Percepfualand~MotorSkills, 1998,87,771-783.

O Perceptual and Motor Skills 1998

INFLUENCE OF EXERCISE ON VISUAL SEARCH:

IMPLICATIONS FOR MEDIATING COGNITIVE MECHANISMS '

DEBORAH J . AKS
Ut~iuerxityof Wiscorzsi~z-Wblirwater

Sutnmary.-The present study investigated the influence of exercise on a feature

and conjunction search performed by 10 women and 8 men. We compared the effect
of 10 min. of low- (65% work load) and high-exertion c y c h g (8 min. at 6.5% work
Load plus 2 min. at 85% work load) on response time, accuracy and response time as
a function of the number of items in the display (slopes). Analysis showed faster
search and increased slopes following exercise, suggesting that exercise improved per-
formance especially under conditions of focused attention, The lowest response times
Followed high-exertion exercise, and most focused search, i.e., largest slopes, followed
exercise of low-exerrion.

Physically active individuals frequently report positive changes in mental

abhties following a period of exercise. Reports include improved alertness
(Tomporowslu & Ehs, 1986), quicker responses (e.g., Gutin, 1973; Schmidt,
1982), and increased energy (Kubitz & Pothakos, 1997; Thayer, 1978). Em-
pirical tests, however, have yielded conflicrmg findings as to the actual bene-
fits of exercise (e.g., Hogervorst, kedel, Jeukendrup, & Jolles, 1996; Kubitz
& Pothakos, 1997; Tomporowski & Ehs, 1986). The seemingly inconsistent
effects are ~Llustratedin the literature on response time and cognitive pro-
cesses following exercise (for comprehensive reviews see Etnier, Salazar, Lan-
ders, Petruzzello, Han, & Nowell, 1997; Tomporowslu & Ehs, 1986). In the
study most similar to the present one, Allard, Brawley, Deaktn, and Elliot
(1989) have shown cycling (at 60% work load for 4 min.) speeds perform-
ance on visual search. Other studies, however, have shown optimal (choice)
reaction time only at intermediate levels of exercise (e.g., Levitt & Gutin,
1971) and stdl other studies have found individuals to be just as quick after
a rested state (e.g., Meyers, Zimmerli, Farr, & Baschnagel, 1969), or even
impaired on (motor) response time (e.g., Sjoberg, 1980; Weingarten & Alex-
ander, 1970), cognitive tasks (e.g., Hancock & McNaughton, 1986; Kubitz
& Pothakos, 1997), and tasks requiring inhibition (AUard, et a/., 1989). Pre-
sumably, such conflicting results reflect ddferences in the types of tasks, e.g.,
psychomotor, sensory, vs cognitive response time, methodology, e.g., type,

'Many thanks to Sceven J. Albrechtsen, Director of UWW Human Performance Laboratory,

for his advise on exercise conditions and loaning us the cycle ergometers. Also, thanks to
Jennifer LaSelva and Ryan Shea for assistance in runnin the experiments. Please address corre-
spondence to Deborah J. Aks, Department of ~sycho!ogy, University of Wisconsin-White-
water, WI 53190 or e-mail (aksd@u~w~ax.uww.edu).
772 D. J. AKS

