When Is Search for a Static Target Among Dynamic Distractors Efficient?

Abstract

Intuitively, dynamic visual stimuli, such as moving objects or flashing lights, attract attention. Visual search tasks have revealed that dynamic targets among static distractors can indeed efficiently guide attention. The present study shows that the reverse case, a static target among dynamic distractors, allows for relatively efficient selection in certain but not all cases. A static target was relatively efficiently found among distractors that featured apparent motion, corroborating earlier findings. The important new finding was that static targets were equally easily found among distractors that blinked on and off continuously, even when each individual item blinked at a random rate. However, search for a static target was less efficient when distractors abruptly varied in luminance but did not completely disappear. The authors suggest that the division into the parvocellular pathway dealing with static visual information, on the one hand, and the magnocellular pathway common to motion and new object onset detection, on the other hand, allows for efficient filtering of dynamic and static information.

Full text

When Is Search for a Static Target Among Dynamic Distractors Efficient?
Yaı¨r Pinto, Christian N. L. Olivers, and Jan Theeuwes
Vrije Universiteit
Intuitively, dynamic visual stimuli, such as moving objects or flashing lights, attract attention. Visual
search tasks have revealed that dynamic targets among static distractors can indeed efficiently guide
attention. The present study shows that the reverse case, a static target among dynamic distractors, allows
for relatively efficient selection in certain but not all cases. A static target was relatively efficiently found
among distractors that featured apparent motion, corroborating earlier findings. The important new
finding was that static targets were equally easily found among distractors that blinked on and off
continuously, even when each individual item blinked at a random rate. However, search for a static
target was less efficient when distractors abruptly varied in luminance but did not completely disappear.
The authors suggest that the division into the parvocellular pathway dealing with static visual informa-
tion, on the one hand, and the magnocellular pathway common to motion and new object onset detection,
on the other hand, allows for efficient filtering of dynamic and static information.
Keywords: attention, visual search, dynamic displays, motion, abrupt onset
From everyday experience, it is clear that dynamic properties of
the visual world can guide people in directing their attention.
Dynamic stimuli are characterized by transient changes in the
visual pattern, such as those induced by motion and abrupt onsets.
For example, when searching for a friend in a crowd, one is aided
if that friend is waving a hand. Another example is provided by the
flashing lights of ambulances, which are explicitly designed to
attract attention and notify people of potential danger. It is exactly
this detection of potential danger that has been thought to be the
underlying reason why the visual system is so sensitive to dynamic
information, because movement or abrupt appearances may signal
the presence of a competitor or predator. Alternatively, for the
predator, dynamic information may reveal something about the
prey (see, e.g., Abrams & Christ, 2003; Hillstrom & Yantis, 1994;
Tipper & Weaver, 1998, for arguments along these lines). In the
present article, we focus on the complementary question, namely,
whether, in a dynamic environment, attention can also be effi-
ciently directed toward static objects.
Previous research has indeed confirmed that participants can
efficiently detect a dynamic target among static distractors. Using
a visual search task in which observers had to detect a moving
target among static distractors, Hillstrom and Yantis (1994) and
Yantis and Egeth (1999) found no effect of the number of distrac-
tors on search reaction times (RTs; i.e., search slopes were flat),
indicating that the moving target could be found efficiently and in
parallel across the displays used. Similarly, McLeod, Driver, and
Crisp (1988) showed that when there is more than one moving
item in the display, search is confined to the moving set, and the
static set is effectively ignored. With standard moving stimuli, the
transients are accompanied by a position change of the object.
However, dynamic targets that do not change position also guide
attention. For instance, looming targets among static distractors are
efficiently found (Hillstrom & Yantis, 1994). This effect is prob-
ably related to motion, because the looming is perceived as a 3D
movement toward the observer. It is interesting to note that Hill-
strom and Yantis also reported efficient search for targets that
remained at their locations but that were defined by a “scintillat-
ing” noise pattern, a characteristic that might not be so clearly
associated with movement.
Another important dynamic property other than motion is the
abrupt onset or appearance of an object. Watson and Humphreys
(1995) found very efficient search when the target was always
defined by an abrupt onset relative to a set of static distractors.
Similarly, Watson and Humphreys (1997) later found that search
can be restricted to a whole set of new onsets among a set of (at
least by then) old static distractors. Both studies provided evidence
that, when relevant, onset differences between stimuli can be used
to effectively guide attention.
In sum, so far studies have shown that a dynamic stimulus is
efficiently selected among static items. However, it is not clear
whether the reverse case—a static target among dynamic distrac-
tors—allows for efficient detection too. McLeod et al. (1988)
showed that a nonmoving item among moving items can be found
independent of the number of moving distractors, although search
was still somewhat less efficient than for a moving target among
stationary distractors. This finding appears to suggest that a static
item may indeed be efficiently found among dynamic items. How-
ever, it remains a question as to whether motion represents a
special case in this scenario or whether static items in general can
be efficiently discriminated from dynamic items. In a recent study,
Theeuwes (2004) sought to explore this issue by presenting par-
ticipants with a search task in which the target was always a static
Yaı¨r Pinto, Christian N. L. Olivers, and Jan Theeuwes, Department of
Cognitive Psychology, Vrije Universiteit, Amsterdam, the Netherlands.
This research was funded by Netherlands Organization for Scientific
Research Grant 400-03-008 to Jan Theeuwes and Grant 451-02-117 to
Christian N. L. Olivers.
Correspondence concerning this article should be addressed to Yaı¨r
Pinto, Department of Cognitive Psychology, Vrije Universiteit, Van der
Boechhorststraat 1, 1081 BT Amsterdam, the Netherlands. E-mail:
y.pinto@psy.vu.nl
Journal of Experimental Psychology: Copyright 2006 by the American Psychological Association
Human Perception and Performance
2006, Vol. 32, No. 1, 59 –72
0096-1523/06/$12.00 DOI: 10.1037/0096-1523.32.1.59
59

item (either a vertical or horizontal bar; the participant’s task was
to detect its orientation), whereas the distractors were all abruptly
changing bars of various orientations. That is, in one condition, all
distractor bars changed, in a single frame, from horizontal or
vertical to either left or right oblique (by 45°; hence, after the
change, the target was the only horizontal or vertical bar). In
another condition, in addition to changing orientation, the distrac-
tors also disappeared from their old locations and abruptly reap-
peared randomly at new locations (hence, the target was also the
only bar that kept its position and was not characterized by an
abrupt new onset). The key finding was that search was much more
efficient (as indicated by small or even absent set size effects) than
in a control condition in which all items (including the distractors)
were static. Theeuwes (2004) argued that the visual system calcu-
lates in parallel across the whole visual field whether an item is
dynamic or static. Attention, then, does not prefer a dynamic item
per se but the item that differs the most from its surroundings—in
this case, the static item.
However, Theeuwes’s (2004) explanation did not go by undis-
puted. Davis and Leow (2005) recently argued that it is actually
not the distinction between dynamic and static that allows for
efficient search but that motion is the crucial factor. They argued
that Theeuwes’s (2004) displays allowed for apparent motion to
operate as the items changed from one orientation or position to
the other. In line with McLeod et al.’s (1988) earlier results, then,
a static among moving items may indeed have been efficiently
found. In contrast, a static item among items with other dynamic
features, like luminance changes or onsets, may not be found
efficiently. In support of this suggestion, Davis and Leow (2005)
found that search was highly inefficient for a static target among
distractors that abruptly changed both color and luminance (with-
out disappearing). Davis and Leow concluded that filtering on the
basis of motion may have a special status, whereas filtering a
single static item from a set of items carrying dynamic properties
other than motion (such as abrupt luminance changes) may be very
difficult.
There may be a good reason for why a static target allows for
more efficient search among moving distractors than among abrupt
onset distractors. Although there is substantial evidence that the
automatic capture is contingent on stimulus-specific or display-
wide attentional settings (e.g., Folk, Remington, & Johnston, 1992;
Gibson & Kelsey, 1998), there is also evidence that abrupt onsets
automatically capture attention even when they are not relevant to
the observer (e.g., Jonides, 1981; Remington, Johnston, & Yantis,
1992; Theeuwes, 1991; Todd & Van Gelder, 1979; Yantis &
Jonides, 1984). In a classic example, Yantis and Jonides (1984)
presented participants with a varying number of items. After 1,000
ms, parts of the items were switched off, revealing the to-be-
searched letters. Simultaneously with these offsets, one new item
appeared through an abrupt onset. The abrupt onset was not
predictive of the target, yet Yantis and Jonides (1984) found search
to be very efficient when the target was the new onset (as indicated
by flat search slopes) compared with when it was one of the
previewed items. Yantis and Jonides (1984) therefore concluded
that onsets capture attention and do so automatically. Important for
the present discussion is the finding that when the very same
paradigm is applied to motion (i.e., one of the items is moving—
occasionally the target), search slopes are not flat when the target
is the only moving item in the display, indicating that motion per
se does not capture attention automatically (Hillstrom & Yantis,
1994; Yantis & Egeth, 1999; note that Franconeri & Simons, 2003,
reported some capture for moving stimuli, but Abrams & Christ,
2003, have argued that it is actually the motion onset that captures
attention, not motion per se). The difference in the ability or
strength with which motion and abrupt onset stimuli can capture
attention may explain why one allows for more efficient search of
a static target than the other: Moving distractors may be effectively
ignored, whereas abruptly onsetting distractors may automatically
capture attention, making search for the static target more difficult.
In the present article, we present six experiments following up
on Theeuwes’s (2004) results. We investigated whether his finding
that a single nonchanging item can be easily found is indeed best
explained by a motion filter, as Davis and Leow (2005) suggested,
or whether there is also support for the idea that static items can be
efficiently detected in dynamic displays in general. To this end, we
introduced dynamic displays in which the distractors would
abruptly blink on and off in a continuous cycle without changing
orientation or position. We argued that this should minimize the
apparent motion in the displays and therefore allow for a more
stringent test of the hypothesis that static targets can guide atten-
tion among dynamic items. This blinking condition was then
compared with an apparent motion condition in which the distrac-
tors did change orientation (comparable with Theeuwes’s, 2004,
original displays) and with a control condition in which all items
were static. If only motion allows for efficient segmentation of
target and distractors, then we should find an improvement relative
to the control condition only for the apparent motion condition.
However, if we find an equal improvement in the blinking condi-
tion, then this lends support to the idea that, more generally, static
items can be efficiently detected among dynamic items.
A similar manipulation was recently reported by Pashler (2001).
In his visual search task, two sets of items were presented, one red,
the other green. Participants were instructed to search one set (e.g.,
the red one), which would remain static throughout the trial. It is
important to note that when search started, the other set (e.g.,
green) could also remain static, or it could start blinking, as it
continuously disappeared and reappeared throughout the trial. Pa-
shler (2001) expected that search in the latter condition might be
more difficult, because the continuous blinking was likely to
capture attention away from the relevant set. To his surprise, he
found that overall RTs in the blinking condition were somewhat
faster, indicating that observers could make use of the differences
in dynamics to detect the static target. However, there may be a
number of caveats in Pashler’s (2001) study. First, in the blinking
condition of his first two experiments, all dynamic distractors
blinked on and off in synchrony. This may have allowed for strong
temporal grouping between blinking elements, allowing for them
to be efficiently rejected as a single set (Alais, Blake, & Lee, 1998;
Blake & Yang, 1997; Lee & Blake, 1999; Leonards, Singer, &
Fahle, 1996; Usher & Donnelly, 1998). This grouping may have
been further strengthened by the fact that Pashler (2001) used
different colors for the static and dynamic sets. In a third experi-
ment, Pashler (2001) abandoned the synchrony and made the
dynamic items appear and disappear at random (within con-
straints). Unfortunately, the results were unclear. There was some
RT benefit in the dynamic condition, but it was relatively small,
limited to target absent trials, and moreover accompanied by
increased errors on target present trials. Note also that Pashler
60
PINTO, OLIVERS, AND THEEUWES

