Enhanced cardiac perception predicts impaired performance

in the Iowa Gambling Task in patients with panic disorder

Julian Wölk1,2, Stefan Sütterlin3,4, Stefan Koch2,5, Claus Vögele4,6 & Stefan M. Schulz1,7
1
Department of Psychology I, University of Würzburg, Würzburg, Germany
2
Schön Klinik Roseneck, Hospital of Behavioral Medicine, Prien am Chiemsee, Germany
3
Lillehammer University College, Department of Psychology, Lillehammer, Norway
4
Research Group on Health Psychology, University of Leuven, Leuven, Belgium
5
Paracelsus Medical University Salzburg, Austria
6
Research Group Self-Regulation and Health, Research Unit INSIDE, University of Luxembourg, Luxembourg, Luxembourg
7
Comprehensive Heart Failure Center, University of Würzburg, Würzburg, Germany

Keywords Abstract
Cardiac perception, decision making,
interoception, Iowa Gambling Task, panic Objective: Somatic marker theory predicts that somatic cues serve intuitive
disorder, somatic marker hypothesis decision making; however, cardiovascular symptoms are threat cues for patients
with panic disorder (PD). Therefore, enhanced cardiac perception may aid
Correspondence intuitive decision making only in healthy individuals, but impair intuitive deci-
Stefan M. Schulz, Department of Psychology sion making in PD patients. Methods: PD patients and age- and sex-matched
I, University of Würzburg, Würzburg,
volunteers without a psychiatric diagnosis (n = 17, respectively) completed the
97070 Germany.
Iowa Gambling Task (IGT) as a measure of intuitive decision making. Interin-
Tel: +49 931 31 80184; Fax: +49 931 31
82733; dividual differences in cardiac perception were assessed with a common men-
E-mail: [email protected] tal-tracking task. Results: In line with our hypothesis, we found a pattern of
opposing associations (Fisher’s Z = 1.78, P = 0.04) of high cardiac perception
Funding Information with improved IGT-performance in matched control-participants (r = 0.36,
Preparation of this manuscript was supported n = 14) but impaired IGT-performance in PD patients (r = 0.38, n = 13).
by the Federal Ministry of Education and
Conclusion: Interoceptive skills, typically assumed to aid intuitive decision
Research, BMBF, project 01EO1004 (S. M. S.,
making, can have the opposite effect in PD patients who experience interocep-
Comprehensive Heart Failure Center [CHFC],
University of Würzburg, Germany). tive cues as threatening, and tend to avoid them. This may explain why PD
patients frequently have problems with decision making in everyday life.
Received: 16 May 2013; Revised: 25 Screening of cardiac perception may help identifying patients who benefit from
November 2013; Accepted: 28 November specifically tailored interventions.
2013

doi: 10.1002/brb3.206

somatic states (e.g., heart rate, skin conductance, muscle

Introduction
tone), which were previously evoked by similar decisions.
The effects of physical sensations on overt behavior have These “emotional marker” signals are represented in the
been subject to extensive research, often based on the anterior insular cortex and embedded in decision-making
somatic marker hypothesis (SMH, Damasio et al. 1991; processes via ventromedial prefrontal pathways (Damasio
Bechara et al. 1994; Damasio 1995). The SMH suggests, 1995).
for example, that somatic cues guide decision making in An established paradigm for the assessment of intuitive
complex situations, which are characterized by little expli- decision-making patterns under time pressure and with
cit information to base a decision on, and/or time pres- incomplete information is the Iowa Gambling Task (IGT,
sure. More specifically, the SMH posits, that responses in Bechara et al. 1994). Implicit learning skills have been
such situations are associated with specific, learned reported to be positively associated with IGT performance

ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc. This is an open access article under the terms of 1
the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
Cardiac Perception and Decision Making J. Wölk et al.

