Cognitive Brain Research 25 (2005) 851 – 861

www.elsevier.com/locate/cogbrainres

Research Report

Individual differences in extraversion and dopamine genetics

predict neural reward responses
Michael X. Cohen a,b,*, Jennifer Young b, Jong-Min Baek c,
Christopher Kessler c, Charan Ranganath b
a
Department of Epileptology, University of Bonn, 53115 Bonn, Germany
b
Center for Neuroscience, University of California, Davis, CA 95616, USA
c
Genomics Facility, University of California, Davis, CA 95616, USA

Accepted 20 September 2005

Available online 11 November 2005

Abstract

Psychologists have linked the personality trait extraversion both to differences in reward sensitivity and to dopamine functioning, but little
is known about how these differences are reflected in the functioning of the brain’s dopaminergic neural reward system. Here, we show that
individual differences in extraversion and the presence of the A1 allele on the dopamine D2 receptor gene predict activation magnitudes in
the brain’s reward system during a gambling task. In two functional MRI experiments, participants probabilistically received rewards either
immediately following a behavioral response (Study 1) or after a 7.5 s anticipation period (Study 2). Although group activation maps
revealed anticipation- and reward-related activations in the reward system, individual differences in extraversion and the presence of the D2
Taq1A allele predicted a significant amount of inter-subject variability in the magnitudes of reward-related, but not anticipation-related,
activations. These results demonstrate a link between stable differences in personality, genetics, and brain functioning.
D 2005 Elsevier B.V. All rights reserved.

Theme: Neural basis of behavior

Topic: Motivation and emotion

Keywords: Extraversion; Personality; fMRI; Dopamine; Genetics; Reward

1. Introduction been linked to clinical disorders such as depression and

addictions [14,30,45], little is known about how individual
Although much is known about the neural substrates of differences in psychological and biological variables affect
reward processing, a complete understanding of the neural the functioning of this system.
mechanisms of reward must take into account how Extraversion, one of the most extensively researched
psychological and biological individual differences affect constructs in personality psychology, has been linked to
these processes [46]. Several lines of evidence show that both normal and abnormal reward processes. Extraversion is
rewards and reward motivation are processed by the nucleus characterized by individual differences in positive emotion,
accumbens, amygdala, and orbitofrontal cortex, an social engagement, and life satisfaction [5,36]. Several
interconnected network of dopaminergic brain regions researchers have hypothesized that these characteristics arise
termed the neural reward system [6,9,20,58,59,68]. Al- from underlying differences in sensitivity to cues of rewards
though individual differences in reward processing have and motivation to obtain future rewards [21,22,25,31,49].
Some researchers have additionally proposed that individual
* Corresponding author. Center for Neuroscience, University of Califor-
differences in the sensitivity of the brain’s reward system
nia, Davis, CA 95616, USA. lead to differences in reward sensitivity, and this in turn may
E-mail address: [email protected] (M.X. Cohen). be a neural mechanism of individual differences in
0926-6410/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.cogbrainres.2005.09.018
852 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861

extraversion [22,50]. In particular, the dopamine D2

receptor (DRD2) has been heavily implicated in reward
processes [7,8,56,57], is found in highest concentrations in
the reward system [57], and thus the functioning of this
system may be related to extraversion [22,23]. Having the
A1 minor allele on the Taq1A DRD2 gene site is associated
with fewer DRD2 receptors [38,56,65,71], and clinical
research has linked this polymorphism with addictive
disorders, reward insensitivity, alcoholism, and personality
Fig. 1. Depiction of trial events. (a) Study 1. On each trial, participants
[8,35,56]. Although this polymorphism has been associated decided whether to choose a high- or low-risk gamble. Soon after the offset
with reduced levels of D2 receptors in several studies using of the cue, participants learned whether they won money on that trial.
diverse methodologies including post-mortem analyses and Following each trial was a variable 2 – 8 s inter-trial interval (ITI). (b) Study
PET, the precise mechanism by which the A1 allele leads to 2. The design was similar to that of Study 1 except for the inclusion of a 7.5
s anticipation period, which allowed separation of hemodynamic responses
lower levels of this receptor remains debated [4,40]. Given
to reward anticipation from reward evaluation.
that it is located at the 3V untranslated region, this
polymorphism may have regulatory functions during gene
expression. 2 – 8 s) separated trials. Participants indicated each decision
Here, we used functional magnetic resonance imaging by either responding or withholding a response, depending
(fMRI) to examine whether individual differences in on the shape of the cue. For example, a white circle meant
extraversion predict the reactivity of the brain’s dopami- that participants would press the button to indicate a low
nergic neural reward system and whether these differences risk gamble or withhold a response to indicate a high risk
are related to the A1 allele on the DRD2 gene. In 2 gamble, whereas a white square would indicate the
experiments, participants were probabilistically rewarded opposite. The purpose of having participants press a
after choosing uncertain gambles. In the first experiment, button during half of the trials and withhold responses
we examined the relationship between individual differ- during the other half of trials was to counterbalance any
ences in self-reported extraversion and variability in neural effects of response selection and inhibition. The reported
responses to rewards. In the second experiment, we effects did not vary as a function of whether a response
replicated these findings in a new sample of participants was made or withheld, so we collapsed across this factor.
and extended them by (1) experimentally separating two Additional control trials were included in which partici-
processes engaged by the reward system – reward anticipa- pants were 100% likely to receive a reward ($2.50) if they
tion, a state of waiting to learn whether a reward would be responded appropriately to the cue. Appropriate responses
received, and reward evaluation, the indication that a were made on 92% (SEM = 3%) of trials in Study 1 and
reward is received – and (2) testing whether neural 89% (SEM = 2%) of trials in Study 2.
responses to reward were additionally related to the Participants were trained extensively on a behavioral
presence of the DRD2 allele. version of this task prior to the experiment and were fully
informed of the probabilities and outcomes associated with
each cue type and response. This training minimized any
2. Methods learning effects over the course of the experiment. Addi-
tionally, because the two decision options (high vs. low risk)
2.1. Experimental design and protocol yielded the same amount of reward in the long run (the
expected value of each option was $1.00), there were no
2.1.1. Study 1 optimal decision strategies participants could learn. We
Seventeen participants completed 520 trials of our tested for sex differences in all of the reported analyses, and
decision-making task while in the MRI scanner. The task none was significant (all F’s < 1).
involved making rapid behavioral responses in order to
win money. On each trial (see Fig. 1a), participants first 2.1.2. Study 2
saw a visual cue for 400 ms and were required to choose Study 2 was identical to Study 1 with the following
one of two economically equivalent (i.e., equal in expected exceptions. First, the cue remained on screen for 500 instead
value) gambles: a low-risk gamble, in which they were of 400 ms. Second, an anticipatory delay of 7.5 s was added
very likely to win a small reward (80% chance of $1.25 after participants made their decision and before they learned
and 20% chance of $0.00), or a high-risk gamble, in which whether they won money on that trial. Third, a longer inter-
they were less likely to win a large reward (40% chance of trial interval (7 –13 s) was added to allow the BOLD response
$2.50 and 60% chance of $0.00). Three hundred milli- from the evaluation period to return to baseline before the
seconds after the offset of the cue, the amount of money start of the next trial. Fourth, there were no withhold-response
participants won in that trial appeared on the screen (for conditions, and participants instead indicated their decision
example, ‘‘+$2.50’’). A variable inter-trial interval (range: using one of two buttons on a response box.
 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861 853

