Review
The somatic marker hypothesis: A critical evaluation
Barnaby D. Dunn*, Tim Dalgleish, Andrew D. Lawrence
MRC Cognition and Brain Sciences Unit, 15 Chaucer Road, Cambridge CB2 2EF, UK
Abstract
The somatic marker hypothesis (SMH; [Damasio, A. R., Tranel, D., Damasio, H., 1991. Somatic markers and the guidance of behaviour:
theory and preliminary testing. In Levin, H.S., Eisenberg, H.M., Benton, A.L. (Eds.), Frontal Lobe Function and Dysfunction. Oxford
University Press, New York, pp. 217–229]) proposes that emotion-based biasing signals arising from the body are integrated in higher brain
regions, in particular the ventromedial prefrontal cortex (VMPFC), to regulate decision-making in situations of complexity. Evidence for the
SMH is largely based on performance on the Iowa Gambling Task (IGT; [Bechara, A., Tranel, D., Damasio, H., Damasio, A.R., 1996. Failure
to respond autonomically to anticipated future outcomes following damage to prefrontal cortex. Cerebral Cortex 6 (2), 215–225]), linking
anticipatory skin conductance responses (SCRs) to successful performance on a decision-making paradigm in healthy participants. These
‘marker’ signals were absent in patients with VMPFC lesions and were associated with poorer IGT performance. The current article reviews
the IGT ﬁndings, arguing that their interpretation is undermined by the cognitive penetrability of the reward/punishment schedule, ambiguity
surrounding interpretation of the psychophysiological data, and a shortage of causal evidence linking peripheral feedback to IGT
performance. Further, there are other well-speciﬁed and parsimonious explanations that can equally well account for the IGT data. Next,
lesion, neuroimaging, and psychopharmacology data evaluating the proposed neural substrate underpinning the SMH are reviewed. Finally,
conceptual reservations about the novelty, parsimony and speciﬁcation of the SMH are raised. It is concluded that while presenting an elegant
theory of how emotion inﬂuences decision-making, the SMH requires additional empirical support to remain tenable.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: Emotion; Iowa gambling task; Decision-making; Ventromedial prefrontal cortex; Body
Contents
1. Development of the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
1.1. Development of the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
1.2. The somatic marker hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
1.3. The Iowa gambling task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
1.4. Review of Iowa laboratory studies using the IGT in support of the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
2. Evaluation of the Iowa gambling task as key evidence for the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
2.1. Strengths of the IGT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
2.2. Cognitive penetrability of the IGT reward/punishment schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
2.3. Interpretation of psychophysiological data on the IGT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
2.4. Variability in control participant performance on the IGT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
2.5. Speciﬁcity and reliability of the IGT in clinical populations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
2.6. The causal relationship between body-state feedback and IGT performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
2.7. Task design issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
2.8. Alternative mechanisms potentially underlying IGT task performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
2.8.1. Working memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
2.8.2. Reversal learning/inhibition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Neuroscience and Biobehavioral Reviews 30 (2006) 239–271
www.elsevier.com/locate/neubiorev
0149-7634/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neubiorev.2005.07.001
* Corresponding author. Tel.: C44 1223 355 294; fax: C44 1223 359 062.
E-mail address: barney.dunn@mrc-cbu.cam.ac.uk (B.D. Dunn).
2.8.3. Risk-taking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
2.8.4. Insensitivity to rewarding and punishing outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
2.8.5. Apathy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
2.8.6. Overview of alternative mechanisms accounting for IGT deﬁcits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
2.9. Current status of the IGT data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
3. Evaluation of the proposed neural substrate of the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
3.1. Lesion data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
3.2. Neuroimaging data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
3.3. Psychopharmacology data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
3.4. Overview of neural substrate underpinning the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4. Further conceptual issues with the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4.1. Novelty of the SMH: a historical overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
4.2. Speciﬁcation of the somatic marker mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
4.3. Parsimony of the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
4.4. Emphasis on the periphery in the SMH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
5. Conclusion: current status of the somatic marker hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Perhaps the most inﬂuential conceptualisation of the
‘emotional brain’ over the last century has been the limbic
system (MacLean, 1949), an anatomical framework that
outlines how emotions and moods are embodied in neural
architecture. A core theoretical aspect of the limbic system
framework is that emotion experiences arise from the
integration of sensations from the external world with
information from the body, speciﬁcally feedback from the
viscera (MacLean, 1949, 1975: for a review, see Dalgleish,
2004). The notion that emotion experience emerges from
feedback from the body in tandem with higher level
representation of the world partially echoes the seminal
theories of emotion put forward by James and Lange (James,
1884, 1894; Lange, 1885). These controversial models
argued that emotion experience arises directly from the
perception of change in the body: when we run from a bear in
the woods, we are afraid because we run rather than we run
because we are afraid (for reviews see Adelmann & Zajonc,
1989; Mandler, 1990; Laird and Bresler, 1990; Izard, 1990;
Lang, 1994; Ellsworth, 1994; Reisenzein et al., 1995;
Damasio, 2004; Prinz, 2004).
Antonio Damasio has recently extended the inﬂuence of
somatic processes to the regulation of decision-making as
well as emotion in his inﬂuential somatic marker hypothesis
(henceforth, SMH; Damasio et al., 1991; Damasio, 1994,
1996, 2004). The SMH proposes that ‘somatic marker’
biasing signals from the body are represented and regulated
in the emotion circuitry of the brain, particularly the
ventromedial prefrontal cortex (VMPFC), to help regulate
decision-making in situations of complexity and uncertainty
(e.g Damasio, 1996; Bechara et al., 2000a,b).
The proposed neural circuitry underlying the SMH
departs from the limbic system in an anatomical sense, in
that it incorporates a variety of brain regions outside of the
classic limbic system structures (MacLean, 1949), including
ventromedial prefrontal cortex (VMPFC), somatosensory
cortices, insula, and basal ganglia (Damasio, 1998). Further,
Damasio extends the function of the limbic system beyond
that of a ‘visceral brain’ (MacLean, 1949) in arguing that
multiple sources of feedback from the periphery (visceral,
somatosensory, and others) shape decision-making (see
Damasio, 2004).
Empirical support for the SMH is largely based on
performance on the Iowa Gambling Task (IGT; Bechara
et al., 1994, 1996), an experimental paradigm designed to
measure decision-making. A correlation has been found
between successful IGT performance and the development
of somatic marker signals (as indexed by the magnitude of
anticipatory skin conductance responses [SCRs]) in healthy
control participants. Crucially, these bodily signals were
found to be absent in people with VMPFC lesions and this
was linked to their poorer performance on the decisionmaking
task.
This article will present a review of the extant research
testing the central claims of the SMH, particularly focusing
on the role of somatic processes in decision-making as
measured by performance on the IGT. Section 1 provides a
descriptive, non-critical account of the evolution of the
SMH and the support the IGT offers to the framework as
described by Damasio’s Iowa laboratory. To examine the
psychological component of the SMH, Section 2 critically
appraises the extent to which the IGT data can validate the
SMH. It is proposed that recent empirical ﬁndings mean that
the IGT, while well validated and extensively applied to a
variety of neurological and psychiatric conditions in its
behavioural form, will no longer sufﬁce as strong evidence
for the SMH. To examine the neural component of the
SMH, Section 3 evaluates the extent to which the neural
substrate Damasio put forward for the SMH has been
empirically validated, concluding that there is reasonable
evidence for this aspect of the model. Section 4 raises and
evaluates some further conceptual concerns about the
novelty and parsimony of the SMH and also discusses
how well the theory has been speciﬁed. Our review of
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271240
the evidence will suggest that additional empirical support
and a clearer conceptualisation are needed to validate
further this intriguing and potentially useful theoretical
framework. Some suggestions for alternative research
designs to aid this process will be put forward.
1. Overview of the SMH and its evolution
1.1. Development of the SMH
The SMH grew from attempts to understand the striking
emotional and everyday decision-making deﬁcits displayed
by patients with damage to VMPFC, the portions of the
frontal lobes above the eye sockets. Damasio (1994)
reviewed how damage to the VMPFC can have profound
effects on work and social function without inducing any
obvious impairments in intellect and cognitive performance,
focusing particularly on the famous cases of Phineas Gage
(see Harlow, 1868 for an overview) and Elliot (EVR; see
Damasio, 1979; Eslinger and Damasio, 1985; Damasio et al.,
1991).
In 1848, Gage was a successful foreman on the railway
and kept a respectable social life. Following a blasting
accident, however, a tamping iron went through his eye
socket and extensively damaged his frontal cortex before
passing out through the top of his skull. Remarkably he
survived and at ﬁrst appearance seemed to have no
intellectual impairment. Gage started to display odd
decision-making and social behaviours, however. Whereas
before he had been a conservative family man, he
subsequently could not hold down a job, made risky ﬁnancial
decisions and his family relations broke down. Modern
neuroimaging techniques were much later applied to identify
the passage of the tamping iron through Gage’s skull
(Damasio et al., 1994) and it was found to ablate a portion
of the frontal lobe centred around the VMPFC (although see
MacMillan, 2000 for a critique of Damasio’s account of
Gage).
Similarly, EVR presented to Damasio’s Iowa laboratory
after he suffered a brain tumour, which led to bilateral
ablation of the VMPFC and related areas (Damasio, 1979,
1994; Eslinger and Damasio, 1985; Damasio et al., 1991).
EVR became unable to make decisions, especially in the
social and personal domain. He could not plan for the future
and tended to choose unsuitable friends, business partners
and activities. Other cases of VMPFC damage that produce
broadly equivalent impairment have also been documented
(e.g. Dimitrov et al., 1999; Barrash et al., 2000). The
syndrome suffered by these patients has been documented as
‘acquired sociopathy’, reﬂecting the fact that the personality
and decision-making effects of damage to this region
resemble a milder form of those seen in sociopathy (Damasio
et al., 1990, 1991; Tranel, 1994).
The Iowa laboratory conducted an elegant series of
studies to attempt to elucidate the cause of the difﬁculties in
day-to-day living displayed by cases such as EVR.
Intriguingly, the peculiar choices EVR made in real life
were at odds with initial laboratory assessments of his
reasoning capabilities, which showed normal, or on many
tasks superior, intellectual performance. For example,
working memory, attention, cognitive estimation, cognitive
ﬂexibility, recency of event judgement and even social
knowledge were all unimpaired (Damasio et al., 1991; Saver
and Damasio, 1991). Later investigations identiﬁed that
EVR and other cases with damage to VMPFC had a
difﬁculty in expressing emotion and experiencing feelings.
Neuropsychological investigation showed that VMPFC
damage altered psychophysiological response and reported
emotion experience to emotional but not neutral stimuli
(Damasio et al., 1990, 1991; Tranel, 1994). This led
Damasio to speculate that these emotional changes were
the cause of decision-making difﬁculties seen following
VMPFC damage.
1.2. The somatic marker hypothesis
In his inﬂuential book ‘Descartes’ Error’, Damasio
(1994) most famously articulated the SMH. Building on
his earlier paper on the consequences of VMPFC damage
(Damasio et al., 1991), he argued that the decision-making
deﬁcits found following VMPFC damage were due to an
inability to use emotion-based biasing signals generated
from the body (or ‘somatic markers’) when appraising
different response options (see also Damasio, 1996, 2004).
In brief, the SMH postulated that reasoning is inﬂuenced
by crude biasing signals arising from the neural machinery
that underlies emotion. For Damasio, emotion is the
representation and regulation of the complex array of
homeostatic changes that occur in different levels of the
brain and body in given situations. When making decisions,
a crude biasing signal (a somatic marker) arising from the
periphery or the central representation of the periphery
indicates our emotional reaction to a response option. For
every response option contemplated, a somatic state is
generated, including sensations from the viscera, internal
milieu, and the skeletal and smooth muscles (see Damasio,
2004). These somatic markers serve as an indicator of the
value of what is represented and also as a booster signal for
continued working memory and attention (Damasio et al.,
1991; Damasio, 1996). Particularly in situations of
complexity and uncertainty, these marker signals help to
reduce the problem space to a tractable size by marking
response options with an ‘emotional’ signal. Only those
options that are marked as promising are processed in a full,
cognitive fashion (Damasio, 1994, 1996; Bechara and
Damasio, in press).
Somatic markers can reﬂect actions of the body proper
(the ‘body’ loop) or the brain’s representation of the action
expected to take place in the body (the ‘as-if’ loop). In other
words, the brain can construct a forward model of changes it
expects in the body, allowing the organism to respond more
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 241
rapidly to external stimuli without waiting for that activity
to actually emerge in the periphery. This is similar to
accounts of how motor control of the periphery is regulated
by advance modelling (e.g. Wolpert and Ghahramani,
2000). Perception of somatic state information makes us
more likely to approach or withdraw from a situation. These
signals can function at an overt level (where the individual
is consciously aware of the emotions and bodily changes
associated with a particular response option) or at a covert
level (where the individual is unaware of his/her emotions
and bodily activity).
Decision-making can be viewed as a combination of
‘high reason’, carrying out a logical cost-beneﬁt analysis of
a given action, and marker signals, indicating how
rewarding or punishing an action is likely to be in complex
situations where more detailed cost-beneﬁt analysis is not
possible (Damasio et al., 1991; Damasio, 1994, 1996, 2004).
Damasio has argued that damage to the VMPFC and
other structures involved in the representation and regulation
of body-state (including amygdala, insula, somatosensory
cortex, cingulate, basal ganglia and brain-stem
nuclei) leads to impaired decision-making because the
somatic marking system can no longer be activated (see
Fig. 1 which outlines the proposed ‘somatic marker’
network and Section 3 evaluating the proposed neural
substrate of the SMH). The VMPFC is believed to be the
crucial area of the brain that integrates actual or predicted
bioregulatory state representations with potential response
options, so is central to the generation of somatic markers.
Cases such as Gage and EVR would therefore only be able
to make decisions based on a logical cost-beneﬁt analysis
and would be unable to utilise prior emotion experience to
guide them towards previous advantageous choices and
away from disadvantageous choices. In uncertain situations,
where a complete logical analysis of the situation is not
possible, such a proﬁle would lead to a marked decisionmaking
impairment characterised by either extreme
procrastination or the selection of inappropriate response
options that would be immediately dismissed by people with
intact marker signals. This proﬁle has been characterised as
‘myopia for the future’, where the individual is unable to
predict long-term punishments and rewards based on
previous experience (e.g. Bechara et al., 1994).
1.3. The Iowa gambling task
As noted already, a major plank of empirical evidence
supporting the SMH stemmed from the IGT, an experimental
paradigm designed to mimic real life decision-making
situations in the way it factors uncertainty, reward and
punishment (see Bechara et al., 1994, 1996, 1999, 2000a for
full details). A key feature of this task is that participants
have to forego short-term beneﬁt for long-term proﬁt.
The task requires participants to select from one of four
decks of cards that are identical in physical appearance for
100 trials. Each card choice leads to either a variable
ﬁnancial reward or a combination of a variable ﬁnancial
reward and penalty. Unknown to participants, the rewards
and punishments on the decks have been ﬁxed by the
experimenter. For each selection from decks A and B
participants win $100 and from each selection from decks C
and D participants win $50. Every so often variable
punishment is also given. On deck A, ﬁve in ten trials
generate a penalty ranging from $35 to $150. On deck B,
one in ten trials incurs a penalty of $1250. On deck C, ﬁve in
ten trials involve a penalty ranging from $25 to $75. Finally,
Body Loop ‘As if’ Loop
Sensory and neurotransmitter nuclei (e.g.
substantia niagra, dorsal raphe nucleus)
Effector structures (e.g. hypothalamus,
autonomic centres, peri-aqueductal grey)
Brainstem Nuclei
VMPFC
Body
SMC/
insula
AM
Brainstem Nuclei
VMPFC
Body
SMC/
insula
AM
Fig. 1. Neuralarchitectureimplicatedin the SMH. The left-handside illustrates a schematic of the ‘body’ loop and the right-handside illustrates a schematic of the
‘as-if’ loop. VMPFC, ventromedial prefrontal cortex; AM, amygdala; SMC, somatosensory cortex. Figure reproduced from Bechara and Damasio (2005).
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271242
on deck D, one in ten trials gives a penalty of $250. Overall,
the high reward decks (A and B) give higher levels of
punishment (so leading to a net loss of $250 every 10 trials),
whereas the low reward decks (C and D) give lower levels
of punishment (so leading to a net gain of $250 every 10
trials). Thus, successful task performance relies on sampling
more from decks C and D than from decks A and B. There is
no advantage for participants in selecting more from the
frequent punishment (A and C) versus infrequent punishment
(B and D) decks, or vice versa. Crucially, it is argued
that this reward/punishment schedule is opaque, such that
participants are unlikely to be able to perform an exact
calculation of net gains and losses. To do well, it is therefore
claimed that participants must rely on more ‘intuitive’
decision-making processes, in particular the activation of
somatic marker biasing signals (Bechara et al., 1994, 1996).
Fig. 2 presents an overview of the rewards and punishments
on each deck of the IGT.
Prior to completing the task, participants are instructed
that the game requires them to choose cards from any one of
the four decks until they are told to stop. They are not given
information about how long the task will go on nor when it
will stop and it is explicitly stated that they can change deck
whenever they wish. The goal of the game is deﬁned as to
win as much money as possible and to avoid losing money
as far as possible. Participants are told that while there is no
way for them to work out when they will lose money, they
will ﬁnd that some decks are worse than others and that to do
well they need to stay away from the worst decks.
In the original manual version, participants are seated in
front of four decks of cards of identical physical appearance
and are given a $2000 loan of money (facsimile US$ bills).
A computerised version of the task displaying the cards on a
computer screen and using a visual representation of proﬁts
and losses (a green bar increases or decreases in size on the
screen to record total money held by the participants) has
also been subsequently developed.
A central aspect of the IGT is that SCRs have been found
to be associated with successful learning. When recording
SCR data, a modiﬁed version of the task is used that has a
longer interval between trials (typically greater than 15 s),
thus allowing SCR activity to return to baseline before the
next selection. Analysis of these data has looked at whether
the SCRs generated in anticipation of deck selection,
following reward, and following combined reward and
punishment vary as a function of which deck is chosen (e.g.
Bechara et al., 2000a).
1.4. Review of Iowa laboratory studies using the IGT
in support of the SMH
Bechara et al. (1994) ﬁrst described the performance of
patients with damage encompassing but not restricted to
VMPFC on the task. The VMPFC patients (nZ6) were
signiﬁcantly worse at the IGT than healthy control volunteers
(nZ44). Over time the control group learned to select more
from the ‘safe’, winning decks (C and D) than the ‘risky’,
losing decks (A and B), whereas the VMPFC lesion group
continued to prefer the disadvantageous decks for the
duration of the task. Interestingly, patient EVR was tested
on multiple occasions of the task and failed to learn, whereas
control participants performance improved on subsequent
testing. The authors concluded that the VMPFC lesion group
decisions were driven more by the immediate reward than the
delayed punishment available on the disadvantageous decks
(‘myopia for the future’ Bechara et al., 1994).
Key support for the hypothesised role of somatic
markers in performance on the IGT derived from
identiﬁcation of a physiological correlation of success
on the decision-making task (Bechara et al., 1996). SCRs
were measured in seven patients with frontal lobe damage
encompassing VMPFC and 12 normal controls during task
performance. Both patients and controls showed reward
and punishment SCRs. After a short period of time,
however, the control group also started to develop
anticipatory SCRs, which were larger for selections from
the ‘risky’ decks than the ‘safe’ decks. The absence of
anticipatory SCRs in the VMPFC lesion group correlated
with impaired task performance. It was speculated that
patients were failing to activate somatic markers to help
them to distinguish between good and bad outcomes
in situations of uncertainty (Bechara et al., 1996). In
particular, failure to activate a negative marking signal for
the disadvantageous decks based on previous punishment
history would make the VMPFC lesion group insensitive
to the possibility of further future punishment on these
decks. Fig. 3 presents the typical behavioural and skin
conductance proﬁles found for VMPFC lesion patients
and healthy control volunteers on the IGT.
A B C D
Pick a card!
$0
Reward Punishment Net Profit
(over 10 trials)
A Win $100
(100% trials)
Lose $150 to $35
(50% trials)
- $250
B Win $100
(100% trials)
Lose $1250
(10% trials)
- $250
C Win $50
(100% trials)
Lose $25 to $75
(50% trials)
+ $250
D Win $50
(100% trials)
Lose $250
(10% trials)
+ $250
$500 $1000 $1500 $2000
Fig. 2. Graphical representation of IGT. On each of 100 trials, participants
must select one of four decks of cards, from which they receive either a
reward or a combined reward and punishment. Decks A and B offer shortterm
rewards but long-term punishments, leading to a net loss. Decks C and
D offer reduced immediate rewards but smaller long-term punishments,
leading to a net proﬁt. Acquisition of the task is measured by number of
selections from the advantageous decks relative to the disadvantageous
deck on each block of 20 trials.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 243
As noted earlier, an important claim is that the
reward/punishment schedule of the IGT is opaque, so
learning is therefore taking place at a non-declarative,
largely implicit level. To test this assertion, Bechara et al.
(1997) asked ten healthy participants and six VMPFC lesion
patients to complete the task and stopped them after every
10 trials to see whether they knew consciously what was
going on in the game. Analysis indicated that participants
went through four phases: pre-punishment (no punishment
yet encountered), pre-hunch (no understanding of what was
going on), hunch (hypotheses generated on which were
‘good’ and ‘bad’ decks), and the conceptual period (clear
idea of what was going on). Around 30% of control
participants did not reach the conceptual period, despite
performing normally on the task. Anticipatory SCR activity
and increased selection from the good decks began to take
place for the control group in the pre-hunch period and was
sustained throughout the task. This was taken to suggest that
implicit learning was taking place prior to explicit understanding
of the paradigm. Notably, 50% of the VMPFC
patients did reach the conceptual period but still performed
disadvantageously on the task.
