The EVect Of Video Game Violence On Physiological Desensitization ...
Journal of Experimental Social Psychology 43 (2007) 489–496
www.elsevier.com/locate/jesp
The eVect of video game violence on physiological
desensitization to real-life violence
Nicholas L. Carnagey a,¤, Craig A. Anderson b, Brad J. Bushman c
a Department of Psychology, Iowa State University, USA
b Department of Psychology, Iowa State University, Center for the Study of Violence, USA
c Department of Psychology, University of Michigan, USA and Vrije Universiteit, Amsterdam, The Netherlands
Received 1 April 2005; revised 15 March 2006
Available online 17 July 2006
Abstract
Past research shows that violent video game exposure increases aggressive thoughts, angry feelings, physiological arousal, aggressive
behaviors, and decreases helpful behaviors. However, no research has experimentally examined violent video game eVects on physiologi-
cal desensitization, deWned as showing less physiological arousal to violence in the real world after exposure to video game violence in the
virtual world. This experiment attempts to Wll this gap. Participants reported their media habits and then played one of eight violent or
nonviolent video games for 20 min. Next, participants watched a 10-min videotape containing scenes of real-life violence while heart rate
(HR) and galvanic skin response (GSR) were monitored. Participants who previously played a violent video game had lower HR and
GSR while viewing Wlmed real violence, demonstrating a physiological desensitization to violence. Results are interpreted using an
expanded version of the General Aggression Model. Links between desensitization, antisocial, and prosocial behavior are discussed.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Violent video games; Violent media; Desensitization; Helping; Prosocial
Over the last three decades the video game industry has
try are rated as considerably violent by parents (Thompson
evolved from oVering a handful of games on bulky home
& Haninger, 2001). In addition, the Federal Trade Commis-
systems to oVering scores of video games on console sys-
sion found that many game manufacturers market violent
tems, personal computers, handheld systems, PDAs, and
games to children (FTC, 2000).
even cell phones. Accompanied with the success of this
Another reason for this debate is the abundance of
thriving industry has been public debate concerning the
research demonstrating negative eVects of violent media
impact of video game exposure. Currently, one of the pri-
exposure. Youth exposed to violent media tend to become
mary public and political issues concerns the eVect of expo-
more aggressive immediately after exposure, and become
sure to excessively violent video games on aggression and
more aggressive adults (e.g., Anderson et al., 2003). The
violence.
eVect of violent television exposure at an early age (between
One reason for this debate is the high prevalence of vio-
6 and 11 years old) on later violent behavior has been
lence in current video games. Over 85% of games contain
shown to be larger than the eVects of low IQ, abusive par-
some violence, and approximately half of video games
ents, exposure to antisocial peers, and being from a broken
include serious violent actions (e.g., Children Now, 2001).
home (US Department of Health & Human Services, 2001).
Video games rated “E” (Everyone: Ages 6+) by the indus-
The smaller video game literature has found that playing
violent video games causes increases in aggressive behavior,
aggressive aVect, aggressive cognitions, physiological
The authors thank Doug Bonett for his help with the HR analyses.
*
arousal, and decreases in prosocial behavior (Anderson
Corresponding author. Fax: +1 515 294 6424.
E-mail address: vasser@iastate.edu (N.L. Carnagey).
et al., 2004). But what does the media violence research
0022-1031/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2006.05.003
490
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
literature have to say about desensitization? The answer is:
Due to ambiguity on how to operationalize desensitization,
both too little and too much.
research on this phenomenon is also somewhat unclear.
In 1982, the US Surgeon General called for research on
One of the earliest studies demonstrating the potential
whether video game violence desensitizes individuals to real
desensitizing eVect of violent media measured GSR of indi-
violence (Koop, 1982). Unfortunately, there is no published
viduals while they watched a documentary Wlm of a tribal
research on this speciWc topic, and the public debate fre-
ceremony that included making incisions on the human
quently generates more heat than light.
body (Lazarus, Speisman, MordkoV, & Davison, 1962).
Participants’ had lower GSRs during incisions at the end of
Desensitizing eVects of violent entertainment media
the Wlm than at the beginning; early scenes of gore appar-
ently reduced physiological arousal to later scenes.
