Neural correlates of affective influence on choice.

Abstract

Making the right choice depends crucially on the accurate valuation of the available options in the light of current needs and goals of an individual. Thus, the valuation of identical options can vary considerably with motivational context. The present study investigated the neural structures underlying context dependent evaluation. We instructed participants to choose from food menu items based on different criteria: on their anticipated taste or on ease of preparation. The aim of the manipulation was to assess which neural sites were activated during choice guided by incentive value, and which during choice based on a value-irrelevant criterion. To assess the impact of increased motivation, affect-guided choice and cognition-guided choice was compared during the sated and hungry states. During affective choice, we identified increased activity in structures representing primarily valuation and taste (medial prefrontal cortex, insula). During cognitive choice, structures showing increased activity included those implicated in suppression and conflict monitoring (lateral orbitofrontal cortex, anterior cingulate). Hunger influenced choice-related activity in the ventrolateral prefrontal cortex. Our results show that choice is associated with the use of distinct neural structures for the pursuit of different goals.

Full text

Page 1

Neural correlates of affective influence on choice
Richard M. Piecha,*, Jade Lewisa, Caroline H. Parkinsona, Adrian M. Owenb, Angela C. Robertsc,
Paul E. Downinga, John A. Parkinsona
aSchool of Psychology, Bangor University, LL57 2AS, UK
bMRC Cognition and Brain Sciences Unit, Cambridge, UK
cDepartment of Physiology, Development and Neuroscience, University of Cambridge, UK
a r t i c l ei n f o
Article history:
Accepted 28 September 2009
Available online xxxx
Keywords:
Food choice
Decision making
Hunger
Motivation
fMRI
Affect
a b s t r a c t
Making the right choice depends crucially on the accurate valuation of the available options in the light
of current needs and goals of an individual. Thus, the valuation of identical options can vary consider-
ably with motivational context. The present study investigated the neural structures underlying context
dependent evaluation. We instructed participants to choose from food menu items based on different
criteria: on their anticipated taste or on ease of preparation. The aim of the manipulation was to assess
which neural sites were activated during choice guided by incentive value, and which during choice
based on a value-irrelevant criterion. To assess the impact of increased motivation, affect-guided
choice and cognition-guided choice was compared during the sated and hungry states. During affective
choice, we identified increased activity in structures representing primarily valuation and taste
(medial prefrontal cortex, insula). During cognitive choice, structures showing increased activity
included those implicated in suppression and conflict monitoring (lateral orbitofrontal cortex, anterior
cingulate). Hunger influenced choice-related activity in the ventrolateral prefrontal cortex. Our results
show that choice is associated with the use of distinct neural structures for the pursuit of different
goals.
Published by Elsevier Inc.
1. Introduction
Our everyday lives consist of a stream of choices. By the time we
arrive at work in the morning, we have already selected a shirt, a
route, and coffee with or without cream. Many choices are made
with some degree of automaticity, but others are preformed in a
consciously controlled manner, as they require a flexible adapta-
tion to the current task. The present study addresses such choices.
Making appropriate choice decisions depends crucially on the
accurate valuation of the available options in the light of attaining
the current goals of an individual. These goals can be manifold:
they can consist of achieving physiological homeostasis like reliev-
ing hunger feelings, or be more abstract, like beating a friend at
chess. Notably, the valuation of – sometimes identical – choice op-
tions depends on the current needs of the individual (Ferguson &
Bargh, 2004). So, for example, hunger might drive or even facilitate
meal preparation, but interfere with completing a chess victory.
Dual-process models of decision making distinguish between
two kinds of conscious choices: humans can arrive at decisions
by virtue of effortful, rule-based reasoning, or by intuitive, heuris-
tic processes (Kahneman, 2003; Sloman, 1996). In agreement with
such a model, Goel and Dolan (2003) dissociated rational, ‘cold’
choices from ‘hot’ ones, influenced by affective value. Their study
included choice options that were logically equivalent but loaded
to a different degree with affective value. They showed reciprocal
responses in the ventral medial prefrontal cortex (ventral mPFC)
and lateral PFC. Increased activity in the ventral mPFC was associ-
ated with ‘hot’ choices, which also decreased activity in the lateral
PFC. The reverse response pattern was associated with ‘cold’
choices. Similarly, Winston and colleagues (Winston, Strange,
O’Doherty, & Dolan, 2002) showed enhanced activity in the ventral
mPFC and the somatosensory cortex when emotional information
processing was the participants’ goal rather than when it happened
incidentally.