To support the ‘myopia for the future’ interpretation of the
deﬁcit displayed by VMPFC lesion patients on the IGT,
Bechara et al. (2000b) examined performance of healthy
control volunteers and lesion patients on two modiﬁed forms
of the task. In the ﬁrst of these, the reward and punishment
contingencies were reversed such that differences in reward
determined the optimal response. The advantageous decks
yielded higher immediate punishment but even greater
delayed reward, whereas the disadvantageous decks gave
low immediate punishment but even lower future reward.
VMPFC lesion patients were impaired on this version of the
task also relative to control volunteers, indicating that the
mechanism underpinning task impairment is unlikely to be
loss of sensitivity to either punishment or reward cues. This
study did not report anticipatory psychophysiology data,
however, so it is unclear if a similar SCR proﬁle emerged in
the variant task. The second variant task investigated whether
decision-making deﬁcits in VMPFC lesion patients could be
normalised by increasing the adverse future consequences
associated with the ‘risky’ decks. Even after increasing the
delayed punishment and decreasing the delayed reward on
the disadvantageous decks, the VMPFC lesion patients were
impaired on the task relative to controls (Bechara et al.,
2000b).
In summary, work from the Iowa laboratory has found
that VMPFC lesions impair the ability to select from the
advantageous decks on the IGT and that this behavioural
deﬁcit is associated with the absence of anticipatory SCR
signals that are known to differentiate between the good and
bad decks in healthy control volunteers. Manipulation of the
reward/punishment schedule used supports the conclusion
that this behavioural deﬁcit following VMPFC lesions
reﬂects ‘myopia for the future’ (Bechara et al., 1994) rather
than altered sensitivity to reward and punishment. Further,
0
5
10
15
20
1 2 3 4 5
Disadvantageous
Decks (A + B)
Advantageous
Decks (C + D)
Block of 20 Trials
No.PicksfromDecks
Control Participants Behavioural Performance
0
5
10
15
20
1 2 3 4 5
Disadvantageous
Decks (A + B)
Advantageous
Decks (C + D)
Block of 20 Trials
No.PicksfromDecks
VMPFC Patients Behavioural Performance
0
0.2
0.4
0.6
0.8
1
A B C D
VMPFC Lesion
Group
Control Group
AnticipatorySCRMagnitude(microsiemens)
Control Participants and VMPFC Patients
Psychophysiological Response
Fig. 3. Typical behavioural and psychophysiology results on the IGT. Healthy control participants (top left panel) increasingly select from the advantageous
decks over time, whereas VMPFC lesion patients (bottom left panel) opt for the disadvantageous decks for the duration of the task. Healthy control participants
show a greater anticipatory SCR to the disadvantageous relative to advantageous decks, but this effect is absent in the VMPFC lesion patients (right panel).
Data reproduced from Bechara et al. (1994, 1996, 1997) and Bechara and Damasio (2005).
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271244
early performance on the IGT appears to be regulated by
non-declarative learning systems, since healthy control
participants have been shown to select from the advantageous
decks prior to being able to explicitly report the
reinforcement contingencies of the decks. All of these
ﬁndings are consistent with the SMH (see Section 3 for a
more detailed review of studies from the Iowa laboratory
investigating the exact neural substrate underpinning these
somatic markers).
2. Evaluation of the Iowa gambling task as key evidence
for the SMH
Work from the Iowa laboratory at ﬁrst glance provides
strong support for the SMH. On closer examination,
however, a number of issues with the IGT come to light
that potentially undermine this evidence. The strengths and
weaknesses of the IGT will now be systematically reviewed
in order to determine the extent to which IGT data can
support the SMH.
2.1. Strengths of the IGT
The IGT, both as a source of evidence for the SMH and
as a measure of broader decision-making, has a number of
strengths. First, impairments on the task following brain
damage and normative performance on the task in healthy
controls have been extensively validated within the Iowa
laboratory. Forty-ﬁve patients with different kinds of frontal
lobe lesions and 35 patients with lesions to the lateral
temporal or occipital cortex have performed the IGT. Of
these patients, only those with damage to neural regions
implicated in the SMH (e.g. VMPFC, amygdala) appear to
be impaired on the task (Bechara et al., 2000a,b). Further,
over 80 healthy control participants have reportedly
completed the task and produced a similar proﬁle of results
as in the original data (Bechara et al., 1994, 1996). In
addition, studies conducted outside the Iowa laboratory
more often than not replicate the behavioural ﬁndings on the
task, although few of these included psychophysiological
measurements (see Table 1 for a summary of studies using
the IGT published before July 2005).
Second, it has been demonstrated that the IGT is robust in
the face of certain changes in its parameters. A similar
performance pattern emerges when the nature of the
incentive used is varied, for example, when giving real
ﬁnancial rewards or facsimile money (Bowman and Turnbull,
2003). Further, behavioural data have been replicated
when using different time delays on the task (Bowman et al.,
2005) or when using a manually administered or computerised
form of the task (Bechara et al., 2000a).
Third, lifespan developmental changes in performance
on the IGT have been examined. Performance on the
paradigm improves with increasing age until adulthood
(Crone and van der Molen, 2004; Overman et al., 2004; Kerr
and Zelazo, 2004; Hooper et al., 2004) and then appears to
tail off in older age (Denburg et al., 2001, 2005; Lamar and
Resnick, 2004). Gender differences have also been
demonstrated, which appear to be modulated by age.
Adolescent women were found to select cards associated
with both immediate wins and long-term outcome, whereas
adolescent men picked decks on the basis of long-term
outcome only (Overman, 2004). Garon and Moore (2004)
report that girls perform better than boys between 3 and 6
years of age, whereas other studies found that adult men
showed superior behavioural performance to adult women
on the task (Reavis and Overman, 2001; Bolla et al., 2004).
Evans et al. (2004) found that higher levels of education and
intelligence are associated with poorer performance on the
IGT. Non-university educated female participants showed
double the improvement in the last 40 trials of the IGT
compared to those with a university education. These
individual difference ﬁndings aid interpretation of the IGT
at the single case level.
Fourth, the behavioural form of the IGT has been shown
to be a highly sensitive measure of impaired decisionmaking
in a variety of neurological and psychiatric
conditions known to be characterised by real world
decision-making impairments (e.g. pathological gambling,
obsessive compulsive disorder [OCD]; see Table 1).
Further, there is increasing evidence that the task has
reasonable predictive validity. For example, IGT performance
has been associated with response to pharmacotherapy
in OCD (Cavedini et al., 2004b) and safety of sexual
practices in substance abusing HIV-positive males (Gonzalez
et al., 2005). These factors offer good support for the
validity of the paradigm and also illustrate that the task has
stimulated a considerable body of research.
Despite its strengths, there are a number of issues
concerning the IGT that undermine the extent of the support
it can offer to the SMH. These will now be discussed in turn.
2.2. Cognitive penetrability of the IGT reward/punishment
schedule
The cognitive impenetrability of the IGT is central to the
claim that it can only be successfully completed through
recourse to emotion-based learning via somatic marker
signals in the early stages. Damasio states that: “the key
ingredient that distinguishes the task of Bechara and
colleagues from other tasks of probabilistic reasoning is
that subjects discriminate choices by feeling; they develop
hunches that certain choices are better than others.
subjects with damage to VMPFC fail this task and they
fail it precisely because they are unable to represent choice
bias in the form of an ‘emotional hunch’” (Damasio et al.,
2003, p. 84). Further, other articles from the Iowa laboratory
make the claim that: “in normal individuals non-conscious
biases guide reasoning and decision-making behaviour
before conscious knowledge does, and without the help of
such biases, overt knowledge may be insufﬁcient to ensure
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 245
Table 1
Summary of studies examining IGT performance in clinical groups organised by presenting condition (published before July 2005)
Authors Groups studied SCR used? Deﬁcit found in clinical group? Control data replicated?
Clark et al.
(2001)
Acutely manic inpatients (nZ15); nonpsychiatric
participants (nZ30)
N Slower learning in manic patients Y
Wilder et al.
(1998a,b)
Schizophrenic patients (nZ12); healthy
control volunteers (nZ30)
N None N, control group preferred
infrequent punishment
(decks B and D) to frequent
punishment (decks A and C)
Beninger
et al. (2003)
Schizophrenic patients on typical medication
(nZ18); schizophrenic patients on
a-typical medication (nZ18)Healthy control
volunteers (nZ18); healthy control
volunteers (nZ18)
N Patients on atypical medication showed
disadvantageous deck preference, whereas
those on typical medication performed
normally on the task
Y
Ritter et al.
(2004)
Chronic schizophrenic patients (nZ20);
non-psychiatric participants (nZ15)
N Disadvantageous deck preference in
schizophrenic patients
Y
Shurman et al.
(2005)
Schizophrenic patients (nZ39); healthy
control volunteers (nZ10)
N No preference for advantageous decks,
preference for infrequent punishment
(decks B and D) rather than infrequent
punishment (decks A and C)
Y
Turnbull et al.
(in press)
Schizophrenic patients (nZ21); healthy
control volunteers (nZ21)
N None on original task, but schizophrenic
patients with marked negative symptoms
showed deﬁcit on subsequent reversal
learning
Y
Bark et al.
(2005)
Catatonic schizophrenic patients (nZ8);
paranoid schizophrenic patients (nZ19);
healthy control volunteers (nZ26)
N Disadvantageous deck preference in
schizophrenic patients
Y
Nielen et al.
(2002)
OCD patients (nZ27); healthy control
volunteers (nZ26)
N None Y
Cavedini et al.
(2002)
OCD patients (nZ34); healthy control
volunteers (nZ34); panic disorder control
patients (nZ16)
N Disadvantageous deck preference in OCD,
related to poor response to drug treatment
Y
Whitney et al.
(2004)
Schiozophrenia with obsessive symptoms
(nZ26); schizophrenic patients without
obessive symptoms (nZ28); OCD patients
(nZ11)
N Trend for both schizophrenic groups to
select more from disadvantageous decks,
compared to OCD patients
No standard control group
used
Cavedini et al.
(2004)
Patients with OCD (nZ34) treated with
ﬂuvoxamine only or ﬂuvoxamine plus
risperidone
N Patients with good IGT performance
responded to ﬂuvoxamine alone, where
patients with impaired IGT performance
responded to ﬂuvoxamine plus risperidone
No standard control group
used
Cavedini et al.
(2002)
Pathological gambling patients (nZ20);
healthy control volunteers (nZ40)
N Disadvantageous deck preference, increasing
over time in pathological gambling
patients
Y
Goudriaan
et al. (2005)
Pathological gambling patients (nZ48);
alcohol-dependent patients (nZ46); Tourette
syndrome patients (nZ47); healthy
control volunteers (nZ49)
N The pathological gambling and alcoholdependent
patients selected less from
advantageous decks overall, relative to
both control participants and Tourette
syndrome patients
Y
Bechara et al.
(2001)
Substance-dependent individuals (SDI: nZ
41); VMPFC patients (nZ5); healthy
control volunteers (nZ40)
N Slower learning in SDI group, disadvantageous
deck preference in VMPFC group
N, 31% of controls performed
in the range of
VMPFC patients
Bechara and
Damasio
(2002)
Substance-dependent individuals (SDI: nZ
41); VMPFC patients (nZ10); healthy
control participants volunteers (nZ49)
Y Slower learning in SDI group, lower
anticipatory SCR in SDI group relative to
controls
N, 37% of controls performed
in the range of
VMPFC patients
Verdejo et al.
(2004)
Substance abusing individuals (nZ104) N 76% of addiction patients showed disadvantageous
deck preference
No standard control group
used
Fein et al.
(2004)
Abstinent alcoholics (nZ44); healthy control
participants (nZ58)
N Abstinent alcoholics made signiﬁcantly
fewer selections from advantageous decks
Y
Fishbein et al.
(2005)
Abstinent drug abusers (nZ21); healthy
control volunteers (nZ20)
Y Trend for abstinent drug abusers to select
less from the infrequent punishment
advantageous deck. No group differences
on SCR or heart rate
Not reported
Whitlow et al.
(2004)
History of heavy marijuana use (nZ10);
healthy control participants (nZ10)
N Marijuana group showed preference for
disadvantageous decks
Y
(continued on next page)
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271246
Table 1 (continued)
Authors Groups studied SCR used? Deﬁcit found in clinical group? Control data replicated?
Rotheram--
Fuller et al.
(2004)
Opiate-dependent smokers (nZ9); opiatedependent
non-smokers (nZ9); control
smokers (nZ9); control non-smokers (nZ
10)
N Methadone maintained smokers selected
less from advantageous decks compared to
all other groups
N, control non-smokers performed
at chance levels
Schmitt et al.
(1999)
Psychopathy offenders classiﬁed as low
(nZ51), medium (nZ68), or high (nZ38)
on psychopathy checklist
N All offender groups showed disadvantageous
deck preference, but this was related
to anxiety and not psychopathy levels
No standard control group
used
van Honk
et al. (2002)
Low psychopathic group (nZ16); high
psychopathic group (nZ16)
N High psychopathic group showed
disadvantageous deck preference
Y
Best et al.
(2002)
Intermittent explosive disorder patients
(IED: nZ24); healthy control volunteers
(nZ22)
N Disadvantageous deck preference in IED
patients
Y
Martin et al.
(2004)
HIV-positive substance-dependent males
(nZ46); HIV-negative substance-dependent
males (nZ47)
N HIV-positive group made fewer selections
from advantageous decks, relative to
HIV-negative group
No standard control group
used
Gonzalez
et al. (2005)
HIV-positive substance-dependent males
(nZ109); HIV-negative substance-dependent
males (nZ154)
N Greater sensation seeking and better
performance on the IGT was associated
with more risky sexual practices
No standard control group
used
Cavedini et al.
(2004a,b)
In-patients with anorexia (nZ59); healthy
control volunteers (nZ82)
N Patients with binge/purge anorexia did not
develop preference for advantageous decks
over time. Patients with restrictive anorexia
showed preference for disadvantageous
decks
Y
Davis et al.
(2004)
Body mass index (BMI) in; healthy adult
women (nZ41)
N Higher BMI related to disadvantageous
performance on the task
Y
van Honk
et al. (2004)
Women given testestorene versus placebo
(within-subjects cross-over design; nZ12
in each condition)
N Less advantageous performance found in
testosterone condition, relative to placebo
condition
Y
Dalgleish
et al. (2004)
Psychosurgery for depression (stereotactic
subcaudate tractotomy): recovered Psychosurgery
group (nZ10); depressed psychosurgery
group (nZ10); recovered
medication group (nZ9); healthy control
participants (nZ9)
N Recovered psychosurgery group insensitive
to negative feedback
Not reported
Campbell
et al. (2004)
Huntington’s disease group (nZ15);
healthy control volunteers (nZ16)
Y Huntington’s patients selected less from
advantageous decks in last 20 trials only
and showed normal anticipatory but
impaired feedback SCRs
Y, both behavioural and
psychophysiological
Levine et al.
(2005)
Patients with Traumatic Brain Injury (TBI;
nZ71)
N Gambling task was sensitive to TBI in
general, but not severity or quantiﬁed
chronic phase atrophy of injury
No standard control group
used
Mongini et al.
(2005)
Women with history of chronic migraine
(nZ23); healthy female control volunteers
(nZ23)
N None Not reported
Apkarian
et al. (2004)
Chronic back pain (CBS: nZ26); complex
regional pain syndrome (CRPS: nZ12);
healthy control volunteers (nZ26)
N Slower learning in CBP group, disadvantageous
deck preference in CRPS group
Y
Kleeberg
et al. (2004)
Patients with multiple sclerosis (nZ20);
healthy control participants (nZ16)
Y Slower learning and reduced anticipatory
SCRs in patients with multiple sclerosis
Y
Ernst et al.
(2003a)
Adolescents with externalising behavioural
disorders (nZ33); healthy control adolescents
(nZ31)
N No group differences when ﬁrst administered
task. Adolescents with externalising
behavioural disorder showed less marked
improvement on second administration,
relative to control adolescents
Y
Ernst et al.
(2003b)
Adults with ADHD (nZ10); healthy
control participants (nZ12)
N None Not reported
Toplak et al.
(2005)
Adolescents with attentional-deﬁcit-hyperactivity
disorder (ADHD: nZ44); healthy
control adolescents (nZ34)
N Adolescents with ADHD made more
selections from the disadvantageous high
magnitude of punishment decks (B) and
fewer selections from the advantageous
high magnitude decks (D). Deck selection
behaviour in ADHD correlated with
hyperactivity/impulsivity symptoms
Y
(continued on next page)
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 247
advantageous behaviour.we believe that the autonomic
responses detected in our experiment (especially those
evident in the pre-hunch period) are evidence for a nonconscious
signalling process” (Tranel et al., 1999, p. 1055).
In other words, learning via emotion-based biasing
signals is believed to precede explicit insight on the IGT.
If the reward/punishment schedule can be consciously
comprehended by participants prior to the development of
somatic markers, this could mean that cognitive outcome
expectancies rather than somatic markers could guide
successful IGT performance (Turnbull et al., 2003) and
seriously undermine the support the paradigm can offer for
covert ‘somatic marker’ activation.
As discussed earlier, preliminary evidence from the Iowa
laboratory appeared to indicate that participants are not fully
aware of the reward/punishment schedule (Bechara et al.,
1997a). Recent data from other laboratories suggest that the
reward/punishment schedule used in the IGT is more
cognitively penetrable than previously thought, however.
Maia and McClelland (2004) argued that the broad, openended
questions used to measure conscious awareness of the
task contingencies in the Iowa laboratory studies (e.g.
Bechara et al., 1997a) were not sufﬁciently sensitive to
identify all conscious knowledge held by participants. Using
more detailed, focused questions in a replication study with
20 healthy participants answering questions after each block
of 20 trials, they found that advantageous performance on
the task was nearly always accompanied by verbal reports of
reasonably accurate quantitative and qualitative knowledge
about the outcomes of the decks that was sufﬁcient to guide
behaviour. Indeed, participants tended to report accurate
knowledge about the reward/punishment schedule more
reliably than they selected from the advantageous decks,
mirroring the VMPFC lesion patients who reached the
conceptual stage in the Iowa laboratory experiments
(Bechara et al., 1997a). To control the possibility that
asking more detailed questions somehow helped participants
to construct a clearer problem space to comprehend
the task structure, another 20 participants completed the task
with the original questions given by Bechara et al. (1997a)
and their behavioural performance was shown to be
identical. The fact that participants have earlier awareness
of the reward/punishment schedule means that the anticipatory
SCR signals found on the IGT could be a
consequence of conscious knowledge of the situation rather
than being causally involved in the decision-making
process.
Bowman et al. (2005) have also reported that participants
are more aware of the IGT reward/punishment schedule
than previously believed, rating the ‘goodness’ and
‘badness’ of decks at above chance levels as early as after
the initial 20 trials.
Maia and McClelland (2004) concluded that the IGT can
be performed through access to conscious, explicit knowledge,
and it is therefore inaccurate to claim that task
acquisition necessarily requires the generation of nonconscious
somatic marker signals. Of course, these ﬁndings
do not rule out the possibility that participants would
sometimes utilise non-conscious bodily biasing signals to
perform the task. A number of features about the IGT design
make it likely to promote explicit rather than implicit
reasoning, however (Maia and McClelland, 2004). First, it
gives participants time to deliberate over each decision if
they so wish and the outcomes are presented in explicit
numerical form. Second, there is little variation in the
magnitude of rewards and punishments used, making it
relatively easy to track deck characteristics consciously and
therefore potentially promoting explicit reasoning. In
particular, participants only need to pay attention to the
punishment delivered on each deck to successfully acquire
the task (see Section 2.7).
In many ways, these criticisms reﬂect a broader literature
debating the validity of the claim that so-called ‘implicit’
learning really can take place without conscious, ‘explicit’
awareness (see Shanks, 2005 for a critical review). Shanks
argues that empirical evidence demonstrating that learning
can take place without conscious knowledge is undermined
by the use of inappropriate measures of awareness. These
measures often do not index all the conscious knowledge the
participant has (the ‘exhaustiveness’ criteria; Shanks and St.
John, 1994) and do not always tap the same stored
knowledge that is actually controlling relevant task
behaviour (the ‘information’ criteria; Shanks and St. John,
1994). For example, the use of forced-choice stimulus
identiﬁcation may be too stringent a measure of conscious
awareness in associative learning paradigms where partially
identiﬁed stimulus features could shape responding and
where participants adopt a conservative response criterion
Table 1 (continued)
Authors Groups studied SCR used? Deﬁcit found in clinical group? Control data replicated?
Harmsen (in
press)
Nicotine-dependent smokers (nZ61); not
nicotine-dependent smokers (nZ47)
N No difference between groups No standard control group
used
Jollant et al.
(2005)
Patients with history of violent suicide
attempts (nZ32), patients with a history of
non-violent suicide attempts (nZ37), control
patients with affective disorders but
with no history of suicidal behaviour (nZ
25), and healthy control participants (nZ
82)
N Both suicide groups showed disadvantageous
deck preference, relative to healthy
control and psychiatric control groups
Y
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271248
(Lovibond and Shanks, 2002; Shanks and Lovibond, 2002).
The measures of awareness included in the original IGT
cognitive penetrability studies (Bechara et al., 1997a)
appear to meet neither Shanks’ exhaustiveness nor
information criteria.
