relationship between

goal neglect and foreperiod effects remains to be worked out.

54

In Experiment 2, a different method was used to manipulate goal neglect in the
antisaccade task. It was expected that the activation of the current goal would be
compromised by requiring participants to switch often between the prepotent and
controlled tasks. In addition, the difficulty of suppressing the reflexive response was
manipulated by using two different types of direction cues, a peripheral onset and a

centrally presented arrow cue.

55

3 Experiment 2

In Experiment 1, the likelihood of goal neglect was manipulated by varying the
RSI. The impact of this manipulation on performance in the antisaccade task was
unexpected and did not support the prediction that goal neglect would be more frequent
following a long RSI than following a short RSI. In this experiment, a different method,
task switching, was employed to observe the effects of reduced goal activation on young
and older adults’ performance on the antisaccade task. In addition, the influence of the
inhibitory mechanism on task performance was examined by using different cues; in the

antisaccade condition one required more inhibition than the other.
3.1 Task switching

Task switching is a condition in which participants must switch between different
tasks within a single block of trials. These task switches may be cued by an instruction
proceeding the stimulus, by the type of stimulus itself, or by the sequence of trials (e. g.,
switch after every second trial). J ersild (1927) was the first to report that performance of
a task within a mixed task block was worse than performance of the same task in single
task block. This effect has been referred to as mixing cost (e. g., Meiran, Gotler, &
Perlman, 2001), alternation cost (e. g., Pashler, 2000), or general switch cost (e. g., Kray &
Lindenberger, 2000). One processing difference between single and mixed task blocks is
the need to activate and switch between different task goals across the mixed task blocks
compared to maintaining only one task goal in the single task blocks. Because
maintenance of task goals requires effort, when there are more goals to be maintained the
likelihood that the current goal will not be active enough to access the correct task set is

increased; the current goal is more likely to be temporarily neglected.

56

Mixing tasks should make more difficult to maintain the current task goal because
more task goals must be maintained at least at a minimal level of activation across the
experiment. Not only is performance slowed in mixed block conditions, but it is slowed
to a greater extent when there is a switch fi'om one task to another. Switch costs are the
difference between performance on trials where the task was changed from the previous
trial, switch trials, to performance on trials where the same task was performed on the
previous trial, nonswitch trials. Typically, performance is slower and more error prone on
switch trials than on nonswitch trials, particularly when short task cue-to-response
stimulus intervals are used. One explanation of switch costs is that they represent the time
needed to reconfigure the task set for the new task to be performed on the switch trial; on
the nonswitch trial task set reconfiguration does not need to be performed (Rogers &
Monsell, 1995). Switch costs are reduced when more time is given to prepare for the
upcoming task switch but significant switch costs (called residual switch costs) are
usually still obtained (e. g., Sudevan & Taylor, 1987). Rogers and Monsell (1995) found
that switch costs were reduced as the preparation interval was increased from 150 to 600
ms, but residual switch costs were not reduced when the interval was further increased
from 600 to 1200 ms.

One explanation of residual switch costs is derived from the goal neglect
hypothesis (De J ong et al., 1999, De Jong, 2000). According to this account, when the
activation of the goal to prepare for the upcoming task is high enough, individuals use the
preparation interval to reconfigure the task-set, but on the remaining trials, when the
preparation goal is not active enough, they fail to engage in advance preparation. This

results in fast responses on some trials for which preparation was engaged, but slower

57

responses on the remaining trials when the task-set reconfiguration has to occur after the
response stimulus is presented. It is the combination of these two types of performance
that leads to residual switch costs.

In task switching experiments, the increased difficulty of responding in mixed
task blocks compared to single task blocks is larger for older adults than young adults
(e.g., Kray & Lindenberger, 2000; Mayr, 2001). Mixing costs are greater in older adults
than in younger adults. Factors found to increase the magnitude of the age difference in
mixing costs include overlap of the response sets for the different tasks and the presence
of stimuli that are ambiguous as to the task that should be performed (Mayr, 2001).
Although rrrixing costs show substantial age differences, the age-related difference in
switch costs tend to be much smaller (e. g., Kray & Lindenberger, 2000; Mayr, 2001; for a
brief review see Meiran et al., 2001), and may even disappear under moderate levels of
practice (Kramer, Hahn, & Gopher, 1999). Even when no age differences in switch costs
are observed, large mixing costs are still present (Kray & Lindenberger, 2000). Taken
together, these findings suggest that with long task instruction to response stimulus
intervals older adults seem able to prepare for the current trial as well as young adults
(Kramer etal., 1999; Kray & Lindenberger, 2000; Meiran et al., 2001).

3.2 Inhibitory Difficulty

In addition to the task switching manipulation, this experiment also varied the
difficulty of the inhibitory task by varying the type of cue used to indicate the correct
looking direction. In the standard antisaccade condition, the inhibitory demands are great
because a peripheral onset indicates to the participant the direction not to look. As was

described in the introduction, peripheral onsets automatically attract attention and an eye

58

movement is programmed to that location (Abrams & J onides, 1988). Thus, in order to
perform the correct task, looking in the opposite direction, the incorrect eye movement
program must be suppressed.

To reduce the inhibitory demand associated with making a saccade in the opposite
direction of a cue, a different type of direction cue, a centrally presented arrow, was used.
Inhibiting responses to peripheral onsets in an antisaccade condition should be more
difficult for peripheral changes than for centrally presented direction cues. Evidence that
it is easier to suppress saccades to peripheral onsets than saccades programmed from
central arrow cues comes from a stop-signal paradigm. In this paradigm participants
respond to a stimulus, but are told that on some trials a signal will occur after the stimulus
and if it does they should try to withhold their response. The onset time of the signal
following the stimulus is varied. In order for saccade initiation to be stopped, the
presentation of a signal to abort the saccade must be presented sooner after programming
has begun for saccades to peripheral onsets than saccades in the direction indicated by a
central arrow (Logan & Irwin, 2000). It is more difficult to withhold a response
programmed from a peripheral onset than from a central arrow.

The instructions were the same for the peripheral and central cue conditions
ensuring that the goal structure of each condition was the same. In the prosaccade
condition, participants were instructed to look toward the peripheral onset or the box the
arrow was pointing at; in the antisaccade condition, the instructions were to look in the
opposite direction from that indicated by the direction cue. In the antisaccade condition,
centrally presented cues (unlike peripheral cues) should not elicit an automatic eye

movement program, reducing the inhibitory demands by eliminating the need to suppress

59

an incorrect eye movement program. Because the goal structure of the task with central
or peripheral cues is the same, any differences between the cue conditions in antisaccade
accuracy rates or the difference between correct antisaccade and prosaccade latencies can

be attributed to differences in inhibitory demand.
3.3 Predictions

In this experiment, participants performed the standard single task blocks of
antisaccade and prosaccade trials and then performed blocks of trials in which they
switched between the two types of eye movement responses. The mixed task condition
used a trial-by-trial task cuing procedure with a long task cue-to-response stimulus
interval of 2800 ms. The long preparation interval was used to ensure that young and
older adults were equally prepared to perform the upcoming task. The trial-by-trial cuing
procedure has produced substantial age differences in mixing costs while minimizing the
working memory demands of keeping track of how many trials of a particular task had
already been performed (Mayr, 2001). Participants either performed these tasks with
central arrow cues or peripheral onset cues.

Goal neglect was measured in two ways: (1) by comparing single and mixed task
block performance and (2) by comparing switch and nonswitch trial performance in the
mixed task condition. If failure to maintain the current goal at a high level of activation
accounts for lower saccade direction accuracy rates and slower saccade latencies in an
antisaccade condition, then correctly performing antisaccades should be more difficult in
the antisaccade condition when participants switch between antisaccade and prosaccade
trials, and when the switch conditions are compared to the single task condition in which

only one goal needs to be maintained.

60

The, firnctioning of the inhibitory mechanism was evaluated by comparing saccade
direction accuracy rates in the antisaccade condition and the difference between
antisaccade and prosaccade latencies in the peripheral and central arrow conditions.
Under both cue conditions in the antisaccade condition, participants were instructed to
look in the opposite direction of the direction specified by the cue, but when a central
arrow cue was used, inhibition of a competing response was not as difficult. Increased
difficulty performing antisaccades in the peripheral cue condition compared to the central
cue conditions will indicate the functioning of the inhibitory mechanism. Saccade
direction errors in the antisaccade condition with central cues would indicate a role of
goal neglect in this task.

Because the goal structures of prosaccade and antisaccade tasks were the same for
the central and the peripheral cue conditions, the increase in the likelihood of goal neglect
from the single to the mixed task blocks should be equivalent for the two conditions. That
is, if goal neglect accounts for difficulty in the antisaccade condition and the goal
structure of the central and peripheral cue conditions are the same, it is expected that the
lower saccade direction accuracy rates and slowed antisaccade latencies will be
equivalent for the both cue conditions.

As has been found previously with peripheral cues, it was predicted that older
adults would make more saccade direction errors in the single task antisaccade condition
and that the difference between their prosaccade and antisaccade latencies would be
larger than found for young adults. If these measures reflect the difficulty of the
inhibitory task, the effects should be smaller with central arrow cues than with peripheral

onsets. If these measures reflect susceptibility to goal neglect than the effects should be

61

larger in the mixed task condition compared to the single task condition. Performing
correct antisaccades should be more difficult in the mixed task blocks than in the single
task block for both age groups.