The representation of affective value itself in the human brain
has been studied utilizing brain imaging recordings of responses,
in most cases, positive stimuli including primary reinforcers like
food, smell and flavor (Anderson et al., 2003; Small & Prescott,
2005; Small, Zatorre, Dagher, Evans, & Jones-Gotman, 2001), but
also pleasant pictures (Garavan, Pendergrass, Ross, Stein, &
Risinger, 2001), and abstract stimuli associated with reward, like
descriptions of food (Arana et al., 2003). These and other studies
suggest a network of structures, including the amygdala, the basal
0278-2626/$ - see front matter Published by Elsevier Inc.
doi:10.1016/j.bandc.2009.09.012
* Corresponding author. Present address: Department of Psychology, Vanderbilt
University, PMB 407817, 2301 Vanderbilt Place, Nashville, TN 37240-7817, USA.
Fax: +1 615 343 8449.
E-mail address: r.piech@vanderbilt.edu (R.M. Piech).
Brain and Cognition xxx (2009) xxx–xxx
Contents lists available at ScienceDirect
Brain and Cognition
journal homepage: www.elsevier.com/locate/b&c
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 2

ganglia, the PFC, the insula and the anterior cingulate to represent
different aspects of value.
An important question provoked by these findings is of how dif-
ferent aspects of value representation guide choice processes, and
which neural structures are involved in their computation. Arana
et al. (2003) found increased activation in the medial orbitofrontal
cortex (mOFC) and the medial striatum in a condition which re-
quired choosing, in addition to evaluating food options. Paulus
and Frank (2003) showed that a beverage preference judgment en-
gaged the ventral mPFC, the anterior insula and sections of the
parietal and cingulate cortices more strongly than a visual discrim-
ination task. While these studies suggest structures which underlie
evaluation and choice processes, critically they do not directly con-
trast choice behavior when salient options were affectively laden
versus choices that were non-affective but otherwise equivalent
in nature. The current study was intended to address this issue.
In particular, it was designed to identify neural sites active during
choice behavior that is guided by an affective process – the assign-
ment of an incentive value to a stimulus, and during choice guided
by a cognitive process – the employment of an effortful, rule-based
computation.
For the present study, we chose to explicitly control the basis of
choice by instructing participants to use a specific criterion – affec-
tive or non-affective – to guide their choice while selecting from
restaurant meal options. The actual stimuli present were identical,
i.e. this study attempted to extract out structures involved specif-
ically in utilizing affective (or cognitive) information in an other-
wise equivalent choice situation. Participants completed the
following task while undergoing an fMRI recording. They were
asked to imagine being in a restaurant and studying the menu to
make a selection. During each experimental trial, they were pre-
sented three menu items. Their task was to read each item descrip-
tion, to imagine it, and to choose one of them. The critical
manipulation was implemented through an instruction prior to
each trial: in half the cases, participants were asked to choose
which item they would like to order and eat if they were actually
in a restaurant (‘eat’ condition). In the other half of the trials, par-
ticipants were instructed to choose which item they consider the
easiest to cook (‘cook’ condition). Thus, both conditions required a
consideration of the properties of the dishes and a selection pro-
cess. In the first condition, however, choices were guided by affec-
tive properties, in the second by cognitive-procedural properties of
the stimuli.
