Content uploaded by Berno Bucker
Author content
All content in this area was uploaded by Berno Bucker on Dec 10, 2014
Content may be subject to copyright.
Distractors that signal reward attract the eyes
Berno Bucker, Artem V. Belopolsky, and Jan Theeuwes
Department of Cognitive Psychology, Vrije Universiteit Amsterdam,
Amsterdam, The Netherlands
(Received 28 May 2014; accepted 20 October 2014)
Salient stimuli and stimuli associated with reward have the ability to attract both
attention and the eyes. The current study exploited the effects of reward on the well-
known global effect in which two objects appear simultaneously in close spatial
proximity. Participants always made saccades to a predefined target, while the colour of
a nearby distractor signalled the reward available (high/low) for that trial. Unlike
previous reward studies, in the current study these distractors never served as targets. We
show that participants made fast saccades towards the target. However, saccades landed
significantly closer to the high compared to the low reward signalling distractor. This
reward effect was already present in the first block and remained stable throughout the
experiment. Instead of landing exactly in between the two stimuli (i.e., the classic global
effect), the fastest eye movements landed closer towards the reward signalling distractor.
Results of a control experiment, in which no distractor-reward contingencies were
present, confirmed that the observed effects were driven by reward and not by physical
salience. Furthermore, there were trial-by-trial reward priming effects in which saccades
landed significantly closer to the high instead of the low reward signalling distractor
when the same distractor was presented on two consecutive trials. Together the results
imply that a reward signalling stimulus that was never part of the task set has an
automatic effect on the oculomotor system.
Keywords: Reward; Attention; Eye movements; Priming; Global effect.
Visual selective attention is crucial in order to function and survive in our visually
rich world. While sampling information from the visual environment, selective
attention determines which parts of the visual scene are prioritized for further
processing and which parts are ignored. Typically, visual attention is categorized
into goal-directed (top-down) attention and stimulus-driven (bottom-up) attention
Please address all correspondence to Berno Bucker, Department of Cognitive Psychology, Vrije
Universiteit Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam, The Netherlands. E-mail:
berno.bucker@vu.nl
This research was supported by an ERC advanced grant [ERC-2012-AdG-323413] to Jan Theeuwes.
© 2014 Taylor & Francis
Visual Cognition, 2014
http://dx.doi.org/10.1080/13506285.2014.980483
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
(Chelazzi et al., 2013; Corbetta & Schulman, 2002; Desimone & Duncan, 1995;
Itti & Koch, 2001; Theeuwes, 2010). Top-down attention is driven by endogenous
factors and is completely under volitional control. Bottom-up attention is driven
by exogenous factors and is determined by the feature properties present in the
environment. An observer can voluntarily choose what to select from the
environment according to specific goals and priorities, while physically salient
stimuli may automatically attract the observer’s attention in a passive way. In
addition to goal-directed and stimulus-driven attention, a recent body of literature
suggests that attention can be influenced by the value coupled to reward predicting
stimuli (see Anderson, 2013; Chelazzi et al., 2013 for reviews). Since the bottom-
up and top-down dichotomy fails to explain a growing number of reported
selection biases due to reward value, it has been proposed that “reward history”is
integrated with task-goals and physical salience to shape an integrated priority
map (Awh, Belopolsky, & Theeuwes, 2012).
While it was already recognized that the knowledge of reward availability
works as an incentive to enhance voluntary attentional processes resulting in
fast and accurate responses (e.g., Bucker & Theeuwes, 2014; Engelmann,
Damaraju, Padmala, & Pessoa, 2009; Small et al., 2005), recent evidence
suggests that reward can automatically influence visual attention beyond and
sometimes even against the strategic control of goal-directed attention (see
Chelazzi et al., 2013, for a review). Several recent studies have shown that a
stimulus that has been previously associated with high monetary reward
captures attention (Anderson, Laurent, & Yantis, 2011a,2011b; Della Libera
& Chelazzi, 2009; Failing & Theeuwes, 2014) and the eyes (Anderson &
Yantis, 2012; Theeuwes & Belopolsky, 2012) more strongly than that same
stimulus when previously associated with low monetary reward. Typically these
studies include an initial training phase, in which participants actively search
for a stimulus that is either associated with high or low reward. In a subsequent
test-phase, when the reward association is no longer in place and the previously
trained stimulus is no longer the target (but a distractor instead), the previously
high rewarded stimulus captures attention and the eyes more strongly than the
low rewarded stimulus.
For example, experiments of Anderson, Laurent, and Yantis (2011a,2011b)
contained a training- and test-phase. In the training-phase, observers actively
searched for either one of two specifically coloured target circles amongst
differently coloured distractor circles and reported the orientation of a bar within
the target circle. One of the target colours was associated with a high chance (80%)
of obtaining a high reward and a low chance (20%) of obtaining a low reward. For
the other colour, the colour-reward contingencies were reversed. Immediately after
the training-phase, there was a test-phase in which no rewards were provided.
Observers were asked to search for a uniquely shaped target amongst differently
shaped and coloured distractors. Crucially, one of the distractors was presented in
the colour that was associated with high or low reward value as was learned during
2BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
the initial training-phase. Although participants were informed that colour was
task-irrelevant and had to be ignored, the results showed that reaction times were
slower if a stimulus previously associated with high reward was present relative to
a stimulus associated with low reward. The authors conclude that the value that
was associated with a specific stimulus feature during training biased attention
towards that feature, even in a different task context, in which it was non-salient,
task-irrelevant and non-rewarded.
In addition to reward-based capture of covert attention, several studies have
investigated whether stimuli also capture the eyes as a consequence of previous
stimulus-reward associations. Emphasizing the importance of eye movement
measures as direct evidence for salience-based attentional capture by learned
reward value, Theeuwes and Belopolsky (2012) used a reward variant of the
oculomotor capture paradigm of Theeuwes et al. (1998). Similar to Anderson,
Laurent, and Yantis (2011a) there was a training-phase in which one stimulus
orientation was associated with high reward and another with low reward.
During the subsequent test-phase, these stimuli served as distractors while
observers searched for a colour singleton. The results showed that saccades were
directed more often towards the stimulus that was previously associated with
high reward compared to the stimulus that was associated with low reward. For
the first time, this study showed that even when a stimulus no longer predicts
reward, the learned value increases oculomotor capture beyond oculomotor
capture that is driven by physical salience alone. Using the same experimental
design, with a training- and testing-phase, Anderson and Yantis (2012) came to a
similar conclusion when they showed oculomotor capture by previously
rewarded stimuli during unconstrained viewing when neither eye movements
nor fixations were required.