The term “desensitization” has been used by scholars,
Other experimental research has yielded similar results.
public policy analysts, politicians, and the lay public to
Participants in one study who viewed a series of “slasher”
mean eVects as varied as: (a) an increase in aggressive
Wlm clips had lower heart rates when shown additional vio-
behavior; (b) a reduction in physiological arousal to real-
lent movie clips than did participants who initially viewed
life violence; (c) a Xattening of aVective reactions to vio-
nonviolent clips (Linz, Donnerstein, & Adams, 1989). Chil-
lence; (d) a reduction in likelihood of helping a violence vic-
dren in another study who saw a violent movie had lower
tim; (e) a reduction in sympathy for a violence victim; (f) a
GSR to a staged “real-life” violent scene than did children
reduction in the sentence for a convicted violent oVender,
who had previously viewed a nonviolent movie (Thomas,
(g) a reduction in the perceived guilt of a violence perpetra-
Horton, Lippincott, & Drabman, 1977). Similar results
tor; and (h) a reduction in judged severity of a violence vic-
were found in an adult population (Thomas et al., 1977). In
tim’s injuries. This hodge-podge of deWnitions—confusing
a related study, college students who had been provoked
lay people and scientists alike—results from a failure to dis-
and had viewed a violent Wlm clip had lower heart rate
tinguish underlying psychological desensitization processes
before and after shocking their provoker than did students
from potential desensitization eVects on other responses.
who viewed a nonviolent Wlm clip (Thomas, 1982).
Too much is included in this broad deWnition.
Other research has demonstrated that past violent
A narrower, clearer deWnition of desensitization to vio-
media exposure correlates with physiological desensitiza-
lence is a reduction in emotion-related physiological reactivity
tion to violence (e.g., Cline, Croft, & Courrier, 1973). For
to real violence. This deWnition Wts well with earlier systematic
example, one study found that past violent video game
desensitization research in cognitive-behavioral treatment of
exposure was related to reduced P300 amplitudes when
phobias (e.g., Wolpe, 1958, 1982). Systematic desensitiza-
exposed to violent photos, even after controlling for ini-
tion—a set of procedures designed to reduce unwanted nega-
tial levels of aggressiveness (Bartholow, Bushman, &
tive emotional reactions to stimuli that initially produce fear
Sestir, 2006).
or anxiety—has been successfully used to treat fear of such
Although these studies are important and insightful,
things as spiders (Bandura, Reese, & Adams, 1982), snakes
none directly address the issue of whether exposure to vio-
(Bandura & Adams, 1977), and blood (Elmore, Wildman, &
lent media physiologically desensitizes individuals to real-
WestWeld, 1980). It has been used to treat anxiety-related dis-
life violence. The main public concern with desensitization
orders such as post-traumatic stress (Pantalon & Motta,
to violence is not that viewing media violence lowers physi-
1998), rape trauma (Frank, Anderson, Stewart, Dancu, &
ological responsiveness to other media violence, but that it
West, 1988), and nightmares (Schindler, 1980). There is also
lowers responsiveness to real world violence.
evidence that the US military has used video games for a
For a direct, causal test of the hypothesis that expo-
variety of training missions, including desensitizing soldiers
sure to violent media can cause physiological desensitiza-
to violence (Grossman & DeGaetano, 1999).
tion to real-life violence, four experiment characteristics
Whether induced intentionally (e.g., therapeutic system-
are necessary: (1) random assignment to violent or nonvi-
atic desensitization) or unintentionally, desensitization can
olent media exposure groups; (2) use of violent and non-
be adaptive, allowing individuals to ignore irrelevant stim-
violent entertainment media that are equivalent (or
uli and attend to relevant stimuli. For example, desensitiza-
statistically controlled) on various nonviolent aspects
tion to distressing sights, sounds, and smells of surgery is
(such as excitement, frustration, involvement level); (3)
necessary for medical students to become eVective sur-
use of emotion-related physiological indicators as the
geons. Desensitization to battleWeld horrors is necessary for
dependent variable (e.g., heart rate, GSR); and (4) use of
troops to be eVective in combat. However, desensitization
real violence as the emotion-provoking stimulus in the
of children and other civilians to violent stimuli may be det-
dependent variable assessment (this provides more gener-
rimental for both the individual and society.
alizable Wndings compared to measuring desensitization
to Wctitious violence). None of the prior studies meet all
Media violence and physiological desensitization
four criteria. Furthermore, no published study has exper-
imentally examined whether exposure to violent video
There are surprisingly few media violence studies exam-
games decreases physiological responsiveness to real-life
ining physiological–emotional indicators of desensitization.
violence.
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
491
GAM, desensitization, and sequela
eVects, visual eVects, rewards for violent actions in the
game), these initial distressing reactions are reduced. One
There are theoretical reasons for expecting violent media
indicator that desensitization has occurred is observation
to desensitize individuals to real-life violence in both short-
of a reduction in physiological arousal (e.g., heart rate,
term (within 1 hour of exposure) and long-term (repeated
GSR) during exposure to real violence after individuals
exposure) contexts. The General Aggression Model (GAM)
have been repeatedly exposed to media violence.
provides a useful social-cognitive framework for under-
Once desensitization has occurred, new presentations of
standing desensitization processes.
real violence instigate diVerent cognitive and aVective reac-
GAM has been described in detail elsewhere (e.g.,
tions than would have occurred in the absence of desensiti-
Anderson & Bushman, 2002; Anderson & Carnagey, 2004;
zation. For example, desensitized people might be less likely
Anderson & Huesmann, 2003), so it will be described only
to notice aggressive events, perceive fewer or less severe
brieXy here. Aggressive behavior is based on the learning,
injuries, feel less sympathy for violence victims, believe that
activation, and application of aggression-related knowl-
the world is a less safe place, and have less negative atti-
edge structures stored in memory. Such learning takes place
tudes towards violence. These cognitive and aVective
through encounters with the physical and social world.
sequela are critical determinants of subsequent episodic
Much learning occurs through observing real and Wctional
decisions and actions.
characters.
