PS YC HOLOGICA L SC IENCE

Research Article

Genetic Gating of Human Fear

Learning and Extinction
Possible Implications for Gene-Environment Interaction in
Anxiety Disorder
Tina B. Lonsdorf,1,2 Almut I. Weike,3 Pernilla Nikamo,4 Martin Schalling,4 Alfons O. Hamm,3,5 and
hman1,2,5
Arne O
Department of Clinical Neuroscience, Psychology Section, Karolinska Institutet; 2Stockholm Brain Institute, Stockholm,
Sweden; 3Department of Clinical and Biological Psychology, Ernst Moritz-Arndt University of Greifswald; 4Department of
Molecular Medicine and Surgery, Neurogenetics Unit, Karolinska Institutet; and 5Center for the Study of Emotion and
Attention, University of Florida

ABSTRACTPavlovian

fear conditioning is a widely used

model of the acquisition and extinction of fear. Neural
findings suggest that the amygdala is the core structure for
fear acquisition, whereas prefrontal cortical areas are
given pivotal roles in fear extinction. Forty-eight volunteers participated in a fear-conditioning experiment,
which used fear potentiation of the startle reflex as the
primary measure to investigate the effect of two genetic
polymorphisms (5-HTTLPR and COMTval158met) on
conditioning and extinction of fear. The 5-HTTLPR
polymorphism, located in the serotonin transporter gene,
is associated with amygdala reactivity and neuroticism,
whereas the COMTval158met polymorphism, which is
located in the gene coding for catechol-O-methyltransferase (COMT), a dopamine-degrading enzyme, affects
prefrontal executive functions. Our results show that only
carriers of the 5-HTTLPR s allele exhibited conditioned
startle potentiation, whereas carriers of the COMT met/
met genotype failed to extinguish conditioned fear. These
results may have interesting implications for understanding gene-environment interactions in the development and
treatment of anxiety disorders.
Pavlovian conditioning is the most basic of all paradigms for the
study of associative learning. It provided a substantial portion of

hman or Tina B. Lonsdorf, SecAddress correspondence to Arne O

tion of Psychology, Department of Clinical Neuroscience, Karolinska
Institutet, SE-171 77 Stockholm, Sweden, e-mail: arne.ohman@ki.se
or tina.lonsdorf@ki.se.

Volume ]]]Number ]]

the empirical basis for the classical theories of learning, which

formed the core of behaviorist psychology. The famous early
work on fear conditioning in Little Albert (Watson & Raynor,
1920) pioneered the analysis of fear and anxiety in terms of
learning, an enterprise that is still vigorously pursued (e.g.,
Lissek et al., 2005; Mineka & Zinbarg, 2006; Ohman & Mineka,
2001). Despite the iconic status of Pavlovian conditioning as an
epitome of learning and environmentalism, we show in this article that Pavlovian fear conditioning and extinction, as measured by fear-potentiated startle reflexes, are modulated by
common genetic polymorphisms in the serotonin and dopamine
systems, respectively.
Pavlovian fear conditioning imbues a relatively neutral
stimulus (the conditioned stimulus, or CS) with fear-evoking
properties as the result of its co-occurrences with an aversive
event (the unconditioned stimulus, or US) that threatens the
well-being of the organism. As a result of fear conditioning,
therefore, remnants of traumatic events will stick to traumaexposed individuals as components of fear memories evoked by
stimuli associated (more or less consciously) with traumas.
Accordingly, fear conditioning has carried a heavy explanatory
burden in psychological theories of anxiety disorders such as
phobias (e.g., Ohman, Dimberg, & Ost, 1985; Seligman, 1971),
panic disorder (e.g., Bouton, Mineka, & Barlow, 2001), and
posttraumatic stress disorder (e.g., Orr et al., 2000). Extinction
(i.e., the waning of fear as a result of the CS being presented in
the absence of the US) has obvious implications for the treatment
of phobias and other anxiety disorders (Anderson & Insel,
2006). Indeed, it inspired the exposure therapies that have
provided highly effective treatments of anxiety disorders (Barlow,