If age differences in performance on the antisaccade task are due to goal neglect,
older adults should make more saccade direction errors in the antisaccade condition and
be slowed more in producing correct antisaccade in both the central and peripheral onset
conditions than young adults. In addition, residual switch costs in both of these measures
should be greater for older adults than for young adults. If an inhibitory deficit is the
source of the difficulty experienced by older adults in an antisaccade task with peripheral
onsets, that deficit should not be apparent when the inhibitory demands of the task are
limited, i.e., when central arrow cues signal the correct direction to look. It is possible
that both an increased susceptibility to goal neglect and a decline in inhibitory

firnctioning will contribute to age-related impairments on this antisaccade task.

3.4 Method

3.4.1 Participants

Thirty-three young adults (Age range = 18 - 23 years) were recruited from
undergraduate psychology courses. Eight were paid $7/hour for their participation and the
remaining twenty-five were given partial course credit. One young subject in the central
cue condition was excluded because his eye movements could not be reliably tracked.

F orty-four community-dwelling older adults (Age range = 61 — 85 years old) were
recruited from the Lansing, MI area and paid $10/hour for their participation. The data
fi'om two older adults were excluded fi‘om the eye movement analyses because they were

not performing the task correctly. In addition, eye movement data werenot obtained from

62

five older adults who could not be reliably tracked because of droopy eyelids and/or
excessive movements during task performance. Young and older adults reporting
learning disabilities, psychoactive medications, or a history of brain trauma (including
stroke) were excluded from the sample. All participants were able to discriminate letters
that subtended .6° degrees of visual angle. Mean ages, vocabulary test scores, and
education levels of the participants in each group are presented in Table 4. The education
levels of the four groups did not differ by age group, F=2.87, or by cue condition, F < 1.
Older adults had higher vocabulary scores than young adults, F (1, 60) = 42.0, MSE =
14.8, p < 0.001, but vocabulary scores did not differ based on cue, F=2.93. Within each
age group, age did not vary based on one condition, F < l and F = 3.0, for young and

older age groups, respectively.

 

 

Young Old

Peripheral Cue ‘

Age (years) 20.9 71.9

Education (years) 14.8 15.6

Vocabulary (mean score) 27.8 34.7

n 16 16
Central Cue

Age (years) 21.1 74.9

Education (years) 15.1 15.9

Vocabulary (mean score) 30.1 35.7

n 16 16

 

Table 4. Experirrrent 2 Participant Characterisitics
3.4.2 Design

This experiment had a two age group (young or old) by two eye movement task
(prosaccade or antisaccade) by two cue type (central or peripheral) by two block type
(single or mixed task) design with age group and cue type as between subjects variables

and eye movement task and block type as within subjects variables.

63

3.4.3 Apparatus & Stimuli

The apparatus was the same as that described for Experiment 1. The size and
locations of the peripheral boxes and fixation cross were the same as was described in
Experiment 1. The central arrow cue subtended the same visual angles as the fixation
stimulus, 08°. The type of eye movement task was indicated by the word “TOWARD” in
the prosaccade condition or “OPPOSITE” in the antisaccade condition presented in the
center of the screen in 30 point Arial Unicode font. The horizontal visual angles
subtended by the words "TOWARD" and "OPPOSITE" were 4.2 and 4.7°, respectively,
and they both subtended .7° vertically. The arrow direction identification task used in
Experiment 1 was changed in this experiment to minimize confusion between the target
task and the eye movement response to the central arrows in the central cue condition.
Instead of pushing a button to indicate whether the target arrow was pointing up or down,
participants were asked to indicate the location of an opening in a circle, top or bottom,
which was presented inside the box at the target location. The circle subtended 0.6°
degrees of visual angle and the hole was 02° in size. A complete circle was displayed in
the box at the nontarget location.

Trials in each experimental block were presented in five strings of eight trials
each as described in Experiment 1. The single task blocks were performed before the
mixed task blocks because recent reports have indicated that older adults take much
longer to return to single task performance levels following task switching than young
adults (Mayr & Liebscher, 2001; Meiran et al., 2001) In the single block condition,
participants completed one block of 40 prosaccade trials followed by one block of 40

antisaccade trials. Each of the single task blocks had an equal number of trials from each

cue direction (left or right) by target (hole on top or bottom) combination. In the mixed
block condition 4 blocks of 40 trials each were presented for a total of 160 trials, 80
prosaccade and 80 antisaccade with an equal number of trials coming from each cue
direction by target combination.

The four mixed task blocks had different trial orders that were the same for each
participant. As was described in Experiment 1 each block of 40 trials was performed as
five strings of eight trials each. To limit the ability of participants to predict when a task
switch would occur within a string of trials, task run lengths of one, two, and three trials
in a row were used. Each string of eight trials contained one task run of one trial, two task
runs of two trials, and one task run of three trials resulting in three task switches in each
string. Two of the four trial orders were created by selecting ten strings of 8 trials that
were unique combinations of 1, 2, and 3 trial task runs. Then half of the strings were
assigned the prosaccade task as the first task to be performed and half were assigned the
antisaccade task. Two strings beginning with the prosaccade task and three strings
beginning with the antisaccade task were assigned to trial order 1 and the remaining five
strings were assigned to trial order 2. Trial orders 3 and 4 were created by using the same
orderings of trial runs but reversing the eye movement task that was performed first.
Thus, there were an equal number of one, two, and three trial runs in each eye movement
task and, across the 160 trials, ten strings began with the prosaccade condition and ten
strings began with the antisaccade condition. The trial orders were presented in a Latin
squares design and each trial order was performed as the first mixed task block for two

participants in each age group by cue type condition.

65

The practice blocks for the single task blocks contained an equal number of trials
from each cue direction by target combination presented in a random order. The practice
blocks for the mixed blocks contained task runs of one, two, or three trials fiom each eye
movement condition.

3.4.4 Materials

The acuity and vocabulary tests completed in Experiment 1 were also completed
by the participants in this experiment.
3.4.5 Procedure

The general procedures through the single task blocks were the same as described
in Experiment 1. Following the single task blocks participants completed the background
questionnaire and read the instructions for the mixed blocks. Participants completed two
practice blocks of mixed trials, a 12 trial block with opportunities to ask questions after
each trial and a 24 trial block run in three strings of eight trials with three task switches in
each string. Next, the participants completed four blocks of mixed trials with 40 trials in
each block.

The timing of events for each trial in a string for the prosaccade, antisaccade, and
mixed blocks is illustrated for the prosaccade condition with a central cue in Figure 7.
The experimenter initiated each string by pressing a button. The fixation display was
presented for 300 ms followed by the task instruction, the word “TOWARD” or
“OPPOSITE” presented in the center of the screen for 800 ms. The task instruction was
replaced by the fixation screen for an additional 1800 ms. The fixation cross was
removed from the screen 200 ms before the direction cue was presented for a task

instruction-to-stimulus interval of 2800 ms. For both cue conditions, the cue was

66

presented for 400 ms followed by the target display for 150 ms. The target and nontarget
stimuli were extinguished and both of the peripheral boxes remained empty until the
button press indicating the location of the opening in the circle was made. Following the
button press, the fixation point flashed for 500 ms at 50 ms intervals to indicate to the
participant to look back to the center of the screen to prepare for the next trial.

In the central one condition, the direction one was a single-headed arrow
presented at fixation. In the peripheral cue condition, the direction cue was a change of
the interior of one of the boxes from black to white. In addition, a double-headed arrow
was presented at fixation simultaneous with the presentation of the peripheral cue. The
double-headed arrow was presented in this condition to equate the peripheral cue
condition with the central cue condition on the need to disengage the oculomotor system
from fixation before the eye movement response can be initiated. Participants were told
to perform the task specified by the task instruction and that in the single task blocks the
task would always be the same, but in the mixed tasks blocks the task would change fiom
trial-to-trial. When the task instruction was “TOWARD”, participants were told to look
toward the box the arrow was pointing at in the central cue condition or to look toward
the box that changed in the peripheral cue condition. When the task instruction was
“OPPOSITE”, participants were told to look at the box opposite from the one the arrow
pointed at in the central cue condition or to look at the box opposite from the change in
the peripheral one condition. For both eye movement task conditions participants were to
indicate the location of the hole or gap in the circle by pressing the button on the top or
bottom. The experiment took approximately 1.5 hours for young and older adults to

complete.

67

 

Displayed until button
press response

 

 

 

 

 

 

 

 

 

 

 

 

Target Display

(150 ms) 0 + O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cue Display _>
(400 ms)

 

 

 

 

 

 

 

 

Time

 

 

 

 

 

Gap Display
(200 ms)

 

 

 

 

 

 

 

 

 

 

 

 

Fixation Display +
(1800 ms)

 

 

 

 

 

 

 

 

 

 

 

Task Instruction TOWARD
(800 ms)

 

 

 

 

 

 

 

 

 

 

 

Fixation Display
(300 ms)

 

 

 

Figure 7. Illustration of a prosaccade trial for the central cue condition of Experiment 2.
In the antisaccade condition, the task instruction was “OPPOSITE” and the target
appeared in the box opposite from the cue. In the peripheral cue condition, a double-
headed arrow was presented at fixation when the peripheral onset occurred. In the actual
displays stimuli were white presented on a black background.

68

3.4.6 Data analysis

Saccade direction accuracy, saccade latencies, button press accuracy and button
press reaction time measures were calculated in the same way as described for
Experiment 1. Because of a ceiling effect in the saccade accuracy measure of young
adults in the single task prosaccade condition in both cue conditions, but only for the
central cue condition for older adults, the omnibus AN OVA was not performed on this
measure. The analyses that were done are described below. Mean correct saccade
latencies, button press accuracy rates, and mean button press reaction times for each
block type and eye movement task were entered as repeated measures into a mixed
AN OVA with age group and one condition as between subjects variables. The power to
detect a moderate size main effect (f =.25, Cohen, 1977) with 32 participants in each cell
was good according to Cohen’s suggestion, power = .79. The power to detect an
interaction 16 participants in each cell was rather low, power = .55.