In this article, we extend GAM to desensitization eVects.
Overview
In our model, “desensitization” is best seen as a process by
which initial arousal responses to violent stimuli are
In this experiment, participants Wrst completed measures
reduced, thereby changing the individual’s “present internal
of video game preferences and trait aggressiveness. Partici-
state.” GAM further speciWes that subsequent decision and
pants then played either a violent or nonviolent video game
behavioral processes will be inXuenced. Fig. 1 details how
for 20 min. Afterwards, they watched a 10-min videotape
exposure to violent video games might produce physiologi-
containing Wlmed scenes of real violence while heart rate
cal–emotional desensitization, and how desensitization
(HR) and galvanic skin response (GSR) were continuously
inXuences other aggression and helping-related variables.
monitored. Finally, participants rated the video game they
The initial response of children and many adults to vio-
played on several dimensions. We predicted that violent
lent media is fear and anxiety (e.g., Cantor, 1998). When
game players would show less physiological arousal (lower
violent stimuli are repeatedly presented in a positive emo-
HR and GSR) in response to real-life violence than would
tional context (e.g., exciting background music, sound
nonviolent game players.
Repeated violent game playing:
Exposure to initially fearful stimuli in
a positive emotional context
Desensitization
*Humorous context
Procedures
*Cartoonish characters
*Rewards for acting violently
*Increasingly intense violent stimuli
-Levels within a game
-"E" to "T" to "M" games
Extinction of fear/anxiety
Desensitization
reactions to violence
(e.g., decreased heart rate reactivity)
Decreased
Decreased
Decreased
Increased
Decreased
Cognitive
perception
attention
sympathy
belief that
negative
& affective
of injury
to violent
for violence
violence is
attitudes
outcomes
severity
events
victims
normative
towards violence
GAM Episodic Processes
Increased aggression
Decreased helping
Behavioral
Higher likelihood of initiating aggression
Lower likelihood of intervening
outcomes
More severe level of aggression
Delay in intervening
More persistence in aggressing
Fig. 1. Media violence desensitization processes: integration of systematic desensitization, helping, and aggression models.
492
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
Method
lating, and addicting, and violent, using 10-point scales
ranging from 1 (strongly disagree) to 10 (strongly agree).
Participants
Finally, they were debriefed.
Participants were 257 college students (124 men and 133
Results
women) who received extra course credit in exchange for
their voluntary participation.
Preliminary analyses
Procedure
Within each type of video game (i.e., violent, nonviolent),
we tested whether the four diVerent games produced diVer-
Participants were tested individually. They were told
ent eVects on HR and GSR. No signiWcant diVerences were
that the purpose of the study was to evaluate diVerent types
found among the four violent or among the four nonviolent
of media. After consent procedures were completed, 5 min
games on HR or GSR. The random-eVects variance esti-
baseline HR and GSR measurements were taken, using
mates for video game exemplars ranged from 0 to 4.43
Wnger electrodes placed on the three middle Wngers of the
(M D 1.06). None of the maximum likelihood random-
non-dominant hand.1 During the baseline period, partici-
eVects variance estimates signiWcantly diVered from zero,
pants reported the number of hours per week they spent
ps > .05. Thus, the data were collapsed across exemplars of
playing video games and the percent of time spent playing
video game types for subsequent analyses.
violent video games (preference for violent video games).2
As expected, violence ratings were higher for the violent
They also completed the nine-item Physical Aggression
video games than for the nonviolent video games,
subscale of the Aggression Questionnaire (Buss & Perry,
Ms D 4.87 and 1.70, F (1, 252) D 529.67, p < .0001, d D 2.90.
1992).3 A sample item is: “Given enough provocation, I
However, violent and nonviolent video games also diVered
may hit another person.” CoeYcient was .84.
on several other dimensions. Compared to the nonviolent
After 5 min, the experimenter removed the electrodes.
games, the violent games were rated as being more action
Participants played a randomly assigned violent or nonvio-
packed, Ms D 5.31 and 3.17, F (1, 252) D 53.48, p < .0001,
lent video game for 20 min. To make the Wndings more gen-
d D 0.92, more frustrating, Ms D 5.79 and 4.63,
eralizable (Wells & Windschitl, 1999), we used four violent
F (1, 252) D 14.30,
p < .001,
d D 0.48, less enjoyable,
games (Carmageddon, Duke Nukem, Mortal Kombat, Future
Ms D 3.95 and 4.95, F (1, 252) D 11.73, p < .001, d D 0.43, and
Cop) and four nonviolent games (Glider Pro, 3D Pinball,
less fun, Ms D 4.00 and 4.81, F (1, 252) D 6.84, p < .01,
3D Munch Man, Tetra Madness).