Copyright r 2009 Association for Psychological Science

Genetic Gating of Fear Learning and Extinction

2002). In direct support of the common assumption that fear

conditioning is a potential mechanism of anxiety disorder, metaanalyses indicate that patients diagnosed with anxiety disorder
show more rapid fear conditioning and slower extinction than
control subjects (Lissek et al., 2005).
Important progress has been made in understanding the neural
mechanisms of Pavlovian fear conditioning (see reviews by Davis
& Whalen, 2001; Fanselow & Poulos, 2005; LeDoux, 2000).
Briefly summarized, research has shown that cortical and subcortical sensory pathways converge on the lateral nucleus of the
amygdala, which houses the molecular machinery for forming
associations between the CS and the US (Fanselow & Poulos,
2005; LeDoux, 2000). As a result of this associative process, the
CS may activate the central nucleus of the amygdala, which
recruits various fear responses through pathways to the striatum,
diencephalon, midbrain, and brainstem (Davis & Whalen, 2001).
The neural basis of extinction is less well understood, but both
animal (e.g., Quirk & Gehlert, 2003) and human (e.g., Phelps,
Delgado, Nearing, & LeDoux, 2004) data suggest that extinction
depends on structures in the medial prefrontal cortex.
Behavior genetic studies suggest that about one third of the
variance in human fear conditioning (Hettema, Annas, Neale,
Kendler, & Fredrikson, 2003) and the risk to develop anxiety
disorders (Gordon & Hen, 2004) can be attributed to genetic
factors. Given that a polymorphism in the serotonin transporter
(5-HTT) gene (SLC6A4) has been convincingly related to
amygdala reactivity (for a meta-analysis, see Munafo, Brown, &
Hariri, 2008), this polymorphism is a likely candidate to be
involved in conditioned fear. A 43-bp insertion/deletion in the
5-HTT promoter, referred to as 5-HTTLPR, yields a long (l) and
short (s) allele, the latter of which reduces 5-HTTexpression and
serotonin uptake by close to 50% (Lesch et al., 1996). In addition to having a relationship to the amygdala, this low activity s allele is associated with higher neuroticism scores (for a
meta-analysis, see, e.g., Sen, Burmeister, & Ghosh, 2004).
The potentially pivotal role of the prefrontal cortex in extinction points to a candidate gene for extinction processes,
the gene coding for catechol-O-methyltransferase (COMT),
which degrades released extracellular dopamine. The COMT
gene harbors an interesting single nucleotide polymorphism:
COMTval158met. Carriers of the met allele have a 4-fold
reduction in enzyme activity compared with carriers of the val
allele, and as a result have higher extracellular dopamine levels,
particularly in prefrontal areas (for a review, see, e.g., Bilder,
Volavka, Lachman, & Grace, 2004). The met allele enhances
prefrontal cognition and working memory (for a review, see, e.g.,
Bilder et al., 2004) and has been associated with negative mood
states (e.g., anxiety, dysphoria) and impaired pain regulation.
Furthermore, the met allele is related to higher responsiveness
in, and connectivity between, brain areas involved in the evocation and regulation of negative affective responsesin particular, the amygdala, hippocampus, and prefrontal areas (for a
review, see, e.g., Heinz & Smolka, 2006).

The purpose of this study was to use psychophysiological

indices of fear to examine whether these polymorphisms are
related to fear conditioning and extinction in humans. Given
some degree of heritability of both fear conditionability and
anxiety disorders, such relationships could provide mechanisms
behind the gene-environment interaction in anxiety disorder.
First, because of the associations between the 5-HTTLPR s
allele and amygdala reactivity, and between the amygdala and
fear conditioning, we hypothesized that carriers of this allele
would exhibit enhanced fear acquisition. Second, we hypothesized a selective effect of the COMT met allele on extinction
because of the prefrontal focus of the actions of COMT and the
suggested role of this brain area in extinction and other forms of
emotion regulation.
Startle blink potentiation and skin conductance responses
(SCRs) were used as measures of conditioning. Fear potentiation
of the startle reflex is one of the most useful indices of defensive
response mobilization (Lang, Davis, & Ohman, 2000). Neurally,
it reflects the influence of direct and indirect connections from
the amygdala to the primary startle-reflex pathway in the
brainstem (Davis & Whalen, 2001). In contrast to the SCR,
which can be dissociated from amygdala activations (Tabbert,
Stark, Kirsch, & Vaitl, 2006) and appears to primarily reflect
cognitive contingency learning (Lovibond & Shanks, 2002),
startle potentiation appears to index a basic, affective level of
fear conditioning largely independent of higher cognition (Hamm
& Weike, 2005; Ohman & Mineka, 2001).
METHOD

Participants
Eighty-one students of the University of Greifswald were recruited by advertisements, filled in an informed-consent form
(approved by the ethics committees at the Karolinska Institutet
and the University of Greifswald), and donated 20 ml of blood for
DNA extraction and genotyping. Thirty-four of these students
completed the experiment before being genotyped, and 47 students were genotyped before being invited to participate in the
experiment. Out of the latter 47 students, 29 were selectively
invited to participate in order to balance out the allele frequencies and sex distribution in the experimental sample. Thus,
63 volunteers participated in the experiment (both participants
and the experimenter were blinded to genotype) and were paid
h15.
We excluded 6 (3 male, 3 female) participants from the data
analysis because of technical problems and 9 (3 male, 6 female)
participants because they could not correctly report the CS-US
contingency at the end of the conditioning phase (see Procedure). Lack of awareness, which was unrelated to our specific
genotype groups, typically compromises SCRs, but not startle
conditioning (Hamm & Weike, 2005; Lovibond & Shanks,
2002), and thus would likely introduce an irrelevant source of
variance. The final sample comprised 48 participants.

Volume ]]]Number ]]

T.B. Lonsdorf et al.