In response to the criticisms of the cognitive impenetrability
of the IGT reward/punishment schedule, in
particular the points raised by Maia and McClelland
(2004), a robust rebuttal was generated by the Iowa
laboratory (Bechara et al., 2005). This argued that the
emotion-based rather than implicit nature of biasing signals
is core to the SMH, so it is not problematic if the IGT is
more transparent than previously believed: “The central
feature of the SMH is not that non-conscious biases
accomplish decisions in the absence of conscious knowledge,
but rather that emotion-related signals assist cognitive
processes even when they are non-conscious” (Bechara
et al., 2005, p. 159). In many ways, this seems a retreat from
earlier arguments emphasising the frequently implicit
nature of somatic markers (e.g. Tranel et al., 1999) and
makes the SMH hard to distinguish from other accounts of
decision-making.
Bechara et al. (2005) also argued that SCRs appear
sufﬁciently early in the IGT to precede the conscious
knowledge identiﬁed by other laboratories, and so may still
reﬂect implicit learning. This assertion is hard to reconcile
with the fact that some understanding of the reward/
punishment schedule was found as early as the ﬁrst block of
20 trials by Maia and McClelland (2004). It is also unclear
whether it is valid to draw a direct comparison between
insight data gathered in different laboratories using very
different insight assessments (Maia and McClelland, 2005).
Finally, Bechara et al. (2005) argued that Maia and
McClelland’s explanation could not account for why correct
knowledge of a situation did not guarantee correct
decisions, instead suggesting that somatic markers need to
be postulated to explain this scenario. An alternative
explanation, however, is that apparently non-optimal
rational decision-making behaviour in the face of partial
explicit knowledge reﬂects exploratory behaviour by the
participant to further garner information about the task at
hand (Maia and McClelland, 2005).
A recent case report has demonstrated that a patient with
hippocampal amnesia was able to acquire the IGT and
actually showed improved performance over multiple
administrations of the paradigm. This was interpreted as
reﬂecting non-conscious emotional learning in the absence
of awareness. It has previously been argued, however, that
dissociations of performance in amnesia cannot be taken as
strong evidence for multiple memory systems (Shanks & St
John, 1994). Even if amnesia is associated with spared nonconscious
and impaired conscious learning systems, it still
cannot be assumed that these non-conscious processes are
necessarily emotional in nature. Finally, there are other
ways to account for acquisition of the task in hippocampal
amnesia other than through recourse to the SMH. For
example, hippocampal lesions in rats facilitate instrumental
learning with delayed reinforcement by reducing competition
from context-reinforcer associations that normally
hinder the formation or expression of response-reinforcer
associations (Cheung & Cardinal, 2005). Therefore, normal
acquisition of the IGT in amnesia is not conclusive proof
that the IGT crucially relies on non-conscious emotional
learning systems.
In summary, there seems little support for the claim that
the reward/punishment schedule of the IGT is fully
cognitively impenetrable. This means that the assertion
that the IGT in the early stages involves covert, nonconscious
somatic markers to regulate decision-making
(Tranel et al., 1999; Bechara et al., 1997a) can no longer be
conﬁdently endorsed. Instead, it appears that participants
have at least some conscious awareness of the reinforcement
contingencies used in the task. It remains unclear whether
this is best conceptualised as a full rational understanding of
the reward/punishment schedule used or simply that
participants are able to consciously label the valence of
the decks (i.e. ‘this one is good’, ‘this one is bad’) in a
heuristic fashion.
The ‘fully explicit’ interpretation would undermine the
utility of the SMH, since if somatic markers arise only after
full conscious knowledge about a situation is available it is
unclear why people would ever need to make use of them. It
would seem more parsimonious to act on the basis of the
explicit knowledge rather than waiting for the emotional
consequences of the decision to be communicated via the
periphery. Further, it would mean that the somatic marker
signal could be interpreted as a consequence of the explicit
knowledge rather than being of causal importance in the
decision-making chain, making the SMH indistinguishable
from other accounts of task performance described later (see
section 2.8). The heuristic interpretation, while challenging
the impenetrability of the IGT, is consistent with the broader
claim of the SMH that emotion (in this case a conscious
verbal affective label) can guide decision-making. Indeed,
Damasio (1994) argues that somatic markers may act at
either an overt or covert level, so some awareness of the
reward/punishment schedule in the IGT is not problematic
for the SMH in its weakest sense.
2.3. Interpretation of psychophysiological data on the IGT
The pivotal ﬁnding in support of the SMH is that
anticipatory SCRs differentiate between the advantageous
and disadvantageous decks over time. The original paper
reported this effect in 15 healthy control participants and
found that it was absent in VMPFC lesion patients (Bechara
et al., 1996). While the IGT has been used in a wide range of
studies, only a subset has included psychophysiological
measures. Studies that do report psychophysiology data
have generally replicated the ﬁndings of elevated anticipatory
SCRs to the disadvantageous decks, relative to the
advantageous decks (for example, Bechara et al., 1999,
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 249
2002; Bechara and Damasio, 2002; Campbell et al., 2004;
Jameson et al., 2004; Hinson et al., 2002; Tomb et al., 2002;
Suzuki et al., 2003; Crone et al., 2004). Further, there is
some evidence that anticipatory marker signals correlate
with both successful performance on the task (Carter and
Smith-Pasqualini, 2004; Oya et al., 2005) and individual
difference measures of neuroticism (Carter and Smith-Pasqualini,
2004). This indicates that the anticipatory SCR
results are reasonably reliable and may relate to meaningful
individual differences in sensitivity to reward and punishment,
but exactly how to interpret them is complicated by a
number of factors.
First, reliable anticipatory SCR differences have been
reported only in a sub-group of the best performing healthy
control participants in some studies. For example, Crone
et al. (2004) gave 96 students a modiﬁed version of the IGT,
while recording participants’ anticipatory and feedback
SCR and heart rate (HR) responses. Participants were split
into three equal sized groups of poor, moderate and good
performers, based on the total number of selections they
made from the advantageous decks during the task.
Anticipatory SCRs were greater for the disadvantageous
than advantageous decks for the good performance group
only. Similarly, anticipatory HR slowing was greater for the
disadvantageous decks, relative to the advantageous decks,
in the good performance group. Crucially, the moderately
performing group (who nevertheless did successfully
acquire the task) did not show any such psychophysiological
differentiation between the decks. These ﬁndings are
potentially problematic for the SMH since they show that
a number of participants can acquire the task without
needing to generate anticipatory HR or SCR signals,
therefore suggesting that somatic markers are not necessary
or sufﬁcient to do well on the paradigm. In partial defence of
this criticism, it can always be argued that other nonmeasured
forms of peripheral feedback (e.g. facial) helped
these participants to perform the task or that there may have
been some important differences between the original IGT
and the modiﬁed decision-making task used in this study.
Second, there appear to be other ways to interpret the
elevation of SCR to the disadvantageous decks. The SMH
claims that anticipatory SCRs on the IGT reﬂect growing
awareness of the negative, long-term consequences of the
‘risky’ decks. Tomb et al. (2002) investigated whether
anticipatory SCR changes on the IGT are correlates of
correct versus incorrect decision-making or correlates of
low-magnitude reward/punishment versus high-magnitude
reward/punishment decision-making. If the anticipatory
SCRs found on the IGT reﬂect the increased variance of
reward and punishment offered on the ‘disadvantageous’
decks rather than their proﬁtability over time, this does not
offer support for the SMH.
To examine this issue, Tomb et al. (2002) contrasted the
original IGT (where the bad decks are associated with a
higher magnitude of punishment and reward than the good
decks) with a modiﬁed version using a different reward/
punishment schedule (linking the advantageous decks with
a higher magnitude of punishment and reward than the
disadvantageous decks). On the original task, participants
picked more from the good than the bad decks and showed
higher anticipatory SCRs for the bad decks. In the altered
version, volunteers picked more from the good than the bad
decks but showed higher anticipatory SCRs for the good
decks, which is consistent with the interpretation that
increased SCRs reﬂect the greater variance in the rewards
and punishments offered rather than their ‘goodness’ over
time. The Tomb et al. interpretation is also consistent with a
number of behavioural studies describing how participants
differed in number of selections made from the low
frequency/high magnitude of punishment decks and the
high frequency/low magnitude of punishment decks (e.g.
Wilder et al., 1998; Shurman et al., 2005).
In response to the Tomb et al. ﬁndings, Damasio et al.
(2002) argued that somatic markers can serve to record the
long-term positive as well as negative consequences of a
particular response option (so a larger SCR to the
advantageous decks in the Tomb et al. study is not
inconsistent with the SMH). Further, they argued that the
larger SCR to the good decks in the Tomb et al. data may
still reﬂect a non-conscious danger signal related to the
likely risk of a large penalty, but that this signal is overridden
by conscious assessment of the overall goodness of
the decks (but see Maia and McClelland, 2004).
Other authors have argued that the psychophysiological
response to feedback rather than in anticipation of deck
selection is more important for regulating task behaviour.
Suzuki et al. (2003) explored the inﬂuence of anticipatory
and feedback SCRs on a Japanese version of the IGT
completed by a population of 40 Japanese students.
Consistent with the SMH, the ‘risky’ decks produced a
greater anticipatory SCR than the ‘conservative’ decks. No
relationship was found between the number of ‘risky’
selections in each block of the task and the amplitude of
anticipatory SCRs, however. Further, there was no
difference in anticipatory SCRs for early and late trials,
which is inconsistent with the claim that anticipatory
markers are acquired through experience (Bechara et al.,
1996, 1997a). Instead, it was found that feedback SCRs
(referred to as appraisal responses by Bechara) may be more
important for mediating task performance. Feedback SCRs
were greater following punishment than reward and on
selections from ‘risky’ rather than ‘conservative’ decks.
Participants who showed greater feedback SCRs tended to
have a greater learning curve on the task (selecting fewer
times from the ‘risky’ decks in late versus early trials).
These results suggested that feedback rather than anticipatory
SCRs may be more crucial in shaping decision-making
on the IGT.
Against this conclusion, Crone et al. (2004) found that
anticipatory rather than feedback physiological responding
was related to task performance on their variant of the IGT.
All participants showed a greater SCR and greater HR
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271250
deceleration following punishment than reward (particularly
on decks that had a low frequency of punishment), but
this did not differentiate between good, average and poor
performers on the task. Further, ﬁndings of elevated
anticipatory SCRs to the disadvantageous decks in the
high performing sub-group stood even when feedback on
the previous trial was entered as an additional factor in the
analysis. In addition, other studies have found a correlation
between anticipatory SCRs and successful acquisition of the
task (Carter and Smith-Pasqualini, 2004; Oya et al., 2005).
Third, the anticipatory marker signals generated on the
IGT may not be directly involved in the decision-making
process. A simpliﬁed but equivalent decision-making task to
the IGT has been used with rhesus monkeys, ﬁnding that
SCRs were associated with anticipation of reward after a
decision had been made rather than before a decision had
been made (Amiez et al., 2003). This was interpreted as
indicating that SCRs are a correlate of anticipatory
appetitive behaviour rather than reﬂecting the decisionmaking
process directly. This is an important ﬁnding and is
potentially extremely problematic for the SMH, since it
suggests that the ‘anticipatory’ changes found on the IGT
could relate to expectancies about reward and punishment
after a deck has been selected rather than driving deck
selection behaviour in the ﬁrst place. In other words, these
signals may not play a causal role in shaping decisionmaking
behaviour.
At face value it is difﬁcult to separate responses prior to
response selection and prior to feedback, due to the
relatively slow time course of SCR signals. One way to
explore this issue, however, would be to acquire SCR data at
a higher temporal resolution and apply various signal
processing strategies to separate the components of the
signal that relate to selection of response versus anticipation
of feedback following selection (e.g. see Lim et al., 1997) or
use other psychophysiological responses with a faster time
course than SCRs.
Finally, another potential problem with the psychophysiology
ﬁndings on the IGT is that investigators typically
only report SCRs (see Crone et al., 2004 for an exception to
this). Work from other laboratories has generally shown that
the SCR is not particularly sensitive in discriminating
between positive and negative valence (e.g. Bradley et al.,
2001a,b), so this measure may not be the best candidate to
index an underlying emotion-based marker ‘signal’ that
indicates if a decision is ‘good’ or ‘bad’. There are a variety
of other sources of bodily feedback that could be measured,
including facial muscle activity with electromyography
(EMG) or heart rate with electrocardiogram (ECG). Given
the uncertain nature of the interpretation of SCRs, stronger
support for the SMH would be provided by evidence of
anticipatory signals emerging across a range of bodily
systems. Further, it is increasingly realised in the
psychophysiology literature that bodily responses to
particular environmental demands cannot be reliably
modelled on a uni-dimensional arousal spectrum and
instead are better conceptualised as a complex of related
responses (see Lacey, 1967 for a critical review of activation
theory). For example, attentional orienting, rather than
leading to a simple increase or decrease in bodily response,
is characterised by decreased motor activity, increased SCR,
delayed respiration, and heart rate deceleration (Graham,
1979). Therefore, the anticipatory SCR changes seen in the
IGT may form part of a broader response complex, which
may not be valid to interpret as a simple increase or decrease
in bodily response.
In summary, there has been limited external replication of
the key anticipatory SCR data on the IGT. Further, when the
pattern of ﬁndings is successfully replicated, it remains
unclear exactly what these SCRs represent. They may be a
response to feedback, an indicator of risk, a marker of postdecision
emotion state, or a signal of how good or bad a
particular response option is. Additional clariﬁcation of the
mechanism by which these anticipatory marker signals act to
bias decision-making is necessary if these data are to be used
to support the SMH. We will return to this point later in the
manuscript (section 4.2 on ‘Speciﬁcation of the somatic
marker mechanism’). To more fully support the SMH
measurement of other bodily responses in addition to SCR
is needed to better model the complex bodily activation
typically seen in response to environmental stimuli. Further,
causal designs that manipulate somatic state and look
at the effects on IGT performance are also required (see
Section 2.6)
2.4. Variability in control participant performance
on the IGT
It is increasingly apparent that substantial minorities of
control participants do not learn to select from the
advantageous decks over time on the IGT. Bechara and
Damasio (2002) reported that not all control participants
perform advantageously on the IGT and this did not clearly
relate to the development of anticipatory SCRs. They found
that around 20% of normal adults performed disadvantageously
on the IGT, selecting more from bad decks than
good decks. Within this poorly performing sub-group, the
variance in anticipatory SCRs was high, with some showing
normal SCRs and some showing a proﬁle similar to VMPFC
lesion patients. Those participants with impaired behavioural
performance but normal anticipatory SCRs were
characterised post hoc as ‘high-risk takers’, choosing to
override ‘somatic marker’ information with conscious
deliberation.
Similarly, a variety of studies have reported that there are
sub-groups of healthy control participants who do not show
a preference for the advantageous decks at the end of the
task (e.g. Crone et al., 2004; Adinoff et al., 2003) or an
overall reduction in the number of selections made from
advantageous decks in healthy populations (Lehto and
Elorinne, 2003: see Table 1). While these ﬁndings are
perhaps not directly problematic for the SMH (since it does
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 251
not claim to be a truly nomothetic model of decisionmaking),
they do make interpretation of IGT data complex.
Where no differences between a patient/lesion group and a
control group are found, does this genuinely reﬂect an
absence of a deﬁcit in the experimental group, or rather that
a poorly performing group of control participants was
selected? This variability in control data suggests that
performance on the IGT should perhaps be compared to
objective criteria (for example, selecting from advantageous
decks at greater than chance levels), as well as relative to the
performance of a comparison group. Further, the variability
in behavioural performance of control participants raises
doubts about the ecological validity of the paradigm. If a
group of control participants can perform extremely poorly
on the task and yet presumably does not always show gross
difﬁculties in day to day decision-making, it is then unclear
to what extent the psychological mechanisms measured by
the paradigm bear any relevance to everyday, real-world
decision-making. Of course, it remains possible that control
participants could exhibit impaired decision-making in
everyday life, so perhaps measures of quality of everyday
decision-making should be included in studies to check for
this possibility.
2.5. Speciﬁcity and reliability of the IGT in clinical
populations
Another issue with the IGT is that performance deﬁcits
do not appear to be speciﬁc to particular neurological or
psychiatric conditions, since a majority of patient groups
show impairments on the measure. This partially undermines
its utility in understanding decision-making difﬁculties
across different disorders. Moreover, there have been
some failures to replicate ﬁndings within a particular
clinical group, suggesting that ﬁndings may not be that
reliable. For example, of six studies examining schizophrenia,
one found no deﬁcit (Wilder et al., 1998), one
found a disadvantageous deck preference (Ritter et al.,
2004), one found deﬁcits were dependent on medication
type (Beninger et al., 2003), one found preference for the
infrequent punishment decks (Shurman et al., 2005), and
two found only sub-types of schizophrenic patients were
impaired on the task (Bark et al., 2005; Turnbull et al., in
press).
2.6. The causal relationship between body-state feedback
and IGT performance
Another potential issue with the IGT is that the
psychophysiology data implicating ‘somatic marker’ generation
with successful performance are correlational only.
This means that no causal conclusions about the role of
body-state feedback on decision-making can be reliably
drawn. Although healthy volunteers seem to show anticipatory
SCR activity when performing well on the task, this
may reﬂect the end product of the decision-making process
rather than being a key feature in its development. For
example, the SCR changes may reﬂect anticipation of
reward and punishment to be delivered after a deck has been
selected (Amiez et al., 2003). Further, it may be the case that
some other mechanism regulates both behavioural selection
and somatic marker generation on the IGT. It could also be
argued that the most parsimonious account of the IGT data
is that greater somatic marker activation actually leads to
impaired behavioural performance (since greatest somatic
marker activation actually temporally precedes selection
from the disadvantageous decks). Causal methodologies are
needed to refute these criticisms.
A number of attempts have been made to demonstrate a
causal link between body-state feedback and IGT performance,
but have so far provided mixed support for the SMH.
The primary methodology has been to look at groups with
impaired feedback from the body and see if this leads to suboptimal
performance on the IGT. Feedback from the body
can arise from multiple routes (e.g. Bechara, 2004;
Damasio, 2004; for a more detailed review see Craig,
2002), including spinal cord, vagus nerve, endocrine
system, feedback from facial muscles, and from the
physiochemical environment of the brain (for example,
temperature).
Preliminary work found that 20 individuals suffering
from peripheral neuropathy, a condition that reduces
afferent feedback to the brain, were mildly impaired on
the IGT compared to control volunteers (Bechara et al.,
1998a). This study has so far only been published as
conference abstract without peer review, so a full analysis of
its methodological merits is not possible. This means it
should not yet be viewed as strong supporting evidence for
the SMH.
Subsequently, IGT performance has been examined in
six patients with pure autonomic failure (PAF), a condition
that leads to a peripheral denervation of autonomic neurons
and therefore an absence of peripheral autonomic responses
(Heims et al., 2004). Contrary to prediction, PAF patients
performed better than a comparable control group on the
IGT (i.e. selected from the advantageous decks to a greater
extent). The fact that PAF patients were signiﬁcantly
superior on the IGT relative to controls means that the
ﬁndings cannot be explained away as a lack of statistical
power due to a small sample size. Further, it has been
demonstrated that long standing PAF also leads to changes
in the morphology of brain regions involved in the
representation and regulation of body state (Critchley
et al., 2003). Voxel-based morphometry on the structural
MRI scans of 15 PAF patients revealed decreases in grey
matter volume and concentration in both the anterior
cingulate and insula cortices, perhaps as a result of loss of
afferent input to brain regions involved in autonomic
representation. This is signiﬁcant for the SMH because it
means that PAF patients have both body-state feedback and
regions of the ‘as-if’ loop compromised, so some kind of
impairment would have been expected. Therefore, the
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271252
ﬁnding that PAF patients are not impaired on the IGT is
difﬁcult for the SMH to explain. Heims et al. (2004) argue,
however, that their data do not disprove the SMH since
other sources of peripheral feedback are still intact in PAF
patients, in particular somatic systems.
There is some tentative evidence that the vagus nerve, a
major feedback pathway for signals from the peripheral
nervous system to the brain, is important in the generation of
somatic markers (Martin et al., 2004a). Patients who had
vagus nerve stimulators implanted to treat epilepsy
performed two repeat versions of the IGT, one when
vagus nerve stimulation was covertly delivered and one
without stimulation. The stimulated group initially selected
more from the disadvantageous decks and then moved to the
advantageous decks, whereas the group without stimulation
initially selected more from the advantageous decks and
then moved away from them as the task progressed. This
improvement in performance over time in the stimulated
relative to the un-stimulated condition was interpreted as
evidence that the vagus nerve inﬂuences decision-making
(Martin et al., 2004a). This cannot be viewed as strong
support for the SMH, however, since there was no overall
group difference.
The study of patients with spinal cord damage has
failed to generate support for the SMH. Patients with
spinal cord sections at the sixth cervical vertebrae, thus
blocking somatic feedback, displayed no deﬁcit on the
IGT relative to healthy controls (North and O’Carroll,
2001). This is despite the fact that spinal cord injury has
been sometimes shown to reduce the intensity of emotion
experience (Hohmann, 1966; Chwalisz et al., 1988;
Montoya and Schandry, 1994; although see Cobos et al.,
2002 for a recent exception). North and O’Carroll (2001)
suggested that their data can be reconciled with
Damasio’s model if it is assumed that feedback from
the hormonal route and nerves outside the spinal cord
(e.g. facial muscles) is more important than the afferent
feedback sent via the spinal cord. It is also possible that
people who have lived with long-term spinal damage have
adapted to the loss of peripheral feedback and make more
extensive use of the ‘as-if’ loop (see Fig. 1). Thus, a
deﬁcit may have been revealed if the patients were studied
immediately after injury. Imaging studies of such patients
could examine whether activity in the ‘as-if’ loop
mediates intact performance on the IGT, as predicted by
the SMH.
An important development for SMH will be to examine
the conditions under which the body loop or ‘as-if’ loop are
involved in decision-making. Bechara (2004) suggests that
the body loop will be engaged in decision-making under
uncertainty, whereas decision-making under certainty (i.e.
where the outcome is explicit and predictable) will engage
the ‘as-if’ loop, although it is currently unspeciﬁed as to
why this should be the case.