2.2. Personality questionnaire combination of the participant’s decision and the outcome
on that trial, relative to a resting fixation baseline (inter-
To measure individual differences in extraversion, trial interval). The variable (2 –8 s) duration of the ITI in
participants completed a paper-and-pencil version of the this experiment allowed us to deconvolve overlapping
Big 5 Personality Inventory [36], a validated and commonly BOLD responses associated with each trial type [52]. For
used self-report measure of 5 personality traits, including Study 2, separate covariates were used to model the
extraversion, after the scanning session. This measure of decision, anticipation, and evaluation phases of each type
extraversion has been shown to be both reliable and stable of event, relative to the variable (7 – 13 s) inter-trial
over time [36]. In our own samples, the reliability estimates interval.
were 0.89 in Study 1 and 0.86 in Study 2. All regression models incorporated empirically derived
estimates of intrinsic temporal autocorrelation [1] and filters
2.3. Genetic collection and analyses to attenuate frequencies above .24 Hz and below .01 Hz.
Hemodynamic response functions (HRFs) were empirically
Genetic assays were performed at the Genomics Facility at derived for each participant using BOLD responses in the
the University of California, Davis. DNA samples were central sulcus during a visuomotor response task [2,34].
obtained from participants in Study 2 using the MasterAmp These HRFs were used to model BOLD responses to each
Buccal Swab kit (Epicentre Inc.). DNA was extracted event type in all analyses. In addition, the mean of each
according to the protocol provided by the company. scanning run, the global signal (orthogonalized with respect
Polymerase chain reaction (PCR) was used to amplify the to the design matrix [24]), and an intercept were included as
DNA sequences of interest. PCR primers used to test for the nuisance covariates.
presence of the A1 allele at the DRD2 Taq1A site were Following single-subject analyses, images of parameter
designed based on previous studies [40]. The primer estimates for each contrast of interest were entered into a
sequences are 5V-CCGTCGACGGCTGGCCAAGTTG- one-sample t test, in which the mean estimate across
TCTA-3V and 5V-CCGTCGACCCTTCCTGAGTGTCA- participants for each voxel was tested against zero.
TCA-3V. These primers amplify a 294 base-pair region of Significant regions of activation were identified using an
the DRD2 gene that contains a Taq1 recognition site, and the uncorrected two-tailed threshold of P < 0.001 and a cluster
allele was detected by digesting the PCR products with Taq1 threshold of 6 contiguous voxels. In the figures, activations
restriction enzyme followed by separation on agarose gel are overlaid on a single subject’s T1-weighted image using
electrophoresis. MRIcro software (http://www.mricro.com).
Regions of interest (ROIs) were defined as contiguous,
2.4. fMRI acquisition significantly active voxels that showed greater activity
during trials in which participants were rewarded versus
MRI data were collected on a 1.5 T GE Signa scanner at those in which they were not (Study 1), receiving versus not
the UC Davis Research Imaging Center. Functional imaging receiving a reward during the evaluation phase (Study 2),
was done with a gradient echo EPI sequence (TR = 2000, and reward anticipation greater than activity during the
TE = 40, FOV = 220, 64  64 Matrix, voxel size = 3.475  inter-trial interval (Study 2). Correlations between brain
3.475  5 mm), with each volume consisting of 22 (24 in activity and extraversion and differences between genotype
Study 2) oblique axial slices (tilt angle: ¨ 15- from ac –pc within these ROIs were performed using a two-tailed alpha
line). In pilot studies, we determined that this set of value of .05. In the scatter plots in Figs. 3 and 5, beta values
parameters maximized signal-to-noise ratios and blood (i.e., parameter estimates) were normalized by dividing all
oxygenation level dependent (BOLD) contrast in the betas by the maximum beta value, which linearly scales the
orbitofrontal cortex and amygdala. Co-planar and high- betas to have a maximum value of 1.
resolution T1-weighted images were also acquired from
each participant. EPI data were realigned to the first volume,
corrected for slice-timing differences, co-registered with the 3. Results
anatomical scan, spatially normalized to MNI template
brain, resampled to 3.5 mm isotropic voxels, and spatially 3.1. Study 1
smoothed with an 8 mm FWHM kernel using SPM99
software (Wellcome Department of Imaging Neuroscience, Participants chose the low-risk option more often than
London, UK). the high-risk option (61% T 3 versus 39% T 3, respectively,
mean T SEM, F(16) = 10.7, P < 0.01). Extraversion was not
2.5. Data analyses significantly related to the number of high risk gambles
chosen (r = 0.24, ns), the probability of using a win-stay/
Event-related BOLD responses were modeled using lose-switch strategy (r = 0.26, ns), or the probability of
multiple regression in voxbo software (http://www.voxbo. choosing a high-risk gamble following a high-risk win (r =
org). For Study 1, separate covariates modeled each 0.20, ns). Thus, individuals who differed in their extraver-
854 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861