(2001) did not vary set size, thus performance could only be
measured in terms of overall RTs and not in terms of search slopes
(reflecting search efficiency). This leaves open the possibility that
search efficiency may actually have benefited substantially from
the differences in dynamics but that observers needed some time to
perform the initial segmentation (leading to increased overall RTs;
i.e., a slope effect vs. an intercept effect, respectively). In the
present experiments, we controlled for these and several other
factors. We consistently find that a static target is efficiently
detected among continuously blinking distractors. Contrary to
Davis and Leow (2005), we concluded that efficient search for a
static among dynamic items is not limited to motion displays. In a
final experiment, we investigated how our results and Davis and
Leow’s results can be reconciled.
Experiment 1: Efficient Search for a Static Target
Among Blinking Distractors
In Experiment 1, participants searched for a static horizontal or
vertical line segment among tilted distractors (see Figure 1). In the
control condition, all distractors were static. We presented partic-
ipants with two dynamic conditions. In the apparent motion con-
dition, they searched for a static item among items that abruptly
flipped back and forth between two orientations, in a continuous
cycle. In the blinking condition, they searched for a static item
among items that continuously blinked on and off (at the same
frequencies as the flipping in the apparent motion condition). We
used four different frequencies and two different phases within
each frequency, so that the distractors would not all flip or blink at
the same time. If the hypothesis that static items in general can
guide attention is correct, then we expected efficient search slopes
for both the blinking as well as the apparent motion condition. If,
on the other hand, motion represents a special case in allowing for
efficient guidance of attention by static items, we would only see
efficient search in the apparent motion condition.
Method
Participants. Six participants, ranging in age from 21 to 31 years (M ⫽
24.5 years), took part as paid (7€ per hour) volunteers. All participants
completed all of the conditions. All had normal or corrected-to-normal
vision.
Apparatus and stimuli. The experiment was conducted on a computer
with a Pentium IV processor, a 17-in. monitor, and a standard QWERTY
keyboard. The software package E-Prime (Psychology Software Tools,
Figure 1. Typical examples of the search displays used in the present
study. Participants searched for a static vertical or horizontal line segment
among slightly tilted line segments. In the actual experiments, the line
segments were white, and the background was black (A). In the apparent
motion condition, the tilted lines flipped back and forth between its original
orientation and a 45° arc deviation from this position (in either direction,
as indicated by the arrows). The tilted lines flipped at a rate evenly
distributed over four frequency groups: Each cycle lasted 150, 200, 250, or
300 ms. In each frequency group, half of the lines flipped in phase, and the
other half flipped in counterphase (B). In the blinking condition, the tilted
lines switched off and on at the same frequencies and phases as in the
apparent motion condition, but they did not change orientation (C). In the
control condition, all line segments remained static.
61
STATIC TARGET AMONG DYNAMIC DISTRACTORS

Pittsburgh, PA) was used for the layout and timing of the experimental
trials. The stimulus field consisted of a 7 ⫻ 6 imaginary matrix (12.68° ⫻
8.26° visual angle). In its cells, white line segments (Commission Inter-
nationale de l’Eclairage [CIE] x, y coordinates: .283, .301, respectively) of
size 0.76° were randomly placed. The distractors could appear anywhere
on the 7 ⫻ 6 matrix, and the target could appear anywhere except in the
middle (row 4 or columns 3 or 4). The luminance of the line segments was
65.62 cd/m
2
, and the background was 0 cd/m
2
, as measured with a
Tektronix photometer. In each display, there was a vertical or horizontal
white line target among white lines that were tilted 22.5° to either side of
the horizontal or vertical.
Procedure. Participants sat at approximately 90 cm from the monitor,
with their fingers resting on the z and m keys, which were used as the
response buttons. The experiment consisted of 15 blocks, each containing
90 trials. The order of the blocks was repeated every three blocks and was
counterbalanced across the participants. Each sequence of three blocks
corresponded to three main conditions: In the apparent motion condition,
participants looked for a static horizontal or vertical white line among tilted
white lines that flipped back and forth between its original orientation and
a 45° arc deviation from this position (in either direction). The tilted lines
flipped at a rate evenly distributed over four frequency groups: each cycle
lasted 150, 200, 250, or 300 ms. In each frequency group, half of the lines
flipped in phase and the other half flipped in counterphase. In the blinking
condition, participants looked for a static horizontal or vertical white line
among blinking tilted white lines. This means that the tilted lines switched
off and on at the same frequencies and phases as in the apparent motion
condition, but they did not change orientation. In the control condition,
participants looked for a static horizontal or vertical white line among static
tilted white lines. In all conditions, set sizes varied randomly within a
block, among 9, 17, and 33 (i.e., 8, 16, or 32 distractors plus one target).
The task was to determine the orientation of the target element. Participants
pressed z for vertical lines and m for horizontal lines. The task was assumed
to require focal attention to be directed to the target element. Before every
block, text appeared on the screen instructing the participants which
condition followed, either apparent motion, blinking,orcontrol. Partici-
pants were instructed that both speed and accuracy were important. The
first three blocks were disregarded as practice. The other 12 blocks were
included in the analyses. The experiment took approximately 120 min, with
breaks between the blocks.
Results
Error percentages were overall low (see Table 1), and an anal-
ysis of variance (ANOVA) revealed no significant effects. We
therefore concentrated on the mean RTs of the correct trials.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of approximately 4% of the trials. See Figure 2 for a graphical
depiction of the findings. A two-way ANOVA on mean RT for
each participant, with condition (control, apparent motion, or
blinking) and set size (9, 17, or 33) as factors, revealed a main
effect for condition, as RTs were elevated in the control condition
compared with the apparent motion and blinking conditions, F(2,
10) ⫽ 22.02, MSE ⫽ 101,194.86, p ⬍ .001, and a main effect for
set size, as RTs increased with set size, F(2, 10) ⫽ 19.03, MSE ⫽
58,566.74, p ⬍ .001. There was also a significant interaction
reflecting the steeper search slope in the control condition com-
pared with the apparent motion and blinking conditions, F(4,
20) ⫽ 15.74, MSE ⫽ 12,014.35, p ⬍ .001. Equivalent overall
effects were present throughout all subsequent experiments and
will not be reported on further. Instead, we concentrated on the
separate comparisons between conditions. These revealed that RTs
were faster and search slopes were shallower in both the apparent
motion condition and the blinking condition than in the control
condition: effects of condition, F(1, 5) ⫽ 26.02, MSE ⫽
128,794.84, p ⬍ .005; F(1, 5) ⫽ 19.55, MSE ⫽ 170,617.65, p ⬍
.01; and Condition ⫻ Set Size, F(2, 10) ⫽ 17.72, MSE ⫽
14,935.69, p ⫽ .001; F(2, 10) ⫽ 15.34, MSE ⫽ 19,629.03, p ⫽
.001, respectively. It is important to note that there was no differ-
ence in RTs, or in search slopes, between the blinking condition
and the apparent motion condition (all ps ⬎ .40).
Figure 2. Mean reaction times (RT) for each condition of Experiment 1
(control, apparent motion, and blink) as a function of set size. For each
condition, the mean search slopes are provided.
Table 1
Average Error Percentages for the Different Conditions and the
Different Set Sizes of Experiments 1– 6
Experiment and condition
Set size
91733
1
Control 5.58 7.30 10.26
Apparent motion 3.20 4.10 4.44
Blink 3.55 3.90 5.33
2
Control 0.56 0.43 0.74
Apparent motion 2.33 1.75 1.44
Blink 1.90 2.22 1.73
3
Control 3.05 3.39 6.92
Random blink 4.10 3.01 4.25
Standard blink 4.26 2.15 3.12
4
Control 3.73 4.71 3.38
Blink 2.68 0.68 3.76
Apparent motion 3.00 3.33 2.69
5
Control 1.22 1.79 3.06
All blink 2.22 1.98 6.25
Standard blink 4.04 2.09 3.23
6
Control 3.99 4.32 7.33
Twinkle 4.54 5.47 4.66
Bright blink 4.52 3.82 4.30
Dark blink 5.19 4.92 3.75
62
PINTO, OLIVERS, AND THEEUWES