Switch costs. Because participants paused in performing the tasks at the end of
each string of trials, each string was treated separately in determining when the task
switched and did not. Within each string of 8 trials, trials were coded as nonswitch or
switch trials based on whether the same or a different task had been performed on the
previous trial. The first trial of each string was not included in either group. The mean
latencies and mean saccade direction accuracy rates were calculated for each type of trial
and submitted to a two age group (older or young) by two eye movement task
(prosaccade or antisaccade) by two cue condition (central or peripheral) mixed AN OVA
with age group and cue condition as between subjects variables and eye movement task

as the within subjects variable.

69

3.5 Results

Trials were excluded from the eye movement analyses if the calibration was
misaligned for that trial, if the participant was not looking at the center of the screen
when the onset occurred, or if a saccade was not made within 1000 ms of the beginning
of the trial. These criteria excluded 20.9% of the older adults’ data and 6.0% of the young
adults’ data. More data were excluded for the older adults than for the young adults, F (1 ,
60) = 35.8, MSE = 0.0393, p < 0.001, and more data were excluded in the peripheral cue
condition than in the central cue condition, F (1, 60) = 9.1, MSE = 0.0393, p = 0.004,
(17.2 and 9.7 % respectively). The larger number of trials excluded in the peripheral cue
condition was probably due to anticipatory eye movements related to directing attention
peripherally to detect the presentation of the cue. In the central cue condition, attention
was directed at fixation. There was no difference in percent of trials excluded based on
the eye movement task or block type and these variables did not enter into any
interactions, Fs < 2.1.

3.5.1 Saccade Direction Accuracy Rates

The saccade direction accuracy rates are presented in Figure 8 by cue condition,
age group and block type. Because the performance of young adults, but not older adults,
was at ceiling in the prosaccade single task blocks two separate analyses were performed
on this measure. In the first analysis, antisaccade accuracy rates were submitted to an age
group by block type by cue condition AN OVA to determine if older adults were less
accurate than young adults and whether this age difference was larger for the peripheral
cue condition compared to the central cue condition (indicating inhibitory difficulty)

and/or if the age difference was more pronounced in the mixed blocks than in the single

70

task blocks (indicating goal neglect). Older adults did have more difficulty making
correct antisaccades than young adults (Ms = 67.9 and 79.4%, respectively), F (l , 60) =
9.1, MSE = 0.0456, p = 0.004, and antisaccade accuracy was less accurate in the mixed
task blocks than in the single task blocks (Ms = 70.6 and 76.7%, respectively), F ( 1, 60) =
16.0, MSE = 0.0076, p < 0.001. Surprisingly, the main effect of cue was not significant
and age and cue type did not enter into any interactions, Fs < 1.6. Older adults had more
difficulty than young adults looking in the opposite direction in response to both a
peripheral onset and a centrally presented arrow cue, but this effect was not modulated by

either the type of cue or the need to switch between tasks.

 

Peripheral Central

     

 

  
   

 

 

 

 

 

 

 

 

 

 

 

E. “77% § Cl Pro-Sngl.
g $5,; § I Anti-Sngl.
8 x \
E § §

 

1' ‘ x a 1‘3
Young Older

 

 

 

Figure 8. Mean saccade direction accuracy rate of the initial saccade in the antisaccade
(gray bars) and prosaccade (white bars) conditions by block type, single task in solids or
mixed tasks in stripes, age group, young or older, and cue type, peripheral or central.
Error bars are equal to 1 standard error.

The second analysis compared performance in the prosaccade and antisaccade
conditions, but only in the mixed task blocks because in these conditions no participants

performed at ceiling. In the mixed blocks, older adults performed worse than young

adults (Ms = 77.5 and 83.8%, respectively), F (l, 60) = 6.9, MSE = 0.01893, p = 0.011,

71

particularly in the antisaccade condition (difference of 11% accuracy compared to 2% in
the prosaccade condition), F (1, 60) = 4.1, MSE = 0.01387, p = 0.048. In addition, the
difference between antisaccade and prosaccade accuracy was greater in the peripheral cue
condition (24%) than in the central cue condition (16%), F (1, 60) = 4.0, MSE = 0.01387,
p = 0.051. Although performance with peripheral cues was more disrupted in the
antisaccade condition this effect was of the same magnitude for young and older adults, F
< 1. The saccade accuracy data do not support either an inhibitory deficit or a goal
neglect account of aging.
3.5.2 Saccade Latencies

As can be seen in Figure 9, antisaccades were slower than prosaccades (Ms = 365
and 315 ms, respectively), F (1, 60) = 136, MSE = 1159, p < 0.001. The difference
between latencies in the two eye movement conditions was larger for older adults (58 ms)
compared to young adults (41 ms), F (l, 60) = 3.9, MSE = 1159, p = 0.052, and it was
larger in the peripheral cue condition (71 ms) than in the central cue condition (28 ms), F
(l, 60) = 25.3, MSE = 1159, p < 0.001. All three of these variables entered into an
interaction indicating that older adults were slowed down to a greater extent in the
antisaccade condition when they were responding based on a peripheral cue (Ms = 340
and 430 ms for older adults and Ms = 260 and 311 ms for the young adults, in the
prosaccade and antisaccade conditions, respectively) compared to when they were
responding based on a central cue, F (1, 60) = 6.4, MSE = 1159, p = 0.014, (Ms = 373
and 399 ms for older adults and Ms = 289 and 320 ms for young adults, respectively).
This finding supports the view that older adults have more difficulty suppressing

reflexive saccades than young adults.

72

To determine if the age difference in the size of the prosaccade and antisaccade
latency difference was greater for older adults than young adults because of the overall
slower responding of the older adults, the difference between the antisaccade and
prosaccade latencies of each subject were converted to a proportion based on the
subjects’ mean prosaccade latency. Different proportions were computed for the single
and mixed task conditions and submitted to a mixed ANOVA with block type as the
repeated measure and age group and cue type as the between subjects variables. Unlike
the results of the proportional analysis of Experiment 1, controlling for the slower
responding of the older adults did not eliminate the interaction although it was reduced, F
(l, 60) = 3.17, MSE = 0.0256, p = 0.080. The results of this analysis indicated that
antisaccades in the peripheral onset condition were initiated 21.6% slower than

prosaccades by young adults, but 28.2% slower by older adults.

 

 

 

 

Peripheral Central
600 El Pro-Sngl.
500 7 I Anti-Sngl.

Pro-Mix
I Anti-Mix

   
 

 

 

    

\‘a

 

Mean Latency (msec)
oo
o
o

 

 

 

 

 

Young Older Young Older

 

Figure 9. Mean latency of the initial eye movement in the correct direction from the
antisaccade and prosaccade conditions by block type, single or mixed task, age group,
young or older, and one type, peripheral or central. Error bars are equal to 1 standard
error. Anti = Antisaccade, Pro = Prosaccade, Sngl. = Single task, Mix = Mixed tasks.

73

The task switching manipulation had an unexpected result. Responses were faster
in the mixed task blocks than in the single task blocks (Ms = 329 and 351 ms,
respectively), F (l, 60) = 15.1, MSE = 2003, p < 0.001, and this difference was larger in
the peripheral cue condition (39 ms) than in the central cue condition (4 ms), F (1, 60) =
9.6, MSE = 2003, p = 0.003. Age did not enter into any interactions with block type, Fs <
1.7. A possible reason that no mixing costs were obtained is that the single task blocks
were performed before the mixed task blocks and extensive practice with the antisaccade
and prosaccade tasks during the mixed task blocks improved performance to a level
better than was observed in the single task conditions. Averaging across the poor
performance in the first mixed task blocks and the improved performance in the latter two
blocks may have resulted in mixed task mean saccade latencies that were slightly faster
than single task latencies. However, a comparison of single task performance with
performance in the first and last blocks of the mixed condition indicates that this is not
the case. In the prosaccade condition, young adults’ performance did improve slightly
from the first to the last mixed task block (Ms = 270 and 255 ms), but even the means in
the first mixed block were faster than the mean prosaccade latency in the single task
condition (M = 290 ms). In addition, for young adults performing antisaccades and older
adults performing both antisaccades and prosaccades there was no indication that
performance improved across the four mixed task blocks (For young adults Ms = 303 and
300 ms on antisaccade trials and for older adults Ms = 418 and 416 ms on antisaccade
and 343 and 344 on prosaccade trials in the first and fourth mixed task blocks).
Increasing the demands on maintaining the current goal did not differentially affect older

adults compared to young adults in the mixed task condition.

74

Much like the RSI manipulation, there is little evidence that having to switch
between the two tasks had any net effect on task performance. In the mixed task blocks,
saccade direction errors were more fi'equent, but that increase in error rate was
accompanied by a decrease in saccade latencies, a speed/accuracy trade-off. Switching
between the two eye movement tasks may not compromise the activation of the current
goal
3.5.3 Residual Switch Costs

In this analysis, the effect of trial type (switch or nonswitch) on saccade latency
was marginally significant, F (1, 60) = 3.0, MSE = 587, p = 0.088, but indicated,
surprisingly, a 5 ms advantage in saccade latency for switch compared to nonswitch trials
(Ms = 325 and 330 ms, respectively). The trial type did not interact with any other
variables, Fs < 1.5. When the saccade direction accuracy rates were considered, switch
trial performance was less accurate than nonswitch trial performance (Ms = 78.8 and
82.5%, respectively), F (1, 60) = 12.8, MSE = 0.0067, p = 0.001. A marginal interaction
of trial type with age suggested that the improvement in performance from switch to
nonswitch trials was smaller for older adults than for young adults, F (1, 60) = 3.7, MSE
= 0.0067, p = 0.059. The type of trial also affected saccade direction accuracy differently
for the two eye movement conditions as well, F ( 1, 60) = 6.8, MSE = 0.0042, p = 0.011.
Prosaccade performance improved 6% from switch to nonswitch trials, but antisaccade
performance only improved 1%. Trial type did not enter into any other significant

interactions, Fs < 2.6.