The ‘eat’ condition in our study represents processes of affect-
guided choice. The investigation of affective influences on choice,
especially in the context of food selection, needs to consider the
relationship between value and motivation. In particular, the affec-
tive value of an option may be susceptible to changes driven by an
altered motivational state. So whilst, in both the ‘eat’ and ‘cook’
conditions, the choice required using appropriate motivational
and cognitive systems to make the response, in the ‘eat’ condition
the choice was based on the projected rewarding quality of the
menu item value, whereas in the ‘cook’ condition the choice had
to be based on cognitive knowledge of procedures required for
the preparation of a dish. Several studies have shown that altering
the motivational state of participants by changing their hunger le-
vel has an impact on decision making and neural responses associ-
ated with relevant stimuli (Hinton et al., 2004; Killgore &
Yurgelun-Todd, 2006; Kringelbach, O’Doherty, Rolls, & Andrews,
2003; Small et al., 2001). A further goal of our study was to inves-
tigate whether neural activity during choice guided by affective
information was susceptible to manipulation of motivation. There-
fore, we conducted the experiment in two otherwise identical ses-
sions, one during which participants were sated, and one in which
they were hungry. We hypothesized that the increased motivation
to eat would have an impact on the processes involved in choosing
food based on its expected reward properties, but not based on the
procedural properties of its preparation.
In summary, participants in the current study made selections
from restaurant menu items presented on a screen, once while
they were hungry and once while sated. They were instructed to
make their selection based on either the desirability of the dishes
or the complexity of their preparation. Thus, two experimental fac-
tors were manipulated, the motivational state of the participants
and the affective value of the information to be processed. We pre-
dicted that motivational state should influence processes involved
in the ‘eat’ food choice, but would not affect cooking complexity
choices.
2. Method
2.1. Design
Eight volunteers (three female; group average age of 27.9,
SD = 4.1) participated in three experimental sessions. These were
the same participants as in a parallel study published elsewhere
(Piech et al., 2009). They underwent fMRI recording during two
one-hour sessions, one in the hungry, and one in the sated condi-
tion. The two recordings happened ca. 1 week apart, and the se-
quence of conditions was balanced for all participants. For the
hungry condition, participants were instructed to not eat for 6 h
prior to the experiment. In an initial session, participants com-
pleted an extended questionnaire indicating their food preferences.
The information from it was then used to design individual menu
choice options for the main experiment, which would include a
variety of items, excluding items evoking negative responses like
disgust. Each session consisted of three blocks of approximately
12 min length. Immediately after the recording, participants re-
ported their hunger level on a visual analogue scale, which was
scored between 0 (not hungry at all) and 100 (extremely hungry).
They were debriefed after the second session and paid for their
time. The study was approved by Research Ethics Committee of
the University of Wales, Bangor.
2.2. Task
Participants were asked to imagine being in a restaurant for an
evening meal. While in the scanner, they were presented with
three dish descriptions from restaurant menus in each trial, with
all three presented on one screen (on top, middle, and bottom of
screen; no actual food was presented). The text was back-projected
on a screen and viewed through a mirror. Participants were in-
structed to study the menu items to make a selection, using a re-
sponse box held in their right hand. Their task depended on an
instruction which appeared on the screen prior to each trial. For
the ‘eat’ condition, participants’ task was to read each item descrip-
tion, and to choose the one they would select in a restaurant. For
the ‘cook’ condition, the selection was to be based on the complex-
ity of preparation, i.e. participants were asked to choose the one
which they thought would be the easiest to cook. The order of trials
was randomized. Participants indicated their choice by pressing
the top, middle or bottom key on a keypad. An example of three
choice options would be: (1) ‘‘Tender roast lamb served with roast
potatoes, cabbage, sweetcorn, and mint sauce.” (2) ‘‘Succulent
chunks of lamb in a thick creamy gravy with chestnut mushrooms,
onions and leeks oven baked to perfection.” (3) ‘‘Bite size chicken
pieces marinated in sherry garlic, soy sauce and lemon juice.
Served with assorted vegetables and rice.” Each session consisted
of 3 scans of 11 min, and 24 menu ratings per scan. Each menu
selection appeared on the screen for 22 s after a trial instruction
(‘eat’ or ‘cook’) of 1 to 2 s. The fixation interval between
2
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 3

presentations was randomly jittered between 1 and 3 s. Only the
duration between item onset and response (i.e. not the entire
22 s) was modeled for the fMRI analysis.