sion scores did not reliably use different strategies during 3.2. Study 2
the experiment.
We hypothesized that extraversion would predict the In Study 1, we demonstrated a link between extraver-
responsiveness to rewards within reward-sensitive regions. sion and reactivity of the reward system to receiving
To test this hypothesis, we first identified regions in the rewards. However, there are at least two neural processes
brain that were more active on trials when participants were linked with rewards: anticipation and evaluation. By
rewarded than on trials when they were not. This contrast anticipation, we refer to the time period after participants
yielded significant bilateral activations in medial orbito- made a behavioral response but before they learned
frontal cortex, amygdala, and nucleus accumbens (see Fig. whether they would be rewarded on that trial; and by
2). Each of these regions has been consistently implicated in evaluation, we refer to the time when participants learned
reward processing in neurophysiological studies of rodents whether they received a reward on that trial (see Fig. 1b).
and monkeys [29,68,72,78] and neuroimaging studies of Structures within the neural reward system, such as
humans [3,9,28,58]. We defined each of these activations as orbitofrontal cortex and striatum, have been identified in
regions of interest (ROIs) for further investigation. Outside both processes [7,44], but it is not known whether
of these ROIs, activations were observed in areas of the extraversion scores would relate to neural activity during
cerebellum and fusiform gyrus, which may reflect the both of these processes. Although there was a brief
increased engagement of attentional processes on trials anticipatory period in Study 1, the rapid event-related
associated with rewards [67]. Coordinates of these and all design did not allow us to differentiate between neural
other activations reported in this paper are presented in activity related to anticipation from that related to
Table 1. evaluation. Thus, Study 2 was designed to separate these
We next investigated whether individual differences in processes and examine whether and how extraversion was
extraversion predicted the magnitude of the reward re- related to neural responses during both of these processes.
sponse, defined as the difference in activation between This was accomplished by adding a 7.5 s delay after the
receiving and not receiving a reward, in each of these ROIs. participants’ decision and before they learned the outcome
We found that participants higher in extraversion exhibited a on each trial (see Fig. 1b). This design allowed us to
significantly greater reward response in 3 out of 6 of these distinguish activity related to anticipating versus experi-
ROIs (left medial orbitofrontal cortex: r = 0.53, P = 0.028; encing the outcome on each trial [18,26,42,62,63]. We
right medial orbitofrontal cortex: r = 0.59, P = 0.012; right additionally collected DNA samples from participants in
nucleus accumbens: r = 0.61, P = 0.01; see Fig. 3). Fig. 3 Study 2 to be analyzed for the presence of the Taq1A
also displays the reward > no reward contrast for two polymorphism on the DRD2 gene.
individual subjects with relatively high and low extraversion Sixteen different participants (9 male, aged 20 –27) were
scores, both thresholded at P > 0.001, two-tailed. included in this study. As with Study 1, participants were
Because participants received different magnitudes of not selected on the basis of personality or genetics. The
rewards on different trials ($1.25 versus $2.50), we further high/low risk decision and probabilities of rewards were the
examined whether the difference between receiving a high same as in Study 1.
reward and a low reward changed as a function of
extraversion and found no significant correlations in any 3.2.1. Extraversion
of our ROIs (all r’s < 0.26). We finally tested whether As in Study 1, extraversion was not significantly related
extraversion scores predicted reward-related brain activity in to the number of high-risk gambles chosen (r = 0.21,
regions outside of the reward system and found no ns), the probability of using a win-stay/lose-switch strategy
significant correlations in these regions (all r’s > 0.28, all (r = 0.08), or the probability of choosing a high-risk
P’s > 0.26). gamble following a high-risk win (r = 0.34, ns). Thus, as

Fig. 2. Activation of the neural reward network in Study 1. Bilateral regions in the medial orbitofrontal cortex, amygdala, and nucleus accumbens exhibited
more activity during trials in which participants were rewarded than during trials in which they were not.
 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861 855