In summary, studies of IGT performance in patients with
altered feedback from the body have not provided strong
support for the SMH, so the causal status of the model
remains unclear. Interpretations of these experiments
perhaps raise issues about the testability of the SMH.
Nearly all of the negative ﬁndings can be explained away
through some aspects of peripheral feedback remaining
intact in the patient groups studied (for example, the somatic
system in PAF, endocrine feedback following spinal injury)
or through recourse to the ‘as-if’ loop. It seems virtually
impossible to test the theory in a scenario where all
peripheral feedback routes are disturbed and the ‘as-if’ loop
cannot be utilised, therefore making it difﬁcult to disprove.
2.7. Task design issues
A number of methodological features of the task design
and psychophysiology analysis of the IGT also complicate
interpretation of the data it generates. First, in the
psychophysiology analysis the deck that participants
eventually select is used to designate each anticipatory
‘somatic marker’. In the deck selection phase, however,
people are free to shift their attentional focus across all of
the decks before settling on one. This means that the
physiological marker generated may not reﬂect attention to
a single deck but a shifting attentional focus across all decks
before arriving at a choice.
Second, the magnitude of rewards given in the task is
predictable (either 50 or 100 points), meaning that to do well
on the tasks participants simply have to attend to the
punishment component of the schedule (see Fig. 2). This
possibility is consistent with the ﬁnding that neuroticism
positively correlates with successful IGT performance
(Carter and Smith-Pasqualini, 2004), given that neuroticism
is characterised by a sensitivity to punishment. This issue
introduces two complications to the interpretation of the
data: the reward/punishment schedule may be easier to
comprehend consciously than previously claimed and the
task only measures variance in response to punishment
rather than reinforcement more generally. This possibly
explains why recent ﬁndings indicate that the task
reinforcement schedule is easier to consciously grasp than
originally assumed (e.g. Maia and McClelland, 2004).
Third, deck position is not counterbalanced in the task,
meaning preferential selection from good or bad decks
could reﬂect a location bias rather than a genuine decisionmaking
deﬁcit.
Fourth, the way in which the decks are classiﬁed as
advantageous or disadvantageous is problematic (Maia and
McClelland, 2004). At any point in the game, participants
have nothing but their prior experience to decide whether
the decks are good or bad. In the early trials, the bad decks
actually are the most advantageous and it is only later that
they become disadvantageous. Therefore, analysis of each
trial should standardly classify each deck based on net
outcomes up until that point in the game, as is typical in
the decision-making literature, rather than on eventual
reinforcement, as currently done in the Iowa laboratory.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 253
Such an approach has recently been taken in a reinforcement
learning model of the IGT (Oya et al., 2005). In this light,
early selection from the disadvantageous decks can be
interpreted as ‘rational’ exploratory behaviour rather than
impaired decision-making. Another way to control for this
issue is to exclude the ﬁrst block of 20 trials from the
analysis and see if a similar behavioural pattern emerges.
Fifth, participants can only select 80 times from each
deck. This raises the possibility that behaviour in the last 20
trials, a point in the game where some participants will run
out of cards in their preferred advantageous deck, may
deteriorate due to this confound. This reduces sensitivity to
detect impairment, since it effectively ‘penalises’ individuals
who learn early in the game. One way round this
possibility would be allow participants to make 100
selections from each deck.
Sixth, there are various statistical issues that undermine
the IGT data. In the behavioural analysis, learning on the
task is indexed in a relatively insensitive fashion by looking
at the number of advantageous deck selections in each block
of 20 trials, which could perhaps mask successful
acquisition in patient populations. A more sensitive analysis
would be to carry out some kind of trend analysis (e.g. a
linear contrast). Further, rate of learning on the task is
always compared relative to the performance of the control
group. It is possible that patient groups are still learning the
task but just at a slower rate than control volunteers. To
explore this possibility it is necessary to look at deck
selection relative to chance as well as relative to control
participant performance.
In the psychophysiological analysis, there is a problem of
an unequal number of means in each cell of the analysis of
variance (ANOVA). In particular, participants make a large
number of selections from the advantageous decks and only
a handful of selections from the disadvantageous decks.
This means that the differences in SCRs to the advantageous
and disadvantageous decks may be partially an artefact of
the statistical assumptions of ANOVA being violated.
Moreover, the elevations in response to the disadvantageous
decks (which are typically only selected early on in the task)
may be an artefact of novelty/habituation. To control for this
possibility, it would be necessary to devise a variant task
where the different decks are equally sampled from across
different periods of time.
To better support the SMH, it needs to be seen whether a
similar pattern of ﬁndings emerges when these methodological
issues are addressed.
2.8. Alternative mechanisms potentially underlying IGT
task performance
The SMH purports that impaired performance on the IGT
relates to an inability to link past emotional response to
punishing stimuli with various future response options via
somatic marker biasing signals, leading to a ‘myopia for the
future’ (Damasio, 1994; Bechara et al., 1996). This account
is derived largely from observation of patients with VMPFC
damage and interpretation of the behavioural deﬁcits they
display on the IGT. Consistent with this interpretation, it has
recently been shown using an ‘expectancy-valence’ model
that the choices of VMPFC patients on the IGT are guided
predominantly by the most recent outcome rather than by
outcomes on all past trials (Busemeyer and Stout, 2002).
Damasio’s account of the deﬁcits displayed by patients
with VMPFC damage has recently been questioned
(MacMillan, 2000), however, and alternative accounts to
explain the deﬁcits displayed by VMPFC lesion cases have
been put forward (e.g. Gomez-Beldarrain et al., 2004;
Camille et al., 2004). Moreover, there is evidence that some
VMPFC patients show impaired performance on the IGT
even when their emotional reactions are in the normal range.
Naccache et al. (2005) report the results of detailed
investigation on patient RMB, who had suffered damage
to left mesio-frontal cortex (encompassing the ACC, the
genu of the corpus callosum, and part of the orbitofrontal
cortex). RMB showed impaired performance on the IGT,
replicating results from the Iowa laboratory (e.g. Bechara
et al., 1994). Problematically for the SMH, RMB showed
impaired performance on the IGT even though the affective
response system of RMB seemed to be largely intact (e.g. a
normal self-report and physiological response to emotion
inducing pictures was found). This suggests that there can
be deﬁcits on the IGT in the absence of deﬁcits in somatic
marker generation, indicating these emotion-based biasing
signals are neither necessary or sufﬁcient to complete the
task.
Given that Damasio’s interpretation of the function of the
VMPFC has come into question, it seems sensible to
examine whether alternative functional explanations may
also be able to account for impaired performance on the
IGT. If other mechanisms provide an equally valid account
of why behavioural deﬁcits may be shown on the IGT, this
undermines support it can offer the SMH. A range of
evidence that suggests other mechanisms could account for
the ﬁndings on the IGT will be evaluated below.
2.8.1. Working memory
The realisation that the reward/punishment schedule on
the IGT is less impenetrable than previously thought (Maia
and McClelland, 2004) raises the possibility that explicit
learning mechanisms are more important for task acquisition
than previously believed. In particular, a number of
studies have examined whether intact working memory is
necessary to do well on the IGT.
A role for working memory on performance of the IGT
has been shown by Hinson et al. (2002). They asked healthy
control participants to perform a modiﬁed, more difﬁcult
variant of the IGT (where only one deck was proﬁtable and
the differences in proﬁtability between the four decks were
less extreme) whilst simultaneously completing a secondary
task with or without a working-memory load. The workingmemory
condition (holding a string of digits in memory),
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271254
compared to the no working-memory condition (repeating
digits ﬂashed up on the screen), led to poorer behavioural
performance and impaired acquisition of anticipatory
somatic markers on the modiﬁed IGT. The authors
concluded that working-memory processes therefore contribute
to the development of somatic markers.
In a follow-up study, Jameson et al. (2004) explored
whether secondary task performance inﬂuenced workingmemory
capacity by reducing central executive capacity or
interefering with short-term capacity in the phonological
loop. They achieved this by comparing the effects of
articulatory suppression (rehearsing the word ‘the’ over and
over again to occupy the phonological loop) to the two
conditions used in the original experiment in a withinsubjects
design using 20 healthy control participants.
More ‘good’ choices were made during the keypad (no
working-memory load) and articulatory suppression conditions,
compared to the digit maintenance task (workingmemory
load). Moreover, performance improved over time
and there was a difference between anticipatory SCRs
between the ‘good’ and ‘neutral/bad’ options on all
conditions except the digit maintenance task. The authors
concluded that secondary tasks interfere with IGT performance
because of the demands they place on central executive
function rather than via blocking the phonological loop.
This is reconciled with the ﬁnding that VMPFC patients
with normal working memory show impaired IGT performance
by stating that central executive resources are
necessary but not sufﬁcient for the development of somatic
markers.
The Hinson et al. (2002) and Jameson et al. (2004)
studies can be criticised on the grounds that they did not use
the original IGT paradigm, however, and it is possible that
the mechanism through which their variant task is learned
differs in some important way. A recent study looking at
dual-task effects on the original IGT has found a less clear
role for working-memory. Turnbull et al. (2005) tested the
contribution of working-memory-dependent and workingmemory-independent
processes to performance on the IGT
using a dual-task methodology. Seventy-ﬁve healthy control
participants were randomly allocated to one of three
experimental conditions: IGT with no secondary task, IGT
with a non-executive secondary task (articulatory suppression),
and IGT with an executive secondary task (random
number generation). Results found that the rate of learning
in the three groups was not signiﬁcantly different, with all
three groups successfully acquiring the task. This was
interpreted as offering support for the claim that IGT
performance is relatively independent of working memory.
Further, both secondary tasks presumably involved some
activation of working memory, so the fact that they did not
impair performance supports the claim that the IGT is
largely independent of working-memory capacity. There
was a trend for the no secondary task group to show superior
performance, however, which perhaps partially undermines
this interpretation.
Other evidence consistent with a working-memory
account of IGT impairment are ﬁndings that task performance
relates to regions of prefrontal cortex believed to
regulate working memory and general intelligence, such as
dorsolateral prefrontal cortex (DLPFC). Bechara et al.
(1998a) compared patients with damage to DLPFC and
VMPFC on the IGT and two tests of working memory. It
was shown that working memory did not depend on the
intactness of decision-making; participants could have
normal working memory in the presence or absence of
decision-making impairment. On the other hand, decisionmaking
was affected by working memory. Performance on
the IGT tended to be worse in those with working-memory
problems. In summary, an asymmetrical relationship
between working memory and decision-making was
found. VMPFC damage led to decision-making impairments,
which were exacerbated by working-memory
problems in those with more extensive lesions. Patients
with right-sided DLPFC damage showed impaired workingmemory,
which also led to low-normal results in the
decision-making task. Damage to broader regions of the
prefrontal cortex including DLPFC has also been found to
impair IGT performance in studies from other laboratories
(e.g. Manes et al., 2002). A similar asymmetric dependence
between working memory and decision-making has been
found in individuals with substance abuse (Bechara and
Martin, 2004). Moreover, IGT acquisition has been found to
correlate with resting state activity in DLPFC (Adinoff
et al., 2003).
In summary, while dual-task methodologies have
provided mixed support for the role of working memory
on the IGT, lesion and neuroimaging ﬁndings suggest that
more dorsal regions of the frontal lobes known to regulate
working-memory do appear to be important for successful
task performance. It is debatable to what extent these
ﬁndings are problematic for the SMH, since Damasio (1994)
explicitly states that one function of somatic markers is to
indicate which options should have working memory and
attentional resources allocated to them. In other words,
successful acquisition of the IGT may involve ﬁrst the
development of somatic markers to distinguish between
good and bad options and second the use of these signals to
devote system resources to the better options. In this light,
there is no need to posit independence between these two
systems.
2.8.2. Reversal learning/inhibition
Another candidate mechanism that could underlie
impaired performance on the IGT is a difﬁculty in reversal
learning. A crucial aspect of the IGT is that participants
have to perform a response reversal: they have to shift their
preference away from the decks that are initially rewarding
in the ﬁrst few trials following subsequent punishment. A
number of studies have found impaired reversal learning in
VMPFC patients, indirectly suggesting this may account for
their deﬁcit on the IGT. Patients with ventral PFC damage
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 255
were shown to have difﬁculty in simple reversal learning
(Rolls et al., 1994), although it has been argued that the
patients studied had lesions extending more laterally in
orbitofrontal cortex than those studied by the Iowa
laboratory (Clark et al., 2003). A later study has now
shown that lesions restricted to VMPFC allow normal
acquisition but impaired reversal on simple reversal
learning tasks (Fellows and Farah, 2003). Similarly, Hornak
et al. (2004) reported that bilateral lesions to OFC but not
DLPFC reliably impair reversal learning on a task
previously found to activate OFC in earlier fMRI studies
of healthy control participants. There is also some evidence
that reversal learning impairments correlate with real life
decision-making problems in patients with VMPFC damage
(e.g. Lawrence et al., 1999).
To directly test the possibility that a reversal deﬁcit
explains impaired IGT performance, Fellows and Farah
(2005a) rearranged the initial reward/punishment schedule
on the task such that the two disadvantageous decks no
longer had an initial advantage in the opening trials.
Following this shufﬂing of the opening trials, the
performance of VMPFC patients was the same as that of
control volunteers, suggesting that it is a difﬁculty in
reversing early learning that is underpinning the behavioural
proﬁle of VMPFC patients on the IGT.
This inability to show reversal learning can be understood
as a deﬁcit in response inhibition (Rescorla, 1996). A
common response to initial feedback is to continue a
response when it is rewarded and to change a behaviour
when it is punished (referred to as win-stay/lose-shift
behaviour in the animal decision-making literature, Restle,
1958; Gaffan, 1979; Reid and Morris, 1992). There is some
evidence to suggest that over time animals are able to
override this pattern to maintain long-term goal-directed
behaviour in the face of short-term setbacks (e.g. Killcross
and Coutureau, 2003). One way to understand successful
acquisition of the IGT is that participants have to be able to
learn over time to inhibit the lose-shift pattern of responding
to the advantageous decks following punishment. Patients
with VMPFC lesions may be unable to show this inhibition,
meaning their responses are driven by feedback in the
moment rather than longer-term proﬁtability. The details of
such inhibitory mechanisms have been well speciﬁed and
may represent a genuine alternative to the SMH to model
impaired performance on the task (Rescorla, 1996).
In defence of the reversal critique, Bechara et al. (2005)
argued that, while the IGT does contain elements of
contingency-reversal, there are other aspects of the task
that need to be learned to explain successful performance.
Consistent with this defence are results from a recent study
examining performance on a variant of the IGT that builds
in a more marked reversal learning component (Turnbull
et al., in press). After completing 100 trials of the standard
task, participants then completed a further series of trials
where the original contingencies are modiﬁed (the decks
that are good and bad are systematically changed) over three
shift phases. Schizophrenic patients with marked negative
symptoms showed no deﬁcit on acquisition of the original
task, but showed chance performance on the shift trials. This
preliminarily suggests that the original IGT measures some
aspects of decision-making that are dissociable from simple
reversal learning (i.e. the task can be acquired despite the
presence of a deﬁcit in reversal learning). It could be the
case, however, that schizophrenic patients have a milder
reversal learning deﬁcit and that this problem builds up over
time on the task. Further work more closely examining
control participant and VMPFC lesion patient performance
on this modiﬁed task is warranted to support this conclusion.
Bechara et al. (2005) further refuted a simple reversal
explanation by asserting that it cannot explain why some
patients perform disadvantageously on the complex reversal
on the IGT, even when they consciously conceptualise the
reward/punishment schedule on the task towards the end of
the experiment. Models of reversal learning (e.g. Rolls,
1999) posit more than one output system, perhaps operating
at different levels of consciousness, however, which can
therefore account for this pattern. Bechara et al. (2005) also
argued that for reversal to take place requires a ‘stop signal’
to be acquired, which could take the form of an emotionbased
biasing signal. In other words, the acquisition of
somatic markers may underpin successful reversal learning.
Against this defence, it has been demonstrated that reversal
learning does not depend on an intact amygdala (Izquierdo
et al., 2004), suggesting reversal is independent of emotion
processing systems. Reversal accounts therefore offer an
explanation of IGT performance distinct from the SMH.
2.8.3. Risk-taking
Another explanation of impaired performance on the IGT
following VMPFC damage is that risk-taking behaviour is
changed. Alterations in deck selection behaviour may
simply reﬂect individual differences in preference for risk
rather than ‘good’ or ‘bad’ decision-making behaviour. For
example, sensation-seeking individuals may prefer decks
that generate the most arousal or interest (i.e. ‘risky’ cards)
rather than are the most proﬁtable (i.e. ‘safe’ cards)
(Zuckerman, 1994). Thus, selection from the ‘disadvantageous’
decks can be perfectly rational depending on the
individuals’ preference structure. This may be particularly
likely to be the case given that play rather than real money is
used. Differences in preference structure, as opposed to
underlying decision-making, could underlie the variation
found in control participant performance on the task and
also account for impaired performance in patient groups.
Indeed, the ‘impaired’ decision-making style shown by
VMPFC lesion patients may in some circumstances be more
adaptive than that shown by healthy control participants.
For example, Shiv et al. (2005) found that VMPFC lesion
patients showed superior performance on a ﬁnancial
investment paradigm, as their decisions were less affected
by previous reward and punishment delivered than were
those of healthy control participants.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271256
One paradigm that has been used to investigate risktaking
is the Cambridge Gambling Task (CGT: Rogers
et al., 1999). On each trial of the CGT an array of 10 red and
blue boxes is presented, behind one of which a token is
randomly hidden. Participants are asked to judge which
colour box the token will be hidden behind and then bet
some of their points total on the outcome. The numbers of
each colour of box are systematically varied across trials,
making it more or less likely the token will be located
behind a particular colour. Therefore, the task measures
accuracy of decision-making via a relatively simple
probabilistic judgement and then measures risk-taking via
analysis of betting patterns. Healthy control subjects have
been found to select the more likely outcome and to
conservatively adjust their betting according to the ratio of
red and blue boxes, for example betting more when the
colour ratio is 9:1 rather than 6:4.
Patient studies have generally found a pattern of normal
but slowed accuracy of decision-making accompanied with
increased, riskier betting behaviour on the CGT in
conditions that damage ventral portions of the prefrontal
cortex (for a review, see Clark and Manes, 2004). This
proﬁle has been shown by patients with frontal variant
Fronto-Temporal Dementia (Rahman et al., 1999), and large
prefrontal lesions including VMPFC (Manes et al., 2002).
These ﬁndings are consistent with the hypothesis that
ventral PFC damage leads to a riskier, more impulsive
decision-making style. Against this conclusion, however, a
study that looked at patients with VMPFC lesions found
impaired decision-making accuracy and decreased betting
(Rogers et al., 1999). One way to reconcile these ﬁndings is
to propose that risky behaviour may be reduced in those
with a severe deﬁcit in decision-making as a compensatory
strategy (Clark and Manes, 2004).
Similarly, Sanfey et al. (2003) looked at VMPFC lesion
performance on another variant of the IGT to test the risk
hypothesis. Participants were asked to select from four
decks of cards that differed in their variance of reward and
punishment over time but not in their overall proﬁtability.
Therefore, participants’ risk preference can be indexed in
isolation from the accuracy of their decision-making.
Control participants showed a marked avoidance of risk,
picking from the more secure, low variance decks. The
VMPFC lesion group could be split into those who were
also risk-averse and those who showed marked risk-taking
behaviour (selecting mostly from the high variance decks).
Intriguingly, these two sub-groups did not differ in any clear
way in terms of lesion location, although the risk-taking
group tended to have lesions extending to the DLPFC. This
ﬁnding contrasts to the IGT data, where VMPFC lesion
patients were insensitive to risk on both the advantageous
and disadvantageous decks. It is important to note, however,
that a number of frameworks have conceptualised risks as
feelings to guide decision-making (e.g. Loewenstein et al.,
2001), so the risk explanation may still be broadly
compatible with the weak form of the SMH.
2.8.4. Insensitivity to rewarding and punishing outcomes
As discussed in the earlier psychophysiology section, yet
another possibility is that it is the response to reward and
punishment, rather than anticipatory change in the body,
that is regulating performance on the task. Differences in
SCRs to winning and losing on the IGT have indeed been
described (see Bechara, 2000b) and may support this
interpretation. Further, differential responses to positive
and negative feedback may explain impaired performance
on the IGT in various patient groups. For example,
Dalgleish et al. (2004) found that an insensitivity to
punishing outcomes on the IGT characterised the performance
of patients who had recovered from severe depression
following psychosurgery in comparison to healthy control
volunteers, depressed patients who had not recovered
following the surgery, and depressed patients who had
recovered with medication. Successful performance on the
IGT has also been found to correlate with self-report of
increased sensitivity to reward (Franken and Muris, 2005).
The Suzuki et al. (2003) data on the IGT discussed earlier
are consistent with a feedback hypothesis, since feedback
SCRs were more clearly related to behaviour performance
on the task than anticipatory SCRs. An alteration in the
response to rewarding and punishing outcomes would be
consistent with a number of other models of decisionmaking.
For example, decision affect theory (see Mellers
et al., 1997) suggests that the emotions people experience
after a decision depend on a comparison between what the
consequences actually were and the consequences that
would have come about if another response option was
chosen. These emotions then determine how likely it is that
a given response option will be repeated or changed (i.e. an
option will not be repeated if it induced regret).
Some data are inconsistent with an insensitivity to
rewarding and punishing outcomes, however. Crone et al.