duration, or intensity of exercise, or mediating mechanisms, e.g., cognitive vs

motor activation, arousal, attentional narrowing or fatigue.
Cognitive mechanisms that can account for discrepant findings include
arousal and attention or a n o r h e a r influence from a single mechanism such
as arousal. One c u w h e a r pattern well-suited to describe many of the arous-
al effects on performance is the classic Yerkes-Dodson's inverted-U hypothe-
sis (1908). The inverted-U theory is frequently reformulated in terms of an
underlying attentional mechanism. Easterbrook's classic theory of attention
(1959) suggested that variations in arousal will produce a change in atten-
tional processes. Increasing arousal WLLI narrow attention and subsequently
facilitate performance up to a certain point, and thereafter, impair perform-
ance. A similar account is offered by Nideffer's (1976) attention control the-
ory. A variety of research supports the inverted-U function (e.g., Davey,
1973; Levitt & Gutin, 1971; Redly & Smith, 1986; Sjoberg, 1980; Salmela &
Ndoye, 1986), but not necessarily attentional narrowing (e.g., Cote, Salmela,
& Papathanasoupou, 1992). These studies taken together suggest that the ef-
fect of exercise on attention has yet to be resolved.
A measure of attention and response time somewhat unique in the ex-
ercise disciphe is visual search. AUard, et al. (1989) introduced this task to
the field after Treisman's Feature Integration Theory (FIT) provided a pop-
ular theoretical explanation of visual-search performance based primarily on
attentional processing (Treisman & Gelade, 1980). Since then, many theoret-
ically driven studies-have been conducted which offer new insights as to
mechanisms underlying visual search performance (see Wolfe, 1997). Some
of the alternatives are presented here to provide a fresh perspective on what
visual search might reveal about the influence of exercise on cognition.
First, consider the information obtained from the visual-search task.
Two of the most fundamental are response times and slopes, i.e., response
times across set size. Accuracy is also important but since in these studies we
aim to restrict performance to fewer than 10% errors, effects of independent
variables are constrained to response time measures. Response times are use-
ful in and of themselves as they provide a measure of general speed or effi-
ciency of processing. Thus, faster response times indicate rapid encodmg
and rapid selection. Slopes summarize response time as a function of the
number of items in a display, thus providing a measure of the time required
per item to detect a target (ms per item).
Interpretation of slopes, however, is somewhat controversial. Treisman's
earliest conception, following Neisser (1967), is that large slopes ( > 10 msec./
item) imply a serial pattern, and shallow slopes ( < 10 msec./item) imply a
parallel search strategy. Today, due to various findings and alternative inter-
pretations (e.g., Townsend, 1971, 1976; Pashler, 1987; see Wolfe, 1998 for a
 EXERCISE AND VISUAL SEARCH IN COGNITION 773

review), the serial vs parallel dichotomy is not considered a literal distinction

but rather a practical one (Wolfe, 1997).
Nevertheless, even with the accumulating evidence that search is not a
strict serial-parallel dichotomy, most researchers still agree that slopes pro-
vide an important measure of relative processing efficiency or attentional
narrowing (Wolfe, 1997). In the present study, we examined relative shifts
across slopes (that exceed 10 msec./item at basehe) to assess whether exer-
cise can influence attentional focus. In addition, we examined two related
theories, attentional narrowing and attentional efficiency. We do not intend
to test the activation based inverted-U theory. Instead, we selected submaxi-
ma1 tests that would likely coincide or precede the peak of the inverted-U
curve. Theories based on attentional narrowing lead us to predict a h e a r re-
lation between intensity of exercise and attentional narrowing, that is, as ex-
ertion increases, attention is expected to become more focused and should
be reflected in increased slopes. The logic of attentional narrowing is sound
in light of the traditional FIT which already presumes a narrowing of the at-
tentional spotl~ghtduring serial search (Treisman & Gelade, 1980). Here we
suggest along with Easterbrook (1959) and Nideffer (1976) that exercise may
further narrow the attentional beam.
Slopes, however, have also been associated with search efficiency (Ash-
by & Townsend, 1986; Wolfe, 1997, 1998). A theory based on attentional
efficiency, as contrasted with attentional narrowing, points to literature on vi-
sual search that shows search for more complex information usually requires
a slower, more deliberate sequential search--one that uthzes a narrower but
more intense focus resulting in a less efficient use of mental effort. Converse-
ly, when searching for simpler information, search becomes faster, parallel,
and more efficient with respect to mental effort.
These two different perspectives of visual search lead to opposing expec-
tations for improved attention. (1) The attentional narrowing view predicts
exercise w d cause slopes to imrease as attention narrows. More time w d be
used to process each additional item added to the display. (2) The efficiency
view predcts slopes to decrease as attention is distributed over multiple
items, i.e., becomes parallel. More items are expected to be processed simul-
taneously following exercise.
AUard, et al. (1989) examined the effect of exercise on slopes and re-
ported no over-all difference across exercise conditions, although there was a
slight decrease on absent trials. However, exercise effects on attention may
have been masked by at least two factors. First, visual search was performed
during exercise, thus potentially introducing influences from fatigue or dis-
traction. Second, their large (15- and 30-item) set-size trials may have intro-
duced a grouping confound known to occur when distractors contain a ho-
mogeneous feature dimension (Humphreys, Quinlan, & hddoch, 1989).
774 D. J. AKS