2.3. fMRI data acquisition and analysis
A 1.5 T Philips MRI scanner was used to acquire 22 T2*weighted
slices per volume (5 mm slices, resulting in 3.75 mm ? 3.75 mm ?
5 mm voxel size), with a repetition time of 2.2 s. The slices were
tilted (anterior up) by 30? from the ACPC axial plane to reduce sus-
ceptibility artifacts. Thus the recorded volume included the entire
brain volume excluding only ventral parts of the cerebellum. The
first five volumes of each scan were discarded to avoid differences
in T1-saturation. Preprocessing and statistical analysis were per-
formed using BrainVoyager 2000 and BrainVoyager QX (Brain Inno-
vation, The Netherlands). The functional images were slice-time
acquisition corrected, subject motion corrected, spatially normal-
ized to Talairach space (Talairach & Tournoux, 1988), and spatially
smoothedwitha4 mmfullwidthathalfmaximumGaussiankernel.
A correction for temporal autocorrelation and a temporal high pass
filter of 0.01 Hz were applied. The two fMRI recordings took place
approximately one week apart. Anatomical scans were acquired
during both sessions to ensure accuracy of the intersession align-
ment of functional data.
The events for the fMRI signal were modeled as follows. For
each event, duration always corresponded to the period from onset
of the menu item to participants’ individual trial choice response
time (roughly 12 s, see results; upper limit 20 s). Since participants
needed this relatively long period of time, and since they could be
assumed to be making the actual choice towards the end of that
period, we included an additional analysis method after initial
inspection of the results. The method employed a stick-function
analysis in which we only used the 10 ms prior to the response
in each trial as the period of interest. The purpose of this analysis
was to increase power to detect interactions between choice con-
dition and motivational state. For the event-related fMRI analysis,
eat and cook trials were modeled as separate events for both the
sated and hungry sessions. This allowed the data to be analyzed
as two factors with two conditions each: hunger state (hungry,
sated) and task (eat, cook). Both sessions were entered in one
analysis using dummy variables which remained empty in the
non-relevant condition (e.g. the eat-hungry variable had no events
for the sated session). The general linear model used for the fMRI
data analysis thus included 11 regressors. These were two trial
types (eat and cook), six motion regressors, and one artefact
regressor. The eat and cook trials were entered twice, for the sated
and the hungry session. The motion predictors included transitions
along the three axes and rotations around them. The artefact
regressor was entered at points where gross head movement was
detected during visual inspection. Columns of the stimulus design
matrix were convolved with a canonical hemodynamic response
function. In order to identify activity peaks corresponding to the
processes of interest, an unconstrained whole-brain random-ef-
fects analysis was conducted. Areas of functional activity were de-
fined as clusters of 10 or more contiguous voxels which exceeded
an uncorrected p-value of .0005. This is an arbitrary, while rela-
tively stringent criterion. In the post hoc, stick-function analysis,
we used a threshold of .005 to explore the activations representing
task by state interactions. Our statistical analysis package allowed
one method of accounting for multiple comparisons, the Bonfer-
roni correction, which has been frequently described as overly con-
servative (Friston, Frith, Liddle, & Frackowiak, 1991). None of the
whole-brain calculated activations reported here survive such a
correction.
The number of participants is small due to the considerable
complexity of the design. While many fMRI studies use larger
sample sizes, random-effects analyses with as few as six subjects
are permitted (Holmes & Friston, 1998). The given sample size
may produce only low statistical power and render null-effects
unreliable. We therefore focus the interpretation of our results
on positive effects.
3. Results
3.1. Behavioral analysis
Confirming the experimental manipulation, participants indi-
cated higher levels of hunger after the scan in the hungry condition
than after the scan in the sated condition (Hungry: 81, SE = 3.1; Sa-
ted: 22, SE = 3.0; t(5) = 9.63, p < .0005). The menu selection was
paced by the participants with the average response times (and
standard errors) being 12.7 s (1.2), 12.4 s (1.3), 12.1 s (1.4), and
12.0 s (1.5) for the ‘hungry & eat’, ‘hungry & cook’, ‘sated & eat’,
and ‘sated & cook’ conditions, respectively. A repeated measures
ANOVA indicated no response time differences between the condi-
tions (F-values < 1.5).