d D 0.33. Violent and nonviolent games did not diVer on
After game play, a second set of 5-min HR and GSR
absorbing, arousing, boring, entertaining, exciting, involv-
measurements were taken. Next, participants watched a
ing, stimulating, or addicting ratings, Fs (1, 252) < 3.10,
10-min videotape of real violence in four contexts: court-
ps > .08, ds < 0.22. Dimensions that yielded nonsigniWcant
room outbursts, police confrontations, shootings, and
eVects were dropped. Others were used as covariates in the
prison Wghts. These were actual violent episodes (not Hol-
main analyses.
lywood reproductions) selected from TV programs and
commercially released Wlms. In one scene, for example,
Main analyses
two prisoners repeatedly stab another prisoner. HR and
GSR were monitored continuously while they watched
Heart rate4
the real-life violence. Participants then rated the video
HR data were analyzed using a 2 (video game exposure:
game they had played on the following dimensions:
violent, nonviolent) £ 2 (participant gender) £ 3 (measure-
action-packed, frustrating, enjoyment, fun, absorbing,
ment time: baseline, after video game, during Wlm) mixed
arousing, boring, entertaining, exciting, involving, stimu-
design, with additional planned contrasts testing the main
hypotheses. The video game exposure and participant sex
1
factors were between-subjects, whereas the measurement
Due to random hand movements during measurement periods, some
data sections containing impossible HR estimates were deleted before av-
time factor was within-subjects. We initially considered
erage HRs were calculated.
trait aggression, preference for violent games, and rated
2 On average, men spent 4.15 h (SD D 5.11) per week playing video
nonviolent game characteristics as possible covariates or
games, whereas women spent 1.43 h (SD D 2.67). On average, men spent
moderators, but they were unrelated to HR during real vio-
42% (SD D 36%) of this time playing violent games, whereas women spent
lence viewing and were therefore dropped.
4% (SD D 12%) of this time playing violent games. Such self-reports likely
underestimate the actual proportion of time spent playing violent games
The time main eVect was signiWcant, F (2, 496) D 9.70,
because most game players do not regard cartoon violence as “violence”
p < .001. Adjusted average HRs were 66.0, 68.8, and 69.6 at
(Anderson, Gentile, & Buckley, in press; Potter, 1999).
3 Preference for violent video games was positively correlated with self-
reported physical aggression, r(255) D .32, p < .0001. The amount of time
4 Two participants in the nonviolent video game group (1 woman, 1
spent playing video games in general was correlated with level of physical
man) and three participants in the violent video game group (2 women, 1
aggressiveness, r(255) D .16, p < .02, but this latter correlation was signiW-
man) were excluded from the HR and GSR analyses due to equipment
cantly smaller than the former one, t(253) D 2.22, p < .05.
malfunctions.
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496 493
baseline, post-gameplay, and during the film, respectively.
Galvanic skin response
As expected, HR increased from baseline to post-game play,
The same procedure was used to analyze GSR, except
F (1,248) = 9.40, p <.05, d = 0.39 (see Fig. 2). The only other
that three of the rated game characteristics (action, involve,
effect that approached significance was the game violence x
fun) were used as covariates because they yielded significant
time interaction, F (2,496) = 2.06, p < .13. Fig. 2 displays the
effects either in the overall analysis or in one of the planned
means. All other Fs< 1.
contrast analyses. Neither trait aggression nor past violent
There were no HR differences between violent and
video game exposure yielded any reliable effects in
nonviolent conditions during baseline [Ms = 66.4 and 65.5,
preliminary analyses, so they were not included in analyses
respectively, F(1,248)=0.43, p>.05, d =0.08] or after game
reported here.
play, Ms=69.3 and 68.4, respectively, F(1,248)=0.55, p> .05,
The 2x2x3 ANCOVA yielded a significant main effect
d = 0.09 (see Fig. 2). Because we expected the interaction
of gender, F (2,244) = 18.06, p < .001, d = 0.54, with
between video game content and measurement time to have
women being higher in GSR than men (Ms = 302.98 and
a specific form (i.e., a spreading interaction rather than a
234.68). In addition, there was an uninteresting time X
cross-over one), we tested it using two planned contrasts.
gender interaction, F (2, 244) = 5.36, p < 005. Women were
The first contrast tested the effect of video game content on
higher than men at baseline [Ms = 297.22 and 243.19, F
heart rate differences at baseline versus after playing the
(2,244) = 10.53, p < .01, d = 0.42], after game play [Ms =
video game. Because we attempted to match the violent and
303.12 and 239.85, F (2,244) = 14.89, p < .001, d = 0.49],
nonviolent video games in terms of how arousing they were,
and during viewing of real-life violence [Ms = 308.62 and
we did not expect any heart differences after playing the
221.01, F (2,244) = 22.51, p < .001, d = 0.61]. Unlike HR,
video game. As expected, the first contrast was small,
there was not an overall change in GSR across time, F (2,
F(1,248) = 0.00, p >.05, d = 0.00. This small contrast
488) = 0.91, p >.40.
suggests that increases in HR during game play were
Similar to HR, there were no GSR differences between
essentially the same for the violent and nonviolent game
violent and nonviolent conditions during baseline [Ms =
conditions.