It is important to note that in all of the above mentioned experiments the
reward-associated distractors that captured attention were always actively
searched for during earlier trials. During the initial training-phase, the stimulus-
reward association was established by reinforcing repeated goal-directed
behaviour towards rewarded targets. Therefore, the reward effects that are
described in these studies might have depended on prior search for reward
associated stimuli. Even though these studies are assumed to provide evidence
for automatic reward effects of task-irrelevant distractors, it is clear that the effect
may not be as automatic as assumed as these distractors were targets during an
initial training-phase. This raises the possibility that automatic capture by highly
rewarding stimuli in these types of paradigms is necessarily dependent on being
task relevant and highly rewarded during an earlier training-phase.
A very recent study by Le Pelley, Pearson, Griffiths, and Beesley (in press)
addressed the question whether stimuli that had never been actively searched
could nevertheless induce reward driven attentional capture. As in the previously
described reward studies, the additional singleton paradigm was used. Partici-
pants searched for a diamond-shaped target among circles and made a response
DISTRACTORS THAT SIGNAL REWARD 3
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
depending on the orientation of a line segment within the diamond. All stimuli
were presented in grey, but on some trials one of the non-target circles (i.e., the
distractor) was coloured. The distractor colour predicted the reward magnitude
(high or low) available on that trial, but obtaining the reward depended on
responding to the line segment within the diamond-shaped target and not to the
distractor. Furthermore, responses to the target had to be faster than a certain
reaction time threshold, so that attending to the distractor hindered performance.
Nevertheless the results show that reaction times on trials in which a high-value
distractor was present were higher than on trials in which a low-value distractor
was present, with the consequence that participants were more likely to miss the
high compared with the low reward. Although participants were never instructed
to search for the distractors before, the simple correlation of the distractor stimuli
with reward was sufficient for attentional capture to occur.
To investigate whether the reward signalling distractors also captured the
eyes, a follow-up eye tracking experiment was conducted. In this experiment,
participants had to make an eye movement towards a filled grey diamond-shaped
target amongst grey distractor circles. Again, on some trials one of the non-target
circles was coloured to signal the reward available on that trial (high or low).
When responses were slower than 600 ms and if any gaze fell inside a predefined
area surrounding the distractor before a response to the target was registered, no
reward was delivered. In accordance with the behavioural results of the first
experiment, saccades were slower in high versus low reward distractor trials and
more rewards were missed in high versus low reward distractor trials, indicating
that high value distractors produced greater oculomotor capture than low value
distractors. Therefore, the authors conclude that reward associated distractors can
capture attention and the eyes, even if this is counterproductive.
The current study expands on this latter finding and investigates whether the
eyes are automatically drawn by reward signalling distractors that never served as
targets. While Le Pelley and colleagues made use of a visual search task with
multiple distractors and a 600 ms response window to obtain reward, we
investigated very fast eye movements in the context of the global effect paradigm
(Coren & Hoenig, 1972; Findlay, 1982; see Van der Stigchel & Nijboer, 2011, for a
review). The notion underlying the global effect is that eye movements typically
land on the centre of gravity within the visual field, which is assumed to reflect the
relative salience of elements present in the visual field. The global effect is
observed when participants initiate an eye movement towards two stimuli
presented simultaneously in close spatial proximity. Instead of moving to one of
the two objects, typically the initial saccade lands in between the two stimuli. This
landing position is likely to reflect the unresolved competition between the
representations of the stimuli (Tipper, Howard, & Jackson, 1997). In line with
saccade generation models based on competitive interactions between subsets of
neurons coding for possible target locations (Fecteau & Munoz, 2006; Godijn &
Theeuwes, 2002), it appears that the global effect can be best described in terms of
4BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
a weighted average of activity in a saccade map. The two stimuli that appear in
close proximity with a simultaneous onset produce two peaks of activity in this
saccade map. The weighted average of activity that is located in between these two
peaks then determines the saccade endpoint on the moment the eye movement is
initiated. While stimulus-driven information can evoke peaks in the activity map,
goal-directed factors can modulate this activity. Importantly, the interaction
between stimulus-driven and goal-directed information determines the pattern of
activity in the saccade map, with stimulus-driven information being dominant
early in time and goal-directed information becoming more dominant with
increasing latency (Van Zoest, Donk, & Theeuwes, 2004).
In the current study, we conducted a reward experiment and a non-reward
control experiment. Both experiments consisted of the same task with and
without stimulus-reward contingencies in place. During a trial, a grey target
circle and a coloured distractor circle were presented simultaneously in close
spatial proximity. Throughout both experiments distractors never served as
targets. Participants were instructed to make a fast saccadic eye movement
towards the target circle, and to ignore the coloured distractor circle. Crucially in
the reward experiment the colour of the distractor circle signalled the reward
magnitude available (high or low) for that particular trial. Reward delivery was
dependent on making a fast saccade towards the circles. In the separate control
experiment, different participants performed the same task only without the
distractor-reward contingencies. The purpose of this experiment was to deter-
mine whether there were any effects of physical salience independent of reward.
On the basis of the competitive integration model (Godijn & Theeuwes,
2002), one expects stimulus-driven information to be dominant early in time and
goal-driven information to become more dominant with increasing saccade
latencies. As participants are provided with goal-directed information to make an
eye movement to the target circle, one expects the landing position to shift more
towards the target location for slower saccades. If reward value has a goal-driven
effect, one expects that the effect of reward value would become evident with
increasing latencies. With increasing latencies there is more and more room for
top-down influences to modulate the peak activity of the reward signalling
distractor and to progressively enhance the representation of the high compared
with the low value distractor in the saccade map. However, if reward value has
an effect even for the fastest saccades, at the time window where bottom-up
processes normally operate (c.f., Godijn & Theeuwes, 2002), then one should
conclude that reward value does not necessarily exert a top-down effect but
instead an effect associated with bottom-up automatic processes. Based on the
physical salience of the target and distractor alone, one expects two equally sized
peaks of activity in the saccade map, resulting in a global effect with an average
landing position perfectly in the middle of the two objects. However, if reward
value exerts an early effect that is similar to classic bottom-up effects, one
expects that the reward signalling distractor would be prioritized in the saccade
DISTRACTORS THAT SIGNAL REWARD 5
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
map. As it is assumed that the presence of a high reward signalling distractor will
evoke a greater peak of activity than a stimulus associated with low reward, it is
expected that the eyes are biased to land closer towards the high compared with
the low reward signalling distractor. To show that the effect is dependent on the
stimulus reward-contingencies introduced in the reward experiment and not on
other task specific features such as physical salience, the control experiment
should show a different pattern of results, without any bias towards the distractor.
