CSG Thesis Final

The Effects of Action Video Game Experience on Perceptual
Decision Making
by
Christopher Shawn Green
Submitted in Partial Fulfillment

of the Requirements for the Degree
Doctor of Philosophy
Advised by
Professor Daphne Bavelier
Department of Brain and Cognitive Sciences

The College
Arts and Sciences
University of Rochester
Rochester, NY
2008
ii
Curriculum Vitae
The author was born in Biloxi, Mississippi on July 19, 1980. His family moved near
Rochester, NY in 1989 and he later attended LeRoy High School in LeRoy, NY. He
attended Genesee Community College in Batavia, NY concurrent with his final two
years of high school and received his A.S. degree in Math and Science in May 1998,
just prior to receiving his high school degree in June 1998. He attended the
University of Rochester from 1999 until 2001 and graduated with a B.A. degree in
Brain and Cognitive Sciences in June 2001. After spending two years as a full-time
research assistant at the University of Rochester, he began graduate studies in the
Department of Brain and Cognitive Sciences in the fall of 2003. He received an M.A.
degree in Brain and Cognitive Sciences in 2006.
iii
Abstract
Action video game players (VGPs) have been shown to outperform their non-game
playing (NVGPs) peers on a number of sensory/cognitive measures. In tasks that
require accurate responses to quickly presented visual stimuli, VGPs typically exhibit
higher levels of accuracy than NVGPs. In particular, VGPs have demonstrated
enhancements in a number of tasks thought to tap reasonably independent aspects of
visual attention (spatial distribution and resolution, temporal characteristics, capacity,
etc). In tasks that require speeded responses, the VGP enhancement is observed as a
large decrease in reaction time (RT) in VGPs compared to NVGPs (accuracy is
typically equivalent in the two groups). Here we put forward the hypothesis that a
single mechanistic explanation, an increase in the rate of sensory integration in VGPs,
can account for the entirety of the data, thus bridging the gap between the accuracy
and RT literatures. To test this hypothesis, two sensory integration tasks were
employed - a standard motion coherence paradigm and a novel auditory localization
task which, in combination with a model developed by Palmer et al (2005), allow for
a more explicit test of the relative contribution of sensory integration rate, criteria,
and motor execution in generating the differences observed between VGPs and
NVGPs. In both the motion and auditory tasks, VGPs demonstrated a large reduction
in RT compared to NVGPs with equivalent accuracy. This pattern was well captured
by the model with an increase in the rate of information accrual and a concurrent
decrease in criteria in the VGPs. Several follow-up experiments provide further
support for the hypothesis that VGPs acquire sensory information more rapidly than
iv
NVGPs. Importantly, similar effects can be induced in NVGPs through extensive

action video game training. Finally, to examine how these changes may be
implemented at the neural level, a model by Ma and colleagues (2007) was utilized,
with the primary difference between VGPs and NVGPs being an increase in the
strength of feed-forward projections between sensory and integration areas in the
VGP group.
Table of Contents
Chapter 1
1.1
Introduction
Observed Differences Between VGPs and NVGPs
1.1.1 Increased Accuracy Observed in VGPs
1.1.1.1 The Spatial Distribution of Visual Attention
(Green and Bavelier, 2006, JEP:HPP)

1.1.1.2 The Spatial Resolution of Visual Attention
65
(Green and Bavelier, 2007, Psychological Science)

1.1.1.3 The Temporal Characteristics of Vision
90
(Green and Bavelier, 2003, Nature)

1.1.1.4 The Capacity of Visual Attention
102
(Green and Bavelier, 2006, Cognition)

1.1.1.5 Increased Accuracy in VGPs: Conclusion
1.2
162
1.1.2 Decreased Reaction Time Observed in VGPs
163
1.1.2.1 Decreased RT in VGPs: Brinley Plot
170
Hypothesis: Faster Integration of Perceptual Information
172
Motion Coherence Decision Task
174
2.1
Introduction
174
2.2
Methods
176
2.2.1 Participants
176
2.2.2 Apparatus
177
Chapter 2
vi
2.2.3 Stimulus/Procedure
177
Results
179
2.3.1 Raw Data
179
2.3.2 Discussion: Raw Data
181
Introduction: Diffusion-to-Bound Model
182
2.4.1 Results: Diffusion-to-Bound Model
185
2.4.2 Discussion: Diffusion-to-Bound Model
186
Discussion: Motion Coherence Decision Task
186
Motion Coherence Presentation Duration Task
188
3.1
Introduction
188
3.2
Methods
188
3.3
Results
189
3.3.1 Results: Raw Data
189
190
3.3.3 Introduction: Model
191
3.3.4 Results: Model
192
Discussion: Motion Coherence Presentation Duration Task
193
Auditory Localization Decision Task
194
4.1
Introduction
194
4.2
Methods
195
2.3
2.4
2.5
Chapter 3
3.4
Chapter 4
vii
4.2.1 Participants
195
4.2.2 Apparatus
195
196
Results
197
197
199
4.3.3 Results: Diffusion-to-Bound Model
199
Discussion: Auditory Localization Decision Task
200
Auditory Localization Presentation Duration Task
201
5.1
Introduction
201
5.2
Methods
201
5.3
Results
202
202
204
Discussion: Auditory Localization Presentation Duration Task
205
Video Game Training
206
6.1
Introduction
206
6.2
Methods
206
6.2.1 Participants
206
6.2.2 Testing Methods
207
4.3
4.4
Chapter 5
5.4
Chapter 6
viii
6.2.3 Training Methods
6.3
6.4
Chapter 7
References
207
6.2.3.1 Training Apparatus
207
6.2.3.2 Training Procedure
207
6.2.3.3 Game Playing Improvement
210
Results
211
6.3.1 Motion Coherence Decision Task: Raw Data
211
6.3.2 Motion Coherence Decision Task: Model Data
213
6.3.3 Auditory Localization Decision Task: Raw Data
214
6.3.4 Auditory Localization Decision Task: Model Data
217
Video Game Training: Discussion
217
Conclusions
219
225
ix
List of Figures
1: Comparison of VGP and NVGP RT Across Tasks
172
2: VGP and NVGP Performance: Motion Direction Discrimination
181
3: VGP and NVGP Performance: Motion Presentation Duration
190
4: VGP and NVGP Performance: Auditory Localization
198
5: VGP and NVGP Performance: Auditory Presentation Duration
203
6: Performance on Motion Discrimination Task Before and After Training
213
7: Performance on Auditory Localization Task Before and After Training
216
8: Correlations Between Auditory and Motion Parameters
222
9: Correlation Between Rate and Bound
223
CHAPTER 1: Introduction*
* partially adapted/excerpted from:
Green, C.S. & Bavelier, D. (2006). The Cognitive Neuroscience of Video Games in Digital Media:
Transformations in Human Communication. Messaris, P. & Humphreys, L. (eds). New York, Peter
Lang.
The Atari video game platform was released in the late 1970s. Nintendo was
born in the early 1980s. Since then, the percentage of Americans who play video
games has grown at an astronomical pace; in fact it has been estimated that 90% of
school-aged individuals currently play video games. However, despite the common
view of video games being for kids, the average age of a video game player is
presently 33 years old with 70% of the heads of households playing video games
(Entertainment Software Association, 2007). This explosion has been spurred on by
advancements in both hardware technology and software development that allow a
more intense and realistic gaming experience. In addition to the improvements in
graphical capability, advances in online gaming now allow users to play with
sometimes hundreds of others, a fact that is slowly changing the perception of video
game play from a solitary to a social activity. Unsurprisingly, both popular and
scientific interest in the potential consequences of game play has been driven by this
dramatic surge in video game use. While the majority of research (and media
attention) has focused on the potential for video game play to negatively affect
temperament and social behavior, or on the potential to harness video games to help
children learn, a subset of cognitive scientists have investigated the more fundamental
question of how video games may alter the way in which people see the world.
In most of the biological sciences, the question of nature versus nurture is
often debated. Researchers constantly strive to determine whether a certain skill
arises from nature (is genetically based), nurture (is completely determined by
experience), or as is usually the case, if the skill reflects a combination of nature and
nurture. In cognitive science the relationship is often quite complex, with the relative
roles of nature and nurture interacting through development with one playing a larger
role in some developmental stages and vice versa. For example, humans require
normal visual experience in infancy and throughout childhood to enjoy normal
vision in adulthood. When infants are deprived of normal experience (by cataracts
for instance), massive and reasonably permanent deficits may result (Birch & Stager,
1996; Lewis & Maurer, 2005). However, the same cataracts experienced later in life
may lead to no enduring deficits once removed. Thus, the effect of experience in this
case is greatest in younger humans and grows progressively weaker through
adulthood.
While there are myriad similar cases in which less than normal experience leads to
deficits in perception and cognition, researchers that investigate the effects of video
game play on perception and cognition ask a slightly different question what is the
effect of more than normal experience? To what extent are our perceptual systems
constrained by genetics? One could argue that evolution is notoriously cost-effective
and thus there is little impetus for our visual system to possess capabilities beyond
those needed in our typical environment. On the other hand, in order for a species to
be successful over an extended time span, members must be capable of adapting to
changes in their environment. One might further note that the skills needed in todays
civilized world are likely far different than those that were required to survive as the
human species evolved, and thus we may retain some additional capacity that unless
tapped, simply goes unused. A particular subset of video games, which we dub action
video games, offers an interesting natural experiment into the effect of more than
normal experience. We define action video games as those games that have many
quickly moving objects, that require effective peripheral processing as items in the
periphery must constantly be localized and identified, and where the number of
independent items that need to be kept track of far exceeds the circumstances
experienced in normal life. In short, an action video game is one that places
extraordinary demands on the visual and visuo-motor systems. Games from a variety
of genres are included as action games, but most typically include first-person
shooters (FPS), third-person shooters (TPS) and some car racing games. Many young
adults are immersed in these environments for more than 20 hours per week. The
question is therefore not whether humans can learn to adapt to these environments
(the very fact that these individuals spend such exorbitant amounts of time playing
the games suggests that they can), but instead the question is how does the system
change to handle these demands? What systems or processing stages are affected and
how do these changes manifest themselves?
1.1. Observed differences between VGPs and NVGPs

The literature on the effects of video game experience on perceptual,
perceptuo-motor, and cognitive skills can be roughly divided into two complementary
branches. The first branch is comprised of tasks wherein accuracy is the primary
dependent measure. These tasks generally require subjects to make perceptual
judgments about quickly flashed stimulus displays. Subjects are instructed to be as
accurate as possible, and reaction time is typically not measured. In these paradigms,
enhancement in the video game player (VGP) population is manifested as an increase
in accuracy as compared to non-video game players (NVGPs). The second branch of
the literature consists of tasks wherein reaction time is the primary dependent
measure. Although accuracy is also recorded in these tasks, it is typically near
ceiling, thus making reaction time the only truly informative measure. In these
paradigms enhancement in the VGP population is manifested as a decrease in reaction
time as compared to NVGPs.
1.1.1. Increased accuracy observed in VGPs

VGPs have demonstrated superior accuracy in a number of tasks designed to
tap reasonably distinct aspects of visual attention including the spatial distribution of
attention, the spatial resolution of attention, the temporal characteristics of visual
attention, and the capacity of visual attention. The following four published papers
explain the tasks and the aspects of visual attention they are thought to evaluate in
detail.
1.1.1.1. The spatial distribution of visual attention

Green, C.S. & Bavelier, D. 2006. Effect of action video games on the spatial
distribution of visuospatial attention. JEP:HPP 32(6), 1465-1478.
- Version below is the final draft version and thus may differ from the final published
version in terms of formatting, precise wording, etc.
Effect of action video games on the spatial distribution

of visuo-spatial attention
C.S. Green and D. Bavelier
Department of Brain and Cognitive Sciences

RC 270268, Meliora Hall
Rochester, NY 14627-0268
(585) 275-0695
[email protected]
Abstract
This paper investigates the effect of action gaming on the spatial distribution
of attention. The flanker compatibility effect was used to separately assess center and
peripheral attentional resources in gamers versus non-gamers. Gamers exhibited an
enhancement in attentional resources compared to non-gamers not only in the
periphery, but also in central vision. A target localization task was then used to
unambiguously establish that gaming enhances the spatial distribution of visual
attention over a wide field of view. Gamers were more accurate than non-gamers at
all eccentricities tested and the advantage held even when a concurrent center task
was added, ruling out a trade-off between central and peripheral attention. By
establishing the causal role of gaming through training studies, this work
demonstrates that action gaming enhances visuo-spatial attention throughout the
visual field.
1. Introduction
Visual acuity, or the ability to discriminate small changes in shape in central

vision, is a key determinant of vision. Ask someone how good their vision is, and
they will typically comment on their ability to read a sign, to recognize faces from
afar, or to score 20/20 on an optometrists eye chart. However, many of the visual
tasks we complete on a day-to-day basis often bear little relation to our ability to read
the bottom line on an eye chart. For instance, driving does not require perfect acuity
(many states in the USA require that vision be only 20/40 to receive a drivers
license). Instead, the most common visual demands present while driving involve
focusing attention on relevant stimuli such as pedestrians, animals, and other cars,
while ignoring the many irrelevant distractors that clutter the visual environment.
The dichotomy between visual acuity and visual attention has been exemplified by
many studies (Ball, Beard, Roenker, Miller, & Griggs, 1988; Ball & Owsley, 1991;
Ball, Owsley, & Beard, 1990; Ball, Owsley, Sloane, Roenker, & Bruni, 1993;
Intriligator & Cavanagh, 2001; Owsley & Ball, 1993; Owsley, Ball, & Keeton, 1995;
Sekuler & Ball, 1986), with the general finding being that simple tests of visual acuity
and perimetry are poor predictors of performance on tasks that demand effective
visuo-spatial attention.
A number of paradigms have been developed with the goal of quantitatively
measuring visual selective attention (Carrasco & Yeshurun, 1998; Eckstein, Pham, &
Shimozaki, 2004; Eriksen & Eriksen, 1974; Lavie & Cox, 1997; Treisman & Gelade,
1980).
In many of these paradigms, targets are presented simultaneously with
distracting objects and the influence of the distracting information on target

processing is measured. Groups thought to have diminished attentional abilities such
as the elderly (Ball et al., 1988; Madden & Langley, 2003; Maylor & Lavie, 1998;
Plude & Hoyer, 1986; Scialfa, Esau, & Joffe, 1998) or young children (Akhtar &
Enns, 1989; Enns & Cameron, 1987; Enns & Girgus, 1985; Plude, Enns, & Brodeur,
1994; Rueda et al., 2004) typically demonstrate larger effects of distracting
information on attentional tasks than normal adult controls, indicating an effect of age
on the determinants of visual selective attention. Similarly, a host of data indicates
that the control of visual selective attention decreases in most pathological
populations, including frontal patients (Husain & Kennard, 1997), Alzheimers
patients (Levinoff, Li, Murtha, & Chertkow, 2004; Tales, Haworth, Nelson, Snowden,
& Wilcock, 2005; Tales, Muir, Jones, Bayer, & Snowden, 2004), children with
ADHD (Shalev & Tsal, 2003) and neglect patients (Russell, Malhotra, & Husain,
2004; Sprenger, Kompf, & Heide, 2002; Vivas, G.W., & Fuentes, 2003). While most
of the studies describing changes in visual selective attention document a decrease in
performance as compared to normal healthy young adults, of interest to us was the
possibility that this type of selective attention may be enhanced (rather than reduced)
from the level typically observed in young adults.
Several researchers have noted enhancements in various aspects of visual attention as
a result of video game play (Castel, Pratt, & Drummond, 2005; Gopher, 1992;
Gopher, Weil, & Bareket, 1994; Green & Bavelier, 2003; Greenfield, DeWinstanley,
Kilpatrick, & Kaye, 1994; Trick, Jaspers-Fayer, & Sethi, 2005). Many of todays
action video games are remarkably visually challenging. They regularly have rather
unnaturally stringent attentional requirements, much more so than any everyday
situation to which one may be exposed. For instance, in many video games, multiple
items must often be processed simultaneously, a task that would be benefited by
additional attentional resources across space. Additionally, many games also require
the efficient rejection of irrelevant objects, a process that would be benefited by a
more proficient selection process. The penalty for either failing to process a target or
allowing non-essential information to interfere with the processing of potential targets
is often great in video games; therefore, those who play should be especially
motivated to develop both capabilities. It is the goal of this paper to assess whether
action video game experience enhances visuo-spatial attention and its allocation over
space. Our previous work led to the hypothesis that action video game play results in
an increase in the amount of available attentional resources as well as an increase in
the selectivity of spatial processing (Green & Bavelier, 2003).
It has remained
unclear however whether the improvements noted were specific to the visual
periphery, possibly occurring at the cost of central vision. The present study reports
on two types of paradigms that test the distribution of attention over space and that
contrast central and peripheral processing. The first, the perceptual load paradigm of
Lavie and colleagues (Lavie, 1995; Lavie & Cox, 1997), offers a measure of the
attentional resources available to both video game players (VGPs) and non-gamers
(NVGPs), and was adapted to compare central and peripheral resources across
10
populations. The second, the Useful Field of View (UFOV) paradigm developed by
Ball and colleagues (Ball et al., 1988; Ball & Owsley, 1992), allows for a measure of
the distribution and selectivity of visual attention across a wide field of view, while
controlling for different levels of central load. Portions of Experiments 1 and 2 were
presented in Green and Bavelier (2003).
2. Experiment 1 the perceptual load paradigm
In order to gain a general understanding of the amount of attentional resources

available to distribute across space in gamers and non-gamers, the perceptual load
paradigm was employed (Lavie & Cox, 1997; Proksch & Bavelier, 2002). In this
paradigm, the effect of task irrelevant distractors on primary task performance is
measured and used as an index of the degree to which these irrelevant distractors are
processed. By contrasting central and peripheral distractors, this paradigm allows us
to compare attentional resources across space in VGPs and NVGPs.
The effect of the distractors is measured using the compatibility effect,
wherein the presence of distractors that lead to the same response as the target
(response compatible) results in faster reaction times than distractors that lead to a
different response as the target (response incompatible) (Eriksen & Eriksen, 1974).
Work by Lavie and colleagues (Lavie, 1995; Lavie & Cox, 1997; Lavie, Hirst,
Viding, & de Fockert, 2004) has shown that the effect of extraneous distractors on
reaction time is largely a function of the perceptual load of the target task display
11
(perceptual load in this case roughly corresponds to the number of items in the visual
search array). When the perceptual load is low (for instance, the visual search array
contains only the target), the effect of an extraneous distractor on performance is
great. However, when the perceptual load is high (the visual search array contains the
target as well as several additional items), the effect of an extraneous distractor on
performance is minimal.
This finding is incorporated in the load theory of selective attention and
cognitive control (Lavie, 2005; Lavie et al., 2004). Relatively easy perceptual tasks
do not require all of ones attentional resources to reach adequate behavioral
performance. In this case, the resources left over from the task are not simply turned
off, but are instead distributed to adjacent locations/items leading to a sizable
compatibility effect. Conversely, challenging perceptual tasks demand a larger
percentage of the available attentional resources, thereby leaving little to spread to
non-task locations/items and resulting in little or no compatibility effect (Lavie, 1995;
Lavie & Cox, 1997). Although the load theory posits that the distribution of attention
is automatic, it is not the case that the exact distribution of attention is identical in all
humans. Proksch and Bavelier (2002) have demonstrated that hearing individuals do
typically allocate more attention to the area around fixation (central vision), but in
contrast deaf individuals appear to allocate more attention to the periphery. If action
video game play primarily affects peripheral visual attention, then VGPs may also
exhibit a proportionally larger compatibility effect for peripheral distractors as noted
in deaf individuals. Conversely, if action video game play similarly affects both
12
central and peripheral vision, then the distribution should be similar to what is
observed in normal hearing subjects, that is greater allocation around fixation than
peripherally (Beck and Lavie, 2005; Proksch & Bavelier, 2002). In Experiment 1, the
compatibility effects induced by peripheral versus central distractors were compared
in VGPs and in NVGPs to gain a measure of the amount of available attentional
resources as a function of eccentricity in each population.
2.1. Method
2.1.1. Participants
Sixteen males with normal or corrected vision were placed into one of two
groups, video or non-video game player, based upon their responses to a
questionnaire given prior to the experiment. Because of the relative difficulty in the
acquisition of females with sufficient video game experience, only males underwent
testing.
The criterion to be considered a video game player (VGP) was a minimum of
3-4 days a week of action video game usage for the previous six months. Eight righthanded males with a mean age of 20.9 years fell into this category. Seven of the eight
reported daily video game usage for at least the previous six months, while the eighth
reported playing several times a week for the same time span. It is important to note
that only action video game players were included in this and all subsequent
experiments. Action video games are those that have fast motion, require vigilant
13
monitoring of the visual periphery, and often require the simultaneous tracking of
multiple targets. The following is a representative sample of the games reported as
played which qualify as action games based on our criteria: Grand Theft Auto, HalfLife, Counter-Strike, Marvel vs. Capcom, Rogue Speare, and Super Mario Kart.
The criterion to be considered a non-video game player (NVGP) was little,
although preferably no, action video game usage in the past six months. Eight males
(six right- and two left-handed) with a mean age of 21.6 years fell into this category.
Seven members reported no video game experience whatsoever in the past year,
while the eighth reported a maximum of five instances of non-action video game play
over the same time frame.
All participants were paid for their participation and provided informed
consent in accordance with the guidelines set by the University of Rochester.
2.1.2. Stimuli
All stimuli/procedures were identical to Proksch & Bavelier (2002). Stimuli

were presented on a 21 Mitsubishi monitor from a standard PC equipped with a
Matrox Millennium video card using OpenGL routines. Each participant viewed the
display at a test distance of 60 cm in a darkened room.
The stimuli consisted of three categories of items (target, filler, and distractor)
presented in light gray on a black background (Figure 1). The target set consisted of
a square and a diamond. The filler set was composed of a house-like shape, an
14
upside-down house shape, a sideways trapezoid, a triangle pointing up, and a triangle
pointing down. Both the target and the filler shapes subtended an average of .6
vertically and .4 horizontally, and were always presented inside circular frames.
Throughout the experiment, the six circular frames were presented in the same
location, arranged around the central fixation point at a distance of 2.1. The distance
between the centers of adjacent circular frames was also 2.1. One, and only one,
member of the target set (square or diamond) always appeared in one of the six
circular frames. Random members of the filler set could occupy zero, one, three, or
five of the remaining circular frames. The frames in which target and fillers appeared
were randomly selected across trials. For all analyses and for the purposes of
subsequent discussion, the two lowest levels of perceptual load (zero fillers or one
filler) were grouped into a single low load condition, while the two highest levels
of perceptual load (three or five fillers) were grouped into a single high load
condition.
The distractor set consisted of a square, a diamond, and an elongated circle.
One member of the distractor set was presented during each trial in one of four
locations. The distractor could appear either centrally (.5 to the right or left of
fixation which falls within the ring of circular frames) or peripherally (4.2 to the
right or left of fixation outside the ring of circular frames).
Although 4.2 from fixation is better described as parafoveal rather than
peripheral, this location was chosen to ensure that peripheral and central distractors
were presented at comparable distances from the target ring. This point is important
15
if differences in effects between central and peripheral distractors are to be interpreted

in terms of eccentricity (central versus peripheral) rather than absolute distance
(Miller, 1991). The size of the distractors was corrected for the known cortical
magnification factor (Rovamo & Virsu, 1979). Accordingly, central distractors
subtended .3 vertically and .2 horizontally, while peripheral distractors subtended
.9 vertically and .5 horizontally.
The design was fully intermixed with all combinations of target (square or
diamond), perceptual load (0,1,3,or 5 fillers), distractor compatibility (compatible,
incompatible, or neutral), and distractor eccentricity (central or peripheral) being
equally represented and presented in pseudorandom order.
16
Figure 1: Perceptual Load Stimuli

The participants task was to determine as quickly and accurately as possible which of two possible
target shapes (square or diamond) appeared in one of the six circular frames. Task difficulty was
manipulated by the addition of other shapes in the circular frames. Low loads correspond to displays
with the target alone or with one other shape in the circular frame, while high loads correspond to
displays with three or five shapes in addition to the target. The distractor shape is the shape that does
not appear in one of the six circular frames. Subjects were explicitly told to ignore the distractor
shape which could be either compatible (led to the same response as the target) or incompatible (led
to the opposite response as the target). The distractors shapes could be presented either centrally, that
is appearing within the ring of circular frames, or peripherally, appearing outside of the ring.
2.1.3. Procedure
Each trial began with a one second fixation point followed by a 100 ms
presentation of the trial shapes. The relative brevity of the presentation time was
chosen to preclude eye movements as a potential source of differences.
17
The task of the participant was to identify which of the two potential target
shapes (square or diamond) appeared in one of the six circular frames as quickly and
accurately as possible. Subjects were reminded to ignore any stimuli that did not
appear in the circular frames (i.e. the distractor). Participants responded to the target
by pressing the key labeled with the corresponding target shape. Feedback was given
after each trial by a change in the color of the fixation point. Trials were grouped into
two halves of 576 trials. Following each block of 48 trials, participants were given a
resting screen that informed them of their performance over the previous block
(reaction time and percent correct). Subjects were instructed to respond as quickly as
possible and to aim for 90% correct.
Before testing began, participants were given two blocks of practice during
which time responses were monitored by the investigator to ensure comprehension of
the task. Following successful training, the participants were left to complete the first
half, followed by a short intermission, and the second half. The entire experiment
lasted approximately one and a half hours.
2.2. Results
As in our past studies, all analyses focused on trials with incompatible or
compatible distractors (Green & Bavelier, 2003; Proksch & Bavelier, 2002).
2.2.1 Reaction Time
18
For the reaction time analysis, incorrect trials were first removed (VGP:
12.1% +/-1.5, NVGP:13.0%+/-1.6) and as well as any trial with reaction times greater
than 1800 ms or less than 300 ms (less than 2% of trials in both VGPs and NVGPs).
Trials were then separated based upon distractor eccentricity (central, peripheral).
For each subject a mean and standard deviation was computed for each of the two
eccentricities; any trial in which the reaction time was more than two standard
deviations away from the mean was excluded (approximately 2% of trials for both
VGPs and NVGPs). This filtered RT data was then analyzed in a 2x2x2x2 omnibus
ANOVA with video game experience (VGP/NVGP), perceptual load (low/high),
distractor eccentricity (central/peripheral), and distractor compatibility
(compatible/incompatible) as factors.
The standard main effects of perceptual load (low load: 629.2 ms +/-9.5, high
load: 712.4 ms +/-11.2; F(1,14) = 119.4, p < .001), reflecting an increase in task
difficulty with increasing perceptual load, and distractor compatibility (compatible
distractors: 664.1 ms +/- 11.5, incompatible distractors: 677.5 ms +/-11.6; F(1,14) =
63.0, p < .001), demonstrating the effect of distractor compatibility on RT, were
observed. Also, as has been consistently reported, an interaction between load and
compatibility was observed (F(1,14) = 7.0, p = .02) with the RT difference between
incompatible and compatible distractors decreasing with increasing perceptual load
(Table 1).
Table 1
19
Center Distractor
Low Load
Peripheral Distractor
High Load
Low Load
High Load
Incompat
Compat
Incompat
Compat
Incompat
Compat
Incompat
Compat
VGP
606 (20)
595 (22)
708 (31)
684 (28)
617 (26)
595 (22)
708 (28)
701 (26)
NVGP
665 (21)
642 (18)
727 (27)
719 (26)
665 (17)
644 (17)
720 (25)
729 (25)
Table 1: Reaction times and (SE) for each of the conditions of Experiment 1
NVGPs show a clear decrease in the size of the compatibility effect with increasing load for
both central and peripheral distractors, while VGPs show a decrease in the size of the compatibility
effect with increasing perceptual load only for the peripheral distractors. The opposite effect, an
increase in the size of the compatibility effect with increasing perceptual load, is observed in the VGP
population for the center distractor condition.
In accord with our previous report (Green & Bavelier, 2003), a video game
experience x perceptual load x distractor compatibility interaction (F(1,14) = 7.4, p =
.02) was also observed, with NVGPs showing a larger decrement in the size of the
compatibility effect with increasing load than the VGPs (Table 1). VGPs instead
demonstrate a consistently high compatibility effect across perceptual load
conditions. This indicates that the VGPs continued to process the extraneous
distractor even at the highest loads, suggesting an increase in available attentional
resources. Importantly, this effect did not further interact with eccentricity (F(1,14) =
1.15, p > .3) signifying the effect of video game experience was similar for both
central and peripheral distractors (see also Figure 2). Finally, consistent with
previous reports demonstrating greater attentional resources in central vision, a
20
perceptual load x distractor compatibility x distractor eccentricity interaction (F(1,14)

= 8.1, p = .01) was observed with the decrease in the size of the compatibility effect
with increasing load being more pronounced for peripheral than central distractors.
Because differences in baseline RTs between groups may produce interactions
with within-subjects measures that reflect only the magnitude of the baseline RT
differences, it is important to note that although the mean VGP reaction time was
approximately 37 ms faster than the mean NVGP reaction time, the main effect of
gaming did not approach significance (VGP: 652.4 ms +/- 13.5, NVGP: 689.2 ms +/8.8; F(1,14) = .8, p > .3) indicating relatively comparable overall reaction times in
VGPs and NVGPs.
To better assess the spatial distribution of attentional resources in VGPs and
NVGPs, the two groups were separated and the size of the compatibility effect
(incompatible compatible RTs) for each group was examined in 2x2 ANOVAs with
perceptual load (low/high) and eccentricity (central/peripheral) as factors. In the
NVGP group, only a main effect of load was found with compatibility effects being
larger for the low load than the high load condition (low: 21.6 ms +/-4.3, high: -0.5
ms +/-4.4; F(1,7) = 28.4, p = .001). As previously reported (Proksch & Bavelier,
2003), the compatibility effect was approximately double for central distractors as
compared to peripheral distractors (central: 14.4 ms +/-4.4, peripheral: 7.0 ms +/-5.7);
however, in this study the main effect was not statistically significant (F(1,7) = .63, p
> .4).
21
In the VGP group no effect of load was found (low: 15.2 ms +/-4.3, high:
15.5 ms +/-4.7; F(1,7) = .001, p > .9 ). However, an eccentricity by load interaction
was observed (F(1,7) = 8.1, p = .03) with compatibility effects being greater for low
load in the periphery and for high load in the center. This indicates that in the VGP
population, the allocation of attention shifts from a more peripherally biased
distribution under conditions of low load to a more centrally biased distribution under
conditions of high load.
Figure 2: Experiment 1 Results