Thus, in these large set-size trials, the test may have tapped
--
into effects of
exercise on grouping rather than attention. An analysis comparing one and
five-item trials shows the greatest exercise effects in the one-item trials. Re-
moving these methodological limitations and improving upon other aspects
of the design (described later in the method section) may yield reliable
exercise effects on visual-search slopes.
One additional difference in design that may be instrumental in uncov-
ering the influence of exercise on attention is the use of a feature and con-
junction task of intermediate difficulty in which neither is subject to a floor
or ceiling effect. Thus, we selected to study a challenging feature task with
low target-distractor discriminability, and a conjunction search task of inter-
mediate difficulty that could be facilitated through dlstractor grouping of
the same colored items (i.e., Egeth, Virzi, & Garbart, 1984). Therefore, the
goal of the present experiment was to extend the assessment of the effect of
exercise on visual search, while using distinct exercise conditions and novel
search tasks that optimize our abdity to detect exercise influences on atten-
tion. We expect that search w d become more efficient following exercise
and that response times and slopes will decrease as attention is distributed
over the items present in the &splay.
METHOD
Subjects
Eighteen members of the university community (10 female, age range =
19-3 1 years; M = 22, SD = 3 1 participated in two sessions. None had previous
experience with visual search tasks, and all reported normal or corrected-to-
normal vision. A health survey, administered prior to participation, indicated
that all subjects were healthy, had no preexisting heart conditions, rated
themselves as average or better than average in fitness, and regularly exer-
cised between 3 and 12 hours per week. This study had University of Wis-
consin Institutional Review Board approval on the use of human subjects.
S&zmuliand Apparatus
Exercise conditions were performed on a cycle ergometer (Monark
Model 818). Display presentation and data collection were controlled by a
Macintosh computer running VSearch software (Enns, Ochs, & Rensink,
1990). The feature task consisted of 2, 6, or 10 red items with a large circu-
lar target (7 mm) and small circular dstractors (6 rnm) that were identical
except for their l-mm difference in size. In the conjunction task, targets
were distinguished from distractors on the basis of both size and color.
Items consisted of the large, red 7-mm targets and any of the following three
dstractors: small and blue, small and red, or large and blue.
Procedure
Subjects performed feature and conjunction search tasks before and af-
 EXERCISE AND VISUAL SEARCH IN COGNITION 775

ter ten minutes of submaximal exercise on the cycle ergometer. Each trial
began with a central fixation symbol (black dot) lit for 720 msec., followed
by the &splay, which remained visible until the subject responded or 2.9
sec. had elapsed. On each of the search tasks, subjects made speeded biman-
ual responses to indicate whether the target was present in the &splay. On a
random half of the displays the target was present, with the total number of
items varying randomly among 2, 6, and 10. Following the response, visual
feedback was presented at the center of the screen ("+" for correct, "-" for
incorrect). Subjects were instructed to maintain fixation throughout the trial
sequence, to respond as rapidly as possible, and to keep errors below 10%.
Exercise exertion was manipulated so that the low-exertion condition
consisted primarily of aerobic processes, and high-exertion consisted prima-
rily of anaerobic processes. The dstinction between these two forms of exer-
cise is far from absolute and typically defined in terms of differing energy
sources. Ln brief, energy for aerobic metabolism where oxygen is readily
available is derived from glycogen and fat. Energy for aerobic metabolism
wherein oxygen is not r e a d y available and must instead be generated by the
muscles is derived from the phosphate and the lactate system.' For compre-
hensive descriptions of these processes see McArdel, Katch, and Katch (1996)
and Plowman and Smith (1997).
In the low-exertion exercise condition, subjects cycled at a steady pace
of 50 revolutions per minute in synchrony with a metronome set at 100
beats per minute; thus one revolution of cycling corresponded to 2 beats on
the metronome. Women exerted 93 Watts of power over the 10-min. exer-
cise period, and men exerted 147 Watts. In the high-exertion condition, the
first 8 min. were the same as in the low-exertion condition. For the last 2
min., however, the work load was increased by a factor of two so women
now exerted 186 Watts, and men exerted 294 Watts. Levels of exertion were
maintained over the designated (8- and 2-rnin.) time periods by having an
experimenter continuously monitor and provide feedback to subjects to syn-
chronize cycling with the metronome. After each exercise condition, the sub-
ject returned to the computer to perform a second feature and conjunction
visual-search task. Changes in response time as a function of Set Size were
used to assess the efficiency of search and changes in attentional focus.
Subjects were tested in two 40-min. sessions. The low- and high-exer-