(2004) showed that anticipatory physiological changes
differed between the advantageous and disadvantageous
decks even when feedback on the previous trial was
controlled for, which undermines the plausibility of
response to, rather than anticipation of, feedback as the
central mechanism for learning the task. Similarly,
performance on the variant IGT task where reward and
punishment were reversed also undermines this feedback
explanation (Bechara et al., 2000b).
2.8.5. Apathy
Another mechanism that could explain impaired performance
on the IGT is apathy (lack of motivation). Rather
than being unable to make the correct decision, VMPFC
patients and other impaired groups may simply not care
enough about the negative outcomes to actively avoid them.
Consistent with an apathy deﬁcit, Barrash et al. (2000)
report that apathy is a symptom exhibited by VMPFC lesion
patients from the Iowa laboratory and the deﬁcit in
emotional response to affective images in VMPFC lesion
patients can be ameliorated when they are directed to look
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 257
carefully (Damasio et al., 1991). Apathy could produce both
the behavioural deﬁcit and the failure to generate
anticipatory SCRs seen on the IGT in VMPFC lesion
patients and it is possible that IGT performance would
similarly improve if engagement levels were raised. Also
consistent with the apathy argument is a recent study
drawing a distinction between different aspects of futuredirected
thinking (Fellows and Farah, 2005b). VMPFC
damage was found to leave temporal discounting (the
subjective devaluation of reward as a function of delay)
intact but to impair future time perspective (a measure of the
length of an individual’s self-deﬁned future). Crucially, this
deﬁcit in future time perspective was found to correlate with
symptoms of apathy rather than impulsivity, suggesting that
the syndrome of apathy may deserve more attention in
understanding impaired future thinking and decisionmaking
following frontal lobe damage.
2.8.6. Overview of alternative mechanisms accounting
for IGT deﬁcits
In summary, there is evidence that a range of
mechanisms other than a ‘myopia for the future’ due to an
inability to acquire somatic markers may explain impaired
performance on the IGT.
One way to measure which of these competing
explanations best explain deﬁcits on the IGT is to correlate
performance within groups on the IGT and measures of the
other related constructs. For example, Monteresso et al.
(2001) compared performance of 32 cocaine-dependent
patients on the IGT, the CGT, and a delayed discounting
procedure (where participants chose between smallersooner
and larger-later rewards). These data allowed
preliminary examination of whether IGT deﬁcits reﬂect
altered risk-taking behaviour (in which case there should be
a correlation with the CGT) or myopia for the future (in
which case there should be a correlation with the delayed
discounting procedure). There were signiﬁcant correlations
between IGT performance and delayed discounting procedure
performance. The link between the IGT and the CGT
was less clear-cut, with reaction times but not behavioural
choices on the CGT correlating with IGT performance. This
suggests that there is some commonality in the construct
that the different measures are indexing, with IGT
impairment perhaps being more clearly linked with a
difﬁculty in considering future outcomes than increased
risk-taking. These data only speak to what impairs IGT
performance in this speciﬁc group of cocaine-dependent
patients, however, and different relationships may exist in a
healthy population. This approach could usefully be
extended to a larger sample of healthy control participants
to test this possibility.
It is important to observe that these explanations of IGT
deﬁcits are not necessarily mutually exclusive. It is possible
that a variety of different mechanisms are involved in
successful task acquisition and that damage to any one of
these can impair task performance. Further, different
mechanisms may be adopted at different points in the task
and across different individuals. It may also be the case that
other cognitive functions form integral parts of the somatic
marker apparatus. For example, the fact that working
memory appears to be needed to perform the task well can
be interpreted either as a factor that confounds the tasks or
as a integral part of the mechanism underlying decisionmaking
via somatic markers (Clark and Manes, 2004). In
many ways, a deﬁcit in somatic marker activation could
underlie a majority of the other explanations put forward to
account for the data (e.g. ‘risks as feelings’; Loewenstein
et al., 2001). What is needed to tease apart these
mechanisms is the development of experimental designs
where the different mechanisms make competing predictions,
plus a more detailed speciﬁcation of the SMH to allow
its predictions to be differentiated from those of other
theories of decision-making.
2.9. Current status of the IGT data
The previous sections have described how the IGT
reward/punishment schedule has been found to be less
opaque than previously believed; how a variety of
mechanisms other than somatic markers could equally
well explain task behavioural performance and psychophysiology
results; that there is an absence of causal evidence
linking disturbed feedback from the body to impaired
performance; that there are a number of task design issues
that complicate interpretation; and that control participant
performance on the task is more variable than previously
believed. For these reasons, it is argued that the IGT is no
longer sufﬁcient to be a major source of evidence for the
SMH and additional empirical support should be sought.
3. Evaluation of the proposed neural substrate
of the SMH
A considerable strength of the SMH is that the neural
substrate considered to mediate such markers has been
speciﬁed in some detail (for an overview, see Damasio,
1994, 2004; Bechara and Damasio, 2005). Damasio draws a
distinction between two different kinds of stimuli that
require a decision-making response, each of which is
believed to be regulated by different regions of the brain.
‘Primary inducers’ are innate or learned stimuli that
generate pleasurable or aversive states, whereas ‘secondary
inducers’ are the thoughts and memories induced by the
recall or imagination of an emotional event.
As noted earlier, the pivotal area in the putative SMH
network is the portion of the frontal lobes above the eye
sockets, referred to as the VMPFC by the Iowa laboratory.
The VMPFC is believed to be the structure that encodes an
association between secondary inducers and the bioregulatory
state linked with that situation in the past experience of
the individual (including the bodily aspects of emotional
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271258
response). It is therefore centrally involved in the generation
of somatic markers when contemplating future response
options. These associations are ‘dispositional’ only, meaning
that they do not hold a representation of the situation or
bioregulatory state directly but can reactivate it through
triggering appropriate other regions of the brain. These
include regions of the brain that effect changes in the body
(hypothalamus, autonomic centres, and periaqueductal
grey) and represent change in the body (somatosensory
cortex, insula, cingulate, basal ganglia and brainstem
sensory/neurotransmitter nuclei). The amygdala is believed
to serve a similar function to the VMPFC for primary rather
than secondary inducers (Bechara et al., 2003).
The biasing action of somatic states on response
selection is proposed to be regulated by neurotransmitter
systems (including dopamine [DA], serotonin [5-HT],
noradrenaline [NA], and acetylcholine [Ach]) in the
brainstem. When the ‘as-if’ loop rather than the body loop
is being utilised, feedback from the body is short-circuited
by direct interactions between brainstem areas and regions
representing body-state (e.g. somatosensory cortex) [Fig. 1].
In more recent accounts, a distinction has been drawn
between posterior and anterior portions of the VMPFC. The
posterior region is believed to hold representations of
tangible, future events close in time that are more certain to
occur, whereas the anterior region is believed to hold
representations of abstract, future events further away in time
that are less likely to occur (Bechara and Damasio, 2005),
although there is little evidence to support this fractionation
at present. Fig. 1 displays a simpliﬁed representation of the
neural systems believed to be involved in the body loop and
‘as-if’ loop proposed by Damasio (Damasio, 1994; Bechara
and Damasio, 2005). For a more detailed account of regions
of the brain that represent and regulate the periphery, see
Craig (2002), Saper (2002), Critchley and Dolan (2004), and
Bernston et al. (2003).
The extent to which there is support for this neural
substrate will now be evaluated, focusing on lesion,
neuroimaging, and psychopharmacology data in turn.
3.1. Lesion data
If the SMH is correct, damage to the areas that regulate
and represent body-state change, in particular the VMPFC,
should impair performance on the IGT and related
paradigms. It is useful to spend some time describing the
anatomical classiﬁcation of the VMPFC before reviewing
available lesion data, as there is some controversy about how
to deﬁne it and this may explain why different studies have
generated a different pattern of ﬁndings. Damasio’s
classiﬁcation of VMPFC includes the entirety of the medial
and aspects of the lateral orbitofrontal cortex (OFC) and
overlaps with Brodmann’s areas 10, 11,12 (medial aspects),
lower 24, 25, and 32 (Bechara et al., 2000a; Bechara, 2004).
Other researchers use the term orbitofrontal cortex (OFC)
damage to refer to similar lesion cases that also extend to the
lateral orbital surface (Rogers et al., 1999; Manes et al.,
2002). The VMPFC/OFC is an anatomically heterogeneous
structure that communicates with a wide variety of other
brain areas. O¨ ngur and Price (2000) argue that a distinction
can be drawn between an orbital and a medial network within
this broad region that serve slightly different functions. The
orbital network is proposed to be a system for sensory
integration, crucially involving Brodmann’s areas 11, 13 and
12/47. The medial network is proposed to be a visceromotor
system crucially involving Brodmann’s areas 9, 10, 11, 13,
14, 24, 25, and 32. For the remainder of this article, the term
VMPFC will be adopted, since this is the one used by
Damasio and colleagues, but we note the lack of speciﬁcity of
this term.
Lesion work from the Iowa laboratory has generally
supported the neural substrate proposed in the SMH,
although some revisions to the original framework have
been made. As discussed earlier, it has been demonstrated in
a number of studies that lesions encompassing VMPFC do
reliably impair performance on the IGT (for example, see
Bechara et al., 1994, 1996, 1997a, 2000). Interestingly, it
appears that the earlier the age of VMPFC damage, the more
severe the deﬁcits in behaviour that emerge. Anderson et al.
(1999) studied two adults who had suffered prefrontal cortex
lesions occurring before age 16 months. They showed the
expected pattern of impaired decision-making despite
otherwise intact cognitive function. Additionally, deﬁcient
social and moral reasoning was found, suggesting that the
acquisition of complex social conventions and moral rules
had been impaired. This resulted in a syndrome resembling
sociopathy.
Also consistent with the SMH, damage to the amygdala
has been shown to alter performance on the task, but in a
subtly different way to VMPFC ablation (Bechara et al.,
1999). As well as being unable to form anticipatory markers
while pondering deck choice, the amygdala group also
failed to show any SCRs in response to winning and losing.
In a classical conditioning paradigm, the amygdala group
failed to generate SCRs to a visual stimulus that reliably
predicted the onset of a loud noise, whereas the VMPFC
group was unimpaired (Tranel et al., 1996). Bechara et al.
(1999) proposed that the amygdala is involved in linking
stimuli to affective attributes, whereas the VMPFC is crucial
for deciding amongst a variety of response options in terms
of the different affective responses they generate.
Consistent with the SMH, there is also preliminary
evidence showing that the insula and somatosensory cortex
need to be intact to acquire the IGT. In a conference abstract
Bechara et al. (1997b) describe performance of patients with
right-sided (nZ12) and left-sided lesions (nZ6) to the
somatosensory/insula cortex, in comparison to age- and
education-matched healthy control participants (nZ13).
The right-sided but not left-sided lesion group were
impaired on the IGT relative to control participants,
showing a behavioural preference for the disadvantageous
decks.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 259
A partial revision to the original neural substrate put
forward in the SMH is that only the right hemisphere
VMPFC appears to be implicated in IGT performance. To
explore the contribution of laterality, Tranel et al. (2002)
contrasted patients with bilateral, left only, and right only
VMPFC lesions. The bilateral and right only VMPFC lesion
group showed impaired everyday decision-making
(assessed by clinical interview) and a marked deﬁcit in
performance on the IGT. The left only VMPFC lesion group
was not severely impaired in everyday decision-making and
their performance on the IGT fell in the low-normal range.
These data imply that the type of decision-making measured
by the IGT crucially depends on the right hemisphere
VMPFC, although this conclusion can only be preliminary
due to the small sample size used.
One explanation for this ﬁnding is that the right
hemisphere has been associated with sensitivity to withdrawal
and punishment learning, whereas the left hemisphere
has been associated with sensitivity to approach and
reward learning by some authors (e.g. Davidson and Irwin,
1999). As discussed earlier, the reward/punishment schedule
on the IGT only varies in terms of punishment. It is
plausible that if reward rather than punishment was key to
the task, then left-sided VMPFC lesions would cause the
greatest impairment. Consistent with this interpretation, in
recent reviews from the Iowa laboratory, it has been
suggested that the positive somatic states are represented in
left-hemisphere VMPFC and negative somatic states are
represented in right-hemisphere VMPFC (Bechara and
Damasio, 2005). Also ﬁtting with a punishment account is
the ﬁnding that successful IGT performance positively
correlates with self-reported neuroticism (Carter and
Smith-Pasqualini, 2004). A slightly different interpretation
is provided by Crucian et al. (2000), who suggest that the
central role of the right hemisphere is to integrate cognitive
interpretation of emotional information and perceived
arousal.
Another change to the neural substrate of the SMH is that
additional regions of the PFC appear to inﬂuence
performance on the IGT. As discussed earlier, performance
on the IGT is often worse in those with damage extending
into the working memory/attentional control system of the
DLPFC (Bechara et al., 1998b; Bechara and Martin, 2004).
This ﬁnding is not particularly problematic for the SMH,
since one way through which ‘somatic markers’ are
believed to operate is by allocating processing resources
via interactions with the attentional control circuitry of the
DLPFC. It does make the predictions of the SMH difﬁcult to
distinguish from other, non-somatic theories, however.
In summary, a consistent performance deﬁcit has been
found on the IGT following damage to the VMPFC,
particularly in the right hemisphere. Further, the contribution
of this region to decision-making can be fractionated
from the role of related structures such as the amygdala and
DLPFC. Lesion work from other laboratories, however, has
provided more mixed support for the neural substrate of the
SMH.
When evaluating the lesion data on the IGT, it is
important to bear in mind that the nature of brain damage
means that lesions described in these studies are very rarely
conﬁned to clearly deﬁned cerebral areas. This means that
most of the patients with VMPFC damage studied on the
IGT (and other decision-making paradigms) often have
lesions extending into other portions of the frontal lobes and
basal forebrain (Clark and Manes, 2004). It is therefore
possible that these other areas, instead of or in addition to
VMPFC, are the crucial neural substrate regulating task
performance.
A range of evidence suggests that other regions of the
PFC may be crucially involved in IGT performance. Manes
et al. (2002) divided patients into categories of discrete OFC
lesions (including VMPFC), discrete dorsolateral PFC
lesions, discrete dorsomedial PFC lesions, and larger lesions
affecting both dorsal and ventral portions of PFC. Patients
with OFC lesions performed in the normal range on the IGT
and two other measures of decision-making, except that
they took longer to choose response options. The group with
large PFC lesions was impaired on the IGT (selecting more
from the disadvantageous decks) and on the two other
measures of decision-making. This would appear to suggest
that the VMPFC is not critical to performance on the IGT as
asserted in the SMH. These results may have been biased by
laterality differences in the lesions, however, since the OFC
group had predominantly left-sided lesions whereas the
large lesion group had predominantly right-sided lesions.
In a later study controlling for laterality confounds, Clark
et al. (2003) found that right frontal lesions (nZ21) severely
impaired IGT performance, whereas left frontal lesions (nZ
20) only led to a slight attenuation in performance relative to
controls (nZ20). Importantly, it was found that the extent of
the performance deﬁcit on the IGT positively correlated
with right-lesion but not left-lesion size. In particular, region
of interest analysis from MRI scans revealed that damage to
the right middle frontal gyrus, right superior frontal gyrus,
and right medial prefrontal cortex was linked to task
performance, indicating a prefrontal contribution to the IGT
extending beyond the VMPFC as deﬁned by Damasio.
3.2. Neuroimaging data
Functional neuroimaging provides an additional source
of evidence to evaluate the neural substrate of the SMH. In
particular, it can be usefully applied to study the activation
of the ‘as-if’ loop proposed by Damasio. A number of
studies have looked at neural activation while participants
perform the IGT. Ernst et al. (2002) asked participants to
perform a modiﬁed version of the IGT while in a positron
emission tomography (PET) scanner, with the blood ﬂow
tracer injection roughly coinciding with the second block of
20 trials on the task. The subtraction task used asked
participants to select cards from four decks in a speciﬁed
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271260
order, thereby isolating the active decision-making component
of the task. A predominantly right-sided network of
prefrontal and posterior cortical regions was activated,
including OFC, anterior cingulate/medial PFC, dorsolateral
PFC, insula, and inferior parietal cortex. Task performance
was found to correlate with regional cerebral blood ﬂow in
right ventrolateral PFC and right anterior insula.
A subsequent PET study using a similar protocol in
cocaine users (Bolla et al., 2003) allowed preliminary
investigation of how this network is altered in groups who
are impaired at behavioural acquisition of the IGT. Cocaine
use was associated with increased activation in right OFC
and decreased activation in right dorsolateral PFC, relative
to control participant performance. Superior performance
on the task correlated with right OFC activation in both
groups.
Adinoff et al. (2003) explored the relationship between
IGT performance and resting state blood ﬂow measured
using PET in healthy control participants (nZ15) and
abstinent cocaine-dependent individuals (nZ13). In contrast
to the ﬁndings discussed above, there was no
relationship between resting state activity in OFC and task
performance in either group. This was not due to a lack of
variability in task performance, since the cocaine-dependent
group showed considerable variance in the extent to which
they selected from the advantageous decks. Instead, IGT
performance correlated with resting state levels in anterior
cingulate and left DLPFC. This study was repeated with
patients in the more acute phase of cocaine abstinence
(Tucker et al., 2004), which found IGT performance
negatively correlated with activity in anterior cingulate
gyrus, middle frontal gyrus, medial frontal gyrus, and
superior frontal gyrus.
Similarly, Bolla et al. (2005) looked at neural activation
in abstinent marijuana users as they performed the IGT. The
marijuana users, particularly those who had a history of
heavy use, showed greater activation in the left cerebellum,
less activation in the right lateral OFC and right DLPFC,
while performing the IGT, relative to a control group with
no drug use history.
Ernst et al. (2003b) looked at neural network recruited by
the IGT in adults with attentional deﬁcit hyperactivity
disorder (ADHD), relative to control participants. In both
groups the task-activated ventral and dorsal PFC and the
insula. Despite no group differences in behavioural
performance, activation in ADHD participants was less
extensive and did not include anterior cingulate and
hippocampus. Further, the ADHD group showed decreased
activation in the insula and increased activity in caudal
portion of the right anterior cingulate.
Fukui et al. (2005) have recently examined the neural
substrate of the IGT in 15 healthy control volunteers using
event-related functional magnetic resonance imaging
(fMRI). The superior temporal resolution of fMRI makes
it possible to focus on the risk anticipation period of the task
only. Subtracting selections from the advantageous decks
from the disadvantageous decks in an event-related fMRI
design, they found that risk anticipation exclusively
activated the medial frontal gyrus. Further, this activation
correlated with the accuracy of behavioural performance
during the task. Surprisingly, the subtraction did not
highlight differences in orbitofrontal regions.
In summary, neuroimaging studies using the IGT provide
mixed support for the SMH. While parts of the neural
network put forward by Damasio for both the body-state
loop and ‘as-if’ loop have been linked to task performance,
there is mixed support for the central role of VMPFC/OFC.
It is important to note that the failure to reliably activate this
area may be partially due to methodological issues
associated with neuroimaging. PET studies have relatively
poor temporal resolution, meaning it is not possible to focus
solely on the risk anticipation period of each trial. MRI data
are know to have a number of areas of signal dropout
(including OFC) due to distortion artefacts (see Cusack
et al., 2005). A role for DLPFC in IGT performance is again
suggested from the neuroimaging data, which can be
integrated with the SMH by the notion that somatic markers
partially inﬂuence behaviour by boosting cognitive
resources allocated to viable response options.
3.3. Psychopharmacology data
Bechara and Damasio (2005) propose that the biasing
action of somatic states on response selection is mediated by
the release of neurotransmitters such as dopamine (DA),
serotonin (5HT) and norepinephrine (NE). For example, one
mechanism through which ‘somatic markers’ could aid
decision-making is through boosting cognitive resources
allocated to advantageous response options via neurotransmitter
modulation of attentional control systems.
The effects of systemic drug manipulations of these
systems on IGT performance have recently been examined,
ﬁnding that dopamine and serotonin are related to IGT
performance but noradrenaline appears not to be. Bechara
et al. (2001a) reported preliminary ﬁndings in a conference
abstract that blockade of dopamine impairs and stimulation
of dopamine improves performance on the early part of the
IGT in healthy control participants (nZ9), whereas
blockade of serotonin impairs and stimulation of serotonin
improves performance only on the latter part of the task in
healthy control participants (nZ9). They concluded therefore
that dopamine might be more important for covert
decision-making, whereas serotonin is more important for
overt decision-making. This interpretation crucially
depends on the cognitive penetrability of the IGT reward/
punishment schedule at different stages of the game,
however, and, as discussed earlier, it now appears that
participants are consciously aware of the good and bad
decks far earlier than previously assumed (Maia and
McClelland, 2004). Bechara et al. (2001a) ﬁndings are
also equally consistent with the idea that dopamine mediates
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 261
exploratory behaviour (Montague et al., 2004), so do not
offer exclusive support for the SMH.
O’Carroll and Papps (2003) compared IGT performance
in 30 healthy adult participants either administered a
placebo, 4 mg of reboxetine (a selective nor-adrenergic
reuptake inhibitor that boosts central noradrenergic [NA]
activity), and 8 mg of reboxetine. There were no differences
in performance between the three groups, suggesting that
NA is not crucially involved in the proposed biasing action
of somatic states on response selection, contra to the SMH.
3.4. Overview of neural substrate underpinning the SMH
Lesion, neuroimaging and psychopharamcology data to
date have to some extent supported the neural substrate put
forward in the SMH (see Fig. 1). Lesions to VMPFC,
amygdala, insula and somatosensory cortex have all been
found to impair performance on the IGT and are activated in
many neuroimaging studies of decision-making and bodystate
representation. A slight revision is needed to take into
account the fact that the right-hemisphere seems to be
implicated to a greater extent than the left-hemisphere and
that other regions of the PFC are also involved in the
acquisition of the IGT.