258.98 and 281.4, respectively, F (1, 244) = 1.19, p >.25, d
The second (key) contrast tested the effect of video game
= 0.14] or after game play, T2 Ms = 261.8 and 281.1,
content on HR during the showing of the real-life violence,
respectively, F (1, 244) = 0.91, p > .30, d = 0.12 (See Fig.
versus the average HR during the other two time periods. As
3). The interaction between video game content and
expected, the second contrast was large, F(1,248)=4.86, p
measurement time was again tested using two planned
<.05, d=0.28 (see Fig. 2). This large contrast suggests that
contrasts. The first contrast testing whether game violence
violent game players were less aroused by the real-life
differentially affected GSR from baseline to post-gameplay
violence than were nonviolent game players. People who had
was again small, F(1, 244)=0.05, p>.80, d=0.03 (see Fig. 3).
recently played a violent video game were less aroused by
This contrast demonstrates that change in GSR was similar
the filmed violence than were those who had recently played
for violent and nonviolent game players.
a nonviolent video game. Indeed, heart rates of nonviolent game
The second (key) contrast comparing the average of
players increased while viewing the film (relative to post-
the GSRs during baseline and post-gameplay to GSR dur-
game HR), F(1,131)=16.60, p<.05, d=.72, whereas heart
ing the viewing of the filmed real violence was large,
rates of violent game players did not change while viewing
the filmed real violence, F(1,116)=.41, p>.05, d=.11.
75
Nonviolent video game
Violent video game
70
Rate
Heart 65
60
Baseline
AfterGame Play
While Viewing
Violence
Measurement Time
Fig. 3. Galvanic skin response at baseline, after playing a video game, and
while watching filmed real-life violence for violent and nonviolent video
Fig. 2. Heart rate at baseline, after playing a video game, and while watching
game players. Mean GSR scores for violent and nonviolent video game
filmed real-life violence for violent and nonviolent video game players. Mean
players during the viewing of filmed real-life violence were 242.9 and 286.8,
heart rates for violent and nonviolent video game players during the viewing
respectively. All means are adjusted for gender, how fun and involving the
of filmed real-life violence were 68.5 and 70.7, respectively. All means are video games were, and how much action the video games contained. Capped
adjusted for gender. Capped vertical bars denote 1 SE. vertical bars denote 1 SE.
494 N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
F (1, 244) =4.67, p <.05, d =0.28 (see Fig. 3). This large
Helping researchers have demonstrated numerous
contrast demonstrates that violent game players were
factors that yield decreases in helping (for a review, see
less aroused by the real-life violence than were
Bat-son, 1998), but the link between desensitization and
nonviolent game players. Participants who had just
helping behavior has not been as carefully examined. Two
played a violent video game experienced relatively less
lines of research are most relevant to GAM and helping: (1)
arousal during the viewing of the filmed violence.
Furthermore, nonviolent game participants had a slight
the work by Latane and Darley (1968), and
(nonsignificant) increase in GSR while viewing the film,
(2) research examining helping as a function of tension
F(1, 127) = 0.52, p >.05, d =0.13, whereas violent game
reduction.
participants showed a decrease in GSR, F (1, 113) =
After the 1964 murder of Kitty Genovese in New York
7.19, p <.05, d = 0.50. Several effects of rated video
City, Latane and Darley were motivated to conduct
game characteristics were also significant: action, F (1,
experimental studies to find out why people did not help
244) = 6.00, p < .05; fun, F (1, 244) = 7.95, p < .05;
her. Newspapers reported that several witnesses watched
involving, F (1, 244) =4.41, p < .05. The higher rated
Genovese being assaulted for 30 min, yet none called the
level of action and involvement in the games, the larger
police or attempted to intervene. Their studies found that
the increase in GSR from the composite baseline to
several factors are required for intervention to occur. Three
during the viewing of real-life violence, r =.13, b =5.96,
of these factors are particularly relevant to GAM’s link
p < .05 and r =.11, b =6.09, p <.05. The less fun the
games were rated, the larger the increase in GSR from
between desensitization and failure to help violence victims
baselines to during the viewing of real-life violence, r
(see Fig. 1). First, the individual must notice or attend to
=.02, b =7.88, p <.05.
the violent incident. Desensitization might reduce attention
to violent incidents involving other people. Second, the
Discussion
individual must recognize the event as an emergency.