Discussion
Experiment 1 shows that participants are equally efficient in
detecting a static target among moving items as in detecting a
static target among blinking items. This is not in accordance with
Davis and Leow’s (2005) explanation that a static target can only
be rapidly found among moving items. Instead, it provides evi-
dence for Theeuwes’s (2004) original account that in general a
static target can be found efficiently among dynamic items. It also
corroborates and extends Pashler’s (2001) earlier findings of faster
overall RTs when half the distractors are blinking synchronously.
Relatively efficient search in the blinking condition is more sur-
prising, given the evidence reviewed in the introduction that abrupt
onsets capture attention automatically. In the blinking condition,
the static target was present among up to 32 distractors continu-
ously blinking on and off at various rates. This constant multitude
of abrupt onsets should have prevented observers from quickly
finding the target. We will return to this issue in the General
Discussion. Before that, we need to exclude several alternative
explanations of our findings. A first alternative explanation may be
based on the average luminance level across display frames within
a trial and was tested in Experiment 2.
Experiment 2: Controlling for Average Luminance Level
In Experiment 2, we investigated the possibility that it was the
average luminance of the target that causes efficient search in the
blinking condition in Experiment 1. Note that across the changing
frames of the blinking condition, the static target (which is always
on) has a higher average luminance than the surrounding dynamic
distractors, because the latter are switched off on half the number
of frames. Furthermore, there is evidence that luminance is per-
ceived differently shortly after stimulus onsets and offsets (Eagle-
man, Jacobson, & Sejnowski, 2004). This may have enabled
participants to efficiently search for an overall luminance differ-
ence instead of for a static item among dynamic items. To control
for this, in Experiment 2, we varied the luminance of all elements
randomly. The luminance of the target varied between 25% and
62.5% of the maximum; the luminance of the distractors, when on
screen, varied between 25% and 100% of the maximum. There-
fore, the average luminance of the distractors across frames varied
between 11.25% and 50% of the maximum, assuring that neither
the average luminance of the target, which on most trials was
lower than 50% of the maximum, nor the momentary luminance of
the target within each frame, which was exactly in between that of
the distractors present, could provide a reliable clue for search. If
average luminance is the cause of the relatively efficient search for
a static among blinking items, then the more efficient search in the
blinking condition compared with the control condition should no
longer be possible in Experiment 2. In contrast, relatively efficient
search in the apparent motion condition should still be possible. If
it does not play any role in causing the relatively efficient search
of a static among on- and offsets, then the search slopes in the
apparent motion and blinking conditions should remain similar, as
was the case in Experiment 1.
Method
Six new participants, ranging in age from 21 to 25 years (M ⫽ 24.3
years), took part as paid (7€ per hour) volunteers. Everything was identical
to Experiment 1, except that now the luminance of the target element
varied randomly between 16.32 cd/m
2
(CIE x, y coordinates ⫽ .290, .300,
respectively) and 41.04 cd/m
2
(CIE x, y coordinates ⫽ .282, .300, respec
-
tively); the luminance of the distractor elements randomly varied between
16.32 cd/m
2
(CIE x, y coordinates ⫽ .290, .300, respectively) and 65.62
cd/m
2
(CIE x, y coordinates ⫽ .283, .301, respectively). Thus, the target
had a luminance between 25% and 62.5%, and the distractors had a
luminance between 25% and 100% of the maximum luminance. This was
to assure that neither average luminance across frames nor momentary
luminance within each frame was a reliable clue for target search.
Results
Error percentages were overall low, as can be seen in Table 1.
However a two-way ANOVA, with condition (control, apparent
motion, or blinking) and set size (9, 17, or 33) as factors, revealed
a main effect for condition, F(2, 10) ⫽ 6.73, MSE ⫽ 1.53, p ⬍ .05,
but no effect for set size nor for the interaction ( ps ⬎ .7). Overall,
fewer errors were made in the control condition than in the
dynamic conditions. However, there was no effect on error slopes,
and most important there was no difference between the apparent
motion and blinking conditions (all ps ⬎ .7). We therefore con-
centrated on RTs.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of approximately 3% of the trials. Figure 3 shows a graphical
depiction of the results, which were analyzed in the same way as
in Experiment 1. Separate comparisons between the conditions
revealed that RTs were faster and search slopes were shallower in
both the apparent motion condition and the blinking condition than
in the control condition: condition, F(1, 5) ⫽ 18.91, MSE ⫽
312,377.36, p ⬍ .01; F(1, 5) ⫽ 19.16, MSE ⫽ 351,068.71, p ⬍
.01; and Condition ⫻ Set Size, F(2, 10) ⫽ 21.61, MSE ⫽
25,522.36, p ⬍ .001; F(2, 10) ⫽ 16.75, MSE ⫽ 42,762.39, p ⫽
.001, respectively. It is important to note that there were no
differences in RTs or slopes between the blinking condition and
the apparent motion condition (all ps ⬎ .10). If anything, there was
a trend for participants to be faster in the blinking condition than
in the apparent motion condition.
Figure 3. Mean reaction times (RT) for each condition of Experiment 2
(control, apparent motion, and blink) as a function of set size. For each
condition, the mean search slopes are provided.
63
STATIC TARGET AMONG DYNAMIC DISTRACTORS

Discussion
The results were essentially the same as in Experiment 1 and
unaffected by the luminance differences. We therefore dismiss the
notion of average or temporary luminance uniqueness as a cause of
the relatively efficient search of a static target among blinking
distractors. Note, however, that overall, search slopes were a bit
increased compared with Experiment 1, but this affected all con-
ditions in the same way. Thus, the variable luminance of target and
distractor elements made search a little more difficult in general,
without affecting the ability to efficiently segment static from
dynamic items.
Experiment 3: Controlling for Long-Range Apparent
Motion and Temporal Grouping
In Experiment 3, we assessed the possibility that apparent mo-
tion might still provide an explanation for the relatively efficient
search of the static target surrounded by on- and offsets. In both
Experiments 1 and 2, the distractors of the blinking condition were
evenly distributed over four frequency groups, and within each
frequency group, half the lines were in phase and the other half
were in counterphase. This may have permitted for long-range
apparent motion to occur (Burt & Sperling, 1981): If two lines
were in the same frequency group but in counterphase, it could
appear that one line was jumping back and forth between two
positions. Although this should have been much weaker than in the
actual apparent motion condition, it may just have been sufficient
for the relatively efficient search found in the blinking condition.
To determine whether this long-range apparent motion is the
explanation of the relatively efficient search in the blinking con-
dition, in Experiment 3, we added a random blinking condition.
Instead of at a fixed frequency, in the random blinking condition,
all blinking elements switched on and off randomly. All blinking
lines had an equal chance of switching (turning from on to off or
vice versa) after 150 ms, 200 ms, 250 ms, or 300 ms. How much
time it took before the previous switch occurred did not affect
chances for when the current switch would occur. Consequently, in
the random blinking condition, the odds are much smaller that at
any given moment two elements are in counterphase at the same
frequency, and long-range apparent motion therefore is unlikely to
occur. If long-range apparent motion is part of the explanation of
why a static target is relatively efficiently found among blinking
distractors, then it is expected that in this experiment, search in the
random blinking condition will not be as efficient as search in the
standard blinking condition.
In addition to acting as a control for long-range apparent motion,
the random blinking condition also allowed us to investigate
whether temporal grouping contributes to the relatively efficient
search of the static target among blinking distractors. In Experi-
ments 1 and 2, the blinking distractors changed at either one of
four frequencies. It may have been the case that items that shared
a frequency were grouped together. Because there were four fre-
quencies, observers may have always distinguished four distractor
groups, regardless of whether the distractor set consisted of 8, 16,
or 32 items (resulting in four temporal groups of 2, 4, or 8 items,
respectively). Relatively efficient rejection of these temporal
groups as a whole would then result in relatively efficient search.
Evidence that observers are able to use temporal differences to
group certain stimuli and segment them from others comes from a
study by Lee and Blake (1999; see also Alais, Blake, & Lee, 1998;
Blake & Yang, 1997; Leonards, Singer, & Fahle, 1996; Usher &
Donnelly, 1998). In their displays, all items moved in random
directions, and most items changed direction at random moments
in time. However, one spatially contiguous patch of items always
changed direction at the same moment. Even though the direction
in which they changed was still random, the fact that these items
changed together was sufficient for them to be perceived as seg-
mented from the background elements. Lee and Blake (1999)
concluded that synchrony per se is a strong segmentation cue. Note
that in our displays, the synchronously blinking items were usually
not spatially contiguous—a factor that has been shown to weaken
grouping by synchrony (Fahle & Koch, 1995; Forte, Hogben, &
Ross, 1999; Kiper, Gegenfurter, & Movshon, 1996). Nevertheless,
it seemed prudent to control for this factor. In the random blinking
condition of the present experiment, there were no frequency
groups, and temporal grouping of the distractors could no longer
contribute to efficient search. Consequently, if temporal grouping
of the distractors caused relatively efficient search of the static
target in the dynamic displays used in Experiments 1 and 2,
relatively efficient search is no longer expected in the random
blinking condition in Experiment 3.
Method
Twelve participants, ranging in age from 20 to 30 years (M ⫽ 23.7
years), took part as paid (7€ per hour) volunteers. This experiment was
identical to Experiment 1, except now there was a random blinking
condition instead of the apparent motion condition. In the random blinking
condition, the tilted elements all stayed on the screen for a random period
between 150 and 300 ms. Every tilted element had a chance of 25% to
switch on or off after 150 ms, 200 ms, 250 ms, and 300 ms. Previous
switching time did not affect following switching times. The experiment
consisted of 9 blocks of 90 trials. The order of the blocks was repeated
every three blocks and was counterbalanced across participants. The first
three blocks were regarded as practice.
Results
Error percentages were overall low (see Table 1), and an
ANOVA, with condition and set size as factors, only revealed a
significant effect for set size, F(2, 22) ⫽ 3.64, MSE ⫽ 9.04, p ⬍
.05. We concentrated on the mean RTs of the correct trials.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of less than 3% of the trials. See Figure 4 for a graphical
depiction of the findings. The results were analyzed in the same
way as in the previous experiments. Separate comparisons between
the conditions revealed that RTs were faster and search slopes
were shallower in both the random blinking condition and the
standard blinking condition than in the control condition: condi-
tion, F(1, 11) ⫽ 60.10, MSE ⫽ 158,449.59, p ⬍ .001; F(1, 11) ⫽
75.07, MSE ⫽ 118,051.12, p ⬍ .001; and Condition ⫻ Set Size,
F(2, 22) ⫽ 28.60, MSE ⫽ 29,929.91, p ⬍ .001; F(2, 22) ⫽ 29.30,
MSE ⫽ 21,412.60, p ⬍ .001, respectively. Furthermore, the search
slope in the random blinking condition was somewhat shallower
than the search slope in the standard blinking condition (11.9 ms
per item compared with 16.3 ms per item): Condition ⫻ Set Size,
F(2, 22) ⫽ 4.26, MSE ⫽ 4,206.86, p ⬍ .05.
64
PINTO, OLIVERS, AND THEEUWES