Size of compatibility effect (RT incompatible RT compatible) as a function of eccentricity: As in
previous work, there is a trend for greater compatibility effects to be present for center distractors.
Importantly, VGPs show compatibility effects for both center and peripheral distractors, suggesting
that the changes in the VGP population are not specific to the visual periphery.
2.2.2 Accuracy
22
Error rates were analyzed in a 2x2x2x2 omnibus ANOVA with video game
experience (VGP/NVGP), perceptual load (low/high), distractor eccentricity
(central/peripheral), and distractor compatibility (compatible/incompatible) as factors.
This analysis revealed only a main effect of perceptual load (low load: 94.3% +/- 1.0,
high low: 80.6% +/- 1.3; F(1,4) = 221.5, p < .001). No other main effects or
interactions were significant including the main effect of video game experience
(F(1,14) = .09, p > .7). The increase in error rate with higher levels of perceptual load
highlights the increase in task difficulty with increasing perceptual load. However,
the lack of a main effect of or interactions with video game experience suggests that
the task was equally difficult for the VGP and the NVGP populations.
2.3. Discussion
Experiment 1 establishes that VGPs continue to be affected by distracting
items at much higher perceptual loads than NVGPs. As previously discussed, the
perceptual load at which compatibility effects disappear provides an estimate of the
amount of available attentional resources. The higher the perceptual load of the task
when this occurs, the greater the attentional resources available. VGPs demonstrate a
clear compatibility effect even under conditions of high load, while NVGPs cease to
show an effect of the distractors at these loads. This indicates an increase in the
attentional resources available in the VGP population.
23
Other potential alternative explanations for this result are not wholly
consistent with the data. One may surmise for instance, that the discrimination task is
less perceptually demanding for the VGPs than the NVGPs. It would then follow that
the perceptual difficulty of, for instance, a load of four for a NVGP is equivalent to a
load of eight for a VGP. However, although the VGPs do show a slight advantage in
both percent correct (VGPs approximately 1% more accurate) and in simple reaction
time (VGPs 37 ms faster) neither is significant, nor are there video game experience
by perceptual load interactions for either dependent variable. Thus, behaviorally one
must assume the tasks are similarly difficult for the two groups. Another possible
explanation is that non-target objects simply more easily distract VGPs than NVGPs.
While it is unintuitive that the advantage in attentional resources manifests itself
through a greater effect of distractors (which would suggest poorer control of visual
selective attention), it should be noted that at a load of one, which elicits the
maximum compatibility effect from NVGPs, the VGP compatibility effect is if
anything smaller than that of the NVGPs (p = .1). This issue will be addressed more
thoroughly in Experiments 2 and 3, but in all, the hypothesis most consistent with the
data is that VGPs have an enhancement in attentional resources compared to NVGPs.
Importantly for the question at hand, the spatial distribution of attention found
in VGPs was similar to that seen in NVGPs. In accord with the previous literature
(Beck & Lavie, 2005; Proksch & Bavelier, 2002), a bias was seen for central vision,
with the size of the compatibility effect decreased more sharply with increasing load
for peripheral than for central distractors reflecting greater attentional resources in
24
central than peripheral locations (Figure 2). It is also significant that the interaction
between video game experience, perceptual load, and distractor compatibility did not
interact further with distractor eccentricity. This suggests that even as load increased,
the VGPs continued to process both central and peripheral distractors to a greater
degree than the NVGPs. Thus, enhanced attentional capacities in VGPs are not
exclusive to the visual periphery, but instead are present in both central and peripheral
vision.
To more conclusively answer the question of whether VGPs can make the
most of this attentional enhancement, a different type of paradigm was employed.
After all, in the perceptual load task distractors are processed, despite the fact that
doing so could be detrimental to the successful completion of the primary task. The
question then arises, are VGPs actually better at filtering out irrelevant items, which
is really the hallmark of visual selective attention? To answer this question we turned
to the Useful Field of View paradigm, which allowed us to measure the effect of
distracting information and changes in central task load on peripheral target
localization.
3. Experiment 2 the Useful Field of View paradigm
The Useful Field of View (UFOV) task (Ball et al., 1988; Ball & Owsley, 1992;
Goode et al., 1998; Mazer, Sofer, Korner-Bitensky, & Gelinas, 2001; Myers, Ball,
Kalina, Roth, & Goode, 2000; Sekuler, Bennett, & Mamelak, 2000) measures the
25
ability to locate a target as a function of the eccentricity of the target, the amount of
distracting elements in the display, and the presence of an added center task.
Performance on the UFOV is poorly correlated with so-called perceptual visual
attributes (contrast sensitivity, acuity, perimetry, etc) and is instead thought to provide
an index of the distribution of visual attention across the visual scene (Ball et al.,
1990; Owsley et al., 1995). Previous results indicate that the ability to localize a
peripheral target decreases with eccentricity, with distraction and as a center task is
made more difficult (Ball et al., 1988).
Three different target eccentricities (10, 20, and 30) were used allowing the
distribution of visual attention to be mapped as a function of eccentricity. Because the
peripheral stimulus in Experiment 1 was within the range of normal video game
playing, we were unable to assess the generality of the learning across space. In the
UFOV paradigm, we can test the effect of action video game experience within, at the
border of, and beyond, the eccentricity games are typically played at (our players
generally reported a viewing angle of 7.5-10 from the center of the screen). If the
effect of action video game play is specific for trained parts of the visual field, there
should be little to no effect of experience at 30, whereas if action video game play
alters processing throughout the visual field, differences should be observed at all
three eccentricities.
To better understand the effect of video game playing on the allocation of
attention over the visual field, the paradigm we used included one condition without a
center task and one with a center task. By contrasting performance with or without a
26
concurrent center task, the UFOV allowed us to test whether enhanced peripheral
localization performance in VGPs may be occurring at the cost of central
performance. If VGPs indeed have greater attentional resources both centrally and
peripherally as suggested in Experiment 1, the detrimental effect of the center task on
peripheral localization should be lesser in the VGPs than NVGPs (while maintaining
equal accuracy on the central task). Alternatively, if enhanced peripheral performance
in the VGPs is at the cost of central attention, we may observe a larger tradeoff
between central and peripheral tasks in VGPs than NVGPs.
Finally, the paradigm we used included displays with and without distractors.
Participants were first asked to perform the task without distractors and then with
distractors. Performance in the distractor condition is thought to reflect the same
processes as in typical visual search; however, because block order was fixed, this
design does not allow us to address the issue of whether there is a discriminative
effect of distractor load. Thus, our paradigm is not suited to address the role of
gaming on the rate of visual search.
3.1. Method
3.1.1. Participants
16 right-handed males with normal or corrected vision, none of whom had
participated in Experiment 1, were again classified as either VGPs or NVGPs
according to the same requirements as those used in Experiment 1. Eight males were
27
classified as VGPs (mean age 19.5, all right handed). The remaining 8 participants
fell into the NVGP category (mean age 20.1, 7 right-handed).
3.1.2. Apparatus
The apparatus consisted of a MacIntosh G3 computer running a program to
present stimuli and collect the data using the Matlab computer language (The Math
Works Inc., Natick, MA) and the Psychophysical Toolbox routines (Brainard, 1997;
Pelli, 1997) (http://psychtoolbox.org). The stimuli were displayed on a 24 Sony
GDM-FW900 driven at 160Hz, 800x600 resolution by a MP 850 video card (Village
Tronic Computer, Sarstedt, Germany).
3.1.3. Stimuli/Procedure:
Each observer viewed the display binocularly with his head positioned in a
chin rest at a test distance of 22 cm. Each trial consisted of four successive displays
presented on a large monitor. The displays were similar to those used by Ball et al.
(Ball et al., 1988), but stimulus size and presentation time were both decreased to
account for the increased ability of comparatively younger subjects.
The initial display consisted of a square outline (4 x 4) that directed fixation
to the center of the screen. One second later the target stimulus, a filled triangle
within a circle outline (subtending 3 x 3), appeared along with the central fixation
box. The target stimulus could appear randomly at one of twenty-four locations on the
28
screen. Each location was positioned on one of eight radial spokes and at one of three
possible eccentricities - 10, 20, or 30. Rapid presentation of the stimulus ensured
that no purposeful change in fixation could be completed during the presentation.
Localization difficulty was roughly equated at all eccentricities by manipulating the
exact stimulus presentation duration to allow a fair comparison of the effects of
gaming across eccentricities. Based on the results from a few pilot VGPs (none of
whom took part in the subsequent experiments), the duration of the stimulus
presentation was chosen to lead to about 80% correct performance in VGPs at all
three eccentricities tested. To achieve this goal, a shorter display presentation was
used at 10 (6.7 ms) than for 20 and 30 (13.4 ms). By preventing ceiling effects in
the VGP group, this manipulation enabled us to assess the true size of the group
effects at each eccentricity. After the test stimulus, a mask screen appeared for 750
ms. The mask screen, designed to eliminate afterimages as a possible source of
information, consisted of randomly spaced vertical and horizontal lines of variable
thickness and luminance, circles and squares of random sizes, and thick lines
(luminance equal to that of the stimulus) that completely covered each possible
stimulus location. The location, size, and contrast of the mask items was randomized
every trial to avoid the creation of potentially confounding consistent local elements.
Finally, a response screen consisting of a radial pattern (eight evenly spaced spokes four cardinal directions as well as four diagonals) appeared to direct the response.
Each spoke was labeled in a one to one stimulus-response mapping with the keyboard
29
number pad (i.e. the number 8 spoke was straight up from center, the number 4 spoke
was straight left) to best facilitate subject response.
Subjects were allowed to respond at any time after the presentation of the
stimulus by pressing the number on the keyboard number pad corresponding to the
radial spoke they believed the stimulus had appeared on. Pilot data from Ball et al.
(Ball et al., 1988) indicated that when subjects could accurately determine the radial
location of the stimulus, they also knew the targets eccentricity more than 90% of the
time. Therefore, subjects were not required to indicate the eccentricity of the target.
While most subjects responded during the mask presentation time, if the subject had
not yet responded, the spoke pattern remained visible until the subject made a
selection. Subjects were made aware that accuracy rather than speed of response was
critical and no penalty was assessed for slow responses. After subject response,
feedback was given and the subject pressed the middle key on the number pad (the
number 5, which was not associated with a spoke) to initiate a new trial.
Two main levels of distraction were tested. Under the no distractor condition
(0-distractor block), the stimulus appeared alone on the screen. In the distractors
present condition, two sub-levels of distraction were tested. In one (23-distractor
block), distractors were present in the twenty-three potential target positions not
occupied by the target (on the eight spokes and at all three possible eccentricities).
The distractors consisted of open squares of the same luminance as the stimulus and
subtending 4 x 4. In the other (47-distractor block), the distractors occupied all of
the same locations as in the half-distraction condition as well as the areas between,
30
thus filling a 60 diameter circle with distractors. Each subject underwent 120 trials
(eight spokes x three eccentricities x five repetitions of each) for each of the three
distraction blocks (0, 23 and 47). The blocks were always tested in a fixed order with
0-distractors, followed by 23-distractors, and then 47-distractors. It should be noted
again that for the purposes of statistics and discussion, because performance
differences have not been observed between the 23- and 47-distractors block either by
our own lab or others (Ball et al., 1988), the data from the 23- and 47-distractor block
were collapsed into the distractors present group. This resulted in twice as many
trials in the distractors present group than the no distractors group. Coupled with the
fact that distractor order was not counterbalanced, separate analyses were performed
for no distractors and distractors present conditions.
In a different set of blocks, one for each block of distractors (0, 23 and 47),
subjects performed the same peripheral localization task, but also performed a center
shape discrimination task as well. The central stimulus was either an isosceles
triangle or a diamond. In these blocks, subjects were asked to determine which of the
two shapes (triangle or diamond) was presented centrally (within the center fixation
box) by pressing the correspondingly labeled key on the keyboard. Subjects then
indicated the spoke upon which the peripheral target fell on the keypad (in the same
manner as previously described).
The experiment therefore consisted of 6 blocks, 0, 23-, 47- distractor blocks
each with and without a simultaneous central task. The level of center task was
counter-balanced as to which was given first, but again, the distractor conditions were
31
always run in the order: 0 distractor first, followed by 23-distractors, and then 47distractors.
To summarize, four main factors were manipulated: the amount of video game
experience of each subject (2 levels VGP/NVGP), the eccentricity of the target (3
levels - 10, 20, 30), the amount of distraction (2 levels no distractors, distractors
present), and the center task (2 levels no center task, center task present).
3.2. Results
Because the design of the experiment did not counterbalance between

distractor blocks (and thus distractor condition is confounded with task experience),
peripheral localization accuracy was analyzed in two separate 2x3x2 ANOVAs, one
for the no distractors condition and one for the distractors present condition with
video game experience (VGP/NVGP), eccentricity (10, 20, 30) and center task (no
center task/center task present) as factors. The peripheral localization data for the
center task present conditions was filtered prior to analysis by removing any trials in
which the center shape was incorrectly identified.
It should be noted that because several of the cell means for the NVGPs
approached floor (and thus may have deviated from normality), we also performed
the same ANOVAs on arcsin transformed data. In no cases did a significant p-value
in the untransformed analyses become non-significant using the arcsin transform or
32
vice versa, and thus for ease of interpretation, only the analyses on untransformed
accuracy are presented.
3.2.1. Peripheral Localization Accuracy - No Distractors Condition

First, although we tried to match performance across eccentricities, a main
effect of eccentricity (F(2,28) = 4.7, p < .05) was still observed. Unlike previous
UFOV studies however, where the main effect of eccentricity represented decreasing
accuracy with increasing eccentricity, the main effect of eccentricity here represents a
failure to equalize the difficulty of each eccentricity by altering the presentation
times. By using different presentation times (7 ms for 10 and 13 ms for 20 and 30)
we had hoped to achieve relatively stable performance across eccentricities.
However, while 10 and 30 did have similar performance with these timings,
performance at 20 was slightly better than both. Second, a main effect of center task
(F(1,14) = 5.2, p < .05) was observed with subjects making more peripheral
localization errors when the center task was present. Finally, as predicted by our
hypothesis, a main effect of video game experience was observed (VGP: 84.3% +/2.5, NVGP: 31.8% +/-3.6; F(1,14) = 44.4, p < .001) (Figure 3A) as the VGP group
outperformed the NVGPs by a large margin. No other effects reached significance.
3.2.2. Peripheral Localization Accuracy - Distractors Present Condition

As in the no distractors condition, a main effect of eccentricity (F(2,28) = 6.5,
p < .01) was observed. The main effect of center task did not reach significance
33
(F(1,14) = 3.8, p = .07), but was in the same general direction as in the previous
analysis. Again, as predicted, a large main effect of video game experience was
observed (VGP: 73.6%+/-3.0, NVGP: 30.0%+/-3.1; F(1,14) = 37.5, p < .001, Figure
3B) indicating superior localization performance by the VGPs. Finally, a video game
experience x eccentricity x center task interaction (F(2,28) = 4.5, p = .02) was
observed and appears to be rooted in the fact that the VGPs performed
disproportionately well in the center task condition at 10 of eccentricity (fastest
presentation time).
Figure 3: Experiment 2 Accuracy of target localization as a function of

eccentricity for gamers and non-gamers
VGPs localize a peripheral target far more accurately than NVGPs at each eccentricity (X-axis), both
without (A) and with (B) distractors present.
3.2.3. Center Identification Task Performance
34
The previous analyses only included trials in which the center shape
identification was correct. However, to conclusively demonstrate that any differences
observed in peripheral localization accuracy were not related to allocation of attention
to the periphery at the expense of the center task, center shape identification was
analyzed in a 2 (video game experience: VGP/NVGP) x 3 (eccentricity: 10, 20, 30)
ANOVA collapsed across all distractor conditions.
VGPs exhibited greater accuracy than NVGPs at the center discrimination
task itself (VGP: 97.2% +/- .8, NVGP: 90.1% +/- 1.1; F(1,14) = 25.4, p < .001). A
main effect of eccentricity (F(2,28) = 15.0, p < .001) highlights the differences in
presentation time. When the peripheral stimulus was presented at 10 of eccentricity,
the presentation time was one screen refresh fewer than when the peripheral stimulus
was at 20 or 30. Thus, the presentation time of the center stimulus was also
decreased by this amount at 10. VGPs were able to achieve the same level of center
identification performance for each eccentricity/presentation time (10: 98.6%+/- .41;
20: 97.9% +/- .6; 30: 95.2% +/- 1.5) whereas NVGPs suffered a cost at the quicker
presentation time (10: 81.1% +/- 2.5; 20: 95.2% +/- 1.0; 30: 96.0% +/-. 8) resulting
in a video game experience x eccentricity interaction (F(2,28) = 6.3, p = .006).
3.2.4. Overall effect of center task on peripheral localization

Because the results of Experiment 1 made the specific prediction that NVGPs
would be more strongly affected by the addition of a concurrent center task than
VGPs, the two groups were separated and the effect of center task on peripheral
35
localization accuracy was analyzed collapsed across eccentricities and distractor

levels. As predicted, only the NVGP group showed a significant decrease in
performance when the center task was added (No Center Task: 33.8%+/-7.1, Center
Task Present: 25.3%+/-5.8, F(1,7) = 7.0, p = .03) while the VGPs showed no such
decrement (No Center Task: 77.9+/-5.%, Center Task Present: 76.3%+/-3.6; F(1,7) =
.2, p = .65). This pattern of results supports the conclusion that VGPs have more
attentional resources available than NVGPs.
3.3. Discussion
VGPs display enhanced target localization abilities under all conditions tested.
VGP performance is superior to NVGPs at all eccentricities, with and without the
addition of distractors and with or without a concurrent center task. Together, these
findings support the results of Experiment 1 and demonstrate an enhancement in
spatial attention in VGPs not only at peripheral but also at central locations.
VGPs more accurately localize the target at all three eccentricities (10, 20,
and 30), demonstrating that video game experience enhances visual processing
across a large portion of the visual field. In particular, the superior performance of
VGPs at 30 suggests that the effects of video game play generalize to untrained
locations, as this eccentricity is beyond the eccentricity at which most gamers play.
VGPs also show a clear advantage in localization with or without the presence
of distracting objects. The superior performance in the no distractors condition
indicates an enhancement at localizing abrupt onsets in the visual periphery. The very
36
brief amount of time the stimulus is displayed (<15 ms) appears sufficient to create a
detectable change in the visual field that is more easily localized by the VGP
population than the NVGP population. While this condition requires the subject to
locate abrupt onsets and so may draw on exogenous attention, it is also possible that
improvement on this condition could be due to more perceptual factors. The
advantage in the distractors present blocks indicates that video game experience
increases the ability to select targets amongst distractors. Therefore, although the
results of Experiment 1 could have been attributed to an increase in distractibility in
VGPs, the findings of Experiment 2 conclusively demonstrate that not only are more
resources available to VGPs, but this enhanced attention can act to increase target
selection. This is consistent with previous reports which have found a positive
relationship between increased attention and enhanced visual selection (Carrasco &
Yeshurun, 1998; Eckstein, Shimozaki, & Abbey, 2002; Palmer, 1994).
Finally, when the center task was added, VGPs continue to substantially
outperform the NVGPs. VGPs perform both tasks easily and in fact, their localization
performance shows no effect of the added center task. Conversely, NVGPs show a
small decrease in task performance with the addition of a center task. The size of the
falloff is consistent with previous work on the UFOV paradigm, namely relatively
modest decreases in peripheral localization performance with the addition of a center
discrimination task in younger observers, with substantially larger effects being seen
in the elderly (Ball et al., 1988).
37
Importantly, VGPs outperform NVGPs on the center task itself suggesting

that no trade-off of attentional distribution is involved (although we note that it could
be the case that the central task was simply easier for the VGPs). Essentially, VGPs
can perform both tasks with near perfect accuracy; this suggests that the load of these
two tasks combined is below their capacity limit for dual-task performance, whereas
NVGPs show lessened performance at both tasks, suggesting their capacity limit is
substantially lower. This mirrors the predictions given by the results of Experiment 1
in which NVGPs were seen to have fewer attentional resources than VGPs both
peripherally and centrally.
While our hypothesis predicts that extensive video game playing leads to
these enhanced skills, it could also be the case that VGPs have inherently better visual
skills and/or were somehow genetically endowed with greater attentional abilities. To
demonstrate a causative role of action video game play in these effects, a group of
NVGPs were trained on an action video game in Experiment 3. If the effects are due
to action video game experience, similar enhancements in localization performance
should be observed following training.
4. Experiment 3 Training Study

NVGPs were divided into two training groups. Half underwent video game
training using an action video game, whereas the others played a game that made
heavy demands on visuo-motor coordination but unlike action video games did not
require the subject to process multiple objects at once at a fast pace. This control
38
group was added to check for another possible explanation for the difference between
VGPs and NVGPs whereby what is learned during video game play is not necessarily
visual in nature, but is instead visuo-motor. Although the use of percent correct, and
not reaction time, should minimize the effect of visuo-motor coordination in our
measures, it is possible that by alleviating the demands of the motor response, video
game playing allows VGPs to have more left-over resources available to process
the stimulus. If the differences observed in Experiment 2 are due to an attentional
enhancement and not due to lightened visuo-motor control or genetically endowed
traits, a notable improvement in UFOV performance should be observed following
training in the action game trainees, but not in the control game trainees. Unlike
Experiments 1 and 2 that only included males, Experiment 3 included half males and
half females, allowing us to test the generality of our findings to both sexes.
Based on the results of Experiment 2 as well as pilot training data, several
modifications were made to the design of the paradigm. Among these were to
remove feedback in order to minimize the amount of task-related learning that
occurred during testing. Also, a more difficult center discrimination task was
employed to avoid potential ceiling effects, as performance on the center
identification task of Experiment 2 was quite high.
4.1. Method
4.1.1. Participants
39
The study enrolled 32 NVGPs, none of whom had taken part in Experiments 1
or 2, who were equally and randomly divided between the experimental and the
control group. The criteria for NVGP remained the same as in all previous
experiments. All subjects underwent training as described below. In all 8 females and
8 males (mean age = 21.3, all right-handed) made up the final experimental group,
while the final control group consisted of 9 females and 7 males (mean age = 21.0, 15
right handed).
4.1.2. Pre-Test
Subjects underwent a slightly modified version of the previous tasks. First,
only four blocks were run - two no center task blocks and two center task present
blocks each with a no distractors and a distractors present condition. Second, a fine
orientation discrimination task was selected for the center task. The difficulty of the
center task was manipulated based on pilot data to lead to around 70% correct
performance making it a far more difficult center task than that used in Experiment 2.
Third, because the by-eccentricity timing manipulations had failed to yield equal
performance across eccentricities in Experiment 2, each eccentricity was tested with a
13 ms stimulus presentation duration. Fourth, a white noise mask was chosen, as
participants found the pattern mask used in the previous experiments especially
disrupting, and there were concerns that this difficult pattern mask may have been
disproportionately disruptive to NVGPs as compared to VGPs. Finally, to minimize
the effect of test-retest improvements, no feedback was given. Because no effect of
40
center task order was found in Experiment 2, and because of the presence of several
other tasks unrelated to the task at hand, subjects were always tested on the no center
task condition first, then the center task condition (making the run order: no center
task/no distractors, no center task/distractors present, center task present/no
distractors, center task present/distractors present). Finally, to minimize any testretest effects each subject underwent only 72 trials (eight spokes x three eccentricities
x three repetitions of each) for each condition.
4.1.3. Apparatus
4.1.3.1. Testing
The apparatus was identical to that described in Experiment 2 except
the monitor was a ViewSonic P817 21-in monitor (ViewSonic, Walnut, CA).
4.1.3.2. Training:
Both groups played on 20 monitors.
4.1.4. Training Stimuli/Procedure

For both groups, training consisted of playing the pre-determined video game
for 30 total hours (maximum of 2 hours per day, minimum of 5 hours per week,
maximum of 8 hours per week). The sixteen members of the experimental group
played the game Unreal Tournament 2004 (henceforth referred to as the action video
41
game). This game was chosen to be similar to those played by our VGPs. It has a
relatively simple interface, uses first-person point of view and requires effective
monitoring of the entire visual field (extent from fixation about 13-height x 16width). Each hour session was divided into three 20 minute blocks. The difficulty of
each block was adjusted based upon the kill/death ratio. If in a block the player
scored twice as many kills as they had deaths, the difficulty level was increased one
level. Players were re-tested on lower difficulty levels on the final two days of
training to quantitatively assess improvement.
The sixteen members of the control group played the game Tetris, which was
displayed to cover the entire extent of the screen. The FOV of the Tetris game was
actually slightly larger than that of the action game - the effective game area extended
18-height x 13-width from fixation. This game was selected to control for the
effect of improved visuo-motor coordination, while putting little demands on the
processing of multiple objects at once. Accordingly, the version of Tetris on which
subjects were trained had the preview block option turned off. In a manner analogous
to the action-trained group, improvement was quantitatively measured by comparing
performance on day 1 versus that on day 30.
4.1.5. Post-Test
After video game training, subjects were re-tested on the same experiment as
in the pre-test, as well as the other aforementioned unrelated tasks.
42
4.2. Results
4.2.1. Game Play:

In order to assess game improvement, several measures were used with a
percent change score being calculated for each. For the action game, the two
measures used were kills and death. For each of five levels of game difficulty (level
five being the highest level that all players attained) the measure taken on their first
playing of the level (which, because of the way in which difficulty was progressed
was not necessarily on the first day of training) was compared with their final playing
of the level on Days 29-30. A substantial increase in number of kills, decrease in
number of deaths, and increase in the ratio of kills/deaths was seen at each difficulty
level (Level 1: 226% increase in kills, 64% decrease in deaths; Level 2: 147%
increase in kills, 38% decrease in deaths; Level 3: 160% increase in kills, 27%
decrease in deaths; Level 4: 80% increase in kills, 33% decrease in deaths; Level 5:
52% increase in kills, 32% decrease in deaths).
For the control game, the mean and median scores from Day 1 were compared
with the same values on Day 30. As in the action game, the control players showed
substantial improvements after training, the mean score improving by 323% and the
median score by 359%.
These results demonstrate that both groups were engaged in their training and
showed improvement on the training task.
43
4.2.2. UFOV Task

Accuracy was analyzed in four 2x2x3 ANOVAs, blocked by distractor level
and center task condition, with game trained (action/control), test (pretest/posttest),
and eccentricity (10, 20, 30) as factors. As in Experiment 2, center task present
trials were first filtered to include only those trials wherein the center task was
correct.
As in Experiment 2, the observation of near-ceiling performance in some cells
led to an analysis with arcsin transformed data. In only one case did a non-significant
p-value in the untransformed analyses become significant in the arcsin transform
analyses (noted in section 4.2.2.1) and in no cases did a significant p-value in the
untransformed analyses become non-significant in the arcsin transformed analyses.
As in Experiment 2, only the untransformed analyses are presented.
4.2.2.1. Peripheral Localization Accuracy - No Distractors, No Center Task

A main effect of eccentricity was observed (F(2,60) = 3.9, p < .05) with
accuracy decreasing with increasing eccentricity. A main effect of test was observed
(F(1,30) = 4.5, p < .05) with accuracy improving from pre- to post-test. However,
although the interaction between game trained and test did not reach significance
(F(1,30) = 3.8, p = .06, Figure 4A), it was in the direction predicted by our
hypothesis, with the action group improving more than the control group. The ability
to see a clear difference between groups was likely hindered by the near ceiling
performance in this condition. It is also worth noting that this effect would be
44
statistically significant assuming a one-tailed test, which would be justified given our
specific prediction of greater improvements in the action trained group, and also that
this effect was significant (F(1,30) = 4.6, p < .05) in the arcsin transformed analysis.
4.2.2.2. Peripheral Localization Accuracy - No Distractors, Center Task Present

Main effects of eccentricity (F(2,60) = 37.2, p < .001) with accuracy
decreasing with increasing eccentricity and of test (F(1,30) = 10.5, p < .01) with
accuracy increasing on the post-test relative to the pre-test were observed. Also
observed was a game trained by test interaction (F(1,30) = 6.8, p < .05) caused by a
greater improvement in the action game than the control game (Figure 4B).
4.2.2.3. Peripheral Localization Accuracy - Distractors Present, No Center Task

A main effect of eccentricity (F(2,60) = 72.4, p < .001) was observed. Main
effects of game trained (F(1,30) = 4.9, p < .05) and of test (F(1,30) = 25.4, p < .001)
were observed with the action group being more accurate than the control group, and
subjects performing better on the post-test than the pre-test. A significant interaction
between game trained and test (F(1,30) = 12.8, p = .001) indicates that the action
group improved significantly more than the control group, which is likely the root of
the main effect of game trained. Finally, a significant interaction between trained
game, test, and eccentricity (F(2,60) = 4.8, p < .05) was caused by the action groups
performance falling off less steeply with eccentricity following training (Figure 4C).
45
4.2.2.4. Peripheral Localization Accuracy - Distractors Present, Center Task Present

Only a main effect of eccentricity (F(2,60) = 70.3, p < .001) and a game
trained by test interaction (F(1,30) = 5.1, p < .05) were observed with again, accuracy
decreasing with increasing eccentricity, and the action group improving by a larger
margin than the control group (Figure 4D).
46
Figure 4: Experiment 3 - Accuracy of target localization as a function of

eccentricity, training group, and test
The action trained group showed a significantly greater improvement in localization accuracy than the
control group following training at each eccentricity for all of the conditions other than the no
distractors/no center task condition where the result was nearly significant (p = .06).
4.2.2.5. Center Identification Task Performance

First, although the center task performance was knocked far from ceiling,
there was no general increase in center discrimination task accuracy following
47
training, and this held true for both groups (action trained: pretest: 61.3%+/-3.0,
posttest: 66.6% +/- 4.3, control trained: pretest: 66.6% +/- 3.1, posttest: 67.5% +/3.7; main effect of test p > .05; interaction between test and game trained, p > .5). A
main effect of distractor level (F(1,30) = 7.6, p = .01) indicates that central task
performance was affected by the demands of the localization task with worse central
task performance for the distractors present than the no distractors condition.
4.3. Discussion
The results from Experiment 3 establish that the act of playing an action video
game improves performance on the UFOV task. Importantly, action trained subjects
showed greater training induced improvements than subjects trained on a control
game that relies heavily on eye-hand coordination. Thus, improvement after action
game training cannot be attributed to a general test-retest advantage or to the fact that
video game training facilitates visuo-motor coordination. Instead, action video game
play appears to truly modify visuo-spatial attention.
As in Experiment 2, the action video game trained group improved their
localization ability at all eccentricities even at 30, which is beyond the maximum
eccentricity of game training. This result confirms that the effects of action video
game play do generalize to untrained locations in the visual field. The action trained
group also improved their localization performance both with and without the
presence of distractors, confirming that action video game play does enhance the
48
ability to monitor the peripheral visual field and also to select targets from within a
field of distractors.
Finally, a strong effect of training was seen on peripheral localization both
without and with the presence of a center task in the action-trained group. The finding
that the action- trained group outperforms the control trained group even when a
center task is added demonstrates that the enhanced peripheral localization
performance of action gamers is not at the cost of central performance. Unlike in
Experiment 2, the central task was equally difficult for both groups. As the
participants in Experiment 2 have many more hours of training than those of
Experiment 3, this pattern of result is consistent with the view that visual performance
may be harder to modify in central than in peripheral vision. Finally, the equivalent
performance on the center task in the two training groups in Experiment 3
demonstrates that the center task was perceptually as demanding in action-trained
and control trained group. It is therefore unlikely that the enhanced performance
induced by action game training could be due to perceptual factors. Rather the
proposal that action video game training enhances attentional resources over the
whole field offers a more parsimonious explanation of the data presented.
5. Relationship between action game improvement and UFOV improvement
The goal of Experiment 3 was to establish a causal link between action video
game play and enhanced performance on the UFOV task. By showing that subjects
49
required to play action video games display greater improvements on a visuo-spatial

attention task than subjects required to play a control game, this study establishes a
causal link between action game play and attentional capacities. What remains
unclear, however, is whether performance on an action video game can be used to
predict how good visual selective attention is. On the one hand, it is clear that
individuals who play action video games are better at these games than those who do
not play. Combined with our finding that action game players outperform non-players
on a visual selection task, it would seem natural to hypothesize that performance on
action game play does predict attentional resources, at least as tested by the UFOV.
On the other hand, action video games are extremely rich, not only visually but also
strategically. There are therefore many ways to excel at action video game playing.
While visual selective attention skills likely contribute strongly to action game
success, other aspects of cognition such as planning, memorizing landmarks and the
lay of the land, and understanding the strengths and weaknesses of the various
characters, weapons, and positions also play a large role in performance. In addition,
the behavioral measures provided by commercially available games are rather coarse.
For instance, shooting accuracy is greatly influenced by the type of gun the player
uses (a fully automatic weapon requires less accuracy than a semiautomatic weapon
to achieve the same ends). Other measures such as kills and deaths are likely much
better correlated with high level strategies (which weapon to use, when to hide and
when to attack, remembering to replenish health/armor, etc) than low-level perceptual
skills. Nevertheless for Experiment 3 we tested whether there was a correlation
50
between improvement in action game performance and improvement on the UFOV

task.
5.1. Correlation Analysis

Percent improvement for each of the individual UFOV conditions in
Experiment 3 (No Distractors/No Center Task, No Distractors/Center Task Present,
Distractors Present/No Center Task, Distractors Present/Center Task Present) was
correlated with percent change in number of kills and in number of deaths for each of
the five levels of game difficulty. If those subjects who improved their scores by the
greatest margin on a particular condition of the UFOV task also demonstrated the
greatest increase in either of these measures, a significant relationship should be
observed. Out of the forty correlations (four UFOV conditions by five kill
improvement scores and five death improvement scores) there were no significant
correlations (using a Bonferoni corrected p-value of .00125)
5.2 Discussion
This analysis indicates a lack of correlation between available measures of
game improvements and UFOV improvement. Although the finding of a such
positive correlation would have nicely complemented the finding of a causal
relationship between number of hours of action game play and UFOV performance, it
still remains that the very act of playing action games at a challenging level enhances
performance on a visual selection task to a greater extent than playing other, similarly
51
challenging, control games. It will be for future research to further assess the link
between the exact level of game expertise of a player and the quality of its visual
selective attention. This may require the development of finer outcome measures for
action video games improvement that do not mix perceptual aspects of gaming with
various strategic decisions.
6. General Discussion
The results of these three experiments demonstrate the positive effect of

action video game play on visuo-spatial attention. First, by measuring the
compatibility effect as a function of perceptual load, we were able to gain a measure
of the attentional resources available to VGPs and NVGPs (Lavie et al., 2004). VGPs
continued to show compatibility effects at greater perceptual loads than NVGPs
confirming the proposal that VGPs have enhanced attentional resources. This effect
held for both central and peripheral distractors, suggesting an overall enhancement in
attentional resources in VGPs rather than some manner of tradeoff.
As seen in several recent studies (Beck & Lavie, 2005; Proksch & Bavelier,
2002), there was less of a decrease in the compatibility effect with increasing load for
central distractors than peripheral distractors. It has been suggested that this is due to
preferential access to attentional resources around fixation, at least in the hearing
population (Beck & Lavie, 2005). Interestingly, the differences seen in VGPs
between low and high perceptual load conditions suggest that they may allocate
52
attention in a more dynamic manner across space. During conditions of low