4. Further conceptual issues with the SMH
Some additional potential problems with the SMH at a
theoretical level will now be considered.
4.1. Novelty of the SMH: a historical overview
The novelty of the SMH has been challenged by some
authors (e.g. McGinn, 2003), based on the fact that there is a
long tradition of theory arguing that emotion-related
feedback from the body can mediate adaptive behaviour,
which predate the SMH. Further, a number of other ﬁelds
have been developing the notion that emotion can bias
decision-making in parallel to Damasio. This section
reviews historical and contemporary parallels to the SMH
and then evaluates the extent to which the SMH advances
the ﬁeld beyond these other accounts.
As discussed in the introduction, the SMH builds on
earlier peripheral feedback theories of emotion that argue
that emotion experience is based on the perception of
changes activity in the body (e.g. James, 1884, 1894; Lange,
1885). Similarly, in the early behavioural literature, twofactor
learning theory (Mowrer, 1947) linked conditioned
visceral activity to the mediation of instrumental responding.
It was observed that when a neutral stimulus is paired
with an unconditionally aversive event, the neutral stimulus
soon produced a conditioned visceral response. The
aversive emotion state associated with visceral activation
(for example, anxiety) was believed to motivate behaviour
in the organism to reduce it (for example, avoidance).
Empirical support for this position was mixed, however
(Mowrer, 1960). A poor correlation was generally found
between directly measurable autonomic responses and the
maintenance of instrumental responding. In addition,
autonomic responses were not always reliably found prior
to the instrumental response, undermining their causal
status. Further, dogs immobilised with the poison Curare
(who therefore cannot modify proprioceptive feedback to
the brain) were shown to be able to still acquire
discriminative avoidance responses on the basis of classical
conditioning, suggesting that peripheral feedback is not
essential to this learning mechanism (Solomon and Turner,
1962). It was therefore concluded that the instrumental
response and visceral response do not directly inﬂuence one
another but instead are both mediated by some other central
nervous system state that follows the rules of Pavlovian
conditioning (Rescorla and Solomon, 1967). Interestingly,
the lack of a causal relationship between visceral activity
and instrumental responding parallels the ﬁndings discussed
earlier that disturbance of feedback from the periphery does
not reliably alter IGT performance.
Damasio accounts for these ﬁndings by arguing that the
‘as-if’ loop rather than the body loop can be activated to
guide decision-making (e.g. see Damasio, 2004). Again, the
notion of the ‘as-if’ loop is also present in earlier accounts
(James, 1884, note 4; Marston, 1928; Tomkins, 1962). For
example, Marston wrote: “Though any given emotion,
experimentally tested, can be shown not to depend upon
sensation, may not the emotion have been built up,
originally, by compounding of sensations containing minute
differences from other major emotional compounds, and
subsequently remembered in connection with that type of
stimulus” (Marston, 1928, p. 56). Damasio extends and
more clearly speciﬁes how the ‘as-if’ loop could operate
than these earlier accounts, however.
False feedback studies, where participants are given
inaccurate information about responses taking place in their
body, also illustrate that body-state feedback had been
thought to inﬂuence decision-making prior to the conception
of the SMH. For example, Dienstbier and colleagues
produced a series of studies looking at the consequences of
false physiological feedback when contemplating acting
immorally (for review see Dienstbeir, 1978). Building on
the seminal attribution of arousal work by Schachter and
Singer (1962), they found that college students were more
likely to cheat if they had been informed that a vitamin pill
they had taken would produce increased physiological
arousal rather than reduced physiological arousal. In a
position very similar to the SMH, Dienstbeir (1978)
concluded that one of the ways in which we make decisions
is by assessing the emotional arousal that follows thinking
about each behavioural option. Batson et al. (1999) gave
people false physiological feedback while listening to
scenarios threatening to personal freedom and equality
values. When asked to make a decision implicating these
values, decisions favoured whichever value had received
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271262
stronger feedback while hearing the original situations.
Feedback did not, however, affect judgements about the
relative importance of the two values in general, which were
believed to derive from cognitive retrieval from memory of
a stored value hierarchy. These results are broadly
consistent with the SMH.
Perhaps the most similar framework to the SMH was
proposed by Nauta (1971), who argued the frontal lobes
integrate adaptive behaviour using feedback from the
periphery. Nauta came to this conclusion on the basis of
the strong reciprocal anatomical connections of the frontal
lobes with sensory processing regions and the limbic system
as described by MacLean (1949, 1975), speculating that a
key role of the frontal lobes was to integrate feedback from
the senses and use this to guide adaptive behaviour. Damage
to the frontal lobes would impair the ability to integrate
internal and external sensory information: “part at least of
the behavioural effects of frontal-lobe destruction could be
seen as the consequence of an ‘interoceptive agnosia’, i.e. an
impairment of the subject’s ability to integrate certain
informations from his internal milieu with the environmental
reports provided by neocortical processing mechanisms”
(p. 182). Further, this could then result in an inability to
regulate behaviour advantageously: “the reciprocal frontolimbic
relationship could be centrally involved in the
phenomenon of behavioural anticipation, and elucidate the
‘loss of foresight’ that has so long been recognised as one of
the most disabling consequences of massive frontal-lobe
lesions...The normal individual decides upon a particular
course of action by a thought process in which .strategic
alternatives are compared..The comparison in the ﬁnal
analysis is one between the affective responses evoked by
each of the various alternatives..It is entirely conceivable
that this anticipatory selection process is severely impaired
in the absence of the frontal cortex” (p. 183). Finally, part of
the affective response for each alternative is based on
feedback from internal sensory information: “a pre-setting
of ...interoceptive information...could be thought to
establish a temporal sequence of affective reference points
serving as ‘navigational markers”. (p. 183). This account,
predating the SMH by over 20 years, seems to have come to
remarkably similar conclusions. There were even some
attempts to experimentally test this framework through
disconnection studies in animal models (e.g. Divac et al.,
1975), although these generally offered little support for the
model. Similarly, Pribram (1970) suggested that feelings, or
sensations arising from lower brain regions and the body,
could serve as ‘monitors’ to co-ordinate behaviour. It should
be noted, however, that the SMH clearly extends these
frameworks by specifying much more clearly how this
system could be implemented in the brain.
There is also a large normative literature looking at the
impact of emotion on decision-making, which has evolved
in parallel to the SMH. In a review of this work,
Loewenstein and Lerner (2003) argued that a consideration
of emotional factors, both in terms of an affect experienced
at the time of the decision (immediate emotions) and the
affect an individual anticipates they will feel following a
particular response option (expected emotions), can help to
better model human decision-making than purely rational
accounts of choice selection. One illustration of a normative
model of decision-making where emotion is central is
‘affect as information’ theory (Clore, 1992), which proposes
that people monitor how they feel at the present time to help
them evaluate a situation. Therefore, if current feelings
happen to be positive, then the evaluation of a speciﬁc
decision-making option being contemplated is likely to be
positive. This appears to be particularly true for consideration
of unfamiliar choices where affect is likely to be
relevant (for example, choosing which movie to see).
Similarly, decision affect theory (see Mellers et al., 1997)
claims that the emotions that people experience after a
decision depend on a comparison between what the
consequences actually were and the consequences that
would have come about if another response option was
chosen. People are disappointed if the option they chose
produced a less beneﬁcial outcome than other alternatives
could have done but are pleased if their response led to a
superior outcome.
The notion that decision-making may be particularly
inﬂuenced by emotion at a more covert level is also present in
the cognitive literature, where a series of theorists have
converged on the notion that at least two parallel but
interacting routes to information processing may exist. One
route may use ‘high reason’ to cognitively appraise situations
at a more explicit level (a ‘cognitive’ loop) and the other
route may bias behaviour at a more implicit, automatic level
(an ‘automatic’ loop). As a representative example,
cognitive-experiential self-theory (CEST; Epstein, 1991)
will be outlined. CEST draws a distinction between rational
and experiential systems for processing information. The
experiential system is inﬂuenced largely by affect and
emotion (or ‘what feels good’) and is believed to be the
‘natural’ or ‘default’ mode of responding to situations
(Epstein et al., 1996). It produces rapid, reﬂexive responses
but is relatively slow to change and acts primarily at an
unconscious level. Its outputs are similar to the heuristics, or
cognitive short-cuts, postulated by Kahneman and Tversky
(1972) to simplify complex decision-making. The rational
system works at a conscious level and functions using
established rules and principles of inference. It is relatively
slow to respond but quick to change, generating complex,
dispassionate analysis. It has evolved more recently than the
experiential system and requires justiﬁcation via logic and
evidence. These two systems act in parallel, but interact with
each other regularly. The experiential system is believed to
be more important in the regulation of everyday behaviour,
automatically encoding, interpreting, and organising experience,
and directing behaviour. In other words, information
exists at both a conscious, verbal level and a pre-conscious,
experiential level. Similar accounts of the interface between
cognition and emotion have been put forward in other
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 263
multiple representation frameworks, including Interactive
Cognitive Subsystems (ICS; Teasdale and Barnard, 1993),
the Schematic Propositional Analogical and Associative
Representation System (SPAARS; Power and Dalgleish,
1997), and Lambie and Marcel’s (2002) account of emotion
experience. It is important to observe, however, that at
present many of these models are less well speciﬁed than the
SMH, particularly at the neural level.
The critical question that this historical synopsis raises
is what is the unique contribution that the SMH makes to
the literature? The notion that emotion, partially via
peripheral mechanisms, inﬂuences decision-making is
reﬂected in a variety of earlier and parallel models
(Mowrer, 1960; Pribram, 1970; Nauta, 1971). Further, the
notion that emotion-based inﬂuences may act at a more
implicit, covert level of decision-making (Epstein, 1991)
and that a representation of expected body-state change
in the brain is sufﬁcient to shape higher level processing
without waiting for actual bodily responses (for example,
see James, 1884 note 4; Marston, 1928) are also present
in earlier accounts.
The resonance of the SMH with earlier work is both a
strength and weakness; it gives the model strong
concurrent validity with over a hundred years of
psychological theory but at the same time it challenges
the novelty of the framework. Perhaps the value of the
SMH is the way it integrates these components from
different models into a unitary framework, although these
earlier inﬂuences are not always explicitly apparent in
Damasio’s writings. Clear advances in the SMH are that it
expands the range of bodily states underlying emotions
beyond visceral feedback (e.g. including hormonal feedback
through the bloodstream) and that it puts forward a
more elaborate psychological mechanism and a clearer
neural substrate to regulate how emotion and peripheral
feedback shape decision-making. Further, Damasio
extends and more clearly speciﬁes the role of the ‘as-if’
loop compared to earlier accounts, and discusses more
fully how feedback from the body is integrated with
cognitive appraisal to mediate both emotion experience
and decision-making (see Damasio, 1994, p. 139). Perhaps
most importantly, Damasio outlines how emotion-based
biasing signals can act at a variety of levels, both
conscious and non-conscious. These advances make the
SMH more robust in the face of the criticisms usually
voiced against Jamesian models (e.g. Cannon, 1927), in
particular allowing the system to be rapid, ﬂexible, and to
subtly differentiate between complex emotion states.
Further, the ‘as-if’ loop can explain why surgical isolation
of the periphery does not always impair emotion
experience or decision-making. Finally, through the
development of novel experimental tools and the study
of lesion groups, Damasio has provided ways to directly
empirically test these ideas. Therefore, the SMH does
appear to be making a novel and valuable contribution to
the literature.
4.2. Speciﬁcation of the somatic marker mechanism
Perhaps more problematic is the fact that the ‘somatic
marker’ mechanism is currently poorly speciﬁed. According
to SMH, a large number of somatic signals, for example
from the viscera, vascular bed, skeletomotor system and
endocrine system need to be integrated into a pattern image
that marks outcomes as either ‘good’ or ‘bad’. It is
unspeciﬁed in the SMH how this complex data reduction
is computed. However, a number of physiological plausibe
data reduction algorithms (such as principal components
analysis) could be used to reduce the high-dimensional
body-state pattern into a low-dimensional ‘somatic marker’.
Work in other ﬁelds has begun to apply these data reduction
approaches to understand other complex problems. For
example, Bar-Gad et al. (2003) propose a model of how the
basal ganglia compresses cortical information, effectively
acting as a central dimensionality reduction system which is
modulated by a reinforcement signal. The mechanism by
which this could be achieved in the SMH is currently underspeciﬁed
and would beneﬁt from more detailed examination.
Other researchers have begun to map out in detail how
signals from the body are relayed to the brain (e.g. the
lamina I spinothalamocortical system; Craig, 2002), which
could aid this process.
4.3. Parsimony of the SMH
Another criticism that has been voiced against the SMH
is that the theory does not provide the most parsimonious
explanation of the phenomenon of interest (e.g. Rolls,
1996). Damasio’s theory neglects to explain what generates
speciﬁc patterns of body-state change in the ﬁrst place and
presumably some kind of central, cognitive appraisal must
be taking place (i.e. the system must ‘know’ that a ‘bad’
decision is about to be made to generate an SCR). Given that
this is occurring, it would seem a more parsimonious
solution for this system to communicate directly with higher
levels of representation, rather than taking the rather
circuitous route through the body. Rolls (1996) argued
that it would be very inefﬁcient for the execution of
behaviour following the appraisal of the value of a stimulus
to have to go through the periphery or a central
representation of the periphery. Instead, it would be much
more adaptive to have a direct connection to output motor
centres to implement the target behaviour.
In fact, rather early on proponents of the SMH
anticipated the parsimony criticism (Damasio et al., 1991).
They essentially argued that the ‘somatic marker’ system is
evolutionarily ancient, and has proved to be highly
effective. Further, they argued that powerful ‘somatic
markers’ are necessary to effectively reduce complex
decision spaces. However, a number of computational
algorithms, in particular in computational reinforcement
learning, have been explicitly designed for decision-making
under uncertainty, essentially by using approximations for
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271264
the value of future actions in order to guide action selection
(Montague et al., 2004). These types of algorithm can be
used to solve n-armed bandit problems, of which the IGT is
a variant, and are also well speciﬁed at an anatomical level.
Therefore, the neural instantiation of such decision-making
mechanisms does not seem to necessitate the use of the
complex somatic and emotion-related machinery described
by Damasio. Interestingly, recent work from the Iowa
laboratory utilises a reinforcement learning algorithm to
model IGT data without making any reference to the SMH
(Oya et al., 2005). A number of other non-somatic
mechanisms for constraining decision-making (e.g. cognitive
heuristics; Kahneman and Tversky, 1972) have also
been proposed. One way to defend this issue would be to
argue that these reinforcement learning algorithms are
simply the instantiation of the central components of the
‘somatic marker’ system, rather than a genuinely alternative
account (see Damasio, 1994).
4.4. Emphasis on the periphery in the SMH
The SMH has also been criticised for being ‘somatocentric’
(Panksepp, 2003). Panksepp suggested that Damasio
has taken the peripheral feedback theories to an extreme
by saying that most mental states are made up purely of
different types of bodily awareness. Instead, Panksepp
suggests that a more moderate conclusion is warranted,
whereby most cognitive states include some kind of
emotional aspect, made up partially of somatosensory
feedback but also other components. For example,
consideration of the different ‘action apparatus’ of the
brain that has evolved to regulate different emotion states
could help to clarify how relatively crude peripheral biasing
signals contribute to distinct, reﬁned emotion states. Craig
(2002) also suggests that the SMH has focused on the
representation of body-state at the cost of other aspects of
emotion processing, in particular neglecting how these
representations of body-state can then motivate behavioural
action. Similar points in relation to the SMH have been
made by McGinn (2003).
These criticisms seem a slight distortion of the SMH,
however, since Damasio (1999) explicitly includes a mental
evaluative component that interprets emotional sensations
of bodily change to lead to more complex emotion
experience. In fact, some kind of speciﬁcation of how
body-state interacts with central mental evaluation seems a
positive advance in the SMH compared to other Jamesian
theories (see also Prinz’s ‘somatic appraisal theory’, 2004).
5. Conclusion: current status of the somatic
marker hypothesis
The SMH (Damasio, 1994) represents an intriguing
model of how feedback from the body may contribute to
successful decision-making in situations of complexity and
uncertainty. This builds on earlier work linking activity in
the body to emotion experience (e.g. James, 1884, 1894;
Lange, 1885) and decision-making (e.g. Pribram, 1970;
Nauta, 1971). Key support for this theory has been largely
drawn from data on the IGT, a decision-making task that has
been claimed to rely on emotion-related feedback from the
body to guide accurate performance (Bechara et al., 1996).
While the IGT has proved to be a sensitive, ecologically
valid measure of decision-making impairment that has
generated a large body of empirical research, three key
assumptions that need to be held for it to offer support for
the SMH may not be tenable. First, the claim that the task
measures implicit learning as the reward/punishment
schedule is cognitively impenetrable (although see Bechara
et al., 2005) is inconsistent with data showing accurate
knowledge of the task contingencies (Maia and McClelland,
2004) and that mechanisms such as working-memory or
cognitive outcome expectancies appear to exert a strong
inﬂuence on task performance. Second, the assertion that
this learning takes place via anticipatory marker signals
arising from the body is not supported by competing
explanations of the psychophysiology proﬁle generated on
the task (Tomb et al., 2002), the failure to establish a clear
causal relationship between disturbed feedback from the
periphery and impaired decision-making (e.g. Heims et al.,
2004), and the possibility that the ‘anticipatory’ changes
may actually reﬂect expectancies about reward and punishment
generated after a decision has been made (Amiez et al.,
2003). Third, the assumption that the task impairment is due
to a ‘myopia for the future’ is undermined by the existence
of a number of other perhaps better speciﬁed and more
plausible psychological mechanisms explaining deﬁcits on
the task (including reversal learning, risk-taking, and
working-memory deﬁcits). In addition, there may be greater
variability in control performance on the task than
previously believed (complicating interpretation of empirical
ﬁndings).
While the psychological mechanism underpinning the
SMH therefore seems to require some revision, the neural
substrate that Damasio has proposed has been reasonably
well supported by the ﬁndings to date. Lesion and
neuroimaging studies have found that VMPFC, amygdala,
somatosensory cortex, insula and related areas are involved
in decision-making processes as suggested in the framework.
This is a clear advance on earlier models of decisionmaking.
The neural substrate of the SMH still needs further
clariﬁcation in the light of ongoing advances in knowledge
about VMPFC anatomy, however (e.g. O¨ ngur and Price,
2000).
Therefore, the SMH seems to have accurately identiﬁed
some of the brain regions involved in decision-making,
emotion, and body-state representation, but exactly how
they interact at a psychological level is still somewhat
unclear.
Of course, none of these reservations falsify the SMH;
they just suggest that other sources of evidence need to be
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 265
gathered to support it. Indeed, we remain open to the idea
that feedback from the periphery can inﬂuence higher-level
cognitive and emotional processes. Bechara et al. (2005)
have recently proposed a series of unanswered questions
about the SMH that could be usefully explored in future
research. These include assessing if different kinds of
decision-making (e.g. under certainty or uncertainty) recruit
different neural networks, identifying when emotions are
unhelpful as well as helpful when making decisions (e.g.
Shiv et al., 2005), further specifying the nature of biasing
signals that guide decision-making, fractionating the
underlying cognitive processing involved in complex
decision-making tasks, and exploring individual differences
in decision-making. We wholeheartedly agree that consideration
of these issues would improve the literature on
both the SMH and broader decision-making processes. Here
we suggest some speciﬁc future studies.
A variety of potential designs could be used to further
test the SMH. One approach is to develop variants of the
IGT that control some of the methodological issues and task
interpretation ambiguities raised. This could include
removing the reversal learning confound and making the
task less easy to consciously comprehend (for example,
using a variant artiﬁcial grammar learning paradigm).
Another approach would be to more aggressively pursue
causal tests of the SMH in a wider range of populations with
altered peripheral feedback (e.g. patients with facial
paralysis, Keillor et al., 2002). As well as looking at patient
groups with impaired somatosensory feedback, pharmacological
challenge studies in healthy control volunteers could
be run. For example, the impact of the peripherally acting
beta-blocker nadalol on IGT performance could be
measured. Moreover, different causal methodologies could
be combined to minimise the amount of feedback from the
body reaching the brain (for example, looking at the impact
of peripheral blockade in patients with PAF). An alternative
body of existing literature that could speak to the validity of
the SMH is false feedback experiments (e.g. see Crucian
et al., 2000). It may be possible to use false feedback while
patient groups with disturbed bodily feedback perform the
IGT (e.g. Linton and Hirt, 1979) to further test the
framework. Finally, an individual differences approach
could be adopted, for example looking at how variance in
interoceptive ability relates to IGT performance (e.g. see
Katkin et al., 2001). As well as looking at behavioural
measures in these studies, it is important for future studies to
simultaneously record a wider range of peripheral variables
(e.g. facial electromyography, heart rate) and central
variables (e.g. fMRI and event-related potentials) wherever
possible (e.g. Oya et al., 2005).
Until a broader range of empirical approaches are used to
test the SMH, the current status of the framework would
appear to be that it is an intriguing idea in need of some
clearer speciﬁcation and some more supporting evidence.
Despite these limitations, the SMH and the IGT have made a
valuable contribution to the literature. They have helped to
reintroduce the idea that emotion can be a beneﬁt as well as
a hindrance when making a decision to the wider
neuroscience community; they have outlined a plausible
neural substrate that could underpin this mechanism, and
they have developed an innovative experimental paradigm
to test the framework. Through the development of the ‘asif’
loop, the SMH also addresses a number of weaknesses in
other Jamesian models and can better account for the speed
and complexity of emotion responses that are typically
observed. The IGT has been an extremely generative
paradigm, applied to a wide range of neurological and
psychiatric conditions and used in numerous experimental
studies. Further, the SMH has importantly extended
MacLean’s pioneering limbic system framework in order
to better understand how emotion is represented in the brain
and to more clearly model the inﬂuence emotion can have
on other cognitive processes.