3.2. Event-related fMRI analysis
3.2.1. Affect-guided and cognition-guided choice behavior
We expected that contrasting the data from the ‘‘cook‘‘ task
from the data from the ‘‘eat” task would emphasize the impact of
appetitive anticipatory processes on choice, while the opposite
subtraction would emphasize utilization of cognitive information.
To this end, we conducted a whole brain contrast comparing
activity elicited by the menu stimuli under the two task conditions,
eat and cook, for both recording sessions. Increased activity was
found in the ‘‘eat” task in the medial prefrontal cortex (mPFC, BA
10, x = ?6, y = 47, z = 22, t(7) = 7.60, Fig. 1A), in the central part of
the insula (x = ?36, y = 8, z = 10, t(7) = 6.33), Fig. 1B), and in an area
of the anterior temporal lobe, in the middle temporal sulcus (BA
21, x = ?57, y = ?7, z = ?8, t(7) = 6.69, Fig. 1C). To confirm that
the specified clusters responded more strongly to the conditions
of interest, we computed averaged time courses of the BOLD signal
for these sites (Fig. 1A–C).
The opposite contrast revealed more activity in the lateral OFC
(BA 47, x = ?42, y = 32, z = ?9, t(7) = 9.13, Fig. 2A), in the anterior
cingulate cortex (ACC, BA 24, x = 0, y = 26, z = 7, t(7) = 6.30,
Fig. 2B), and in an area of the posterior temporal lobe (Middle tem-
poral lobe, BA 20, x = ?63, y = ?34, z = ?11, t(7) = 8.48, Fig. 2C). The
averaged time courses for those clusters are shown in Fig. 2A–C.
To investigate the impact of motivational state on task related
activity, we conducted an interaction analysis, with a contrast
which would reveal areas activated more strongly for the eat con-
dition, but only in the hungry state. That was achieved through a
contrast which assigned positive weighs to the eat condition and
negative weights to the cook condition for the hungry session, as
well as the reverse weights for the sated session. A whole brain
analysis with these contrasts followed by inspection of the activa-
tion patterns of potentially involved areas showed that no such
area could be identified, even at a liberal threshold of p < .01, with
our original analysis. Our whole-brain exploratory stick-function
analysis yielded a single significant activation, in the ventrolateral
PFC, at a statistical threshold of p < .005 with a cluster size of more
than 10 contiguous voxels (BA 47, x = 44, y = 44, z = ?2, t(7) = 5.1,
Fig. 3).
Overall, we found that affect-guided choice behavior activates a
different set of areas compared to equivalent choice behavior when
it is solely based on procedural information not related to value. An
area of the ventrolateral PFC showed activation consistent with an
interaction function between task and hunger state.
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
3
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 4

4. Discussion
The affective choice condition in our study activated the mPFC,
the insula, and an anterior site in the lateral temporal lobe (middle
temporal gyrus, MTG), while the cognitive choice condition acti-
vated the lateral OFC, the anterior cingulate, and a site in the pos-
terior part of the lateral temporal lobe (middle temporal sulcus,
MTS). We also designed our experiment to be conducted in two
sessions, while the participants were either sated or hungry in or-
der to allow us to identify regions that show an interaction re-
sponse pattern. We showed such an activation pattern in an area
of the ventrolateral prefrontal cortex.
4.1. Affect-guided choice task
The function of the medial prefrontal cortex (mPFC) is the sub-
ject of much debate (Christoff & Gabrieli, 2000; Gilbert et al., 2006;
Ramnani & Owen, 2004). Amodio and Frith (2006) describe the sec-
tion of mPFC activated in our study as anterior rostral medial fron-
tal cortex and suggest it plays a role in three categories of
processes, involving self-knowledge, perception of other persons
and mentalizing. The last of these is also referred to as theory of
mind (ToM) – thinking about what others think (Mitchell, Heather-
ton, & Macrae, 2002; Völlm et al., 2006). Likewise, Gilbert et al.