perceptual load, VGPs distributed attention proportionally more to the visual
periphery, but as the perceptual load of the primary task was increased, attention was
shifted to central vision. One can easily imagine how such a configuration would be
useful during action video game play. During low-load conditions, such as when
the player walks down an empty path, it would be beneficial for attention to be shifted
toward the periphery in order to best detect any incoming enemies. However, during
high-load conditions, such as when the player is being attacked by charging
enemies, it would be beneficial for those resources to be shifted to the area around the
player (near fixation). In all, these results suggest that while the sizing of the
attentional window may be largely automatic, it need not occur with the same spatial
distribution in every population or in a static manner across levels of perceptual load.
A more direct measure of the effect of gaming on visuo-spatial attention was
obtained by using the UFOV paradigm (Ball et al., 1988) in which subjects were
asked to localize a very briefly presented target stimulus at multiple eccentricities,
with and without the presence of distracting items, and with and without the presence
of a concurrent center task. VGPs showed increased localization abilities both within
and outside of their typical game playing field of view, indicating that the effect of
action video game play generalizes to untrained locations. VGPs also showed a
decided advantage in localization accuracy over NVGPs both when distractors were
absent and when they were present. The former demonstrates that VGPs display a
general enhancement in the ability to detect an abrupt onset target in the visual
53
periphery. While it is known that this type of task draws on exogenous attention
(Yantis & Jonides, 1990), it is also possible that improvement on this condition could
also be accounted for by perceptual, rather than purely attentional factors. The VGP
advantage in the distractors present condition indicates that VGPs are better able to
select targets amongst distractors than NVGPs. The fact that the distractors present
condition requires the successful selection of the target from amongst competing
alternatives suggests that VGPs do indeed display an enhancement in visual selective
attention, which has been shown to increase the spatial resolution of visual processing
(Carrasco, Williams, & Yeshurun, 2002; Talgar & Carrasco, 2002; Yeshurun &
Carrasco, 1998).
Alternative hypotheses that have purely perceptual factors at the root of the
differences observed in the distractors present condition are unlikely for several
reasons. First, as mentioned briefly in the introduction, performance on the UFOV is
quite poorly correlated with basic perceptual skills (acuity, contrast sensitivity,
perimetry) (Ball et al., 1990; Owsley et al., 1995). In fact, the UFOV was initially
designed by Ball and colleagues specifically because driving accidents in the elderly
were found to be poorly predicted by basic perceptual abilities, and a test was
required that better tapped the visual attentional requirements present while driving
(peripheral monitoring, target selection, distractor rejection). Of special note is the
fact that the relationship between low-level perceptual skills and UFOV performance
is particularly weak for the distractors present condition (Owsley, et al., 1995).
Second, the type of low-level sensory enhancements that could potentially underlie
54
increased performance in the no distractors condition, such as signal enhancement,

would be of much less use in the distractors present condition in which target
selection is of the essence.
Unfortunately, our design does not allow us to address the issue of search
efficiency in which the rate of visual selection is measured by systematically varying
set size. Although the UFOV task manipulates the number of distractors, our design
used steps in distractor number that are too large (0, 23 or 47 distractors) as well as
blocked order of distractor presentation, preventing any estimation of the rate of
visual search proper. However, a recent study comparing visual search skills in action
video game players and non players by Castel et al (2005) speaks to this issue. In the
Castel et al (2005) paper, video game players were found to have faster response
times for both easy (feature/parallel) and difficult (conjunction/serial) visual search
displays - an advantage that held for displays ranging from four to twenty-six items.
The difference was quite large, with VGPs performing the difficult search task with a
set-size of twenty-six in less time than the NVGPs performed the task with a set-size
of eighteen. Importantly, as would be predicted by an increase in search efficiency in
VGPs, Castel et al (2005) observed a significant interaction between group
(VGP/NVGP) and set-size. The authors state that this interaction indicates that
videogame players are more efficient in searching through displays than the nonvideogame players (Castel et al, 2005, pp. 226). However, because the set-size by
group interaction disappeared when only the largest set-sizes were included in the
analysis (set-sizes of 10+) the authors concluded that the VGP RT advantage might
55
be better understood as a change in stimulus-response mappings rather than a change

in visual selective attention. Based on the results of the current study, it appears as
though the initial interpretation of increased search efficiency in VGPs may have
some validity. This proposal does not discount the possibility that some proportion of
the difference between VGP and NVGP reaction time in their experiment could be
accounted for by stimulus-response mappings or other mechanisms. One might
hypothesize that the mechanism(s) leading to enhanced search in VGPs is capacity
limited, the effect of which on total reaction time would necessarily be diminished
with increasing set size. It will be for future research to tease out the roles of
sensitivity, criteria, and residual response time as it pertains to VGPs in visual search.
Finally, the use of the UFOV paradigm also allowed us to study the effect of
action game play on dual-task performance. When a concurrent center task was added
to the peripheral localization task, NVGPs showed a clear cost of the added center
load in the form of decreases in peripheral localization ability, while VGPs showed
no such effect suggesting an improvement in the ability to dual-task in VGPs. These
results are well predicted by the increased capacity of visual attention in VGPs seen
in Experiment 1. In addition, they unambiguously rule out the possibility that VGPs
may show increased peripheral performance at the cost of central vision. The
attentional resources of VGPs are sufficient to perform both central and peripheral
tasks at high accuracy. In contrast, the center task depletes the resources of NVGPs
and therefore decreases peripheral localization. Experiment 3 demonstrated that the
effects noted in Experiment 2 can be induced by training, establishing a causative
56
relationship between action video game experience and performance enhancement on

the UFOV. Along with the results of Experiment 1, these findings establish that
action video game experience improves both central and peripheral visuo-spatial
attention by increasing attentional resources and facilitating visual selective attention.
57
References
Akhtar, N., & Enns, J. T. (1989). Relations between covert orienting and filtering in
the development of visual attention. Journal of Experimental Child
Psychology, 48(2), 315-334.
Ball, K., Beard, B., Roenker, D., Miller, R., & Griggs, D. (1988). Age and Visual
Search: Expanding the Useful Field of View. J. Optical Society of America,
A., 5(10), 2210-2219.
Ball, K., & Owsley, C. (1991). Identifying correlates of accident involvement for the
older driver. Human Factors, 33(5), 583-595.
Ball, K., & Owsley, C. (1992). The useful field of view test: a new technique for
evaluating age-related declines in visual function. J Am Optom Assoc, 63, 7179.
Ball, K., Owsley, C., & Beard, B. (1990). Clinical visual perimetry underestimates
peripheral field problems in older adults. Clinical Vision Sciences, 5, 113-125.
Ball, K., Owsley, C., Sloane, M., Roenker, D., & Bruni, J. (1993). Visual attention
problems as a predictor of vehicle crashes in older drivers. Investigative
Ophthalmology, 34(11), 3110-3123.
Beck, D. M., & Lavie, N. (2005). Look here but ignore what you see: effects of
distractors at fixation. Journal of Experimental Psychology: Human
Perception and Performance, 31(3), 592-607.
Brainard, D. G. (1997). The Psychophysics Toolbox. Spatial Vision, 10, 433-436.
58
Carrasco, M., Williams, P. E., & Yeshurun, Y. (2002). Covert attention increases
spatial resolution with or without masks: support for signal enhancement.
Journal of Vision, 2(6), 467-479.
Carrasco, M., & Yeshurun, Y. (1998). The contribution of covert attention to the setsize and eccentricity effects in visual search. Journal of Experimental
Psychology: Human Perception and Performance, 24(2), 673-692.
Castel, A. D., Pratt, J., & Drummond, E. (2005). The effects of action video game
experience on the time course of inhibition of return and the efficiency of
visual search. Acta Psychologica, 119, 217-230.
Eckstein, M., Pham, B. T., & Shimozaki, S. S. (2004). The footprints of visual
attention during search with 100% valid and 100% invalid cues. Vision
Research, 44, 1193-1207.
Eckstein, M., Shimozaki, S. S., & Abbey, C. K. (2002). The footprints of visual
attention in the Posner cueing paradigm revealed by classification images.
Enns, J. T., & Cameron, S. (1987). Selective attention in young children: the relations
between visual search, filtering, and priming. Journal of Experimental Child
Psychology, 44(1), 38-63.
Enns, J. T., & Girgus, J. S. (1985). Developmental changes in selective and
integrative visual attention. Journal of Experimental Child Psychology, 40(2),
319-337.
59
Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters upon the
identification of a target letter in nonsearch task. Perception &
Psychophysics, 16(1), 143-149.
Goode, K. T., Ball, K. K., Sloane, M., Roenker, D. L., Roth, D. L., Myers, R. S., et al.
(1998). Useful field of view and other neurocognitive indicators of crash risk
in older adults. Journal of Clinical Psychology in Medical Settings, 5(4), 425441.
Gopher, D. (1992). The skill of attentional control: acquisition and execution of
attention strategies. In D. E. Meyer & S. Kornblum (Eds.), Attention and
Performance XIV (pp. 299-322). Cambridge, Massachusetts: The MIT Press.
Gopher, D., Weil, M., & Bareket, T. (1994). Transfer of skill from a computer game
trainer to flight. Human Factors, 36(3), 387-405.
Green, C. S., & Bavelier, D. (2003). Action video game modifies visual selective
attention. Nature, 423, 534-537.
Greenfield, P. M., DeWinstanley, P., Kilpatrick, H., & Kaye, D. (1994). Action video
games and informal education: effects on strategies for dividing visual
attention. Journal of Applied Developmental Psychology, 15, 105-123.
Husain, M., & Kennard, C. (1997). Distractor-dependent frontal neglect.
Neuropsychologia, 35(6), 829-841.
Intriligator, J., & Cavanagh, P. (2001). The spatial resolution of visual attention.
Cognitive Psychology, 43(3), 171-216.
60
Lavie, N. (1995). Perceptual load as a necessary condition for selective attention.

Journal of Experimental Psychology: Human Perception and Performance,
21, 451-468.
Lavie, N. (2005). Distracted and confused?: Selective attention under load. Trends in
Cognitive Sciences, 9(2), 75-82.
Lavie, N., & Cox, S. (1997). On the efficiency of visual selective attention: Efficient
visual search leads to inefficient distractor rejection. Psychological Science,
8(5), 395-398.
Lavie, N., Hirst, A., Viding, E., & de Fockert, J. W. (2004). Load theory of selective
attention and cognitive control. Journal of Experimental Psychology: General,
133(3), 339-354.
Levinoff, E. J., Li, K. Z., Murtha, S., & Chertkow, H. (2004). Selective attention
impairments in Alzheimer's disease: evidence for dissociable components.
Neuropsychology, 18(3), 580-588.
Madden, D. J., & Langley, L. K. (2003). Age-related changes in selective attention
and perceptual load during visual search. Psychol Aging, 18(1), 54-67.
Maylor, E. A., & Lavie, N. (1998). The influence of perceptual load on age
differences in selective attention. Psychol Aging, 13(4), 563-573.
Mazer, B., Sofer, S., Korner-Bitensky, N., & Gelinas, I. (2001). Use of the UFOV to
evaluate and retrain visual attention skills in clients with stroke: A pilot study.
American Journal of Occupational Therapy, 55, 552-557.
61
Miller, J. (1991). The flanker compatibility effect as a function of visual angle,

attentional focus, visual transients, and perceptual load: A search for boundary
conditions. Perception and Psychophysics, 49(3), 270-288.
Myers, R. S., Ball, K. K., Kalina, T. D., Roth, D. L., & Goode, K. T. (2000). Relation
of useful field of view and other screening tests to on-road driving
performance. Percept Mot Skills, 91(1), 279-290.
Owsley, C., & Ball, K. (1993). Assessing visual function in the older driver. Clinics
in Geriatric Medicine, 9(2), 389-401.
Owsley, C., Ball, K., & Keeton, D. M. (1995). Relationship between visual sensitivity
and target localization in older adults. Vision Research, 35(4), 579-587.
Palmer, J. (1994). Set-size effects in visual search: the effect of attention is
independent of the stimulus for simple tasks. Vision Research, 34(13), 17031721.
Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics:
Transforming numbers into movies. Spatial Vision, 10, 437-442.
Plude, D. J., Enns, J. T., & Brodeur, D. (1994). The development of selective
attention: a life-span overview. Acta Psychologica, 86, 227-272.
Plude, D. J., & Hoyer, W. J. (1986). Age and the selectivity of visual information
processing. Psychol Aging, 1(1), 4-10.
Proksch, J., & Bavelier, D. (2002). Changes in the spatial distribution of visual
attention after early deafness. Journal of Cognitive Neuroscience, 14, 1-5.
62
Rovamo, J., & Virsu, V. (1979). An estimation and application of the human cortical
magnification factor. Experimental Brain Research, 37, 495-510.
Rueda, M. R., Fan, J., McCandliss, B. D., Halparin, J. D., Gruber, D. B., Lercari, L.
P., et al. (2004). Development of attentional networks in childhood.
Neuropsychologia, 42(8), 1029-1040.
Russell, C., Malhotra, P., & Husain, M. (2004). Attention modulates the visual field
in healthy observers and parietal patients. Neuroreport, 15(14), 2189-2193.
Scialfa, C. T., Esau, S. P., & Joffe, K. M. (1998). Age, target-distractor similarity,
and visual search. Experimental Aging Research, 24(4), 337-358.
Sekuler, R., & Ball, K. (1986). Visual localization: age and practice. J. Optical
Society of America, A., 3(6), 864-867.
Sekuler, R., Bennett, P., & Mamelak, M. (2000). Effects of aging on the useful field
of view. Experimental Aging Research, 26, 103-120.
Shalev, L., & Tsal, Y. (2003). The wide attentional window: a major deficit of
children with attention difficulties. Journal of Learning Disabilities, 36(6),
517-527.
Sprenger, A., Kompf, D., & Heide, W. (2002). Visual search in patients with left
visual hemineglect. Prog Brain Res, 140, 395-416.
Tales, A., Haworth, J., Nelson, S., Snowden, R. J., & Wilcock, G. (2005). Abnormal
visual search in mild cognitive impairment and Alzheimer's disease.
Neurocase, 11(1), 80-84.
63
Tales, A., Muir, J., Jones, R., Bayer, A., & Snowden, R. J. (2004). The effects of
saliency and task difficulty on visual search performance in ageing and
Alzheimer's disease. Neuropsychologia, 42(3), 335-345.
Talgar, C. P., & Carrasco, M. (2002). Vertical meridian asymmetry in spatial
resolution: visual and attentional factors. Psychonomic Bulletin and Review, 9,
714-722.
Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention.
Cognitive Psychology, 12, 97-136.
Trick, L. M., Jaspers-Fayer, F., & Sethi, N. (2005). Multiple-object tracking in
children: The "Catch the Spiers" task. Cognitive Development, 20(3), 373387.
Vivas, A. B., G.W., H., & Fuentes, L. J. (2003). Inhibitory processing following
damage to the parietal lobe. Neuropsychologia, 41(11), 1531-1540.
Yantis, S., & Jonides, J. (1990). Abrupt visual onsets and selective attention:
voluntary versus automatic allocation. Journal of Experimental Psychology:
Human Perception and Performance, 16, 121-134.
Yeshurun, Y., & Carrasco, M. (1998). Attention improves or impairs visual
performance by enhancing spatial resolution. Nature, 396(6706), 72-75.
64
Acknowledgements: We thank T. Monacelli, D. McColgin, J. Cohen for help with

data acquisition. This research was supported by grants from the NIH, the ONR, and
the James S. McDonnell-Pew Foundation to D.B.
65
1.1.1.2. The spatial resolution of visual attention

Green, C.S. & Bavelier, D. 2007. Action-video-game experience alters the spatial
resolution of vision. Psychological Science 18(1), 88-94.
version in terms of formatting, precise wording, etc
Action video game experience alters the spatial resolution of vision
Department of Brain and Cognitive Sciences, University of Rochester

Telephone: (585) 273-4863
Fax: (585) 442-9216
[email protected]
66
Abstract
Action video game play enhances several different aspects of visual processing
(Green & Bavelier, 2003); however the mechanisms underlying this improvement
remain unclear. Here we show that action video game play can alter fundamental
characteristics of the visual system such as the spatial resolution of visual processing
across the visual field. To determine the spatial resolution of visual processing we
measured the smallest distance a distractor can be from a target without
compromising target identification. This approach exploits the fact that visual
processing is hindered as distractors are brought nearer to the target, a phenomenon
known as crowding. Video game players could tolerate smaller target-distractor
distances, establishing an enhancement of the spatial resolution of visual processing
in this population. Critically, similar effects were observed in non-video game players
that were trained on an action video game, thus verifying a causative relationship
between video game play and augmented spatial resolution.
67
Video game players appear to outperform non -game players on several

different visual tasks (Castel, Pratt, & Drummond, 2005; Gopher, Weil, & Bareket,
1994; Green & Bavelier, 2003, 2006, In Press; Greenfield, DeWinstanley, Kilpatrick,
& Kaye, 1994; Trick, Jaspers-Fayer, & Sethi, 2005). For example, avid action video
game players (VGPs) were found to localize a peripheral target in a field of
distracting objects more accurately than NVGPs, to process a visual stream of briefly
presented objects more efficiently and to track more objects at once than NVGPs.
Non-game players trained on an action video game exhibited similar improvements
establishing that the very act of action video game play results in an increased ability
to perform such complex visual tasks (Green & Bavelier, 2003, In Press). Although
the effect of game play on complex visual tasks is striking, it remains unclear whether
the improvements previously described may be mainly due to strategic changes or to
changes in more fundamental aspects of visual processing. We show here using a
crowding paradigm that video game play alters the spatial resolution of vision - one
of the fundamental characteristics of vision.
Crowding refers to the general finding that it is substantially more difficult to
identify a target object when other distracting objects are present in its immediate
vicinity than when the target object is presented in isolation. This decrement in
performance is a function of the number of distractors and the distance between the
target and the distractor(s), with performance decreasing as the number of distractors
increases, as well as when targets and distractors are moved closer together (Leat, Li,
& Epp, 1999; Miller, 1991; Orbach & Wilson, 1999). The region in space around a
68
target wherein the presence of distracting objects leads to decreased sensitivity for the
target is known as the crowding region or the zone of spatial interaction (Bouma,
1970; Flom, Weymouth et al., 1963; Intriligator & Cavanagh, 2001; Jacobs, 1979;
Toet & Levi, 1992), and in our everyday life limits our ability to identify letters or
words embedded in text for example.
Several lines of explanation for visual crowding have been advanced. Some
have suggested important roles for interactions between the facilitatory and inhibitory
regions within neuronal receptive fields in early visual areas or interactions between
neurons via long-range horizontal connections (Flom, Heath, & Takahashi, 1963;
Polat & Sagi, 1994; Tripathy & Levi, 1994). Others have suggested that crowding,
particularly foveal crowding, is a form of spatial frequency masking or contrast
masking (Chung, Levi, & Legge, 2001; Levi, Klein, & Hariharan, 2002) or is related
to the physical properties of the stimulus (Hess, Dakin, & Kapoor, 2000).
Alternatively, it has been proposed that the size of the crowding region offers a
measure of the resolution of visual attention (Intriligator & Cavanagh, 2001; Tripathy
& Cavanagh, 2002). Importantly for our purpose, whichever interpretation of
crowding may prevail all parties agree that crowding reflects a fundamental limitation
on the spatial resolution of the visual system.
In this paper we test the hypothesis that action video game play leads to
enhanced spatial resolution, by using a crowding paradigm modeled after Toet and
Levi (1992). Experiment 1 establishes that avid action game players (VGPs) exhibit
smaller crowding regions than non-game players (NVGPs), meaning that VGPs
69
maintain high performance at target-distractor separations that hinder NVGPs

performance. In Experiment 2 we use a training paradigm to demonstrate a causative
relationship between action video game experience and performance in a training
study.
Experiment 1
The proposal of enhanced spatial resolution in VGPs predicts that the size of
the crowding region will be smaller in VGPs than NVGPs. The size of the crowding
region was measured by testing the ability of participants to discriminate between a
right-side up and an upside-down T as a function of the distance between this target
object and two flanking distractor T shapes presented above and below the target
(Figure 1) (Toet & Levi, 1992). The size of the crowding region was assessed at
three different eccentricities (0, 10, and 25) chosen to allow the measurement of
potential changes both well within (0 and 10) and just at the limit of (25) the field
of view of normal game playing (most of our gamers reported playing on screens that
subtended an average of +/-15 from the center of the screen). If changes in the size
of the crowding region do arise as a result of action video game experience, but the
learning is specific for the trained region of space, one would predict greater
enhancements for 0 and 10 than for 25. Conversely, if similar changes were
observed across eccentricities, it would be evidence for generalization of the learning
beyond the more highly trained regions of space.
70
Figure 1 - Stimuli
The stimuli consisted of three T shapes randomly oriented either right-side up or upside down. The
subjects task was to indicate the orientation of the center T. In separate blocks, three eccentricities
were tested--0, 10, and 25. The size of the Ts was set to be 1.5 times each individual subjects Talone threshold at each given eccentricity.
Method
Participants
Twenty right-handed males (all undergraduates at the University of
Rochester) with normal or corrected-to-normal vision were placed into one of two
groups, VGP or NVGP, based on the outcome of an interview about their video game
playing habits (only males were tested in Experiment 1 because of the relative
scarcity of female action video game players). Participants were asked about different
types of video games in turn (action, sports, fantasy, role playing, other), and for each
type, were asked to name all the games they had played in the past 12 months. For
each video game the participants reported playing, they were asked how often they
played that game in the previous 12 months (daily, weekly, monthly or less), and if
applicable, how many hours they spent playing per week (0, 0-1, 1-2, 3-5, 5-10, 10+)
and for how long they played it during a typical session.
71
To be considered a video game player, a subject needed to report a minimum

of 5 hours a week of action video game usage for the previous six months (N=10,
mean age= 19.7 years). The criterion to be considered a non-video game player was a
report of zero hours per week of action games for at least the previous six months
(N=10, mean age =20.3 years; note: playing other games was allowed). Subjects
provided written informed consent and were paid $8/hour.
Apparatus
The apparatus consisted of a MacIntosh G3 computer running a program to
present stimuli and collect the data using the Matlab computer language (The Math
Works Inc., Natick, MA) and the Psychophysical Toolbox routines (Brainard, 1997;
Pelli, 1997- http://psychtoolbox.org). The stimuli were displayed on a 24 Sony
GDM-FW900 driven at a resolution of 1920x1440 at 75Hz by a MP850 video card
(Village Tronic Computer, Sarstedt, Germany).
Stimuli
The stimuli/procedure were modeled after Toet & Levi (1992; Figure 1). The
stimuli consisted of three vertically aligned Ts formed by black pixels (3cd/m2) with
stroke widths of 2 pixels, which were presented against a uniform white background
(70 cd/m2). The Ts were randomly presented either right-side up (T) or upside down.
For each eccentricity tested, the size of the Ts was set to 1.5 times each individual
subjects T-alone discrimination threshold. The T-alone discrimination threshold was
72
derived by averaging the threshold obtained in two blocks prior to the experimental
condition where the T was presented in isolation.
Viewing conditions
The three eccentricities tested, 0, 10 and 25, called for three different
viewing distances (300cm, 90cm, 50cm respectively). For the two peripheral
conditions, the stimuli were always centered on the horizontal meridian to the right of
fixation, and subjects were eye-tracked. The eye-tracking analysis was conducted
offline; thus, trials in which an eye movement occurred could not be removed from
the final analyses. However, the number of eye movements made was quite few (<
4% of trials) and did not differ between groups (F(1,18) = .7, prep < .58) or among
eccentricities (F(1,18) = .98, prep < .65). The order of eccentricity blocks was
counterbalanced across subjects. No effect of run-order was found and will not be
discussed further.
Procedure
Each trial began with a short auditory tone followed by a 150ms ISI. For the
0 stimulus the fixation dot was extinguished during this interval to diminish any
possible forward masking. For the two peripheral conditions the fixation dot
remained visible throughout the trial. The stimuli were then presented for 100 ms.
The subjects task was to indicate the orientation (up/down) of the center T by
pressing the corresponding key on the keyboard. The subject was free to respond at
73
any time following stimulus presentation. Subjects were told that accuracy rather
than speed of response was critical. Following response, auditory feedback was given
(low tone incorrect, high tone correct).
The critical manipulation was the center-to-center spacing between the two
flanking Ts and the target T (initial spacing: 30/400/600 minutes of arc for 0/10/25
conditions respectively), which was controlled by a three up, one down staircase
(step-size of 5%). Rather than ending the staircase procedure after a certain number
of reversals, in order to ensure that subjects were given equal experience (as three
different eccentricities were tested and compared), all subjects underwent 200 trials
per condition. The final crowding threshold value was calculated by averaging the
center-to-center target/distractor spacing across the final 10 trials.
Results and Discussion

The crowding threshold varied from approximately 10 minutes of arc for the
0 condition, to 120 minutes of arc at 10 to more than 300 minutes of arc for the 25
condition. The finding that the crowding region increases dramatically with
eccentricity mirrors previous reports in the field (Leat et al., 1999; Toet & Levi,
1992). To be able to compare the effect of video game experience across
eccentricities, the distance thresholds were first converted to log10 values thus
allowing a relative equalization of the means/variances across conditions (Leat et al.,
1999).
74
The effect of VGP status (VGP/NVGP) and eccentricity (0/10/25) were

analyzed in a 2x3 ANOVA run on log distance thresholds. A main effect of
eccentricity was observed (F(2,36)=331.7, prep > .99, partial eta-squared=.95), with
the size of the crowding region increasing with increasing eccentricity as expected.
Importantly, a main effect of VGP status was also observed (F(1,18) = 22.4, prep >
.99, partial eta-squared=.55), with VGPs demonstrating smaller crowding regions
than NVGPs (Figure 2A). Eccentricity did not interact with VGP status (F(2,36) = .9,
prep < .58) suggesting that the effect of VGP status was consistent across the
eccentricities tested.
Although our primary goal was to ascertain the size of the crowding zones in
VGPs and NVGPs, another interesting pattern of results was observed in the T-alone
condition. Analyses of the T-alone thresholds as a function of VGP status
(VGP/NVGP) and eccentricity (0/10/25) in a 2x3 ANOVA on log10 transformed
values revealed, as expected, a main effect of eccentricity (F(2,36) = 412.6, prep > .99,
partial eta-squared=.96) with the size of the T at threshold increasing with increasing
eccentricity. Unexpectedly though, a main effect of VGP status was also observed
(F(1,18) = 15.9, prep > .99, partial eta-squared=.47) with VGPs being able to
discriminate smaller Ts than NVGPs. VGP status did not interact with eccentricity
(F(2,36) = 1.2, prep < .65) suggesting the effect was consistent across the eccentricities
tested.
Together these results demonstrate that video game players have both smaller
regions of spatial interaction and better visual acuity thresholds. It is worth noting that
75
the performance improvement seen in gamers is consistent across all eccentricities

tested. A difference between groups at 0 is somewhat surprising as it is generally
assumed that foveal vision is near optimal, and therefore more difficult to enhance
through training than peripheral vision (Neville & Bavelier, 2002). Nonetheless, the
present study establishes a central performance enhancement of a comparable
magnitude as that observed in the periphery. In addition, the presence of a group
difference at 25 of eccentricity suggests that some transfer of learning has occurred,
as this eccentricity is not within the well-trained region of space. If this finding were
truly due to experience with action video games, it would be in contrast with much of
the perceptual learning literature, which has demonstrated exquisite spatial specificity
for learning in many training tasks (Fahle, 2004, 2005; Karni & Sagi, 1991).
Before any conclusions can be reached about the effects of action video game
experience, self-selection needs to first be ruled out as a potential factor. It is
possible that by selecting avid game players we have chosen a population that would
have superior visual skills even without the difference in training. It may be, for
example that individuals with better visual skills are more likely to excel at game
playing and therefore more likely to become avid players.
76
Figure 2 - Results
A) VGPs demonstrated smaller crowding regions than NVGPs. However, while the size of the
crowding region increased with increasing eccentricity, the size of the VGP effect was consistent
across eccentricities.
B) A significant decrease in crowding threshold was observed in NVGPs following action video game
training.
C) No reliable change in threshold was seen in the control group following training suggesting that the
effect seen in the action group was not due to test-retest effects or changes in visuo-motor factors.
# - SEM for all data points was less than the size of the squares denoting the values
## - (*: prep > .92, **: prep > .97)
Experiment 2
Experiment 2 employed a training paradigm to investigate whether a causative
relationship exists between action video game experience and higher spatial
resolution. Thirty-two non-gamers were divided into two training groups. One group
was then trained on an action game that was similar to those played by the VGPs in
Experiment 1 (described below), while the other was trained on a game that was less
visually intense, but that did require substantial visuo-motor coordination (also
described below). In addition to controlling for any effects of improved visuo-motor
77
coordination (better hand-eye coordination could conceivably reduce the resources

necessary for response execution, leaving more resources free for target
identification), the control group also acted as a control for test-retest improvements
(improvements due to practice on the tests themselves), and for any Hawthorne-like
effects (any improvements due to the fact that the experimenters had paid attention
to the subjects during the training - Benson, 2001). Crowding threshold were
measured pre- and post-training (30 hours of training over a span of 4-6 weeks). If
action video game experience is sufficient to cause a decrease in the size of the
crowding region, a greater pre-post reduction will be seen in the group trained on the
action game than on the control group.
Method
Participants
Thirty-two NVGPs were equally and randomly divided between the
experimental and the control group. The criteria for NVGP remained the same as in
Experiment 1. All subjects underwent training as described below. In all, 8 females
and 8 males (mean age = 21.3; all right-handed) made up the final experimental
group, while the final control group consisted of 9 females and 7 males (mean age =
21.0, 15 right handed).
78
Testing: Apparatus/Stimuli/Procedure
The apparatus/stimuli/procedure were identical to that described in
Experiment 1 with three exceptions. First, because subjects underwent several
unrelated experiments, the total testing duration was minimized and thus only one Talone block was completed prior to the experimental block. Also, to decrease running
time, the step size of the change in T size was set at 20% until three reversals were
observed and then decreased to 5%, allowing stable thresholds to be reached in 120
trials. Finally, because no effect of run order was found in Experiment 1, the three
eccentricities were run in the fixed order: 0, 10, 25. Eye movements were again
measured for the two peripheral conditions and as in Experiment 1, were quite rare
(approximately 3% of trials) and did not differ as a function of test, game, or
eccentricity nor were there interactions between any of these variables.
Training: Apparatus/Stimuli/Procedure
for a total of 30 hours (maximum of 2 hours per day, minimum of 5 hours per week,
game). This game was chosen to be similar to those played by our VGPs; it has a
relatively simple interface, uses first-person point of view and requires effective
monitoring of the entire visual field (extent from fixation about 13-height x 16width). Each hour session of the action game was divided into three 20-minute
79
blocks. The difficulty of each block was adjusted based upon the kill/death ratio. If
in a block the player scored more than twice as many kills as they had deaths, the
difficulty level was increased one level. Also, players were periodically re-tested on
lower difficulty levels to quantitatively assess improvement.
displayed to cover the entire extent of the screen (18-height x 13-width from
fixation). As such, the field of view of the Tetris game was actually slightly larger
than that of the action game. Tetris was selected to control for the effect of improved
visuo-motor coordination, while putting little demands on the processing of multiple
objects at once. Accordingly, the version of Tetris on which subjects were trained had
the preview block option turned off. In a manner analogous to the action-trained
group, improvement was quantitatively measured by comparing performance on Day
1 versus that on Day 30.
Both groups played their respective games on 20 monitors. The action game
group played on Dell FlatPanel displays, whereas the control group played on CRT
monitors.
Results and Discussion

Game Play
Game improvement was assessed using several measures. For the action
game, the two measures used were kills and death. For each of five levels of game
difficulty (level five being the highest level that all players attained) the measure
80
taken on their first playing of the level (which, because of the way in which difficulty
was increased was not necessarily on the first day of training) was compared with
their final playing of the level on Days 29-30. A substantial increase in number of
kills and decrease in number of deaths was seen at each difficulty level (Level 1:
226% increase in kills, 64% decrease in deaths; Level 2: 147%, 38%; Level 3: 160%,
27%; Level 4: 80%, 33%; Level 5: 52%, 32%).
For the control game, the mean and median scores from Day 1 were compared
with the same values on Day 30. As in the action game, the control players showed
substantial improvements after training, the mean score improving by 323% and the
median score by 359%.
These results demonstrate that both groups were engaged in their training and
Crowding Thresholds
As in Experiment 1 the crowding thresholds were first converted to log10
values and then analyzed in a 2x2x3 ANOVA with trained game (action/control), test
(pre/post), and eccentricity (0/10/25) as factors. A strong main effect of
eccentricity was observed (F(2,60) = 2506.5, prep > .99, partial eta-squared=.98) with
crowding thresholds increasing with increasing eccentricity. A main effect of test
was also observed (F(1,30) = 4.8, prep = .93, partial eta-squared=.14) with crowding
thresholds being lower post-test than pre-test. Importantly, and confirming the results
from Experiment 1, an interaction between trained game and test was observed
81
(F(1,30) = 11.6, prep = .99, partial eta-squared=.28) with the action group showing
larger decreases in crowding threshold from pre-test to post-test than the control game
(Figure 2B and 2C respectively). This effect did not interact with eccentricity
(F(2,60) = .03, prep < .1) suggesting that the effect of training was similar across the
eccentricities tested.
Unlike in Experiment 1, where VGPs showed lower thresholds than NVGPs,
no effect of trained game or test was detected in the T-alone data in Experiment 2.
Only a main effect of eccentricity (F(2,60) = 918.8, prep > .99) was observed, with, as
typical, thresholds increasing with increasing eccentricity.
The results of Experiment 2 establish that subjects trained on an action video
game have significantly greater decreases in crowding threshold than those subjects
trained on the control game. Unlike in Experiment 1, visual acuity (T-alone threshold)
was not detectably modified by 30 hours of training. Although it is unclear at this
point whether more training could produce a reliable change in visual acuity,
Experiment 2 unambiguously demonstrates that visual crowding can be altered
through action video game training.
Discussion and Conclusions

Action video game experience was shown to lead to an increase in the spatial
resolution of vision as measured by crowding. VGPs could tolerate smaller center-tocenter spacing between target and distractors than NVGPs. This improvement was
noted not only at peripheral locations but also in central vision, indicating that even
82
the high central resolution can be enhanced with proper training. In addition,
improvements in the periphery were seen at locations well within the range of playing
as well as at more eccentric locations, indicating transfer of training beyond the
highly trained regions of space. NVGPs specifically trained on an action video game
for 30 hours showed similar improvement in the size of the visual crowding region
establishing a causative relationship between video game experience and a reduction
in the size of the crowding region. These results establish that video game play may
alter basic properties of the visual system.
This study extends previous findings on the impact of video game and visual
skills by showing that video game play can alter visual performance even in a task
where the location and time of arrival of the stimulus are fixed and known ahead of
time to the subject. In contrast, previous work in gamers focused on complex visual
tasks that by design relied on uncertainty, such as in visual search tasks where the
target location is uncertain and has to be found. Although the finding that video game
play changes performance on such complex visual tasks is interesting, a number of
other manipulations are known to also affect performance on such tasks. For example,
much previous research documents enhanced visual spatial resolution when
exogenous cues are used to reduce spatial and temporal uncertainty (Carrasco,
Williams, & Yeshurun, 2002; Yeshurun & Carrasco, 1999), or alternatively poorer
performance when a difficult secondary task is added (Zenger, Braun, & Koch, 2000).
Here VGPs demonstrated reductions in crowding thresholds even though where and
when the stimulus would appear was fixed allowing subjects to focus at their optimal
83
level. The finding of improved performance under such conditions cannot easily be
attributed to strategic factors. Rather, one of the mechanisms by which action video
game play may enhance visual processing is by increasing the spatial resolution of
visual processing across the visual field.
The present study highlights the potential of action video game training for
rehabilitation of visual deficits. Indeed a common feature in visually impaired
patients is greater vulnerability to crowding, such as in the cases of amblyopia or
normal aging (Ball et al., 1988; Ball & Owsley, 1992; Hariharan, Levi, & Klein,
2005; Levi, Hariharan, & Klein, 2002; Levi & Klein, 1985; Polat, Sagi, & Norcia,
1997). Of course, much work remains to characterize the level(s) of processing at
which video game playing may be acting; however, by establishing that a video game
training regimen can reduce the detrimental effects of crowing, this research opens
new avenues for the development of rehabilitation software.
84
Acknowledgements: We thank W. Makous for much helpful advice and J.