Acknowledgements
This review was supported by the Medical Research
Council of the United Kingdom. Address reprint requests to
Dr Barnaby Dunn, MRC CBU, 15 Chaucer Road,
Cambridge, CB2 2EF, UK; barney.dunn@mrc-cbu.cam.ac.
uk (e-mail)
References
Adelmann, P.K., Zajonc, R.B., 1989. Facial efference and the experience
of emotion. Annual Review of Psychology 40, 249–280.
Adinoff, B., Devous Sr., M.D., Cooper, D.B., Best, S.E., Chandler, P.,
Harris, T., et al., 2003. Resting regional cerebral blood ﬂow and
gambling task performance in cocaine-dependent subjects and healthy
comparison subjects. American Journal of Psychiatry 160 (10),
1892–1894.
Amiez, C., Procyk, E., Honore, J., Sequeira, H., Joseph, J.-P., 2003. Reward
anticipation, cognition, and electrodermal activity in the conditioned
monkey. Experimental Brain Research 149, 267–275.
Anderson, S.W., Bechara, A., Damasio, H., Tranel, D., Damasio, A.R.,
1999. Impairment of social and moral behaviour related to early damage
in human prefrontal cortex. Nature Neuroscience 2 (11), 1032–1037.
Apkarian, A.V., Sosa, Y., Krauss, B.R., Thomas, P.S., Fredrickson, B.E.,
Levy, R.E., et al., 2004. Chronic pain patients are impaired on an
emotional decision-making task. Pain 108 (1–2), 129–136.
Bar-Gad, I., Morris, G., Bergman, H., 2003. Information processing,
dimensionality reduction and reinforcement learning in the basal
ganglia. Progress in Neurobiology 71, 439–473.
Bark, R., Dieckmann, S., Bogerts, B., Northoff, G., 2005. Deﬁcit in decision
making in catatonic schizophrenia: an exploratory study. Psychiatry
Research 134, 131–141.
Barrash, J., Tranel, D., Anderson, S.W., 2000. Acquired personality
distrubances associated with bilateral damage to the ventromedial
prefrontal region. Developmental Neuropsychology 18 (3), 355–381.
Batson, C.D., Engel, C.L., Fridell, S.R., 1999. Value judgements: testing
the somatic-marker hypothesis using false physiological feedback.
Personality and Social Psychology Bulletin 25 (8), 1021–1032.
Bechara, A., 2004. The role of emotion in decision-making: evidence from
neurological patients with orbitofrontal damage. Brain and Cognition
55 (1), 30–40.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271266
Bechara, A., Damasio, H., 2002. Decision-making and addiction (part I):
impaired activation of somatic states in substance dependent
individuals when pondering decisions with negative future consequences.
Neuropsychologia 40 (10), 1675–1689.
Bechara, A., Damasio, A.R., 2005. The somatic marker hypothesis: a neural
theory of economic decision. Games and Economic Behaviour 52(2),
336–372.
Bechara, A., Martin, E.M., 2004. Impaired decision making related to
working memory deﬁcits in individuals with substance addictions.
Neuropsychology 18 (1), 152–162.
Bechara, A., Damasio, A., Damasio, H., Anderson, S., 1994. Insensitivity to
future consequences following damage to human prefrontal cortex.
Cognition 50, 7–15.
Bechara, A., Tranel, D., Damasio, H., Damasio, A.R., 1996. Failure to
respond autonomically to anticipated future outcomes following
damage to prefrontal cortex. Cerebral Cortex 6 (2), 215–225.
Bechara, A., Damasio, H., Tranel, D., Damasio, A.R., 1997a. Deciding
advantageously before knowing the advantageous strategy. Science
275, 1293–1295.
Bechara, A., Tranel, D., Damasio, H., Damasio, A.R., 1997b. An
anatomical system subserving decision-making. Society for Neuroscience
Abstracts 23, 495.
Bechara, A., Damasio, H., Tranel, D., Anderson, S.W., 1998a. Dissociation
of working memory from decision-making within the human prefrontal
cortex. The Journal of Neuroscience 18 (1), 428–437.
Bechara, A., Tranel, D., Wilson, J., Heberlein, A.S., Ross, M., Damasio,
A.R., 1998b. Impaired decision-making in peripheral neuropathy.
Society for Neuroscience Abstracts 24, 1176.
Bechara, A., Damasio, H., Damasio, A.R., Lee, G.P., 1999. Different
contributions of the human amygdala and the ventromedial prefrontal
cortex to decision-making. Journal of Neuroscience 19 (13), 5473–
5481.
Bechara, A., Damasio, H., Damasio, A.R., 2000a. Emotion, decision
making and the orbitofrontal cortex. Cerebral Cortex 10, 295–307.
Bechara, A., Tranel, D., Damasio, H., 2000b. Characterization of the
decision-making deﬁcit of patients with ventromedial prefrontal cortex
lesions. Brain 123, 2189–2202.
Bechara, A., Damasio, H., Damasio, A.R., 2001a. Manipulation of
dopamine and serotonin causes different effects on covert and overt
decision-making. Society for Neuroscience Abstracts 27, 126.
Bechara, A., Dolan, S., Denburg, N., Hindes, A., Anderson, S.W., Nathan,
P.E., 2001b. Decision-making deﬁcits, linked to a dysfunctional
ventromedial prefrontal cortex, revealed in alcohol and stimulant
abusers. Neuropsychologia 39 (4), 376–389.
Bechara, A., Dolan, S., Hindes, A., 2002. Decision-making and addiction
(part II): myopia for the future or hypersensitivity to reward?
Neuropsychologia 40 (10), 1690–1705.
Bechara, A., Damasio, H., Damasio, A.R., 2003. Role of the amygdala in
decision-making. Annals of the New York Academy of Science 985,
356–369.
Bechara, A., Damasio, H., Tranel, D., Damasio, A.R., 2005. The Iowa
Gambling Task and the somatic marker hypothesis: some questions and
answers. Trends in Cogntive Sciences 9 (4), 159–162.
Beninger, R.J., Wasserman, J., Zanibbi, K., Charbonneau, D., Mangels, J.,
Beninger, B.V., 2003. Typical and atypical antipsychotic medications
differentially affect two nondeclarative memory tasks in schizophrenic
patients: a double dissociation. Schizophrenia Research 61, 281–292.
Bernston, G.G., Sarter, M., Cacioppo, J.T., 2003. Ascending visceral
regulation of cortical affective information processing. European
Journal of Neuroscience 18, 2103–2109.
Best, M., Williams, J.M., Coccaro, E.F., 2002. Evidence for a dysfunctional
prefrontal circuit in patients with an impulsive aggressive disorder.
Proceedings of the National Academy of Science USA 99 (12),
8448–8453.
Bolla, K.I., Eldreth, D.A., London, E.D., Kiehl, K.A., Mouratidis, M.,
Contoreggi, C., et al., 2003. Orbitofrontal cortex dysfunction in
abstinent cocaine abusers performing a decision-making task. Neuroimage
19 (3), 1085–1094.
Bolla, K.I., Eldreth, D.A., Matochik, J.A., Cadet, J.L., 2004. Sex-related
differences in a gambling task and its neurological correlates. Cerebral
Cortex 14 (11), 1226–1232.
Bolla, K.I., Eldreth, D.A., Matochik, J.A., Cadet, J.L., 2005. Neural
substrates of faulty decision-making in abstinent marijuana users.
Neuroimage 26, 480–492.
Bowman, C.H., Turnbull, O.H., 2003. Real versus facsimile reinforcers on
the Iowa Gambling Task. Brain and Cognition 53 (2), 207–210.
Bowman, C.H., Evans, C.E., Turnbull, O.H., 2005. Artiﬁcial time
constraints on the Iowa Gambling Task: the effects on behavioural
performance and subjective experience. Brain and Cognition 57 (1),
21–25.
Bradley, M.M., Codispoti, M., Cuthbert, B.N., Lang, P.J., 2001a. Emotion
and motivation I: defensive and appetitive reactions in picture
processing. Emotion 1 (3), 276–298.
Bradley, M.M., Codispoti, M., Sabatinelli, D., Lang, P.J., 2001b. Emotion
and motivation II: sex differences in picture processing. Emotion 1 (3),
300–319.
Busemeyer, J.R., Stout, J.C., 2002. A contribution of cognitive decision
models to clinical assessment: decomposing performance on the
Bechara gambling task. Psychological Assessment 14, 253–262.
Camille, N., Coricelli, G., Sallet, J., Pradat-Diehl, P., Duhamel, J.R.,
Sirigu, A., 2004. The involvement of the orbitofrontal cortex in the
experience of regret. Science 304, 1167–1170.
Campbell, M.C., Stout, J.C., Finn, P.R., 2004. Reduced autonomic
responsiveness to gambling task losses in Huntington’s disease. Journal
of the International Neuropsychological Society 10, 239–245.
Cannon, W.B., 1927. The James–Lange theory of emotions: a critical
examination and an alternative theory. American Journal of Psychology
39, 106–124.
Carter, S., Smith-Pasqualini, M.C., 2004. Stronger autonomic response
accompanies better learning: a test of Damasio’s somatic marker
hypothesis. Cognition and Emotion 18 (7), 901–911.
Cavedini, P., Riboldi, G., D’Annucci, A., Belotti, P., Cisima, M., Bellodi,
L., 2002a. Decision-making heterogeneity in obsessive compulsive
disorder: ventromedial prefrontal cortex function predicts different
treatment outcomes. Neuropsychologia 40, 205–211.
Cavedini, P., Riboldi, G., Keller, R., D’Annucci, A., Bellodi, L., 2002b.
Frontal lobe dysfunction in pathological gambling patients. Biological
Psychiatry 51 (4), 334–341.
Cavedini, P., Bassi, T., Ubbiali, A., Casolari, A., Giordani, S., Zorzi, C.,
2004a. Neuropsychological investigation of decision-making in
anorexia nervosa. Psychiatry Research 127, 259–266.
Cavedini, P., Bassi, T., Zorzi, C., Bellodi, L., 2004b. The advantages of
choosing antiobsessive therapy according to decision-making functioning.
Journal of Clinical Psychopharmacology 24 (6), 628–631.
Cheung, T.H., Cardinal, R.N., 2005. Hippocampal lesions facilitate
instrumental learning with delayed reinforcement but induce impulsive
choice in rates. BMC Neurosci 6(1), 36.
Chwalisz, K., Diener, E., Gallagher, D., 1988. Autonomic arousal feedback
and emotional experience: evidence from the spinal cord injured.
Journal of Personality and Social Psychology 54, 820–828.
Clark, L., Manes, F., 2004. Social and emotional decision-making
following frontal lobe injury. Neurocase 10 (5), 398–403.
Clark, L., Iversen, S.D., Goodwin, G.M., 2001. A neuropsychological
investigation of prefrontal cortex involvement in acute mania.
American Journal of Psychiatry 158 (10), 1605–1611.
Clark, L., Manes, F., Antoun, N., Sahakian, B.J., Robbins, T.W., 2003. The
contributions of lesion laterality and lesion volume to decision-making
impairment following frontal lobe damage. Neuropsychologia 41 (11),
1474–1483.
Clore, G.L., 1992. Cognitive phenomenology: feelings and the construction
of judgement. In: Martin, L.L., Tesser, A. (Eds.), The Construction of
Social Judgements. Erlbaum, Hillsdale, NJ, pp. 133–163.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 267
Cobos, P., Sanchez, M., Garcia, C., Vera, M., Vila, J., 2002. Revisiting the
James versus Cannon debate on emotion: startle and autonomic
modulation in patients with spinal cord injuries. Biological Psychology
61, 251–269.
Craig, A.D., 2002. How do you feel? Interoception: the sense of the
physiological condition of the body. Nature Reviews Neuroscience 3,
655–666.
Critchley, H.D., Dolan, R.J., 2004. Central representation of autonomic
states. In: Frackowiak, R., Friston, K., Frith, C., Dolan, R., Price, C.,
Zeki, S., Ashburner, J., Penny, W. (Eds.), Human Brain Function,
second ed. Elsevier, London, UK, pp. 397–418.
Critchley, H.D., Good, C.D., Ashburner, J., Frackowiak, R.S., Mathias,
C.J., Dolan, R.J., 2003. Changes in cerebral morphology consequent to
peripheral autonomic denervation. Neuroimage 18, 908–916.
Crone, E.A., van der Molen, M.W., 2004. Developmental changes in real
life decision making: performance on a gambling task previously shown
to depend on the ventromedial prefrontal cortex. Developmental
Neuropsychology 25 (3), 251–279.
Crone, E.A., Somsen, R.J., Van Beek, B., Van Der Molen, M.W., 2004.
Heart rate and skin conductance analysis of antecendents and
consequences of decision making. Psychophysiology 41 (4), 531–540.
Crucian, G.P., Hughes, J.D., Barrett, A.M., Williamson, D.J.G., Bauer,
R.M., Bowers, D., 2000. Emotional and physiological responses to false
feedback. Cortex 36, 623–647.
Cusack, R., Russell, B., Cox, S.M.L., De Panﬁlis, C., Schwarzbauer, C.,
Ansorge, R., 2005. An evaluation of the use of passive shimming to
improve frontal sensitivity in fMRI. Neuroimage 24, 82–91.
Dalgleish, T., 2004. The emotional brain. Nature Neuroscience Reviews 5
(7), 583–589.
Dalgleish, T., Yiend, J., Bramham, J., Teasdale, J.D., Ogilvie, A.D., Malhi,
G., et al., 2004. Neuropsychological processing associated with
recovery from depression after stereotactic subcaudate tractotomy.
American Journal of Psychiatry 161 (10), 1913–1916.
Damasio, A.R., 1979. The frontal lobes. In: Heilman, K.M., Valenstein, E.
(Eds.), Clinical Neuropsychology. Oxford University Press, New York.
Damasio, A.R., 1994. Descartes Error: Emotion, Reason and the Human
Brain. Avon, New York, pp. 350–412.
Damasio, A.R., 1996. The somatic marker hypothesis and the possible
functions of the prefrontal cortex. Philosophical Transactions of the
Royal Society of London (series B) 351 (1346), 1413–1420.
Damasio, A.R., 1998. Emotion in the perspective of an integrated nervous
system. Brain Research Reviews 26, 83–86.
Damasio, A.R., 1999. The Feeling of What Happens: Body and Emotion in
the Making of Consciousness. Harcourt, New York.
Damasio, A.R., 2004. William James and the modern neurobiology of
emotion. In: Evans, D., Cruse, P. (Eds.), Emotion, Evolution and
Rationality. Oxford University Press, Oxford, pp. 3–14.
Damasio, A.R., Tranel, D., Damasio, H., 1990. Individuals with sociopathic
behavior caused by frontal damage fail to respond autonomically to
social stimuli. Behaviour Brain Research 41 (2), 81–94.
Damasio, A.R., Tranel, D., Damasio, H., 1991. Somatic markers and the
guidance of behaviour: theory and preliminary testing. In: Levin, H.S.,
Eisenberg, H.M., Benton, A.L. (Eds.), Frontal Lobe Function and
Dysfunction. Oxford University Press, New York, pp. 217–229.
Damasio, H.T., Grabowski, R., Frank, A.M., Galaburda, A.M., Damasio,
A.R., 1994. The return of Phineas Gage: the skull of a famous patient
yields clues about the brain. Science 264, 1102–1105.
Damasio, H., Bechara, A., Damasio, A., et al., 2002. Nature Neuroscience 5
(11), 1104 (Reply to Tomb et al.).
Damasio, A.R., Adolphs, R., Damasio, H., 2003. The contributions of the
lesion method to the functional neuroanatomy of emotion. In:
Davidson, R.J., Scherer, K.R., Goldsmith, H.H. (Eds.), Handbook of
Affective Sciences. Oxford University Press, Oxford, pp. 66–92.
Davidson, R.J., Irwin, W., 1999. The functional neuroanatomy of emotion
and affective style. Trends in Cognitive Science 3 (1), 11–21.
Davis, C., Levitan, R.D., Muglia, P., Bewell, C., Kennedy, J.L., 2004.
Decision-making deﬁcits and overeating: a risk model for obesity.
Obesity Research 12 (6), 929–935.
Denburg, N.L., Tranel, D., Bechara, A., Damasio, A.R., 2001. Normal
ageing may compromise the ability to decide advantageously. Brain and
Cogntion 47, 156–159.
Denburg, N.L., Tranel, D., Bechara, A., 2005. The ability to decide
advantageosuly declines prematurely in some normal older persons.
Neuropsychologia 43 (7), 1099–1106.
Dienstbeir, R.A., 1978. In: Harvey, J.H., Ickes, W.J., Kidds, R.F. (Eds.),
Attribution, Socialization, and Moral Decision-Making New Directions
in Attribution Research, vol. 2. Lawrence Erlbaum, Hillsdale NJ,
pp. 181–206.
Dimitrov, M., Phipps, M., Zahn, T., Grafman, J., 1999. A thoroughly
modern Gage. Neurocase 5, 345–354.
Divac, I., Gade, A., Wikmark R.G., 1975. Taste aversion in rats with lesions
in the frontal lobes: No evidence for intoreceptive agnosia.
Physiological Psychology 3, 43–46.
Ellsworth, P.C., 1994. William James and emotion: is a century of fame
worth a century of misunderstanding. Psychological Review 101 (2),
222–229.
Epstein, S., 1991. Cognitive-experiential self-theory: an integrative theory
of personality. In: Curtis, R. (Ed.), The Self with Others: Convergences
in Psychoanalytic, Social and Personality Psychology. Guilford, New
York, pp. 111–137.
Epstein, S., Pacini, R., Denes-Raj, V., Heier, H., 1996. Individual
differences in intuitive–experiential and analytical–rational thinking
styles. Journal of Personality and Social Psychology 71, 390–405.
Ernst, M., Bolla, K., Mouratidis, M., Contoreggi, C., Matochik, J.A.,
Kurian, V., 2002. Decision-making in a risk-taking task: a PET study.
Neuropsychopharmacology 26, 682–691.
Ernst, M., Kimes, A.S., London, E.D., Matochik, J.A., Eldreth, D., Tata, S.,
2003a. Neural substrates of decision making in adults with Attentional
Deﬁcit Hyperactivity Disorder. American Journal of Psychiatry 160 (6),
1061–1070.
Ernst, M., Grant, S.J., London, E.D., Contoreggi, C., Kimes, A.S.,
Spurgeon, L., 2003b. Decision making in adolescents with behaviour
disorders and adults with substance abuse. American Journal of
Psychiatry 160 (1), 33–40.
Eslinger, P.J., Damasio, A.R., 1985. Severe disturbance of higher cognition
after bilateral frontal lobe ablation: patient EVR. Neurology 35,
1731–1741.
Evans, C.E., Kemish, K., Turnbull, O.H., 2004. Paradoxical effects of
education on the Iowa Gambling Task. Brain and Cognition 54 (3),
240–244.
Fein, G., Klein, L., Finn, P., 2004. Impairment on simulated gambling task
in long term long-term abstinent alcoholics. Alcoholism: Clinical and
Experimental Research 28 (10), 1–5.
Fellows, L.K., Farah, M.J., 2003. Ventromedial frontal cortex mediates
affective shifting in humans: evidence from a reversal learning
paradigm. Brain 126 (Pt 8), 1830–1837.
Fellows, L.K., Farah, M.J., 2005a. Different underlying impairments in
decision-making following ventromedial and dorsolateral frontal lobe
damage in humans. Cerebral Cortex 15 (1), 58–63.
Fellows, L.K., Farah, M.J., 2005b. Dissociable elements of human
foresight: a role for the ventromedial frontal lobes in framing the
future, but not in discounting future rewards. Neuropsychologia 43,
1214–1221.
Fishbein, D., Eldreth, D., Matochik, J., Isenberg, I., Hyde, C., London, E.,
2005. Cognitive performance and autonomic reactivity in abstinent
drug abusers and nonusers. Experimental and Clinical Psychopharmacology
13 (1), 25–40.
Franken, I.H.A., Muris, P., 2005. Individual differences in decisionmaking.
Personality and Individual Differences 39, 991–998.
Fukui, H., Murai, T., Fukuyama, H., Hayashi, T., Hanakawa, T., 2005.
Functional activity related to risk anticipation during performance of
the Iowa Gambling Task. Neuroimage 24 (1), 253–259.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271268
Gaffan, D., 1979. Acquisition and forgetting in monkeys’ memory of
informational object-reward associations. Learning and Motivation 10,
419–444.
Garon, N., Moore, C., 2004. Complex decision-making in early childhood.
Brain and Cognition 55, 158–170.
Gomez-Beldarrain, M., Harries, C., Garcia-Monco, J.C., Ballus, E.,
Grafman, J., 2004. Patients with right frontal lesions are unable to
assess and use advice to make predictive judgements. Journal of
Cognitive Neuroscience 16 (1), 74–89.
Gonzalez, R., Vassileva, J., Bechara, A., Grbesic, S., Sworowski, L.,
Novak, R.M., 2005. The inﬂuence of executive functions, sensation
seeking, and HIV serostatus on the risky sexual practices of substancedependent
individuals. Journal of the International Neuropsychological
Society 11, 121–131.
Goudriaan, A.E., Oosterlaan, J., de Beurs, E., van den Brink, W., 2005.
Decision making in pathological gambling: a comparison between
pathological gamblers, alcohol dependents, persons with Tourette
syndrome, and normal controls. Cognitive Brain Research 23, 137–151.
Graham, F.K., 1979. Distinguishing among orienting, defense and startle
reﬂexes. In: Kimmel, H.D., Olst, E.H.V., Orlebeke, J.F. (Eds.), The
Orienting Reﬂex in Humans. Erlbaum, Hillsadale, NJ, pp. 137–167.