(Bechara et al. 1997). The main body of literature has previous studies (Domschke et al. 2010; Grosche et al.
considered skin conductance response as a proxy for vis- 2011), we expected to find enhanced cardiac perception
ceral somatic markers (Dunn et al. 2006), although the in PD patients as compared to matched controls without
SMH suggests that cardiac cues may play a similar role. psychiatric diagnosis. In control participants we further
In line with this assumption, interindividual differences in hypothesized cardioceptive perception accuracy to be pos-
trait cardiac perception accuracy have been found to itively associated with performance in intuitive decision
affect emotional bias on speeded reactions in healthy vol- making. In PD patients, we expected to find the opposite
unteers (Sütterlin et al. 2013). Moreover, at least one pattern of results, that is high cardioceptive accuracy to
study has demonstrated that healthy participants with impair intuitive decision making due to PD patients’ cat-
particularly high cardiac perception outperform those astrophic interpretation of such cues.
with lower accuracy in the IGT (Werner et al. 2009).
The perception and cognitive evaluation of physical
symptoms is considered to play a crucial role in the Methods
development and maintenance of panic disorder (PD).
The psychophysiological model of PD (Ehlers and Marg- Sample
raf 1989) describes a vicious circle of perception of physi- The patient sample consisted of 17 inpatients (eight
cal cues and their catastrophizing evaluation, which female patients; M = 41.59 years, SD = 13.30), admitted
increases the probability of panic attacks. While healthy to a psychosomatic hospital (Schön Klinik Roseneck,
individuals attribute the experience of physical changes Hospital of Behavioral Medicine, Prien, Germany). All
(e.g., beating heart, shortness of breath, etc.) to a variety patients had a principal diagnosis of PD as assessed by
of internal or external stressors, patients with PD habitu- trained clinical psychologists. Diagnoses were based on
ally associate such sensations with imminent threat (Clark DSM-IV-TR criteria (American Psychiatric Association
et al. 1988; Hofmann et al. 2008). 2000), and verified by J. W., based on the PD-related
Increased sensitivity to physical cues (Barsky 1992; parts of the SKID-I (Wittchen et al. 1997). The control
Ehlers and Breuer 1996; Eley et al. 2004; Hoehn-Saric group comprised of 17 volunteers (eight female volun-
et al. 2004) and their catastrophizing appraisal are typical teers; M = 36.53 years; SD = 12.10) without mental dis-
features of PD and are often principal targets for PD orders, matched for gender and age. As indicated by self-
treatment (Hofmann et al. 2008). A large body of disclosure, none of the control participants had a psychi-
research further supports the role of biased perception atric diagnosis or any cardiac and/or neurological disor-
and interpretation of physical symptoms in the develop- der and, therefore, no related medication. Moreover,
ment (Bouton et al. 2001) and maintenance of PD (Ehlers none of the participants had taken benzodiazepines within
1993; Richards et al. 2003). There is evidence for 2 weeks prior to the experimental assessment. Patients
increased perception of physical symptoms in PD patients with an additional diagnosis of somatoform disorder were
(Domschke et al. 2010). Physical symptom perception is excluded, due to the suggested role of abnormal percep-
often part of PD patients’ reported symptomatology tion of physical symptoms in this diagnostic category.
(Zoellner and Craske 1999) with cardiac symptoms such Nevertheless, those meeting criteria for secondary anxiety
as heart rate playing a prominent role (Hartl 1995). disorders (social phobia n = 1, 11.76%), generalized anxi-
In addition to increased symptom perception and its ety disorder (n = 1, 5.88%), or major depression (n = 16,
biased attribution to impending threat, there is also 94.10%) were included, thus representing a typical sample
evidence for intolerance of uncertainty in patients with of PD patients in clinical practice (Kaufmann and Char-
PD (Carleton et al. 2012; Mahoney and McEvoy 2012), ney 2000; Brown et al. 2001).
reduced risk-taking behavior (Giorgetta et al. 2012),
increased latency in speeded decision making (Kaplan
et al. 2006) and heightened sensitivity to errors (Ludewig Materials
et al. 2003). Yet, whether increased perception of physical Equipment
cues would impact upon intuitive decision making in PD
patients remains unclear. Given PD patients’ habitual ECG was recorded with the NeXus-10 system (Mind
catastrophizing interpretation of physical cues, it could be Media BV/Roermond-Herten, Netherlands) using Eintho-
argued that increased interoceptive awareness is detrimen- ven lead I configuration with Ambu Blue Sensor VL
tal for intuitive decision making in patients with PD. (Ambu GmbH/Bad Nauheim/Germany) electrodes. Data
The aim of this study was to examine the effects of were sampled at 512 Hz. A freeware IGT application was
increased perception and processing of somatic markers run on a personal computer (ASUS, Taipeh, Taiwan)
on decision-making processes in PD patients. In line with with Windows XP operating system and presented on a

2 ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc.
J. Wölk et al. Cardiac Perception and Decision Making

Table 1. Characteristics of the Iowa Gambling Task.

Deck A Deck B Deck C Deck D

Gain per draw 100 100 50 50

Batch of possible {0,0,0,0,0,150,200,250,300,350} {0,0,0,0,0,0,0,0,0,1250} {0,0,0,0,0,0,0,0,0,55,55,55,55,55} {0,0,0,0,0,0,0,0,0,250}
losses per draw
Range of net loss 250 to +100 1150 to +100 5 to +50 200 to +50
versus gain
per draw
Probability of 0.5 0.9 0.5 0.9
positive net gain
per draw

Each time a card is drawn from one of the decks, the associated gain is won but counterbalanced by a potential loss that is selected at random
from the respective batch of losses. The net amount per draw results from subtracting the loss from the gain. Furthermore, achieving a positive
net gain only occurs at a certain probability.