Cohen for help with subjects. Supported by NIH grants EYO16880 and DC04418 to
D.B.
85
References
Ball, K., Beard, B., Roenker, D., Miller, R., & Griggs, D. (1988). Age and Visual
Search: Expanding the Useful Field of View. J. Optical Society of America,
A., 5(10), 2210-2219.
Ball, K., & Owsley, C. (1992). The useful field of view test: a new technique for
evaluating age-related declines in visual function. J Am Optom Assoc, 63, 7179.
Benson, P. G. (2001). Hawthorne Effect. In W. E. Craighead & C. B. Nemeroff
(Eds.), The Corsini encyclopedia of psychology and behavioral science (3rd
ed., Vol. 2, pp. 667-668). New York: John Wiley and Sons.
Bouma, H. (1970). Interaction effects in parafoveal letter recognition. Nature, 226,
177-178.
Carrasco, M., Williams, P. E., & Yeshurun, Y. (2002). Covert attention increases
spatial resolution with or without masks: support for signal enhancement.
Chung, S. T., Levi, D. M., & Legge, G. E. (2001). Crowding: a classical spatialfrequency masking effect? Vision Research, 41, 1833-1850.
86
Fahle, M. (2004). Perceptual learning: a case for early selection. Journal of Vision,
4(10), 879-890.
Fahle, M. (2005). Perceptual learning: specificity versus generalization. Current
Opinion in Neurobiology, 15(2), 154-160.
Flom, M. C. (1991). Contour interactions and the crowding effect. Problems in
Optometry, 3, 237-257.
Flom, M. C., Heath, G. G., & Takahashi, E. (1963). Contour interaction and visual
resolution: contralateral effects. Science, 142, 979-980.
Flom, M. C., Weymouth, F. W., & Kahneman, D. (1963). Visual resolution and
contour interaction. Journal of the Optical Society of America, 53, 1026-1032.
Green, C. S., & Bavelier, D. (2006). The cognitive neuroscience of video games. In
L. Humphreys & P. Messaris (Eds.), Digital Media: Transformations in
Human Communication: Peter Lang Pub Inc.
Green, C. S., & Bavelier, D. (In Press). Enumeration versus multiple object tracking:
The case of action video game players. Cognition.
87
Hariharan, S., Levi, D. M., & Klein, S. A. (2005). "Crowding" in normal and
amblyopic vision assessed with gaussian and gabor C's. Vision Research, 45,
617-633.
Hess, R. F., Dakin, S. C., & Kapoor, N. (2000). The foveal 'crowding' effect: physics
or physiology. Vision Research, 40, 365-370.
Intriligator, J., & Cavanagh, P. (2001). The spatial resolution of visual attention.
Cognitive Psychology, 43(3), 171-216.
Jacobs, R. J. (1979). Visual resolution and contour interaction in the fovea and
periphery. Vision Research, 19(1187-1195).
Karni, A., & Sagi, D. (1991). Where practice makes perfect in texture discrimination:
evidence for primary visual cortex plasticity. Proc Natl Acad Sci, 88(11),
4966-4970.
Leat, S. J., Li, W., & Epp, K. (1999). Crowding in central and eccentric vision: the
effects of contour interaction and attention. Investigative Ophthalmology &
Visual Science, 40(2), 504-512.
Levi, D. M., Hariharan, S., & Klein, S. A. (2002). Suppressive and facilitatory spatial
interactions in amblyopic vision. Vision Research, 42, 1379-1394.
Levi, D. M., & Klein, S. A. (1985). Vernier acuity, crowding, and amblyopia. Vision
Research, 40, 1775-1783.
Levi, D. M., Klein, S. A., & Hariharan, S. (2002). Suppressive and facilitatory spatial
interactions in foveal vision: Foveal crowding is simple contrast masking.
Journal of Vision, 2, 140-166.
88
Miller, J. (1991). The flanker compatibility effect as a function of visual angle,

attentional focus, visual transients, and perceptual load: A search for boundary
conditions. Perception and Psychophysics, 49(3), 270-288.
Neville, H. & Bavelier, D. (2002). Human brain plasticity: evidence from sensory
deprivation and altered language experience. Prog Brain Res., 138, 177-188.
Orbach, H. S., & Wilson, H. R. (1999). Factors limiting peripheral pattern
discrimination. Spatial Vision, 12, 83-106.
Poder, E. (2005). Effect of attention on the detection and identification of masked
spatial patterns. Perception, 34, 305-318.
Polat, U., & Sagi, D. (1994). The architecture of perceptual spatial interactions.
Vision Research, 34, 73-78.
Polat, U., Sagi, D., & Norcia, A. M. (1997). Abnormal long-range spatial interactions
in amblyopia. Vision Research, 37, 737-774.
Toet, A., & Levi, D. M. (1992). The two-dimensional shape of spatial interaction
zones in the parafovea. Vision Research, 32(7), 1349-1357.
Trick, L. M., Jaspers-Fayer, F., & Sethi, N. (2005). Multiple-object tracking in
children: The "Catch the Spiers" task. Cognitive Development, 20(3), 373387.
Tripathy, S. P., & Cavanagh, P. (2002). The extent of crowding in peripheral vision
does not scale with target size. Vision Research, 42, 2357-2369.
89
Tripathy, S. P., & Levi, D. M. (1994). Long-range dichoptic interactions in the human
visual cortex in the region corresponding to the blind spot. Vision Research,
34, 1127-1138.
Yeshurun, Y., & Carrasco, M. (1999). Spatial attention improves performance in
spatial resolution tasks. Vision Research, 39, 293-306.
Zenger, B., Braun, J., & Koch, C. (2000). Attentional effects on contrast detection in
the presence of surround masks. Vision Research, 40, 3717-3724.
90
1.1.1.3. The temporal characteristics of vision

Excerpt from:
Green, C.S. & Bavelier, D. Action video game modifies visual selective attention.
Nature 423, 534-537.
Action video game modifies visual selective attention
C. Shawn Green & Daphne Bavelier
Department of Brain and Cognitive Sciences, Center for Visual Science, University
of Rochester, Rochester, New York 14627, USA
91
As video-game playing has become a ubiquitous activity in todays society, it is

worth considering its potential consequences on perceptual and motor skills. It is well
known that exposing an organism to an altered visual environment often results in
modification of the visual system of the organism. The field of perceptual learning
provides many examples of training-induced increases in performance. But perceptual
learning, when it occurs, tends to be specific to the trained task; that is, generalization
to new tasks is rarely found110. Here we show, by contrast, that action-video-game
playing is capable of altering a range of visual skills. Four experiments establish
changes in different aspects of visual attention in habitual video-game players as
compared with non-video-game players. In a fifth experiment, non-players trained on
an action video game show marked improvement from their pre-training abilities,
thereby establishing the role of playing in this effect
We next examined the temporal characteristics of visual attention and asked

whether the pressure to act rapidly on several visual items, which is inherent to most
action games, alters the ability to process items over time, particularly the ability to
avoid bottlenecks of attention that often occur in temporal processing. The
experimental paradigm that we used to test the temporal aspects of visual attention is
the attentional blink task (refs2022 and Fig. 4a). Attentional blink refers to the
phenomena wherein subjects have difficulties reporting a second target when it
appears a few hundreds of milliseconds after the onset of a first target. In our case, we
used a variant of the attentional blink, in which participants were asked to identify a
92
first target and then detect a second target. This task includes two distinct attentional
bottlenecks. First is the attentional blink per se, which is the difficulty of processing a
second target that comes 200500 ms after the onset of the first one. Second is the
cost of switching tasks between the first and second target (from identification to
detection); unlike the attentional blink per se, this effect is most pronounced when the
two targets are temporally adjacent and then decreases slowly as the time between the
two targets increases. This attentional bottleneck is not specific to vision but rather
appears amodal21,23,24. Thus, by using an identification/detection attentional blink task,
we could test whether the enhanced capacities after video-game training not only
applied to a purely visual bottleneck but also generalized to an amodal one.
VGPs outperformed NVGPs on second-target correct detections from lag 1 to
lag 5, indicating less attentional blink (Fig. 4b). The finding of a population
difference as early as lag 1 indicates that video-game training enhances taskswitching abilities as well as decreasing the attentional blink. Thus, both the visual
and amodal bottlenecks identified during temporal processing of visual information
are reduced in VGPs. Clearly, these individuals have an increased ability to process
information over time; however, whether this is due to faster target processing, such
as faster selection and stabilization of information in memory, or to an increased
ability to maintain several attentional windows in parallel, cannot be determined from
our current data.
93
Figure 4
A.
B.
Figure 4. Measure of attention over time.

a. Attentional blink task.
Black letters were rapidly presented at fixation in a standard rapid, serial, visual presentation manner.
At a random time in the stream, a white letter was presented (first target). After this first target,
an X (second target) was presented among the following letters 50% of the time. After
the trial, the subject gave the identity of the first target and then indicated whether the X
was presented. Of interest is the performance of subjects on the X detection task, given
that they have correctly identified the white letter.
b. Attentional blink performance.
At early lags, VGPs performed better (less blink) than NVGPs; as lag increased, the effect of
the attentional bottlenecks decreased and, as expected, the two populations became
comparable (lag by population P , 0.02). Error bars denote s.e.m. Points without error
bars indicate that the s.e.m. was smaller than the size of the square (**P , 0.01).
We also carried out a training experiment to alleviate concerns over the source
of the VGP and NVGP population differences. By selecting VGPs, we may have
94
selected individuals with inherently better attentional skills than NVGPs. The
consequence of this potential split would be that VGPs, with their greater natural
ability, did well at video games and so played often, whereas NVGPs, whose
inherent abilities limited their success, avoided playing games as a result. Another
possible confound was that, because of relatively superior visuo-motor coordination,
VGPs were better able to perform the motor aspect of the tasks and thus had more
spare resources to devote to the visually demanding part of the tasks. To rule out
these potential confounds, a group of NVGPs underwent action-video-game training,
in which they were asked to play Medal of Honor (a game similar to those reported
being played by the VGP population) for 1 h per day for 10 consecutive days. A
control group was trained, over the same time span, on the game Tetris. This game
contains a challenging visuo-motor component but, whereas action games require that
attention is distributed and/or switched around the field, Tetris demands focus on one
object at a time. Tetris, therefore, would not be expected to change the aspects of
visual attention described above and thus affords an excellent control for monitoring
improvement due to enhanced visuo-manual expertise and testretest improvements.
Before and after training we tested each group on the enumeration, useful-field-ofview and attentional-blink experiments. The training was successful as all participants
improved their scores on the video game on which they were trained. Notably, actiongame training led to greater performance improvement than did the control game on
all three experimental tasks. Enumeration increased by 1.7 items in individuals
trained on action games, whereas the control group showed no improvement
95
population by improvement, p < 0.05). Training on action video games enhanced the
useful field of view and lead to faster recovery from the attentional blink (Fig. 5)
Figure 5
Figure 5: Training Attentional Blink

The group trained on an action video game recovered faster from the attentional blink than did the
control group trained on a non-action video game.
Thus, 10 days of training on an action game is sufficient to increase the

capacity of visual attention, its spatial distribution and its temporal resolution. By
forcing players to simultaneously juggle a number of varied tasks (detect new
enemies, track existing enemies and avoid getting hurt, among others), action-videogame playing pushes the limits of three rather different aspects of visual attention. It
leads to detectable effects on new tasks and at untrained locations after only 10 days
of training. Therefore, although video-game playing may seem to be rather mindless,
96
it is capable of radically altering visual attentional processing. There are several ways
by which video-game training could lead to such enhancements. Changes in known
attentional bottlenecks is certainly a possibility; however, speeded perceptual
processes and/or better management of several tasks at the central executive level are
also likely to contribute. It will be for future studies of the effect of video-game
practice to determine the relative contribution of these different factors to skill
learning.
Methods
Participants
Subjects were aged between 18 and 23 years. The VGPs had played action video
games on at least 4 days per week for a minimum of 1 h per day for the previous 6
months. The games included Grand Theft Auto3, Half-Life, Counter-Strike, Crazy
Taxi, Team Fortress Classic, 007, Spider-Man, Halo, Marvel vs Capcom,
Roguespeare and Super Mario Cart. The NVGPs had little, and preferably no, videogame usage in the past 6 months. Experiments 14 included only males; in
experiment 5, both male and female NVGPs underwent training.
For experiment 4attentional blink (Fig. 4)the stimuli and procedure described
previously20 were used except that eight lags instead of nine, and eight trials per
condition instead of ten, were used. VGPs had almost significantly higher-secondtarget detection accuracy in the control condition (detecting the X in the absence of a
first target, 95.6% to 87.9%, p = 0.061). Conservatively, all analyses were done on
97
the difference between the variable of interest and the baseline detection condition to
eliminate different baseline abilities as a potential confound in interpreting the depth
of the attentional blink between populations.
For experiment 5training (Fig. 5)all subjects were tested on the useful-field-of
view, attentional-blink and enumeration tasks (in this order) before training; they then
underwent training for 1 h per day for 10 days with one of two possible kinds of
video game. After training they were retested on the same three tasks (in the same
order). The experimental group trained on Medal of Honor: Allied Assault (Electronic
Arts, 2002). This game simulates Second World War combat situations and was
chosen to be similar to those played by our VGPs. It has a relatively simple interface,
uses first-person point of view and requires effective monitoring of the whole visual
field. Subjects played the game straight through for the first 8 days. On days 9 and 10,
they returned to the beginning of the game to quantitatively measure their
improvement (comparing performance over mission 1 during their first and second
playings). In all available measures of improvement taken (accuracy and deaths to
complete mission), subjects improved their performance after training (accuracy, 17%
improvement; deaths to complete, 42% improvement).
For the control group (n = 8), Tetris (Original Tetris, AcademySoft Elorgwhich,
1987; Tengen, 1988) was displayed to cover the entire extent of the screen. This
game was selected to control for the effect of improved visuo-motor coordination,
while putting few demands on the processing of several objects at once. Accordingly,
the version of Tetris on which subjects were trained had the preview block option
98
turned off. In the two measures of improvement available (high score,maximum level
reached), all subjects improved after training (high score, 71% improvement; mean
level improvement, 67% improvement).
Received 4 December 2002; accepted 8 April 2003; doi:10.1038/nature01647.

References
1. Sagi, D. & Tanne, D. Perceptual learning: learning to see. Curr. Opin. Neurobiol. 4,
195199 (1994).
2. Ball, K. & Sekuler, R. A specific and enduring improvement in visual motion
discrimination. Science 218, 697698 (1982).
3. Fiorentini, A. & Berardi, N. Perceptual learning specific for orientation and spatial
frequency. Nature 287, 4344 (1980).
4. Karni, A. & Sagi, D. Where practice makes perfect in texture discrimination:
evidence for primary visual cortex plasticity. Proc. Natl Acad. Sci. USA 88, 4966
4970 (1991).
5. Poggio, T., Fahle, M. & Edelman, S. Fast perceptual learning in visual hyperacuity.
Science 256, 10181021 (1992).
6. Fahle, M. & Edelman, S. Long-term learning in vernier acuity: effects of stimulus
orientation, range, and of feedback. Vision Res. 33, 397412 (1993).
7. Ramachandran, V. S. & Braddick, O. Orientation-specific learning in stereopsis.
Perception 2, 371376 (1973).
8. Ahissar, M., Laiwand, R., Kozminsky, G. & Hochstein, S. Learning pop-out
99
detection: building representations for conflicting target-distractor relationship.

Vision Res. 38, 30953107 (1998).
9. Ahissar, M. & Hochstein, S. The spread of attention and learning in feature search:
effects of target distribution and task difficulty. Vision Res. 40, 13491364 (2000).
10. Sirenteanu, R. & Regina, R. Perceptual learning in visual search generalizes over
task locations and eyes. Vision Res. 40, 29252949 (2000).
11. Eriksen, B. A. & Eriksen, C. W. Effects of noise letters upon the identification of
a target letter in nonsearch task. Percept. Psychophys. 16, 143149 (1974).
12. Lavie, N. & Cox, S. On the efficiency of visual selective attention: efficient visual
search leads to inefficient distractor rejection. Psychol. Sci. 8, 395398 (1997).
13. Kaufman, E., Lord, M., Reese, T. & Volkman, J. The discrimination of visual
number. Am. J. Psychol. 62, 498525 (1949).
14. Trick, L. M. & Pylyshyn, Z. W. What enumeration studies can show us about
spatial attention: evidence for limited capacity preattentive processing. J. Exp.
Psychol. Human Percept. Perform. 19, 331351 (1993).
15. Trick, L. M. & Pylyshyn, Z. W. Why are small and large numbers enumerated
differently? A limited capacity preattentive stage in vision. Psychol. Rev. 101, 80
102 (1994).
16. Tuholski, S. W., Engle, R. W. & Baylis, G. C. Individual differences in working
memory capacity and enumeration. Mem. Cogn. 29, 484492 (2001).
17. Ball, K. K., Beard, B. L., Roenker, D. L., Miller, R. L. & Griggs, D. S. Age and
visual search: expanding the useful field of view. J. Optic. Soc. Am. 5, 22102219
100
(1988).
18. Myers, R. S., Ball, K. K., Kalina, T. D., Roth, D. L. & Goode, K. T. Relation of
useful field of view and other screening tests to on-road driving performance. Percept.
Mot. Skills 91, 279290 (2000).
19. Ball, K. K., Owsley, C., Sloane,M., Roenker, D. & Bruni, J. Visual attention
problems as a predictor of vehicle crashes in older drivers. Invest. Ophthalmol. 34,
31103123 (1993).
20. Raymond, J. E., Shapiro, K. L. & Arnell, K. M. Temporary suppression of visual
processing in an RSVP task: an attentional blink? J. Exp. Psychol. Human Percept.
Perform. 18, 849860 (1992).
21. Chun, M. M. & Potter, M. C. A two-stage model for multiple target detection in
rapid serial visual presentation. J. Exp. Psychol. Human Percept. Perform. 21, 109
127 (1995).
22. Broadbent, D. & Broadbent,M. Fromdetection to identification: response to
multiple targets in rapid serial visual presentation. Percept. Psychophys. 42, 105113
(1987).
23. Arnell, K. & Jolicoeur, P. The attentional blink across stimulus modalities:
evidence for central processing limitation. J. Exp. Psychol. Human Percept. Perform.
25, 630648 (1999).
24. Potter, M. C., Chun, M. M., Banks, B. S. & Muckenhoupt, M. Two attentional
deficits in serial target search: the visual attentional blink and an amodal task-switch
deficit. J. Exp. Psychol. Human Percept. Perform. 24, 979992 (1998).
101
25. Proksch, J. & Bavelier, D. Changes in the spatial distribution of visual attention
after early deafness. J. Cogn. Neurosci. 14, 15 (2002).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank T. Monacelli, D. McColgin, J. Cohen and K. Schneider

for help with subjects and software support; and R. Aslin, A. Pouget and D. Knill for
feedback on the manuscript. This research was supported by the NIH and the James
S. McDonnell-Pew Foundation.
102
1.1.1.4. The capacity of visual attention

Green, C.S. & Bavelier, D. Enumeration versus multiple object tracking: the case of
action video game players. Cognition 101, 217-245.
Enumeration versus multiple object tracking:

The case of action video game players
Department of Brain and Cognitive Sciences, University of Rochester

(585) 275-0695
[email protected]
103
Abstract
Here we demonstrate that action video game play enhances subjects' ability in
two tasks thought to indicate the number of items that can be apprehended. Using an
enumeration task, in which participants have to determine the number of quickly
flashed squares, accuracy measures showed a near ceiling performance for low
numerosities and a sharp drop in performance once a critical number of squares was
reached. Importantly, this critical number was higher by about two items in video
game players (VGPs) than in non-video game players (NVGPs). A following control
study indicated that this improvement was not due to an enhanced ability to instantly
apprehend the numerosity of the display, a process known as subitizing, but rather
due to an enhancement in the slower more serial process of counting. To confirm that
video game play facilitates the processing of multiple objects at once, we compared
VGPs and NVGPs on the multiple object tracking task (MOT), which requires the
allocation of attention to several items over time. VGPs were able to successfully
track approximately two more items than NVGPs. Furthermore, NVGPs trained on an
action video game established the causal effect of game playing in the enhanced
performance on the two tasks. Together, these studies confirm the view that playing
action video games enhances the number of objects that can be apprehended and
suggest that this enhancement is mediated by changes in visual short-term memory
skills.
Keywords: Video games; Subitizing; Enumeration; Multiple object tracking
104
1. Introduction
1.1. Video Games and Visual Skills
Video game play leads to a number of alterations in visual attention, visuomotor coordination, and other perceptual/cognitive processes (Dorval & Pepin, 1986;
McClurg & Chaille, 1987; Orosy-Fildes & Allan, 1989; Gopher, 1992; Gopher, Weil,
& Bareket, 1994; Greenfield, DeWinstanley, Kilpatrick, & Kaye, 1994;
Subrahmanyam & Greenfield, 1994; Yuji, 1996; Green & Bavelier, 2003; Li &
Atkins, 2004). For example, several researchers have noted that video game play
decreases the reaction time (RT) of subjects asked to respond to visual stimuli
(Orosy-Fildes & Allan, 1989; Yuji, 1996). In the case of visual attention, Greenfield
and colleagues (Greenfield et al., 1994) demonstrated that in a simple target detection
task, expert video game players showed a diminished attentional cost (measured by
RT) when presented with low probability targets compared to that observed in nonplayers, indicating enhancements in the ability to divide attention. Gopher and
colleagues, working in collaboration with the Israeli military, demonstrated that
cadets trained on a video game performed significantly better than their untrained
peers on measures of flight performance (Gopher et al., 1994). Finally, our own work
(Green & Bavelier, 2003) has demonstrated that video game players outperform nonplayers on different aspects of visual attention, in particular in the flexibility and
efficiency with which they distribute attention over space and time. Such increases in
visual attention in video game players may have real-world practical implications.
105
The goal of the present study is to assess the claim that, in addition to its
effect on the spatial and temporal aspects of visual attention, action video game
playing also modifies the number of objects that can be apprehended. In this paper,
we compare the performance of video game players (VGPs) and non-video game
players (NVGPs) on both an enumeration task and the multiple object tracking
(MOT) task. These two particular tasks were chosen because they allow the separate
assessment of the number of items that can be enumerated in parallel (also termed
immediate apprehension), that can be enumerated serially (also termed counting), and
that can be successfully tracked over time. Our results suggest a relationship between
the number of items that can be accurately counted and the number of items that can
be tracked. In contrast, the number of items that can be immediately apprehended
seems to be under the control of distinct constraints. In addition, our results reveal
for the first time a dissociation between reaction time measures and accuracy
measures during the enumeration task. The consequences of these findings for our
understanding of the mechanisms at play in enumeration and MOT studies are
considered in the General Discussion.
1.2. The Enumeration Task

The enumeration task has classically been used to study the number of items
that can be readily attended. This task requires participants to report the number of
briefly flashed items in a display as quickly and as accurately as possible. Participant
performance on this task appears best captured by two distinct processes, easily seen
106
when RT is plotted against the number of items presented. When viewing

enumeration performance in this manner a clear discontinuity, or elbow, is seen,
giving the appearance of a bilinear function. For low numbers of items, usually in the
range of one to four items (Atkinson, Campbell, & Francis, 1976; Oyama, Kikuchi, &
Ichihara, 1981; Trick & Pylyshyn, 1993; Trick & Pylyshyn, 1994), subject
performance is extremely fast. The RT slopes are near zero over this range also
termed the subitizing range. As numerosity increases above this range, each
additional item has a substantially greater cost in terms of RT. The cost to
performance is evident in the steep slope observed beyond about four items. The
change in reaction time slopes is mirrored by a similar change in accuracy, with
accuracy remaining stable and high within the subitizing range and exhibiting a linear
decrease with increasing numerosity beyond the subitizing range. Although reaction
times and accuracy are typically correlated in enumeration studies, the subitizing
range has been defined in terms of reaction time as the range of numerosities that can
be apprehended without a significant increase in reaction time as numerosity is
increased.
The discontinuity in the enumeration curve has been the subject of much
debate and has been posited to have various explanations. One proposal holds that the
discontinuity marks a fundamental difference in the perceptual quality of the display
when few versus many items are presented. Following this line of thought, some have
proposed that performance when few items are present is exceptionally efficient
because it relies on canonical pattern recognition or configurational effects, which are
107
not available when many items are present (Mandler & Shebo, 1982; van Oeffelen &
Vos, 1982). Along the same line, density differences have also been proposed to lead
to the observed discontinuity (Atkinson et al., 1976). A second view holds that the
discrepancy can be accounted for by a single cognitive process, counting, measured at
different points of a continuum (Gallistel & Gelman, 1992). In this model, the elbow
in the curve is thought to reflect a switch from a fast non-verbal mode of counting, to
a slower verbal mode of counting (Gallistel & Gelman, 1992). This view is
supported by the fact that many studies have demonstrated that the RT curve is not
truly flat, even for low numerosity; instead there is often a small slope (between five
and ten times shallower than what is seen for greater numbers of items) (Akin &
Chase, 1978; Oyama et al., 1981). It has also been speculated that increases in
memory load, light for few items but increasingly heavy for many items, may play a
role. An alternative view posits that the discontinuity reflects two fundamentally
different cognitive processes. From this viewpoint, the nearly flat region has been
taken to reflect an automatic, parallel, and immediate process, which has been dubbed
subitizing (Kaufman, Lord, Reese, & Volkman, 1949). For simplicity, throughout
the paper, the region over which performance is immediate will, for ease of
exposition, be referred to as the subitizing range. According to this model, the
mechanism(s) that underlie the subitizing process are hypothesized to be severely
capacity limited (on the order of three to four items). When the number of items to
enumerate exceeds the capacity of this automatic system, subjects must use a separate
process wherein serial attention is employed to count the items from visual short-
108
term memory. As it is well known that information in STM decays over time, the
process is slower and more error-prone. This is clearly reflected in the steep region of
the performance graph, which will be referred to as the counting range throughout the
remainder of the paper.
While there is much dissent as to whether the subitizing and counting range
reflect two separate cognitive processes or a more continuous process with added
constraints from short term memory and indexing as the numerosity increases, all
parties agree that the subitizing range provides an estimate of the number of items
that can be concurrently apprehended. We therefore decided to compare video game
players (VGPs) and non-video game players (NVGPs) on the enumeration task.
2. Experiment 1
In Experiment 1, we asked whether playing action video games would alter
enumeration ability. Experiment 1 also tests the view that video game play primarily
affects peripheral vision. Although our own work (Green & Bavelier, 2003), as well
as that of others (Orosy-Fildes & Allan, 1989), has shown improvements occurring in
both central and peripheral vision, it is still commonly thought that video game play
predominantly modifies peripheral vision. This view partly finds its support in the
fact that many video games require subjects to distribute attention peripherally, as
enemies can appear at any location. To test the effect of video game play on both
the central and peripheral visual field, VGPs and NVGPs underwent two different
enumeration experiments, one with a field of view restricted around fixation (5
109
square; henceforth referred to as the narrow field of view condition) and the other
with a much wider field of view (20 square; the wide field of view condition). If
video game play disproportionately affects far peripheral vision, any VGP advantage
over NVGPs should be magnified in the wide field of view condition. However, if
the effects of video game play are consistent across retinal eccentricities, at least for
the task under study, then performance should be somewhat equivalent between the
two populations at both fields of view.
Using two fields of view also allows us to test the effect of field of view on
the two components of the enumeration curve. The subitizing hypothesis suggests
that the flat region of the performance graph reflects a preattentive spatially parallel
mechanism and therefore predicts a similar subitizing span under both conditions. In
contrast, models with canonical patterns or density as the primary causative
mechanism predict that performance should be affected by this change in sparseness.
In addition, the counting portion of the enumeration task could be adversely affected
by the wider field of view as attending to a larger field of view reduces the amount of
resources available at any particular location in space and increases the probability
that the items will be inaccurately enumerated. The comparison of the two conditions
may therefore help in distinguishing between the various accounts of enumeration
performance. Experiments 1 and 2 were partially reported in Green and Bavelier
(2003).
110
2.1. Method
2.1.1. Subjects
Twenty-six males with normal or corrected-to-normal vision were placed into
one of two groups, VGP or NVGP, based upon their responses to a questionnaire
given prior to the experiment. Only males underwent testing because of the relative
paucity of females with sufficient video game experience.
The criterion to be considered a VGP was a minimum of 3-4 days a week of
action video game usage for the previous six months. Thirteen right-handed males
with a mean age of 19.4 years fell into this category. Ten of the thirteen subjects
reported daily video game usage for at least the previous six months, while the
remaining three reported playing several times a week for the same time span. A
highly abridged list of the games reported as played includes: Grand Theft Auto,
Half-Life, Counter-Strike, Marvel vs. Capcom, Rogue Speare, and Super Mario Kart.
The criterion to be considered a NVGP was little, although preferably no,
video game usage in the past six months. Thirteen males (eleven right- and two lefthanded) with a mean age of 19.3 years fell into this category. Eleven of the thirteen
members reported no video game experience whatsoever in the past year. The
remaining two quantified their video game experience as once per month or less.
Written informed consent was obtained from each subject and each subject
was paid $7.50 for each hour of participation.
111
2.1.2. Stimuli and Procedure

Subjects viewed the display binocularly with their head positioned in a chin
rest at a test distance of 60 cm. Each trial began with the presentation of a small
white fixation cross in the center of a dark screen. After 500 ms, a stimulus
consisting of a random number of white squares was presented for 50 ms (between 1
and 12 squares each subtending .5 x .5 degrees). Subjects were then allowed to
respond. This study did not use the typical vocal response of enumeration studies;
instead subjects were asked to press the number on the keyboard corresponding to the
number of squares they believed were presented. Subjects were instructed to respond
by a keyboard press as quickly as possible while maintaining accuracy.
For the response, the numbers 1-12 were placed on the keyboard above their
respective keys (10 on the 0, 11 on the -, and 12 on the +). Subjects were instructed
to use whatever key pressing strategy was most comfortable. Most used the four
fingers of the left hand on 1-4, the four fingers of the right hand for 5-8 and moved
the right hand for numbers above 8. However, some subjects used different
strategies.
Each subject underwent two experimental blocks of 360 trials each (1-12
items presented, 30 repetitions of each number of items, pseudorandom presentation).
The only difference between the two experimental blocks was the extent of the visual
field over which the white squares could be presented. In the narrow field of view
condition, the white squares were presented in a random location within an invisible
boundary square in the center of the screen measuring 5 x 5. It was also
112
constrained such that there was at least a .5 separation between adjacent squares. In
the wide field of view condition, the squares could be presented over a much wider
field of view (20 x 20). It should be noted that because the squares were kept at the
same size in the two field of view conditions, the square density was greater in the
narrow than in the wide field of view condition.
The order of the two conditions was counterbalanced. No effect of order was
found (p > .7) and thus will not be discussed further.
2.2. Results
Four measures of performance will be discussed for each experiment
accuracy breakpoint, percent correct, average response and RT.
Enumeration studies rely typically on reaction time analyses. In our case,
however, the use of a keyboard response, rather than a vocal response, makes the
interpretation of RT difficult. Although subjects were trained in advance in typing the
key that corresponded to a given number, it is likely that this method of response
nevertheless introduced additional variability in the measurement of RTs that may not
be consistent across the two groups. Video game players are known to have faster key
press RTs (Orosy-Fildes & Allan, 1989) and because there were 12 possible response
keys to choose from, issues such as working memory (the ability of a subject to
remember which key each finger was resting upon) and strategy (where the fingers
were centered, use of both hands versus only one hand, etc.) could have played a role.
With these caveats in mind, the RT data will be presented last.
113
The accuracy breakpoint refers to the point at which the discontinuity in

accuracy occurs (the point where the elbow forms in the function). As previously
discussed, accuracy performance during enumeration paradigms appears well
approximated by a bilinear model. To determine an individual subjects accuracy
breakpoint, their percent correct data was fit to a bilinear model using the least
squares method. As our goal was to gain a quantitative measure of the breakpoint, as
well as a measure of the slope of the two linear components that intersect at the
breakpoint, only the data points for 1-8 items were entered into the model. This range
was chosen based on previous evidence that the breakpoint lies approximately in the
middle of this span, as well as to minimize the contribution of processes such as
guessing, estimation biases, and strategy that may come into play for greater numbers
of items. Each subjects curve was modeled as an intersection between two linear
components; the first component was constrained to have a slope very near zero
(maximum slope of 3% per item) while the second component was modeled as
linearly increasing (as the data is plotted in terms of error rate, not percent correct)
with the slope allowed full room to vary in order to best fit the data. The output of
the model was therefore the slope of the two lines as well as the accuracy breakpoint
the point at which the two lines intersected. Although the accuracy breakpoint has
classically been well mirrored by the subitizing span, or the shallow near zero slope
of the reaction time data curve, the use of the term subitizing will be reserved for RT
analyses in this paper, as Experiment 3 will reveal for the first time a dissociation
between accuracy and RT measures of breakpoints.
114
The dependent variable percent correct is self-explanatory.