'Although it is not clear how changes in glycogen, lactate, and phosphate levels relate to cogni-
tive funcrion, it is reasonable to expect moderare increases in oxy en transport and metabolic
rate to be associated with speeded and perhaps parallel visual searcl since such exercise condi-
tions have sbo\lv large increases in cerebral blood flow (e.g., Jorgensen, Perko, & Secher,
1992), nerve rr.rnsn~issionspeed (Astrand & Rod.~hl,1977), and reduced response rime (Travlos
& Marisjsi. 1996) O n the other hand, the o\ygc.n deficit associated with anaerobic exercise
may lead to rhc opposite effect of slowed cognlrl!c performance (F. D. Aks, personal commu-
nication, A u g ~ ~20,
s r 1997).
776 D. J. AKS

tion sessions were run at the same time on separate days within a 2-wk. pe-
riod. In both sessions, subjects practiced p e d a h g to the metronome for 30
sec. Immediately aftenvards, subjects rested for 2.5 min. while completing a
brief health questionnaire containing questions about their exercise habits
and perceived level of fitness. Following the rest period, subjects partici-
pated in the Feature and Conjunction search tasks with each consisting of
three sets of 40 trials. Order of exercise condtions, search tasks and hand of
response (left vs right) were counterbalanced across subjects.
Statirtical Procedure
A repeated, 5-way analysis of variance was conducted on five factors,
type of exercise (low vs high exertion) x search test (pre vs post) x task (fea-
ture vs conjunction) x target (present vs absent) x set size (2, 6, 10). An al-
pha level of .05 was used for all statistical procedures. Estimated power
prior to the experiment was .51.

Table 1 shows subjects were both faster and more accurate on visual-
search performance after exercise.' Fig. 1 further shows mean correct re-
sponse time and % errors across all five factors noted above. The fastest
search on trials with the fewest items resulted in steeper slopes following ex-
ercise as is shown in Table 2. The greatest increases in slopes occurred on
the target-absent trials following low-exertion exercise, and on conjunction
target-present trials following low- and high-exertion exercise.

TABLE 1
OVER-ALL EFFECTSO F EXERCISEON VISUALSEARCHPERFORMANCE BEFOREAND
A F ~ E REXERCISE:
MEANS A N D STAND~RD
DEVIATIONSFOR RESPONSE
TIMES(MSEC.).SLOPES(MSEC./JTEM),
A N D % ERRORS

Target Preexercise Postexercise F1.1,

M SD IM SD
Response Time 924 248 888 244 6.6"
Slope 40 25 46 27 9.6t
Errors, % 9 11 7 9 6.4t
*p<.05. tp<.Ol.

Additional effects included those that are standard to visual search-

relative to target-absent trials, target-present trials were faster and more ac-
curate. Response times and errors increased with set size and the interaction

'Two of the 18 subjects participated only in the high-exertion exercise conditions. Thus, in the
combined analysis, the coral number of subjects was 16. Also, missing dare were observed in
0.13% of the cells in this design (representing one subjecr on 3 of 48 conditions). These cells
were filled b the grou mean for statistical analyses, but in no case was the reported signifi-
cance level a&ecred by ttis procedure.
 EXERCISE AND VISUAL SEARCH IN COGNITION

-2
0 0
1200
LOW ~xertion 1 Feature Search
a;

Present Absent Present Absent

Target

Conjunction Search 16

Present Absent Present Absent

Target

FIG. 1. Mean correct response time and % Error for Feature (1A) and Conjuncrion (1B)
search trials before and after low- and high-exertion exercise. Error bars indicate SEM. Faster
search after exercise was significant in nwo- ( ~ , ~ = 6 .p<.001)
7, and six-item (t,>=5.0,p<.001)
(pre-post search x set size: F,,,,=7.5, p < ,01). Errors were fairly consistent across set
size (pre-post search x set size: F,,,,=0.9, ns). Exercise: pre ( ) and post ( )
778 D. J. AKS

between target presence and set size reflect the typical 2:1 ratio in slopes be-
tween target-absent ( M = 57 msec./item) and target-present trials (M = 29
msec./item; F,,,, = 106, p < .001)-a trend representative of a serial self-termi-
nating search.