15-inch LCD-monitor at 1024 9 768 pixel screen resolu- fixed order. Instructions were given in written form to
tion with ~40 cm head-to-screen distance. standardize the instruction (Ehlers et al. 1995). We added
an initial warm-up trial to allow sufficient time for the
transition from the instructional phase to the different
Iowa Gambling Task
mode of processing during interoception. The warm-up
The IGT consists of four decks of cards (A, B, C, D). trial was not included in the analysis, as we aimed for an
Drawing cards from deck A or B results in large gains but optimal compromise between keeping the scores compa-
high losses, leading to an overall loss. In comparison, rable with previous reports as much as possible and the
drawing cards from decks C and D results in small gains added benefit of reducing task-irrelevant training effects
but similarly small losses and an overall net profit (see in a situation unfamiliar for most participants (Sütterlin
Table 1 for details). Participants are instructed to draw et al. 2013).1
100 cards from these decks, with the aim to maximize
their profit. Typically, control participants begin by select-
Self-report data
ing cards more or less randomly, followed by a period of
implicit learning with a preference for the net gain option The German version of the State–Trait Anxiety Inventory
and finally explicit knowledge resulting in a clear prefer- (Laux et al. 1981) was used to assess habitual trait and
ence for decks C and D. state anxiety on two scales comprising 20 items each.
Items are rated from 0 (not at all) to 3 (very much so),
resulting in a scale range of 0–60 for each measure.
Mental-tracking task
Higher scores indicate higher anxiety. Cronbach’s a is
Following Herbert et al. (2012), the participants were about 0.90 for both scales (Laux et al. 1981). The current
instructed to quietly count the heartbeats, which they sample achieved a Cronbach’s a of 0.92 for the STAI-
experienced in the time interval between a start and a State and 0.96 for the STAI-Trait.
stop signal without any supplementary aids such as taking The German 21-item version of the Beck Depression
their pulse or estimating the expected number of heart- Inventory (BDI) was used to assess dysphoric mood and
beats based on the (estimated) elapsed time. This task depression. Items are rated on a four-point scale from 0
was performed for four time intervals with 20, 25, 35, to 3, resulting in an overall score ranging from 0 to 63.
and 45-sec duration and a 60-sec rest time between the Higher scores represent more severe symptoms of depres-
time intervals. During this procedure, participants were sion. Internal consistency is good with Cronbach’s
asked to close their eyes, to sit relaxed, and to breathe a = 0.89 (Hautzinger et al. 2006).
consistently. Start and stop of each interval was indicated Trait anxiety sensitivity was assessed with the German
verbally by the experimenter. Particularly in small sam- version of the Anxiety Sensitivity Index (ASI, Peterson
ples, randomization often does not produce comparable and Plehn 1999). Notably, high scores on the ASI have
distributions of conditions across groups. Hence, the
order of time intervals was not randomized, to increase
procedural comparability between the two groups. Impor- 1
It may be noted that including these data in the analysis did
tantly, the individual participants were not aware of the not affect the results.

ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc. 3
Cardiac Perception and Decision Making J. Wölk et al.

been shown to predict both the frequency of panic attacks

in PD patients and increased cardioceptive accuracy
(Domschke et al. 2010). Cronbach’s a of the ASI has been
reported to exceed at least 0.75 (Peterson and Plehn Figure 1. Formula for computing the cardioceptive accuracy index
1999); the current sample achieved an internal consistency across three time intervals (Werner et al. 2009).
of 0.95.
Positive and negative affect were assessed with the Ger-
man version of the Positive and Negative Affect Schedule
Data reduction
(PANAS, Krohne et al. 1996), assessing positive and nega-
tive affect on separate scales comprising 10 items each. First, interbeat intervals were extracted from the raw
Items are rated from 1 = “little” to 5 = “most” and each ECG using ARTiiFACT (Kaufmann et al. 2011). Next,
scale score ranges from zero to 50. Higher scores repre- for each time interval of 25-, 35-, and 45-sec cardiocep-
sent higher positive/negative affect. Cronbach’s a is 0.85 tive accuracy was calculated with the formula presented
for both scales (Krohne et al. 1996). The current sample in Figure 1. The result is an index that ranges from 0
achieved a Cronbach’s a of 0.79 for the PA subscale and to 1, with 1 indicating perfect accuracy of heartbeat
0.88 for the NA subscale. detection.
In addition, the PD group completed the Panik und
Agoraphobie Skala (PAS, Bandelow 1997), which mea-
Statistical analysis
sures severity of PD on four scales comprising two items
(panic attacks, agoraphobic avoidance, constraints in daily All data were checked for normal distribution with Kol-
life and worry on healthiness), one scale with three items mogorov–Smirnov tests and Lilliefors Significance Cor-
(anticipatory anxiety), plus one additional item to assess rection. The various group characteristics were compared
whether panic attacks are mostly unexpected or related to using independent sample t-tests. To assess associations
feared situations. The PAS overall score ranges from 0 to of cardioception with IGT parameters and other vari-
52 with a cutoff of 9 indicating slight PD. Psychometric ables, Pearson’s correlations were computed. Group dif-
properties are overall sound with a test–retest reliability ferences in cardioceptive accuracy and IGT were
of 0.73, Cronbach’s a of 0.86 and high convergent validity compared with independent samples t-test. Fisher’s Z test
ranging from 0.58 to 0.76 (Bandelow 1997). Cronbach’s a for independent samples was used to compare correla-
for the individual subscales ranged between 0.70 and 0.94 tions between cardioceptive accuracy and IGT perfor-
in the current sample. mance in both groups.
There are reports on associations between performance A priori sensitivity analysis (G*Power 3.1) resulted in a
in the mental heartbeat-tracking task and participant’s critical r = 0.48 to achieve test power of 0.80 (a = 0.05)
gender (Ludwick-Rosenthal and Neufeld 1985) and body in two-tailed bivariate testing given the available sample
mass index (BMI, Montgomery et al. 1984; Jones et al. size. Varying group sizes in single tests may occur due to
1987), as well as between educational level and IGT per- missing items in self-report data.
formance (Davis et al. 2008). We assessed gender and
educational level as control variables via self-report ques-
Results
tionnaires. BMI was assessed at a medical examination
during admission.
Sample characteristics
Patients with PD did not differ from matched controls
Procedure
in terms of age, BMI, educational level, and state and
The study was carried out in compliance with the Code trait anxiety. Significant differences between groups,
of Ethics of the World Medical Association (Declaration however, occurred for positive and negative affect,
of Helsinki) and was approved by the ethics committee of depression, and anxiety sensitivity (see Table 2 for
the Schön Klinik Roseneck. Participants participated vol- details). Depression and anxiety sensitivity did not
untarily and received no compensation for taking part in correlate with other variables in both groups. Both
the study. Signed informed consent was obtained for sub- groups consisted of nine male and eight female partici-
jects after the nature of the procedures was explained. pants, their educational level was high. Of these, 58.8%
Next, participants completed the questionnaires and sub- of PD patients and 58.9% of control participants
sequently performed the mental-tracking task. After reached a higher education entrance qualification. Both
assessment of cardiocepetive accuracy, the participants groups were comparable with regard to physical
performed the IGT. activity.