Desensitization might reduce the perceived seriousness of
The results demonstrate that playing a violent video
injury, and thereby decrease the likelihood of perceiving
game, even for just 20 min, can cause people to become
the situation as an emergency. Third, the individual must
less physiologically aroused by real violence.
feel personally responsible to help. Desensitization might
Participants randomly assigned to play a violent video
reduce sympathy for the victim, increase beliefs that
game had relatively lower HR and GSR while watching
violence is normative, and decrease negative attitudes
actual footage of people being beaten, stabbed, and shot
towards violence, thereby decreasing feelings of personal
than did those randomly assigned to play a nonviolent
responsibility.
video game.
Similarly, research has shown that people are more
One issue that arises frequently in the media violence
likely to help when they are highly aroused (Dovidio,
literature concerns individual differences in
Piliavin, Gaertner, Schroeder, & Clark, 1991). Other
susceptibility to media violence effects. If there are
helping researchers have proposed that observation of a
large individual differences in susceptibility to short-
victim produces a state of aversive tension that can
term desensitization effects, they would be revealed in
motivate helping behavior (e.g., Cialdini & Kendrick,
the present study as significant interactions between the
1976). In both cases, individuals desensitized to violence
individual difference variables (violent video game
should be less likely to help a violence victim.
preference; trait aggressiveness; gender) and the
Several studies have examined the effects of violent
experimental manipulation of game violence. We found
television on children’s helping behavior (e.g., Drabman &
no such interactions, suggesting that the results are quite
Thomas, 1974). Children were exposed to a short violent or
robust across individuals.
nonviolent television program, and then were exposed to a
Relevance of results to GAM
situation in which they could intervene in a fight they
believed was occurring in another room by calling an adult.
Although the present experiment did not measure
Participants exposed to a violent TV program took longer
aggressive or helping behaviors, we think our extension
to intervene than did those exposed to a nonviolent TV
of the GAM in Fig. 1 can explain such behaviors (see
program. But the researchers did not test whether the
Fig. 1). Desensitization to violence can increase
violent TV program used could produce physiological
aggression in several ways. For example, when
desensitization in the target population.
considering several possible behavioral scripts for
Similarly, several studies have found decreases in
guiding action, children who have strong negative
prosocial behavior as a result of exposure to violent video
reactions to a violent script are less likely to use it than
games (e.g., Silvern & Williamson, 1987). But again, there
those who have either a neutral or a positive reaction to
was no evidence in these studies that the violent games
it (Anderson & Huesmann, 2003; Huesmann, 1998).
produced physiological desensitization in the target
Exposure to violent video games can increase aggression
population. Furthermore, none of these studies examined
in ways that are largely unrelated to desensitization as
prosocial behavior towards a violence victim.
well, but because the present study focuses on
In sum, GAM is relevant to media violence desensitiza-
desensitization our discussion (and Fig. 1) emphasize
tion in two ways. First, the social-cognitive-learning
effects on helping behavior.
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
495
aspects predict that exposure to violent entertainment
Conclusion
media will decrease normal negative emotional-physiolog-
ical reactions to real violence. The current experiment
The present experiment demonstrates that violent video
tested this hypothesis, using violent video games as the
game exposure can cause desensitization to real-life vio-
entertainment medium, heart rate and galvanic skin
lence. In this experiment, violent game players were less
response as the physiological measures, and Wlmed real-
physiologically aroused by real-life violence than were non-
world violence as the stimulus during dependent measure
violent game players. It appears that individuals who play
assessment. Second, GAM predicts that violent entertain-
violent video games habituate or “get used to” all the vio-
ment stimuli that produce physiological desensitization
lence and eventually become physiologically numb to it.
will also decrease helping a violence victim. A recent study
The integration of systematic desensitization processes
has found such results (Bushman, Shlecter, Anderson, &
and models of helping behavior with GAM is heartening in
Carnagey, in preparation).
the insights provided to a long-standing and somewhat
muddled research literature. But it is also frightening in
Future research
some of its implications. The existing rating systems (Bush-
man & Cantor, 2003), the content of much entertainment
Numerous important theoretical questions remain for
media, and the marketing of those media combine to yield a
future research. Several features of violent video games sug-
powerful desensitization intervention on a global level.
gest that they may have even more pronounced eVects on
Children receive high doses of media violence. It initially is
users than violent TV programs and Wlms. Violent video
packaged in ways that are not too threatening, with cute
game players are more actively involved, more likely to
cartoon-like characters, a total absence of blood and gore,
identify with violent characters, more directly reinforced for
and other features that make the overall experience a pleas-
violent acts, and more frequently exposed to violent scenes.
ant one, arousing positive emotional reactions that are
In the past, the level of realism in video game images might
incongruent with normal negative reactions to violence.
have reduced their ability to create physiological desensiti-
Older children consume increasingly threatening and realis-
zation. But recent technological advances have removed this
tic violence, but the increases are gradual and always in a
obstacle from video games. Consequently, desensitization
way that is fun. In short, the modern entertainment media
and decreases in helping might well progress more quickly
landscape could accurately be described as an eVective sys-
and eYciently in violent video game players than in violent
tematic violence desensitization tool. Whether modern soci-
TV/Wlm viewers. Future research should investigate how
eties want this to continue is largely a public policy
these diVerences between types of entertainment media
question, not an exclusively scientiWc one (Anderson et al.,
inXuence desensitization to real violence. Future research
2003; Gentile & Anderson, 2006).
also should investigate who is most likely to become desensi-
tized as a result of exposure to violent video games.