The last dependent variable is average response, as previous models have
hypothesized a prominent role of estimation ability in enumeration performance
(Krueger, 1982). Average response refers to the average response the subject made
when presented with a given number of squares. Perfect performance corresponds to
a line starting at the origin with a slope of one. Overestimation leads to a deviation
above the perfect line, whereas underestimation leads to a digression below the
perfect line.
Accuracy breakpoint was analyzed in a 2(VGP status: VGP/NVGP) x 2(field
of view: narrow/wide) ANOVA while percent correct, RT, and average response
were each initially analyzed in a 2(VGP status: VGP/NVGP) x 2(field of view:
narrow/wide) x 12 (number of squares) ANOVA.
2.2.1. Overall Analyses
2.2.1.1. Accuracy Breakpoint

The analysis of accuracy breakpoint revealed a clear main effect of VGP
status with VGPs switching about 2 items beyond the accuracy breakpoint of NVGPs,
VGP: 5.0 +/- .2 squares, NVGP: 3.0 +/- .3 squares, F(1,24) = 27.7, p < .001 (Figure
1A). No interactions with field of view were observed. Separate analysis of the slope
of each component yielded no main effect of VGP status, in particular, the lack of
effect for the second slope indicates that both groups fall-off at a similar rate beyond
115
their inflection point in both field of view conditions. While there was no effect of
field of view in the analysis of the slope of the first component, a main effect of field
of view was observed in the analysis of the second component, with the wide field of
view having a significantly greater slope than the narrow field of view, wide: 13.7+/0.7%, narrow: 11.1 +/- 0.7%, F(1,24) = 7.3, p = .013. Since the accuracy breakpoint
was equivalent in the two field of view conditions, it appears that the ability to
apprehend low numerosity is unperturbed by the size and density of the display,
whereas the greater slope of the second component in the wide field of view condition
suggests that the ability to count higher numbers of squares falls off more quickly in
the wide field of view condition.
2.2.1.2. Percent Correct

The expected strong main effect of number of squares was observed, with
performance decreasing with increasing numerosity, F(11,264) = 202.8, p < .001.
Importantly, a main effect of VGP status was observed with VGPs outperforming
NVGPs, VGP: 75.5%+/-1.4% correct, NVGP: 62.4%+/-1.8% correct, F(1,24) = 9.1, p
= .006 (Figure 1A). Furthermore, the predicted VGP status x number of squares
interaction was also found with both groups performing similarly well for low
numbers (1-3), but VGPs outperforming NVGPs as the number of squares exceeded
3, F(11,264) = 3.3, p < .001.
Also, performance was better in the narrow field of view condition than in the
wide field of view condition, narrow: 71.5+/-1.5% correct, wide: 66.4 +/- 1.8%
116
correct, F(1,24) = 9.1, p = .006. This main effect can largely be attributed to
differences in performance for the very large numbers (10-12) between the two fields
of view which was further reflected in the observed interaction of field of view and
number of squares, F(11,264) = 2.5, p = .005.
Finally, a VGP status x field of view x number of squares interaction,
F(11,264) = 2.02, p = .03 will be further explored in analyses separated by field of
view, but appeared to be rooted in disproportionately poor performance by NVGPs on
even relatively low numbers of squares during the wide field of view condition.
2.2.1.3 Average Response

The expected main effect of number of squares, with average response
increasing with increasing numerosity was observed, F(11,264) = 1398.3, p < .001.
VGP status and number of squares interacted with both groups having similar average
responses for low number of items, but NVGPs beginning to consistently
underestimate the number of items before VGPs did, F(11,264) = 7.1, p < .001
(Figure 1B). Field of view also interacted with number of squares, as subjects began
to consistently underestimate the number of items earlier in the wide field of view
(around 8 items) than in the narrow field of view (around 10 items) F(11,264) = 5.3, p
< .001. Finally, a VGP status x field of view x number of squares interaction
prompted analyses separated by field of view discussed subsequently, F(11,264) =
2.4, p = .007.
117
2.2.1.4 RT
Despite the previously mentioned potential pitfalls with the RT data, the
statistics were in harmony with the percent correct data. A main effect of number of
squares was observed with subjects taking longer for greater numbers of squares,
F(11,264) = 101.8, p < .001. The main effect of VGP status on RT was nonsignificant (p > .05). However, a VGP status x number of squares interaction was
observed, rooted in the fact that VGPs were faster than NVGPs for low number of
items (1-2) but then became much slower for greater number of items, F(11,264) =
3.3, p < .001 (Figure 1C).
A main effect of field of view was also observed with subjects taking longer
in the narrow field of view condition than the wide field of view condition, narrow:
1.91+/-.05 s, wide: 1.7 +/- .05 s, F(1,24) = 14.7, p = .001. The effect of field of view
further interacted with number of squares, as RTs were relatively equivalent for low
numbers of squares, but much longer for larger numbers of squares in the narrow
field of view condition, F(11,264) = 6.2, p < .001. Finally, a VGP status x field of
view x number of squares interaction F(11,264) = 1.9, p < .05, was caused by VGPs
taking a disproportionately long time to respond in the narrow field of view condition
when many squares were presented.
Because of the observed interactions outlined above, the two field of view
conditions were further analyzed separately by field of view along the three
components revealing field of view by video game status interactions (percent
correct, average response, RT).
118
Figure 1 - Results
A. Enumeration accuracy - % correct and breakpoint: VGPs clearly outperform NVGPs on
the enumeration task. The VGP breakpoint (the elbow in the regression line) is approximately
2 items beyond that seen in the NVGPs. Furthermore, the VGP advantage continues to hold
even for high numerosity.
B. Average response: NVGPs begin to underestimate the true number of items (bar above zero)
about 2 items before VGPs. The relative inaccuracy of the NVGPs compared to VGPs can
also be seen for all numbers of squares as the VGP bar does not begin to deviate from zero
(on average perfectly correct) until around 10 items.
C. RTs: NVGPs and VGPs have very similar RTs for low numerosity, VGPs being slightly
faster, but as the number of items to enumerate increases, the VGP RT becomes much greater
than NVGPs. This is unlikely to represent a speed/accuracy trade-off in the conventional
sense, in that longer RTs allow for more information to decay from visual memory. Instead,
these results may indicate a more stable visual memory representation in the VGP population.
Note: (in all figures: error bars denote SEM, * = p < .05, ** = p < .001)
2.2.2 Narrow Field of View Analyses

Main effects of number of squares and VGP status were again seen, number of
squares: F(11,264) = 148.7, p < .001; VGP status: F(1,24) = 11.9, p = .002, with
119
accuracy decreasing with increasing numerosity and VGPs again outperforming

NVGPs by a large margin, VGP: 78.2+/-1.9% correct, NVGP: 64.8+/-2.3% correct.
A VGP status x number of squares interaction, F(11,264) = 2.6, p = .004, outlined
identical effects as seen in the main ANOVA and is captured by the difference in
accuracy breakpoint, with both groups performing similarly well for low numbers of
items (1-3 items p > .07) but VGPs significantly (p < .05) outperforming NVGPs for
all subsequent numbers of items except for 10 squares (p = .055).
2.2.2.2. Average Response

A main effect of number of squares was again observed, F(11, 264) = 1358.2,
p < .001. A VGP status x number interaction closely followed the main ANOVA
with both groups giving very similar average responses for low numbers of items, but
with VGPs making better estimates for higher numbers, F(11,264) = 4.2, p < .001.
2.2.2.3. RT
A main effect of number of squares was observed F(11,264) = 86.9, p < .001.
A VGP status x number interaction with VGPs responding faster for small numbers
of squares, but becoming slower than NVGPs when greater numbers of squares were
presented, F(11, 264) = 2.8, p < .001.
120
2.2.3 Wide Field of View Analyses

A main effect of number of squares was observed with performance as always
decreasing with increasing numbers of squares, F(11,264) = 135.3, p < .001. A main
effect of VGP status was again observed, VGP: 72.9+/-2.1% correct, NVGP:60.0+/2.7% correct, F(1,24) = 5.9, p = .02. And a VGP status x number of squares
interaction again highlights the differences seen in accuracy breakpoint with the VGP
advantage in accuracy only becoming evident for items above 3, F(11,264) = 3.3, p <
.001.

Main effects of number of squares, F(11,264) = 1069.4, p < .001 and a VGP
status x number of squares, F(11,264) = 8.3, p < .001 were observed.
2.2.3.3. RT
A main effect of number of squares was observed, F(11,264) = 76.8, p < .001.
Also, as in the narrow field of view analysis, a VGP status x number of squares
interaction, F(11,264) = 3.4, p < .001, reflected the fact that VGPs were faster than
NVGPs for low numbers, but slower for higher numbers.
121
2.3. Discussion
The main finding of Experiment 1 is that VGPs enumerate more accurately
than NVGPs. VGPs are able to enumerate with high accuracy about two items more
than NVGPs, as exemplified by the reliable difference in accuracy breakpoint
between the two populations. Beyond the accuracy breakpoint, the estimate of the
number of objects presented remained more accurate in VGPs than for NVGPs. Both
groups underestimated the true value at high numerosities, an expected result since
subjects were limited in their response to up to 12 items (although see Mandler and
Shebo, 1982, who also report systematic underestimation of numerosity beginning at
set size of 9 when the maximum number of items was 20). However, the results
reveal an additional pattern of underestimation that differs between the two groups.
At high numerosities (9 and above), NVGPs systematically underestimated the
number of squares presented to a greater extent than VGPs, and were consistently
faster than VGPs. While this result takes the form of a simple speed/accuracy tradeoff, there are other possible mechanistic explanations for it, such as faster loss of
information in memory in NVGPs than in VGPs. If the ability to count accurately
from short term memory decays faster in NVGPs than in VGPs, NVGPs responses
would show greater underestimation of numerosity as well as faster RTs as the
number of items they could enumerate is smaller. This point will be discussed further
in the General Discussion.
Experiment 1 also reveals a systematic effect of the size of the field of view
on performance at high numerosity. The wide field of view condition led to more
122
errors, systematic underestimation and faster RTs than the narrow field of view
condition suggesting an important role for the spread of attention in the enumeration
of high numerosities. Importantly, display size did not affect performance at low
numerosity, as exemplified by the comparable accuracy breakpoints found in the two
field of view conditions. The different sensitivity of low and high numerosity displays
to the field of view condition is further documented by the interactions between field
of view and number of squares observed in percent correct, average response, and RT
analyses. This finding lends support to the view that two distinct sets of constraints
are underlying performance for low and high numerosity displays in the enumeration
task. In this paradigm, field of view co-varies with density, with the wide field of
view condition corresponding to lower density displays than the narrow field of view
condition. The data pattern, however, runs counter to that predicted by models that
suppose that canonical patterns or density may be at the source of the low/high
numerosities split. Indeed, such models predict better performance for low density
than high density displays, a prediction which runs contrary to the finding of greater
error in the wide condition and equal accuracy breakpoint across conditions.
Accuracy breakpoints and RTs for low numerosities were similar for VGPs
and NVGPs across fields of view, indicating that the advantage conferred by video
game playing applies equally to narrow and wide displays. Thus, the effects of game
playing are not narrow to peripheral locations, but are also visible more centrally.
This is consistent with the fact that video game play, in addition to necessitating
constant peripheral monitoring, also commonly requires the subject to actively attend
123
to the center of the visual field, which normally contains the primary object of
interest.
Experiment 1 demonstrates that expert video game players outperform nonplayers in the enumeration task by having both an extended accuracy breakpoint and
greater accuracy in the counting range. It is unclear, however, whether the very act of
playing is a causative factor in the improved enumeration performance of VGPs or
whether selecting for video game players inherently biased us in selecting individuals
with better visual skills. After all, individuals with good visual abilities probably have
an advantage when it comes to playing video games, and thus may be more prone to
become video game players than individuals with poor visual skills. This issue is
addressed in Experiment 2.
3. Experiment 2
Results from Experiment 1 indicate enhanced enumeration performance in

VGPs. While our hypothesis predicts that extensive video game playing leads to this
enhanced skill, it could also be the case that VGPs have inherently better perceptual
skills and/or were somehow genetically endowed with greater attentional abilities.
Another explanation is that what is learned during video game play is not necessarily
perceptual in nature, but is instead a perceptual-motor skill. Although the use of
percent correct as our primary dependent measure in Experiment 1 was chosen to
minimize the effect of visuo-motor facilitation in our measures, it is possible that by
124
alleviating the demands of the motor response, video game playing allows VGPs to
have more left-over resources available to process the sensory stimulus.
To control for these two possibilities, NVGPs underwent video game training
in Experiment 2. Some underwent video game training using an action video game,
whereas others played a game that placed heavy demands on visuo-motor
coordination but did not tap aspects of visual attention of interest in Experiment 1. If
the differences observed in Experiment 1 are due to the demands of action video
game playing and not due to better visuo-motor control or genetically endowed traits,
a notable improvement in the enumeration task should be observed following training
in the action game trainees, but not in the control game trainees.
Finally, while Experiment 1 employed only males as subjects, by using an
equal number of males and females in Experiment 2, we can test whether the effects
of video game play are similar across gender.
3.1. Method
3.1.1. Subjects
The study initially enrolled 20 NVGPs that were equally and randomly
divided between the experimental and the control group (5 males/5 females in each
group). The criteria for NVGP remained the same as in Experiment 1. All subjects
underwent training as described below. One male from the experimental group and
one male and one female from the control group did not finish training. Thus 5
females and 4 males (mean age = 20.4, 8 right-handed) made up the final
125
experimental group, while the final control group consisted of 4 females and 4 males
(mean age = 19.7, all right handed).
3.1.2. Pre-Test
As differences between the two fields of view were minimal with respect to
the effect of video game play, and because of two additional experimental tasks, not
relevant to this paper, subjects completed only the narrow field of view condition.
3.1.3.Apparatus
Testing:
The apparatus was identical to that described in Experiment 1.
Training:
The control group played on the same experimental setup (computer,
monitor, refresh, and resolution) the experiment itself was conducted on. The action
group played on one of two Dell computers each equipped with 20 flat-panel LCD
monitors.
3.1.4. Training Stimuli and Procedure

for one hour per day for ten out of fifteen days. The nine members of the
experimental group played the game Medal of Honor: Allied Assault (henceforth
referred to as the action video game). This game was chosen to be similar to those
126
played by our VGPs from Experiment 1. It has a relatively simple interface, uses
first-person point of view and requires effective monitoring of the entire monitor
display (extent from fixation about 13-height x 16-width). Subjects played the game
straight through for the first eight days, beginning each day at the point where they
had finished the previous day. On days nine and ten they returned to the beginning of
the game in order to quantitatively measure their improvement by comparing
performance over mission 1 during their first (days 1-2) and last (days 9-10) playing.
The eight members of the control group played the game Tetris, which was displayed
to cover the entire extent of the screen. However, because Tetris adds graphics on the
side of the screen, the effective game area extended 13-height x 9-width from
fixation. This game was selected to control for the effect of improved visuo-motor
coordination, while placing little demand on the simultaneous processing of multiple
items. Accordingly, the version of Tetris on which subjects were trained had the
preview block option turned off. Furthermore, this group serves as a control for any
possible effects due to familiarity with the task (test-retest). In a manner analogous to
the action-trained group, improvement was quantitatively measured by comparing
performance on day 1 versus that on day 10.
3.1.5. Post-Test
in the pre-test, as well as the other two aforementioned unrelated tasks (Green &
Bavelier, 2003).
127
3.2. Results
3.2.1. Game Play

In order to assess game improvement, two different measures were taken for
each group during the first and last playing. Improvement measures were determined
by computing the difference between post-training performance and pre-training
performance divided by pre-training performance.
For the action game, the two measures were shooting accuracy (number of
targets hit/total number of shots fired) and the number of deaths before completing
the first mission. In both measures, subjects showed improved performance
following training. For shooting accuracy a 68% improvement was seen. In terms of
the number of deaths to complete mission one, a 42% improvement was observed.
For the control game, the two measures were high score (highest score
achieved in one level) and highest level reached. Again, all subjects improved
following training (high score = 71% improvement, highest level reached = 67%
improvement).
These results establish that both groups were engaged in their training and
128
3.2.2 Enumeration Results

Similar analyses as in Experiment 1 were carried out. In the case of percent
correct, average response, and RT each of the factors was initially analyzed in a
2(trained game: action/control) x 2 (test: pre/post) x 12 (number of squares) ANOVA
and in the case of accuracy breakpoint 2 (trained game: action/control) x 2 (test:
pre/post) ANOVA. Gender was not included as a factor, as a preliminary analysis
indicated there was no main effect of gender, nor were there any interactions with
gender in any of the effects observed.
3.2.2.1. Accuracy Breakpoint

A main effect of test was observed indicating a slight increase in accuracy
breakpoint at post-test, pre:3.0+/-.15 squares, post:3.8+/-.31 squares, F(1,15) = 6.4, p
= .02. Importantly for our hypothesis, a trained game x test interaction was also
observed, caused by a greater post-test improvement in the action game than in the
control game, F(1,15) = 6.1, p = .03. In analyses separated by trained game, a main
effect of test was observed in only the action group, F(1,8) = 10.6, p = .01, not in the
control group (p > .9), indicating that training had a significant effect on accuracy
breakpoint in the action game alone. No effect was observed on the slope of either
component (all ps > .2) (Figures 2A and 2B).
129
Figure 2
A. Enumeration accuracy - % correct and breakpoint Action game: The action group
demonstrates a clear increase both in breakpoint (by around 1.5 squares) and in overall
accuracy.
B. Enumeration accuracy - % correct and breakpoint Control game: The control group
showed no sign of improvement after training, thus ruling out test-retest improvements or
improvements in visuo-motor control as possible hypotheses to explain the improvement in
the action group.

A main effect of number of squares was observed as expected, F(11,165) =
111.6, p < .001. A main effect of test was also observed with subjects being more
accurate post-test than pre-test, pre: 66.9+/-2.1% correct, post: 72.2+/-1.9% correct,
F(1,15) = 23.0, p < .001. Importantly, an interaction was observed between trained
game and test caused by the action group showing greater improvement than the
130
control group, F(1,15) = 15.4, p < .001. Furthermore, significant interactions between
test and number of squares, F(11,165) = 2.1, p = .02, and between trained game, test
and number of squares, F(11,165) = 2.4, p = .008 reveal a similar pattern as in
Experiment 1 (Figures 2A and 2B). Following training, the action group was as
accurate as the control group for low numbers of squares, but outperformed the
control group on higher numbers.

A main effect of number of squares was observed, F(11,165) = 5473.0, p <
.001, as well as the expected interaction between test and number of squares,
F(11,165) = 4.2, p < .001, indicating more accurate average responses following
training. Importantly, a trained game x test x number of squares, F(11,165) = 2.4, p =
.008 revealed that the bulk of the changes in accuracy were in the action group. As
was seen with the VGPs in Experiment 1, the trained group became better estimators
of large numbers of items following training.
3.2.2.4. RT
The only effect was a main effect of number of squares, F(11,165) = 48.4, p <
.001. This is unsurprising however because of the way we required subjects to
respond.
131
3.3. Discussion
Experiment 2 establishes that even relatively little action video game play (10
hours) is sufficient to alter enumeration performance. In fact, similar changes to
those described in Experiment 1, albeit of lesser amplitude, were observed in action
game trainees for the main aspects of the task. These include a shift of accuracy
breakpoint as well as greater accuracy and better estimation for numerosities above
the accuracy breakpoint. This conclusively demonstrates that action video game play
is at the source of the improved performance on the enumeration task. Furthermore,
the fact that no effect of gender was observed, indicates that the consequences of
video game play are not sex dependent.
Taken together, Experiments 1 and 2 indicate that action video game play
induces two main changes in performance on the enumeration task. First, measures of
accuracy breakpoint show that gamers switch from the shallow to the steep
component of the enumeration accuracy curve beyond the point where control
subjects switch. Second, beyond the accuracy breakpoint, measures of percent correct
and average response indicate greater accuracy in gamers than their controls.
When considering the underlying mechanism for these changes, it is important
to recognize that the estimation of the subitizing range in enumeration studies has
typically relied on RT measures rather than accuracy. In fact, the number of items that
can be apprehended in a fast and parallel manner classically defines the subitizing
range. Although accuracy measures have typically resulted in a similar pattern of
132
results to RT measures in previous studies, we cannot rule out the possibility that the
greater accuracy breakpoint in gamers may be due to more accurate counting rather
than an increase in the number of objects that can be immediately apprehended.
Whether gaming does modify the ability to apprehend more objects at once remains
an open question that will be addressed in Experiment 3.
4. Experiment 3
A few caveats in the interpretation of Experiment 1 were addressed in

Experiment 3. First, subjects gave vocal responses thus allowing precise
measurement of RT. Second, afterimages may have played a role in the effects of
gaming reported in Experiments 1 and 2, as the stimuli display was not masked.
Although Simon and Vaishnavi (1996) reported no substantial increase in the
subitizing range when the items were presented as afterimages (thus allowing up to
60 seconds of viewing), because their primary dependent measure was accuracy
rather than reaction time, caution dictated that in Experiment 3, a backwards pattern
mask be employed, specifically designed to eliminate afterimages as a viable source
of information that could be used unequally by the two groups.
133
4.1. Method
4.1.1. Subjects
Twenty-two males (none of which had participated in Experiment 1 or 2) with
normal or corrected vision were again placed into one of two groups, VGP or NVGP,
based upon their responses to a questionnaire, slightly modified from that used in
Experiment 1. The questionnaire and criteria to be considered a VGP were altered
slightly to allow a more accurate measure of the amount of time each subject spent
playing specific types of video games. The questionnaire asked the number of hours
(0, 0-1, 1-2, 3-5, 5-10, 10+) spent playing each of several types of video games
(action, sports, fantasy, role playing, other) per week. To be considered a video game
player, a subject needed a minimum of 5 hours a week of action video game usage for
the previous six months. Eleven right-handed males with a mean age of 19.1 years
fell into this category.
The criterion to be considered a non-video game player was zero hours per
week of action games. Eleven right-handed males with a mean age of 20.3 years fell
into this category.
134

The stimulus/procedure was identical to that described in the narrow field of
view portion of Experiment 1 except for four changes. First, to rule out any role of
afterimages, a backwards pattern mask was presented for 500 ms following the
presentation of the squares. The mask consisted of a black and white checkerboard
pattern of the same contrast/luminance as the stimulus, designed to eliminate
afterimages as a potential source of information. The second change was that the
presentation time was increased to 100 ms a change necessitated by the addition of
the mask. Third, the task was shortened by including only 1-10 squares. Finally, the
most important change was the method of response. The voice onset time was used to
measure RT. The trials were recorded and scored for accuracy offline. If the subject
hesitated or changed their response (threnofour), the trial was omitted from RT
and accuracy results (this circumstance occurred on less than 1% of all trials).
4.2. Results
The results were analyzed using the same four dependent measures (accuracy
breakpoint, percent correct, average response, RT) as well as a new component RT
breakpoint, which was computed in the same manner as the accuracy breakpoint.
4.2.1. Accuracy Breakpoint

There was a main effect of VGP status on the breakpoint for accuracy with
VGPs again having a 2.5 item advantage over NVGPs, VGP: 5.0+/-.3 squares,
135
NVGP: 2.5+/-.32 squares, F(1,20) = 33.2, p < .001 (Figure 3A). Again, there was no
significant difference in the slope of either component of the model (ps > .08). This
result replicates those of Experiments 1 and 2.
136
Figure 3 - Results
A. Enumeration accuracy - % correct and breakpoint: As was seen in Experiment 1, VGPs
show a clear advantage both in breakpoint as well as overall accuracy.
B. RT: Also as was seen in Experiment 1, VGPs and NVGPs have extremely similar RTs for
low numbers of items, but VGPs begin to take significantly longer than NVGPs as the number
of items to be enumerated increases. The RT breakpoint, which offers an index of the number
of items that can be instantaneously apprehended, is similar in both groups.
4.2.2. Percent Correct

A main effect of increased errors with increasing number of squares was
observed, F(9,180) = 119.1, p < .001. More importantly, there was a main effect of
VGP status, VGP: 77.4+/-2.5% correct, NVGP: 60.5+/-3.2% correct, F(1,20) = 27.8,
p < .001 and a VGP status x number of squares interaction, F(9,180) = 5..0, p < .001
(Figure 3A). As seen in Experiment 1, this is due to the two groups being roughly
137
equivalent at small numbers of squares (p > .05 for 2 and 3 squares), but diverging for
larger numbers with VGPs retaining high accuracy for even large numbers of squares
(VGPs more accurate than NVGPs, p < .05 for 4-9 squares).
4.2.3. Average Response

There was only a main effect of squares, F(9,180) = 1792.6, p < .001. There
was no main effect of VGP status or any interactions with VGP status, indicating that
on average the two groups performed quite similarly. This difference with the
previous experiments is not surprising as Experiment 3 included up to 10 squares
while the significant difference in average response were most marked at very high
numerosity (11-12 squares) in the two previous experiments.
4.2.4. RT
An ANOVA on the simple RT scores revealed a main effect of number of
squares, F(9,180) = 155.5, p < .001 and a VGP status x number of squares interaction,
F(9,180) = 12.0, p < .001 with the VGPs again starting slightly (although not
significantly) faster for low numbers of squares but becoming significantly slower (p
< .05) than NVGPs for high numbers of squares (8-10) (Figure 3B).
4.2.5. RT Breakpoint
In contrast to the accuracy breakpoint, no main effect of VGP status was
found for RT breakpoint, VGPs: 3.4 +/- .1 items, NVGPs: 3.0+/-.2 items, F(1,20) =
138
2.8, p = .1. Although the trend appears in the same direction as the accuracy
breakpoint, the size in terms of number of squares is certainly much smaller than has
been observed in accuracy (Figure 3B). No effect of VGP status was seen in the
slope of either component (ps > .5).
4.3. Discussion
The main finding of Experiment 3 is that there is a difference in accuracy
breakpoint between gamers and non-gamers despite similar RT breakpoints. Clearly
the accuracy breakpoint of the VGP population does not reflect the true number of
items that can be immediately apprehended. Instead, it is apparent that despite near
perfect accuracy for up to five items in the VGP group, the RTs for four and five
squares are slower than that for one to three items. This finding provides the first
large dissociation of RT and accuracy on an enumeration task. It has generally been
taken for granted that when subjects begin to count, they begin to lose information
(Sperling, 1960). The more items that are presented, the longer subjects need to
count, and the less accurate they become. However, VGPs are apparently counting
for four and five dots (as demonstrated by the RT data), while their accuracy
continues to be nearly perfect.
This enhancement in counting ability is also seen in the large differences in
accuracy and RT between VGPs and NVGPs for greater numbers of items. These
results suggest that the VGPs continue to successfully read-off the items from their
visual memory after the NVGPs have made their best attempt. The correlation
139
between enumeration performance and working memory ability has been previously
demonstrated in the literature (Tuholski, Engle, & Baylis, 2001). Subjects that
performed poorly on a working memory task also performed poorly on the counting
portion of the enumeration paradigm while no differences were observed in the
subitizing span. It therefore seems most likely that what is truly being altered by
VGP play is at the level of counting, rather than at the level that mediates immediate
apprehension of numerosity.
While it is clear that the capacity of the immediate apprehension mechanism
(or in other words the subitizing range defined by the RT breakpoint) in the
enumeration task is similar in VGPs and NVGPs, what is left open is whether VGPs
can simultaneously track more items than NVGPs. While some researchers have
suggested that there may be a connection between the number of items that can be
automatically enumerated and the number of items that can be tracked, it is not
necessary that they be rooted in the same mechanism. To more directly address this
question we made use of the multiple object tracking (MOT) paradigm (Pylyshyn,
1989) which requires subjects to dynamically allocate attention to multiple objects
and sustain that attention for several seconds.
5. Experiment 4
In our version of the MOT paradigm, subjects view a number of

randomly moving circles. At the beginning of the trial, some subset of the circles is
140
cued. The cues then disappear and subjects are required to keep track of the circles
that were cued (now visually indistinguishable from uncued circles) as they continue
to move randomly about the screen. After several seconds of tracking, one of the
circles is highlighted and the subject must make a yes (was cued)/ no (was not cued)
decision. This method of response, rather than the more typical method of asking the
subject to indicate each of the initially cued objects, was employed to minimize the
role of working memory in the response process, and in doing so gain a cleaner
measure of the number of items that can be successfully tracked.
While previous theories have suggested a preattentive link between subitizing
and MOT performance (Pylyshyn, 1989), it is generally accepted that there is a large
dynamic attentional component to the MOT task as well (Scholl, Pylyshyn, &
Feldman, 2001). The task requires active allocation of visual attention in order to
successfully track targets embedded in a field of competing, and visually identical,
distracting elements. Several studies have demonstrated that attention is actually split
between the items during tracking (Sears & Pylyshyn, 2000). Furthermore,
neuroimaging has revealed activation in what are thought of as attentional areas
parietal and frontal regions when subjects perform a MOT task (Culham et al.,
1998; Culham, Cavanagh, & Kanwisher, 2001).
While the subitizing span offers a glimpse at the number of items that can be
immediately apprehended, the MOT paradigm offers a good measure of the number
of items than can be simultaneously tracked, and some have suggested,
simultaneously attended, over a period of time. We therefore decided to use the MOT
141
task to clarify whether VGPs can actually track more items at once as well as
examine what, if any, relationship exists between multiple object tracking and
enumeration performance.
5.1. Method
5.1.1. Participants
Twenty males (none of whom had participated in previous
experiments reported in this paper) with normal or corrected vision were again placed
into one of two groups, VGP or NVGP, in a manner identical to that used in
Experiment 3. Ten right-handed males with a mean age of 19.4 years were placed
into the VGP category, while ten right-handed males with a mean age of 20.6 years
were placed into the NVGP category. None of the participants was red/green
colorblind.