4 ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc.
J. Wölk et al. Cardiac Perception and Decision Making

Table 2. Means (M), standard deviations (SD), or frequencies (n) and percent (%) significance level (P) and effect size (Cohen’s d, calculated on
basis of control group’s standard deviation) of group characteristics for patients with panic disorder (PD) versus matched controls.

PD patients Matched controls

M or n (SD or %) M or n (SD or %) Test statistic P Effect size

Age 41.59 (13.26) 36.53 (12.61) t(32) = 1.14 0.263 d = 0.40

BMI (kg/m2) 24.58 (2.69) 23.70 (3.06) t(31) = 0.87 0.383 d = 0.29
Educational level1
Hauptschule (2) 2 11.76 2 11.76 V(df) = 0.09 0.966 U = 0.09
Realschule (2) 5 29.41 5 29.41
(Fach-)Abitur (3) 3 17.65 2 11.76
Hochschulabschluss (5) 7 41.18 8 47.06
Participants practicing 12 70.59 14 82.35 Fisher’s exact test P > 0.069 –
any kind of sports (in
all cases this included
endurance training)
Frequency of training 2.67 .88 2.71 1.33 t(24) = 0.11 0.92 d = 0.04
per week
Duration of training per week 1 h 32 min 2 h 21 min 4 h 24 min 7 h 9 min t(24) = 0.11 0.92 d = 0.04
STAI-state 47.19 (3.71) 49.88 (3.26) t(31) = 2.22 0.034* d = 0.83
STAI-trait 47.41 (2.92) 49.59 (2.94) t(32) = 2.17 0.038* d = 0.74
Positive affect 22.47 (5.03) 30.59 (2.91) t(32) = 5.76 <0.001*** d = 1.98
Negative affect 19.06 (6.92) 10.88 (0.93) t(32) = 0.83 <0.001*** d = 1.66
BDI 34.43 (7.92) 3.40 (3.60) t(27) = 13.74 <0.001*** d = 8.62
ASI 35.47 (12.90) 9.94 (6.40) t(32) = 7.31 <0.001*** d = 3.99

BMI, body mass index (kg/m2); STAI, state trait anxiety inventory; ASI, anxiety sensitivity index; Hauptschule, Realschule, and (Fach-)Abitur can be
considered different levels of high school degrees (in ascending order).
1
In parentheses: international standard classification of education (ISCE) according to the UNESCO guidelines from 2011. *P < 0.05,
***P < 0.001.