(2006) conducted a meta-analysis of over a hundred functional
imaging studies which activated the mPFC (specifically BA 10)
and concluded that this site is often associated with tasks that in-
volve mentalizing and attending to one’s own emotions, as op-
posed to more lateral and more rostral sites, involving memory
processes and multi-tasking, respectively.
Montague, King-Casas, and Cohen (2006) review of literature on
reward and valuation considers the mPFC (together with the ven-
tral striatum and the OFC) a component of the ventral valuation
network, and suggests that mPFC response often reflects the value
of rewards, including a prospective valuation of predicted reward.
In the current study, mPFC activation occurred during affectively
guided choice, when representation of the anticipated reward va-
lue is necessary to make the correct decision. This is consistent
with a role of that area in predictive valuation of stimuli (Monta-
gue et al., 2006). Within decision making literature, evaluating re-
wards is often associated with ventral sites of the mPFC,
sometimes referred to as medial orbitofrontal cortex (Knutson,
Fig. 1. Sites activated by affect-guided choice. Averaged time courses of the response to conditions ‘eat’ and ‘cook’ (left side of panels, arrows indicate stimulus onset, bars
show the standard error), and respective activated clusters. (A) Medial prefrontal cortex, (B) the insula and (C) middle temporal gyrus. Activation maps are shown overlaid
over averaged anatomical scans for all subjects. Activations were defined as clusters of 10 or more contiguous voxels which exceeded an uncorrected p-value of .0005. The
above figure uses a more lenient threshold of p < .005 for demonstration purposes. The shown cluster extent threshold is 10 voxels. Averaged time courses illustrate the
activity of the selected clusters, they are not an independent analysis. The arrows at time point 0 indicate the stimulus onset, arrows marked ’RT’ indicate the mean response
time.
4
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 5

Fig. 2. Sites activated by cognition-guided choice. Averaged time courses of the response to conditions ‘cook’ and ‘eat’ (left side of panels, arrows indicate stimulus onset, bars
show the standard error), and respective activated clusters. (A) Lateral orbitofrontal cortex, (B) anterior cingulate cortex and (C) middle temporal sulcus. Activation maps are
shown overlaid over averaged anatomical scans for all subjects. Activations were defined as clusters of 10 or more contiguous voxels which exceeded an uncorrected p-value
of .0005. The above figure uses a more lenient at a threshold of p < .005 for demonstration purposes. The shown cluster extent threshold is 10 voxels. Averaged time courses
illustrate the activity of the selected clusters, they are not an independent analysis. The arrows at time point 0 indicate the stimulus onset, arrows marked ’RT’ indicate the
mean response time.
Fig. 3. Site showing an interaction between the choice task and hunger state. The activated cluster is in the ventrolateral PFC, likely within BA 47, the maximum peak at
x = 44, y = 44, z = ?2. Activation maps are shown overlaid over averaged anatomical scans for all subjects. Activation was defined as a cluster of 10 or more contiguous voxels
which exceeded an uncorrected p-value of .005. The above figure uses a more lenient threshold of p < .019 for demonstration purposes.
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
5
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 6

Fong, Adams, Varner, & Hommer, 2001; Plassmann, O’Doherty,
Shiv, & Rangel, 2008). Such activations are typically localized when
stimuli of great value are contrasted with ones of lesser value, or as
correlations with stimulus ratings. The mPFC activation reported
here is localized at a more dorsal site (Fig. 1A). We believe that is
because our contrast does not reflect sole value representation,
as in each trial three stimuli of different values are present. Rather,
our tasks portrays an affective evaluation process, possibly based
on attending to one’s own emotions and motivational needs
(Christoff & Gabrieli, 2000; Gilbert et al., 2006; Steele & Lawrie,
2004). Another site which is activated in our study during affec-
tively guided choice is the anterior temporal lobe. Activations of
this site have also previously been implicated in evaluating events
related to the self or others (Mitchell et al., 2002; Völlm et al.,
2006), and may reflect judgments about the personal value of a
prospective meal, including attempts to infer the consequences
and hence mental and physiological state of the self following a
particular meal.