75

3.5.4 Reflexive saccades (correct prosaccades and incorrect antisaccades)

For the antisaccade condition saccade direction errors occurred when the
participant looked toward the peripheral onset before looking at the opposite box. In the
peripheral onset condition, these errors should be fast, reflexive saccade programs that
were not inhibited, in the central arrow condition these errors should be slower
intentional saccades. The mean saccade latencies for incorrect responses in the
antisaccade condition were compared to the saccade latencies of prosaccades. As can be
seen in Figure 10, incorrect antisaccades were faster than prosaccades, F(1, 59) = 26.0,
MSE = 1575, p < 0.001, and this effect was larger for the peripheral cue condition than
for the central cue conditions, F(1, 59) = 9.1, MSE = 1575, p = 0.004. Age also interacted
with type of eye movement and cue condition, F(1, 59) = 4.5, MSE = 1575, p = 0.038, to
indicate that for older adults the difference between looking direction error latencies and
prosaccade latencies was much larger in the peripheral cue condition (Ms = 280 and 340
ms, respectively) than in the central cue conditions (Ms = 365 and 373 ms, respectively),
but for young adults this difference was much smaller (Ms = 238 and 260 ms in the
peripheral cue condition and Ms = 275 and 288 ms in the central cue condition). Block
type did interact with cue condition, F (1, 59) = 4.9, MSE = 1995, p =0.031, and also
interacted with the type of eye movement, F (1, 59) = 40.1, MSE = 1040, p < 0.001, but
did not enter into any interactions with age group.

Prosaccade latencies for both older and young adults in the peripheral cue
conditions were slower than their antisaccade error responses. If the antisaccade error
response latencies reflect the time needed to program a reflexive saccade, this finding

suggests that older adults may have been using more controlled processing in the

76

prosaccade condition than young adults. For example, a larger proportion of young
adults’ prosaccades may have been reflexively programmed compared to older adults’
prosaccades. If this is true it actually suggests that the difference between prosaccade and
correct antisaccade latencies is underestimated to a larger degree for older adults than

young adults.

 

Peripheral Central

1] Pro-Sngl.
500 , IAnti_Er-Sngl.
Pro-Mix

400 4 I Anti_Er-Mix

 

600

 

 

300 7

200 7

 

Mean Latency (ms)

100 7

\

 

 

 

   

0_

 

 

Young Old Young Old

Figure 10. Mean latencies of the initial eye movement in the correct direction for the
prosaccade condition and for incorrect eye movements in the antisaccade condition by
cue, peripheral and central, block type, pure and mixed tasks, and age group, young and
old. Error bars are equal to 1 standard error. Pro = Prosaccade, Anti_Er = Incorrect eye
movements in the antisaccade condition, Sngl = Single task blocks, Mix = mixed task
blocks.
3.5.5 Target Identification Responses

The mean target identification accuracy rates and reaction times are presented in
Table 5. Young adults correctly identified the location of the gap in the circle 94.0% of
the time, while older adults only identified it correctly on 82.2% of the trials, F (1, 60) =
35.8, MSE = 0.025, p < 0.001, and target identification was better in the prosaccade

condition (M = 90.7%) than in the antisaccade condition (M = 85.5%), F (l, 60) = 29.1,

77

MSE = 0.0061, p < 0.001. The difference between identification accuracy in the
prosaccade and antisaccade conditions was greater for older adults than young adults
(differences of 8.8 and 1.9%, respectively, F (1, 60) = 12.8, MSE = 0.0061, p = 0.001.
These findings are not surprising because there were differences in mean saccade latency
between these conditions. In conditions where the eyes moved faster and more often in
the correct direction, participants had more time to view the target and therefore were
better at identifying it. There were also two interactions that barely missed the
conventional significance or level of p < 0.05. An eye movement task by block type
interaction, F (1 , 60) = 3.9, MSE = 0.0026, p = 0.054, indicating that prosaccade
performance was more disrupted when both tasks were performed together in the mixed
task blocks than antisaccade performance. An age by block type by cue interaction, F (1 ,
60) = 3.8, MSE = 0.0043, p = 0.057, suggests that whereas young adults’ target
identification accuracy in the peripheral and central cue conditions was similarly affected
by mixing the tasks together, older adults were more disrupted by mixing the tasks in the
central cue condition compared to disruption in the peripheral cue condition. This seems
to be because whereas older adults’ initial saccades in the central cue condition were
slower and less accurate in mixed compared to the single task blocks of trials, their initial
saccades in the peripheral cue conditions were faster and less accurate. When they moved
their eyes in the mixed blocks they were about equally likely to move them in the correct
direction for the two cue conditions, but they moved them more slowly in the central
condition than in the peripheral condition and therefore did not have as much time to

view the target in the central one condition.

78

The mean button press reaction times of young adults were faster than the
reaction times of older adults (Ms = 890 and 1195 ms, respectively), F (1, 60) = 61.5,
MSE = 96812, p < 0.001, and the targets were identified faster in the prosaccade than in
the antisaccade task (1020 and 1064 ms, respectively), F (1, 60) = 12.3, MSE = 10292, p
= 0.001. As with the button press accuracy data, the slowing of saccade latencies from the
prosaccade to the antisaccade condition was greater for older adults (81 ms) than young
adults (9 ms), F (1, 60) = 8.1, MSE = 10292, p = 0.006. The only effect of block type was
a three way interaction with age group and cue type, F (l, 60) = 5.1, MSE = 12923, p =
0.028, indicating that, like in the accuracy data, older adults were disrupted more by

switching between the tasks than young adults, but only in the central cue condition.

 

 

Accuracy Reaction Time
Pro Anti Pro Anti
Young Adults
Peripheral
Pure 0.970 0.949 873 875
Mixed 0.922 0.912 897 904
Central
Pure 0.976 0.942 890 888
Mixed 0.929 0.922 882 908
Older Adults
Peripheral
Pure 0.873 0.747 1202 1267
Mixed 0.855 0.760 1153 1251
Central
Pure 0.904 0.823 1098 1 169
Mixed 0.830 0.782 1164 1253

 

Table 5. Experiment 2 Target Identification Responses
3.6 Discussion

In this experiment the likelihood of goal neglect was manipulated using a task

switching procedure and the difficulty of the inhibitory task was varied by using direction

79

cues that differed in the difficulty of inhibiting a response to them. Although the effects
of these manipulations were not as dramatic as might have been expected, the saccade
latency data suggested that older adults had more difficulty inhibiting responses to
peripheral onsets than young adults even when the slower responding of older adults was
controlled. There was no evidence that they were more susceptible to goal neglect in the
antisaccade task than young adults.

In fact, faster reaction times coupled with higher error rates in the mixed blocks
compared to the single task blocks and on the switch trials compared to the nonswitch
trials suggest that antisaccade performance was affected minimally, if at all, by the task
switching demands. This result was quite surprising. Switching between the prosaccade
and antisaccade conditions has the characteristics that lead to large mixing costs and large
age differences in those costs, response-set overlap and stimulus ambiguity (Mayr, 2001).
There was complete response set overlap between the antisaccade and prosaccade
conditions; the tasks shared the same eye movement responses, to look to the left or to
the right. In addition, it was impossible to know which task was supposed to be
performed based on the direction cues because they were identical for the antisaccade and
prosaccade conditions. Therefore, it was expected that substantial mixing costs would be
obtained and that these costs would be larger for older adults than young adults.

The failure to find a consistent decline in performance under switch instructions
suggests that, although the mixed task blocks had the properties attributed to large mixing
costs, performing antisaccades and prosaccades under single task and mixed task
instructions put similar demands on information processing. It is tempting to say that the

preparation interval in the eye movement tasks was sufficient to allow complete

80

preparation for the upcoming task and that the small and inconsistent patterns in residual
switch costs were due to this long interval. This explanation does not account for why the
demands of switching between two tasks did not slow performance compared to the
single task conditions though. Even with complete preparation for the upcoming task in
the mixed task conditions, there should have still been some dual task conditions created
by the need to be prepared to perform a different task on an upcoming trial; performance
should have been slowed in the mixed task blocks compared to the single task blocks.

Another possibility is that the antisaccade condition may, even under single task
instructions, require a similar level of prosaccade goal activation as is needed in the
mixed task blocks. For the antisaccade condition, it is always the case that the location of
the direction cue, either where it onsets or where it is pointing to, needs to be encoded in
order for the correct response to be determined. In essence, the prosaccade response must
be encoded and rejected on every antisaccade trial under both single task and mixed task
conditions and so the requirement to be prepared to switch to performing prosaccades on
some trials may not increase the demands associated with antisaccade goal maintenance
on antisaccade switch trials.

Although this explanation seems quite sensible for the antisaccade condition it
does not explain why prosaccade performance was not more disruptedunder switch
instructions. There was a small residual switch cost associated with accuracy of
performance on prosaccade switch trials compared to nonswitch trials, but no mixing
costs in this condition. One might argue that the reflexive nature of the prosaccade
response precludes the need for goal maintenance and so prosaccade performance is

insensitive to the requirement to be prepared to switch to the antisaccade task. This

81

explanation is consistent with the lack of mixing costs in the peripheral onset prosaccade
condition and the small switch cost in accuracy, but it falls short because it predicts that
in the central cue condition when the prosaccade goal would have to be maintained there
would be significant mixing and residual switch costs, a result that was not obtained.