METHOD
Participants
For the reward experiment, 18 participants (five males, 19–28 years of age, mean
= 23.8 years, SD = 2.4 years) were tested at the Vrije Universiteit Amsterdam.
These participants received €9.00 to compensate for participation and could earn
up to a maximum of approximately €15.00 extra reward, which was delivered
based on task performance. For the control experiment 14 new participants (four
males, 19–34 years of age, mean = 24.4 years, SD = 4.6 years) were recruited
and received €9.00 to compensate for participation. All participants reported
having normal or corrected-to-normal vision and gave informed consent before
participation. All research was approved by the Vrije Universiteit Faculty of
Psychology ethics board and was conducted according the principles of the
Declaration of Helsinki.
Apparatus
All participants were tested in a sound-attenuated, dimly-lit room, with their head
resting on a chinrest at a viewing distance of 58 cm. A Pentium IV computer
(2.3 GHz) generated all stimuli on a 21-inch SVGA monitor (resolution 1024 x 768
pixels, refreshing at 100 Hz). Monocular movements were tracked using the Eyelink
1000 system (Tower model, SR Research Ltd, Canada), an infrared video-based eye
tracker that has a 1000 Hz temporal resolution and a 0.01° spatial resolution.
Design
We conducted two experiments. In the reward experiment participants made
saccades as fast as possible towards a grey target circle, while the colour (red or
green) of the distractor circle signalled the reward magnitude (high or low)
available on that trial. In the control experiment, a group of different participants
performed the same task as in the reward experiment only without the stimulus-
reward contingencies.
6BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Stimuli
Throughout both experiments a white (CIE: x = .270, y = .309; 32.7 cd/m
2
)
fixation dot was presented on a black (CIE: x = .240, y = .323; 1.5 cd/m
2
)
background at the centre of the screen. The stimulus display contained two filled
circles that had the same size (r= 0.14°), were located within the same quadrant
of the screen (boarders: 0°, 90°, 180°, 270°) at the same distance (7.7°) from
fixation and always positioned 3.4° visual degrees (25° polar angle) apart from
each other. Instead of placing these two circles at fixed locations around the four
principal axes (45°, 135°, 225°, 315°), they were placed at a random location
within a quadrant. This is unlike other studies investigating the global-effect
(e.g., Heeman, Theeuwes, & Van der Stigchel, 2014; Silvis & Van der Stigchel,
2013; Van der Stigchel, Heeman, & Nijboer, 2012) as we intended to reduce the
influence of top-down expectancy (or guessing) of where the stimuli could
appear within the visual field. Furthermore, this procedure circumvented
reactivation (or planning) of eye movement trajectories from previous trials.
The two circles always consisted of a grey (CIE: x = .264, y = .316; 6.2 cd/m
2
)
target circle and a red (CIE: x = .534, y = .328; 5.9 cd/m
2
) or green (CIE: x = .295, y
= .530; 8.5 cd/m
2
) distractor circle. In the reward experiment the colour of the
distractor circle signalled the reward magnitude (high or low) available on that trial,
whereas there were no associations with reward in the control experiment. In the
reward experiment, one of the distractor colours signalled high reward (10
eurocents), while the other colour signalled low reward (1 eurocent). Colour-
reward contingencies were counterbalanced across participants. If saccades were
accurate (see Latency Threshold and Accuracy) and faster than the saccade latency
threshold (which was dynamically adjusted, see Latency Threshold and Accuracy),
the white (CIE: x = .270, y = .309; 32.7 cd/m
2
) feedback text “+€10 ct”(∼.4° x
1.0°) or “+€1ct”(∼.4° x .9°) was shown for high and low reward distractor trials,
respectively. Note that no reward feedback was shown in the control experiment. If
the landing position of the saccade was considered inaccurate a 500 Hz warning
tone was played for 100 ms simultaneously with the visually presented feedback
text “More accurate”(∼.4° x 1.6°). If the saccade latency exceeded the saccade
latency threshold a 500 Hz warning tone was played for 100 ms simultaneously
with the visually presented feedback text “Too slow”(∼.4° x 1.0°).
Procedure
Participants signed the informed consent and the eye tracker was calibrated. If at
any point eye movements would drift, another calibration procedure would be
performed. Participants were asked to keep their head still during the trials, but
were free to move their head during the breaks between blocks.
Both experiments began with a practice block followed by 10 experimental
blocks. In the reward experiment, during the practice block, the distractor circle
DISTRACTORS THAT SIGNAL REWARD 7
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
was coloured yellow (CIE: x = .386, y = .518; 13.0 cd/m
2
) or blue (CIE: x =
.173, y = .123; 4.5 cd/m
2
) in order to prevent the formation of any bias with
regard to the reward associated red and green coloured distractors used in the
main part of the experiment. In the control experiment, during the practice
block, the distractors were coloured red and green, as these colours were never
associated with reward. Furthermore, in the control experiment, during practice,
participants were informed on a trial-by-trial basis towards which circle their
eye movement landed closest to. The main part of both experiments consisted
of 320 trials divided into 10 equally sized blocks. Within a block, reward
distractor type (high/low), the quadrant in which the stimuli appeared (1–4) and
distractor position relative to the target (clockwise/counterclockwise) were
balanced.
A trial (see Figure 1) started with a drift correction in which participants were
required to press the spacebar while fixating the fixation cross. To indicate the
start of the trial the fixation cross was replaced by the fixation dot. After a
random variable interval of 500–1000 ms the fixation dot was removed,
immediately followed by the simultaneous appearance of the two circles.
Subjects were instructed to move their eyes towards the grey target circle as
fast as possible while ignoring the distractor circle. Before the reward
experiment, participants were informed that the distractor colour was predictive
of the available reward for that trial. It was also stressed that the distractor had to
be ignored and that an eye movement had to be made as fast as possible towards
the grey target. Once the first saccade was made, the circles remained present on
the screen for 100 ms followed by the visually presented feedback for 400 ms.