Table 3. Means (M), standard deviations (SD), minimum (Min), and

Cardioceptive perception
maximum (Max) of cards drawn from the four decks (A, B, C, D) for
In contrast to our expectations there were no significant the panic disorder (PD) patient group as compared to matched
group differences in scores of cardiac perception controls.
(t(28) = 0.85, P = 0.20, d = 0.31). PD patients showed a PD patients Matched controls
tendency for lower accuracy (M = 0.61, SD = 0.20) as
Deck M SD Min Max M SD Min Max
compared to matched controls (M = 0.68, SD = 0.22).
A 23.71 6.30 12 33 16.44 10.10 3 41
B 29.50 10.80 11 53 35.25 11.70 16 53
IGT performance C 20.57 5.80 9 29 20.31 11.60 6 37
D 26.21 7.60 15 40 28.00 14.50 5 56
On average, PD patients chose a successful card in
46.79 trials out of 100 trials (SD = 8.61, min = 29,
max = 64), as compared to the matched controls, select-
Associations between cardioceptive skill
ing winning cards in 48.31 trials out of 100 trials
and IGT performance
(SD = 11.88, min = 24, max = 65). More precisely, the
PD group chose deck B most of the time followed by In line with our hypothesis, the correlations between car-
decks D, A, and C. The control group chose deck B dioceptive accuracy and IGT performance differed signifi-
most often, followed by decks D, C, and A. For details cantly between groups (Fisher’s Z = 1.78, P = 0.04).
see Table 3. Nevertheless, in contrast to previous stud- When tested against zero, there was a positive but not
ies, participants did not switch their strategy after an significant correlation between cardioceptive accuracy
initial period from exploration of all decks to selective score and positive deck selection in the control group
preference of decks with a positive yield (C and D), but (r = 0.36, n = 14, P = 0.10; see Figure 2), and an almost
continued switching decks unpredictably throughout the equally strong negative correlation in the PD group
experiment. (r = 0.38, n = 13, P = 0.10; see Figure 3), suggesting

ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc. 5
Cardiac Perception and Decision Making J. Wölk et al.

complex decision task, requiring implicit and explicit

learning (i.e., the IGT). In line with predictions derived
from somatic marker theory (Damasio et al. 1991), we
expected that utilization of interoceptive cues aids intui-
tive decision-making in age and sex-matched control par-
ticipants without psychiatric diagnosis. However, in PD
patients we expected the opposite, because interoceptive
information, in particular when related to cardiac symp-
toms, comprises a major source of threat to them, some-
times triggering panic attacks (Ehlers and Margraf 1989;
Hofmann et al. 2008). Therefore, rather than utilizing
interoceptive – in particular cardioceptive information,
we assumed that PD patients may rather try to avoid it,
hence being distracted rather than guided by
somatic cues. Therefore, we expected that high-cardiac
perception, would rather impair decision making, and
Figure 2. Scatter plot for correlations of cardioceptive skill with IGT
hence, IGT performance in PD patients. In line with our
performance in the group of PD patients, including linear regression
hypothesis, we found significantly different and opposing
line plus lines for margins of one standard deviation.
patterns of association between cardioceptive accuracy2
and IGT performance in patients with PD and matched
controls.
Control participants tended to benefit from increased
cardioceptive accuracy in terms of better IGT perfor-
mance. First, this replicates evidence for an association
between enhanced cardiac perception and intuitive deci-
sion making (Werner et al. 2009). Second, this is in line
with a study indicating that enhanced cardiac perception
is associated with avoidance of seemingly risky choices in
a framing task, where trials with objectively equal options
were framed emotionally by suggesting that one would
either have the chance to win or face the risk to lose an
equal amount of money in a given trial (Sütterlin et al.
2013).
This is the first study to show that cardiac perception
can also have the opposite effect: In PD patients, the asso-
ciation of high-cardiac perception with IGT-task perfor-
Figure 3. Scatter plot for correlations of cardioceptive skill with IGT
performance in the control group, including linear regression line plus mance was significantly different. High cardiac perception
lines for margins of one standard deviation. rather impaired than supported intuitive decisions in the
IGT. This result suggests qualitative differences between
that PD patients with high-cardioceptive accuracy selected control participants and PD patients in the processing of
advantageous decks less often. interoceptive information.
Cardioceptive accuracy did not correlate significantly It could be argued that enhanced cardiac perception
with depression (r = 0.45, n = 12, P = 0.14), state anxi- may feed into dysfunctional cognitive appraisal. This can
ety (r = 0.27, n = 14, P = 0.36) or trait anxiety in the be well integrated into classical vicious-circle models of
PD group (r = 0.01, n = 15, P = 0.97) or the control PD (Ehlers and Margraf 1989). According to these mod-
group (depression: r = 0.37, n = 13, P = 0.21; state anxi- els, perception of symptoms leads to catastrophic inter-
ety: r = 0.37, n = 15, P = 0.18; trait anxiety: r = 0.01, pretations, thus increasing autonomic arousal and
n = 15, P = 0.97). physical symptoms that can be perceived as threatening.
Avoidance of associated eliciting cues then leads to
Discussion
2
Please note the current discussion on related terminology and
In this study, we investigated whether cardioceptive what exactly is assessed with the mental-tracking task (Ceunen
accuracy in patients with PD predicts performance in a et al. 2013).