Genotyping
DNA extraction from whole blood was performed using standard
methods (Autopure LS system, Gentra Systems, Minneapolis,
MN; see Lindblom & Holmlund, 1988). For 5-HTTLPR genotyping, primers (Thermo Scientific, Ulm, Germany) flanking the
5-HTTLPR of the 5-HTT gene (forward 5 0 -TGAATGCCAG
CACCTAACCCCTAA-3 0 and reverse 5 0 -GAATACTGGTAGG
GTGCAAGGAGA-3 0 ) were used. Polymerase chain reactions
(PCRs) were carried out using the following cycling conditions:
an initial 5-min step at 95 1C; followed by 10 cycles consisting of
denaturation at 95 1C for 1 min, annealing at 65 1C (decrease of
0.5 1C with each cycle) for 1 min, and elongation at 72 1C for 1
min; followed by 39 cycles consisting of denaturation at 95 1C
for 45 s, annealing at 60 1C for 1 min, and elongation at 72 1C for
1 min; followed by a final step at 72 1C for 4 min. Reactions were
performed in 10  Reaction buffer (Finnzymes, Espoo, Finland), 1.5-mM MgCl2, 20 ng of genomic DNA, 0.75 ml of dNTPs
(Larova, Teltow, Germany; 10 mM), 0.625 ml of each primer
(10 mM), 0.0375 ml of 7-deaza-dGTP (Roche, Mannheim, Germany), and 0.5 ml of TaqPolymerase (Dynazyme, Finnzymes,
Espoo, Finland). The PCR products (10 ml) were separated into
short (336 bp) and long (379 bp) fragments by electrophoresis on
2.5% agarose gels with 2% normal agarose (Certifiedt Molecular Biology Agarose, Bio Rad, Hercules, CA) and 0.5% lowmelting agarose (Sea Plaques GTG Agarose, Cambrex Bio
Science, Rockland, ME) with ethidium bromide in TRIS-BoratEDTA-Buffer (TBE). After 2.5 hr of electrophoresis in TBE,
the products were visualized by ultraviolet illumination.
COMTval158met (rs4680) genotype was determined on an
ABI HT7900 (Applied Biosystems, Foster City, CA) using the
TaqMans allelic discrimination (5 0 nuclease assay; Livak, 1999).

Stimulus Materials
Four different color pictures depicting male faces were selected
from the Karolinska Directed Emotional Faces (Lundqvist,
Flykt, & Ohman, 1998) to serve as CSs (two pictures depicted
neutral expressions, and two depicted angry facial expressions1). Which picture was coupled to the US, the valence of the
expressions presented (each participant saw only neutral faces
or only angry faces), and the stimulus sequence used (eight
different sequences) were balanced across the genotype groups;
Table 1 shows the distribution of the variables of interest in the
final sample of 48 participants. The pictures (visible size: 126 cm
 93 cm) were projected onto a screen approximately 2 m in
front of the participant using a projector (Sanyo PLC-XU86) that
was situated in an adjacent room.
The US was a 500-Hz monopolar DC-pulse electric stimulation applied above the participants right ankle in a 10-ms train
1
CS valence was not included in the statistical analyses because of a nonsignificant Stimulus  Valence interaction for startle responses and SCR
magnitudes (see the Data Analysis section for further information on the factors
in these analyses).

Volume ]]]Number ]]

of 1-ms single pulses. It was generated by a commercial stimulator (Grass Instruments S48K, West Warwick, RI), isolated
(SIU5), and transmitted via a constant-current unit (CCU1) to a
bipolar electrode (F-E10S2). The mean physical intensity chosen during the intensity adjustment for the experiment (see
Procedure), the sensitization effect (increase in startle and SCRs
after US intensity adjustment), and unconditioned SCRs to the
US did not differ between the 5-HTTLPR groups or between the
COMTval158met genotype groups.
Startle probes were 50-ms bursts of 95-dB[A] white noise (rise
time < 1 ms). They were presented binaurally over Sony (MDRCD 170) headphones.
Physiological Recordings
Startle responses were measured by recording electromyographic (EMG) activity over the orbicularis oculi muscle beneath the left eye using miniature Ag/AgCl surface electrodes.
The raw EMG signal was amplified and filtered through a 30-Hz
high-pass filter (Coulbourn S75-01) and a 400-Hz low-pass filter
(Kemo KEM-VBF8-03; Beckenham, Kent, United Kingdom),
rectified, and integrated with a time constant of 10 ms. Skin
conductance was recorded using Hellinge Ag/AgCl standard
electrodes placed adjacently on the hypothenar eminence of the
right hand (see Weike, Schupp, & Hamm, 2007, for a more
detailed description of the recording procedures).
Procedure
Experimental Groups
Carriers of the 5-HTTLPR s allele (s/s or s/l genotype) were
combined to form the s-allele-carrier group and were compared
with l/l homozygotes. Similarly, carriers of the COMT val allele
were combined to form the val-allele-carrier group and compared with met/met homozygotes2 (see Table 1 and Table 2 for
further description of these groups).
Day 1 (Fear Acquisition)
The experimental procedure on Day 1 included a baseline phase
(5 min of familiarization with the lab situation and presentation
of four startle probes for habituation), individual adjustment of
the US to a level described as highly annoying but not painful,
and conditioning. During conditioning (see Fig. 1a), participants

2
During conditioning and extinction, carriers of one 5-HTTLPR s allele
(s/l genotype) and carriers of two 5-HTTLPR s alleles (s/s genotype) showed
comparable CS1 potentiation, but significantly larger CS1 potentiation than
the l/l homozygotes. The three COMT genotype groups did not differ from each
other in CS1 potentiation during conditioning. During extinction, the met/met
group showed larger CS1 potentiation than both the heterozygotes and the val/
val group. However, the latter trend did not reach significance, likely because of
the small sample size for this contrast (n 5 9 for met/met and n 5 12 for val/
val). Thus, we considered it empirically justified to combine carriers of at least
one 5-HTTLPR s allele into a single group and to combine carriers of at least
one COMT val allele into a single group.