The insular cortex subserves a variety of functions related to
interoception (Craig, 2002; Critchley, Wiens, Rotshtein, Oehman,
& Dolan, 2004) and the perception of taste and its reward qualities
(de Araujo & Rolls, 2004). Chaudhry, Parkinson, Hinton, Owen, and
Roberts (2009) have recently shown that affective evaluation of
more abstract stimuli than ours (holidays), activates a similar area
of the insula (albeit in the right hemisphere). Most notably, it has
been shown to represent interactions of reward value with motiva-
tional states (Hinton et al., 2004; Small et al., 2001). The activation
of the insula found during affect-guided choice in our study is
likely to represent expected sensory gustatory qualities of the pre-
sented items, upon which the decision regarding which item
would be most preferred to eat must be based.
4.2. Cognition-guided choice task
In the current task, the cognitive contrast yielded activity in the
lateral OFC. The OFC is broadly thought to represent reward value
(Rolls, 2000), as it has been shown to respond to a number of po-
sitive stimuli, among them tastes and odors. Besides that, the
OFC has also been shown to represent negative stimuli, and reward
value change (Kringelbach, O’Doherty, Rolls, & Andrews, 2003;
Small et al., 2001). This combination of observations has led to
claims that the OFC represents both primary reinforcer value and
processes associated with incorporating changes in value brought
about by fluctuating motivation, and hence in the control and cor-
rection of behavior driven by reward (Rolls, 2000).
The area in the lateral OFC activated in the current study maps
well onto previous findings: Small et al. (2001) observed in their
study that as the value of a food (chocolate) decreased with con-
sumption, activity in the lateral OFC increased, suggesting that it
represented aversive value. O’Doherty and colleagues (O’Doherty,
Kringelbach, Rolls, Hornak, & Andrews, 2001) came to a similar
conclusion, as their value reversal learning task produced lateral
OFC responses to punishing stimuli. Using a logical reasoning task,
Goel and Dolan (2003) dissociated lateral from medial PFC on the
basis of emotional salience: purely logical judgments engaged lat-
eral PFC, whilst those involving emotional content activated med-
ial PFC. Elliott, Dolan, and Frith (2000) reviewed work on OFC
function and concluded that the results are consistent with the
suppression of a previously rewarded response in the lateral OFC.
In the current study, lateral OFC was activated during cognitively
guided choice, when it is appropriate to ignore or suppress the re-
ward value of the food items and the choice decision needs to be
based on a different criterion. A parsimonious explanation of the
current findings therefore is that lateral OFC supports cognitively
driven judgments and may operate antagonistically with affect-
based medial circuitry.
The ACC has been implicated in an array of functions, including
processing of sensory, motor, emotional and cognitive information.
A common denominator in ACC function appears to be the moni-
toring of conflict (Botvinick, Cohen, & Carter, 2004). In an attempt
to specify the function of the ACC and its subdivisions, Bush and
colleagues (Bush, Luu, & Posner, 2000) reviewed findings from neu-
roimaging, lesion and physiological studies. They suggested a sep-
aration of the ACC in dorsal and ventral areas, corresponding to
Brodmann areas 24 and 32, with the dorsal division frequently ac-
tive in cognitive tasks, and the (rostral-) ventral areas involved in
affective tasks. The activation peak we identified in the ACC for
cognition-guided choice falls into the rostral-ventral division of
BA 32. In agreement with the cited work, it might represent an
affective conflict monitoring process. This is a plausible explana-
tion if one assumes that the affective representation of the choice
options potentially occurs automatically, and that it is a process
which during the cognitively guided choice might create a conflict.
This interpretation of the ACC activity would converge with the
activation in the lateral OFC found in the same condition.
In the light of the dual-process model of decision making
(Kahneman, 2003; Sloman, 1996), our ‘eat’ condition could be re-
ferred to as an intuitive choice, based on one’s liking of a dish.