Discussion
We found that participants were neither slower nor less efficient
in the random blinking than in the standard blinking condition.
There may have been some long-range apparent motion in the
standard blinking conditions of Experiments 1 to 3, but if there
was, it should have been reduced substantially in the present
random blinking condition. If long-range apparent motion was a
major contributor to the relatively efficient search of a static
among dynamic items, we should therefore have found increased
slopes in the random blinking condition. However, if anything, we
found slightly (but significantly) decreased slopes. Therefore, the
findings from Experiment 3 imply that long-range apparent motion
does not contribute to the relatively efficient search for a static
element among on- and offsets.
Another account tested in Experiment 3 was temporal grouping.
It could be argued that four temporal groups of distractors were
created in Experiments 1 and 2 and that this caused the relatively
efficient search for the static among dynamic items. In the random
blinking condition, temporal synchrony between any of the dis-
tractors was eliminated, yet relatively efficient search for the static
target was still observed. We concluded that temporal grouping of
the distractors cannot be the cause of the relatively efficient search
for a static target among dynamic distractors. Instead, we proposed
the fact that it is static grants the target a unique status among the
dynamic distractors, resulting in relatively efficient search.
Experiment 4: Search for a Static Target Without
Preknowledge
Because in all previous experiments conditions were blocked
and participants knew beforehand whether the distractors were
dynamic or static, it could well be that the top-down expectations
of participants influenced the efficiency with which they selected
a static target among dynamic items. To assess this possibility, in
the present experiment, we randomly mixed all conditions. The
target was static, whereas the distractors could undergo apparent
motion, blink, or remain static as well. Because conditions ap-
peared in random order, participants did not know beforehand
what kind of distractors they would be presented with. If pre-
knowledge is crucial for the relatively efficient search of a static
target among dynamic distractors, it is expected that this search is
no longer efficient in this experiment.
Method
Ten participants, ranging in age from 18 to 27 years (M ⫽ 20.7 years),
took part as paid (7€ per hour) volunteers. The apparatus, stimuli, and
procedure were the same as in Experiment 3, except for the following
changes. There were three conditions: the control condition, the blinking
condition, and the apparent motion condition. The control condition and
the blinking condition were the same as the control condition and the
random blinking condition, respectively, in Experiment 3. The apparent
motion condition was the same as the apparent motion condition in Ex-
periment 1, except now the items moved at random rather than at a given
frequency. The tilted elements stayed in one orientation for a random
period between 150 and 300 ms. Every tilted element had a chance of 25%
to flip 45° arc deviation from its current orientation (in either direction)
after 150 ms, 200 ms, 250 ms, and 300 ms. Previous switching time did not
affect following switching times. The experiment consisted of five blocks,
and one block consisted of 90 trials. Within one block, all conditions were
randomly mixed. The first two blocks were disregarded as practice.
Results
Error percentages were overall low, as can be seen in Table 1.
A two-way ANOVA, with condition (control, apparent motion, or
blinking) and set size (9, 17, or 33) as factors, revealed no
significant effects. Therefore, we concentrated on RTs.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of less than 4% of the trials. Figure 5 shows a graphical
depiction of the results, which were analyzed in the same way as
in the previous experiments. Separate comparisons between the
conditions revealed that RTs were faster and search slopes were
shallower in both the apparent motion condition and the blinking
condition than in the control condition: condition, F(1, 9) ⫽ 33.37,
MSE ⫽ 163,180.86, p ⬍ .001; F(1, 9) ⫽ 38.22, MSE ⫽
236,706.26, p ⬍ .001; and Condition ⫻ Set Size, F(2, 18) ⫽ 33.27,
MSE ⫽ 33,206.08, p ⬍ .001; F(2, 18) ⫽ 30.34, MSE ⫽ 52,100.45,
p ⬍ .001, respectively. Moreover, participants were faster and had
Figure 5. Mean reaction times (RT) for each condition of Experiment 4
(control, apparent motion, and blink) as a function of set size. For each
condition, the mean search slopes are provided.
Figure 4. Mean reaction times (RT) for each condition of Experiment 3
(control, standard blink, and random blink) as a function of set size. For
each condition, the mean search slopes are provided.
65
STATIC TARGET AMONG DYNAMIC DISTRACTORS

shallower search slopes in the blinking condition compared with
the apparent motion condition: condition, F(1, 9) ⫽ 42.31, MSE ⫽
10,738.61, p ⬍ .001, and Condition ⫻ Set Size, F(2, 18) ⫽ 5.84,
MSE ⫽ 9,188.81, p ⬍ .05, respectively.
Discussion
The results of Experiment 4 were highly comparable with Ex-
periments 1 and 2. Again participants were slower in the control
condition than in the apparent motion condition and the blinking
condition. Because in the current experiment, conditions were
mixed, these results show that participants do not need to know
before the start of the trial that the distractors are dynamic to use
this information as a cue to search for the static target. Note that
although the general pattern of results was the same, overall search
slopes were increased compared with Experiment 1. Thus, the lack
of preknowledge regarding the nature of the target made search a
little more difficult in general, without affecting the ability to
segment static from dynamic items.
In accordance with the trend observed in Experiment 2, we
found participants to be slightly faster in the blinking condition
than in the apparent motion condition. This tendency might be due
to the fact that on- and offsets provide a stronger dynamic cue than
movement. We will elaborate on this in the General Discussion.
Experiment 5: Search for a Dynamic Target Among
Dynamic Distractors
There is the possibility that there is nothing special about static
items. A static among dynamic items may be unique in the sense
that it changes at an infinitely slow frequency, but perhaps observ-
ers can efficiently direct their attention at any unique frequency.
To test this possibility, we introduced the all blinking condition in
which not only the distractors blink on and off but the target does
too—at a unique frequency. Furthermore, there were three ver-
sions of this condition, presented in separate blocks. In the slow
blinking target condition, the target would show the lowest blink-
ing rate of all items in the display. In the medium blinking target
condition, the target blinked at a frequency in between the fre-
quencies of the distractors. In the fast blinking target condition, the
target blinked at the highest rate of all items. We included these
different versions because previous evidence has indicated that the
unique target feature may have to be linearly separable from the
distractor features for efficient search to occur (Bauer, Jolicoeur, &
Cowan, 1996; D’Zmura, 1991; Saumier & Arguin, 2003; Wolfe,
Friedman-Hill, Stewart, & O’Connell, 1992). Thus, efficient
search for a unique frequency may occur for the low- and high-
frequency targets (because they are linearly separable from the
distractors) but not for the medium-frequency target. For the
purpose of comparison, we also included a standard blinking
condition in which only the distractors blinked (at frequencies
matched to those in the all blinking conditions) and a control
condition in which all items were static. If static targets have a
special status, then we should only see efficient search in the
standard blinking condition. If any unique frequency allows for the
efficient search, then we should also find improved performance in
the all blinking conditions compared with the control condition.
Method
Seven participants, ranging in age from 16 to 29 years (M ⫽ 20.1 years),
took part as paid (7€ per hour)volunteers. There were seven conditions.
Three all blinking conditions, three standard blinking conditions, and a
control condition. In the slow blinking target condition, the target element
switched on or off every 350 ms, and the tilted elements switched on or off
every 150, 200, 250, or 300 ms. As in Experiments 1 and 2, the tilted
elements were evenly distributed over frequency groups, and within one
frequency group, half switched in phase and half switched in counterphase.
The medium blinking target condition was the same as the slow blinking
target condition, except that now the target switched on or off every 250
ms, and the distractors switched on or off every 150, 200, 300, or 350 ms.
In the fast blinking target condition, the target switched on or off every 150
ms, and the distractors switched on or off every 200, 250, 300, or 350 ms.
For comparison, there were also three standard conditions. In these con-
ditions, the target was always static, but the distractors behaved in the same
way as in the slow, medium, and fast blinking target conditions. Finally,
the control condition was the same as the control condition in Experiments
1, 2 and 3, in which all items were static. Before every block, there
appeared a text on the screen instructing the participants which of the seven
conditions followed. The experiment consisted of five clusters of seven
blocks, and each block consisted of 36 trials. The order of the blocks was
the same within each cluster. The order was determined by a 7 ⫻ 7 Latin
square design.
Results
Error percentages were overall low (see Table 1), and an
ANOVA, with condition and set size as factors, revealed no
significant effects. We therefore concentrated on the RTs.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of less than 4% of the trials. See Figure 6 for a graphical
depiction of the findings. First, we looked at whether there were
any differences among the different versions of the all blinking
conditions. A two-way ANOVA, with condition (slow, medium, or
fast blinking target) and set size (9, 17, or 33) as factors, revealed
a trend for a main effect for condition, F(2, 12) ⫽ 3.36, MSE ⫽
179,481.76, p ⫽ .07. Participants tended to be overall somewhat
faster in the slow blinking target condition. However, there was no
interaction with set size, F(4, 24) ⫽ 1.25, MSE ⫽ 95,323.25, p ⬎
Figure 6. Mean reaction times (RT) for each condition of Experiment 5
(control, all blink, and standard blink) as a function of set size. For each
condition, the mean search slopes are provided.
66
PINTO, OLIVERS, AND THEEUWES