Genetic Gating of Fear Learning and Extinction

TABLE 1
Mean Age (in Years) and Sex Distribution of the Participants and
Assignment of Stimuli to the Different Genotype Groups

Genotype group

Mean
age

CS
Sex: valence:
male/ angry/
CS1:
female neutral #09/#28

5-HTTLPR
s-allele carriers
30 23.7 (0.5) 15/15
l/l homozygotes
18 24.3 (0.6) 10/8
COMTval158met
val-allele carriers
39 24.0 (0.4) 22/17
met/met homozygotes 9 23.7 (1.0) 3/6

12/18
8/10

16/14
10/8

16/23
4/5

22/17
4/5

Note. For age, standard errors of the means are given in parentheses. The
CS valence column indicates how many participants in each group viewed
faces with angry expressions and how many viewed faces with neutral expressions. The CS1 column indicates the number of participants for whom
Picture 09 (in the Karolinska Directed Emotional Faces; Lundqvist, Flykt, &
hman, 1998) was coupled with the unconditioned stimulus and the number of
O
participants for whom Picture 28 was coupled with the unconditioned stimulus.

viewed each of two CS pictures nine times, in a mixed order.

Each picture was presented for 6 s. One of the pictures (CS1)
was always paired with the 10-ms US, which occurred simultaneously with the offset of the picture (100% reinforcement,
delay conditioning); the other picture (CS) was never coupled
to the US. Acoustic startle probes were presented 4 or 5 s after
picture onset for six of the nine presentations of each CS and
during six intertrial intervals (ITIs; ITI 5 1018 s). Participants
were instructed to attend to the pictures, but no information
about the CS-US contingencies was given. The conditioning
phase ended with a standardized postexperimental awareness
interview (cf. Bechara et al., 1995), so we could assess awareness of the CS-US contingency. Participants also retrospectively
rated the US, acoustic startle probe, CS1, and CS for valence
and arousal using the Self-Assessment Manikin (Lang, 1980);
these ratings did not differ between the 5-HTTLPR groups or
between the COMTval158met genotype groups.
Day 2 (Extinction)
The experimental procedure on Day 2 (approximately 24 hr
later; see Fig. 1b) included a baseline phase (presentation of four
startle probes for habituation) and extinction. During extinction,
the CS1 and CS were presented 18 times each without
TABLE 2
Distribution of Genotypes in the Sample
5-HTTLPR
COMTval158met

s-allele
carriers

l/l
homozygotes

met/met homozygotes
val-allele carriers
Total

n56
n 5 24
n 5 30

n53
n 5 15
n 5 18

Total
n59
n 5 39

administration of any further USs. Startle probes were presented

4 or 5 s after picture onset for 12 of the 18 presentations of each
CS and during 12 ITIs. Again, participants were instructed to
attend to the pictures, but no information about the CS-US
contingencies was given. After completing the experiment,
participants were debriefed and paid.
Data Reduction and Response Definition
The magnitude of the startle eyeblink (in microvolts) was measured from onset to peak, as described previously by Weike et al.
(2007). Blink magnitudes were normalized using z-standardization and converted to T scores to ensure that all participants
contributed equally to the group means. The T-score calculation,
50 1 (z  10), results in a distribution with an overall mean of 50
and a standard deviation of 10 for each participant.
SCR magnitude (in microsiemens) was scored as the largest
response occurring 0.9 to 4.0 s after picture onset. Logarithms
were computed for all values, to normalize the distribution (Venables & Christie, 1980), and these log values were range-corrected
(individual score/individual maximum response) to account for
interindividual variability (Lykken & Venables, 1971).
Startle and SCR measurements that showed recording artifacts or excessive baseline activity were discarded. Thirty-four
of the 3,120 startle measurements were discarded (1.1%; 06
per participant), as were 70 of the 3,360 SCR measurements
(2.1%; 013 per person).
Data Analysis
Data were analyzed separately for the conditioning and extinction phases using SPSS 15 for Windows. Fear-potentiated startle
was measured by subtracting the mean magnitude of startle
responses elicited by probes during ITIs from the mean magnitude of startle responses elicited by the same probes presented
during the CSs. We calculated the mean startle potentiation to
the CS1 and the CS using individually standardized T-score
differences. Furthermore, we calculated scores for mean CS
discrimination by subtracting the mean T-score startle magnitude elicited during the CS from the mean T-score startle
magnitude elicited during the CS1. To make sure that T scores
reflected the group differences in stimulus effects, rather than
baseline differences, we examined group differences in rawscore startle responses to ITI probes separately. SCR conditioning was assessed as mean CS discrimination.
We performed repeated measures analyses of variance (ANOVAs)
with stimulus (CS1 vs. CS vs. ITI for startle; CS1 vs. CS
for SCR) as a within-subjects variable and 5-HTTLPR genotype
(s carriers vs. l/l homozygotes) or COMTval158met genotype (val
carriers vs. met/met homozygotes) as a between-subjects variable (N 5 48). If an ANOVA revealed a significant main effect
for genotype or a significant Stimulus  Genotype interaction,
simple contrasts (CS1 potentiation, CS potentiation, and CS
discrimination) were calculated to specify this effect. For an ex-