6 ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc.
J. Wölk et al. Cardiac Perception and Decision Making

generalization and maintenance of PD. In complex deci- due to error variance. Future studies should therefore pay
sion-making tasks this may have detrimental effects, when particular attention to such moderating factors. Third, it
attention to associated information is withdrawn due to has been suggested that the mental-tracking task may
generalized avoidance of somatic cues as described in the reflect cognitive ability rather than interoceptive skill.
SMH. Although the present task was designed to prevent the
Future studies should examine real-life decision making respective strategy of estimating the time passed during
in PD patients based on such models. This could well counting of heartbeats, we cannot completely rule out
lead to better explanations why PD patients’ history is that cognitive ability is confounded with cardiac percep-
often characterized by decision difficulties (Ludewig et al. tion scores. Nevertheless, the reverse association of high-
2003; Lorian and Mahoney 2012). cardiac perception and decision-making performance
Although the group difference was not significant, con- found in panic patients would still be at odds with this
trols exercised about twice as long per week as panic alternative explanation. Moreover, it appears quite hard
patients. Although cardiac perception was similar in both to find a plausible explanation why high-cognitive skill
groups and the correlations with decision making were may predict impaired decision making. Hence, we con-
not stronger in the control group, this indicates that sider our interpretation as the more parsimonious one.
panic patients may be less familiar with experiencing car- Fourth, the small sample size limits the generalizability of
diac symptoms in a safe context. In patients with high our results, therefore, warranting future replications in
cardiac perception, this may further add to the presumed larger samples. An independent replication of these results
detrimental effect of experiencing cardiac somatic cues on with a larger sample size and consequently more hetero-
decision making. From a clinical point of view, it may geneous sample (e.g., including typical comorbidities)
therefore be interesting to address such a potential associ- could increase the generalizability of our conclusions.
ation of (cardiac) somatic cues with panic-related (nega- Nevertheless, several observations support the reliability
tive) associations. Symptom-focused exposure (e.g., of the current results. First, the correlations are similar to
elicited by physiological provocation tasks and discrimi- previously published results on interoceptive accuracy and
nation learning) could help to weaken these associations. anxiety (Pollatos et al. 2007; Domschke et al. 2010) or
Once cardiac symptoms are not experienced as threaten- those achieved with a similar task (Sütterlin et al. 2013).
ing anymore, this may also withdraw the basis for the Second, the sample was homogeneous and carefully
side effects of (cardiac) somatic markers on decision mak- selected to minimize effects of comorbidity and medica-
ing as delineated above. Screening for cardiac perception tion, and third the results do not depend on outliers, sin-
may help identify patients who may profit from such an gle participants or small groups of individuals (see
approach. Figs. 2, 3). Fifth, the test situation in the laboratory may
have induced moderate stress in all participants. This may
have induced particular attention to or salience of cardiac
Limitations
cues in some PD patients. On the other hand, a similarly
First, it should be noted, that almost no patient reached stressful context is present in many real-life situations
the stage of explicitly understanding the effects of select- requiring decision making, hence, this could also be seen
ing a particular deck in the IGT. Therefore, our findings as a factor increasing the ecological validity of this study,
only apply to the stage where participants decide ran- improving generalizability of these findings to real-life
domly or rely on a hunch. However, it is clear from situations.
somatic marker theory that this is the stage where cardio-
ception would be considered to have the largest impact
Conclusion
on behavior. Second, we did not find a main effect of
group. This is at odds with previous studies suggesting This is the first study to demonstrate a detrimental effect
that PD patients may generally have higher cardiac per- of enhanced cardiac perception on intuitive decision mak-
ception (for a review see, Domschke et al. 2010). How- ing in PD patients. These findings extend previous results
ever, not all previous studies have found this difference, indicating that PD patient experience heightened aware-
and the small to medium effect size in our study could ness and dysfunctional cognitive processing of somatosen-
become significant in a larger sample. Comparing these sory cues, particularly cardiac activity (Hofmann et al.
studies, it is interesting that both participants recruited 2008) and associated bias (Amrhein et al. 2005). Together
from inpatient versus outpatient settings occur. Maybe with cognitive models comprising of a vicious circle
differences in symptom severity, comorbidity, treatment involving vigilance-anxiety-avoidance (Clark et al. 1988),
intensity, time since diagnosis, etc. contribute to a larger our findings provide new insight into the basis for detri-
heterogeneity in our patient group, reducing effect size mental decision making in PD patients.

ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc. 7
Cardiac Perception and Decision Making J. Wölk et al.