Table 1 prefrontal cortex (left BA 45 and right BA 6/44), and

Region BA t x y z parietal cortex (BA 40). Extraversion scores did not
Study 1: reward – no reward correlate with activity in any of these regions, regardless
R. orbitofrontal cortex 11 5.45 21 39 21 of the decision chosen (all r’s < 0.21, ns).
L. orbitofrontal cortex 11 5.07 10 42 24 We next identified regions in the brain that were more
R. amygdala 4.77 27 1 21 active when participants received a reward than when they
L. amygdala 5.38 23 0 16
R. nucleus accumbens 8.12 14 14 10 did not (during the evaluation phase). Consistent with
L. nucleus accumbens 6.94 21 14 10 findings from Study 1, we observed significantly greater
R. hippocampus 5.12 28 35 3 activity in bilateral medial orbitofrontal cortex, right
R. hippocampus 4.85 39 21 14 amygdala, bilateral nucleus accumbens, and left hippocam-
L. fusiform gyrus 20 6.27 38 24 17 pus (see Fig. 4) when participants were rewarded compared
R. calcarine sulcus 17 5.41 7 87 0
L. cerebellum 5.25 28 45 28 with when they were not rewarded. Participants higher in
R. cerebellum 5.21 14 45 14 extraversion exhibited a significantly greater reward re-
R. supplementary motor 6 4.75 25 10 67 sponse (the difference in activation between receiving and
not receiving a reward) in 5 out of these 6 ROIs (left medial
Study 2: anticipation
orbitofrontal cortex: r = 0.60, P = 0.013; right medial
R. parietal 40 10.23 56 38 49
L. dorsolateral PFC 45 7.69 52 35 14
orbitofrontal cortex: r = 0.58, P = 0.018; right amygdala:
L. posterior temporal 22 6.09 59 52 21 r = 0.52, P = 0.038; right nucleus accumbens: r = 0.56, P =
R. insula 48 5.95 49 14 7 0.025; left nucleus accumbens: r = 0.61, P = 0.011; see Fig.
L. temporal/insula 21/48 5.88 45 7 10 5a). Thus, about 33% of inter-subject variability of reward
R. parietal 40 5.84 60 45 28 responses in these ROIs was explained by individual
R. thalamus 5.12 7 10 11
R. ventrolateral PFC 6 5.02 53 7 11 differences in extraversion. Finally, as in Study 1, we tested
R. cerebellum 4.81 49 49 35 whether the difference in activation between high and low
L. temporal pole 38/21 4.70 56 7 14 rewards was correlated with extraversion. We found that this
L. thalamus 4.69 10 10 14 correlation was significant in left medial orbitofrontal cortex
R. orbitofrontal cortex 11 4.54 25 49 7 (r = 0.53, P = 0.03) and marginal in the right nucleus
L. caudate 4.45 17 21 21
L. superior temporal gyrus 21 4.29 63 0 7 accumbens (r = 0.47, P = 0.06), indicating that, in some
regions, not only did individuals higher in extraversion
Study 2: reward – no reward exhibit increased activity when receiving a reward, they also
R. amygdala 6.58 18 7 28 exhibited a larger difference in activation between relatively
L. amygdala 6.47 17 3 14
larger compared to smaller rewards.
L. hippocampus 5.40 28 24 14
L. orbitofrontal cortex 11 5.31 21 21 25
R. precuneus 23 5.21 4 56 35 3.2.2. DRD2
R. orbitofrontal cortex 11 5.11 25 25 21 Nine out of 16 participants had one or two copies of the
R. posterior insula 4.79 39 21 14 A1 allele. The presence of this allele was not statistically
R. nucleus accumbens 4.60 11 0 14 related to extraversion scores (t = 1.48, ns) and was not
R. nucleus accumbens 4.52 14 11 14
R. postcentral gyrus 4 4.44 53 17 46 related to race (66% Caucasian [34% Asian/other] with A1
L. hippocampus 4.23 31 14 21 allele, 57% Caucasian without A1 allele; t = 0.13, ns).
L. nucleus accumbens 4.30 7 4 8 Participants with the A1 allele did not take significantly
Brodmann (BA), MNI coordinates (x, y, z), and t values of regions active in more high-risk gambles than participants without the
each contrast. L = left; R = right. polymorphism (t = 0.9, ns), utilize a win-stay/lose-shift
strategy to a different extent (t = 0.1, ns), or choose a high-
in Study 1, individuals who differed in their extraversion risk gamble following a high-risk win more often (t = 0.3,
scores did not reliably use different strategies during the ns). Thus, the presence of this allele did not affect the kinds
experiment. of behavioral strategies that participants used during the
As in Study 1, we first identified regions responding to task.
reward-related processes and then examined whether To test whether the presence of the DRD2 A1 allele
individual differences in neural responses to rewards were affected neural responses to rewards, we compared
correlated with extraversion scores. In this experiment, activity in each functionally defined ROI identified above
however, we were able to separately identify regions between participants who had versus did not have this
responding to reward anticipation and reward evaluation. allele. We first examined responses during anticipation
We first identified regions of the brain that were and found that the two groups did not differ in activation
significantly active during the anticipation phase of the magnitudes during anticipation in any of the regions that
trials. We observed significant activations in dorsal caudate showed increased activity during anticipation, regardless of
nucleus, right ventral prefrontal cortex (BA 11/10), insula/ whether participants anticipated a high-risk or low-risk
ventral prefrontal cortex (BA 48) (see Fig. 4), bilateral reward (all F’s < 4, ns).
856 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861

Fig. 3. Extraversion scores significantly predicted individual differences in the magnitude of BOLD responses to rewards in several of the regions identified in
the reward contrast in Study 1 (see Fig. 2). Extraversion scores (higher scores indicate greater extraversion) are plotted on the x axis, and the maximum value
normalized activation magnitudes from the reward > no reward contrast are plotted on the y axis. Activation maps of the reward > no reward contrast are
displayed for two individual subjects at y = 20, along with their extraversion scores.

Next, we examined whether the groups differed in acti- low-risk wins differed according to whether participants
vation magnitudes when receiving rewards. A 2 (group)  6 had or did not have the A1 allele and found that they did
(region) ANOVA revealed that that the reward response was not (all F’s < 2, ns). Thus, the presence of the A1 allele on
significantly less in participants who had the allele (F 1,14 = the DRD2 gene has a significant effect on our participants’
6.41, P = 0.024). As shown in Fig. 5b, this difference was neural responses to reward processing.
significant in bilateral medial orbitofrontal cortex (left:
F 1,14 = 5.99, P = 0.028; right: F 1,14 = 5.49, P = 0.034),
right amygdala (F 1,14 = 7.11, P = 0.018), left hippocampus 4. Discussion
(F 1,14 = 6.60, P = 0.022), and marginal in right nucleus
accumbens (F 1,14 = 4.12, P = 0.062). Finally, we tested In this study, we found that individual differences in
whether the difference in activation between high-risk and extraversion and the presence of the DRD2 A1 allele

Fig. 4. Activation related to reward anticipation and reward evaluation in Study 2. (a) Regions in the striatum, insula, and prefrontal cortex were active during
the anticipation period prior to receiving a reward. (b) Regions in the orbitofrontal cortex, amygdala, and nucleus accumbens showed greater activity when
receiving rewards than when not receiving rewards (evaluation period).
 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861 857

Fig. 5. Extraversion and DRD2 variability predict neural reward responses during the evaluation phase of Study 2. (a) Scatter plots depict the relationship
between extraversion scores and reward responses in regions showing greater activity in the reward > no reward contrast during the evaluation phase (Fig. 4).
Conventions are as in Fig. 3. (b) Bar graphs show reward responses in subjects with (left) and without (right) the DRD2 gene polymorphism. Error bars
represent one standard error of the mean across subjects.