.3, indicating that search efficiency was very similar across these
conditions. Therefore, we pooled the three versions of the all
blinking conditions together. A similar analyses on the three ver-
sions of the standard blinking conditions (in which only the dis-
tractors blinked) revealed no main effect for condition, F(2, 12) ⫽
0.73, MSE ⫽ 22,709.42, p ⬎ .50, but a marginally significant
interaction of condition and set size, F(4, 24) ⫽ 2.80, MSE ⫽
7,069.59, p ⫽ .049. Closer analyses revealed that search was
somewhat less efficient when the distractors were blinking at the
same rate as in the medium blinking target condition (28.8 ms/item
vs. 20.9 ms/item and 23.0 ms/item, respectively, for the equivalent
controls for the slow and fast blinking target conditions). Because
of this difference, we also compared each of the slow, medium,
and fast standard blinking conditions with their equivalent coun-
terparts of the all blinking conditions. These pairwise comparisons
revealed that observers were always faster and always more effi-
cient when the target was static (standard blinking condition)
compared with when it was blinking (all blinking condition; all
ps ⬍ .015). Therefore, because all three standard blinking condi-
tions appeared to behave in more or less the same way relative to
the all blinking conditions, we decided to pool them together to
form one standard blinking condition.
A two-way ANOVA, with these pooled conditions (all blinking,
standard blinking, or control) and set size (9, 17, or 33) as factors,
revealed a main effect for condition, as RTs were shortest in the
standard blinking condition followed by the control and the all
blinking conditions, F(2, 12) ⫽ 18.20, MSE ⫽ 239,352.54, p ⬍
.001, a main effect for set size, as RTs increased with increasing
set size, F(2, 12) ⫽ 25.37, MSE ⫽ 301,088.77, p ⬍ .001, and a
significant interaction, indicating shallower search slopes in the
standard blinking condition than in the all blinking and control
conditions, F(4, 24) ⫽ 9.04, MSE ⫽ 58,809.39, p ⬍ .001. Separate
comparisons between the conditions revealed that RTs were faster
and search slopes were shallower in the standard blinking condi-
tion compared with the control and with the all blinking condition:
condition, F(1, 6) ⫽ 23.58, MSE ⫽ 262,164.93, p ⬍ .005; F(1,
6) ⫽ 17.73, MSE ⫽ 387,533.10, p ⬍ .01; and Condition ⫻ Set
Size, F(2, 12) ⫽ 6.90, MSE ⫽ 89,844.89, p ⬍ .05; F(2, 12) ⫽
18.44, MSE ⫽ 49,712.61, p ⬍ .001, respectively. RTs and search
slopes did not differ significantly between the all blinking condi-
tion and the control condition ( ps ⬎ .24).
Discussion
The results indicate that participants could not make use of the
unique target frequencies when all items blinked, even when these
frequencies were linearly separable from the distractor frequen-
cies. This finding indicates that the unique frequency aided little to
nothing. We concluded that it is not just any temporal difference
that may guide attention but that it is specifically the target being
static that allows for efficient search.
Of course, our results may heavily depend on the range of
frequencies we have used. Perhaps the difference between the
target frequency and the distractor frequencies in the all blinking
conditions was simply not large enough? The fastest rate contained
one blink every 150 ms (a frequency of 6.67 Hz), and the slowest
rate contained one blink every 350 ms (a frequency of 2.86 Hz),
with the other rates falling in between at steps of 50 ms. The
problem is that if we would extend this range much further, the
dynamic items become practically undistinguishable from static
items. That is, if we made the rate much slower, then in its on
period, an item is on for so long that it might be regarded as a static
item. In this respect, we were especially careful not to let items be
on for longer than one would normally expect observers to start
generating a response (i.e., around 400 ms). If we would make the
rate much faster, then subsequent frames will be merged into a
single percept, again making it virtually static. In this respect, we
were careful not to choose a frequency at which the blinking object
might be perceptually treated as one and the same object across
frames (i.e., faster than once every 100 ms; Yantis & Gibson,
1994), a point to which we will return in the General Discussion.
Finally, psychophysical studies have shown that observers are
quite sensitive to frequency and phase differences below 10 Hz (as
used in the present study), with relative difference thresholds of
around 0.1 (Mandler, 1984; Mowbray & Gebhard, 1955; see also
Forte, Hogben, & Ross, 1999). This means that observers should
be able to distinguish the 50-ms differences between rates. Thus, it
deserves pointing out that our conclusions are limited to the
frequency range used here but that this range is not unreasonable.
Experiment 6: Luminance Offsets Versus Complete
Object Offsets
In Experiments 1–5, we showed that participants are able to find
a static among blinking items, and by elimination of other expla-
nations, we concluded that it is indeed the static nature of the target
that allows for relatively efficient search. This is in line with
Theeuwes’s (2004) account but contradicts Davis and Leow’s
(2005) claim that search for a static target among dynamic dis-
tractors is not possible unless the distractors exhibit some form of
motion. They based this claim on their finding that search was
inefficient for a static target among distractors that abruptly
changed both color and luminance from frame to frame. Thus, a
question remains how our results can be brought into accordance
with Davis and Leow’s findings. An important difference between
our experiments and Davis and Leow’s experiments is that in our
blinking conditions, the blinking distractors completely disap-
peared and reappeared. In Davis and Leow’s experiment, the
distractors changed in luminance and color but did not disappear.
A possible explanation for the discrepancy between our results and
Davis and Leow’s results, then, could be that luminance change of
the distractors is not enough for relatively efficient search but that
complete on- and offsets of the distractors are required. There is
substantial evidence that the effectiveness of luminance transients
depends not only on the relative increase or decrease in luminance
but also on whether a new perceptual object is being created (Cole,
Kentridge, & Heywood, 2004; Enns, Austen, DiLollo, Rauschen-
berger, & Yantis, 2001; Yantis & Hillstrom, 1994). A second
reason why search may have been less efficient in Davis and
Leow’s displays is that the luminance changes were accompanied
by a color change. It has been found that simultaneous changes in
color, on the one hand, and dynamic properties such as motion or
orientation changes, on the other hand, are not always perceived as
simultaneous (Clifford, Arnold, & Pearson, 2003; Moutoussis &
Zeki, 1997). The same asynchrony may apply to color and lumi-
nance changes, possibly obscuring the dynamic signal. Moreover,
the color changes are likely to have activated the color-sensitive
parvocellular pathway. There is evidence that the parvocellular
67
STATIC TARGET AMONG DYNAMIC DISTRACTORS