Volume ]]]Number ]]

T.B. Lonsdorf et al.

a
US

Startle Probe

Conditioning
CS

CS+

CS+

CS

CS+

CS

6s

6s

6s

6s

6s

6s

b
Extinction (24 hr later)
CS+

CS

CS

CS+

CS

CS+

6s

6s

6s

6s

6s

6s

Fig. 1. The fear-conditioning paradigm. During conditioning (a), participants viewed each of two conditioned stimuli (CSs) nine times, in a mixed order. One of the pictures (CS1) was always paired with the
10-ms unconditioned stimulus (US); the other picture (CS) was never coupled to the US. Acoustic startle
probes were presented 4 or 5 s after picture onset for six of the nine presentations of each CS and during
six intertrial intervals (ITIs). During extinction (b), the CS1 and CS were presented 18 times each
without administration of any further USs. Startle probes were presented 4 or 5 s after picture onset for 12
of the 18 presentations of each CS and during 12 ITIs.

ploratory within-subjects analysis in s-allele carriers, the same

analyses were applied with COMT genotype as a between-subjects
factor (N 5 30). We adopted a significance level of .05, and
Greenhouse-Geisser adjustments of degrees of freedom were used
when appropriate. We report Zp 2 as the estimate of effect size.
RESULTS

A 3  2 ANOVA revealed a significant interactive effect of

stimulus and 5-HTTLPR genotype on startle blink responses
during conditioning, F(2, 92) 5 3.78, p 5 .026, Zp 2 :076, in
the absence of significant group differences in blink response
during the ITI (raw-score magnitudes in microvolts), F(1, 46) 5
1.07, p 5 .307. Contrasts showed that 5-HTTLPR s-allele carriers exhibited significantly stronger startle potentiation to the
CS1 than did l/l homozygotes, p 5 .01, Zp 2 :132 (Fig. 2a);
these two genotype groups did not differ significantly in CS
discrimination or CS potentiation. Within-group analyses revealed robust CS1 and CS potentiation and significant CS
discrimination in s-allele carriers, whereas the l/l group did not

Volume ]]]Number ]]

show significant CS1 or CS potentiation or significant CS

discrimination. COMTval158met genotype did not affect fearpotentiated startle during conditioning (Fig. 2c).
During extinction, a significant Stimulus  5-HTTLPR
Genotype interaction was again found, F(2, 92) 5 8.80,
p < .001, Zp 2 :161, in the absence of group differences in
startle response magnitudes during the ITI (uncorrected magnitudes in microvolts). As during conditioning, 5-HTTLPR
s-allele carriers showed significantly more CS1 potentiation
than did l/l homozygotes ( p < .001, Zp 2 :221). In addition,
the l/l group showed stronger CS inhibition than did the sallele carriers (Fig. 2b). CS discrimination was comparable in
the two groups and significant for both.
In contrast to the conditioning phase, the extinction phase
showed a significant Stimulus  COMT Genotype interaction,
F(2, 92) 5 5.09, p 5 .008, Zp 2 :100, in the absence of
significant group differences in blink response during the ITI
(uncorrected blink magnitudes in microvolts). The met/met
homozygous group showed clearly more pronounced CS1 potentiation than did the val-allele carriers (p 5 .005, Zp 2 :159;

Genetic Gating of Fear Learning and Extinction

CS+

c
Startle Blink Magnitude
(Difference From ITI; T Scores)

CS

Conditioning:
COMTval158met

Conditioning:
5-HTTLPR
10

6
4

long/long

Extinction:
5-HTTLPR

10

***

***

***

s-allele
carriers

val-allele
carriers

long/long

**

val-allele
carriers

met/met

Extinction:
5-HTTLPR s Carriers

**

10
8

**

***

met/met

Extinction:
COMTval158met

10

***

s-allele
carriers

0
2

10

***

Conditioning:
5-HTTLPR s Carriers

10

b
Startle Blink Magnitude
(Difference From ITI; T Scores)

**

met/met

val-allele
carriers

*
met/met

val-allele
carriers

Fig. 2. Potentiation of startle-response magnitudes as a function of genotype and stimulus. Black bars show the difference between magnitude of the
startle response elicited during the CS1 (the conditioned stimulus coupled to the unconditioned stimulus) and magnitude of the startle response
elicited during the intertrial interval (ITI); white bars show the difference between magnitude of the startle response elicited during the CS (the
conditioned stimulus never coupled to the unconditioned stimulus) and magnitude of the response elicited during the ITI. Results are shown for 5HTTLPR genotype groups (a) during conditioning and (b) during extinction, for COMTval158met genotype groups (c) during conditioning and (d)
during extinction, and for COMTval158met genotype groups within 5-HTTLPR s-allele carriers (e) during conditioning and (f) during extinction.
Error bars represent standard errors. Asterisks indicate significant differences, np < .05, nnp < .01, nnnp < .001.