predicted the responsiveness to rewards in orbitofrontal Prior to the present report, a few neuroimaging studies
cortex, amygdala, and nucleus accumbens. Importantly, investigated how differences in extraversion relate to brain
decision-making strategies did not vary as a function of activation. For example, Canli and colleagues [12,13]
extraversion or the presence of the DRD2 allele, suggesting correlated brain responses during passive picture-viewing
that our observed correlations were not driven by differ- tasks with extraversion scores. Consistent with our results,
ences in task performance. We note that, although we did these researchers found that extraversion was associated
not collect on-line behavioral indices of subjective enjoy- with increased activity in the amygdala, among other
ment of the reward, results from previous neuroimaging regions, when viewing positively valenced pictures. In
studies have shown that reward-related activations are another study by Johnson and colleagues [37], positron
correlated with subjective ratings of enjoyment of the emission tomography was used to correlate extraversion
reward [3,43]. Accordingly, the association between neural scores with measures of glucose metabolism when
reward responsiveness and extraversion and the DRD2 A1 participants were resting and not performing any tasks.
allele in the present study was likely related to individual Their results showed that extraverts had higher resting
differences in subjective reward enjoyment. levels of metabolism in the amygdala, among other
regions, than did introverts. Our findings, and the findings
4.1. Extraversion of the previously described fMRI studies [12,13], seem
inconsistent with this result because extraversion was only
Our finding that individual differences in extraversion related to activity in the brain during events that were
predicted variability in the reactivity of the reward system likely interpreted by the participants as being rewarding or
when receiving monetary rewards supports a growing pleasurable. However, it is important to note that fMRI
literature that highlights the importance of reward sensitivity does not measure absolute levels of brain activity, only
to extraversion. Diener and colleagues have argued that the relative levels. Thus, it is possible that extraverts might
core feature of extraversion is reward sensitivity and that have relatively more tonic activity than introverts as well
other facets of extraversion, such as social engagement and as enhanced phasic responses to rewarding events. This
subjective well-being, are by-products of this sensitivity remains an open question that could be addressed in future
[25]. Depue and Collins [22] provided a neurobiological studies.
framework for this hypothesis by suggesting that this reward We additionally observed that the relationship between
sensitivity is mediated by differences in the efficacy of extraversion and brain activity was specific to reward
dopaminergic inputs from the brainstem to regions including evaluation and was not observed during anticipation. It is
the nucleus accumbens, orbitofrontal cortex, and amygdala. important to note that reward anticipation has been
Our results offer direct support for this theory by demon- operationalized differently in different studies. Here, as in
strating that individuals who scored relatively high on some other studies [18], anticipation was operationalized
measures of extraversion exhibited increased activation in as a state of waiting to learn whether a reward will be
reward-sensitive regions when receiving rewards. received, whereas other researchers have operationalized
858 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861

reward anticipation as a state that produces motivation to individuals with this allele are more likely to develop
engage in reward-seeking behaviors. For example, antici- addictive [45,55,75] or reward deficiency disorders [8,56].
pating a drug-induced reward motivates addicted individ- Variation on this allele may also relate to differences in
uals to engage in drug-seeking behaviors [7]. It is thus subjective enjoyment from rewards, which has been shown
possible that extraversion is related to brain activity during to correlate with activity in regions including the ventral
reward anticipation if that anticipatory state could induce striatum [3,43]. Future studies could help elucidate this by
behaviors that increase the probability of receiving a having subjects explicitly report subjective pleasure
reward [22]. This or other methodological differences obtained from different sized rewards.
(e.g., smoothing kernels, scan parameters, etc.) might also Given that DRD2 is found in highest concentrations in
explain differences in the localization of anticipation- the ventral striatum, it is surprising that the effects of the
related activity across studies. For example, we observed allele on reward-related activity in nucleus accumbens
anticipation-related activations in the dorsal striatum, were not robust. However, previous reports of DRD2
whereas others have reported anticipation-related activity concentration and addiction have reported robust findings
in other regions such as the ventral striatum [43,44] or the in orbitofrontal cortex [76], and it is possible that
amygdala [10,39]. orbitofrontal functioning in this task is more sensitive to
The present report focused on extraversion, but a wide this allele’s phenotypic expression than is functioning of
range of personality measures have been used to investigate the nucleus accumbens. It is also important to note that the
reward sensitivity and reward motivation. For example, presence or absence of the DRD2 Taq1A allele is only one
traits such as reward dependency, the behavioral activation of many variables that factor into the functioning of the
system (BAS), and positive emotionality are correlated with dopamine system, and thus the nonsignificant differences
extraversion and have been linked behaviorally to reward- in nucleus accumbens between groups do not imply that
related processes [15,22,36]. Although the terminology and dopaminergic neurons in this region are not involved in the
precise definitions of these traits differ, they likely measure task.
somewhat overlapping constructs [22,36,77]. Although it is The fact that differences were not observed during
unknown whether the correlations observed here would anticipation may also be related to the fact that, in our
generalize to other similar personality measures, the present study, anticipation could not lead to increased probability of
results, along with those from other groups [12,32,33,61], rewards, as discussed above. Indeed, the DRD2 system is
suggest that individual differences in personality are related heavily involved in reward anticipation and reward predic-
to meaningful individual differences in neural responses tion [29,70,73], but it is possible that the presence of the
during emotional and cognitive processes. Accordingly, DRD2 Taq1A allele only affects experiential or consuma-
functional neuroimaging might be a promising source of tory reward processes, and not all reward processes. This
evidence for contrasting competing models of fundamental pattern of results also argues against an alternative
personality dimensions. explanation that the differences in reward activation are
simply due to the polymorphism affecting the BOLD
4.2. DRD2 response.
In our sample, the presence of the DRD2 A1 allele was
Over half (9 of 16) of our sample had at least one copy not significantly related to extraversion scores, although
of the A1 allele on the DRD2 gene, a proportion that is there was a weak trend for individuals with the A1 allele
slightly higher than reported in previous studies [56]. to have lower extraversion scores. One might actually
Participants were self-selected from a university popula- predict the opposite association because both having the
tion and were not prescreened for their genetic make-up or A1 allele and scoring high on measures of extraversion
any behavioral factors that might be related to genetics. have been associated with reward-seeking behaviors,
fMRI data showed that participants with this allele although two variables being related to a third does not
exhibited relatively less differentiation in activation necessitate that they be related to each other. Although a
between receiving and not receiving rewards in orbito- link between the A1 allele and extraversion has been
frontal cortex, amygdala, and, to a lesser extent, nucleus reported previously [48,60], other studies have failed to
accumbens. This allele did not affect activation magni- find such an association [41], even with large sample sizes.
tudes during anticipation. Noble has suggested that the relationship between these
Although clinical research has linked the DRD2 A1 two variables is moderated by environmental factors, such
allele with alcoholism and other addictive disorders that emotional stress during childhood can change the
[19,35,54], ours is the first study to investigate how the direction of the relationship between the presence of the
presence of this allele affects brain activity in a nonclinical allele and extraversion scores [60]. Thus, a link between
population when receiving rewards. Our findings support this allele and extraversion might be subtle and moderated
the idea that a reduced concentration of DRD2 in the by environmental or other genetic factors. It is important to
reward system leads to reduced sensitivity to rewards [73 – note that, as previously mentioned, the DRD2 Taq1A gene
75] and that this may help form the basis of why site is only one variable that affects the functioning of the
 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861 859