Damasio, A. R. 1995. Descartes’ error: emotion, reason, and

Acknowledgments the human brain. Putman, New York, NY.
We thank all volunteers for their participation, the staff Damasio, A. R., D. Tranel, and H. Damasio. 1991. Somatic
at Schön Klinik Roseneck as well as Professor Paul Pauli, markers and the guidance of behaviour: theory and
Department of Psychology I, University of Würzburg, preliminary testing. Pp. 217–229 in H. S. Levin, H. M.
Germany for making this study possible. Preparation of Eisenberg, and A. L. Benton, eds. Frontal lobe function and
this manuscript was supported by the Federal Ministry of dysfunction. Oxford University Press, New York, NY.
Education and Research, BMBF, project 01EO1004 (S. M. Davis, C., J. Fox, K. Patte, C. Curtis, R. Strimas, C. Reid, et al.
S., Comprehensive Heart Failure Center [CHFC], Univer- 2008. Education level moderates learning on two versions of
sity of Würzburg, Germany). the Iowa Gambling Task. J. Int. Neuropsychol. Soc.
14:1063–1068.
Domschke, K., S. Stevens, B. Pfleiderer, and A. L. Gerlach.
Conflict of Interest 2010. Interoceptive sensitivity in anxiety and anxiety
None declared. disorders: an overview and integration of neurobiological
findings. Clin. Psychol. Rev. 30:1–11.
References Dunn, B. D., T. Dalgleish, and A. D. Lawrence. 2006. The
somatic marker hypothesis: a critical evaluation. Neurosci.
Amrhein, C., P. Pauli, W. Dengler, and G. Wiedemann. 2005. Biobehav. Rev. 30:239–271.
Covariation bias and its physiological correlates in panic Ehlers, A. 1993. Interoception and panic disorder. Adv. Behav.
disorder patients. J. Anxiety Disord. 19:177–191. Res. Ther. 15:3–21.
American Psychological Association. 2000. Diagnostic and Ehlers, A., and P. Breuer. 1996. How good are patients with
statistical manual of mental disorders. DSM-IV-TR. panic disorder at perceiving their heartbeats? Biol. Psychol.
American Psychological Association, Washington, DC. 42:165–182.
Bandelow, B. 1997. Panik und agoraphobie-skala (PAS). Ehlers, A., and J. Margraf. 1989. The psychophysiological
Hogrefe, Göttingen. model of panic attacks. Swets, Amsterdam, Netherlands.
Barsky, A. J. 1992. Palpitations, cardiac awareness, and panic Ehlers, A., P. Breuer, D. Dohn, and W. Fiegenbaum. 1995.
disorder. Am. J. Med. 92:31S–34S. Heartbeat perception and panic disorder: possible
Bechara, A., A. R. Damasio, H. Damasio, and S. W. Anderson. explanations for discrepant findings. Behav. Res. Ther.
1994. Insensitivity to future consequences following damage 33:69–76.
to human prefrontal cortex. Cognition 50:7–15. Eley, T.C., L. Stirling, A. Ehlers, A. M. Gregorya, and D. M.
Bechara, A., H. Damasio, D. Tranel, and A. R. Damasio. 1997. Clark. 2004. Heart-beat perception, panic/somatic symptoms
Deciding advantageously before knowing the advantageous and anxiety sensitivity in children. Behav. Res. Ther. 42:439–
strategy. Science 275:1293–1295. 448.
Bouton, M. E., S. Mineka, and D. H. Barlow. 2001. A modern Giorgetta, C., A. Grecucci, S. Zuanon, L. Perini, M. Balestrieri,
learning theory perspective on the etiology of panic N. Bonini, et al. 2012. Reduced risk-taking behavior as a
disorder. Psychol. Rev. 108:4–32. trait feature of anxiety. Emotion 12:1373–1383.
Brown, T. A., L. A. Campbell, C. L. Lehman, J. R. Grisham, Grosche, C., S. S€utterlin, S. Koch, S. M. Schulz, and C. V€ogele.
and R. B. Mancill. 2001. Current and lifetime comorbidity 2011. Interozeption und Inhibitionskapazit€at bei
of the DSM-IV anxiety and mood disorders in a large Panikst€ orung und Somatoformen St€ orungen.
clinical sample. J. Abnorm. Psychol. 110:585–599. Verhaltenstherapie (Basel) 21(S1):1–39.
Carleton, R. N., M. K. Mulvogue, M. A. Thibodeau, R. E. Hartl, L. 1995. A clinical approach to cardiac perception. Pp.
McCabe, M. M. Antony, and G. J. Asmundson. 2012. 251–263 in D. V. R. Schandry, ed. From the heart to the
Increasingly certain about uncertainty: intolerance of brain. The psychophysiology of circulation–brain
uncertainty across anxiety and depression. J. Anxiety Disord. interaction. Frankfurt, Lang.
26:468–479. Hautzinger, M., F. Keller, and C. Kühner. 2006. BDI-II:
Ceunen, E., I. Van Diest, and J. W. Vlaeyen. 2013. Accuracy and depressions-inventar revision. Harcourt Test Services,
awareness of perception: related, yet distinct (commentary on Frankfurt am Main, Germany.
Herbert et al., 2012). Biol. Psychol. 92:426–427. Herbert, B. M., C. Herbert, O. Pollatos, K. Weimer, P. Enck,
Clark, D. M., P. M. Salkovskis, M. Gelder, C. Koehler, M. H. Sauer, et al. 2012. Effects of short-term food deprivation
Martin, P. Anastasiades, et al. 1988. Tests of a cognitive on interoceptive awareness, feelings and autonomic cardiac
theory of panic. Pp. 149–158 in H. H. U. Wittchen, ed. activity. Biol. Psychol. 89:71–79.
Panic and phobias 2: treatments and variables affecting Hoehn-Saric, R., D. R. McLeod, F. Funderburk, and
course and outcome. Springer, Berlin, Heidelberg, New P. Kowalski. 2004. Somatic symptoms and physiologic
York, London, Paris, Tokyo. responses in generalized anxiety disorder and panic disorder:

8 ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc.
J. Wölk et al. Cardiac Perception and Decision Making

an ambulatory monitor study. Arch. Gen. Psychiatry 61: Ludwick-Rosenthal, R., and R. W. Neufeld. 1985. Heart beat
913–921. interoception: a study of individual differences. Int. J.
Hofmann, S. G., G. W. Alpers, and P. Pauli. 2008. Psychophysiol. 3:57–65.
Phenomenology of panic and phobic disorders. Pp. 34–46 in Mahoney, A. E., and P. M. McEvoy. 2012. Trait versus
M. M. Antony, and M. B. Stein, eds. Oxford handbook of situation-specific intolerance of uncertainty in a clinical
anxiety and related disorders. Oxford University Press, New sample with anxiety and depressive disorders. Cogn. Behav.
York, NY. Ther. 41:26–39.
Jones, G. E., K. R. Jones, C. H. Rouse, D. M. Scott, and J. A. Montgomery, W. A., G. E. Jones, and J. G. Jr Hollandsworth.
Caldwell. 1987. The effect of body position on the 1984. The effects of physical fitness and exercise on cardiac
perception of cardiac sensations: an experiment and awareness. Biol. Psychol. 18:11–22.
theoretical implications. Psychophysiology 24:300–311. Peterson, R., and K. Plehn. 1999. Measuring anxiety sensitivity.
Kaplan, J. S., K. Erickson, D. A. Luckenbaugh, P. Pp. 61–82 in S. Taylor, ed. Anxiety sensitivity–theory,
Weiland-Fiedler, M. Geraci, B. J. Sahakian, et al. 2006. research, and treatment of the fear of anxiety. Lawrence
Differential performance on tasks of affective processing and Erlbaum Associates Publishers, Mahwah, NJ.
decision-making in patients with panic disorder and panic Pollatos, O., K. Gramann, and R. Schandry. 2007. Neural
disorder with comorbid major depressive disorder. J. Affect. systems connecting interoceptive awareness and feelings.
Disord. 95:165–171. Hum. Brain Mapp. 28:9–18.
Kaufmann, J., and D. Charney. 2000. Comorbidity of mood Richards, J. C., A. J. Cooper, and J. H. Winkelman. 2003.
and anxiety disorders. Depression and Anxiety 12:69–76. Interoceptive accuracy in nonclinical panic. Cogn. Ther. Res.
Kaufmann, T., S. Sütterlin, S. M. Schulz, and C. Vogele. 2011. 27:447–461.
ARTiiFACT: a tool for heart rate artifact processing and heart Sütterlin, S., S. M. Schulz, T. Stumpf, P Pauli, and C V€ ogele,
rate variability analysis. Behav. Res. Methods 43:1161–1170. et al. 2013. Enhanced cardiac perception is associated with
Krohne, H. W., B. Egloff, C. W. Kohlmann, and A. Tausch. increased susceptibility to framing effects. Cogn. Sci. 37:922–
1996. Positive and negative affect schedule (PANAS) – 935.
German version. Diagnostica 42:139–156. Werner, N. S., K. Jung, S. Duschek, and R. Schandry.
Laux, L., P. Glanzmann, P. Schaffner, and C. Spielberger. 1981. 2009. Enhanced cardiac perception is associated with
State-trait-angstinventar (STAI): theoretische grundlagen benefits in decision-making. Psychophysiology 46:1123–
und handanweisungen. Beltz, Weinheim, Germany. 1129.
Lorian, C. N., A. Mahoney, and J. R. Grisham. 2012. Playing it Wittchen, H.-U., M. Zaudig, and T. Fydrich. 1997. SKID-I
safe: an examination of risk-avoidance in an anxious strukturiertes klinisches interview für DSM-IV: achse I:
treatment-seeking sample. J. Affect. Disord. 14:63–71. psychische störungen. Hogrefe, Göttingen, Germany.
Ludewig, S., M. P. Paulus, K. Ludewig, and F. X. Zoellner, L. A., and M. G. Craske. 1999. Interoceptive accuracy
Vollenweider. 2003. Decision-making strategies by panic and panic. Behav. Res. Ther. 37:1141–1158.
disorder subjects are more sensitive to errors. J. Affect.
Disord. 76:183–189.

ª 2013 The Authors. Brain and Behavior published by Wiley Periodicals, Inc. 9