Fig. 2d); CS discrimination and CS potentiation were comparable in these two groups. Furthermore, although CS1 potentiation remained at the same level as during conditioning for
the met/met group, it decreased to a nonsignificant level for valallele carriers.
SCRs showed reliable CS discrimination in the absence of genotype effects during both conditioning and extinction. This result
suggests that the participants, irrespective of genotype, successfully learned the CS-US contingencies on a cognitive level.
As the l/l group did not show reliable startle potentiation during
conditioning, only s-allele carriers were selected for an exploratory
within-subjects analysis assessing the effect of COMTval158met
genotype on extinction (see Table 2 for the distribution of
COMTval158met genotype groups within 5-HTTLPR s-allele

carriers). As in the overall analysis, COMTval158met genotype

did not modulate startle responses during conditioning (Fig. 2e).
However, during extinction, a significant Stimulus  COMT Genotype interaction was found, F(2, 56) 5 6.00, p 5 .004,
Zp 2 :177, in the absence of significant differences in startle
responses during the ITI (uncorrected magnitudes in microvolts).
Those s-allele carriers who were homozygous for the COMT met
allele showed significantly more pronounced CS1 potentiation
than did those s-allele carriers who also carried a val allele (p 5
.002, Zp 2 :302), even though significant CS1 potentiation was
observed for both groups (Fig. 2f). No significant differences in CS
discrimination or CS potentiation were found.
Analyses of SCRs within s-allele carriers revealed significant
CS discrimination during conditioning, and a trend for CS

Volume ]]]Number ]]

T.B. Lonsdorf et al.

discrimination during extinction, in the absence of an effect of

COMTval158met genotype.
DISCUSSION

In summary, our results suggest the rather strong conclusion that

the 5-HTTLPR and COMTval158met polymorphisms gate fear
learning and extinction, respectively, as measured by fearpotentiated startle. On the one hand, only carriers of the
5-HTTLPR s allele acquired potentiated startle reactions to
environmental stimuli associated with an aversive event through
Pavlovian conditioning. On the other hand, homozygosity for the
COMT met allele selectively blocked the ability to extinguish
conditioned fear when the CS no longer was predictive of the US.
Our exploratory analysis on s-allele carriers alone confirmed on
a within-subjects basis that the combination of a 5-HTTLPR s
allele and COMT met-homozygosity conferred an enhanced risk
for acquiring fear that resisted extinction.
These findings undermine the commonly held (but mistaken)
belief in an impenetrable barrier between genes and environment. According to this belief, genes are inherited, intraorganismic, causal factors, and the environment provides the arena for
learned influences on behavior. Our results imply a more dynamic relationship by suggesting that genes may act through the
environment by making carriers of particular gene combinations
more likely than other individuals to easily pick up and retain
fear of stimuli associated with threat and trauma. Thus, people
carrying at least one 5-HTTLPR s allele and two COMT met
alleles are likely to expand their sets of fear- and anxietyevoking stimuli through facilitated fear conditioning and poor
extinction. This process might be further accelerated by stimulus generalization and second-order conditioning, and hence
such individuals may end up fearful of many stimuli that they are
exposed to in their everyday environment. As a consequence,
they might have frequent experiences of negative affect, which is
a core characteristic of neuroticism (Clark, Watson, & Mineka,
1994). This could explain the relationship between the 5HTTLPR s allele and neuroticism. If the fear elicited by many
stimuli is intense enough to promote coping attempts in the form
of avoidance, the result could be the restrictions in life options
that characterize people with anxiety disorder. This could also
explain why negative affect is a risk factor for anxiety disorder
(Clark et al., 1994).
The two polymorphisms we studied will act synergistically in
this process, one by promoting fear acquisition, and the other by
slowing extinction. Nevertheless, they are independent of each
other, as shown by the double dissociation, with the 5-HTTLPR
polymorphism affecting acquisition but not extinction of fear,
and the COMT polymorphism affecting extinction but not acquisition.
It is noteworthy that our conclusions were based on the results
of conditioned startle potentiation and were not valid for SCRs.
Our results show that the amount of CS1 startle potentiation

Volume ]]]Number ]]