dopamine system. Thus, it is possible that there is a strong References

link between extraversion and dopaminergic neurotrans-
[1] G.K. Aguirre, E. Zarahn, M. D’Esposito, Empirical analyses of BOLD
mission but that it is weak or not apparent when
fMRI statistics: II. Spatially smoothed data collected under null-
examining this particular allele. hypothesis and experimental conditions, NeuroImage 5 (3) (1997)
199 – 212.
4.3. Caveats and conclusions [2] G.K. Aguirre, E. Zarahn, M. D’Esposito, The variability of human,
BOLD hemodynamic responses, NeuroImage 8 (4) (1998) 360 – 369.
[3] I. Aharon, et al., Beautiful faces have variable reward value: fMRI and
There are a few caveats about our findings that we
behavioral evidence, Neuron 32 (3) (2001) 537 – 551.
emphasize. First, we only investigated our pre-determined [4] T. Arinami, et al., A functional polymorphism in the promoter region
ROIs for effects related to DRD2 genotype or extraversion. of the dopamine D2 receptor gene is associated with schizophrenia,
It is therefore possible that activity in other regions of the Hum. Mol. Genet. 6 (4) (1997) 577 – 582.
brain may also be modulated by DRD2 genetic variability [5] M.C. Ashton, K. Lee, S.V. Paunonen, What is the central feature of
extraversion? Social attention versus reward sensitivity, J. Pers. Soc.
or personality. However, focused ROI analyses are
Psychol. 83 (1) (2002) 245 – 252.
commonly used [53,64], and we chose to limit our ROI [6] M.G. Baxter, E.A. Murray, The amygdala and reward, Nat. Rev.,
analyses to regions identified as exhibiting enhanced Neurosci. 3 (7) (2002) 563 – 573.
responses to monetary rewards. Second, studies linking [7] K.C. Berridge, T.E. Robinson, Parsing reward, Trends Neurosci. 26
genetic mutations to behavior or brain activity must always (9) (2003) 507 – 513.
[8] K. Blum, et al., The D2 dopamine receptor gene as a determinant
be cautious of spurious effects being driven by population
of reward deficiency syndrome, J. R. Soc. Med. 89 (7) (1996)
differences that might be related to genetic polymorphisms 396 – 400.
(e.g., ethnic groups). We observed no differences in race [9] H.C. Breiter, B.R. Rosen, Functional magnetic resonance imaging of
(classified as Caucasian vs. Asian/other) between our brain reward circuitry in the human, Ann. N. Y. Acad. Sci. 877 (1999)
subjects with and without the A1 allele, and other studies 523 – 547.
[10] H.C. Breiter, et al., Functional imaging of neural responses to
have demonstrated that ethnic differences are not a likely
expectancy and experience of monetary gains and losses, Neuron 30
confounding factor with DRD2 [27,57]. Nonetheless, it is (2) (2001) 619 – 639.
possible that there were other differences between our [11] P. Brieger, et al., The relationship between five-factor personality
groups that we did not measure or that our effects were measurements and ICD-10 personality disorder dimensions: results
due to other functional mutations, perhaps in the promoter from a sample of 229 subjects, J. Pers. Disord. 14 (3) (2000)
282 – 290.
region of the gene, that covary with the presence of the D2
[12] T. Canli, et al., An fMRI study of personality influences on brain
A1 allele [40]. reactivity to emotional stimuli, Behav. Neurosci. 115 (1) (2001)
Our findings demonstrate clear links between stable 33 – 42.
individual differences in personality, genetics, and func- [13] T. Canli, et al., Amygdala response to happy faces as a function of
tioning of the brain’s reward system. This anatomically extraversion, Science 296 (5576) (2002) 2191.
[14] B.J. Carroll, Brain mechanisms in manic depression, Clin. Chem. 40
and functionally interconnected network [16] is critical for
(2) (1994) 303 – 308.
both biological and social survival, and our findings [15] C.R. Cloninger, R. Przybeck, D. Svrakic, The tridimensional
suggest that variation in the functioning of this system is personality questionnaire: U.S. normative data, Psychol. Rep. 69
linked with socially and emotionally relevant conditions. (1991) 1047 – 1057.
For example, both low extraversion scores and the DRD2 [16] M.X. Cohen, A.S. Heller, C. Ranganath, Functional connectiv-
ity with anterior cingulate and orbitofrontal cortices during
A1 allele have been linked to psychiatric disorders
decision-making, Brain Res. Cogn. Brain Res. 23 (1) (2005)
including depression [47,51], drug dependency [11,66, 61 – 70.
76], and compulsive gambling [17,69]. Thus, an important [17] D.E. Comings, et al., A study of the dopamine D2 receptor
question for future research is whether the relationships gene in pathological gambling, Pharmacogenetics 6 (3) (1996)
between individual differences in personality, genetics, and 223 – 234.
[18] H.D. Critchley, C.J. Mathias, R.J. Dolan, Neural activity in the human
brain activation observed here might relate to clinical
brain relating to uncertainty and arousal during anticipation, Neuron
disorders and whether neuroimaging could be used to 29 (2) (2001) 537 – 545.
guide new approaches for diagnosis and treatment of such [19] C. Cruz, et al., The dopamine D2 receptor gene TaqI A1 polymor-
conditions. phism and alcoholism in a Mexican population, Arch. Med. Res. 26
(4) (1995) 421 – 426.
[20] R.J. Davidson, W. Irwin, The functional neuroanatomy of emotion and
affective style, Trends Cogn. Sci. 3 (1) (1999) 11 – 21.
Acknowledgments [21] K.M. DeNeve, H. Cooper, The happy personality: a meta-analysis of
137 personality traits and subjective well-being, Psychol. Bull. 124 (2)
This work was supported by an Extramural Research (1998) 197 – 229.
Grant from the Institute for Research on Pathological [22] R.A. Depue, P.F. Collins, Neurobiology of the structure of personality:
dopamine, facilitation of incentive motivation, and extraversion,
Gambling and Related Disorders and a faculty research
Behav. Brain Sci. 22 (3) (1999) 491 – 517 (discussion 518-69).
grant from the University of California at Davis. The [23] R.A. Depue, et al., Dopamine and the structure of personality: relation
authors thank Tonya Jacobs and Eveline Crone for their of agonist-induced dopamine activity to positive emotionality, J. Pers.
help. Soc. Psychol. 67 (3) (1994) 485 – 498.
860 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861