TABLE 2
MFANCORRECT
SLOPES A N D STANDARD D ~ ~ O N S
(MSEC./ITEM)
SearchExertion Target
Present /I, Absent '15
Pretest Posttest Pretest Posttest
Feature
Low M 33.5 34.8 0.3 55.8 67.9" 2.1"
SD 16.8 17.6 32.8 39.6
High IM 30.8 35.2 1.0 61.1 62.9 0.2
SD 14.8 17.6 28.8 30.4
Conjunction
Low M 23.9 31.5" 2.07 43.1 57.1" 5.2*
SD 12.0 14.0 13.6 14.0
High M 19.3 29.5" 2.2$ 59.1 57.5 0.4$
SD 12.4 11.6 23.2 18.8
*p< .05. t p < .O6. t d f = 17. Pre-postsearch: F,,15=9.6, p < .01. Type of exercise x pre-post-
search x target: F1,,,=5.4,pc.05. Errors: pre.postsearch x set size: FI,,,=0.9, ns).

There were also three unexpected findings in the data set.4 The most
important was the stronger high-exertion exercise effects on target-present
trials containing either two (t,,=4.0, p < ,001) or six items (t,,=4.7: p < ,001).
This 4-way interaction also reflects the reversed effect on 10 item target-ab-
sent trials following low-exertion exercise (t,, = 2.8, p < .01). Analyses based
on simple effects of Exercise Type (low-exertion vs high-exertion) and
Search Task (feature vs conjunction) support these trends. One final analysis
presented below further confirms these trends, ensures the four-way interac-

'Subjects were substantially slower and slightly less accurate in detecting the Feature than the
,,,,
Conjunction target RT: F =30.0, pc.001; Slope: F I , 1 5 = 17.3, p<.01; Error: FI,IS=4.2, p=.06.
Faster and more accurate search for the Conjunction search task is likely due to groupin of
the same colored items and inhibition of the irrelevant color feature (i.e., Egeth, Virzi, & t a r -
barc, 1984). This interpretation is corroborated by the compressed effects in rhe 10-item targer-
present pretesr (baseline) condition. There was also some variation across low- and high-exer-
tion prerest b a s e h e target-absent conditions with slower performance in high-exertion trials
(MdiB=67msec./item, f 5=2.4, pc.05). However, this difference wis not significant as a main
effect (F, ,,=2.4, ns). \jarlability within individuals can be accounted for b the fact that low-
and high-exertion exercise were performed on separate days. Nevertheless, tKe critical compari-
son between y,n- and. postexercise conditions w m run on the same day; rhus, rhese were not
subject to suc var~abll~ty. One final analysis examined the effect o l sex which was a nonsignifi-
cant effect (F,,,=0.5, ns): yet h e r e was a significant 4 - w interaction
~ among sex xerercise
t pe x re vs postexercise x set size (F2,,,,=3.6, p < ,051, re ectlng rhe tendency lor women to
SLOW ;st;r RTs from high-exertion exerclse on 6- and 10-irern trials. This interaction should
be interpreted cautiously as additional confounding variables such as subjects' physical firness
were not balanced across all conditions.
 EXERCISE AND VISUAL SEARCH IN COGNITION 779

tion is not spurious and by collapsing across exercise conditions has the
benefit of reducing the number of factors in the analysis.
The Exercise Effect: Difference Scores
Differences between exercise and no exercise conditions shown in Fig.
2 were calculated to isolate and compare further effects of low- vs high-exer-
tion exercise. A positive score indicates faster search from exercise; a nega-
tive score indicates slowed search; and a zero score indicates no effect of ex-
ercise. High-exenion exercise poduced consistently high scores (Merr=60
msec.) that significantly deviated from a zero basehe (t,,,=4.0, p<.001).
The weakest effects occurred on target present ten-item conditions (t,, = 0.0,

Set Size 2
Set Size 6
Set Size 10

Low Exertion High Exertion

Feature

Present Absent Present Absent

Target

FIG. 2. Difference in response time before and after exercise. Error bars indicate SEM.
Significance of univariate f tests: ' p < .06, * p < .05, **p< .01.
780 D.J. AKS