differed between the 5-HTTLPR genotype groups and between

the COMTval158met genotype groups during acquisition and
extinction, respectively, whereas CS1/CS differentiation was
comparable between the two groups (the finding for s-allele
carriers echoes results with anxiety-disorder patients; Lissek
et al., 2005). Consistent with the lack of differences between
groups in discriminative startle potentiation to the CS1 and the
CS, SCRs showed reliable CS1/CS discrimination that was
independent of genotype. The dissociation between our startle
and SCR results in response to the CS1 probably reflects a
selective impact of 5-HTTLPR and COMTval158met on a basic
emotional level of fear that is manifested in startle potentiation,
but not in conditioned SCRs (e.g., Hamm & Weike, 2005;
Ohman & Mineka, 2001).
However, in an earlier study (Garpenstrand, Annas, Ekblom,
Oreland, & Fredrikson, 2001) that measured only SCRs, s-allele
carries were significantly overrepresented among individuals
showing very good fear conditioning. The discrepancy between
our results and those reported by Garpenstrand et al. may be
attributed to different research strategies: Whereas the latter
authors selected their participants on the basis of conditioning
performance and compared the genotypes of people who exhibited extremely good and poor conditioning, we followed the
hypothesized causal sequence and grouped participants a priori
by genotype and subjected them to a fear-conditioning procedure.
The results for the COMTval158met genotype appear consistent with the tonic-phasic dopamine hypothesis for COMT
(Bilder et al., 2004). Thus, the low-activity met allele should
facilitate cognitive stability but jeopardize the cognitive
flexibility (updating or resetting working memory content) needed to adapt to the change in conditioning contingencies during
fear extinction. This emotional perseveration, observed as a
failure of extinction in the met/met group, most likely reflects
impaired cognitive control over emotional reactions. Our results
are thus in line with the fact that the COMT met allele has been
associated with both cognitive control and anxiety proneness.
Although our data suggest that the 5-HTTLPR s allele serves
as a gate for fear conditioning and that carrying a COMT val
allele serves as a gate for fear extinction, the generality of these
conclusions across experimental conditions and measures remains to be determined. For example, one interesting possibility
is that our results depended on our use of facial stimuli as CSs
(Canli & Lesch, 2007). Another potentially critical factor is the
intensity of the US. For ethical reasons, our US was of moderate
intensity; therefore, the present findings should not be taken to
imply that homozygosity for the 5-HTTLPR l allele precludes
fear conditioning regardless of circumstances. Similarly, more
prolonged extinction or the intense CS exposure used in singlesession treatment of specific phobia (Ost, 1997) may overcome
the apparent limitation to extinction we observed among met
homozygotes. Examining these limitations, as well as realizing
the promise of a deeper understanding of the dynamics inherent

Genetic Gating of Fear Learning and Extinction

in vulnerability-stress conceptualizations of anxiety disorder,

must await future research.

AcknowledgmentsThis work was supported by grants from

the Swedish Science Research Council, the Nordic Research
Council for the Humanities and Social Sciences (NOS-HS)
Nordic Centre of Excellence in Cognitive Control, and the National Institute of Mental Health Center for the Study of Emotion
and Attention to A.O., and by a grant from the Swedish Research
Council to M.S. T.B.L. was supported by the German Academic
Exchange Service (Deutscher Akademischer Austauschdienst).
We thank Heino Mormann for technical assistance, Carmen
Hamm for blood sampling, and Andreas Olsson and Armita
Golkar for comments on earlier versions of this manuscript.
REFERENCES
Anderson, K.C., & Insel, T.R. (2006). The promise of extinction research for the prevention and treatment of anxiety disorders.
Biological Psychiatry, 60, 319321.
Barlow, D.H. (2002). Anxiety and its disorders: The nature and treatment of anxiety and panic (2nd ed.). New York: Basic Books.
Bechara, A., Tranel, D., Damasio, H., Adolphs, R., Rockland, C., &
Damasio, A.R. (1995). Double dissociation of conditioning and
declarative knowledge relative to the amygdala and hippocampus
in humans. Science, 269, 11151118.
Bilder, R.M., Volavka, J., Lachman, H.M., & Grace, A.A. (2004).
The catechol-O-methyltransferase polymorphism: Relations to the
tonic-phasic dopamine hypothesis and neuropsychiatric phenotypes. Neuropsychopharmacology, 29, 19431961.
Bouton, M.E., Mineka, S., & Barlow, D.H. (2001). A modern learning
theory perspective on the etiology of panic disorder. Psychological Review, 108, 432.
Canli, T., & Lesch, K.P. (2007). Long story short: The serotonin
transporter in emotion regulation and social cognition. Nature
Neuroscience, 10, 11031109.
Clark, L.A., Watson, D., & Mineka, S. (1994). Temperament, personality, and the mood and anxiety disorders. Journal of Abnormal
Psychology, 103, 103116.
Davis, M., & Whalen, P.J. (2001). The amygdala: Vigilance and
emotion. Molecular Psychiatry, 6, 1334.
Fanselow, M.S., & Poulos, A.M. (2005). The neuroscience of mammalian associative learning. Annual Review of Psychology, 56,
207234.
Garpenstrand, H., Annas, P., Ekblom, J., Oreland, L., & Fredrikson, M.
(2001). Human fear conditioning is related to dopaminergic and
serotonergic biological markers. Behavioral Neuroscience, 115,
358364.
Gordon, J.A., & Hen, R. (2004). Genetic approaches to the study of
anxiety. Annual Review of Neuroscience, 27, 193222.
Hamm, A.O., & Weike, A.I. (2005). The neuropsychology of fear
learning and fear regulation. The International Journal of Psychophysiology, 57, 514.
Heinz, A., & Smolka, M.N. (2006). The effects of catechol O-methyltransferase genotype on brain activation elicited by affective
stimuli and cognitive tasks. Reviews in the Neurosciences, 17,
359367.