References
Another question concerns the duration of desensitizing
eVects of violent video games. Although no studies have
Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual
been conducted on this topic, some violent Wlm studies
Review of Psychology, 53, 27–51.
show that without repeated exposure, the laboratory eVects
Anderson, C. A., Berkowitz, L., Donnerstein, E., Huesmann, L. R., Johnson,
of media violence exposure on perceived victim injury can
J., Linz, D., Malamuth, N., & Wartella, E. (2003). The inXuence of media
deteriorate in a matter of days (Mullin & Linz, 1995). Of
violence on youth. Psychological Science in the Public Interest, 4, 81–110.
Anderson, C. A., & Carnagey, N. L. (2004). Violent evil and the general
course, most youths and many adults are exposed to violent
aggression model. In Chapter A. Miller (Ed.), The Social Psychology of
media on a regular basis.
Good and Evil (pp. 168–192). New York: Guilford Publications..
Finally, although our main focus has been on
Anderson, C. A., Carnagey, N. L., Flanagan, M., Benjamin, A. J., Eubanks,
unintended desensitization and helping eVects of violent
J., & Valentine, J. C. (2004). Violent video games: SpeciWc eVects of vio-
video games, a better understanding of video game eVects
lent content on aggressive thoughts and behavior. Advances in Experi-
mental Social Psychology, 36, 199–249.
can be put to good use in other contexts. For example, if
Anderson, C. A., Gentile, D. A., & Buckley, K. E. (in press). Violent video
violent video games designed primarily to entertain are
game eVects on children and adolescents: Theory, research, and public
good at producing physiological desensitization, then it
policy. Oxford University Press.
should be feasible to design games to produce such
Anderson, C. A., & Huesmann, L. R. (2003). Human aggression: A social-
desensitization in desired populations and contexts. Can
cognitive view. In M. A. Hogg & J. Cooper (Eds.), Handbook of social
psychology (pp. 296–323). London: Sage Publications.
we make better combat soldiers by desensitizing them to
Bandura, A., & Adams, N. E. (1977). Analysis of self-eYcacy theory of
some of the sights and sounds of combat? Can we help
behavioral change. Cognitive Therapy and Research, 1, 287–310.
medical students become comfortable with the types of
Bandura, A., Reese, L., & Adams, N. E. (1982). Microanalysis of action
physical and emotional trauma they will experience in
and fear arousal as a function of diVerential levels of perceived self-
emergency rooms? Can we use video games to systemati-
eYcacy. Journal of Personality and Social Psychology, 43, 5–21.
Bartholow, B. D., Bushman, B. J., & Sestir, M. A. (2006). Chronic violent
cally desensitize individuals who need to be desensitized
video game exposure and desensitization to violence: Behavioral and
to speciWc stimuli that cause them problems (e.g., auto
event-related brain potential data. Journal of Experimental Social Psy-
accident victims afraid of riding or driving again)?
chology, 42, 532–539.
496
N.L. Carnagey et al. / Journal of Experimental Social Psychology 43 (2007) 489–496
Batson, D. C. (1998). Altruism and prosocial behavior. In D. T. Gilbert, S.
Koop, C. E. (1982). Surgeon General sees danger in video games. New
T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (pp.
York Times, A16.
282–316). Boston: McGraw-Hill.
Latané, B., & Darley, J. M. (1968). Group inhibition of bystander interven-
Bushman, B. J., Shlecter, L. R., Anderson, C. A., & Carnagey, N. L. (in
tion in emergencies. Journal of Personality and Social Psychology, 10,
preparation). Violent video game eVects on helping violence victims.
215–221.
Bushman, B. J., & Cantor, J. (2003). Media ratings for violence and sex:
Lazarus, R. S., Speisman, M., MordkoV, A. M., & Davison, L. A. (1962). A
Implications for policy makers and parents. American Psychologist, 58,
laboratory study of psychological stress produced by a motion picture
130–141.
Wlm. Psychological Monographs: General and Applied, 34 Whole No.
Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of
553.
Personality and Social Psychology, 63, 452–459.