The explanation discussed above indicates that performance was hard in the
antisaccade condition for both the single and mixed task blocks because the stimuli and
initial steps in response coding were the same. Conversely, similarity of task sets has
been shown to decrease switch costs. When the tasks to be performed require attending
to the same dimension of a stimulus or when the response labels (but not response keys)
are shared, switch costs are reduced by about 80 ms compared to switching to a task that
does not share properties along these dimensions (Arrington, Altrnann, & Carr, 2001).
Substantial switch costs, on the order of 300 ms, are still observed. The antisaccade and
prosaccade conditions were similar because they required attending to the same
dimension of the stimuli and because they used the same response set (i.e., look left). In
addition, both tasks included a target identification component and a returning the eyes to
the center component. Perhaps because the antisaccade and prosaccade conditions shared
several operations and because long preparation intervals were used, residual switch costs
were eliminated. In addition, the reduction in switch costs associated with task similarity
might have equated single and mixed task block performance. Further research is
required to answer this question.

A third potential explanation for the failure to find an effect of either mixing the
tasks or switching to a new task is that the responses being performed on each trial should

not be categorized as a single task. For the purposes of this experiment, all the responses

82

associated with a single trial were considered to be part of a single task set, but it may be
the case that task switching occurred within each trial for both the single and mixed task
blocks. For example, after moving their eyes in the direction specified by the cue,
participants were also required to identify the target by making a button press and to
move their eyes back to the center of the screen to prepare for the next trial. Performing
each of these steps on a particular trial might have involved switching between task
schemas associated with different response types (i.e., button press for target
identification). J ersild (1927) found that when participants switched between two very
different operations (e. g., subtraction and opposite naming) mixing costs were eliminated.
Perhaps, mixing costs were not present in this experiment because on every trial three
tasks, two of which involved different operations (target identification vs. eye
movements), were performed. Likewise, residual switch costs may have been so
nriniscule because the target identification and preparation tasks preceded every eye
movement trial and so every eye movement trial was a switch trial. Further research is
required to determine if the component operations of a task must be considered as
separate tasks in a task switching paradigm.

A second unexpected finding from this experiment was the lack of an overall
effect of cue type, central or peripheral, on the eye movement measures. Typically eye
movement responses to peripheral onsets are initiated much faster than responses to
symbolic cues, like central arrows, because in the former case the responses are reflexive,
but in the latter the responses are controlled (e.g., Logan & Irwin, 2000). The lack of a
significant latency difference was probably due to a change made to the peripheral cue

procedure to equate it more closely with the central cue condition. In the central cue

83

condition the central arrow was presented at fixation to indicate the correct looking
direction. The effect of a stimulus at fixation on eye movement latencies has been widely
studied (e.g., Saslow, 1967) and, it is clear that, when the eyes are fixating a stimulus it
takes longer to initiate an eye movement response away from that stimulus, than when
the fixation stimulus disappears before the saccade is initiated. In part, this latency
difference results from the additional time needed by the oculomotor system to disengage
from fixation when there is a stimulus there (Kingstone & Klein, 1993). In order to
equate the peripheral and central cue conditions on the need to disengage from fixation
before saccade initiation, a noninformative double-headed arrow was presented at
fixation at the same time that the peripheral onset occurred.

The onset of a distractor at fixation in the peripheral cue condition did slow
saccade initiation in the peripheral cue condition; saccades in the peripheral single task
blocks of this experiment were 76 ms slower than saccades in the long RSI condition of
Experiment 1.2 Walker, Deubel, Schneider, and Findlay (1997) also found dramatic
increases in saccade latencies caused by the presentation of a stimulus at fixation
simultaneous with the onset of a peripheral saccade target. The slowing of prosaccades in
the peripheral onset condition actually resulted in no significant difference in prosaccade
latencies for the peripheral and central cue conditions. Although these latencies were the
same the direct comparison of these conditions was not warranted because saccades to
peripheral onsets and saccades in response to central arrows are mediated by different
processes. For example, the distance and direction parameters needed to execute a

saccade are programmed differently for peripheral onsets and symbolic cues. For

 

2 Although these two peripheral cue conditions differed slightly (there was no task cue in
Exp. 1), the difference is unlikely to account for slower responding in this experiment.

84

reflexive saccades, like those to peripheral onsets, the parameters are programmed
simultaneously whereas these parameters are specified sequentially in response to
symbolic cues, like a central arrow (Abrams & Jonides, 1988). In addition, reflexive
saccades are programmed in superior colliculus whereas saccades programmed using
symbolic cues require cortical involvement (Pierrot-Deseilligny, Rivaud, Gayrnard, Muri
& Verrnersch, 1995).

The presentation of the fixation cross simultaneous with the peripheral cue
disrupted reflexive saccade programming. However, fast incorrect antisaccades in the
peripheral onset condition and saccade direction accuracy rates of less than 75% suggest
that an incorrect reflexive saccade was being programmed in the antisaccade condition
and suppressing the reflexive response to the peripheral onsets was still quite difficult.

In fact, comparing the central and peripheral cue condition on saccade direction
errors indicated that the central cues might have elicited responses similar to the those
generated by peripheral cues. The saccade direction error rates were as high in the central
cue condition as they were in the peripheral cue condition. Although it might seem that a
central arrow would be easily reinterpreted under opposite looking direction instructions,
recent evidence suggests that attentional responses to these symbolic cues are automatic.
Central arrow cues elicit attentional shifts even when the direction of the arrow is
irrelevant to the current task and when attention is biased to be oriented to a different
location (Hommel, Pratt, Colzato, & Godijn, 2001). These automatic attentional shifts
seem to have caused saccade direction errors to be as frequent in the central cue condition
as in the peripheral one condition but did not result in as much slowing of the initiation of

antisaccades (28 vs. 71 ms in the central and peripheral conditions, respectively). In

85

addition, the amount of slowing observed varied by age. Older adults were slowed to the
same extent as young adults when making antisaccades inconsistent with the direction of
a central arrow, but older adults had more difficulty making antisaccades inconsistent
with a peripheral onset.

Although there were no differences between the cue types when performance was
collapsed across the prosaccade and antisaccade conditions, when prosaccade
performance was taken as a baseline, antisaccades were more difficult to make in
response to peripheral onsets than central arrows. Older adults had more difficulty
making antisaccades than young adults did in both cue conditions, but the effect was
larger in the peripheral onset condition. Because the goal structure of the central and
peripheral cue procedures was the same, the larger age difference in performance in the
antisaccade condition can be to attributed to a deficit in the ability to inhibit the prepotent

response.

86

4 General Discussion

4.1 Summary of Experimental Results

In two experiments the sources of difficulty experienced by older adults on tasks
that require suppressing a reflexive response were examined. In Experiment 1, the role of
goal activation was assessed by using short and long RSIs in the manner reported by De
Jong et al. (1999) on both a spatial Stroop and an antisaccade task. As was previously
found, the long RSI produced a larger interference effect in the performance of young
adults on a spatial Stroop task than the short RSI condition supporting the idea that the
goal of the current task is less accessible following a long RSI than a short RSI. This
pattern was the same in older adults. The goal neglect hypothesis had predicted that older
adults would be more susceptible to goal neglect and, as an extension, under conditions
that induced goal neglect in young adults, older adults would show larger interference
effects. This was not the case. Older adults were not more susceptible to goal neglect than
young adults.

In the antisaccade task, the RSI manipulation did not affect the performance of
older or younger adults in a way that was consistent with a goal neglect explanation.
Performance was more error prone in the long RSI condition, as the goal neglect
hypothesis predicted, but it was also faster, suggesting that the differences in accuracy
rates and latencies between the two conditions were due to a speed/accuracy trade-off
rather than an increase in goal neglect. The speed/accuracy trade-off in the antisaccade
task in Experiment 1 precludes any consideration of how goal neglect plays a role in
antisaccade task performance in both older and younger adults. It is unknown why the

RSI manipulation increased interference in the spatial Stroop task, but had no net effect

87

on performance in the antisaccade task. Possible explanations for this difference are
considered below.

In Experiment 2, a different manipulation, task switching, was employed to vary
the difficulty of goal maintenance, and thus, the likelihood of goal neglect. Participants
performed the prosaccade and antisaccade conditions in single task and mixed task
blocks. Although under some conditions mixing costs are not observed with task
switching (e.g., when the different tasks are distinguished by different stimuli (J ersild,
1927; Spector & Biederrnan, 1976)), typically, mixing costs are significant, particularly
when the response stimuli do not indicate the task to be performed as was the case with
the eye movement tasks used here (Mayr, 2001; for a review see Pashler, 2000). In
addition, large age differences in mixing costs are found when the response sets for the
two tasks are exactly the same, another condition present in Experiment 2. Although an
initial reading of the literature predicted large mixing costs, particularly for older adults,
they were not obtained. The failure to find a large effect of task switching may have been
due to task similarity reducing switch costs, although the size of reductions that have
been reported do not come close to eliminating the switch costs (Arrington et al., 2001).
Altemately, the different component operations that were required on each trial may have
reduced the effect of task switching (J ersild, 1927). It seems counterintuitive to predict
that making the task to be performed on each trial more complex should reduced the
effect of having to switch between the different tasks. As with the RSI manipulation the
task switching manipulation created a speed/accuracy trade-off. Saccade direction
accuracy was reduced for both the antisaccade and prosaccade conditions in the mixed

task blocks, but saccade latencies were also faster indicating no net effect of the task

88

switching manipulation. In addition, residual switch costs were also not observed. As in
Experiment 1, no evidence of goal neglect was found on the eye movement tasks using a
manipulation thought to make goal activation more difficult.