Harlow, J.M., 1868. Recovery from the passage of an iron bar through the
head. Publications of the Massachusetts Medical Society 2, 327–347
(also in booklet form published in Boston in 1869 by D.H. Clapp).
Harmsen, H., in press. The relationship between impaired decision making,
sensation seeking and readiness to change in cigarette smokers.
Addictive Behaviours (2005; Jun 24 [Epub ahead of print]).
Heims, H.C., Critchley, H.D., Dolan, R., Mathias, C.J., Cipolotti, L., 2004.
Social and motivational functioning is not critically dependent on
feedback of autonomic responses: neuropsychological evidence from
patients with pure autonomic failure. Neuropsychologia 42, 1979–1988.
Hinson, J.M., Jameson, T.L., Whitney, P., 2002. Somatic markers, working
memory, and decision making. Cognitive Behavioural and Affective
Neuroscience 2 (4), 341–353.
Hohmann, G.W., 1966. Some effects of spinal cord lesions on experienced
emotional feelings. Psychophysiology 3, 143–156.
Hooper, C.J., Luciana, M., Conklin, H.M., Yarger, R.S., 2004. Adolescents
performance on the Iowa Gambling Task: implications for the
development of decision making and ventromedial prefrontal cortex.
Developmental Psychology 40 (6), 1148–1158.
Hornak, J., O’Doherty, J., Bramham, J., Rolls, E.T., Morris, R.G., Bullock,
P.R., 2004. Reward related reversal learning after surgical excisions in
orbito-frontal or dorsolateral prefrontal cortex in humans. Journal of
Cognitive Neuroscience 16 (3), 463–478.
Izard, C.E., 1990. The substrates and functions of emotion feelings:
William James and current emotion theory. Personality and Social
Psychology Bulletin 16, 626–635.
Izquierdo, A., Suda, R.K., Murray, E.A., 2004. Bilateral orbital prefrontal
cortex lesions in rhesus monkeys disrupt choices guided by both reward
value and reward contingency. The Journal of Neuroscience 24 (34),
7540–7548.
James, W., 1884. What is an emotion? Mind 19, 188–205.
James, W., 1894. The physical basis of emotion. Psychological Review 1,
516–529.
Jameson, T.L., Hinson, J.M., Whitney, P., 2004. Components of working
memory and somatic markers in decision making. Psychonomic
Bulletin and Review 11 (3), 515–520.
Jollant, F., Bellivier, F., Leboyer, M., Astruc, B., Torres, S., Verdier, R.,
2005. Impaired decision making in suicide attempters. American
Journal of Psychiatry 162 (2), 304–310.
Kahneman, D., Tversky, A., 1972. Judgement under uncertainty: Heuristics
and biases. Psychological Review 80, 237–251.
Katkin, E.S., Wiens, S., O¨ hman, A., 2001. Nonconscious fear conditioning:
Visceral perception and the development of gut feelings. Psychological
Science 12 (5), 366–370.
Keillor, J.M., Barret, A.M., Crucian, G.P., Kortenkamp, S., Heilman, K.M.,
2002. Emotional experience and perception in the absence of facial
feedback. Journal of the International Neuropsychological Society 8,
130–135.
Kerr, A., Zelazo, P.D., 2004. Development of ‘hot’ executive function: the
children’s gambling task. Brain and Cognition 55 (1), 148–157.
Killcross, S., Coutureau, E., 2003. Coordination of actions and habits in the
medial prefrontal cortex of rats. Cerebral Cortex 13, 400–408.
Kleeberg, J., Bruggimann, M.A., Annoni, J., Melle, G., Bougousslavsky,
M.D., Schluep, M., 2004. Altered decision-making in multiple
sclerosis: a sign of impaired emotional reactivity. Annals of Neurology
56 (6), 787–795.
Lacey, J.L., 1967. Somatic response patterning and stress: some revisions
of activation theory. In: Appley, M.H., Trumbull, R. (Eds.),
Psychological Stress. Appleton Century Crofts, New York, pp. 14–37.
Laird, J., Bresler, C., 1990. William James and the mechanisms of
emotional experience. Personality and Social Psychology Bulletin 16,
630–651.
Lamar, M., Resnick, S.M., 2004. Aging and prefrontal functions:
dissociating orbitofrontal and dorsolateral abilities. Neurobiology of
Aging 25, 553–558.
Lambie, J.A., Marcel, A.J., 2002. Consciousness and the varieties of
emotion experience: A theoretical framework. Psychological Review
109 (2), 219–259.
Lang, P.J., 1994. The varieties of emotional experience: a meditation on
James-Lange theory. Psychological Review 101 (2) 211–221.
Lange, C., 1885. The mechanism of the emotions. In: Dunlap, E. (Ed.), The
Emotions. Williams & Wilkins, Baltimore, Maryland, pp. 33–92.
Lawrence, A.D., Sahakian, B.J., Rogers, R.D., Hodges, J.R., Robbins,
T.W., 1999. Discrimination, reversal, and shift learning in Huntington’s
disease: mechanisms of impaired response selection. Neuropsychologia
37, 1359–1374.
Lehto, J.E., Elorinne, E., 2003. Gambling as an executive function task.
Applied Neuropsychology 10 (4), 234–238.
Levine, B., Black, S.E., Cheung, G., Campbell, A., O’Toole, C., Schwartz,
M.L., 2005. Gambling task performance in traumatic brain injury.
Cognitive and Behavioural Neurology 18 (1), 45–54.
Lim, C.L., Rennie, C., Barry, R.J., Bahramali, H., Lazzaro, H., Manor, B.,
1997. Decomposing skin conductance into tonic and phasic components.
International Journal of Psychophysiology 25, 97–109.
Linton, J.C., Hirt, M., 1979. A comparison of predictions from peripheral
and central theories of emotion. British Journal of Medical Psychology
52, 11–15.
Loewenstein, G., Lerner, J.S., 2003. The role of affect in decision-making.
In: Davidson, R.J., Scherer, K.R., Goldsmith, H.H. (Eds.), Handbook of
Affective Sciences. Oxford University Press, Oxford, pp. 619–642.
Loewenstein, G.F., Weber, E.U., Hsee, C.K., Welch, N., 2001. Risk as
Feelings. Psychological Bulletin 127 (2), 267–286.
Lovibond, P.F., Shanks, D.R., 2002. The role of awareness in Pavlovian
conditioning: Empirical evidence and theoretical interpretations.
Journal of Experimental Psychology: Animal Behaviour Processes 28,
3–26.
MacLean, P., 1949. Psychosomatic disease and the visceral brain; recent
developments bearing on the Papez theory of emotion. Psychosomatic
Medicine 11 (6), 338–353.
MacLean, P., 1975. Sensory and perceptive factors in emotional functions
of the triune brain. In: Levi, L. (Ed.), Emotions—Their Parameters and
Measurement. Raven Press, New York, pp. 71–92.
MacMillan, M., 2000. An Odd Kind of Fame: Stories of Phineas Gage. MIT
Press, Cambridge, Massachusetts.
Maia, T.V., McClelland, J.L., 2004. A reexamination of the evidence for
the somatic marker hypothesis: what participants really know in the
Iowa gambling task. Proceedings of the National Academy for Science
USA 101 (45), 16075–16080.
Maia, T.V., McClelland, J.L., 2005. The somatic marker hypothesis: still
many questions but no answers. Trends in Cogntive Science 9 (4), 162–
164.
Mandler, G., 1990. William James and the construction of emotion.
Psychological Science 1, 179–180.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 269
Manes, F., Sahakian, B., Clark, L., Rogers, R., Antoun, N., Aitken, M.,
et al., 2002. Decision-making processes following damage to the
prefrontal cortex. Brain 125 (Pt 3), 624–639.
Marston, W.M., 1928. Emotions of Normal People. Harcourt, Brace, New
York.
Martin, C.O., Denburg, N.L., Tranel, D., Granner, M.A., Bechara, A.,
2004a. The effects of vagus nerve stimulation on decision-making.
Cortex 40 (4–5), 605–612.
Martin, E.M., Pitrak, D.L., Weddington, W., Rains, N.A., Nunnally, G.,
Nixon, H., 2004b. Cognitive impulsivity and HIV serostatus in
substance dependent males. Journal of the International Neuropsychological
Society 10, 931–938.
McGinn, C., 2003. Fear factor: a review of looking for Spinoza by Antonio
Damasio. New York Times Book Review, 2011.
Mellers, B.A., Schwartz, A., Ho, K., Ritov, I., 1997. Decision affect theory:
emotional reactions to the outcomes of risky options. Psychological
Science 8 (6), 423–429.
Mongini, F., Keller, R., Deregibus, A., Barbalonga, E., Mongini, M., 2005.
Frontal lobe dysfunction in patients with chronic migraine: a clinicalneuropsychological
study. Psychiatry Research 133, 101–106.
Montague, P.R., Hyman, S.E., Cohen, J.D., 2004. Computational roles for
dopamine in behavioural control. Nature 431, 760–767.
Monteresso, J., Ehrman, R., Napier, K.L., O’Brien, C.P., Childress, A.R.,
2001. Three decision-making tasks in cocaine-dependent patients: do
they measure the same construct? Addiction 96, 1825–1837.
Montoya, P., Schandry, R., 1994. Emotional experience and heartbeat
perception in patients with spinal cord injury and control subjects.
Journal of Psychophysiology 8, 289–296.
Mowrer, O.H., 1947. On the dual nature of learning—a reinterpretation of
‘conditioning’ and ‘problem solving’. Harvard Educational Review 17,
102–148.
Mowrer, O.H., 1960. Learning Theory and Behaviour. Wiley, New York.
Naccache, L., Dehaene, S., Cohen, L., Habert, M., Guichart-Gomez, E.,
Galanaud, D., et al., 2005. Effortless control: executive attention and
conscious feeling of mental effort are dissociable. Neuropsychologia
43, 1318–1328.
Nauta, W.J.H., 1971. The problem of the frontal lobe: a reinterpretation.
Journal of Psychiatric Research 8, 167–187.
Nielen, M.M.A., Veltman, D.J., de Jong, R., Mulder, G., den Boer, J.A.,
2002. Decision making performance in obsessive compulsive disorder.
Journal of Affective Disorders 69, 257–260.
North, N.T., O’Carroll, R.E., 2001. Decision making in patients with spinal
cord damage: afferent feedback and the somatic marker hypothesis.
Neuropsychologia 39 (5), 521–524.
O’Carroll, R.E., Papps, B.P., 2003. Decision making in humans: the effect
of manipulating the central noradrenergic system. Journal of
Neurology, Neurosurgery and Psychiatry 74, 376–378.
O¨ ngur, D., Price, J.L., 2000. The organisation of networks within the orbital
and medial prefrontal cortex of rats, monkeys and humans. Cerebral
Cortex 10, 206–209.
Overman, W.H., 2004. Sex differences in early childhood, adolescence, and
adulthood on cognitive tasks that rely on orbital prefrontal cortex. Brain
and Cognition 55 (1), 134–147.
Overman, W.H., Frassrand, K., Ansel, S., Trawalter, S., Bies, B., Redmond,
A., 2004. Performance on the IOWA card task by adolescents and
adults. Neuropsychologia 42 (13), 1838–1851.
Oya, H., Adolphs, R., Kawasaki, H., Bechara, A., Damasio, A.R., 2005.
Electrophysiological correlates of reward prediction error recorded in
the human prefrontal cortex. Proceedings of the National Academy of
Sciences USA 102 (23), 8351–8356.
Panksepp, J., 2003. Review article: “Looking for Spinoza: Joy, Sorrow, and
the Feeling Brain” by A. Damasio. Consciousness & Emotion 4(1),
111–134.
Power, M., Dalgleish, T., 1997. Cognition and Emotion: From Order to
Disorder. Psychology Press, Hove.
Pribram, K.H., 1970. Feelings as monitors. In: Arnold, M.B. (Ed.), Feelings
and Emotions: The Loyola Symposium. American Psychiatric Press,
New York, 41–53.
Prinz, J.J., 2004. Gut Reactions. Oxford University Press, Oxford.
Rahman, S., Sahakian, B.J., Hodges, J.R., Rogers, R.D., Robbins, T.W.,
1999. Speciﬁc cognitive deﬁcits in mild frontal variant frontotemporal
dementia. Brain 122 (Pt 8), 1469–1493.
Reavis, R., Overman, W.H., 2001. Adult sex differences on a decisionmaking
task previously shown to depend on the orbital prefrontal
cortex. Behavioural Neuroscience 115 (1), 196–206.
Reid, I.C., Morris, R.G.M., 1992. Smells are no surer: rapid improvements
in olfactory discrimination is not due to a learning set. Proceedings of
the Royal Society of London, Series B 247, 137–143.
Reisenzein, R., Meyer, W.U., Schutzwohl, A., 1995. James and the physical
basis of emotion: a comment on Ellsworth. Psychological Review 102,
757–761.
Rescorla, R.A., 1996. Response inhibition in extinction. Quarterly Journal
of Experimental Psychology 50, 238–252.
Rescorla, R., Solomon, R.C., 1967. Two-process learning theory:
relationships between Pavlovian conditioning and instrumental conditioning.
Psychological Review 74, 151–182.
Restle, F., 1958. Toward a quantitative description of learning set data.
Psychological Review 65, 77–91.
Ritter, L.M., Meador-Woodruff, J.H., Dalack, G.W., 2004. Neurocognitive
measures of prefrontal cortical dysfunction in schizophrenia. Schizophrenia
Research 68, 65–73.
Rogers, R., Everitt, B.J., Baldacchino, A., Blackshaw, A.J., Swainson, R.,
Wynne, K., Baker, N.B., Hunter, J., Carthy, T., Booker, E., London, M.,
Deakin, J.F., Sahakian, B.J., Robbins, T.W., 1999. Dissociable deﬁcits
in the decision-making cognition of chronic amphetamine abusers,
opiate abusers, patients with focal damage to prefrontal cortex, and
tryptophan-depleted normal volunteers: evidence for monoaminergic
mechanisms. Neuropsychopharmacology 20 (4), 322–339.
Rolls, E.T., 1996. The orbitofrontal cortex. Philosophical transactions of
the Royal Society of London (Series B) 351, 1433–1444.
Rolls, E.T., 1999. The Brain and Emotion. Oxford University Press,
Oxford.
Rolls, E.T., Hornak, J., Wade, D., McGrath, J., 1994. Emotion-related
learning in patients with social and emotional changes associated with
frontal lobe damage. Journal of Neurology Neurosurgery and
Psychiatry 57 (12), 1518–1524.
Rotheram-Fuller, E., Shoptaw, S., Berman, S.M., London, E.D., 2004.
Impaired performance in a test of decision-making by opiate-dependent
tobacco smokers. Drug and Alcohol Dependence 73 (1), 79–86.
Sanfey, A.G., Hastie, R., Colvin, M.K., Grafman, J., 2003. Phineas gauged:
decision-making and the human prefrontal cortex. Neuropsychologia 41
(9), 1218–1229.
Saper, C.B., 2002. The central autonomic nervous system: conscious
visceral perception and autonomic pattern generation. Annual Review
of Neuroscience 25, 433–469.
Saver, J.L., Damasio, A.R., 1991. Preserved access and processing of social
knowledge in a patient with acquired sociopathy due to ventromedial
frontal damage. Neuropsychologia 29, 1241–1249.
Schachter, S., Singer, J.E., 1962. Cognitive, social, and physiological
determinants of emotional state. Psychological Review 69, 379–399.
Schmitt, W.A., Brinkley, C.A., Newman, J.P., 1999. Testing Damasio’s
somatic marker hypothesis with psychopathic individuals: risk takers or
risk averse? Journal of Abnormal Psychology 108 (3), 538–543.
Shanks, D.R., 2005. Implicit learning. In: Lamberts, K., Goldstone, R.
(Eds.), Handbook of Cognition. Sage, London, pp. 202–220.
Shanks, D.R., Lovibond, P.F., 2002. Autonomic and eyeblink conditioning
are closely related to contingency awareness. Journal of Experimental
Psychology: Animal Behaviour Processes 28 (1), 38–42 (Reply to
Wiens and Ohman (2002) and Manns et al. (2002)).
Shanks, D.R., St. John, M.F., 1994. Charcteristics of dissociable learning
systems. Behavioural and Brain Sciences 17, 367–447.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271270
Shiv, B., Loewenstein, G., Bechara, A., 2005. The dark side of emotions in
decision-making: when individuals with decreased emotional reactions
make more advantageous decisions. Cognitive Brain Research 23, 85–92.
Shurman, B., Horan, W.P., Nuechterlein, K.H., 2005. Schizophrenia patients
demonstrate a distinctive pattern of decision-making impairment on the
Iowa Gambling Task. Schizophrenia Research 72, 215–224.
Solomon, R.L., Turner, L.H., 1962. Discriminative classical conditioning in
dogs paralysed by Curare can later control discriminative avoidance
responses in the normal state. Psychological Review 69 (3), 202–219.
Suzuki, A., Hirota, A., Takasawa, N., Shigemasu, K., 2003. Application of
the somatic marker hypothesis to individual differences in decision
making. Biological Psychology 65 (1), 81–88.
Teasdale, J.D., Barnard, P.J., 1993. Affect, Cognition and Change.
Lawrence Erlbaum Associates, Hove.
Tomb, I., Hauser, M., Deldin, P., Caramazza, A., 2002. Do somatic markers
mediate decisions on the gambling task? Nature Neuroscience 5 (11),
1103–1104 (author reply 1104).
Tomkins, S.S., 1962. Affect, Imagery, Consciousness, vol. 1. The Positive
Affects. NewYork: Springer.
Toplak, M.E., Jain, U., Tannock, R., 2005. Executive and motivational
processes in adolescents with attention-deﬁcit-hyperactivity disorder
(ADHD). Behaviour and Brain Functions 27 (1).
Tranel, D., 1994. In: Fowles, D.C., Sutker, P., Goodman, S.H. (Eds.).
Acquired sociopathy: The development of sociopathic behaviour
following focal brain damage progress in experimental psychology
and psychopathology Research, vol. 17, pp. 258–311.
Tranel, D., Bechara, A., Damasio, H., Damasio, A.R., 1996. Fear
conditioning after ventromedial frontal lobe damage in humans.
Society for Neuroscience Abstracts 22, 1108.
Tranel, D., Bechara, A., Damasio, A.R., 1999. Decision making and the
somatic marker hypothesis. In: Gazzaniga, M.S. (Ed.), The New
Cognitive Neurosciences. MIT Press, Cambridge, Massachusetts,
pp. 1047–1061.
Tranel, D., Bechara, A., Denburg, N.L., 2002. Asymmetric functional roles
of right and left ventromedial prefrontal cortices in social conduct,
decision-making, and emotional processing. Cortex 38 (4), 589–612.
Tucker, K.A., Potenza, M.N., Beauvals, J.E., Browndyke, J.N., Gottschalk,
P.G., Kosten, T.R., 2004. Perfusion abnormalities and decision making
in cocaine dependence. Biological Psychiatry 56, 527–530.
Turnbull O.H., Evans C.E., 2005. Preserved complex emotion-based
learning in amnesia. Neuropsychologia [Epub ahead of print].
Turnbull, O.H., Berry, H., Bowman, C.H., 2003. Direct versus indirect
emotional consequences on the Iowa Gambling Task. Brain and
Cognition 53 (2), 389–392.
Turnbull, O.H., Evans, C.Y., Bunce, A., Carzolio, B., O’Connor, J., 2005.
Emotion based learning and central executive resources: an investigation
of intuition and the Iowa Gambling Task. Brain and Cognition 57
(3), 244–247.
Turnbull, O.H., Evans, C.Y., Kemish, K., Park, S., Bowman, C.H., in press.
A novel ’set-shifting’ modiﬁcation of the Iowa Gambling Task: ﬂexible
emotion-based learning in schizophrenia. Neuropsychology.
van Honk, J., Hermans, E.J., Putman, P., Montagne, B., Schutter, D.J.,
2002. Defective somatic markers in sub-clinical psychopathy.
Neuroreport 13 (8), 1025–1027.
van Honk, J., Schutter, D.J., Hermans, E.J., Putman, P., Tuiten, A.,
Koppeschaar, H., 2004. Testosterone shifts the balance between
sensitivity for punishment and reward in healthy young women.
Psychoneuroendocrinology 29 (7), 937–943.
Verdejo, A., Aguilar de Arcos, F., Perez Garcia, M., 2004. Alterations in the
decision making processes linked to the ventromedial prefrontal cortex
in drug-abusing patients. Review of Neurology 38 (7), 601–606.
Whitlow, C.T., Liguori, A., Livengood, L.B., Hart, S.L., Mussat-Whitlow,
B.J., Lamborn, C.M., 2004. Long-term heavy marijuana users make
costly decisions on a gambling task. Drug and Alcohol Dependence 76,
107–111.
Whitney, K.A., Fastenau, P.S., Evans, J.D., Lysaker, P.H., 2004.
Comparative neuropsychological function in obsessive-compulsive
disorder and schizophrenia with and without obsessive-compulsive
symptoms. Schizophrenia Research 69, 75–83.
Wilder, K.E., Weinberger, D.R., Goldberg, T.E., 1998. Operant
conditioning and the orbitofrontal cortex in schizophrenic patients:
unexpected evidence for intact functioning. Schizophrenia Research
30, 169–174.
Wolpert, D.M., Ghahramani, Z., 2000. Computational principles of
movement neuroscience. Nature Neuroscience 3, 1212–1217.
Zuckerman, M.M., 1994. Behavioural Expressions and Biosocial Bases of
Sensation Seeking. Cambridge University Press, Cambridge.
B.D. Dunn et al. / Neuroscience and Biobehavioral Reviews 30 (2006) 239–271 271