After each block the mean saccade latency of that particular block was displayed
along with the saccade latency threshold for the first trial of the following block.
In addition the amount of reward obtained in that particular block and the
accumulated reward amount over all blocks was displayed on the screen in the
reward experiment only. The feedback screen that appeared between blocks
remained visible until a key was pressed. Including the calibration procedure and
breaks, participants were able to finish the experiment within approximately 50
minutes.
Latency threshold and accuracy
At stimulus display onset, participants were instructed to make an eye movement
towards the grey target circle and ignore the coloured distractor circle. It was
emphasized that eye movements had to be made as fast as possible and in the
reward experiment participants were made explicitly aware that in order to obtain
the reward for that trial, eye movements had to be faster than a dynamically
adjusting saccade latency threshold. For each participant, the saccade latency
threshold was based on the latencies in the 20 previous trials. From the latencies
of these 20 trials, the 75th percentile was calculated and set as the latency
8BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Landing position (φ)
_
-1
-1
+1
+1
0
+
φ
= .45
= .45
Drift correction
(key press)
Fixation
(500-1000 ms)
Stimulus display
(wait for saccade)
Feedback
(400 ms)
+ € 10 ct
Saccade interval
(100 ms)
φ
Figure 1. Schematic representation of trial sequence and timing on the left. Schematic representation of how saccade landing position was calculated on the right. The
“midline angle”exactly in the middle of both circles served as the null (ϕ= 0.0) reference. Saccades landing towards the target circle (ϕ= 1.0) were defined as having a
positive landing position and saccades landing towards the distractor circle (ϕ=−1.0) as having a negative landing position. In the figure an example saccadic angle of
ϕ= .45 is shown. Dashed lines were not visible in the actual stimulus display. Note that the reward feedback screen was not present in the control experiment.
DISTRACTORS THAT SIGNAL REWARD 9
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
threshold for the next trial. Obtaining reward depended on saccade latency and
accuracy in landing position. In order to be considered accurate, the first saccades
had to land in one of three imaginary circles (r=1.7°) (one around the target, one
around the distractor and one around the point exactly in the middle of the target
and the distractor). We implemented this accuracy constraint so that participants
made saccades towards the stimuli presented on the screen. Saccades landing
closer to the distractor instead of the target were considered accurate and not
explicitly punished, because this would slow down overall latencies making it
more difficult to investigate the global effect. In the reward experiment, if the first
saccade was accurate and faster than the dynamic saccade latency threshold,
participants received the reward that was signalled by the colour of the distractor
circle on that trial.
Preprocessing
An eye movement was considered a saccade when either eye velocity exceeded
35°/s or eye acceleration exceeded 9500°/s
2
and end points were defined as the
location where velocity fell below this threshold. First, all trials in which the first
saccade was not accurate were not further analyzed. Second, trials with a landing
position of more than two and a half standard deviations away from the
participants’mean were excluded from the analysis. Third, trials were filtered on
saccadic latency with a minimum of 80 ms (anticipatory saccades) and a
maximum of 450 ms (too slow saccades). Saccade latency was defined as the
interval between the presentation of the circles and the initiation of the first
saccadic eye movement.
Landing position of the first eye movement was calculated as a proportion of
the angle between the target and the distractor. The geometric point exactly in
the middle of the two circles served as the null reference for the landing
position (φ= 0.0). Saccades that landed towards the target were defined as
having a positive landing position and saccades that landed towards the
distractor were defined as having a negative landing position. The grey target
circle had position one (φ= 1.0) and the reward signalling distractor had
position minus one (φ=−1.0). To compensate for small drift (< 1°) of the eye
movements from fixation at the start of the saccade, the actual starting point of
the saccade was used to calculate the landing position (φ). A schematic
representation of how saccade landing position was calculated is shown in
Figure 1.
Statistical analysis
All trials after preprocessing were categorized into high and low reward
distractor type trials. Two tailed t-tests were performed for landing position
and saccade latency to examine the effect of reward distractor type (high/low).
10 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
To investigate the time course of our reward manipulation on saccade landing
position, the trials of each participant were divided into four latency quartiles for
the high and low reward trials separately (i.e., data were Vincentized). A repeated
measures ANOVA with reward distractor type (high/low) and latency bin (1–4)
was performed on landing position.
In relation to reward learning effects, we examined the landing position for
high and low reward distractor trials over time (blocks). In order to retain a
sufficient number of trials per cell we grouped consecutive blocks such that the
analysis was performed on five blocks. A repeated measures ANOVA with
reward (high/low) and block (1–5) as factors was performed on landing position.
Furthermore we investigated the more transient, trial-by-trial, priming effect of
the reward signalling distractors. For this analysis, we examined the difference in
landing position given that the identity of the distractor changed or remained the
same in two consecutive trials. The first trial of each block was excluded, since no
trial directly preceded these trials. A repeated measures ANOVA with distractor
repetition (same/different) and current reward distractor type (high/low) as factors
was performed on landing position.
With regard to the control experiment, we compared the landing position and
the saccade latency with the reward experiment by means of two-tailed
independent samples t-tests. Furthermore, a repeated measures ANOVA with
factor latency bin (1–8) was performed on landing position. Because there was
no reward assigned to either of the colours, we collapsed over red and green
distractor trials to retain a sufficient amount of data per cell to divide the data of
the control experiment into eight quantiles.
RESULTS
Exclusions
In the reward experiment, the exclusion criteria led to a total loss of 6.8% of the
trials. With regard to landing position, 5.9% of the data were discarded due to
inaccurate eye movements and 0.01% due to a difference angle of more than two
and a half standard deviations away from the calculated mean. With regard to
saccade latency, 0.73% of the trials were discarded because latency onset was
shorter than 80 ms and 0.16% of the trials were discarded because latency onset
was longer than 450 ms.
In the control experiment the exclusion criteria led to a total loss of 6.1% of
the data. More specifically, 3.8% of the data were discarded due to inaccurate
eye movements, 1.2% due to a difference angle of more than 2.5 standard
deviations away from the participants mean, 0.57% because of anticipatory
(latency < 80 ms) saccades and 0.61% because of too slow (latency > 450 ms)
saccades.
DISTRACTORS THAT SIGNAL REWARD 11
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Reward
Participants earned between €9.59 and €13.41 (mean = €12.23, SD =€0.92)
monetary reward. Paired samples t-test showed that there was no difference in
the likelihood of receiving a reward for high versus low reward distractor trials
(t(17) = 1.47, SE = 1.97, p= .16).