Experiment 2 also varied the difficulty of the inhibitory task while maintaining a
consistent goal structure by using different types of direction cues, a peripheral onset and
a central arrow. Because the task to be performed was the same in the different cue
conditions, goal neglect was equally likely in both. Although the goal structure was the
same for the peripheral onset and central arrow cue conditions performance in the
antisaccade condition was different for the two cue conditions. The difference between
antisaccade and prosaccade latencies was greater in the peripheral onset condition and
this effect was larger for older adults than for younger adults. When the inhibitory
demands were high, in the peripheral onset condition, older adults had more difficulty
than young adults, but when the role of inhibition was reduced, in the central arrow one
condition, age-related changes in the difference between antisaccade and prosaccade
performance were reduced as well. Age differences in performance on the antisaccade
task are related to an inhibitory deficit.

4.2 Implications for Theories of Cognitive Aging
4.2.1 Inhibitory Deficit Hypothesis

The results of Experiment 2 provided support for the inhibitory deficit hypothesis
of aging (Hasher & Zacks, 1988; Zacks & Hasher, 1994; Hasher et al., 1999). Older
adults had more difficulty performing antisaccades as measured by the slower responding
when the cue elicited a prepotent response in the wrong direction that needed to be

suppressed, than when the response was associated with a cue that was not prepotent. The

89

additional difficulty in performance with peripheral onsets can not be attributed to more
difficulty maintaining the current task goal, however, because the goal structure of the
two tasks was the same.

Older adults also had more difficulty than young adults performing antisaccades
in the central cue condition. In this condition, it can be assumed that attention was
automatically moved to a location consistent with the direction of the central arrow cue
(Hommel et al., 2001), the direction inconsistent with the required response. The
premotor theory of attention asserts that patterns of performance that indicate attentional
shifts are actually caused by the programming of a motor response to a particular location
(Rizzolatti et al., 1994, but see Klein & Pontefract, 1994). In this view, in order to have a
movement of attention in the central arrow condition, a saccade to that location must be
programmed. Older adults appear to have more difficulty recovering from the
inconsistent attentional shift than young adults do.

Ifboth peripheral and central arrow cues lead to automatic attentional shifts and,
according to premotor theory, to the programming of a saccade that is incorrect, then why
would the requirements for suppression be higher in the peripheral condition than in the
central arrow condition? As was described above, both types of cues may automatically
elicit an incorrect saccade program, but the types of saccades would be different.
Saccades consistent with the direction of an arrow, a symbolic cue, would still require the
involvement of cortical structures and would be considered intentionally generated,
whereas saccades toward a peripheral onset could be programmed in superior colliculus
(Pierrot-Deseilligny et al., 1995). In addition, suppression of a program for an intentional

saccade appears to be easier than the suppression of a program for a reflexive saccade

90

(Logan & Irwin, 2000). The inhibitory deficit of older adults has more of an impact when
a reflexive saccade program must be suppressed.

Older adults had more difficulty suppressing incorrect responses in an eye
movement task, but they did they not show a larger interference effect in the spatial
Stroop task for either RSI condition. Why would older adults show a pattern consistent
with an inhibitory deficit in one task, but not the other? This finding adds to a growing
literature indicating that the inhibitory deficits of older adults are task specific (Burke,
1997; Kramer et al., 1994; MacCrae & Abrams, 2001; McDowd, 1997). For example,
Connelly and Hasher (1993) found that different inhibitory tasks show different patterns
of age decrements. In a negative priming experiment they found that both young and
older adults demonstrated negative priming for a previously ignored location, but that
only young participants showed negative priming for the identity of a previously ignored
letter.

Many of the tasks that demonstrate equivalent inhibitory function for older and
younger adults require the inhibition of information at a particular spatial location (Zacks
& Hasher, 1997). When the spatial location of distracting information is predictable, age-
differences in the impact of the distracting information can be eliminated (e. g., Connelly
& Hasher, 1993; Carlson, Hasher, Connelly, & Zacks, 1995). It may be that older adults
were as able as young adults to ignore the meaning of the word while making the spatial
location information because it always appeared in the same position, above or below the
asterisks. However, this task analysis is tenuous because it would seem that in order to
make the location judgement the location of the word would have to be attended allowing

the meaning of the word to be automatically extracted. Ifthe meaning of the word is

91

processed, then it would be this meaning that would have to be suppressed. Whether the
failure to find a larger interference effect for the older adults in the spatial Stroop task
supports the idea that older adults are spared in their ability to suppress information at a
particular location hinges on whether this Stroop task is a test of the ability to suppress
information at a particular location or the meaning of a word. This is a direction for
future research.
4.2.2 Goal Neglect

The experiments reported do not provide support for the idea that older adults
have more difficulty maintaining the current task goal and, in situations where the
stimulus may be associated with an alternative response, may show larger interference
effects. On the one task that did elicit a pattern of performance consistent with goal
neglect, the spatial Stroop task, the performance of older adults was not more disrupted
than the performance of young adults when the likelihood of goal neglect was increased.
Although the goal neglect hypothesis predicted a larger age difference in performance
when the likelihood of goal neglect was higher, the failure to find this result provides
evidence against the goal neglect hypothesis. Admittedly, this evidence is weak because
it argues from a null statistical finding fiom one experiment and the power of the
manipulation to detect a moderate size interaction was low (power = .55). For example,
one could argue, as has been done for the inhibitory deficit hypothesis, that older adults
are spared increased susceptibility to goal neglect on some tasks, and the spatial Stroop
task is one of them.

More troubling for the goal neglect hypothesis of interference effects was the

failure of manipulations that should have affected the likelihood of goal neglect. Both the

92

RSI and task switching manipulation had no overall effect on antisaccade performance
when speed/accuracy trade-offs in performance were considered. Although the
manipulation of RSI in the Stroop task led to a pattern of data that supports a goal neglect
account of interference effects, this same manipulation did not affect performance in the
antisaccade task in a consistent manner. The different patterns of results for the two tasks
may indicate that the two tasks differed in a critical characteristic that obscured the
impact of goal neglect on antisaccade performance. Altemately, given that the task
switching manipulation also did not have the predicted impact on performance in the
antisaccade task, it may be that the processes involved in antisaccade performance are not
explained well by the goal neglect model. Both of these suggestions have some merit.
Although the spatial Stroop performance was affected by the RSI manipulation
such that interference was more pronounced in the long RSI condition than in the short
RSI condition, this pattern of results was not obtained in the antisaccade task. It might be
argued that the manipulation of RSI used in the antisaccade task was not sufficient to
induce goal neglect. In the spatial Stroop task RSIs of 21 8 and 2028 ms were used,
whereas in the antisaccade task the RSIs were 1200 and 3200 ms. The longer values were
chosen for the eye movement tasks because the RSI was based on the button press to the
target stimulus, but at this point in task performance the eyes were usually still fixating
the target stimulus in one of the peripheral boxes. Before the next trial began the eyes had
to be back at fixation, however. The longer intervals allowed participants enough time to
return their eyes to the center of the screen before the next trial began. In essence, the
response on each trial was not completed until the eyes returned to fixation and so the

difference between the RSIs used in the two conditions was not as great as it might have

93

seemed. Even at slightly longer intervals, the discrepancy between the lengths of the RSIs
on the two tasks probably does not account for the failure to find goal neglect in the
antisaccade task. If the RSI is affecting goal activation in a similar way for the two tasks,
then the cumulative response distribution for the 1200 ms RSI condition in the eye
movement task and the 2000 ms RSI condition in the spatial Stroop task should be
comparable, but they are not. The different RSIs used in the eye movement and spatial
Stroop tasks does not account for the failure to find an effect of RSI in the eye movement
task.

Another feature of the antisaccade task that may have distinguished it from the
spatial Stroop task and eliminated the effect of goal neglect was the warning signal that
unconfounded RSI from foreperiod. In both the short and the long RSI conditions of the
antisaccade task the fixation cross disappeared 200 ms before the peripheral onset was
presented. In the spatial Stroop task the RSI was not interrupted by any events for either
the short or the long RSI condition. As was discussed previously, warning signals aid in
nonspecific preparation processes, and it may be assumed that part of nonspecific
preparation is to increase the activation of the current task goal. If these assumptions are
correct, the warning signal in the antisaccade task may have boosted goal activation 200
ms before stimulus presentation for both RSI conditions eliminating differences in goal
activation that would have been apparent before the warning signal was presented.
Although the preparedness argument is compelling, the failure to find evidence of goal
neglect in the antisaccade task with the task switching manipulation as well suggests that

the goal neglect account may not adequately explain antisaccade performance.

94

In applying the goal neglect hypothesis to performance in the antisaccade task it
was assumed that the goal of the task is to make a correct saccade, toward the cue in the
prosaccade condition and away from the one in the antisaccade condition (see also De
Jong, 2001). In this View, competing task sets only lead to interference if the current task
goal is not sufficiently activated. Once the task goal reaches an activation threshold the
correct task set is selected and other related task sets are inhibited through lateral
inhibition (De J ong et al. 1999, De Jong 2001). This view implies that in the antisaccade
condition, correct antisaccade performance is executed independently of the prosaccade
task set.