ns). Low-exertion exercise had a more variable influence on performance

(MdZf=14 msec.; t,,,= 1.2, ns) but did consistently speed performance when
subjects detected the absence of a conjunction target in the two item condi-
tions (Mdlii=X1msec.; t,,,=3.9, p<.01).
The lour-way analysis of variance yielded no over-all difference between
the two types of exercise (Response time: F,,,, = 2.4, ns; Error: F,,,, = 1.I, ns).
However, the three-way interaction among type of exercise x target x set
size (response time: F,,,, = 3.9, p < .05) resulced from high-exertion exercise
producing more reliable exercise effects than low-exertion exercise as is
shown in Fig. 2 . Differences in errors are not reported since the interaction
was not significant (Error: F,,,,=3.3, p < .09). The three-way interaction that
emerged on response times supports the consistent exercise effect on high-
exertion trials across set size (ts > 2.0, p < .05) and target-present trials (ts >
2.0). Note, two exceptions include (1) a marginal effect when subjects search-
ed for a conjunction in six-item trials and (2) no effect on ten item condi-
tions (ts > 1.8, ns). Both support earlier demonstrations that the most reliable
exercise effects occurred in trials with the fewest number of items.
Ln low-exertion conditions, only target-absent condtions showed faster
search from exercise in two-item trials. Interestingly, this caused significant
effects of exercise on Slopes (Dlff: t,, = 3.9, p < .002). Trials containing six
and ten items did not show a significant effect (ts< 1.5, ns) which further
supported our finding of faster search in minimal item trials and impaired
performance in multiple item trials. These interactions also reflect the con-
sistent increase in Slopes following low-exertion exercise in the conjunction
condition (Diff: t,, =4.6, p < .001) and a marginal increase in the feature con-
&tion (Dlff: t , , = 1.7, p = .09). Within conjuncrion conhtions, target present
trials showed a marginal dtfference (Slope Diff: t,, = -2.0, p = .06) and target
absent trials showed a substantial difference (Slope Diff: t,, = -5.2, p < .001).
Within feature trials only absent trials were significant (Slope Diff: t,, = -2.1,
p < .05). A significant interaction between type of exercise and target pres-
ence on Slope performance (F,,,,=5.4, p < .05) indicated that, while there
was no difference in the effect of low vs high-exertion exercise on target-
present trials (t,, =0.3, ns), there was a dfference on target-absent trials (t,,
=5.4, p < .05). These response time differences corroborate earlier response
time, slope, and error trends.
Drscussro~
Over-all, our results show a reliable effect of exercise on visual search-
both response times and errors decreased following ten minutes of exercise
(see Fig. 1). These trends are consistent with previous work showing that ex-
ercise of intermediate exertion fachtates performance (e.g., Davey, 1973;
Levitt & Gutin, 1971). Since both exercise conditions in the present experi-
 EXERCISE AND VISUAL SEARCH IN COGNITlON 781

ment lasted only 10 minutes, it is likely that they quahfy as intermediate ex-
ercise intensities. The relatively small work load difference between the low-
and high-exertion conditions (20% dfiference for 2 of 10 min.), did show
subtle influences on performance. The most pronounced difference is that
high-exertion exercise ~ r o d u c e dfaster and more reliable search. Low-exer-
tion exercise, on the other hand, while still speeding search (albeit less in
magnitude), produced steeper slopes indicative of a narrower attentional fo-
cus.
The few existing studes assessing the effects of exercise on attention
have reported conflicting effects (e.g., Allard, et a/., 1989; Cote, Salmela, &
Papathanasoupou, 1992 vs Salmela & Ndoye, 1986). Ln the present experi-
ment, we were surprised to find slopes increased following both forms of
exercise. One interpretation of the steeper slopes is that attention may have
become less efficient following exercise (i.e., Ashby & Townsend, 1986;
Wolfe, 1997, 1998), perhaps to provide a more careful sequential examina-
tion of indvidual items. One problem with this interpretation is that a less
efficient search strategy seems to contradict the findmg that absolute re-
sponse times were faster after exercise.
Alternatively, improved performance under two-item conditions may be
the result of either an exogenous or endogenous attentional mechanism (e.g.,
Easterbrook, 1959; Nideffer, 1979; Treisman & Gelade, 1980; Treisman,
1992). In exogenous attention, stimuli drive the distribution of attention. In
this view, exercise may only improve efficiency when attention is already
narrowly focused. This may occur when a subject detects a target in a local
region of the display as in the case of trials with a small number of items
(LaBerge & Brown, 1989). Two-item trials, since they most frequently clus-
ter together in a small region of the display, possess the local stimulus
property needed to narrow attention. Analysis of search performance before
and after exercise supports this theory by showing that exercise helps per-
formance on trials containing the fewest number of items. As mentioned ear-
lier, a careful analysis of Allard, et al.'s (1989) work shows a similar trend-
performance was better following exercise on single-item trials than five-item
trials.
A second interpretation of performance gains in two-item conditions is
that exercise may directly influence attention (e.g., Easterbrook, 1959). Ac-
cordmg to this endogenous view of attention, internal cognitive factors drive
or modulate the distribution of attention. Exercise may narrow attention re-
sulting in further performance gains in small set-size trials. A distinguishing
characteristic of an endogenous process is that impaired performance is ex-
pected when stirnd are more widely dstributed across the display as occurs
in the ten-item conditions.
In addition to finding faster search when there are fewer items in the
782 D. J. AKS