Distractor reward effect
A two tailed paired-samples t-test on landing position for high versus low
reward distractor type showed a significant effect (t(17) = 2.35, SE = 0.79,
p< .05), with saccades landing closer to the high versus low reward signalling
distractor. Even though overall saccades landed significantly closer to the
target than to the distractor (for high mean φ=0.11,SD =0.17andforlow
mean φ=0.26,SD = 0.18), in the high reward condition the eyes landed
significantly closer to the distractor than in the low reward condition. Figure 2
shows the frequency distributions of the landing positions for the high and low
reward signalling distractor conditions separately. A two tailed paired-samples
t-test for high and low reward distractor type saccade latency showed a
significant effect (t(17) = 4.45, SE = 0.67, p< .001). Although the absolute
difference between conditions was only 3.0 ms, saccades in the high reward
signalling distractor condition (mean = 202.0 ms, SD =15.1ms)werereliably
faster than in the low reward signalling distractor condition (mean = 205.0 ms,
SD =15.0ms).
To investigate the time course of the high and low reward distractor type
landing positions, for each participant we divided the high and low reward trials
in four latency bins. A repeated measures ANOVA on landing position with
reward distractor type (high/low) and latency bin (1–4) as factors showed a
significant main effect of reward distractor type (F(1,17) = 5.64, p< .05) and
latency bin (F(3,51) = 127.03, p< .001). There was no significant interaction
between reward distractor type and latency bin (F(3,51) = 2.31, p= 0.11
Greenhouse-Geisser corrected). The main effect of reward showed a sustained
difference in landing position between the high and low reward distractor
conditions (see Figure 3). The main effect of latency bin showed that saccades
with the shortest latencies (bin 1), landed closest to the distractor and saccades
with the longest latencies (bin 4) landed closest to the target. From bin 1 to bin 4
the landing position gradually shifted from the distractor location towards the
target location as indicated by a significant linear trend (F(1,17) = 204.48,
p< .001). Planned two tailed t-tests indicated that the fastest saccades did not
show the typical global effect (i.e., a mean landing position on the midline,φ=
0). The landing positions of the fastest saccades in the high reward distractor
condition (mean φ=−0.13, SD = 0.15) deviated significantly (t(17) = 3.70,
p< .01, d= 1.8) away from the midline (i.e., φ= 0) and the landing position of
12 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
the fastest saccades in the low reward distractor condition (mean φ=−0.08,
SD = 0.17) showed a trend (t(1,17) = 1.90, p= .07, d= .92) for deviating away
from the midline. These results indicate that with increasing latency the landing
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-3 03
Landing position (φ)
Percentage of trials
Percentage of trials
High reward
21
target
-1
distractor
-2
4.5
4.5
Low rewardLow reward
1
target
-1
distractor
Figure 2. Frequency plots of landing position for the low (above) and high (below) distractor reward
condition. The target was presented at ϕ= 1 and the reward signalling distractor was presented at ϕ=−1.
Note that in the high reward condition, the distribution is skewed more to the distractor than in the low
reward condition.
DISTRACTORS THAT SIGNAL REWARD 13
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
position of the first saccade shifted from a location near the distractor to a
location near the target.
Furthermore, we determined how the difference in landing position developed
over the course of the experiment, as participants were more exposed to the stimulus-
reward contingencies. A repeated measures ANOVA with reward distractor
type (high/low) and block (1–5) as factors showed a significant effect of block
(F(4,68) = 6.52, p<.001)andreward(F(1,17) = 5.70, p< .05). Notably there was no
significant interaction between reward and block (F(1,17) = 1.13, p=.35),
indicating that the difference between saccades made in the high and low reward
conditions remained constant over the course of the experiment (see Figure 4).
200150 250
High reward
Low reward
Saccade latency (ms)
Landing position (φ)
175 225 275
1
-0.5
-0.25
0
0.25
0.5
0.75
Target
Distractor
-0.75
-1
Midline
Figure 3. Mean landing position of the first saccade divided into four latency bins for the high and low
reward distractor condition. Error bars represent 95% within-subject confidence intervals.
14 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Interestingly, the reward information already showed a near significant effect in the
first block of the experiment (t(17) = 1.88, p=.077,d=0.91),withsaccadeslanding
closer to the high compared to the low reward signalling distractor. With regard to
the main effect of block (1–5), as indicated by a significant linear trend (F(1,17) =
10.28, p< .01), the landing position shifted more and more away from the target.
This shift towards the midline (i.e., the global effect becoming more global) was
probably related to our dynamically adjusted latency threshold, which made
participants make faster and faster saccades towards the end of the experiment.
Trial-by-trial reward priming
To investigate inter-trial reward priming we examined the difference in landing
position given that the identity of the distractor changed or remained the same
in two consecutive trials. A repeated measures ANOVA with distractor
repetition (same/different) and current reward distractor type (high/low) as
factors showed a main effect of current reward distractor type (F(1,17) = 5.26,
p< .05), reflecting the reward effect on landing position. Yet crucially for the
Time (block)
Landing position (φ)
High reward
Low reward
1
-0.25
0
0.25
0.5
0.75
Midline
Target
target
distractor
53421
Figure 4. Mean landing position (ϕ) plotted over time (blocks) for the high and low reward distractor
condition. Error bars represent 95% within-subject confidence intervals.
DISTRACTORS THAT SIGNAL REWARD 15
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
present analysis, there was a significant interaction between distractor
repetition and current reward distractor type (F(1,17) = 8.82, p<.01),
indicating that the landing position for high and low reward distractor trials is
modulated by whether the distractor on the current trial is the same or different
with regard to the previous trial. Subsequent two-tailed t-tests showed that the
landing position in a high reward distractor trial resulted in a significantly
larger deviation towards this high reward distractor (t(17)= 2.64, SE = 0.03,
p< .05) when it was preceded by a high reward distractor trial (same)
compared to the condition in which it was preceded by a low reward distractor
trial (different). This suggests that a high reward signalling distractor has a
strong inter-trial priming effect on landing position of the current trial.
Crucially, landing position did not significantly differ on current low reward
signalling distractor trials (t(17) = 1.29, p= .21), depending on whether these
were preceded by high (different) or low (same) reward signalling distractors.