Each observer viewed the display binocularly with his head positioned in a
chin rest at a test distance of 57 cm. Subjects were instructed to fixate within a center
ring (radius = .25 deg). Subjects pressed a key to begin each trial. Each trial began
with 16 circles (radius .5 deg) moving randomly on a circular gray background
(radius of circular background = 10 deg) at a rate of 5 deg/sec. The circles repelled
142
one another before contact (.5 deg minimum separation), and were repelled by the
outer edges of the background and by the center fixation circle. At the start of the
trial, some number (1-7) of these circles were cued (colored red) while the rest were
green. During this time the subject was instructed to attend to the red circles, as they
would shortly change to green after which time the subject had to track the circles
that were previously cued. After 2 seconds the cued circles changed to green, leaving
all 16 circles visually indistinguishable. The subject had to track the cued circles for
5 seconds after which one of the 16 circles turned white (probe). The subject had to
press either a yes or no key in response to whether the probe circle was one of the
originally cued circles. The probe circle was one of the originally cued circles 50%
of the time. Each number of cued circles (1-7) was presented 20 times (10 yes, 10
no) for a total of 140 trials. In many other instantiations of this paradigm subjects
were not eye-tracked (Pylyshyn, 2004), or eye-movements were found to have few
implications (Pylyshyn & Storm, 1988). However, because of the use of more objects
than most MOT paradigms (up to 7) and the possibility that the two groups could
differ in eye movement strategies, subjects were eye-tracked and trials where they
made an eye-movement greater than 1 from center were omitted from later analyses.
This happened fairly rarely (around 6% of trials) and did not differ between groups
nor did the occurrence of eye-movements appear to affect accuracy or affect accuracy
differently between the two groups (p > .4).
143
5.2. Results
The results were analyzed in a 2(VGP status: VGP/NVGP) x 7 (number of
circles to track) ANOVA as earlier analyses had indicated there was no effect (p > .7)
of whether the answer was yes or no (i.e. no response bias). As expected, a main
effect of number of circles to track was found with accuracy decreasing with
increasing number of circles, F(6,108) = 60.6, p < .001. Importantly, a main effect of
VGP status, VGP:84.3+/-1.6% correct, NVGP:78.2+/-1.8% correct, F(1,18) = 9.2, p =
.007 was observed indicating better performance in VGPs (Figure 4). As VGP status
and number of circles to track did not interact, we can assume the VGP advantage
was relatively equal across number of circles. Although there was no interaction of
VGP status with the number of circles to track, individual analyses separated by
number of circles to track were performed for comparison with following analyses
and indicated a significant advantage (p < .05) for VGPs only for three to five circles
to track and a marginally significant advantage at 6 circles to track (p = .07). At the
ends of the spectrum (one circle to track or seven circles to track) the two groups
were equivalent (ps > .1).
144
Figure 4. Multiple object tracking performance

VGPs demonstrate a substantial increase in the accuracy with which multiple items can be tracked
compared to NVGPs. The effect is most pronounced for 3-5 items to track.
5.3. Discussion
The results indicate that VGPs outperform NVGPs when it comes to
tracking several objects over time. Unlike for enumeration accuracy and RT, there
was no significant interaction between VGP status and number of items on MOT
accuracy. However, the results of the two paradigms do appear qualitatively similar,
with VGPs and NVGPs having comparable performance with relatively few items,
the differences only emerging after some critical threshold in load is exceeded. The
data is furthermore quite consistent with the recent findings of Trick and colleagues
(Trick, Jaspers-Fayer, & Sethi, 2005) who reported that children (6-19 years old) who
played action video games performed significantly better than non-action game
playing children on a version of the multiple object tracking task. Before drawing
145
further conclusions, Experiment 5 investigates causation through a controlled training

study.
6. Experiment 5
As in Experiment 2, it is critical to show a causative effect of video game play

on MOT performance.
In this experiment, a larger sample of NVGPs than in
Experiment 2 was trained for 30 hours, three times as long as the training in
Experiment 2. The choice of training time was determined by other tasks not reported
in this paper.
6.1. Method
6.1.1. Subjects
The study enrolled 32 NVGPs that were equally and randomly divided
between the experimental and the control group. The criteria for NVGP remained the
same as in Experiment 4. All subjects underwent training as described below. In all
8, females and 8 males (mean age = 21.3, all right-handed) made up the final
experimental group, while the final control group consisted of 9 females and 7 males
(mean age = 21.0, 15 right handed).
146
6.1.2. Apparatus
Testing:
The apparatus was identical to that described in Experiment 4.
Training:
Both groups played their respective games on 20 monitors. The
action game group played on Dell FlatPanel displays, whereas the control group
played on CRT monitors.
6.1.3. Training Stimuli and Procedure

for a total of 30 hours (maximum of 2 hours per day, minimum of 5 hours per week,
game), a different action game than previously used. This game was chosen to be
similar to those played by our VGPs; it has a relatively simple interface, uses firstperson point of view and requires effective monitoring of the entire visual field
(extent from fixation about 13-height x 16-width). One source of confounding in
the previous game used was the fact that players could learn the development of the
story and develop efficient wait and ambush strategies. Unreal Tournament 2004
was chosen because there is no script. Instead, the game is controlled by the action
of 32 AI agents rather than linear story development. Each hour session of the action
game was divided into three 20-minute blocks. The difficulty of each block was
147
adjusted based upon the kill/death ratio. If in a block the player scored more than
twice as many kills than they had deaths, the difficulty level was increased one level.
Also, players were periodically re-tested on lower difficulty levels to quantitatively
assess improvement.
displayed to cover the entire extent of the screen. As such, the field of view of the
Tetris game was actually slightly larger than that of the action game (which was the
same as in Experiment 2 - 13x16). The effective control game area extended 18height x 13-width from fixation. This game was selected to control for the effect of
improved visuo-motor coordination, while putting little demands on the processing of
multiple objects at once. Accordingly, the version of Tetris on which subjects were
trained had the preview block option turned off. In a manner analogous to the actiontrained group, improvement was quantitatively measured by comparing performance
on Day 1 versus that on Day 30.
6.1.4. Post-Test
in the pre-test, as well as the other aforementioned unrelated tasks.
148
6.2. Results
6.2.1. Game Play
In order to assess game improvement, several measures (slightly different than
those collected in Experiment 2) were used. However, as in Experiment 2, a percent
change score was calculated for each of the measures.
For the action game, the two measures used were kills and deaths. For each of
five levels of game difficulty (level five being the highest level that all players
attained) the measure taken on their first playing of the level (which because of the
way in which difficulty was progressed was not necessarily on the first day of
training) was compared with their final playing of the level on Days 29-30. A
substantial increase in number of kills, and decrease in number of deaths was seen at
each difficulty level (Table 1).
For the control game, the average and median scores from Day 1 were
compared with the same values on Day 30. As in the action game, the control players
showed substantial improvements after training, the mean score improving by 323%
and the median score by 359%.
As in Experiment 2, these results demonstrate that both groups were engaged
in their training and showed improvement on the training task.
149
Table 1.
% Change
Difficulty
1
2
3
4
5
Kills
226.3
147.6
160.3
79.8
52.0
Deaths
-64.1
-38.0
-27.4
-32.5
-32.3
6.2.2. MOT Performance

As in Experiment 2, no effect of gender was found in preliminary analyses,
and as such, a 2(trained game: action/control) x 2(test: pre/post) x 7 (number of
circles to track) ANOVA was run on accuracy. Again, only the trials where subjects
did not break fixation were used (as in Experiment 4, eye-movements occurred in
only approximately 5% of trials). Also, as previously observed, there was no effect of
trained game or test on the number of eye movements (p > .6) and again eyemovements did not appear to affect accuracy.
The expected main effect of number of circles was found, F(6,180) = 155.6, p
< .001 with accuracy decreasing with increasing numbers of circles to track. Also a
main effect of test was found. pre: 75.3+/-1.2% correct, post:78.1+/-1.1% correct,
F(1,30) = 5.1, p = .03. Most importantly for our hypothesis, however, an interaction
between trained game and test was found, F(1,30) = 13.4, p = .001, indicating an
unequal effect of training with the action group improving approximately 7.5%
whereas the control group remained stable (Figures 5A and 5B).
150
Separating the groups, and running a 2(test:pre/post) x 7(number of circles)

ANOVA revealed that only the action group showed a main effect of test, F(1,15) =
15.5, p = .001 as well as a test x number of circles interaction, F(6,90) = 2.3, p = .04.
No effect of test or any interactions with test were found in the control group.
Figure 5
A. Multiple object tracking performance Action game: Those subjects trained for 30 hours
on an action video game show a marked improvement in their MOT performance as
compared to their pre-test scores.
B. Multiple object tracking performance Control game: Performance was identical before
and after training in the control group. These results rule out test-retest as a source of
confounding as well as increases in visuo-motor coordination.
151
6.3. Discussion
Experiment 5 demonstrates that relatively little video game play leads to
substantial differences between groups and further demonstrates that the effects
observed in Experiment 4 were not due to an inherent population bias.
7. General Discussion
The five experiments presented demonstrate that action video game play
increases the number items that can be enumerated and tracked simultaneously over
time. In Experiment 1, habitual action video game players display enhanced
enumeration accuracy as compared to non-players. Experiment 2 establishes a causal
role for action video game play, as NVGPs specifically trained on an action video
game show similar enhancements. Experiment 3 demonstrates for the first time a
dissociation between accuracy and reaction time measures of the subitizing range and
establishes that action game playing does not modify the number of items that can be
immediately apprehended, but rather enhances accurate counting. By making use of
the multiple object tracking paradigm, Experiment 4 demonstrates an effect of VGP
status on the ability to simultaneously track multiple objects over an extended period
of time. The significant improvement in MOT performance seen in NVGPs after
action game training in Experiment 5 demonstrates that action video game playing
has a causal role in the measured effects.
Taken together, these five experiments suggest that action video game play
may enhance some aspects of visual working memory. Several lines of evidence
152
point to this conclusion. First, VGPs demonstrate enhanced enumeration accuracy

even at very high numerosities. Second, this enhanced accuracy is accompanied by
an increase in RTs. Although this pattern would be the expected speed/accuracy
trade-off in a system in which evidence accumulates over time but does not decay, the
behavior under study relies on visual short-term memory in which representations are
known to decrease in fidelity over time (Lee & Harris, 1996; Nilsson & Nelson,
1981; Sperling, 1960; Vogels & Orban, 1986). Delaying responses in studies of
short-term memory does not lead to increased accuracy, but rather decreased accuracy
as the memory representations have more time to fade. Thus, an alternative
explanation seems warranted in which video game experience leads to enhancements
in some aspect(s) of visual short-term memory. At least two alternatives are possible,
one based on the durability of the memory trace and another on the speed of cycling
through the memory trace. In the first case, action video game experience may lead to
a more durable visual memory trace. This view would be consistent with the
accuracy and reaction time data as well as the average response data where NVGPs
begin to underestimate the number of squares well before the VGPs. One may
speculate that after a certain period of time, NVGPs begin to drop items from visual
memory, and at this point they simply make their best guess (from viewing Figures
1C and 3B, one can surmise that the NVGPs RTs appear to plateau at around eight
items). Conversely, if it were the case that VGPs possess a more durable memory
trace, they would be able to continue counting beyond the point where the NVGPs
have stopped, which would account for both the greater accuracy and longer RTs. In
153
addition, this process may also sustain better tracking ability in the MOT by allowing
more durable indexing of the dynamics of the objects to be tracked. However, a
change in the fidelity of working memory representations in gamers is only one
possible explanation for the observed results. A possible alternative hypothesizes that
items in working memory are not necessarily kept simultaneously active, but instead
one or a few items are constantly refreshed by a visit from a single focus of attention
that moves from item to item in a cyclical fashion. As the speed of cycling through
the items increases, the number of items that could be successfully maintained in
short term memory would correspondingly increase. It is therefore possible to
capture the present findings by assuming that the speed of cycling through memory
traces is faster in VGPs than NVGPs, thus accounting for both the better counting and
multiple object tracking performance. It should be further noted that factors unrelated
to visual short-term memory, such as estimation ability and response bias, may also
be at work in the enumeration paradigm, particularly for high numerosities. For
instance, VGPs may be better able to judge when the most number of squares were
presented, without necessarily being able to explicitly count each item. Also, as
previously mentioned, because the maximum response was capped at some maximum
value, a bias toward underestimation for the larger numerosities is created that may
not be exactly equal in the two populations. A role for these differences in
estimation/bias cannot be ruled out in interpreting some of the current results,
especially at high numerosities, but they remain an unlikely explanation at lower
154
numerosities where the accuracy breakpoint is seen to shift between NVGPs and
VGPs.
Beyond the effect of action video game play, these findings also lend strong
support to models of enumeration performance that propose relatively distinct
constraints for the two components of the enumeration performance curve. The
dissociation between the accuracy and RT breakpoints in gamers is probably the most
robust indication to-date of separate mechanisms, one that is sensitive to gaming
(counting) and one that is not (subitizing). Similarly, the comparison of two different
fields of view in Experiment 1 indicates that the mechanism(s) behind subitizing are
less malleable that those behind counting. Indeed, in Experiment 1, performance over
the subitizing range was quite similar across visual field conditions. Only in the
counting range did performance differ with more accurate performance for the
smaller field of view and denser displays. Models that suppose fundamental
differences in the characteristics of the display (density, patterns, etc) between low
and high numerosity stimuli cannot readily account for the overall pattern of results
reported here, be it the effect of gaming or that of visual field size. Models of
enumeration studies which posit two separate mechanisms - a fast and parallel one for
subitizing and a more serial process for counting more naturally capture the main
findings. Under this view, the mechanism underlying subitizing would show little to
no sensitivity to gaming or visual field size, and be highly specific to the enumeration
of low numerosities. In contrast, the mechanism underlying counting would be much
155
more plastic, showing enhancement with gaming and be facilitated by the use of a
small visual field.
Although some have suggested a link between immediate apprehension in the
enumeration task and performance on the MOT paradigm (for instance, that they may
both utilize preattentive mechanisms, or FINSTs - Pylyshyn, 1989; Trick & Pylyshyn,
1994), our results suggest that the subitizing range does not index the same process as
the MOT. VGPs demonstrate no enhancement in subitizing range as measured by RT,
but do demonstrate an enhancement in MOT ability. Also, while there is virtually no
cost in terms of speed or accuracy moving from one to three items in the enumeration
paradigm, a clear decrease in accuracy is observed with each additional item in the
MOT paradigm (even moving from one to two items). Thus, it appears that the
number of items that can be immediately apprehended as measured by RT measures
in enumeration studies is not necessarily a good predictor of the number of items than
can be simultaneously tracked. Although our data do not allow us to draw strong
conclusions, our findings suggest that the number of items that can be accurately
counted may be a better correlate of tracking capabilities, as both of these measures
are found to improve with gaming.
This study establishes that when it comes to the number of objects that can be
attended, a distinction should be drawn between a fast, parallel behavior that displays
little plasticity and a more serial behavior that displays a range of plastic behaviors.
As such these studies make several contributions, both to our understanding of the
processes indexed by the enumeration and MOT paradigms, as well as to our
156
understanding of the nature of the changes that occur as a result of action video game
play. It will be, however, for future experiments to fully characterize the
consequences of these results for models of attention and working memory.
157
Acknowledgements: We thank J. Cohen, N. Fernandez, D. McColgin, and K.

Schneider for help with subjects, data analysis, and software support, M. Dye for
comments on earlier drafts, and S. Dehaene for discussion. This research was
supported by NIH grants EY O16880 and DC 04418 to D.B.
158
References
Akin, O., & Chase, W. (1978). Quantification of three-dimensional structures.
Journal of Experimental Psychology: Human Perception and Performance, 4,
397-410.
Atkinson, J., Campbell, F., & Francis, M. (1976). The magic number 4+/-0: A new
look at visual numerosity judgments. Perception, 5, 327-334.
Culham, J. C., Brandt, S., Cavanagh, P., Kanwisher, N. G., Dale, A. M., & Tootell, R.
B. (1998). Cortical fMRI activation produced by attentive tracking of moving
targets. Journal of Neurophysiology, 80, 2657-2670.
Culham, J. C., Cavanagh, P., & Kanwisher, N. G. (2001). Attention response
functions: characterizing brain areas using fMRI activation during parametric
variations of attentional load. Neuron, 32(4), 737-745.
Dorval, M., & Pepin, M. (1986). Effect of playing a video game on a measure of
spatial visualization. Perceptual Motor Skills, 62, 159-162.
Gallistel, C. R., & Gelman, R. (1992). Preverbal and verbal counting and
computation. Cognition, 44, 43-74.
Gopher, D. (1992). The skill of attentional control: acquisition and execution of
attention strategies. In D. E. Meyer & S. Kornblum (Eds.), Attention and
Performance XIV (pp. 299-322). Cambridge, Massachusetts: The MIT Press.
159
Kaufman, E., Lord, M., Reese, T., & Volkman, J. (1949). The discrimination of
visual number. American Journal of Psychology, 62, 498-525.
Krueger, L. E. (1982). Single judgements of numerosity. Perception and
Psychophysics, 31, 175-182.
Lee, B., & Harris, J. (1996). Contrast transfer characteristics of visual short-term
memory. Vision Research, 36, 2159-2166.
Li, X., & Atkins, M. S. (2004). Early childhood computer experience and cognitive
and motor development. Pediatrics, 113(6), 1715-1722.
Mandler, G., & Shebo, B. J. (1982). Subitizing:An analysis of its component
processes. Journal of Experimental Psychology: General, 111(1), 1-22.
McClurg, P. A., & Chaille, C. (1987). Computer games: Environments for developing
spatial cognition. Journal of Educational Computing Research, 3(1), 95-111.
Nilsson, T. H., & Nelson, T. M. (1981). Delayed monochromatic hue matches
indicate characteristics of visual memory. Journal of Experimental
Psychology: Human Perception and Performance, 7, 141-150.
160
Orosy-Fildes, C., & Allan, R. W. (1989). Psychology of computer use: XII.

Videogame play: Human RT to visual stimuli. Perceptual and Motor Skills,
69, 243-247.
Oyama, T., Kikuchi, T., & Ichihara, S. (1981). Span of attention, backward masking
and RT. Perception and Psychophysics, 29, 106-112.
Pylyshyn, Z. W. (1989). The role of location indexes in spatial perception: A sketch
of the FINST spatial-index model. Cognition, 32, 65-97.
Pylyshyn, Z. W. (2004). Some puzzling findings in multiple object tracking (MOT):
I. Tracking without keeping track of object identities. Visual Cognition, 11(7),
801-822.
Pylyshyn, Z. W., & Storm, R. W. (1988). Tracking multiple independent targets:
Evidence for a parallel tracking mechanism. Spatial Vision, 3(3), 179-197.
Scholl, B. J., Pylyshyn, Z. W., & Feldman, J. (2001). What is a visual object?
Evidence from target merging in multiple object tracking. Cognition, 80, 159177.
Sears, C. R., & Pylyshyn, Z. W. (2000). Multiple object tracking and attentional
processing. Canadian Journal of Experimental Psychology, 54, 1-14.
Simon, T. J., & Vaishnavi, S. (1996). Subitizing and counting depend on different
attentional mechanisms:Evidence from visual enumeration in afterimages.
Perception and Psychophysics, 58(6), 915-926.
Sperling, G. (1960). The information available in brief visual presentations.
Psychological Monographs, 74, 1-29.
161
Subrahmanyam, K., & Greenfield, P. M. (1994). Effect of video game practice on

spatial skills in girls and boys. Journal of Applied Developmental Psychology,
15, 13-32.
Trick, L.M., Jaspers-Fayer, F., & Sethi, N. (2005). Multiple-object tracking in
children: The Catch the Spies task. Cognitive Development, 20, 373-387.
Trick, L. M., & Pylyshyn, Z. W. (1993). What enumeration studies can show us about
spatial attention: Evidence for limited capacity preattentive processing.
Journal of Experimental Psychology: Human Perception and Performance,
19(2), 331-351.
Trick, L. M., & Pylyshyn, Z. W. (1994). Why are small and large numbers
enumerated differently? A limited-capacity preattentive stage in vision.
Psychological Review, 101(1), 80-102.
Tuholski, S. W., Engle, R. W., & Baylis, G. C. (2001). Individual differences in
working memory capacity and enumeration. Memory and Cognition, 29(3),
484-492.
van Oeffelen, M. P., & Vos, P. G. (1982). Configurational effects on the enumeration
of dots: Counting by groups. Memory and Cognition, 10(4), 396-404.
Vogels, R., & Orban, G. A. (1986). Decision processes in visual discrimination of
line orientation. Journal of Experimental Psychology: Human Perception and
Performance, 12, 115-132.
Yuji, H. (1996). Computer games and information processing skills. Perceptual and
Motor Skills, 83, 643-647.
162
1.1.1.5. Increased accuracy observed in VGPs: Conclusion

In each of the tasks above, VGPs were seen to exhibit superior accuracy
compared to NVGPs, with each individual result suggesting an enhancement in some
independent aspect of visual attention. In the UFOV, superior localization accuracy
at all tested eccentricities, with and without distractors, suggests an enhancement of
the spatial distribution of visual attention. In the crowding experiment, smaller
center-to-center distances could be tolerated between the target and the distractors in
the VGP group, suggesting an increase in the resolution of visual attention. In the
AB, VGPs more accurately detected the presence of the second target, particularly at
short T1-T2 lags, suggesting an enhancement in the temporal dynamics of visual
attention. Finally, in the MOT, VGPs more accurately tracked numerous moving
objects, suggesting an increase in the capacity of visual attention.
Each of these paradigms was designed to test what are thought of as relatively
independent aspects of visual attention (for instance, one would expect that you could
theoretically alter capacity without affecting resolution). However, the question is
nevertheless begged, rather than positing a separate independent mechanism
underlying each of the observed differences, could there instead be a single common
underlying mechanism that can account for all of the results? However, before
considering what form such a mechanism may take, it is worth considering the second
complementary branch in the literature on video game effects, reaction time studies,
as such a mechanism should account for changes in performance between VGPs and
NVGPs regardless of the dependent measure that is used to assess said performance.
163
1.1.2. Decreased reaction time observed in VGPs

While the work from our lab has focused almost exclusively on tasks where
accuracy is the primary dependent measure, one could argue that the most consistent
finding in the gaming literature is that VGPs have substantially faster RTs than
NVGPs. This basic result has been reported and repeatedly replicated for nearly 20
years using a wide variety of paradigms. In fact, it is interesting to note that besides
RT being the primary dependent measure (and as will be described, VGPs
demonstrating faster RTs than NVGPs), there are few obvious links between the
experiments. Perhaps even more so than in the accuracy literature, each of these RT
tasks was designed to test a very different aspect of perceptual or cognitive
performance and as such, the results have been interpreted as enhancements in a
range of different processes in VGPs.
Greenfield and colleagues (Greenfield, DeWinstanley, Kilpatrick, & Kaye,
1994) were among the first researchers to test the effects of video game experience on
cognitive performance. As many games present multiple objects that must be
attended to, or switched between, Greenfield and colleagues tested the ability to
divide attention between two locations and/or to reorient attention to a relevant
location after it has been improperly drawn to another location. Subjects were told
they should press a button as soon as they saw a briefly flashed target stimulus that
could only appear at one of two locations, A or B. They were warned that on 80% of
the trials the target would appear at Location A while the target would only appear
10% of the trials at Location B (the remaining 10% were catch trials). The logic of
164
this type of experiment is that by manipulating the probability of occurrence at each

location, subjects will become biased to allocate more attention to the high
probability location and less attention to the low probability location. Accordingly,
subjects are generally faster to respond to high probability targets and much slower to
respond to low probability targets, a fact that is taken to reflect the difference in
attentional allocation.
In an initial experiment, Greenfield and colleagues pitted VGPs against
NVGPs on this divided attention task (along with a control condition in which the
stimulus appeared at each location equally often). First, replicating previous findings
in the video game literature, VGPs were found to have an overall faster RT than
NVGPs. Also, as is consistently found in the attentional literature, they found that
NVGPs responded faster to stimuli presented at the 80% location and slower to
stimuli presented at the 10% location compared to their reaction times when the
probabilities were equal (a condition that theoretically represents an even division of
attention). Interestingly, while the VGPs showed a benefit (decreased reaction time)
at the 80% location, their reaction time for the 10% location was the same as what
was observed during the control condition. One item of particular note is that the
VGPs responded at least as fast, if not faster, in the 10% condition (the hardest
condition) as did NVGPs in the 80% condition (the easiest condition). Importantly,
no difference in accuracy was observed, which may have indicated a simple speedaccuracy trade-off. This finding indicated to the authors that the efficiency with
which attention is divided is greatly increased in VGPs.
165
Castel and colleagues (Castel, Pratt, & Drummond, 2005) used the widely
known Posner exogenous cueing paradigm to examine the effects of video game
experience on the temporal dynamics of attentional orienting. The task was a
standard inhibition-of-return type task in which two boxes were presented, one to the
left and one to the right of fixation. A target would appear in either of the two boxes
and the subjects task was to simply press the spacebar as quickly as possible once the
target appeared. Performance was manipulated by the presentation of an exogenous
(flashed) cue prior to the targets appearance. The cue could be either compatible (on
the same side as the target) or incompatible (on the opposite side as the target), and
several stimulus-onset asynchronies (SOA - time between the onset of the cue and the
onset of the target) were tested, as performance on this task is known to be a function
of both cue-target compatibility and SOA. Subjects are typically faster to respond to
compatible cues than incompatible cues when the SOA is short (< 100 ms), while the
opposite is true when the SOA is long (> 100 ms). This finding is thought to capture
a fundamental principle of human orienting if attention is allocated to a location,
but no target appears, attention is disengaged from that location and is inhibited from
returning to that same location for several hundred milliseconds. The authors found
that both VGPs and NVGPs displayed this typical pattern of results. However, VGPs
were found to respond far faster than the NVGPs without a corresponding increase in
error rate. This effect was of such a size that the slowest condition for the VGPs
(compatible cue, 600 ms SOA) was faster than the quickest condition for the NVGPs
(incompatible cue, 400 ms SOA). However, the lack of a 3-way interaction between
166
VGP experience, SOA, and compatibility, led the authors to conclude that although
VGPs respond much faster, the basic orienting mechanisms in VGPs and NVGPs are
similar.
Castel and colleagues (Castel et al., 2005) also examined the rate of visual
search in VGPs and NVGPs. In this task, some number of letters (4-26) was
presented, with one of the letters always being a b or a d. The subjects task was
to find the b or d and respond accordingly, with the primary dependent measure
being reaction time. Two conditions were run that differed in the composition of the
remaining (non-target/distracting) letters. One condition was what is known as an
easy search (also known as parallel/feature/efficient search). In this condition, each
of the non-target letters was the letter k. Performance on this task is known to be
very fast, accurate, and relatively independent of set-size (number of distracting
letters). The second condition was what is known as a hard search (also known as
serial/inefficient search). In this condition, each of the non-target letters was a
random member of the alphabet (other than b or d, non-repeating). Performance
on this task is known to be a function of the set size with speed (and to some extent
accuracy) decreasing with increasing set sizes. As predicted, both VGPs and NVGPs
performed according to previous results in both the easy and hard searches.
However, as predicted, the VGPs responded far faster than the NVGPs without a
decrease in accuracy. In the hard search, VGP RT for a set-size of 26 was
considerably below that of NVGPs for a set-size of only 18. While a significant
interaction between set-size and VGP status suggested an increase in the efficiency of
167
search in the VGPs, the authors argued that their results might also be understood as a
difference in stimulus-response mapping.
Bialystok (Bialystok, 2006) examined the effects of video game experience on
the Simon task. The Simon effect is a well-known phenomena wherein subjects are
faster to make responses when the stimulus appears on the same side as the response
key and slower when the stimulus appears on the opposite side as the response key
(for instance, pressing the left key to respond to a stimulus on the left side of the
screen is faster than pressing the left key to respond to a stimulus on the right side of
the screen). In one version of the task, colored squares were presented either to the
left or right of fixation, with the two different colors corresponding to left/right key
presses. In another version, arrows pointing left/right appeared on the left or right of
fixation, with the direction of the arrow indicating the correct key press. As in the
previous literature, VGPs were seen to respond faster than NVGPs in all conditions
(compatible, incompatible, and control) with equivalent error rates. The fact that the
RT advantage was reasonably consistent regardless of display type led the author to
conclude that video game experience leads to a general increase in the speed of
processing, rather than any specific enhancement in executive function (as might have
been suggested if the VGP advantage was particularly large for incompatible trials).
Work from our lab has addressed the common belief that video game players
may show more impulsivity than NVGPs. To assess this issue, the Test of Variables
of Attention (T.O.V.A.; (Leark, Wallace, & Fitzgerald, 2004)) was administered to
VGPs and NVGPs. Briefly, this test requires subjects to monitor a display and make a
168
timed response to a stimulus if it appears at one location, while withholding a

response to the same stimulus if it appears at another location (a go/no-go task). In
different blocks of trials, the target (go) can appear either frequently or infrequently.
The T.O.V.A. therefore offers a measure of both impulsivity (Is the subject able to
withhold a response to a non-target when most of the stimuli are targets?) and a
measure of sustained attention (Is the subject able to stay on task and respond quickly
to a target when most of the stimuli are non-targets?). VGPs responded more quickly
than did NVGPs on both task components, with equivalent accuracy. The fact that
the advantage was equivalent regardless of block (mostly go trials or mostly no-go
trials) suggests that VGPs are faster but not more impulsive than NVGPs and equally
capable of sustaining their attention.
We have also examined the efficiency of what some authors have called the
three basic attentional systems alerting, orienting, and executive (Fan, McCandliss,
Sommer, Raz, & Posner, 2002). We administered the Attentional Network Test
(ANT) to action game players and non-playing controls aged between 7 and 22 years.
The subjects task was to indicate the direction (left/right) either a fish (childrens
version) or an arrow (adult version) was facing with the corresponding arrow key.
The target could appear above or below fixation, either alone (neutral) or flanked by
distractors (which may point the same way as the target or the opposite direction),
and several types of cues that indicate the position or stimulus onset time could
precede the target. By comparing performance in these various conditions, one can
examine the effect of knowing the exact onset time of the target versus having no
169
temporal information (alerting), knowing both the onset time and position of the
target versus knowing only the onset time (orienting), and the influence of
incompatible distractors (executive). Again as predicted, VGPs responded
significantly faster than NVGPs, while making an equivalent number of errors.
However, the lack of any relevant interactions with cue or distractor type suggested
that VGPs were similar in terms of the performance of these basic attentional
systems.
Clark and colleagues examined the possibility that video game training could
reverse the age-related decline in performance on speeded tasks (Clark, Lanphear, &
Riddick, 1987). Seven subjects with a mean age of 65 years underwent a 7-week
video game training treatment. Prior to and after video game training the subjects (as
well as a group of control seniors) underwent a reaction time experiment, testing the
effect of stimulus-response compatibility. The seniors were seated in front of two
lights, under which were two buttons. In one block of trials the seniors were to press
the button below the light that went on as fast as possible (compatible condition). In
the second block, they were to press the button beneath the light that did not go on
(incompatible condition). Normal individuals are generally much quicker when the
response is compatible with the cue than when the response is incompatible. The fact
that this cost is particularly magnified in the elderly is taken to reflect an age-related
decline in response selection (mapping a stimulus onto an appropriate action).
During 7 weeks of video game training, the seniors in the experimental group
practiced one of two games (Pac Man and Donkey Kong) for at least 2 hours per
170
week. The effect of this video game training on reaction time was indeed significant
as the experimental groups average reaction time dropped approximately 25
milliseconds in the compatible condition and an even more impressive 80 ms in the
incompatible condition, while the control group (which received no video game
experience) did not improve. The authors argue that because the largest improvement
was in the incompatible condition, the video game experience led to enhancements
particularly in response selection in addition to the widely reported decreases in
simple reaction time.
Many other authors have reported speeded responses in VGPs or as a result of
video game experience in a wide variety of tasks from simple RT (Orosy-Fildes &
Allan, 1989; Yuji, 1996), to tasks used to train jet pilots (Gopher, Weil, & Bareket,
1994) and laparoscopic surgeons (Rosser Jr., Lynch, Cuddihy, Gentile, & Klonsky,
2007).
1.1.2.1. Decreased reaction time observed in VGPs: Brinley Plot

Throughout the RT literature, VGPs have been seen to exhibit faster RTs
compared to NVGPs. As was true of the accuracy studies, each of the reaction time
studies was designed to test a different aspect of perception or cognition, including
the ability to divide attention, the temporal dynamics of attentional orienting,
impulsivity, stimulus-response conflict resolution, alerting, executive functioning,
distractor rejection, serial search ability, and so on. However, unlike the common
thread of visual attention in the accuracy studies, no correspondingly obvious link
171
pops out that binds the reaction time studies together. To examine the possible
connection between VGP and NVGP performance on these tasks in more detail, we
employed what is known as a Brinley plot (Cerella, 1991). Commonly used in the
gerontology literature, a Brinley plot allows one to examine the relationship between
what may seem to be very disparate reaction time tasks, with the goal of uncovering a
factor common to all tasks. To construct this plot, each condition of each experiment
is plotted as a single point, with the x-coordinate corresponding to the mean NVGP
RT in that condition, and the y-coordinate corresponding to the mean VGP RT in that
same condition. The logic of the plot is that if a single underlying variable explains
the difference in performance between two groups across many types of tasks, a clear
correlation should be visible. If no such single variable exists, one would predict a
more scattered plot, wherein one group has more or less of an advantage depending
on the specific task or condition within a task. As is clearly evident in Figure 1, the
data strongly favors the former interpretation. Regardless of the task, condition
within the task, or total RT, VGP RT is approximately 12% faster than NVGP RT.
This is particularly notable because a priori there would be no reason to suspect that
because VGPs are 12% faster than NVGPs at a go/no-go task, that they would also be
12% faster at visual search, resolution of stimulus-response conflict, or orienting.
However, the fact that such a clear relationship does exist is strongly indicative of a
common underlying change in VGPs that explains the difference in performance
across all of the different tasks.
172
Figure 1. Comparison of VGP and NVGP RT across tasks