Overlap between the processes required for correct antisaccade performance and
the processes required for correct prosaccade performance may prevent the task sets from
being represented as separate schemas, however. Lateral inhibition following antisaccade
schema selection may be insufficient for correct antisaccade performance because, at the
very least, the initial stages of prosaccade response programming are essential in the
antisaccade condition. In order to correctly perform an antisaccade the location of the
peripheral onset must be explicitly attended, as it is in the prosaccade condition. This
explicit attention to the location of a peripheral onset may allow the prosaccade task set
access to response generation and on all trials the response programming must be
explicitly suppressed. This may, in part, account for the failure to find an effect of the
task switching manipulation on antisaccade performance. Perhaps even under single task
conditions, the prosaccade task set is activated to some extent making it difficult to
maintain the antisaccade task goal. Unfortunately, this idea does not account for why the

task switching manipulation did not affect prosaccade performance. There would be no

95

reason for a different task set to be active during single task prosaccade blocks
particularly when that task was introduced before the antisaccade task.

Other research has suggested that older adults have more difficulty on tasks that
require the explicit inhibition of an inappropriate response than young adults, but when
the task includes a inhibitory component that is not conscious older and young adults are
equally able to inhibit irrelevant task sets (Kramer, Hahn, Irwin, & Theeuwes, 2000;
Nieuwenhuis et al., 2000). Kramer and colleagues have found that older and younger
adults were equally likely to incorrectly look toward an irrelevant peripheral onset when
their goal was to look toward a target defined by its color, but only when they were
unaware that the peripheral onset was occurring (Kramer, Hahn, Irwin, & Theeuwes,
1999). When the onset of the nontarget stimulus was made more salient, older adults
were more likely and young adults were less likely to look toward it suggesting that older
adults have more difficulty suppressing responses when the need to suppress is made
explicit, but they perform like younger adults when the suppression goal is not explicit

(Kramer et al., 2000).
4.3 Implications for Theories of Cognitive Control

A goal neglect account of cognitive aging was not supported by this research and
the unexpected findings from the manipulations thought to affect goal neglect may call
into question some of the assumptions of the hypothesis. A recent article also questioned
whether the goal neglect theory was specified adequately to make predictions about how
older adults performance may be affected in a particular situation (De Jong, 2001).

Indeed, concepts such as goal selection, goal maintenance, internal control, and

environmental support are presently so poorly defined that it is on one hand often

96

difficult to derive precise predictions from the goal-neglect hypothesis of

cognitive ageing, while on the other hand it is often relatively easy to provide

some reasonable post hoc account for findings that would at first sight seem to fit
uncomfortably with the hypothesis. (pp.87)

Despite these misgivings about a goal neglect hypothesis of cognitive aging, there
may still be merit in the idea that the current task goal must be actively maintained and
interference in task performance will occur if the goal activation is not sufficient. The
theory is supported by the increase in the interference effect with long RSIs compared to
short RSIs found by De Jong et al. (1999) and replicated here with young and older
adults.

The results fi'om Experiment 2 can be captured by a model of cognitive control
that allows the strength of the competing response to influence task performance. For
example, consider Kimberg and F arah’s production system (Kimberg et al., 1997;
Kimberg & Farah, 1993; 2000). In this view, the strengthening of associations between
stimuli and responses (or goals and tasks) can carry out the work of internal control. It is
the strength of connections that a stimulus shares with different responses that determines
what task is executed. This model can capture the difference between antisaccade
performance in the peripheral cue and central arrow conditions of Experiment 2. A
peripheral onset would have a strong connection to the response of looking toward that
onset, the peripheral-prosaccade strength. A central arrow cue would also have a strong
association with the response of looking in the direction that the arrow is pointing, the
central-prosaccade strength, although the strength of this association should be less than

the peripheral-prosaccade strength because of less experience responding to arrows than

97

responding to changes in our peripheral vision. The two antisaccade association strengths
should be equivalent because the instructions of the experiment would set up these
associations and participants would be expected to have little experience responding in
the opposite direction from these cues. Performing the antisaccade in the peripheral onset
condition would be more difficult because the strength of the association to the
prosaccade response is greater than in the central arrow condition.

This theory can explain the age difference in performance in the antisaccade
condition with central arrow cues as well. Age decrements in performance are thought to
result fi'om weakened associations between stimuli and responses. Because the
association strength of the antisaccade response to peripheral onsets and central arrow
cues should be equivalent, then the weakening of those associations caused by age should
be equivalent. The weakened antisaccade response associations would not lead to
equivalent impairments in antisaccade performance because the competing prosaccade
response associations have different strengths. Weaker central-antisaccade strength in
older adults compared to younger adults should result in age differences in performance
in the central arrow condition, a pattem consistent with the finding of Experiment 2.
Further research is required to determine whether cognitive control in the face of a
competing response requires an inhibitory mechanism. One way to examine this question
would be to look at whether inhibiting a response on one trial makes perfomring that

response more difficult on the next trial.
4.4 Conclusions

The results of this research support the inhibitory deficit hypothesis of aging, but

not the goal neglect hypothesis. Older adults’ performance was slowed to a greater degree

98

when the inhibitory demands of the task were increased, but the goal structure of the
tasks was the same. In addition, an increase in the likelihood of goal neglect was not
sufficient to cause greater interference in older adults. In fact, what appeared to be
straightforward predictions on how goal neglect would affect performance were not

supported.

99

References

Abrams, R. A. & Jonides, J. (1988). Programming saccadic eye movements.
Journal of Experimental PsychologyaHuman Perception and Perforraance. 14. 428-443.

Arrington, C. M., Altrnann, E. M., & Carr, T. H. (2001). Task switching and task
similarity: Old-task inhibition and new-task facilitation. Poster presented at the 2001
annual meeting of the Psychonomic Society in Orlando, FL.

Bertelson, P. (1967). The time course of preparation. Quarterly Journal of
Experimental Psychology, 19, 272-279.

Block, R. A., Zakay, D., & Hancock, P. A. (1998). Human aging and duration
judgments: A meta-analytic review. Psychology & Agiag,L3_, 584-596.

Burke, D. (1997). Language, aging, and inhibitory deficits: Evaluation of a
theory. Journal of Geronotology: Psychological Sciences, 52B(6), P254-P264.

Butler, K. M., Henderson, J. M., & Zacks, R. T. (1999). Suppression of reflexive
saccades in younger and older adults: age comparisons on an antisaccade task. Memopy
& Cognition. 27, 584-589.

Carlson, M. C., Hasher, L., Connelly, S. L., & Zacks, R. T. (1995) Aging,
distraction, and the benefits of predictable location. Psychology & Aging, 10, 427-436.

Cohen, J. (197 7). Sartistical Power Analysis for the Behavioral Sciences.
Academic Press Inc. New York.

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

Connelly, S. L., Hasher, L., & Zacks, R. T. (1991). Age and reading: The impact
of distraction. Psychology and Aging, 6, 533-541.

Craighero, L., Fadiga, L., Rizzolatti, G., & Umilta, C. (1999). Action for
perception: A motor-visual attentional effect. Journal of Experimeml Psychologjp
Human Perception and Performance, 25, 1673-1692.

Craik, F. I. M. & Kerr, S. A. (1996). Commentary: Prospective memory, aging,
and lapses of intention. In M. Brandimonte, G. O. Einstein, & M. A. McDaniel (Eds)
Prospective memory: Theory and applications (pp. 227-238). Mahwah, NJ: Erlbaum.

Daneman, M., & Carpenter, P. A. (1980). Individual differences in working
memory and reading. Journal of Verbal Legningand Verpal Behavior. 19, 450-466.

100

De Jong, R. (2000). An intention-activation account of residual switch costs. In S.
Monsell & J. Driver (Eds), Attention and Performance XVIII: Control of Cognitive
Processes (pp. 357-376). Cambridge, MA : MIT Press.

De Jong, R. (2001). Adult age differences in goal activation and goal
maintenance. European Journal of Cognitive Psychology, 13, 71-89.

De Jong, R., Berendsen, E., & Cools, R. (1999). Goal neglect and inhibitory
limitations: Dissociable causes of interference effects in conflict situations. Acta

Psychologica 101 , 379-394.

Duncan, J. (1995). Attention, intelligence, and the frontal lobes. In M. S.
Gazzaniga (Ed.), The Cognitive Neurosciences (pp. 721-733). Cambridge, MA: MIT
Press.

Duncan, J ., Emslie, H., Williams, P., Johnson, R., & Freer, C. (1996). Intelligence
and the frontal lobe: The organization of goal-directed behavior. Cognitive Psychology,
3_Q, 257-303.

Einstein, G. O., McDaniel, M. A., Manzi, M., Cochran, B., & Baker, M. (2000).
Prospective memory and aging: Forgetting intentions over short delays. Psychology &

Aging, 15, 671-683.

Everling, S., & Fischer, B. (1998). The antisaccade: A review of basic research
and clinical studies. Neuropsychologia, 36, 885-899.

Fischer, B., Biscaldi, M., & Gezeck, S. (1997). On the development of voluntary
and reflexive components in human saccade generation. Brain Research. 754. 285-297.

Guitton, D., Buchtel, H. A., & Douglas, R. M. (1985). Frontal lobe lesions in man
cause difficulties in suppressing reflexive glances and in generating goal-directed
saccades. Experimental Brain Research. 58, 455-472.

Hallet, P. E. (1978). Primary and secondary saccades to goals defined by
instructions. Vision Reseafih, 18, 1279-1296.

Hallet, P. E. & Adams, B. D. (1980). The predictability of saccadic latency in a
novel voluntary oculomotor task. Vision Regarch. 20. 329-339.

Hartman, M., & Hasher, L. (1994). Aging and suppression: Memory for
previously relevant information. Psychology & Aging, 6, 587-594.

Hasher, L., & Zacks, R. T. (1988). Working memory, comprehension, and aging:
A review and a new view. The Psychologyof Learning and Motivation. 22. 193-225.

101

Hasher, L., Zacks, R. T., & May, C. P. (1999). Inhibitory control, circadian
arousal, and age. In D. Gopher, A. Koriat, et al. (Eds), Attention and performance XVII:

Cogpitive regplation of performflce: Interaction of theory and application. (pp. 653-
675). Cambridge, MA: MIT Press.