As is clear from Figure 5, priming only occurred for the high and not for the
low reward signalling distractor. The high reward signalling distractor colour
in the previous trial, caused saccades to land significantly closer to the high
reward signalling distractor in the current trial compared to the situation in
which a low reward signalling distractor was presented in the previous trial.
Possibly the high reward value that was coupled to the high reward signalling
distractor colour caused reward priming over trials, such that the eyes were
attracted significantly stronger to the high reward signalling distractor if the
0.15
0.1
0.05
0
-0.05
-0.1
0.4
0.35
0.3
0.25
0.2
Landing position (φ)
ns
*
target
distractor
Current high reward
Current low reward
Low - Low
High - Low
Low - High
High - High
Condition (previous-current)
Figure 5. Mean landing position (ϕ) for current low (green) and high (red) reward distractor present
trials, following previous low (Low-Low and Low-High) and high (High-Low and High-High) reward
distractor trials. Note that reward priming is only observed for repeated presentation of high reward
distractors (High-High). Error bars represent 95% within-subject confidence intervals.
16 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
distractor identity remained the same compared with a distractor identity
switch.
Control experiment
In the control experiment, a new group of 14 participants performed the same
task as the main experiment without the stimulus-reward contingencies. The
purpose of this experiment was to ensure that the global bias towards
distractors in the main experiment was due to their association with reward
and not to differences in physical salience. Note that in the main experiment
the target colour was always grey while the distractors were always coloured.
As in the main experiment, we asked participants in the control experiment to
make saccades as fast as possible towards the grey target circle. However,
unlike in the main experiment, no rewards were given indicating that
there was no association between the colour of the distractor and a possible
reward.
A between experiment t-test showed that in the control experiment saccades
landed significantly closer to the target (t(30) = 18.0, SE = 0.05, p< .001) than
in the reward experiment. Also, participants were overall slower in the control
experiment (t(30) = 4.00, SE = 6.44, p< .001). The saccade latencies of the
control experiment were divided into eight quantiles which allowed us to
compare the saccades latencies of the fastest saccades in the main experiment
with those in the control experiment. An ANOVA on landing position with
latency bin (1–8) as a factor showed a significant linear trend (F(1,13) = 95.77,
p< .001) with saccades landing progressively closer towards the target location
with increasing latencies. Crucially, a planned two-tailed t-testshowedthatthe
fastest saccades in the control experiment already showed a significant bias
(t(13)= 5.98, SE = 0.25, p< .001) away from the midline (i.e., φ=0)towards
the grey target circle (see Figure 6). This is unlike the results of the main
experiment, where we observed a bias for the fastest saccades to land closer to
the reward signalling distractor. The effect in the control experiment was
robust, with 12 out of 14 participants showing a strong bias for landing closer
towards the grey target circle for the fastest saccades (mean landing position
between φ=0.22andφ= 0.82), while the two other subjects showed the
typical global effect (mean φ=−0.05 and φ=−0.06). Together these results
suggest that in the main experiment it was the reward association and not
physical salience that caused the eyes to be attracted more to the reward
signalling distractors.
DISTRACTORS THAT SIGNAL REWARD 17
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
DISCUSSION
In the present study we investigated whether reward-signalling distractors, that
never served as targets, were able to attract the eyes in the context of the global
effect paradigm. Participants were instructed to make an eye movement to the
target circle as fast as possible, while the colour of the distractor circle signalled
the reward available for that trial. We show that even though participants made
fast saccades towards the target, their eyes landed significantly closer to the
stimulus that signalled the availability of a high compared to the low reward.
This reward effect was already present in the first block and remained stable
throughout the experiment. The time course analysis indicated that this effect did
not change with increasing latency, suggesting that the reward information
carried by the distractors was present over the full latency range. Crucially,
instead of landing exactly in between the equally salient target and distractor
(i.e., the classic global effect), the fastest eye movements (around 165 ms) landed
closer to the reward signalling distractor than to the equally salient target. Results
of a control experiment without any reward signalling stimuli provided strong
evidence that it was the reward association and not another specific task feature
(e.g., physical salience) that caused the eyes to be attracted more to the reward
signalling distractors. In addition to this sustained reward effect, we show trial-
by-trial reward priming with saccades landing significantly closer to the high
200 052051
Saccade latency (ms)
175 225 275 300 325
Tar g e t
Midline
Landing position (φ)
1
0
0.25
0.5
0.75
target
distractor
-0.25
Figure 6. Mean landing position (ϕ) of saccades in the control experiment divided into eight latency
bins. Note that the fastest eye movements already show a bias towards the target (i.e., ϕ> 0). Error bars
represent 95% within-subject confidence intervals.
18 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
reward signalling distractor but not to the low reward signalling distractor if the
same value distractor was presented for two consecutive trials. Together these
results imply that reward signalling stimuli elicit a rather automatic and
involuntary effect on the oculomotor system that manifests itself very early
in time.
Our results are consistent with other studies demonstrating that reward
learning influences oculomotor behaviour (Anderson & Yantis, 2012; Theeuwes
& Belopolsky, 2012). These studies have shown that, following a training-phase,
distractor stimuli associated with high reward attracted the eyes more strongly
than the very same distractor stimuli when associated with low reward. However,
unlike these previous studies, the reward signalling distractor stimuli in the
present study were never targets and were never actively searched in a preceding
training-phase. Nevertheless, we show a sustained reward effect, with the eyes
landing significantly closer to the high compared with the low reward signalling
distractor. Crucially, this implies that oculomotor capture by reward associated
distractor stimuli is not dependent on whether the distractor was previously a
target in an earlier training-phase or not.
This is consistent with Le Pelley et al. (in press) who showed that in a visual
search task with multiple distractors, relative to a low reward signalling distractor
a high reward signalling distractor captured attention and the eyes more, even
though the distractors were not actively searched. In multiple behavioural
experiments, Le Pelley et al. (in press) showed that participants responded slower
in a visual search task, when a high versus a low reward signalling distractor was
present. In a follow-up eye tracking experiment, initial saccades went
significantly more often to the high compared with the low value distractor,
while participants were searching for a target stimulus. Consistent with Le Pelley
et al. (in press) we also observed an effect of reward on the initial saccades.