Several research studies, using a variety of experimental paradigms, have now observed decreased RTs
in individuals who play action video games. In this Brinley plot, eighty-four experimental conditions
from nine studies are plotted, with the x-coordinate of each point corresponding to the mean NVGP RT
in a given task condition and the y-coordinate corresponding to the mean VGP RT in the same task
condition. Plotting the data in this fashion demonstrates a clear linear relationship, wherein VGPs
respond approximately 12% faster than NVGPs regardless of task or task condition. This pattern
strongly suggests that a common underlying factor is responsible for much of the VGP RT advantage.
1.2. Hypothesis: Faster integration of perceptual information

Based on the totality of the gaming literature, any possible single mechanistic
explanation must be able to predict an increase in VGP accuracy in accuracy-based
experiments (which largely consist of making a perceptual judgment about a quickly
flashed stimulus display) and decreased RT in reaction time tasks (which must be
proportional to the total task RT - VGPs 12% faster than NVGPs). Perhaps the
simplest single mechanism that could potentially explain these results is an increase
in the rate at which perceptual information is integrated. According to this view,
173
VGPs would be more accurate than NVGPs on accuracy-based tasks because they are
able to extract more information from the flashed displays, while in reaction-time
tasks this faster integration would be manifested as quicker RTs.
However, none of the previous tasks are ideal to examine this hypothesis. A
change in the rate of information accrual will lead to both a decrease in response time
and an increase in accuracy RT and accuracy should be explicitly linked. If one
dependent measure changes without the other, or if they change in opposing
directions (for instance, faster RTs, but lower accuracy), then other hypotheses need
to be considered. Therefore, to fully evaluate the hypothesis, one would preferably
want a task where the accumulation of information over time can be measured by
examining both RT and accuracy. The experiments in the current gaming literature
are lacking in one dimension or the other. The accuracy literature is obviously
lacking in RT information, while in the RT literature accuracy is typically at ceiling
levels and thus not informative.
Therefore, to specifically test the hypothesis that VGP experience results in
faster integration of sensory information, we made use of a perceptual decision
making task, known to require the integration of information over time, and where
both RT and accuracy measures provide information about the rate at which this
integration occurs.
174
CHAPTER 2 Motion Coherence Decision Task
2.1. Introduction
A well-studied model of decision-making in the field of neuroscience is one in
which subjects accumulate evidence about a sensory stimulus over time, with the goal
of executing the proper motor action for that particular sensory stimulus and task
requirement. In particular, the coherent dot motion direction discrimination task has
been used extensively both in the human (Palmer, Huk, & Shadlen, 2005) and animal
(Roitman & Shadlen, 2002) literatures to assess the rate at which sensory information
is accumulated over time. In this task subjects are asked to determine the motion
direction of many simultaneously moving dots. RT and accuracy in this task are
known to reflect the information that is accumulated until the subject makes a
decision and executes a motor response. When many of the dots move in a consistent
direction (high coherence), RTs are generally very fast and accuracy is high.
Conversely, when very few dots move in a consistent direction (low coherence), RTs
are slow and as the percentage of consistently moving dots approaches zero, accuracy
approaches chance-level.
It is important to stress the fact that RT and accuracy in this task are tightly
coupled. Any processing change will alter the shape of both the RT and accuracy
functions in a predictable fashion. If the difference between VGPs and NVGPs is
truly a simple increase in the rate at which information accumulates in the VGP
group, one would predict both a decrease in RT as well as an increase in accuracy in
175
the VGP group. Graphically, as a function of coherence, this would be visualized as a

shift of both the VGP RT and accuracy plots to the left (faster/more accurate for
lower levels of coherence). This is intuitive if one considers that an increase in
integration rate/sensitivity is functionally equivalent to an increase in stimulus
strength; for the same reason that all subjects respond faster and are more accurate
when, for instance, contrast is increased in a visual display, a subject that accrues
sensory information more quickly will also respond faster and more accurately.
Another possible mechanistic explanation, at least for the reduction in RT
seen in the VGP population, is a reduction in decision criterion (speed-accuracy
trade-off). While a tendency to trade accuracy for increased speed could not explain
the increase in VGP accuracy in those tasks wherein accuracy was the primary
dependent measure, it could possibly explain the reduction in VGP RT seen in tasks
wherein RT was the primary dependent measure. Although accuracy was never
observed to differ statistically between VGPs and NVGPs in those tasks, accuracy
was also typically near ceiling, where statistically significant differences would be
difficult to obtain. Graphically, as a function of coherence, a simple speed-accuracy
tradeoff would be visualized as a shift of both the VGP RT plot to the left (faster for
lower levels of coherence), but a shift of the accuracy plot to the right (less accurate
for lower levels of coherence).
While the qualitative predictions regarding the raw accuracy and RT that
could be expected are clear, this motion direction discrimination task is of particular
interest, as psychometric models of this decision task indicate that performance (both
176
accuracy and RT) on the task can be captured by three main variables: (1) the rate at
which information is accumulated over time, which is a function of both the quality
of the stimulus itself as well as the sensitivity of the system to the stimulus (how well
the system is able to detect the given stimulus) (2) the stopping rule, or the threshold
at which the system stops accumulating evidence and the motor decision is made and
(3) a residual amount of time that is common to all tasks and reflects motor planning
and execution (independent of the stopping rule and accumulation rate). This
formalism allows us to not only examine the qualitative pattern of results, but also to
ask in a quantitative fashion which component of the decision making process is
modified by action video game experience.
2.2. Methods
2.2.1. Participants
Twenty-three males with normal or corrected-to-normal vision were placed
into one of two groups, VGP or NVGP, based upon their responses to a questionnaire
given prior to the experiment. Only males underwent testing because of the relative
paucity of females with sufficient video game experience.
The criterion to be considered a VGP was a minimum of 5 hours per week of
action video game usage for the previous six months. Eleven males with a mean age
of 18.8 years fell into this category. A highly abridged list of the games reported as
177
played includes Grand Theft Auto: San Andreas, Counterstrike, Dead Rising, Splinter
Cell, F.E.A.R., and Halo 2.
The criterion to be considered a NVGP was little, although preferably no,
action video game usage in the past six months. Twelve males with a mean age of
20.6 years fell into this category. All twelve males reported no action video game
experience in the past year, and little to no video game experience of any type (two
subjects reported playing sports games for 1 hour per month).
was paid $8 for each hour of participation.
2.2.2 Apparatus
The apparatus consisted of an Apple G4 computer running a program to
present stimuli and collect the data using the MATLAB computer language (The
Math Works Inc., Natick, MA) and the Psychophysical Toolbox routines (Brainard,
1997; Pelli, 1997) (http://psychtoolbox.org). The stimuli were displayed on a 20 flat
screen NEC Dimaondtron MultiSync FP2141SB CRT driven at 75Hz, 1600x1200
resolution.
2.2.3. Stimulus/Procedure
Each observer viewed the display binocularly at a test distance of 59 cm. The
stimuli were designed to be qualitatively the same as Palmer et al (Palmer et al.,
2005). The motion stimulus was created by presenting a sequence of frames
178
containing small white dots (0.1 square, density = 16.7dots/deg2/sec) within a gray
circular aperture (5 diameter). On each trial, although much of the dot motion was
random, coherent motion in a certain direction (left/right) and of certain strength was
always present. Seven levels of motion coherence were fully intermixed (0.8%, 1.6%,
3.2%, 6.4%, 12.8%, 25.6%, 51.2%) corresponding to the probability that a given dot
would be replotted in motion, as opposed to randomly replotted. In accord with
previous work on this paradigm, 3 interlaced sequences were used, meaning that a dot
in frame 1that was selected to be replotted in motion, would be replotted 0.2 in the
given direction 40 ms later in frame 4. This method ensures a limited dot lifetime and
also eliminates the ability to perform the task by comparing a given dots position
across successive frames. The subject was instructed to press the key on the keyboard
corresponding to the direction of the coherent motion (left/right arrow keys) as
quickly and accurately as possible. Subjects were told to respond at any point after
the onset of the motion up to a maximum of 2 seconds (after 2 seconds the motion
stimulus stopped and subjects were prompted for a response). After each trial,
auditory feedback was given (high tone = correct, low tone = incorrect) and after a
delay of 1 second the next trial began.
As this task is extremely challenging for novice observers, subjects first
completed three practice sessions (over the course of no more than 5 days). Each
practice session was composed of 100 trials per level of motion coherence (split
evenly between left and right motion) for a total of 700 trials per session, run in
pseudorandom fashion. Subjects then completed the experimental session within two
179
days of their final training session (also 100 trials per level of motion coherence, 700
total trials). Only the results of this final experimental session will be considered in
the analyses that follow.
2.3. Results
2.3.1. Raw Data

Accuracy was analyzed in a 2 (VGP Status: VGP/NVGP) x 7 (Coherence: 0.8,
1.6, 3.2, 6.4, 12.8, 25.6, 51.2%) ANOVA (see Figure 2, top panel). The expected
strong main effect of coherence was observed (F(6,126) = 219.3, p < .001, partial etasquared = .91) with accuracy increasing with increasing coherence (.52, .55, .57, .65,
.76, .93, .98). However, no main effect of VGP Status was observed (VGP = .71+/.02, NVGP = .71+/- .02; F(1,21) = 0.2, p = .65, partial eta-squared = .01) nor was an
interaction between VGP Status and coherence (F(6,126) = 0.57, p = .76, partial etasquared = .03). Although one obviously cannot accept the null hypothesis, there is no
statistical indication that the groups differ in accuracy.
Mean reaction time for correct trials was also analyzed in a 2 (VGP Status:
VGP/NVGP) x 7 (Coherence: 0.8, 1.6, 3.2, 6.4, 12.8, 25.6, 51.2%) ANOVA. As in
the accuracy analysis, a strong main effect of coherence was observed (F(6,126) =
48.5, p < .001, partial eta-squared = .70) with mean reaction time decreasing with
increasing motion coherence (.788, .803, .776, .757, .691, .596, .505). However,
180
unlike the accuracy analysis, both a strong main effect of VGP status (VGP = .592+/.018, NVGP = .803+/-.020; F(1,21) = 18.9, p < .001, partial eta-squared = .47) and a
VGP status x coherence interaction (F(6,126) = 3.5, p = .002, partial eta-squared =
.15) were observed. The main effect of VGP status is in agreement with the majority
of the gaming literature, as VGPs were seen to have significantly shorter mean RTs
than NVGPs, while the interaction suggests that this difference is largest at low
coherences when total RTs are slowest (see Figure 2, bottom panel).
181
Figure 2. VGP and NVGP Performance: Motion Direction Discrimination

VGPs and NVGPs demonstrate nearly identical accuracy (Top Panel) across levels of motion
coherence (x-axis), while VGPs respond substantially faster than NVGPs (Bottom Panel). The VGP
RT advantage is largest at low levels of coherence, but remains quite significant even at the highest
level of coherence.
*Note For illustration only, model fit corresponds to best fit to the mean data rather than the mean of
the individual fits.
2.3.2. Discussion: Raw Data

As predicted by the gaming RT literature, VGPs responded much more
quickly than NVGPs. However, in terms of accuracy, there was no evidence to
suggest either a shift of the function to the left (toward higher accuracy, which would
182
have been indicative of faster sensory integration) or a shift of the function to the
right (toward lower accuracy, which would have been indicative of a reduction in the
decision criterion). This does not appear to be due to a lack of statistical power
(improperly rejecting the alternative hypothesis). Rather, it appears that the two
groups truly do have very nearly identical accuracy (within 1% at 6 of 7 coherence
levels). Therefore the data are inconsistent with any single simple interpretation
(faster sensory integration or reduced criteria). The data are also inconsistent with
any interpretation positing a simple reduction in motor execution or stimulusresponse mapping, as the significant VGP status x coherence level interaction in the
RT analysis suggests that the RT difference between the groups depends on the total
task difficulty. Instead, the data suggest a more complicated pattern wherein multiple
factors play a role.
2.4. Introduction: Diffusion-to-bound model

In order to quantitatively assess the individual contribution of integration rate
and criterion, individual data were fit with the proportional-rate diffusion model
proposed by Palmer and colleagues (2005). This model visualizes the motion
coherence task as a simple noisy diffusion process, wherein information accumulates
over time until the amount of accumulated information hits a bound, which then
triggers a response. Importantly, the model predicts that both RT and accuracy are a
function of this diffusion process and should therefore be controlled by the same set
of parameters. The model has two main parameters that control how quickly and
183
accurately a decision is reached the mean drift rate (how quickly information
accumulates) and the position of the bound. The mean drift rate grows proportionally
with two factors the stimulus coherence and the individual subjects sensitivity. If
either coherence or sensitivity increases, the rate at which information is accrued will
increase correspondingly. This increased rate will in turn lead to faster RTs and more
accurate responses. As the stimulus coherence is known, the only free parameter
needed to fit the rate of accumulation is the individual subjects sensitivity. The
second main parameter, the position of the bound, controls the amount of information
that must be accumulated before the subject initiates a response. Decreasing the level
of the bound will lead to faster RTs, but less accurate responses (as the likelihood that
noise will lead to the incorrect bound being reached will increase). As both accuracy
and RT are hypothesized to be a function of the same process, the psychometric and
chronometric functions are fit simultaneously to give the parameters that best capture
overall performance.
Accuracy is modeled as a simple logistic function:

x: % coherence (motion strength)
k: sensitivity
A: bound/criterion
PC(x) = 1/(1+e-2Akx)
184
with x being the stimulus strength, A being the bound, and k the sensitivity. If
any of the three are equal to zero, percent correct goes to chance level (50%). This is
quite intuitive; if there is no motion present, the subject has no ability to detect
motion, or they wait for no motion to appear, the best they will be able to do is
chance. Conversely, as any of the three parameters increase, percent correct will
increases. Again, this is quite intuitive; it is much easier to discriminate the direction
of 100% coherent motion than 1% coherent motion, a subject who is extremely
sensitive to motion will be more accurate than a subject with little motion sensitivity,
and the higher one sets their bound, the less likely the bound will be hit by chance.
RT is modeled as a hyperbolic tangent:

x: % coherence (motion strength)
k: sensitivity
A: bound/criterion
TR: residual RT
RT(x) = A/kx * tanh (Akx) + TR
Unlike in the accuracy equation, wherein changes in bound and sensitivity lead to
similar directional changes in accuracy, for reaction time, the rate and bound
parameters come in opposite to one another. As the rate of integration is increased
(either by increasing motion coherence or sensitivity), RT will decrease. However,
RT increases as the bound parameter increases. Again the intuitions are clear.
185
Higher levels of motion coherence or higher motion sensitivity means more motion
information available per unit time. Thus, increases in either parameter will lead to
faster RTs. Conversely, increasing the bound means more information needs to be
accrued before the bound is reached and thus will slow RTs. The final parameter
(residual time TR) is a post-decision additive component that reflects the portion of
the total reaction time that is independent of the task parameters (motor execution,
stimulus-response mapping, etc).
Each subjects psychometric and chronometric functions were fit to the model
simultaneously by maximizing the average variance explained in the two functions.
The resulting best fitting parameters then offer a quantitative measure of the
contribution of sensitivity/rate of integration, bound, and non-decision components in
creating the observed pattern of differences between the VGPs and NVGPs.
2.4.1. Results: Diffusion-to-bound model

The model fits were good (r2VGP = .93, r2NVGP = .90) and equivalent in the two
groups (t(21) = 0.93, p = .36). Each of the three model parameters (A/bound,
k/sensitivity, and TR/residual time was analyzed in a t-test comparing the values in
VGPs and NVGPs. The bound parameter A, was seen to be significantly lower in the
VGPs than the NVGPs (VGP: .533+/-.05, NVGP: .732+/-.03; t(21) = 3.6, p = .002).
The rate/sensitivity parameter k was seen to be significantly larger in the VGPs than
the NVGPs (VGP: 10.94 +/- .9, NVGP: 7.6+/-.9; t(21) = 2.6, p = .02). Finally, no
186
significant difference was seen between the groups in the residual reaction time
parameter TR (VGP: .345+/-.02, NVGP: .356+/-.03; t(21) = 0.3, p = .76).
2.4.2. Discussion: Diffusion-to-bound model

As predicted by the analysis of the raw data, the pattern of results (VGPs
significantly faster than the NVGPs, but equally accurate) could not be captured by
manipulating a single parameter (rate of integration or bound), but instead appears to
reflect an interaction between the main parameters. A larger value of the
rate/sensitivity parameter is observed in the VGPs, indicating that the VGPs
accumulate sensory information more quickly than the NVGPs. All other parameters
being equal, this would tend to lead to both faster RTs and higher levels of accuracy.
However, all other parameters are not equal; VGPs have a smaller value of the bound
parameter, indicating that they make a decision sooner than the NVGPs. This
reduction in bound in VGPs contributes in the same direction as the increase in rate
for RTs, but directly counters the effect of sensitivity for accuracy. At the population
level, the rate and bound parameters trade-off nearly perfectly, which leads to
substantially faster RTs in the VGP population but nearly identical accuracy in the
two groups.
2.5. Discussion: Motion Coherence Decision Task

As has been observed throughout the gaming literature, VGPs were seen to
respond considerably faster than NVGPs. Alone, this reduction in RT would be
187
consistent with at least two simple mechanistic explanations, (1) an increase in

sensitivity/rate at which sensory information is accrued or (2) a reduction in decision
criterion/speed-accuracy trade-off. The first explanation, increased sensitivity, would
predict an increase in accuracy in the VGP population, while conversely the second
would predict a decrease in accuracy. Interestingly however, neither possibility was
observed, with instead VGP accuracy matching NVGP accuracy nearly identically
across all levels of motion strength. This pattern of drastically faster RTs, but
equivalent accuracy is indicative of a more complicated explanation wherein changes
in both sensitivity and criterion contribute to the final level of performance. In order
to quantify the relative involvement of sensitivity and criterion, the data was fit using
the proportional rate diffusion-to-bound model put forth by Palmer and colleagues
(2005). As indicated by the raw data, neither a change in integration rate nor a
reduction in bound could account for the data. Instead, it is the confluence of the two
factors (an increase in the rate of integration along with a reduction in bound) that
results in the difference in observed performance in the two populations.
The most obvious testable hypothesis based on these results is therefore that if
criterion was removed from the equation, VGPs should demonstrate an advantage in
accuracy compared to the NVGPs. In other words, if VGPs are truly able to extract
more sensory information per unit time, then if the same motion stimulus was
presented for the same amount of time to both the VGPs and NVGPs, the VGPs
should demonstrate a higher level of accuracy than the NVGPs.
188
CHAPTER 3 Motion Coherence Presentation Duration Task
3.1. Introduction
In an attempt to only measure the rate of sensory integration uncontaminated
by criterion, the stopping point was set experimentally, rather than allowing the
subjects to determine for themselves when enough information had accrued.
By presenting the motion stimulus for short periods of time and assessing accuracy as
a function of stimulus duration, the resulting function offers an excellent
measurement of the rate at which a subject acquires information. This type of task
has been used in the contrast sensitivity literature to assess the rate at which low
contrast information is acquired (Rovamo, Leinonen, Laurinenn, & Virsu, 1984).
3.2. Methods
This experiment was always run within two days of the completion of the
fourth (experimental) session of the motion decision task (Chapter 2). The subjects,
apparatus, and stimuli were identical to the previous experiment, with the exception
of two changes. First, rather than allowing the subjects to decide when to respond,
stimuli were presented for fixed durations and the subjects were instructed to watch
the entire motion stimulus and only make a decision after the dots stopped. Seven
such presentation durations were presented in fully intermixed fashion (50, 100, 200,
400, 800, 1600, 3200 ms). Also, only three coherence levels were tested 6.4%,
189
12.8%, and 25.6% - as performance at these coherence levels in the decision task was
neither at floor nor at ceiling, and would thus be maximally informative. Subjects
completed 50 repetitions (25 left, 25 right) for each of the duration/coherence pairs
for a total of 1050 trials presented in pseudorandom order.
3.3. Results
Accuracy was analyzed in a 2 (VGP Status: VGP/NVGP) x 3 (Coherence: 6.4,
12.8, 25.6%) x 7 (Presentation Duration: 50, 100, 200, 400, 800, 1600, 3200 ms)
ANOVA. As was seen in the decision task, a strong main effect of coherence was
observed with accuracy increasing with increasing levels of coherence (6.4%: .62,
12.8%: .72, 25.6%: .85; F(2,42) = 341.5, p < .001, partial eta-squared = .94). As
predicted, a main effect of presentation duration was also observed, with accuracy
increasing with increasing presentation duration up until around 800 ms where it
reached an asymptote (.54, .60, .70, .78, .82, .82, .85; F(6, 126) = 209.0, p < .001,
partial eta-squared = .91). An interaction between coherence and presentation
duration indicated that accuracy grew more quickly with time as coherence increased
(F(12, 252) = 17.1, p < .001, partial eta-squared = .45). Finally, an interaction in the
predicted direction was observed between presentation duration and VGP status with
accuracy growing faster with time in the VGPs (F(6, 126) = 2.1, p = .03: one-tailed,
partial eta-squared = .09). The main effect of VGP status was not significant,
190
although it was in the predicted direction (VGP: .87+/-.03, NVGP: .83+/-.04; F(1,21)
= 2.4, p = .13, partial eta-squared = .10).
Figure 3. VGP and NVGP Performance: Motion Presentation Duration

VGP and NVGP accuracy is plotted as a function of presentation duration. While VGP and NVGP
accuracy is reasonably equivalent at both extremely short presentation durations (both groups near
chance-level performance) and at extremely long presentation durations (both groups reach a common
asymptotic level of performance), VGP accuracy grows more quickly than NVGP accuracy in the
critical region between chance and asymptote. This pattern is particularly evident in the region
between .04 seconds and .5 seconds wherein VGP accuracy is easily distinguished as higher than
NVGP accuracy for both levels of motion coherence.
*Note Model fit corresponds to best fit to the mean data rather than the mean of individual fits.

Qualitatively, the data is quite consistent with the diffusion-to-bound model
from Chapter 2. Accuracy grew as a function of time and grew more quickly for
191
higher levels of coherence. The interaction between VGP status and presentation
duration is also consistent with the results of Chapter 2, in which VGPs were seen to
accumulate information more rapidly than NVGPs. However, in order to obtain a
more quantitative measure of the actual rate at which information accrues in the two
populations, a standard model in the psychophysical field was fit to the data.
3.3.3 Introduction: Model

Performance is modeled as a simple exponential rise to an asymptote (see (Carrasco,
Giordano, & B., 2004; Dosher, 1979) although note that the procedure and
interpretation of the function is quite different from the current design):
: asymptote
: rate of increase
: intercept
%Correct(t) = (1- e-(t- ))+50%
where lambda () is the level of asymptotic performance, beta () is the rate at which
accuracy grows as a function of time, and delta () is the intercept or the time at
which accuracy rises above chance levels. 50% is added to the function to convert
from percent above chance to percent correct. The main prediction based on the
earlier findings is a greater rate () value in the VGPs.
192

As the model fits were generally quite poor (r2 < .7) for the lowest level of
coherence (6.4%) only the two higher levels of coherence (12.8% & 25.6%) were
included in the analysis. Also, one NVGP was excluded because of poor model fits at
each of the coherence levels (r2<.6). For the models included in the analysis, fits
were good (r212.8%: .88, r225.6%: .97) and did not differ between the groups (R2VGP: .94,
R2NVGP: .92; F(1,20) = 0.5, p = .48). Each of the included model parameters
(asymptote/, rate/, intercept/) was entered into a 2 (VGP Status: VGP/NVGP) x 2
(Coherence: 12.8, 25.6%) ANOVA. For the asymptote parameter , a main effect of
coherence was observed, with asymptotic performance being higher for 25.6%
coherence than 12.8% coherence (12.8%: .87, 25.6%: .99; F(1,20) = 103.1, p < .001,
partial eta-squared = .84). However, neither the main effect of VGP status, nor the
interaction between coherence and VGP status approached significance. For the rate
parameter , a main effect of coherence was also observed, with the rate being higher
for the higher level of coherence (12.8%: 4.4, 25.6%: 8.4; F(1,20) = 13.0, p = .002,
partial eta-squared = .39). It is interesting to note that the rate parameter () in the
current model behaved almost exactly as would be predicted by the diffusion-tobound model of Chapter 2. In particular, the diffusion-to-bound model predicts that if
% coherence is doubled, the rate at which information accrues should double as well,
which is precisely what is observed. The significant main effect of VGP status
indicates a higher rate in the VGP group (VGP: 8.2+/-1.3, NVGP: 4.8+/-0.6; F(1,20)
193
= 4.5, p = .045, partial eta-squared = .19). The interaction between coherence and
VGP status did not approach significance. Finally, for the intercept parameter ,
neither the main effects of coherence and VGP status, nor the interaction approached
significance.
3.4 Discussion: Motion Coherence Presentation Duration Task

As predicted, VGP accuracy exceeded NVGP accuracy when short clips of
motion were presented. The accuracy advantage was particularly evident for
intermediate motion durations, between 50 ms (when little to no motion information
is present) and 800 ms (when subjects receive enough information to reach
asymptotic levels of accuracy). Fitting the raw data with an exponential approach to
an asymptote quantified the increased rate of accrual in the VGP population. These
results, together with the findings in Chapter 2, offer strong support for the hypothesis
that the rate at which sensory information accrues is greater in VGPs than NVGPs.
194
CHAPTER 4 Auditory Localization Decision Task
4.1. Introduction
The data from the coherent motion direction discrimination task suggests an
increase in the rate at which VGPs extract information from a visual display.
However, it is unknown whether this increase is specific to the visual modality or
whether it also generalizes to other senses. Indeed, the entire body of literature on the
effect of video game experience on perception has focused exclusively on the visual
system. While audio in early video games was largely inconsequential (i.e.
background music that was often uncorrelated with game conditions/performance), in
todays games it is often reasonably informative (primarily in orienting players to
unseen foes). However, it is certainly still the case that the primary focus of video
games continues to be vision (most games can be completed without any sound
whatsoever). Therefore, whether action video game experience affects auditory
integration in a manner similar to what was observed in the visual modality is
unclear.
To determine whether the increase in the rate at which sensory information is
accrued is seen in audition as well as vision, an auditory analog of the motion
direction task was developed. A pure tone embedded in a white noise mask was
presented in one ear, while white noise alone was presented in the other (both were
normalized to the same mean amplitude). The subjects task was to indicate the ear in
which the tone was presented as quickly and accurately as possible. In a manner
195
consistent with adjusting the coherence level of the motion stimulus, the ratio of the
amplitude of the target tone to the white noise mask was manipulated in order to test
performance across the range of possible accuracy levels and reaction times. As in
the motion task, performance on the task requires the accumulation of information
over time, the intuitions regarding sensitivity and criterion are similar and thus will be
modeled in the same manner.
4.2. Methods
4.2.1. Participants
The subjects were the same as those in the previous two experiments (Chapter
2 and Chapter 3).
4.2.2 Apparatus
The apparatus consisted of an Apple G4 computer running a program to
present stimuli and collect the data using the MATLAB computer language (The
Math Works Inc., Natick, MA) and the Psychophysical Toolbox routines (Brainard,
1997; Pelli, 1997) (http://psychtoolbox.org). The stimuli were presented binaurally
using Sennheiser HD 250 II headphones.
196
On each trial the subject was presented with, in one ear, a white noise
stimulus, and in the other, a 1000Hz tone of a variable amplitude added to the same
white noise stimulus. The absolute amplitude of the stimuli in both ears and across
sine wave amplitudes was fixed at 50 dB with the real modification being the ratio of
maximum noise amplitude to sine wave amplitude, which ranged from 1:1 (maximum
noise amplitude = maximum sine wave amplitude) to 64:1 (maximum noise
amplitude = 64x maximum sine wave amplitude). Eight signal-to-noise ratios
(SNRs) were tested in log spaced steps (1:64, 1:32, 1:16, 1:8, 1:4, 1:2, 1:1). The
subjects task was to press the arrow on the keyboard corresponding to the ear that
contained the sine wave as quickly as accurately as possible. The auditory stimulus
was present for a maximum of 1.5 seconds during which time the subject was free to
respond. After 1.5 seconds the auditory stimulus ceased and subjects were prompted
for a response. Visual feedback was given with a rectangle in the center of the screen
(green for a correct response and red for an incorrect response).
As with the motion task, this task is extremely challenging for novice
observers. As such two practice sessions were completed (over the course of no more
than 4 days) prior to the experimental session. Each session contained 100 trials per
SNR (split evenly between left and right ear) for a total of 800 trials presented in
pseudorandom fashion. As was the case in the motion experiment, only the data from
the final experimental session will be considered in the analyses that follow.
197
4.3. Results
4.3.1. Results: Raw Data
Accuracy was analyzed in a 2 (VGP Status: VGP/NVGP) x 8 (SNR: 1:64,
1:32, 1:16, 1:8, 1:4, 1:2, 1:1) ANOVA (see Figure 4, top panel). The expected strong
main effect of SNR was observed (F(7, 147) = 411.7, p < .001, partial eta-squared =
.95) with accuracy increasing with increasing SNR (.52, .51, .60, .83, .95, .98, .98,
.98). However, no main effect of VGP Status was observed (VGP = .78+/-.04,
NVGP = .80+/- .04; F(1,21) = 1.0, p = .32, partial eta-squared = .05) nor was an
interaction between VGP Status and SNR (F(7,147) = 0.4, p = .90, partial eta-squared
= .02). As was the case of motion accuracy, one cannot accept the null hypothesis,
however, there is no statistical indication that the groups differ in accuracy.
Mean reaction time for correct trials was also analyzed in a 2 (VGP Status:
VGP/NVGP) x 8 (SNR: 1:64, 1:32, 1:16, 1:8, 1:4, 1:2, 1:1) ANOVA. As in the
accuracy analysis, a strong main effect of SNR was observed (F(7, 147) = 77.0, p <
.001, partial eta-squared = .79) with mean reaction time decreasing with increasing
SNR (.808, .789, .742, .629, .495, .448, .425, .418). However, unlike the accuracy
analysis, both a strong main effect of VGP status (VGP = .488+/-.023, NVGP =
.802+/-.021; F(1,21) = 20.6, p < .001, partial eta-squared = .50) and a VGP status x
SNR interaction (F(7,147) = 5.2, p < .001, partial eta-squared = .2) were observed.
The main effect of VGP status is in agreement with the majority of the gaming
literature (along with the results of Chapter 2), as VGPs were seen to have
198
significantly shorter mean RTs than NVGPs, while the interaction suggests that this
difference is largest at low SNRs, when total RTs are slowest (see Figure 4, bottom
panel).
Figure 4. VGP and NVGP Performance: Auditory Localization

VGPs and NVGPs demonstrate nearly identical accuracy (Top Panel), while VGPs respond
substantially faster than NVGPs (Bottom Panel). The VGP RT advantage is largest at low SNRs, but
remains quite significant even at the highest SNR.
*Note Model fit corresponds to best fit to the mean data rather than the fit given by the mean of
individual fits.
199
4.3.2. Discussion: Raw Data

The observed pattern of results is remarkably similar to what was observed in
the motion analog of the task (Chapter 2) and therefore, the intuitions that can be
drawn are also similar. As was the case in the motion task, the VGPs and NVGPs
demonstrate nearly identical levels of accuracy, while the VGPs perform the task
substantially faster than the NVGPs, a relationship that is inconsistent with a single
simple interpretation (i.e. faster integration rate or reduced criteria). In order to
elucidate the relative contributions of integration rate and decision bound, the same
diffusion-to-bound model as was employed in Chapter 2 was fit to the data.
4.3.3. Results: Diffusion-to-bound model

The model fits were good (r2VGP = .90, r2NVGP = .91) and equivalent in the two
groups (t(21) = 0.91, p = .37). Each of the three model parameters (A/bound,
k/sensitivity, and TR/residual time) was analyzed in a t-test comparing the values in
VGPs and NVGPs. The bound parameter A, was seen to be significantly lower in the
VGPs than the NVGPs (VGP: .538+/-.05, NVGP: .712+/-.05; t(21) = 2.6, p = .02),
while the rate/sensitivity parameter k was seen to be significantly larger in the VGPs
than the NVGPs (VGP: 23.9 +/- .98, NVGP: 19.2 +/-.80; t(21) = 3.8, p = .001). No
significance difference was seen between the groups in the residual reaction time
parameter TR (VGP: .303+/-.013, NVGP: .378+/-.040; t(21) = 1.9, p = .07).
200
4.4. Discussion: Auditory Localization Decision Task