[24] A.E. Desjardins, K.A. Kiehl, P.F. Liddle, Removal of confounding [48] H.J. Lee, et al., D2 and D4 dopamine receptor gene polymorphisms
effects of global signal in functional MRI analyses, NeuroImage 13 (4) and personality traits in a young Korean population, Am. J. Med.
(2001) 751 – 758. Genet. 121B (1) (2003) 44 – 49.
[25] E. Diener, S. Oishi, R.E. Lucas, Personality, culture, and subjective [49] R.E. Lucas, E. Diener, Understanding extraverts’ enjoyment of social
well-being: emotional and cognitive evaluations of life, Annu. Rev. situations: the importance of pleasantness, J. Pers. Soc. Psychol. 81 (2)
Psychol. 54 (2003) 403 – 425. (2001) 343 – 356.
[26] T.J. Druzgal, M. D’Esposito, Dissecting contributions of prefrontal [50] R.E. Lucas, et al., Cross-cultural evidence for the fundamental features
cortex and fusiform face area to face working memory, J. Cogn. of extraversion, J. Pers. Soc. Psychol. 79 (3) (2000) 452 – 468.
Neurosci. 15 (6) (2003) 771 – 784. [51] J.H. Meyer, et al., Lower dopamine transporter binding potential
[27] M.F. Egan, et al., The BDNF val66met polymorphism affects activity- in striatum during depression, NeuroReport 12 (18) (2001)
dependent secretion of BDNF and human memory and hippocampal 4121 – 4125.
function, Cell 112 (2003) 257 – 269. [52] F.M. Miezin, et al., Characterizing the hemodynamic response: effects
[28] R. Elliott, K.J. Friston, R.J. Dolan, Dissociable neural responses in of presentation rate, sampling procedure, and the possibility of
human reward systems, J. Neurosci. 20 (16) (2000) 6159 – 6165. ordering brain activity based on relative timing, NeuroImage 11 (6
[29] C.D. Fiorillo, P.N. Tobler, W. Schultz, Discrete coding of reward Pt. 1) (2000) 735 – 759.
probability and uncertainty by dopamine neurons, Science 299 (5614) [53] A. Nieto-Castanon, et al., Region of interest based analysis of
(2003) 1898 – 1902. functional imaging data, NeuroImage 19 (4) (2003) 1303 – 1316.
[30] A.E. Goudriaan, et al., Pathological gambling: a comprehensive [54] E.P. Noble, Polymorphisms of the D2 dopamine receptor gene and
review of biobehavioral findings, Neurosci. Biobehav. Rev. 28 (2) alcoholism and other substance use disorders, Alcohol Alcohol.,
(2004) 123 – 141. Suppl. 2 (1994) 35 – 43.
[31] J.A. Gray, The psychophysiological basis of introversion – extraver- [55] E.P. Noble, Addiction and its reward process through polymorphisms
sion, Behav. Res. Ther. 8 (3) (1970) 249 – 266. of the D2 dopamine receptor gene: a review, Eur. Psychiatry 15 (2)
[32] J.R. Gray, T.S. Braver, Personality predicts working-memory-related (2000) 79 – 89.
activation in the caudal anterior cingulate cortex, Cogn. Affect. Behav. [56] E.P. Noble, D2 dopamine receptor gene in psychiatric and neurologic
Neurosci. 2 (1) (2002) 64 – 75. disorders and its phenotypes, Am. J. Med. Genet. 116B (1) (2003)
[33] D.A. Gusnard, et al., Persistence and brain circuitry, Proc. Natl. Acad. 103 – 125.
Sci. U. S. A. 100 (6) (2003) 3479 – 3484. [57] E.P. Noble, et al., Addiction and its reward process through poly-
[34] D.A. Handwerker, J.M. Ollinger, M. D’Esposito, Variation of morphisms of the D2 dopamine receptor gene: a review, Eur.
BOLD hemodynamic responses across subjects and brain regions Psychiatry 15 (2) (2000) 79 – 89.
and their effects on statistical analyses, NeuroImage 21 (4) (2004) [58] J. O’Doherty, et al., Abstract reward and punishment representa-
1639 – 1651. tions in the human orbitofrontal cortex, Nat. Neurosci. 4 (1) (2001)
[35] S. Higuchi, et al., Association of structural polymorphism of the 95 – 102.
dopamine D2 receptor gene and alcoholism, Biochem. Biophys. Res. [59] J. O’Doherty, et al., Dissociating valence of outcome from behavioral
Commun. 204 (3) (1994) 1199 – 1205. control in human orbital and ventral prefrontal cortices, J. Neurosci. 23
[36] O.P. John, S. Srivastava, The big five trait taxonomy: history, (21) (2003) 7931 – 7939.
measurement, and theoretical perspectives, in: L.A. Pervin, O.P. John [60] T. Ozkaragoz, E.P. Noble, Extraversion. Interaction between D2
(Eds.), Handbook of Personality: Theory and Research, Guilford dopamine receptor polymorphisms and parental alcoholism, Alcohol
Press, New York, 1999. 22 (3) (2000) 139 – 146.
[37] D.L. Johnson, et al., Cerebral blood flow and personality: a [61] M.P. Paulus, et al., Increased activation in the right insula during risk-
positron emission tomography study, Am. J. Psychiatry 156 (2) taking decision making is related to harm avoidance and neuroticism,
(1999) 252 – 257. NeuroImage 19 (4) (2003) 1439 – 1448.
[38] E.G. Jonsson, et al., Polymorphisms in the dopamine D2 receptor gene [62] C. Ranganath, M. D’Esposito, Medial temporal lobe activity associ-
and their relationships to striatal dopamine receptor density of healthy ated with active maintenance of novel information, Neuron 31 (5)
volunteers, Mol. Psychiatry 4 (3) (1999) 290 – 296. (2001) 865 – 873.
[39] I. Kahn, et al., The role of the amygdala in signaling prospective [63] C. Ranganath, et al., Inferior temporal, prefrontal, and hippocampal
outcome of choice, Neuron 33 (6) (2002) 983 – 994. contributions to visual working memory maintenance and associative
[40] R. Kaiser, et al., Relationship between adverse effects of antipsychotic memory retrieval, J. Neurosci. 24 (16) (2004) 3917 – 3925.
treatment and dopamine D(2) receptor polymorphisms in patients with [64] C. Ranganath, et al., Inferior temporal, prefrontal, and hippocampal
schizophrenia, Mol. Psychiatry 7 (7) (2002) 695 – 705. contributions to visual working memory maintenance and associative
[41] S. Katsuragi, et al., Lack of association between a polymorphism in memory retrieval, J. Neurosci. 24 (16) (2004) 3917 – 3925.
the promoter region of the dopamine D2 receptor and personality [65] T. Ritchie, E.P. Noble, Association of seven polymorphisms of the D2
traits, Psychiatry Res. 105 (1 – 2) (2001) 123 – 127. dopamine receptor gene with brain receptor-binding characteristics,
[42] P. Kirsch, et al., Anticipation of reward in a nonaversive differential Neurochem. Res. 28 (1) (2003) 73 – 82.
conditioning paradigm and the brain reward system: an event-related [66] B. Rodgers, Addiction 95 (2000) 1833.
fMRI study, NeuroImage 20 (2) (2003) 1086 – 1095. [67] M.R. Roesch, C.R. Olson, Impact of expected reward on neuronal
[43] B. Knutson, et al., Anticipation of increasing monetary reward activity in prefrontal cortex, frontal and supplementary eye fields and
selectively recruits nucleus accumbens, J. Neurosci. 21 (16) (2001) premotor cortex, J. Neurophysiol. 90 (3) (2003) 1766 – 1789.
RC159. [68] E.T. Rolls, The orbitofrontal cortex and reward, Cereb. Cortex 10 (3)
[44] B. Knutson, et al., Dissociation of reward anticipation and outcome (2000) 284 – 294.
with event-related fMRI, NeuroReport 12 (17) (2001) 3683 – 3687. [69] A. Roy, J. De Jong, M. Linnoila, Extraversion in pathological
[45] G.F. Koob, M. Le Moal, Drug abuse: hedonic homeostatic dysregu- gamblers. Correlates with indexes of noradrenergic function, Arch.
lation, Science 278 (5335) (1997) 52 – 58. Gen. Psychiatry 46 (8) (1989) 679 – 681.
[46] S.M. Kosslyn, et al., Bridging psychology and biology. The analysis [70] W. Schultz, et al., Neuronal activity in monkey ventral striatum
of individuals in groups, Am. Psychol. 57 (5) (2002) 341 – 351. related to the expectation of reward, J. Neurosci. 12 (12) (1992)
[47] R. Larisch, et al., In vivo evidence for the involvement of dopamine- 4595 – 4610.
D2 receptors in striatum and anterior cingulate gyrus in major [71] J. Thompson, et al., D2 dopamine receptor gene (DRD2) Taq1 A
depression, NeuroImage 5 (4 Pt. 1) (1997) 251 – 260. polymorphism: reduced dopamine D2 receptor binding in the human
 M.X. Cohen et al. / Cognitive Brain Research 25 (2005) 851 – 861 861