display, the data from high-exertion trials is most consistent with the exoge-
nous interpretation since performance was not impaired in multiple-item con-
ditions following exercise. On the other hand, the data from low-exertion
trials is most consistent with the endogenous interpretation since perform-
ance is both improved when fewer items are present, and more importantly,
impaired as more items are present in the &splay. This cross-over interac-
tion involving exercise and set-size supports the possibility of two mecha-
nisms, each being activated by varying intensities of exercise. Nevertheless,
ddferences between the two exercise (and search type) conditions appear to
be subtle and require additional testing to v e r b unique effects.
On the whole, this study dustrates that ten minutes of exercise results
in faster response times, fewer errors and steeper slopes. These trends sug-
gest that ten minutes of moderate exercise speeds us up and makes us more
accurate when we are engaged in a search task. This gain in performance is
especially helpful in improving cognitive processing within focused attention.
REFERENCES
ALIARD,F., BRAWLEY, L., DEAKIN, I., &ELLIOT,F. (1989) The effect of exercise on visual atten-
tion performance. Human Performance, 2, 131-145.
ASHBY.F. G., & TOWNSEND, J. T. (1986) Varieties of perceptual independence. Psychological
Review, 93, 154.179.
RAND, F! O., &RODAHL. K. (1977) Test of working physiology: physiological bases of exercise.
New York: McGraw-Hill.
COTE.1.. SALMEIA, J., & PAPATHANASOUPOU, K. I? (1992) Effects of progressive exercise on ar-
rentional focus. Perceptzral and Motor Skills, 75, 351-354.
DAVEY,C. l? (1973) Physical exertion and mental performance. Ergoizoinics, 16, 595-599.
EASTERBROOK. J. A. (1959) The effects OF emotion on cue utilization and the organization of
behavior. Psychological Review, 66, 183-281.
EGETH,H., VIRZI,R.. &GARBART. H. (1984) Searching for conjuncrively d e f i e d rargets. ]orrnzal
of Experierztal Pxychology: Hzlmarz Perceptzot~O Perfonnonce, 10, 32-39.
ENNS.J. 1. OCHS,E. F!, &&NSINK,R. A. (1990) VSearch: Macintosh sofnvare for experiments
in visual search. Behavior Research Methods, lnsfru?nenfs& Compzrters, 22. 1 18- 122.
ETNIER,J. L., SALAZAR. W., LANDERS, D. M., PETRUZZELLO, S. I., HAN, M.. & NOWELL. l? (1997)
The influence of physical fitness and exercise upon cognitive Functioning. 4 rneta-analysis.
Jozrnral of Spori G Exercise Psychology, 19, 249-277.
GUTIN,B. (1973) Exercise-induced activation and human performance: a review. Review Quar-
terly, 44, 256-268.
HANCOCK, S., & MCNAUGHTON. L. (1986) Effects of Fatigue on abhty to process visual infor-
mation by experienced orienteers. Perceptual and Motor Skills, 62, 491 -498
HECKLER, B., &GROCE,R. (1992) Effects of time of posttesr after two durdr~onsOF exercise on
speed and accuracy OF addirion and subrraction by fit and less-fit women. Percepttral and
motor Skills, 75, 1059-1065.
HOGERVORST, E., RIEDEL,W., JEUKENDRUP, A., & JOLLES,J. (1996) Cognitive performance after
strenuous physical exercise. Perceptual and Motor Skills, 83, 479-488.
HUMPHREYS, G . W., QUINIAN,I? T., & RIDDOCH,M. 1. (1989) Grouping processes in visual
search: effects with single and combined-feature targets. Jounlaf ofExperimenta1 Psycholo-
gy: General, 118, 258-279.
JORGENSEN, L. G., PERKO.M., &SECHER, N. H. (1992) Regional cerebral artery mean flow ve-