As was the case in Chapter 2, VGPs were seen to respond considerably faster
than NVGPs, but with equivalent accuracy, a pattern of results that suggests changes
in both sensitivity and criterion. In order to quantify the relative involvement of
sensitivity and criterion, the data were again fit using the proportional rate diffusionto-bound model put forth by Palmer and colleagues (2005) and again, the difference
in VGP performance was well captured by an increase in the rate of information
accrual along with a concomitant decrease in decision criteria. As in Chapter 3, the
next step is to directly test the rate at which information accrues, uncontaminated by
differences in criterion.
201
CHAPTER 5 Auditory Localization Presentation Duration Task
5.1. Introduction
The results of the auditory localization experiment suggest that VGPs acquire
sensory information more rapidly than NVGPs. All other parameters being equal,
this would lead to an increase in accuracy in the VGP group. However, as the VGP
group exhibits a reduction in decision criteria in addition to the increase in sensory
integration rate, an increase in accuracy relative to the NVGPs is not observed.
Therefore, if, as in Chapter 3, the stopping point is set experimentally, rather than
being under subject control, the VGPs should exhibit greater accuracy than the
NVGPs.
5.2. Methods
All subjects underwent the experiment within two days of completing the final
auditory decision task session. The subjects, apparatus, and stimuli were identical to
the previous experiment, with the exception of two changes. First, rather than
allowing the subjects to decide when to respond, stimuli were presented for fixed
durations and the subjects were instructed to watch the entire motion stimulus and
only make a decision after the dots stopped. Seven such presentation durations were
presented in fully intermixed fashion (20, 40, 80, 160, 320, 640, 1280 ms). Also,
only three SNR levels were tested 1:32, 1:16, 1:8 - as performance at these
202
coherence levels in the decision task was neither at floor nor at ceiling, and would
thus be maximally informative. Subjects completed 50 repetitions (25 left, 25 right)
for each of the duration/SNR pairs for a total of 1050 trials.
5.3. Results
Accuracy was analyzed in a 2 (VGP Status: VGP/NVGP) x 3 (SNR 1:32,
1:16, 1:8): x 7 (Presentation Duration: 20, 40, 80, 160, 320, 640, 1280 ms) ANOVA
(see Figure 5). As was seen in the decision task, a strong main effect of SNR was
observed with accuracy increasing with increasing levels of coherence (1:32: .60,
1:16: .76, 1:8: .88; F(2,42) = 444.7, p < .001, partial eta-squared = .96). As predicted
a main effect of presentation duration was also observed, with accuracy increasing
with increasing presentation duration up until around 320 ms where it reached an
asymptote (.56, .63, .72, .77, .82, .86, .87; F(6, 126) = 235.4, p < .001, partial etasquared = .92). An interaction between SNR and presentation duration indicated that
accuracy grew more quickly with time as SNR increased (F(12, 252) = 17.0, p < .001,
partial eta-squared = .45). Finally, an interaction in the predicted direction was
observed between presentation duration and VGP status with accuracy growing faster
with time in the VGPs (F(6, 126) = 1.9, p = .04 - one-tailed, partial eta-squared =
.08). The main effect of VGP status was also significant with VGPs being more
accurate than NVGPs (VGP: .78+/-.02, NVGP: .72+/-.012; F(1,21) = 6.5, p = .018,
partial eta-squared = .24).
203
Together, these results are consistent with the hypothesis that sensory
information is integrated more rapidly in the VGP population. However, to more
explicitly model the rate at which the auditory information accrues, the same model
as was used for the motion task (exponential approach to an asymptote) was fit to the
data.
Figure 5. VGP and NVGP Performance: Auditory Presentation Duration

VGP and NVGP accuracy is plotted as a function of presentation duration. While VGP and NVGP
accuracy is reasonably equivalent at both extremely short presentation durations (both groups near
chance level performance) and at extremely long presentation durations (both groups reach a common
asymptotic level of performance), VGP accuracy grows more quickly than NVGP accuracy in the
critical region between chance and asymptote. This pattern is particularly evident in the region
between .04 seconds and .2 seconds wherein VGP accuracy is easily distinguished as higher than
NVGP accuracy for both levels of motion coherence.
*Note For illustration purposes, model fit corresponds to best fit to the mean data rather than the fit
given by the mean of individual fits.
204

As the model fits were generally quite poor for the lowest level of SNR
((r21:32< .7) only the two higher levels of SNR (1:16 & 1:8) were included in the
analysis. For the models included in the analysis, fits were good (r21:16: .92, r21:8: .94)
and did not differ between the groups (R2VGP: .94, R2NVGP: .92; F(1,21) = 1.7, p = .21).
Each of the included model parameters (asymptote/, rate/, intercept/) was entered
into a 2 (VGP Status: VGP/NVGP) x 2 (SNR: 1:16, 1:8) ANOVA. For the asymptote
parameter , a main effect of SNR was observed, with asymptotic performance being
higher for the higher level of SNR (1:16: .93, 1:8: .98; F(1,21) = 20.8, p < .001). A
main effect of VGP status was also significant, with VGPs demonstrating higher
asymptotic accuracy (VGP: .98+/-.01, NVGP: .94+/-.01; F(1,21) = 4.8, p = .039).
However, the interaction between VGP Status and SNR did not approach
significance. For the rate parameter , a main effect of SNR was also observed, with
the rate being higher for the higher level of SNR (1:16: 10.9, 1:8: 26.1; F(1,21) =
27.2, p < .001). A main effect of VGP status indicates a higher rate in the VGP group
(VGP: 24.5+/-4.7, NVGP: 12.9+/-1.7; F(1,21) = 4.5, p = .045). The interaction
between coherence and VGP status did not approach significance. Finally, for the
intercept parameter , only a main effect of SNR was observed with performance
rising above chance levels earlier for the higher SNR (1:16: .015, 1:8: .004, F(1,21) =
9.9, p = .005).
205
5.4. Discussion: Auditory Localization Presentation Duration Task

As predicted, VGP accuracy exceeded NVGP accuracy when short clips of the
auditory stimulus were presented. The accuracy advantage was particularly evident
for intermediate presentation durations, between 20 ms (when little to no localization
information has become available) and 320 ms when subjects have reached
asymptotic levels of accuracy. Fitting the raw data with an exponential approach to
an asymptote quantified the increased rate of accrual in the VGP population. These
results, together with the findings in Chapter 4, offer strong support for the hypothesis
that the rate at which auditory information accrues is greater in VGPs than NVGPs.
206
CHAPTER 6 Video Game Training
6.1. Introduction
The previous 4 chapters have indicated an increase in the rate of sensory
integration in VGPs. While our hypothesis is that extensive video game experience is
at the root of this enhanced skill, it could also be the case that VGPs are individuals
who have been born with better perceptual skills. Such individuals would tend to
succeed at action video games, and therefore find such games rewarding, whereas
individuals born with poorer skills would tend to eschew those games that exceed
their capacity. Therefore, in order to establish that video game experience is
sufficient to drive an increase in the rate of sensory integration, a selection of NVGPs
underwent extensive video game training on either an action video game or a control
video game. If action video game experience does enhance the rate of sensory
integration, larger improvements should be noted in the action-trained group than in
the control group.
6.2. Methods
6.2.1. Participants
The study enrolled 25 NVGPs, none of whom had taken part in previous
experiments. The criteria for being considered an NVGP remained the same as in
207
previous experiments. All subjects underwent training as described below. In all, 7

females and 7 males (mean age = 25.7) made up the final experimental group, while
the final control group consisted of 7 females and 4 males (mean age = 24.7).
6.2.2. Testing Methods

All subjects underwent the same two decision tasks described previously
(motion coherence decision task - Chapter 2, auditory localization decision task
Chapter 4) as well as several other tasks not relevant to the work at hand, both before
and after video game training (described below).
6.2.3. Training methods
6.2.3.1. Training apparatus

Both groups played on 20 Dell LCD monitors.
6.2.3.2 Training procedure

for 50 total hours. The subjects were allowed to play a maximum of 2 hours per day
and a maximum of 10 hours per week. No minimum amount of game play per week
was enforced, but subjects were required to finish the 50 hours training in no more
than 12 weeks. The subjects completed the 50 hours in an average of 44 days. All
208
training games covered the entire extent of the screen (approximately 15 height x
18 width from fixation).
The eleven members of the control group played the game The Sims 2. The
Sims 2 is a simulation-style game, wherein the player takes complete control of the
life of a character (or the lives of several characters). The player must ensure that the
character meets all of his/her basic needs (cooking meals/eating, sleeping, using the
restroom, bathing, etc), they must guide the character toward short-term wants (which
can be buying a new curtain, kissing a neighbor, having a phone conversation with a
parent, etc) and away from short-term fears (such as losing a job or starting a fire),
and finally they must work the character toward the characters long-term aspiration
(such as having a family or making a fortune). Along the way, characters have jobs
(which require the character to acquire various skills to be promoted), relationships
with other characters (which with positive interactions improve and negative
interactions decline), are married, have children, change houses, grow old, and die essentially most of the things that characterize normal human life. As characters are
added to the players the household, the player takes control of and is responsible for
those characters as well. The control group played an average of ten different
characters and lived in an average of four different houses during the 50 hours of
training.
In an attempt to minimize the large inherent difference in the number of
characters, goals, and available environments between the control game and a
standard action game, the fourteen members of the experimental group played two
209
different action games (both chosen to be similar to those played by our VGPs). Both
action games have a relatively simple interface, use first-person point of view, and
require effective monitoring of the entire visual field. Unfortunately, even by using
two action games, the control group played five times as many characters, had
infinitely more (both in number and variety) immediate, short-term, and long-term
goals, and more diversity in environment; the nature of video games simply made it
impossible to fully equalize these parameters. As our hypothesis was that greater
learning would be observed in the action group than the control group, we chose to
have the action game trainees be at a disadvantage for these factors to ensure that
amount of novelty could not be used to explain any improvements in the action group
beyond what is seen in the control group.
During the first half of training, the experimental group played the game
Unreal Tournament 2004 in Death Match mode. In this mode the character is in an
abandoned warehouse with 32 computer-controlled artificial agents. The sole goal is
for the player to kill as many of these agents as possible, while minimizing the
number of times the player dies. Each training session was divided into 20-minute
blocks. The difficulty of the block was adjusted based on the kill/death ratio. If in a
block the player scored twice as many kills as they had deaths, the difficulty level was
increased one level. Players were retested on lower difficulty levels in the middle and
at the end of training to quantitatively assess improvement. During the second half of
training, the experimental group played the game Call of Duty 2. This game puts the
player into fictional or fictionalized World War II combat situations. Although Call
210
of Duty 2 has more goals than Unreal Tournament 2004, they remained somewhat
crude when compared with the control game (both in terms of variety nearly all
goals in Call of Duty 2 involved things like taking a bunker, a farmhouse, or a ridge and in terms of the possible ways to reach the goal all goals were accomplished
simply by killing every possible enemy soldier). Because Call of Duty 2 did not
provide quantitative data with which to assess improvement, subjects were retested on
Unreal Tournament 2004 during the final two hours of training.
6.2.3.3. Game playing improvement

In order to quantitatively assess game improvement and engagement in
training, several measures were used. For the control game, the best measure was
money accumulated (which increases with positive actions such as being promoted or
adding a member to the household and decreases with negative actions such as
burning down ones house or having a character die due to neglect). All subjects
showed an exponential increase in accumulated wealth over the course of training.
The time course of the accumulation was well fit by a polynomial function (wealth =
77(training hour)2+1319(training hour) + 9191).
For the action game, kills and deaths in each block were used to calculate a
skill metric ([Kills Deaths]/[Kills+Deaths]). This score increased dramatically for
all subjects at all levels of difficulty. For the easiest level of difficulty, the score went
from .53 (around three times as many kills as deaths) to .97 (nearly zero deaths). For
211
the hardest difficulty level the score went from -.74 (around four times as many
deaths as kills) to -.21 (approximately the same number of deaths and kills).
These results demonstrate that both groups were engaged in their training and showed
improvement on the training task.
6.3. Results
6.3.1. Motion coherence decision task: Raw Data

Accuracy was analyzed in a 2 (Test: Pre/Post) x 2 (Group: Action/Control) x 7
(Coherence: 0.8, 1.6, 3.2, 6.4, 12.8, 25.6, 51.2%) ANOVA (see Figure 6, top panels).
The expected strong main effect of coherence was observed (F(6,138) = 513.7, p <
.001, partial eta-squared = .96) with accuracy increasing with increasing coherence
(.51, .53, .55, .61, .73, .89, .98). However, neither the main effect of test (Pre = .68+/.05, Post = .69+/-.04, F(1,23) = 1.3, p = .26) nor any interactions with test approached
significance. A significant interaction between group and coherence was observed
(F(6, 138) = 2.4, p = .03, partial eta-squared = .08) with the action group having
slightly higher accuracy at the 12.8% and 25.6% coherence levels, but as this was
advantage was small and similar across tests it will not be considered further.
Mean reaction time for correct trials was also analyzed in a 2 (Test: Pre/Post)
x 2 (Group: Action/Control) x 7 (Coherence: 0.8, 1.6, 3.2, 6.4, 12.8, 25.6, 51.2%)
ANOVA (see Figure 6, bottom panels). As in the accuracy analysis, a strong main
effect of coherence was observed (F(6,138) = 31.0, p < .001, partial eta-squared =
212
.57) with mean reaction time decreasing with increasing motion coherence (.834,
.834, .821, .804, .759, .655, .551). However, unlike in the accuracy analysis, a strong
main effect of test was observed (Pre = .823+/-.06, Post = 691+/-.05, F(1,23) = 12.2,
p = .002, partial eta-squared = .35) as well as an interaction between test and group,
with the action group demonstrating a larger decrease in RT than the control group
between pre- and post-testing (Action: Pre = .809+/-.06, Post = .612+/-.05; Control:
Pre = .691+/-.05, Post = .679+/-.05, F(1,23) = 9.6, p = .005, partial eta-squared = .29).
While a significant interaction was also observed between test and coherence
(F(6,138) = 2.6, p = .02, partial eta-squared = .05) with reaction time decreasing by a
larger margin at low coherences than high coherences, the three-way interaction
between test, group, and coherence was not significant (F(6,138) = 1.7, p = .12,
partial eta squared = .07).
The qualitative pattern of results is consistent with the VGP/NVGP data from
Chapter 2. No change in accuracy was observed in either group, however, the actiontrained group demonstrated a large reduction in reaction time relative to the control
trained group. As in Chapter 2, the diffusion-to-bound model was fit to the data in
order to quantify the relative changes in integration rate and decision criteria.
213
Figure 6. Performance on Motion Discrimination Task Before and After Training

A. While no change in accuracy was observed in the group trained on the action video game (Top
Panel), a sizeable reduction in RT was observed (Bottom Panel). B. The group trained on the control
video game showed no change in either accuracy (Top Panel) or RT (Bottom Panel).
*Note Model fit corresponds to best fit to the mean data rather than the fit given by the mean
individual fits.
6.3.2. Motion coherence decision task: Model Data

The same diffusion-to-bound model as in Chapter 2 was fit to the data of the
trainees. The model fits were good (R2>.9) and did not differ as a function of group
or test (all p > .1). Each of the three model parameters (A/bound, k/sensitivity, and
TR/residual time) was analyzed in a 2(Test: Pre/Post) x 2 (Group: Action/Control)
214
ANOVA. For the bound parameter A, both a main effect of test (F(1,23) = 7.4, p =
.01, partial eta-squared = .24) and a test x group interaction (F(1,23) = 6.7, p = .02,
partial eta-squared = .23) were observed. The bound parameter decreased in both
groups (Action: Pre = .75+/-.05, Post = .57+/-.06; Control: Pre = .61+/-.06, Post =
.60+/-.05), but by a larger margin in the action group. It should be noted that
although the bound parameter was larger in the action group than the control group at
pre-test, this difference was not significant in a post-hoc analysis (p = .11).
For the rate/sensitivity parameter k, both a main effect of test (F(1,23) = 17.0,
p = .02, partial eta-squared = .2) as well as an interaction between test and group
(F(1,23) = 19.6, p = .02, partial eta-squared = .23) were observed with the action
group improving significantly at post-test while the control group did not change
(Action: Pre = 6.9+/0.6, Post = 9.3+/-0.6; Control: Pre = 7.1+/-.8, Post = 7.0+/-0.7).
Finally, neither the main effect of test nor a group x test interaction
approached significance for the residual reaction time parameter TR (all ps > .8).
6.3.3. Auditory localization decision task: Raw Data

Accuracy was analyzed in a 2 (Test: Pre/Post) x 2 (Group: Action/Control) x 8
(SNR: 1:64, 1:32, 1:16, 1:8, 1:4, 1:2, 1:1) ANOVA (see Figure 7, top panels). The
expected strong main effect of SNR was observed (F(7,161) = 608.1, p < .001, partial
eta-squared = .96) with accuracy increasing with increasing SNR (.52, .52, .57, .78,
.94, .97, .98). However, no other effects were significant.
215
Mean reaction time for correct trials was also analyzed in a 2 (Test: Pre/Post)
x 2 (Group: Action/Control) x 8 (SNR: 1:64, 1:32, 1:16, 1:8, 1:4, 1:2, 1:1) ANOVA
(see Figure 7, bottom panels). As in the accuracy analysis, a strong main effect of
SNR was observed (F(7,161) = 62.0, p < .001, partial eta-squared = .73) with mean
reaction time decreasing with increasing SNR (.855, .855, .824, .710, .565, .498, .465,
.457 secs). However, unlike in the accuracy analysis, a strong main effect of test was
observed (Pre = .704+/-.07, Post = .602+/-.06, F(1,23) = 15.3, p = .001, partial etasquared = .40) as well as an interaction between test and group, with the action group
having a larger decrease in RT than the control group between pre- and post-testing
(Action: Pre = .659+/-.09, Post = .542+/-.07; Control: Pre = .644+/-.06, Post =
.612+/-.05, F(1,23) = 5.4, p = .03, p-eta2 = .19). A significant interaction was also
observed between test and SNR (F(7,161) = 4.1, p < .001, partial eta-squared = .15)
with reaction time decreasing by a larger margin at low SNR than high SNR, but the
three-way interaction between test, group, and coherence was not significant
(F(7,161) = 0.7, p = .64, partial eta squared = .03). As was true in the motion task,
the qualitative pattern of results is consistent with the pattern observed in the
VGP/NVGP populations in Chapter 4. Mean accuracy did not change as a function of
training or group. Mean RT decreased in both groups, particularly at the lower levels
of SNR, but overall RT decreased by a larger margin in the action group. Again, the
diffusion-to-bound model was fit to the data to quantify the changes in criteria and
integration rate.
216
Figure 7. Performance on Auditory Localization Task Before and After Training

A. While no change in accuracy was observed in the group trained on the action video game (Top
Panel), a sizeable reduction in RT was observed (Bottom Panel). B. The group trained on the control
video game showed no change in accuracy (Top Panel) and a much smaller reduction in RT, mainly
for those levels of SNR where accuracy was near chance (Bottom Panel).
*Note Model fit corresponds to best fit to the mean data rather than the mean of the individual fits.
6.3.4. Auditory localization decision task: Model Data

The same diffusion-to-bound model as in Chapter 4 was fit to the data of the
trainees. The model fits were good (R2>.9) and did not differ as a function of group
or test. Each of the three model parameters (A/bound, k/sensitivity, and TR/residual
time) was analyzed in a 2(Test: Pre/Post) x 2 (Group: Action/Control) ANOVA. For
217
the bound parameter A, only a main effect of test (F(1,23) = 6.1, p = .02, partial etasquared = .2) was observed. The test x group interaction was not significant (F(1,23)
= 1.7, p = .2, partial eta-squared = .07) as the bound parameter in both groups
decreased at post-test (Action: Pre = .65+/-.06, Post = .50+/-.06; Control: Pre =
.66+/.05, Post = .62+/-.05).
For the rate/sensitivity parameter k, a main effect of test (F(1,23) = 18.5, p <
.001, partial eta-squared = .44) was observed. The interaction between test and group
was marginally significant (F(1,23) = 4.2, p = .026 one-tailed, partial eta-squared =
.16) as both groups had an increase in k, but the difference was larger in the action
group (Action: Pre = 17.6+/-.1.0, Post = 22.6+/-1.0; Control: Pre = 15.8+/-1.1, Post =
17.5+/-1.5).
Finally, neither the main effect of test nor a group x test interaction
approached significance for the residual reaction time parameter TR (all ps > .7).
6.4. Video Game Training: Discussion

As was the case in the VGPs of Chapters 2 and 4, following 50 hours of action
video game experience, reaction time dropped substantially without a concurrent drop
in accuracy in both the motion direction discrimination task as well as the auditory
integration task. This pattern was well fit in the diffusion-to-bound model by an
increase in the rate of sensory integration along with a decrease in criterion.
Significant interactions in both tasks between test and group indicated that the
increase in the rate of integration was significantly greater in the action-trained group
218
than the control trained group. Therefore, these findings cannot be attributed to
simple test-retest or Hawthorne-like effects. Although the bound parameter
decreased in both tasks for the action-trained group, the group by test interaction was
significant only in the case of the motion task. Therefore, although it appears that a
reduction in criterion is associated with action video game experience, it is less clearcut than in the case of an increase in the rate of sensory information accrual.
219
7. Conclusions
Over the past twenty-five years, a growing body of literature has indicated
that video game experience has the potential to enhance basic perceptual, motor,
and/or cognitive processes. While each report in the literature posited a different
independent enhancement, whether in the capacity of visual attention, the spatial
resolution or temporal resolution of visual attention, the ability to divide attention, the
general speed of processing, the efficiency of visual search, the susceptibility to
distractors, or the formation of stimulus-response mappings, we have put forth the
hypothesis that a single mechanistic change, an increase in the rate at which sensory
information accrues, can account for the majority of the findings in the literature. By
making use of two sensory integration tasks, one in the visual modality and one in the
auditory modality, we have demonstrated that indeed, video game experience is
correlated with an increase in the rate at which sensory information accumulates. By
training NVGPs on an action video game and observing similar changes, we have
further demonstrated that the relationship between video game experience and this
increased rate of sensory information accrual is causative.
However, while we have good evidence for this mechanistic change as a result
of video game experience, it would be beneficial to explore how such a change in the
rate of sensory integration could be exhibited neurally. A model developed by Pouget
and colleagues attempts to model performance on the dot motion decision task in a
neurally plausible manner. In this model, a layer of independent Poisson neurons
with Gaussian tuning curves for motion direction represent area MT. These neurons
220
feed onto an integration layer composed of 100 linear/non-linear Poisson (LNP)

neurons representing area LIP. These LIP neurons have lateral connections and a
long time constant, which in essence makes them near perfect integrators of the
sensory information being passed from MT (at least over the time scale of a typical
trial). The weight on the feed-forward connection between area MT and area LIP
determines the rate at which sensory information accrues (essentially acting similar to
the sensitivity). A decision is made when the peak activity in a given LIP neuron
reaches a certain rate in Hz. Interestingly, in a manner consistent with what was
observed in the diffusion-to-bound model of Palmer and colleagues, the difference
between VGP and NVGP performance can be well captured in this model by
increasing the weight on the feed-forward connections as well as decreasing the
stopping bound in the VGP group. Because more sensory information reaches area
LIP from area MT, the rate at which sensory information grows over time is
obviously greater in the VGP population. However, as was the case in the diffusionto-bound model, a decrease in the stopping bound is required to match the particular
pattern of results, with VGPs faster but with the same accuracy as the NVGPs.
Although the exact brain areas that contribute to performance in the auditory
integration task are not known, the model of Pouget and colleagues suggests that a
change in the rate of integration rate would likely be manifested as a change in the
connections between sensory and integration areas (rather than for instance a change
in the lateral connections between integration neurons). If the changes in rate are
truly manifested at the level of connections between sensory and integration areas, it
221
would suggest that the rate at which auditory information is accrued might be
reasonably independent of the rate at which motion information is accrued (there
being no a priori reason to believe that, among those with normal vision and hearing,
individuals with extremely good hearing would also have extremely good vision and
vice versa). Conversely, one might imagine that the bound is set by a higher level
structure that controls the integration of information across modalities. In this case
one would expect to find a significant correlation between the bound parameters in
the two integration tasks. Individuals who are willing to make decisions with less
information would likely be willing to do so regardless of the task. These predictions
are indeed born out by the data. No correlation is observed between the motion and
auditory integration rates. Although VGPs tend to have higher integration rates for
both motion and auditory information, there is no correlation in individual subject
performance (see Figure 8A Top Panel). Indeed, the VGP with the highest rate of
auditory integration also showed the lowest rate of motion integration. The same
pattern (or lack thereof) is observed in the data from the video game trainees (Figure
8A Bottom Panel). Conversely, a strong linear relationship with a slope of
approximately 1 is observed between the bound parameters (see Figure 8B Top
Panel). Again, while VGPs tend to have lower bound parameters than NVGPs for
both the motion and auditory tasks, there is also a strong correlation in the bounds of
individual subjects. For instance, the NVGP with the highest motion bound also
demonstrated the highest auditory bound and the VGP with the lowest auditory bound
222
demonstrated the second lowest motion bound. Again, the same pattern was observed
in the video game trainees (Figure 8B Bottom Panel).
Figure 8. Correlations between auditory and motion parameters

A. No relationship is observed between the rate of auditory integration and the rate of motion
integration in either the VGPs/NVGPs (Top Panel) or in the trainees (Bottom Panel). B. A strong
linear relationship is observed between the auditory and motion bounds in the VGPs/NVGPs (Top
Panel) as well as the trainees (Bottom Panel).
One might further wonder whether the bound and the integration rate are truly
independent factors or whether they might be correlated. While this would not be
223
surprising (it makes intuitive sense that individuals who acquire information very
rapidly might tend to cut off sooner), this is not born out by the data. No systematic
relationship is observed between an individual subjects integration rate and the same
subjects bound (Figure 9). This is an important point because it demonstrates that
the tradeoff between bound and rate of integration observed in the VGP and NVGP
populations is truly only applicable at the population level.
Figure 9. Correlation between rate and bound

No relationship is observed between an individual subjects rate of integration and their bound.
It will be for future work to determine why the VGPs, who acquire
information more rapidly than NVGPs, manifest this enhancement through extremely
fast RTs rather than extremely accurate responses. Theoretically, the VGPs could
have chosen to wait as long as the NVGPs before making a decision and in doing so
would have instead demonstrated an accuracy advantage.
224
One possible hypothesis is that subjects are attempting to maximize reward

per unit time (Gold & Shadlen, 2002). If this is indeed the case, once given the rate
of sensory integration, the residual time, the time between trials, and some
quantization of the reward (i.e. the utility of the reward), the value of the bound that
maximizes reward per unit time can be calculated. When a simple linear mapping
between correct beeps and reward (i.e. one beep = one util with no temporal
discounting) is employed, the trends are in the predicted direction with VGPs being
closer to the optimal reward rate than the NVGPs. However, both VGPs and NVGPs
demonstrate bounds that are substantially higher than those which would correspond
to the optimal reward rate. This may reflect either a simple failure to maximize or
may indicate that the simple linear mapping does not truly capture the utility of the
rewards (correct/incorrect beeps). Our experimental design does not allow us to
measure how rewarding subjects find the correct beep or correct green square,
how punishing they find the incorrect beep or incorrect red square, or how they
temporally discount those values (i.e. a reward right now is typically worth more
than the same reward sometime in the future). Future directions in this field will
therefore likely include quantitative measurements of the utility function, such that
the amount of reward acquired per unit time can be more precisely calculated.
225
8. References
Bialystok, E. (2006). Effect of bilingualism and computer video game experience on
the Simon task. Canadian Journal of Experimental Psychology, 60, 68-79.
Birch, E. E., & Stager, D. R. (1996). The critical period for surgical treatment of
dense congenital unilateral cataract. Investigative Ophtalmology and Visual
Science, 37(8), 1532-1538.
Carrasco, M., Giordano, A. M., & B., M. (2004). Temporal performance fields: visual
and attentional factors. Vision Research, 44, 1351-1365.
Cerella, J. (1991). Age effects may be global, not local: comment on Fisk and Rogers
(1991). Journal of Experimental Psychology: General, 120(2), 215-223.
Clark, J. E., Lanphear, A. K., & Riddick, C. C. (1987). The effects of videogame
playing on the response selection processing of elderly adults. Journal of
Gerontology, 42(1), 82-85.
Dosher, B. A. (1979). Empirical approaches to information processing: Speedaccuracy tradeoff functions or reaction time - a reply. Acta Psychologica,
43(347-359).
226
Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the
efficiency and independence of attentional networks. Journal of Cognitive
Neuroscience, 14, 340-347.
Gold, J. I., & Shadlen, M. N. (2002). Banburismus and the brain: Decoding the
relationship between sensory stimuli, decisions, and reward. Neuron, 36, 299308.
Leark, R. A., Wallace, D. R., & Fitzgerald, R. (2004). Test-retest reliability and
standard error of measurement for the Test of Variables of Attention
(T.O.V.A) with healthy school-age children. Assessment, 11(4), 285-289.
Lewis, T. L., & Maurer, D. (2005). Multiple sensitive periods in human visual
development: evidence from visually deprived children. Developmental
Psychobiology, 46(3), 163-183.
Orosy-Fildes, C., & Allan, R. W. (1989). Psychology of computer use: XII.
Videogame play: Human reaction time to visual stimuli. Perceptual and
227
Palmer, J., Huk, A. C., & Shadlen, M. N. (2005). The effect of stimulus strength on
the speed and accuracy of a perceptual decision. Journal of Vision, 5, 376404.
Roitman, J. D., & Shadlen, M. N. (2002). Response of neurons in the lateral
intraparietal area during a combined visual discrimination reaction time task.
Journal of Neuroscience, 22(21), 9475-9489.
Rosser Jr., J. C., Lynch, P. J., Cuddihy, L., Gentile, D. A., & Klonsky, J. (2007). The
impact of video games on training surgeons in the 21st century. Archives of
Surgery, 142, 181-186.
Rovamo, J., Leinonen, L., Laurinenn, P., & Virsu, V. (1984). Temporal integration
and contrast sensitivity in foveal and peripheral vision. Perception, 13(6),
665-674.
Yuji, H. (1996). Computer games and information processing skills. Perceptual and

CSG Thesis Final

Uploaded by

Copyright:

Available Formats

CSG Thesis Final

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CSG Thesis Final

Uploaded by

Copyright:

Available Formats

The Effects of Action Video Game Experience on Perceptual

Submitted in Partial Fulfillment

Department of Brain and Cognitive Sciences

NVGPs. Importantly, similar effects can be induced in NVGPs through extensive

Observed Differences Between VGPs and NVGPs

1.1.1 Increased Accuracy Observed in VGPs

1.1.1.1 The Spatial Distribution of Visual Attention

(Green and Bavelier, 2006, JEP:HPP)

(Green and Bavelier, 2007, Psychological Science)

(Green and Bavelier, 2003, Nature)

(Green and Bavelier, 2006, Cognition)

1.1.2 Decreased Reaction Time Observed in VGPs

1.1.2.1 Decreased RT in VGPs: Brinley Plot

Hypothesis: Faster Integration of Perceptual Information

Motion Coherence Decision Task

2.3.1 Raw Data

2.3.2 Discussion: Raw Data

Introduction: Diffusion-to-Bound Model

2.4.1 Results: Diffusion-to-Bound Model

2.4.2 Discussion: Diffusion-to-Bound Model

Discussion: Motion Coherence Decision Task

Motion Coherence Presentation Duration Task

3.3.1 Results: Raw Data

3.3.2 Discussion: Raw Data

3.3.3 Introduction: Model

3.3.4 Results: Model

Discussion: Motion Coherence Presentation Duration Task

Auditory Localization Decision Task

4.3.1 Results: Raw Data

4.3.2 Discussion: Raw Data

4.3.3 Results: Diffusion-to-Bound Model

Discussion: Auditory Localization Decision Task

Auditory Localization Presentation Duration Task

5.3.1 Results: Raw Data

5.3.2 Results: Model

Discussion: Auditory Localization Presentation Duration Task

Video Game Training

6.2.2 Testing Methods

6.2.3 Training Methods

6.2.3.1 Training Apparatus

6.2.3.2 Training Procedure

6.2.3.3 Game Playing Improvement

6.3.1 Motion Coherence Decision Task: Raw Data

6.3.2 Motion Coherence Decision Task: Model Data

6.3.3 Auditory Localization Decision Task: Raw Data

6.3.4 Auditory Localization Decision Task: Model Data

Video Game Training: Discussion

1: Comparison of VGP and NVGP RT Across Tasks

2: VGP and NVGP Performance: Motion Direction Discrimination

3: VGP and NVGP Performance: Motion Presentation Duration

4: VGP and NVGP Performance: Auditory Localization

5: VGP and NVGP Performance: Auditory Presentation Duration

6: Performance on Motion Discrimination Task Before and After Training

7: Performance on Auditory Localization Task Before and After Training

8: Correlations Between Auditory and Motion Parameters

9: Correlation Between Rate and Bound

1.1. Observed differences between VGPs and NVGPs

1.1.1. Increased accuracy observed in VGPs

1.1.1.1. The spatial distribution of visual attention