striatum associated with the A1 allele, Pharmacogenetics 7 (6) (1997) circuits in drug addiction: insight from imaging studies, Neurobiol.
479 – 484. Learn. Mem. 78 (3) (2002) 610 – 624.
[72] L. Tremblay, W. Schultz, Relative reward preference in primate [76] N.D. Volkow, et al., Dopamine in drug abuse and addiction: results
orbitofrontal cortex, Nature 398 (6729) (1999) 704 – 708. from imaging studies and treatment implications, Mol. Psychiatry
[73] N. Volkow, Drug dependence and addiction: III. Expectation and brain (2004) 557 – 569.
function in drug abuse, Am. J. Psychiatry 161 (4) (2004) 621. [77] C. Watson, L. Clark, Extraversion and its positive emotional core, in:
[74] N.D. Volkow, et al., Expectation enhances the regional brain metabolic S. Briggs, W. Jones, R. Hogan (Eds.), Handbook of Personality
and the reinforcing effects of stimulants in cocaine abusers, Psychology, Academic Press, 1996.
J. Neurosci. 23 (36) (2003) 11461 – 11468. [78] D.J. Woodward, et al., Mesolimbic neuronal activity across behavioral
[75] N.D. Volkow, et al., Role of dopamine, the frontal cortex and memory states, Ann. N. Y. Acad. Sci. 877 (1999) 91 – 112.