Home], B., Pratt, J ., Colzato, L., & Godijn, R. (2001). Symbolic control of
visual attention. Psycholog'cal Science. 12. 360-365.

Jersild, A. T. (1927). Mental set and shift. Archives of Psychology, 89.

Kane, M. J ., Bleckley, M. K., Conway, A. R. A., & Engle, R. W. (2001). A
controlled-attention view of working memory capacity. Journal of Experimental
Psychology: Generaly 130. 169-183.

Kimberg, D. Y., D'Esposito, M., & Farah, M. J. (1997). Cognitive functions in the
prefrontal cortex--Working memory and executive control. Current Directiona_i_n_
Psychological Science. Q 185-192.

Kimberg, D. Y. & Farah, M. J. (1993). A unified account of cognitive
impairments following frontal lobe damage: The role of working memory in complex,
organized behavior. J ournal of Experimentaa Psychology: GeneraL 122, 411-428.

Kimberg, D. Y. & Farah, M. J. (2000). Is there an inhibitory module in the
prefi'ontal cortex? Working memory and the mechanisms underlying cognitive control. In
S. Monsell & J. Driver (Eds), Attention and Performance XVIII: Control of Cognitive
Processes (pp. 739-751). Cambridge, MA : MIT Press.

Kingstone, A., & Klein, R., M. (1993). Visual offsets facilitate saccadic latency:
does predisengagement of visuospatial attention mediate this gap effect? Journal of
Experimental Psychology: Human Perception and Performance. 19(6), 1251-1265.

Klein, C., Fischer, 3., Hartnegg, K., Heiss, W. H., & Roth, M. (2000). Optomotor
and neuropsychological performance in old age. Experimental Brain Research, 135, 141-
154.

Klein, R. M., & Pontefract, A. (1994). Does oculomotor readiness mediate
cognitive control of visual attention? Revisited! In C. Umilta & M. Moscovitch (Eds),
Attentiona and Performance XV - Conacious aad Nonconacious Information (pp. 333-
350). Cambridge, MA: MIT Press.

Kramer, A. F., Hahn, S., & Gopher, D. (1999). Task coordination and aging:

Explorations of executive control processes in the task switching paradigm. Acta
Psychologica. 101, 339-378.

102

Kramer, A. F., Hahn, S., Irwin, D. E., & Theeuwes, J. (2000). Age differences in
the control of looking behavior: Do you know where your eyes have been? Psychological
Science 11(3), 210-217.

 

Kramer, A. F., Humphrey, D. G., Larish, J. F., Logan, G. D., & Strayer, D. L.
(1994). Aging and inhibition: Beyond a unitary view of inhibitory processing in attention.
Psychology and Aging, 9, 491-512.

Kray, J. & Lindenberger, U. (2000). Adult age differences in task switching.
Psychology and Aging, 15, 126-147.

Logan, G. D., & Irwin, D. E. (2000). Don't look! Don't touch! Inhibitory control
of eye and hand movements. Psychonomic Bulletin & Review. 7. 107-112.

Los, S. A., & Van Den Heuvel, E. (2001). Intentional and unintentional
contributions to nonspecfic preparation during reaction time foreperiods. Journal of
Erarerimental Psychology: Human Perception and Performance. 27. 370-386.

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

May, C. P. & Hasher, L. (1998). Synchrony effects in inhibitory control over
thought and action. Journal of Experimental Paycholagy: Human Perception a_ngl_
Performance. 24. 363-379.

Maylor, E. A. (1996). Aging-related impairment in an event-based prospective
memory task. lfichology & Aging, 11, 74-78.

Mayr, U. (2001). Age differences in the selection of mental sets: A role of
inhibition, stimulus ambiguity, and response-set overlap. Psychology & Aging, 16, 96-
109.

Mayr, U. & Keele, S. W. (2000). Changing internal constraints on action: The
role of backward inhibition. J ournaaa of Experimental Psychology: Generaa 129. 4-26.

Mayr, U., & Liebscher, T. (2001). Is there an age deficit in the selection of mental
sets? European Journal of Cognitive Psychology, 13, 47-69.

McDowd, J. M. (1997). Inhibition in attention and aging. Joumala of Gerontology
Psychological Sciences, 523(6), P265-P273.

Meiran, N., Gotler, A., & Perhnan, A. (2001). Old age is associated with a pattern
of relatively intact and relatively impaired task-set switching abilities. Journal of
Gerontology: Psychological Sciences. 56B. P88-P102.

103

Munoz, D. P., Broughton, J. R., Goldring, J. E., & Armstrong, 1. T. (1998). Age-
related performance of human subjects on saccadic eye movement tasks. Experimental
Brain Research. 121. 391-400.

 

Niemi, P., & Naatanen, R. (1981). Foreperiod and simple reaction time.
Psychological Bulletin. 89. 133-162.

Nieuwenhuis, S., Ridderinkhof, R. R., de Jong, R., Kok, A., & van der Molen, M.
W. (2000). Inhibitory inefficiency and failures of intention activation: Age-related
decline in the control of saccadic eye movements. Psychology & Aging, 15, 635-647.

Olincy, A., Ross, R. G., Youngd, D. A., & Freedman, R. (1997). Age diminishes
performance on an antisaccade eye movement task. Neurobiology of Aging, 18(5), 483-
489.

Pashler, H. (2000). Task switching and multitask performance. In S. Monsell & J.
Driver (Eds), Control of Cognitive Processes : Attention and Performance XVIII (pp.
277-307). Cambridge, MA: MIT Press.

Pierrot-Deseilligny, C., Rivaud, Gayrnard, Muri & Verrnersch. (1995). Cortical
control of saccades. Neurological ProgressJ7(5), 557-567.

Psychological Software Tools, Inc. (2000). E-prime (Version 1.0 Beta 5).
Pittsburgh, PA.

Rizzolatti, G., Riggio, L., & Sheliga, B. (1994). Space and selective attention. In
C. Umilta & M. Moscovitch (Eds), Attention and Perfonnflce XV (pp. 231-265).
Cambridge, MA: MIT Press.

Roberts, R. J., Hager, L. D., & Heron, C. (1994). Prefrontal cognitive processes:
Working memory and inhibition in the antisaccade task. J ournalaof Experimental
Psychology: Genfirl, 123, 374-393.

Rogers, R., & Monsell, S. (1995). The costs of a predictable switch between
simple cognitive tasks. Journal of Experimental Psycholggy: General. 124, 207-231.

Saslow, M. G. (1967). Latency for saccadic eye movement. Journal of the Optical
Sociem of America, 57(8), 1030-1033.

Shipley, W. C. (1940). A self-administered scale for measuring intellectual
impairment and deterioration. J ournIal of Psychology, 9, 371-377.

Spector, A., & Biederman, I. (1976). Mental set and mental shift revisited.
American Journal of Psychology, 89, 669-679.

104

Stoltzfus, E. R., Hasher, L., & Zacks, R. T. (1996). Working memory and aging:
Current status of the inhibitory view. In J. T. Richardson, R. W. Engle, L. Hasher, R. H.
Logic, E. R. Stoltzfus, & R. T. Zacks (Eds), Working Memory and Humaa Cognition
(pp. 66-88). New York: Oxford University Press.

Sudevan, P., & Taylor, D. A. (1987). The cueing and priming of cognitive
operations. J ouanal of Experimental Psychology: Human Perception and Performance.
.12, 89-103.

Verhaeghen, P., & De Meersman, L. (1998). Aging and the Stroop effect: A meta-
analysis. Psychology and Aging, 13, 120-126.

Walker, R., Deubel, H., Schneider, W. X., & Findlay, J. M. (1997). Effect of
remote distractors on saccade programming: Evidence for an extended fixation zone.
Journal of Neurophysiology, 78, 1108-1 1 19.

West, R. (1999). Age differences in lapses of intention in the Stroop task. Journals
of Gerontolagy: Series B: Psychological Sciences and Social Sciences,54B, P34-P43.

West, R. (2001). The transient nature of executive control processes in younger
and older adults. European Journal of Cognitive PsycholagyJ3 ( V2). 91-105.

West, R. & Alain, C. (2000). Evidence for the transient nature of a neural system
supporting goal-directed action. Cerebral Cortex. 10. 748-752.

West, R. & Craik, F. I. M. (1999). Age-related decline in prospective memory:
The roles of cue accessibility and one sensitivity. Psychology & Aging, 14, 262-272.

Williams, B. R., Ponesse, J. R., Schachar, R. J., & Logan, G. D. (1999).
Development of inhibitory control across the life span. Developmental Psychology, 35,
205-213.

Zacks, R. T., & Hasher, L. (1994). Directed ignoring: Inhibitory regulation of
working memory. In D. Dagenbach & T. H. Carr (Eds) Inhibitogi mechanisms in
attention. memory. and langaage (pp. 241-264). San Diego, CA: Academic Press.

Zacks, R. T., & Hasher, L. (1997). Cognitive gerontology and attentional
inhibition: A reply to Burke and McDowd. Journal of Gerontology: Psychological
Sciences, 52B(6), P274-P283.

Zacks, R. T., Hasher, L., Doren, B., Hamm, V., & Attig, M. S. (1987). Encoding
and memory of explicit and implicit information. Journal of Gerontology. 42, 418-422.

Zacks, R. T., Radvansky, G. A., Hasher, L. (1996). Studies of directed forgetting
in older adults. Journal of Experiments; Psychology: LearningaMemcg, and Cogpition,
Q, 143-156.

105

illililiiliiiiiiiiifliiiiil