pathway inhibits the magnocellular pathway thought to be sensi-
tive for dynamic information such as luminance transients (Breit-
meyer & Williams, 1990; Tassinari, Marzi, Lee, DiLollo, & Cam-
para, 1999; Yeshurun, 2004; Yeshurun & Levy, 2003).
To investigate these possibilities, in Experiment 6, we presented
participants with four conditions: a twinkle condition in which the
distractors underwent luminance changes but never disappeared
from the display (comparable with Davis and Leow’s, 2005, dis-
plays but without a color change); two blinking conditions; and a
control condition. In the twinkle condition, distractors abruptly
changed luminance between 33% and 100% of the maximum
against a black background of zero luminance. In the bright blink-
ing condition, the distractors also abruptly switched between 33%
and 100% of the maximum luminance but now against a gray
background of 33% luminance. In the dark blinking condition, the
distractors abruptly changed between 0% and 67% of the maxi-
mum luminance against a zero luminance background. This way,
both the absolute luminance change and the luminance change
relative to the background were controlled for.
If complete object disappearances and reappearances are impor-
tant, then we would expect efficient search only in the blinking
conditions and not in the twinkle condition, where luminance
changes were equivalent but the distractors did not disappear.
Furthermore, unlike in Davis and Leow’s (2005) study, the dis-
tractors in our twinkle condition did not undergo a color change. If
color changes were the main reason for inefficient search in their
study, then we might expect efficient search in our twinkle
condition.
Method
Sixteen participants, ranging in age from 18 to 33 years (M ⫽ 12.0
years), took part as paid (7€ per hour) volunteers. The apparatus, stimuli,
and procedure were the same as in Experiment 1, except for the following
changes. There were four conditions: the control condition in which all
elements stayed on the screen with a luminance of 59.99 cd/m
2
(CIE x, y
coordinates ⫽ .286, .305, respectively), and the background had zero
luminance; the bright blinking condition in which the target remained on
the screen with a luminance of 59.99 cd/m
2
(CIE x, y coordinates ⫽ .286,
.305, respectively), the background had a luminance of 20.13 cd/m
2
(CIE
x, y coordinates ⫽ .291, .305, respectively), and the tilted elements alter-
nated between these two luminances; the dark blinking condition in which
the target remained on the screen with a luminance of 40.13 cd/m
2
(CIE x,
y coordinates ⫽ .285, .305, respectively), the background had zero lumi-
nance, and the tilted elements alternated between these two luminances;
and the twinkle condition in which the target remained on the screen with
a luminance of 59.99 cd/m
2
(CIE x, y coordinates ⫽ .286, .305, respec
-
tively), the background had zero luminance, and the tilted elements alter-
nated between a luminance of 59.99 cd/m
2
(CIE x, y coordinates ⫽ .286,
.305, respectively) and a luminance of 20.13 cd/m
2
(CIE x, y coordinates ⫽
.291, .305, respectively). The experiment consisted of five clusters of four
blocks; each block consisted of 36 trials. The first cluster of four blocks
was disregarded as practice. Within each cluster, the blocks had a fixed
order. This order was counterbalanced among participants.
Results
Error percentages were overall low (see Table 1). An ANOVA,
with condition and set size as factors, revealed no significant main
effects but a significant effect for the interaction, F(6, 90) ⫽ 2.35,
MSE ⫽ 9.35, p ⬍ .05. Further analyses revealed that errors
increased more with set size in the control condition, especially in
comparison with the dark blink condition. Because this pattern did
not go against the pattern of RTs, we excluded the possibility of a
speed–accuracy trade-off, and we concentrated on RTs.
Trials on which RTs were two and a half standard deviations
away from the mean were excluded from analysis, resulting in a
loss of less than 3% of the trials. See Figure 7 for a graphical
depiction of the findings. The results were analyzed in the same
way as in the previous experiments. Separate comparisons between
the conditions revealed that RTs were faster and search slopes
were shallower in the dark blinking condition, the bright blinking
condition, and the twinkle condition compared with the control
condition: condition, F(1, 15) ⫽ 93.94, MSE ⫽ 179,734.26, p ⬍
.001; F(1, 15) ⫽ 101.56, MSE ⫽ 203,462.42, p ⬍ .001; F(1, 15) ⫽
93.82, MSE ⫽ 60,607.43, p ⬍ .001, and Condition ⫻ Set Size,
F(2, 30) ⫽ 68.53, MSE ⫽ 31,922.70, p ⬍ .001; F(2, 30) ⫽ 71.62,
MSE ⫽ 37,418.55, p ⬍ .001; F(2, 30) ⫽ 17.59, MSE ⫽ 26,071.62,
p ⬍ .001. Also RTs were faster and search slopes were shallower
in both the dark blinking condition and the bright blinking condi-
tion compared with the twinkle condition: condition, F(1, 15) ⫽
53.12, MSE ⫽ 55,985.96, p ⬍ .001; F(1, 15) ⫽ 65.39, MSE ⫽
71,416.13, p ⬍ .001, and Condition ⫻ Set Size, F(2, 30) ⫽ 26.89,
MSE ⫽ 23,970.92, p ⬍ .001; F(2, 30) ⫽ 34.03, MSE ⫽ 27,118.90,
p ⬍ .001. Overall, participants were somewhat faster in the bright
blinking than in the dark blinking condition, F(1, 15) ⫽ 32.53,
MSE ⫽ 5,680.51, p ⬍ .001, and participants had shallower search
slopes (by 4 ms/item) in the bright blinking than in the dark
blinking condition, F(2, 30) ⫽ 10.32, MSE ⫽ 2,420.00, p ⬍ .001.
Discussion
The results from Experiment 6 again showed that search for a
static target among on- and offsets is relatively efficient. The main
finding is that search for a static target among distractors changing
only in luminance without offsets was considerably less efficient,
even when the extent of the luminance change was the same as in
the on- and offset conditions in either relative or absolute terms.
This finding implies that the change of luminance of distractors is
not enough for efficient search. Instead, the distractors need to
Figure 7. Mean reaction times (RT) for each condition of Experiment 6
(control, twinkle, bright blink, and dark blink) as a function of set size. For
each condition, the mean search slopes are provided.
68
PINTO, OLIVERS, AND THEEUWES

completely disappear and reappear to allow the visual system to
fully separate the distractors from the static target.
Our results are in accordance with the hypothesis that the crucial
difference between our experiments and Davis and Leow’s (2005)
search task is the complete on- and offsets of the distractors. When
items change in luminance but produce no on- or offsets, partici-
pants are less able to efficiently detect a static target. In contrast to
Davis and Leow’s experiment, in our task, there were no color
changes, yet the results were very much comparable. This evi-
dence suggests that the color change probably played little to no
role in causing the inefficient search in Davis and Leow’s displays.
This result furthermore implies that Theeuwes’s (2004) original
account is not entirely correct. Participants are not always able to
rapidly find a static target among dynamic distractors. They can do
so when the distractors are moving or blinking but not when the
distractors are only changing in luminance. Note, though, that
there were nevertheless some benefits in the twinkle condition
relative to the control condition, suggesting that the observers
made some (but limited) use of the luminance changes. The
implications of these findings are elaborated on in the General
Discussion.
General Discussion
This work started with the question of whether observers are
able to efficiently detect a static target among dynamic distractors.
Others have claimed that such efficiency may only be confined to
the case of moving distractors (Davis & Leow, 2005). However,
here we have shown that relatively efficient search may also be
achieved when the distractors are characterized by multiple asyn-
chronous abrupt onsets and that this search is equally efficient or
even slightly more efficient than when the distractors are moving,
lending support to Theeuwes’s (2004) claim that, in general, static
information may be detected relatively rapidly from among dy-
namic information. In Experiment 1, we showed that search for a
static target among blinking distractors is performed efficiently
and that this search is as efficient as for a static target among
moving distractors. In Experiment 2, average or momentary lumi-
nance differences between the target and the distractors were ruled
out as an explanation. In Experiment 3, long-range apparent mo-
tion was also dismissed as a possible cause, as was temporal
grouping on the basis of common onset frequencies. Experiment 4
showed that preknowledge of the dynamic nature of the distractors
is not needed for efficient search of the static target. The results of
Experiment 5 showed that it is not just any unique frequency that
results in efficient search. Experiment 6 made clear that a lumi-
nance change alone is not enough for efficient search. It does
improve search efficiency, but the largest improvement in search
efficiency arises when the blinking objects fully disappear and
reappear. The results of this last experiment suggest that neither
Theeuwes (2004) nor Davis and Leow were entirely correct. Par-
ticipants cannot efficiently find a static target among dynamic
distractors, regardless of the dynamic nature of the distractors.
They can do so when the distractors move or blink on and off
completely but not when the distractors only change in luminance
(though luminance changes per se appear to contribute too).
Note that so far we have discussed improvements in perfor-
mance in the dynamic conditions in terms of relative efficiency. It
deserves mentioning that, even though search became substantially
more efficient when distractors were dynamic in nature, search
slopes were not completely flat, indicating that the target was
usually not found entirely in parallel across the display. This
finding contrasts with the flat search slopes usually found when
participants search for a dynamic target among static distractors
(Hillstrom & Yantis, 1994; Watson & Humphreys, 1995; Yantis &
Egeth, 1999). However, Mu¨ller and Found (1996) showed that
participants after intensive practice were as efficient in finding a
static target among moving distractors as the reverse. It could be
that in our experiments, with more extensive practice, participants
would have shown parallel search.
In any case, contrary to Davis and Leow’s (2005) claim, the
present results show that motion is not unique in allowing for
relatively efficient search. Instead, we agree with Pashler (2001)
that dynamic differences between stimuli, such as abrupt onsets,
are just another aspect on which these stimuli can be discriminated
and prioritized for attentional selection. The question remains as to
how the visual system performs this discrimination.
Memory
One way the visual system might discriminate static from dy-
namic stimuli is by building up a memory representation of the
successive frames of the changing displays. By taking a series of
complete snapshots and comparing them, the visual system may
then filter out the only item that is present in every single snapshot.
Such snapshots may be taken from iconic memory, a large-
capacity initial storage of visual information (Sperling, 1960).
However, various studies have now shown that not much visual
information survives from one snapshot to the next, at least not
across brief interruptions or eye movements (Irwin, 1991;
O’Regan, Rensink, & Clark, 1999; Pashler, 1988; Phillips, 1974;
Simons, 1996). Illustrative in this respect are the change blindness
studies showing that observers often fail to detect large changes
between two separate displays when the transients accompanying
these changes are eliminated (see Rensink, 2002, for a review).
Closer to the present study, Theeuwes (2004) presented two suc-
cessive displays in which all elements changed, except the target,
which remained identical. When the second display followed the
first without interruption, the target was easily found. However,
when a blank display was inserted between the two frames, search
for the one nonchanged element became very effortful. Taken
together, the evidence indicates that a high-level memory repre-
sentation of the displays offers an implausible explanation of the
efficient detection of static among dynamic elements. Instead, just
as efficient detection of a changing target appears to rely on the
presence of transients at its location, efficient detection of a static
target appears to rely on the presence of transients at the distractor
locations. This evidence appears to point to an important role of
relatively earlier visual transient detection mechanisms.
Transient Versus Sustained Channels
The dynamic properties of the stimulus are already distin-
guished at the level of retinal ganglion cells. One group of cells,
the so-called X cells (or
␤
/B cells), have relatively slow conduction
velocity, and they show sustained firing to predominantly station-
ary stimuli. The Y cells (or
␣
/A cells) have faster conduction
velocities, and they show transient bursts of firing in response to
69
STATIC TARGET AMONG DYNAMIC DISTRACTORS