While Le Pelley and colleagues utilized a visual search task with multiple
distractors (of which the reward signalling distractors were more salient than the
other non-target elements) and a fixed latency threshold of 600 ms, we
investigated very fast eye movements (as fast as 165 ms) in the context of the
global effect paradigm. Our results showed that the eyes were more attracted by
high compared to low reward signalling distractors and that this effect remained
stable over latency bins and the course of the experiment. Notably, the fastest eye
movements landed closer towards the reward signalling distractor instead of
landing in the middle (i.e., typical global effect) or landing closer towards the
target. Furthermore, we showed that high and not low reward signalling
distractors elicited a reward priming effect, with the eyes being attracted more
by high value signalling distractors after repeated presentation. Together these
studies provide converging evidence that previous selection of reward-associated
items in a training-phase may not be the driving force behind reward capture in a
later testing-phase. Rather, the simple correlation of stimuli with reward causes
DISTRACTORS THAT SIGNAL REWARD 19
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
the eyes to be drawn more to those stimuli, even if they never served as targets
before.
With regard to the time course of saccades, the results of the present study are
in line with a recent study investigating the influence of top-down control on the
global effect (Heeman et al., 2014). Similar to the present study, participants
were instructed to make an eye movement to a specifically coloured target circle,
while ignoring a differently coloured distractor circle. A significant linear effect
was observed for saccades landing closer to the target circle with increasing
latency. A comparable linear effect, with saccades landing closer to the target
with increasing latency, was observed in both the reward and the control
experiment. This increasing target bias is in line with the idea that goal-driven
information becomes more dominant with increasing latencies (Van Zoest, Donk,
& Theeuwes, 2004). However, the reward values introduced in the reward
experiment cause a notable difference with regard to the landing position of the
fastest saccades. Compared with a baseline condition in which participants had
not received a target instruction, Heeman et al. show that the fastest saccades
were already biased towards the target (i.e., φ> 0), results which are strikingly
similar to the results of our control experiment. In the reward experiment, we
observe a bias towards the reward signalling distractors (i.e., φ< 0) for the
fastest saccades. Although we cannot rule out the possibility that there are goal-
driven effects at the 165 ms saccade latencies in the reward experiment, a
possible top-down effect at these early time intervals is out weighted by the
automatic effect of the reward signalling stimulus. This implies that reward can
exert an effect very early in time that might be considered involuntary, since it
counteracted the current task-goal.
In terms of the earlier described competitive integration model (Godijn &
Theeuwes, 2002), these results indicate that early in time the reward signalling
distractor evoked a larger peak of activity in the saccade map compared with the
target. Crucially, based on the physical stimulus properties of the target and
distractor alone, one expects two equally sized peaks of activity in the saccade
map, resulting in a global effect with an average landing position perfectly in the
middle of the two circles (i.e., φ= 0). However, although both the target and
the distractors had the same physical salience, the reward value associated with
the distractor made it more “salient”compared to the target. The reward value
coupled to the distractors possibly altered their representation on a salience map
and increased the priority of the distractor over that of the target very early
in time.
The sustained reward effect with saccades landing closer to the high
compared with the low reward signalling distractor can similarly be explained
in the context of the competitive integration framework. That is, the high reward
distractor signals a higher value than the low reward distractor and therefore
evokes a greater peak of activity in the saccade map. This relative difference
caused the average activity in the saccade map to be located closer the distractor
20 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
on trials during which the high compared with the low value distractor was
present. Furthermore, the enlarged peak evoked by the high compared with the
low reward distractor is responsible for an absolute increase of average activity
in the saccade map, causing the saccade threshold to be reached earlier. This
might explain the observed latency difference. As suggested by Theeuwes and
Belopolsky (2012), it is possible that reward value lowered the threshold for
making a saccade, especially since we made use of a dynamic threshold to
ensure that saccades were made progressively faster in order to obtain reward. In
addition to the sustained reward effect, the competitive integration model can
also explain why saccades landed significantly closer to the target with
increasing saccade latency. Since the model assumes goal-driven control to
improve with increasing latency, the peak of activity that is evoked by the target
will be progressively enhanced with increasing saccade latency. Overall, the
current results can be well explained in the context of the competitive integration
model, with a very early factor enhancing higher reward value associated stimuli
and a goal-driven factor enhancing target activity with increasing latency.
In addition to the sustained reward effect, we investigated the transient process
of reward priming by examining the difference in landing position given that the
identity of the distractor changed or remained the same in two consecutive
trials. As is clear from Figure 5, saccades landed significantly closer to the high
reward signalling distractor when it was preceded by a high compared with a
low reward signalling distractor. Crucially, no such effect was observed for the low
reward signalling distractor, indicating that inter-trial priming was driven by the
reward value of the stimulus. Furthermore, it is important to realize that the
priming that we observed is not the more often described type of priming during
which repeated presentation of attended stimulus features facilitates detection of
such features (see Kristjánsson & Campana, 2010, for a review). Instead, here it
was the distractor that drew the eyes when it was repeatedly presented, although
participants were constantly making eye movements to the same target stimulus. In
the context of the global effect paradigm, Meeter and Van der Stigchel (2013) also
showed inter-trial priming, demonstrating that repeating the object colours
improved the target representation such that after target repetition, the eyes tended
to land closer to the target. They claimed that inter-trial priming caused a bottom-
up boost of the target representation. In our study however, repeating a trial with a
high reward signalling distractor caused the eyes to move away more from the
target and more towards the distractor. This suggests that in line with the target
boost mechanism of Meeter and Van der Stigchel (2013), inter-trial reward priming
caused a similar bottom-up boost of the distractor representation, increasing the
priority of the reward signalling distractor.
This hypothesis regarding the origin of inter-trial priming is consistent with a
reward study by Hickey, Chelazzi, and Theeuwes (2010), who showed that high
versus low reward associated colours automatically captured selective attention
on a trial-by-trial basis. In this study, the additional singleton paradigm was used
DISTRACTORS THAT SIGNAL REWARD 21
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
and participants received either a high or a low reward for correct answers.
Reward delivery was random on a trial-by-trial basis and the colours of items in
the display could remain the same or swap, with the colour of the target
becoming the colour of the distractors. The results showed that only after
receiving a high reward, responses were fast when the target colour remained the
same, but slow when the colours swapped. The high reward value associated
with the target features caused visual attention to be biased towards those
features in the upcoming trial. Even in a second experiment, where it was most
beneficial for participants to abandon the current reward attentional set, they
counterproductively continued to select the stimulus characterized by the colour
previously associated with the high reward value. Thus, despite being counter-
productive, the association of a colour with the high reward value automatically
changed the visual salience of stimuli in a way that is strong enough to negate
the impact of endogenous strategic deployment of attention. These results are
very much in line with the present study, since we show an increased bias
towards the high reward signalling distractor after repeated presentation,
although participants maintained a strategic attentional set to make an eye
movement towards the target at all times. Although here, the high and low
reward value were consistently coupled to the same colour and these colours
never served as target colours, we also show that a high and not a low reward
signalling stimulus automatically attracts attention and the eyes when it is shown
on two consecutive trials.