locity and blood flow during dynamic exercise in humans. ]ourizal of Applied Physiology.
73, 1825-1830.
 EXERCISE AND VISUAL SEARCH IN COGNITION 7 83
KUBITZ, K. A., &POTHAKOS, K. (1997) Does aerobic exercise decrease brain activation. Journal
of Sport & Exernie Psychology, 19, 291-301.
LABERCE, D., &BROWN,V. (1989) Theory of attentional operations in shape idendication. Psy-
chological Review, 96, 101-124.
LEVITK,S., &GUTIN,B. (1971) Multiple choice reaction time and nlovement tinie during physi-
cal exertion. Research Quarterly. 42, 405-411.
MCARDEL,W. D., KATCH,E I., & KATCH.V. L. (1996) Exercise Physlblogy (4th ed.) Baltimore,
MD: Wdhams & Wilkins.
MEYERS, C. R, ZIMMERLI, W., FARR,S. D.. & BASCHNAGEL, N. A. (1969) Effect of strenuous
physical activity upon reaction time. Research Quarterly, 40, 332-337.
NCISSER, U. (1967) Cognitive psychology. New York: Appleton-Century-Crofts.
NIDEFFER, R. M. (1976) Test of Attentional and hterpersonal Style. Journal of Persotzaliiy and
social Psychofogy, 34, 3 94-404.
PASHLER, H. (1987) Detectin con'unctions of color and form: reassessing the serial search hy-
pothesis. Perception ond:~sycLo~hysics, 41, 191-201.
PLOWMAN, S. A,, &SMITH.S. L. (1997) Exercise physiology for health, fitrzess, and perforronce.
Needham, MA: Prent~ceHall.
REILLY, T., &SM[TH,D. (ITS&) Effect of work intensity on performance in a psychomotor task
during exercise. Ergorzo~nics,29, 601-606.
SALMELA, J . H., &NDOYE,0. D. (1986) Cognitive distoaions during progressive exercise. Per-
ceptual and motor Skills, 63, 1067-1072.
SCHMIDT, R. A. (1982) Motor control and learning: a behavioral emwphosis. C h a r n p ~ ~ p IL:
1 . Hu-
man Kineucs.
SIOEERC, H. (1980) Physical fitness and mental performance during and after work. Ergonom-
ics, 23, 977-985.
THAYER, R. E. (1978) Toward a psychological theory of multidimensional acuvation (arousal).
Motivation and Emotion, 2, 1-34.
TOMPOROWSKI. I? D.. &ELLIS,N.R. (1986) Effects of exercise on cognitive processes: a review.
Psychological Btrlletin, 99, 338-346.
TOWNSEND, J. T. (1971) A note on the idendication of parallel and serial processes. Perception
and Psychophysics. 10, 161-163.
TOWNSEND, J. T (1976) Serial and within-sta e independent XaraUel model equivalence on the
minimum completion time. Jot~rnolof dthernatical Pryc ology, 14, 219-239.
TRAVI.OS. A K., & MARLSI, D. Q. (1995) Information processin and concentration as a func-
[Ion of fitness level and exercise-induced activation to exaaustion. Perceptual and Motor
Skrll\. 80, 15-26.
TREISMAN, A (1992) Spreading su pression or feature integration? A reply to Duncan and
Humphreys (1992). Journal o f ~ x ~ e r i m e n t aPsychology:
l Human Perception and Perform-
ance, 18, 589-593.
TREISMAN. A,, &GELADE, G. (1980) A fearure-incegrauon theory of attention. Cognifiue Prychol-
0 0 , 12, 97-136.
WEINCARTEN. G., &ALEXANDER, J . F. (1970) Effects of physical exertion on mental performance
of college males of different physical fitness levels. Perceptual and Motor Skifls, 31, 371-
378.
WOLFE,J . M. (1997) Visual search. In H. Pashler (Ed.), A!ferrtion. London, UK: Univer. Col-
lege London Press.
WOLFE,1. M. (1998) What can one million trials tell us about visual search? Psychological Sci-
ence, 9, 33-39.
YERKES. R., & DODSON, . D. (1908) The relation of stren rh of stimulus rapidity of habit for-
mation. ]ot,rml o)CotBparafivc Neurology and ~ s ~ i b i i f &18.
v . 459-482.

Accepted Azdgztsr 9, 1998.