abrupt changes in the stimulus, such as onset and motion (Cleland,
Dubin, & Levick, 1971; Enroth-Cugell & Robson, 1966; Lev-
enthal, Rodieck, & Dreher, 1981; Stone & Fukuda, 1974). This
division is thought to lie at the basis of what have been referred to
as the magno- and parvocellular pathways at the physiological
level, or transient and sustained channels of visual processing at a
more functional level, and it extends into (and probably beyond)
the primary visual cortex (e.g., Breitmeyer & Ganz, 1976; Living-
stone & Hubel, 1988; Todd & Van Gelder, 1979).
The relatively independent transient and sustained subsystems
may provide a direct explanation for the efficient discrimination of
static and dynamic information. When searching for a dynamic
target, the visual system only has to look for activity in the
transient (magno) channel. When looking for a static target, as in
the present experiments, the registration of activity in the sustained
(parvo) channel is sufficient. The fact that dynamic targets are
usually somewhat faster and more efficiently detected than static
targets may then be explained by a slight preference of the visual
system for the transient channel or by the fact that the transient
neurons have faster conduction velocities. Note, however, that the
transient-sustained dichotomy at a pure retinal level cannot ac-
count for the data. In the twinkle condition of Experiment 6, the
distractors featured abrupt luminance changes equivalent to those
in the blinking conditions but without completely disappearing
from the displays. Retinal cells should have responded equally in
these conditions, yet search was much more effortful when the
distractor offsets were not complete. Thus, the crucial distinction
between changing and nonchanging elements appears to be
whether a new object has been created relative to its background,
rather than a simple luminance change. Such new object compar-
isons relative to its surroundings are more likely to be made at
higher levels, for instance in the primary visual cortex (V1), where
center-surround cells provide important information about the
background. Moreover, V1 cells respond to both moving and
blinking stimuli but do not distinguish between them (Andersen,
1997). This would explain why performance in our motion and
blinking conditions was so similar. In contrast, the higher up
motion-sensitive area, the middle temporal area of the extrastriate
visual cortex (MT), responds only weakly or not at all to blinking
stimuli (Andersen, 1997). Thus, V1 may provide the necessary
initial mechanisms for both motion processing and temporal seg-
mentation of static and dynamic stimuli (Fahle, 1993; Forte et al.,
1999).
Attentional Capture
A somewhat surprising aspect of our findings is that search for
a static item among blinking items is efficient, despite the fact that
the distractors are characterized by repeated abrupt onsets. As
mentioned in the introduction, there is substantial evidence that,
under the right circumstances, abrupt onsets automatically capture
attention (Yantis & Jonides, 1984). As also mentioned earlier,
moving stimuli appear to be less strong attentional captors (Hill-
strom & Yantis, 1994; Yantis & Egeth, 1999). So in our blinking
conditions, why were observers not continuously distracted by the
abrupt onsets of the blinking distractors?
One possible explanation is that after the first abrupt onset,
subsequent onsets become much less salient. As has been sug-
gested before, what appears to make an abrupt onset so salient is
the appearance of a new object (Yantis & Hillstrom, 1994). Rel-
evant to this finding is a study by Kahneman, Treisman, and Gibbs
(1992), who found that features were recognized faster when they
had been part of one and the same object across time relative to
when they had switched objects. Kahneman et al. suggested that
whenever a new object appears, an object file is created, storing
the object properties as long as the spatiotemporal continuity is
preserved. The creation of this object file requires attention. In our
experiments, continuously appearing items failed to attract atten-
tion. Perhaps, then, the blinking line segments were not seen as
new objects but as one and the same object simply disappearing
and reappearing. However, a study by Yantis and Gibson (1994)
argued against this account. In one of their experiments, they used
a visual search task in which one of the items briefly disappeared.
When it reappeared, it could be the target, although it was more
likely to be a distractor. The amount of attentional capture was
determined as the relative efficiency of search when the blinking
item was indeed the target. Yantis and Gibson (1994) found that
the blinking item automatically captured attention as long as the
temporal interval between the offset and onset was more than
about 100 ms. In the same study, the same interval was also found
crucial in determining the percept of bistable apparent motion
displays (so-called Ternus displays, in which part of the array may
be seen as stationary or moving depending on the interval between
frames). Yantis and Gibson (1994) concluded that a spatiotemporal
discontinuity of around 100 ms is sufficient for an object to be
regarded as new. In the present study, the distractor items were
always switched off for at least 150 ms (and up to 350 ms). Thus,
according to Yantis and Gibson’s (1994) measure, these items
should be regarded as new and, in principle, capable of capturing
attention.
Another possibility is that abrupt onsets only capture attention
when they are the only onset in the display. In the same vein, the
creation of new object files may only be limited to a small number
of items (perhaps because of the limited attentional resources
available for such creation). In other words, single onsets may
capture attention, but multiple onsets, when distributed evenly
across the visual field, may not. Like other features such as color
or orientation, abrupt onsets may become more salient the more
unique they are. Some support for this idea comes from a study by
Chastain and Cheal (1999), who found that a single onset precue
shows all the characteristics of involuntary attentional capture
(rapid attentional build-up followed by rapid attentional decay
across longer precue to target delays) but that an onset of multiple
precues shows the characteristics of voluntary attentional control
(slow attentional build-up and no attentional decay with longer
time between precue and target). However, other studies suggest
that multiple onsets can still capture attention. For instance, Yantis
and Johnson (1990; see also Yantis & Jones, 1991) found just as
strong attentional capture in displays of up to 16 items, half of
which were defined by abrupt onsets. Similarly, Donk and Theeu-
wes (2003) found that in a visual search task, participants priori-
tized up to 14 new elements (as defined by an abrupt onset) over
up to 14 interspersed old elements, even when the target was twice
as likely to be old. Thus, the presence of multiple onsets in itself
does not appear to necessarily prevent attentional capture. Note
further that, in our displays, the blinking items appeared and
disappeared at different rates, making them at least locally rela-
tively unique.
70
PINTO, OLIVERS, AND THEEUWES

It appears, then, that observers can at least exert some control
over the attentional capture by abrupt onsets (see also Pashler,
2001; Yantis & Jonides, 1990). Attention may be initially captured
by one of the blinking items (or sometimes even the target, because
it too is initially defined by an onset), but soon observers are able
to ignore them and actually use the difference between transient
and sustained signals to direct their attention to the target. This
account may be regarded as in between the automatic capture
account (Theeuwes, 1992; Yantis & Jonides, 1984), which states
that some stimuli capture attention regardless of the tasks and
goals of the observer, and the contingent capture account (Folk et
al., 1992; Folk, Remington, & Wright, 1994), which states that
capture is dependent on the attentional set of the observer. In the
case of multiple onsets, attention may initially be captured auto-
matically, but it is then quickly overridden by top-down goal
settings to prevent further interference.
A final finding worth returning to, now in relation to attentional
capture, is the fact that search for a static target was relatively
inefficient when the dynamic distractors only changed luminance,
without disappearing (Experiment 6). Recently, Enns et al. (2001)
reported a related finding: An item featuring a maximum lumi-
nance change (i.e., a polarity reversal: changing from black to
white on a gray background) did not get as much priority in search
as an item that newly appeared in an empty location (i.e., whose
luminance changed from the gray background to either black or
white). Enns et al. concluded, as we do here, that the visual system
is biased toward new object appearances rather than luminance
changes. In our experiments, in terms of attentional capture, one
might actually have expected search to be easier when the distrac-
tors only changed luminance, because they would draw less atten-
tion. The fact that search was actually more difficult means that the
effect of new object appearances cuts both ways: They draw more
attention when they are unique in the display, but they can also be
more easily discriminated and rejected when they constitute the
distractors.
References
Abrams, R. A., & Christ, S. E. (2003). Motion onset captures attention.
Psychological Science, 14, 427– 432.
Alais, D., Blake, R., & Lee, S. H. (1998). Visual features that vary together
over time group together over space. Nature Neuroscience, 1, 160 –164.
Andersen, R. A. (1997). Neural mechanisms of visual motion perception in
primates. Neuron, 18, 865– 872.
Bauer, B., Jolicoeur, P., & Cowan, W. B. (1996). Visual search for colour
targets that are or are not separable from distractors. Vision Research,
36, 1439 –1465.
Blake, R., & Yang, Y. (1997). Spatial and temporal coherence in percep-
tual binding, Proceedings of the National Academy of Sciences, USA, 94,
7115–7119.
Breitmeyer, B. G., & Ganz, L. (1976). Implications of sustained and
transient channels for theories of visual pattern masking, saccadic sup-
pression, and information processing. Psychological Review, 83, 1–36.
Breitmeyer, B. G., & Williams, M. C. (1990). Effects of isoluminant-
background color on metacontrast and stroboscopic motion: Interactions
between sustained (P) and transient (M) channels. Vision Research, 30,
1069 –1075.
Burt, P., & Sperling, G. (1981). Time, distance, and feature trade-offs in
visual apparent motion. Psychological Review, 88, 171–195.
Chastain, G., & Cheal, M. (1999). Time course of attention effects with
abrupt-onset and offset single- and multiple-element precues. American
Journal of Psychology, 112, 411– 436.
Cleland, B. G., Dubin, M. W., & Levick, W. R. (1971). Sustained and
transient neurons in the cat’s retina and laterate geniculate nucleus.
Journal of Physiology, 217, 473– 496.
Clifford, C. W. G., Arnold, D. H., & Pearson, J. (2003). A paradox of
temporal perception revealed by a stimulus oscillating in colour and
orientation. Vision Research, 43, 2245–2253.
Cole, G. G., Kentridge, R. W., & Heywood, C. A. (2004). Visual salience
in the change detection paradigm: The special role of object onset.
Journal of Experimental Psychology: Human Perception and Perfor-
mance, 30, 464 – 477.
Davis, G., & Leow, M. C. (2005). Blindness for unchanging targets in the
absence of motion filtering: A response to Theeuwes (2004). Psycho-
logical Science, 16, 80 – 82.
Donk, M., & Theeuwes, J. (2003). Prioritizing selection of new elements:
Bottom-up versus top-down control. Perception & Psychophysics, 65,
1231–1242.
D’Zmura, M. (1991). Color in visual search. Vision Research, 31, 951–966.
Eagleman, D. J., Jacobson, J. E., & Sejnowski, T. J. (2004, April). Per-
ceived luminance depends on temporal context. Nature, 428, 854 –856.
Enns, J. T., Austen, E. L., DiLollo, V., Rauschenberger, R., & Yantis, S.
(2001). New objects dominate luminance transients in setting attentional
priority. Journal of Experimental Psychology: Human Perception and
Performance, 27, 1287–1302.
Enroth-Cugell, C., & Robson, J. G. (1966). The contrast sensitivity of
retinal ganglion cells of the rat. Journal of Physiology, 187, 517–552.
Fahle, M. (1993). Figure-ground discrimination from temporal informa-
tion. Proceedings of the Royal Society of London Series B, Biological
Sciences, 254, 199 –203.
Fahle, M., & Koch, C. (1995). Spatial displacement, but not temporal
asynchrony, destroys figural binding. Vision Research, 35, 491– 494.
Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary
covert orienting is contingent on attentional control settings. Journal of
Experimental Psychology: Human Perception and Performance, 18,
1030 –1044.
Folk, C. L., Remington, R. W., & Wright, J. R. (1994). The structure of
attentional control: Contingent attentional capture by apparent motion,
abrupt onset, and color. Journal of Experimental Psychology: Human
Perception and Performance, 20, 317–329.
Forte, J., Hogben, J. H., & Ross, J. (1999). Spatial limitations of temporal
segmentation. Vision Research, 39, 4052– 4061.
Franconeri, S. L., & Simons, D. J. (2003). Moving and looming stimuli
capture attention. Perception & Psychophysics, 65, 999 –1010.
Gibson, B. S., & Kelsey, E. M. (1998). Stimulus-driven attentional capture
is contingent on attentional set for displaywide visual features. Journal
of Experimental Psychology: Human Perception and Performance, 24,
699 –706.
Hillstrom, A. P., & Yantis, S. (1994). Visual motion and attentional
capture. Perception & Psychophysics, 55, 399 – 411.
Irwin, D. E. (1991). Information integration across saccadic eye move-
ments. Cognitive Psychology, 23, 420 – 456.
Jonides, J. (1981). Voluntary vs. automatic control over the mind’s eye’s
movement. In J. B. Long & A. D. Baddeley (Eds.), Attention and
performance IX (pp. 187–203). Hillsdale, NJ: Erlbaum.
Kahneman, D., Treisman, A., & Gibbs, J. G. (1992). The reviewing of
object files: Object-specific integration of information. Cognitive Psy-
chology, 24, 175–219.
Kiper, D. C., Gegenfurtner, K. R., & Movshon, J. A. (1996). Cortical
oscillatory responses do not affect visual segmentation. Vision Research,
36, 539 –544.
Lee, S. H., & Blake, R. (1999, May). Visual form created solely from
temporal structure. Science, 284, 1165–1168.
Leonards, U., Singer, W., & Fahle, M. (1996). The influence of temporal
71
STATIC TARGET AMONG DYNAMIC DISTRACTORS