Hettema, J.M., Annas, P., Neale, M.C., Kendler, K.S., & Fredrikson,
M. (2003). A twin study of the genetics of fear conditioning. Archives of General Psychiatry, 60, 702708.
Lang, P.J. (1980). Behavioral treatment and bio-behavioral assessment: Computer applications. In J.B. Sidowski, J.H. Johnson, &
T.A. Williams (Eds.), Technology in mental health care delivery
systems (pp. 119137). Norwood, NJ: Ablex.
Lang, P.J., Davis, M., & Ohman, A. (2000). Fear and anxiety: Animal
models and human cognitive psychophysiology. Journal of Affective Disorders, 61, 137159.
LeDoux, J.E. (2000). Emotion circuits in the brain. Annual Review of
Neuroscience, 23, 155184.
Lesch, K.-P., Bengel, D., Heils, A., Sabol, S.Z., Greenberg, B.D., Petri,
S., et al. (1996). Association of anxiety-related traits with a
polymorphism in the serotonin transporter gene regulatory region.
Science, 274, 15271531.
Lindblom, B., & Holmlund, G. (1988). Rapid DNA purification for
restriction fragment length polymorphisms analysis. Gene Analysis Techniques, 5, 97101.
Lissek, S., Powers, A.S., McClure, E.B., Phelps, E.A., Woldehawariat,
G., Grillon, C., & Pine, D.S. (2005). Classical fear conditioning in
the anxiety disorders: A meta-analysis. Behaviour Research and
Therapy, 43, 13911424.
Livak, K.J. (1999). Allelic discrimination using fluorogenic probes and
the 5 0 nuclease assay. Genetic Analysis: Biomolecular Engineering, 14, 143149.
Lovibond, P.F., & Shanks, D.R. (2002). The role of awareness in
Pavlovian conditioning: Empirical evidence and theoretical implications. Journal of Experimental Psychology: Animal Behavior
Processes, 28, 326.
Lundqvist, D., Flykt, A., & Ohman, A. (1998). KDEF: The Karolinska
Directed Emotional Faces [CD-ROM]. Stockholm: Karolinska Institutet, Department of Clinical Neuroscience, Psychology Section.
Lykken, D.T., & Venables, P.H. (1971). Direct measurement of skin
conductance: Proposal for standardization. Psychophysiology, 8,
656672.
Mineka, S., & Zinbarg, R. (2006). A contemporary learning theory
perspective on the etiology of anxiety disorders: Its not what you
thought it was. American Psychologist, 61, 1026.
Munafo, M.R., Brown, S.M., & Hariri, A.R. (2008). Serotonin transporter (5-HTTLPR) genotype and amygdala activation: A metaanalysis. Biological Psychiatry, 63, 852857.
Ohman, A., Dimberg, U., & Ost, L.-G. (1985). Animals and social
phobias: Biological constraints on learned fear responses. In S.
Reiss & R.R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp. 123175). New York: Academic Press.
Ohman, A., & Mineka, S. (2001). Fears, phobias, and preparedness:
Toward an evolved module of fear and fear learning. Psychological Review, 108, 483522.
Orr, S.P., Metzger, L.J., Lasko, N.B., Macklin, M.L., Peri, T., & Pitman,
R.K. (2000). De novo conditioning in trauma-exposed individuals
with and without posttraumatic stress disorder. Journal of Abnormal Psychology, 109, 290298.
Ost, L.-G. (1997). Rapid treatment of specific phobias. In G.C.L.
Davey (Ed.), Phobias: A handbook of theory, research and treatment (pp. 227247). Chichester, England: Wiley.
Phelps, E.A., Delgado, M.R., Nearing, K.I., & LeDoux, J.E. (2004).
Extinction learning in humans: Role of the amygdala and vmPFC.
Neuron, 43, 897905.
Quirk, G.J., & Gehlert, D.R. (2003). Inhibition of the amygdala: Key to
pathological states. In P. Shinnick-Gallagher, A. Pitkanen, A.

Volume ]]]Number ]]

T.B. Lonsdorf et al.

Shekhar, & L. Cahill (Eds.), The amygdala in brain function:

Basic and clinical approaches (Annals of the New York Academy
of Sciences Vol. 985, pp. 263272). New York: New York
Academy of Sciences.
Seligman, M.E.P. (1971). Phobias and preparedness. Behavior Therapy, 2, 307320.
Sen, S., Burmeister, M., & Ghosh, D. (2004). Meta-analysis of the
association between a serotonin transporter promoter polymorphism (5-HTTLPR) and anxiety-related personality traits.
American Journal of Medical Genetics Part B: Neuropsychiatric
Genetics, 127B, 8589.
Tabbert, K., Stark, R., Kirsch, P., & Vaitl, D. (2006). Dissociation of
neural responses and skin conductance reactions during fear
conditioning with and without awareness of stimulus contingencies. NeuroImage, 32, 761770.

Volume ]]]Number ]]

Venables, P.H., & Christie, M.J. (1980). Electrodermal activity. In I.

Martin & P.H. Venables (Eds.), Techniques in psychophysiology
(pp. 367). Chichester, England: Wiley.
Watson, J.B., & Raynor, R. (1920). Conditioned emotional reactions.
Journal of Experimental Psychology, 3, 114.
Weike, A.I., Schupp, H.T., & Hamm, A.O. (2007). Fear acquisition
requires awareness in trace but not delay conditioning. Psychophysiology, 44, 170180.

(RECEIVED 3/20/08; REVISION ACCEPTED 7/11/08)