In sum, the present study shows that the global effect is influenced by the
mere presence of a stimulus that signals the availability of reward. Our data show
the effect of the reward signalling stimulus is already present at the fastest
saccade latencies suggesting an early, involuntary and automatic modulation of
the oculomotor system by reward signalling stimuli.
REFERENCES
Anderson, B. A. (2013). A value-driven mechanism of attentional selection. Journal of Vision,
13(3), 7.
Anderson, B. A., & Yantis, S. (2012). Value-driven attentional and oculomotor capture during goal-
directed, unconstrained viewing. Attention, Perception, & Psychophysics,74, 1644–1653.
Anderson, B. A., Laurent, P. A., & Yantis, S. (2011a). Learned value magnifies salience-based
attentional capture. PLoS One,6(11), e27926.
Anderson, B. A., Laurent, P. A., & Yantis, S. (2011b). Value-driven attentional capture. Proceedings
of the National Academy of Sciences,108, 10367–10371.
Awh, E., Belopolsky, A. V., & Theeuwes, J. (2012). Top-down versus bottom-up attentional control:
A failed theoretical dichotomy. Trends in Cognitive Sciences,16, 437–443.
Bucker, B., & Theeuwes, J. (2014). The effect of reward on orienting and reorienting in exogenous
cuing. Cognitive, Affective, & Behavioral Neuroscience,14, 635–646. doi:10.3758/s13415-014-
0278-7
22 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Chelazzi, L., Perlato, A., Santandrea, E., & Della Libera, C. (2013). Rewards teach visual selective
attention. Vision Research,85,58–72.
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the
brain. Nature Reviews Neuroscience,3(3), 201–215.
Coren, S., & Hoenig, P. (1972). Effect of non-target stimuli upon length of voluntary saccades.
Perceptual and Motor Skills,34, 499–508.
Della Libera, C., & Chelazzi, L. (2009). Learning to attend and to ignore is a matter of gains and
losses. Psychological Science,20, 778–784.
Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective visual attention. Annual Review
of Neuroscience,18, 193–222.
Engelmann, J. B., Damaraju, E., Padmala, S., & Pessoa, L. (2009). Combined effects of attention and
motivation on visual task performance: Transient and sustained motivational effects. Frontiers in
Human Neuroscience,3, 4. doi:10.3389/neuro.09.004.2009.
Failing, M. F., & Theeuwes, J. (2014). Exogenous visual orienting by reward. Journal of Vision,14(5), 6.
Fecteau, J. H., & Munoz, D. P. (2006). Salience, relevance, and firing: A priority map for target
selection. Trends in Cognitive Sciences,10, 382–390.
Findlay, J. M. (1982). Global visual processing for saccadic eye movements. Vision Research,22,
1033–1045.
Godijn, R., & Theeuwes, J. (2002). Programming of endogenous and exogenous saccades: Evidence
for a competitive integration model. Journal of Experimental Psychology: Human Perception and
Performance,28, 1039.
Heeman, J., Theeuwes, J., & Van der Stigchel, S. (2014). The time course of top-down control on
saccade averaging. Vision Research,100,29–37.
Hickey, C., Chelazzi, L., & Theeuwes, J. (2010). Reward changes salience in human vision via the
anterior cingulate. The Journal of Neuroscience,30, 11096–11103.
Itti, L., & Koch, C. (2001). Computational modelling of visual attention. Nature Reviews
Neuroscience,2, 194–203.
Kristjánsson, Á., & Campana, G. (2010). Where perception meets memory: A review of repetition
priming in visual search tasks. Attention, Perception, & Psychophysics,72(1), 5–18.
Le Pelley, M. E., Pearson, D., Griffiths, O., & Beesley, T. (in press). When goals conflict with values:
Counterproductive attentional and oculomotor capture by reward-related stimuli. Journal of
Experimental Psychology: General.
Meeter, M., & Van der Stigchel, S. (2013). Visual priming through a boost of the target signal: Evidence
from saccadic landing positions. Attention, Perception & Psychophysics,75, 1336–1341.
Silvis, J. D., & Van der Stigchel, S. (2013). How memory mechanisms are a key component in the
guidance of our eye movements: Evidence from the global effect. Psychonomic Bulletin &
Review,21, 357–362.
Small, D. M., Gitelman, D., Simmons, K., Bloise, S. M., Parrish, T., & Mesulam, M. M. (2005).
Monetary incentives enhance processing in brain regions mediating top-down control of attention.
Cerebral Cortex,15(12), 1855–1865.
Theeuwes, J. (2010). Top-down and bottom-up control of visual selection. Acta Psychologica,135(2),
77–99.
Theeuwes, J., & Belopolsky, A. V. (2012). Reward grabs the eye: Oculomotor capture by rewarding
stimuli. Vision Research,74,80–85.
Theeuwes, J., Kramer, A. F., Hahn, S., & Irwin, D. E. (1998). Our eyes do not always go where we
want them to go: Capture of the eyes by new objects. Psychological Science,9, 379–385.
Tipper, S. P., Howard, L. A., & Jackson, S. R. (1997). Selective reaching to grasp: Evidence for
distractor interference effects. Visual Cognition,4(1), 1–38.
Van der Stigchel, S., & Nijboer, T. C. (2011). The global effect: What determines where the eyes land.
Journal of Eye Movement Research,4(2), 1–13.
DISTRACTORS THAT SIGNAL REWARD 23
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
Van der Stigchel, S., Heeman, J., & Nijboer, T. C. W. (2012). Averaging is not everything: The
saccade global effect weakens with increasing stimulus size. Vision Research,62, 108–115.
Van Zoest, W., Donk, M., & Theeuwes, J. (2004). The role of stimulus-driven and goal-driven control
in saccadic visual selection. Journal of Experimental Psychology: Human Perception and
Performance,30, 746.
24 BUCKER, BELOPOLSKY, THEEUWES
Downloaded by [Vrije Universiteit Amsterdam] at 03:27 10 December 2014