Linz, D., Donnerstein, E., & Adams, S. M. (1989). Physiological desensiti-
Cantor, J. (1998). “Mommy, I’m scared”: How TV and movies frighten chil-
zation and judgments about female victims of violence. Human Com-
dren and what we can do to protect them. San Diego, CA: Harvest/Har-
munication Research, 15, 509–522.
court.
Mullin, C. R., & Linz, D. (1995). Desensitization and resensitization to vio-
Children Now. (2001). Children and the media. Retrieved July 1, 2001,
lence against women: EVects of exposure to sexually violent Wlms on
from <http://www.childrennow.org/>.
judgments of domestic violence victims. Journal of Personality and
Cialdini, R. B., & Kendrick, D. T. (1976). Altruism as hedonism: A social
Social Psychology, 69, 449–459.
development perspective on the relationship of negative mood state
Pantalon, M. V., & Motta, R. W. (1998). EVectiveness of anxiety manage-
and helping. Journal of Personality and Social Psychology, 34, 907–914.
ment training in the treatment of posttraumatic stress disorder: A pre-
Cline, V. B., Croft, R. G., & Courrier, S. (1973). Desensitization of children
liminary report. Journal of Behavior Therapy and Experimental
to television violence. Journal of Personality and Social Psychology, 27,
Psychiatry, 29, 21–29.
360–365.
Potter, W. J. (1999). On media violence. Thousand Oaks, CA: Sage Publica-
Dovidio, J. F., Piliavin, J. A., Gaertner, S. L., Schroeder, D. A., & Clark, R.
tions.
D., III (1991). The arousal: Cost-reward model and the process of
Schindler, F. E. (1980). Treatment by systematic desensitization of a recur-
intervention: A review of the evidence. In M. S. Clark (Ed.), Prosocial
ring nightmare of a real life trauma. Journal of Behavior Therapy and
behavior (pp. 86–118). Newbury Park, CA: Sage.
Experimental Psychiatry, 11, 53–54.
Drabman, R. S., & Thomas, M. H. (1974). Does media violence increase
Silvern, S. B., & Williamson, P. A. (1987). The eVects of video game play on
children’s toleration of real-life aggression? Developmental Psychology,
young children’s aggression, fantasy and prosocial behavior. Journal of
10, 418–421.
Applied Developmental Psychology, 8, 453–462.
Elmore, R. T., Wildman, R. W., & WestWeld, J. S. (1980). The use of sys-
Thomas, M. H. (1982). Physiological arousal, exposure to a relatively
tematic desensitization in the treatment of blood phobia. Journal of
lengthy aggressive Wlm, and aggressive behavior. Journal of Research in
Behavior Therapy and Experimental Psychiatry, 11, 277–279.
Personality, 16, 72–81.
Federal Trade Commission. (2000). Marketing violent entertainment to
Thomas, M. H., Horton, R. W., Lippincott, E. C., & Drabman, R. S. (1977).
children: A review of self-regulation and industry practices in the
Desensitization to portrayals of real life aggression as a function of
motion picture, music recording, & electronic game industries. Report
television violence. Journal of Personality and Social Psychology, 35,
of the Federal Trace Commission. Available online at: <http://
450–458.
www.ftc.gov/bcp/congress/index.html/>.
Thompson, K. M., & Haninger, K. (2001). Violence in E-rated video
Frank, E., Anderson, B., Stewart, B. D., Dancu, C., & West (1988). EYcacy
games. Journal of the American Medical Association, 286, 591–598.
of cognitive behavior therapy and systematic desensitization in the
United States Department of Health and Human Services. (2001). Youth
treatment of rape trauma. Behavior Therapy, 19, 403–420.
Violence: a report of the Surgeon General. Rockville, MD: US Depart-
Gentile, D. A., & Anderson, C. A. (2006). Violent video games: EVects on
ment of Health and Human Services, Centers for Disease Control and
youth and public policy implications. In N. Dowd, D. G. Singer, & R.
Prevention, National Center for Injury Prevention and Control; Sub-
F. Wilson (Eds.), Handbook of children, culture, and violence (pp. 225–
stance Abuse and Mental Health Services Administration, Center for
246). Thousand Oaks, CA: Sage.
Mental Health Services; and National Institutes of Health, National
Grossman, D., & DeGaetano, G. (1999). Stop teaching our kids to kill : A
Institute of Mental Health.
call to action against TV, movie & video game violence. New York:
Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social psy-
Crown Publishers.
chological experimentation. Personality and Social Psychology Bulletin,
Huesmann, L. R. (1998). The role of social information processing and
25, 1115–1125.
cognitive schema in the acquisition and maintenance of habitual
Wolpe, J. (1958). Psychotherapy by reciprocal inhibition. Stanford, CA:
aggressive behavior. In R. Geen & E. Donnerstein (Eds.), Human
Stanford University Press.
aggression: Theories, research and implications for policy (pp. 73–109).
Wolpe, J. (1982). The practice of behavior therapy (3rd ed.). New York:
New York: Academic Press.
Pergamon Press.
Document Outline