phase differences on texture segmentation. Vision Research, 36, 2689 –
2697.
Leventhal, A., Rodieck, R. W., & Dreher, B. (1981, September). Retinal
ganglion cell classes in the old world monkey: Morphology and central
projections. Science, 213, 1139 –1142.
Livingstone, M. S., & Hubel, D. H. (1988, May). Segregation of form,
color, movement, and depth: Anatomy, physiology, and perception.
Science, 240, 740 –749.
Mandler, M. B. (1984). Temporal frequency discrimination above thresh-
old. Vision Research, 24, 1873–1880.
McLeod, P., Driver, J., & Crisp, J. (1988, March). Visual search for a
conjunction of movement and form is parallel. Nature, 332, 154 –155.
Moutoussis, K., & Zeki, S. (1997). A direct demonstration of perceptual
asynchrony in vision. Proceedings of the Royal Society of London Series
B, Biological Sciences, 264, 393–399.
Mowbray, G. H., & Gebhard, J. W. (1955, February). Differential sensi-
tivity of the eye to intermittent white light. Science, 121, 173–175.
Mu¨ller, H. J., & Found, A. (1996). Visual search for conjunctions of
motion and form: Display density and asymmetry reversal. Journal of
Experimental Psychology: Human Perception and Performance, 22,
122–132.
O’Regan, J. K., Rensink, R. A., & Clark, J. J. (1999, March). Change-
blindness as a result of “mudsplashes.” Nature, 398, 34.
Pashler, H. (1988). Familiarity and visual change detection. Perception &
Psychophysics, 44, 369 –378.
Pashler, H. (2001). Involuntary orienting to flashing distractors in delayed
search? In C. L. Folk & B. Gibson (Eds.), Attraction, distraction, and
action: Multiple perspectives on attentional capture (pp. 77–92). New
York: Elsevier Science.
Phillips, W. A. (1974). Distinction between sensory storage and short-term
visual memory. Perception & Psychophysics, 16, 283–290.
Remington, R. W., Johnston, J. C., & Yantis, S. (1992). Involuntary
attentional capture by abrupt onsets. Perception & Psychophysics, 51,
279 –290.
Rensink, R. A. (2002). Change detection. Annual Review of Psychology,
53, 245–277.
Saumier, D., & Arguin, M. (2003). Distinct mechanisms account for the
linear non-separability and conjunction effects in visual shape encoding.
Quarterly Journal of Experimental Psychology: Human Experimental
Psychology, 56(A), 1373–1388.
Simons, D. J. (1996). In sight, out of mind: When object representations
fail. Psychological Science, 7, 301–305.
Sperling, G. (1960). The information available in brief visual presentations.
Psychological Monographs, 74(11, Whole No. 498), 1–29.
Stone, J., & Fukuda, Y. (1974). Properties of cat retinal ganglion-cells—
Comparison of w-cells with x-cells and y-cells. Journal of Neurophys-
iology, 37, 722–748.
Tassinari, G., Marzi, C. A., Lee, B. B., Di Lollo, V., & Campara, D. (1999).
A possible selective impairment of magnocellular function in compres-
sion of the anterior visual pathways. Experimental Brain Research, 127,
391– 401.
Theeuwes, J. (1991). Exogenous and endogenous control of attention: The
effect of visual onsets and offsets. Perception & Psychophysics, 49,
83–90.
Theeuwes, J. (1992). Perceptual selectivity for color and form. Perception
& Psychophysics, 51, 599 – 606.
Theeuwes, J. (2004). No blindness for things that do not change. Psycho-
logical Science, 15, 65–70.
Tipper, S. P., & Weaver, B. (1998). The medium of attention: Location-
based, object-centered, or scene-based? In R. Wright (Ed.), Visual at-
tention (pp. 77–107). London: Oxford University Press.
Todd, J. T., & van Gelder, P. (1979). Implications of a transient-sustained
dichotomy for the measurement of human performance. Journal of
Experimental Psychology: Human Perception and Performance, 5, 625–
638.
Usher, M., & Donnelly, N. (1998, July). Visual synchrony affects binding
and segmentation in perception. Nature, 394, 179 –182.
Watson, D. G., & Humphreys, G. W. (1995). Attention capture by contour
onsets and offsets: No special role for onsets. Perception & Psychophys-
ics, 57, 583–597.
Watson, D. G., & Humphreys, G. W. (1997). Visual marking: Prioritizing
selection for new objects by top-down attentional inhibition of old
objects. Psychological Review, 104, 90 –122.
Wolfe, J. M., Friedman-Hill, S. R., Stewart, M. L., & O’Connell, K. M.
(1992). The role of categorization in visual search for orientation.
Journal of Experimental Psychology: Human Perception and Perfor-
mance, 18, 34 – 49.
Yantis, S., & Egeth, H. E. (1999). On the distinction between visual
salience and stimulus-driven attentional capture. Journal of Experimen-
tal Psychology: Human Perception and Performance, 25, 661– 676.
Yantis, S., & Gibson, B. S. (1994). Object continuity in apparent motion
and attention. Canadian Journal of Experimental Psychology, 48, 182–
204.
Yantis, S., & Hillstrom, A. P. (1994). Stimulus-driven attentional capture:
Evidence from equiluminant visual objects. Journal of Experimental
Psychology: Human Perception and Performance, 20, 95–107.
Yantis, S., & Johnson, D. N. (1990). Mechanisms of attentional priority.
Journal of Experimental Psychology: Human Perception and Perfor-
mance, 16, 812– 825.
Yantis, S., & Jones, E. (1991). Mechanisms of attentional selection: Tem-
porally modulated priority tags. Perception & Psychophysics, 50, 166 –
178.
Yantis, S., & Jonides, J. (1984). Abrupt visual onsets and selective atten-
tion: Evidence from visual search. Journal of Experimental Psychology:
Human Perception and Performance, 10, 601– 621.
Yantis, S., & Jonides, J. (1990). Abrupt visual onsets and selective atten-
tion: Voluntary versus automatic allocation. Journal of Experimental
Psychology: Human Perception and Performance, 16, 121–134.
Yeshurun, Y. (2004). Isoluminant stimuli and red background attenuate the
effects of transient spatial attention on temporal resolution. Vision Re-
search, 44, 1375–1387.
Yeshurun, Y., & Levy, L. (2003). Transient spatial attention degrades
temporal resolution. Psychological Science, 14, 225–231.
Received January 11, 2005
Revision received June 13, 2005
Accepted June 20, 2005 䡲
72
PINTO, OLIVERS, AND THEEUWES