The ‘cook’ condition however, required more effortful reasoning
to allow choosing between the complexity of different cooking pro-
cedures, and thus constitutes an effortful, rule-based choice. With-
in the PFC, we found a medial activation for the intuitive choice,
and a lateral activation for the rule-based choice. This is consistent
with the results of other studies, which also found the medial PFC
to be activated by processes related to heuristic processes incorpo-
rating emotional evaluation (Goel & Dolan, 2003; Winston,
Strange, O’Doherty, & Dolan, 2002) and it shows that employment
of the rule-based and intuitive systems can be evoked by direct
instructions.
4.3. Choice task and hunger state interaction
Our stick-function analysis revealed a site in the ventrolateral
PFC, likely within BA 47, which showed an interaction between
the task and state factors. Inspection of the beta-values suggested
that activation was increased in the eat condition when partici-
pants were hungry, while the reverse pattern appeared for the sa-
ted state. The ventrolateral PFC has been previously reported to be
active during affectiveevaluation
Johnson, Gatenby, Gore, & Banaji, 2003; Cunningham, Raye, & John-
son, 2004; Jacobsen, Schubotz, Hoefel, & Cramon, 2006; Maddock,
Garrett, & Buonocore, 2003), but to our knowledge no influences
of motivational state on these have been reported to this date. As
such, we report a novel and interesting neural locus reflecting
the relationship between motivation and evaluative choice, as pre-
dicted by theories of goal-directed action Ferguson (de Wit &
Dickinson, 2009; Ferguson & Bargh, 2004).
of stimuli (Cunningham,
5. Conclusion
In the current study, we instructed participants to choose food
menu items based on their anticipated taste or ease of preparation,
aiming to elicit affect-guided and cognition-guided choice. During
affective choice, structures representing primarily taste and valua-
tion (mPFC, the insula) as well as an area of anterior temporal lobe
were active. During cognitive choice, the structures used involved
ones implicated in suppression and conflict monitoring. These re-
sults shed important insight into neural mechanisms of prospective
choice behavior and support dual-process models of decision-mak-
ingbasedon the affectivecontentof the material.The relative activ-
ity of neural circuitry engaged in processing text-based food
6
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012

Page 7

descriptions differed depending on the nature of the choice to be
madeandsuggeststhatdecisionsrequiringinputfrommotivational
systemsrecruitsmedialprefrontalstructureswhilstthoserequiring
thesuppressionofthisinformationinfavorofrule-basedprocessing
activates lateral prefrontal structures and the ACC. Finally, we indi-
cate a role of the ventrolateral prefrontal cortex in the motivational
modulation of choice processes.
Acknowledgments
We thank Tony Bedson and radiography staff at the Magnetic
Resonance Unit at Ysbyty Gwynedd in Bangor, an anonymous re-
viewer for analysis suggestions, and Maureen McHugo for assis-
tance with the analysis.
References
Amodio, D. M., & Frith, C. D. (2006). Meeting of minds: The medial frontal cortex and
social cognition. Nature Reviews Neuroscience, 7(4), 268–277.
Anderson, A. K., Christoff, K., Stappen, I., Panitz, D., Ghahremani, D. G., Glover, G.,
et al. (2003). Dissociated neural representations of intensity and valence in
human olfaction. Nature Neuroscience, 6(2), 196–202.
Arana, F. S., Parkinson, J. A., Hinton, E., Holland, A. J., Owen, A. M., & Roberts, A. C.
(2003). Dissociable contributions of the human amygdala and orbitofrontal
cortex to incentive motivation and goal selection. Journal of Neuroscience,
23(29), 9632.
Botvinick, M. M., Cohen, J. D., & Carter, C. S. (2004). Conflict monitoring and anterior
cingulate cortex: An update. Trends in Cognitive Sciences, 8(12), 539–546.
Bush, G., Luu, P., & Posner, M. (2000). Cognitive and emotional influences in anterior
cingulate cortex. Trends in Cognitive Sciences, 4, 215–222.
Chaudhry, A., Parkinson, J., Hinton, E., Owen, A., & Roberts, A. (2009). Preference
judgements involve a network of structures within frontal, cingulate and insula
cortices. European Journal of Neuroscience, 29(5), 1047.
Christoff, K., & Gabrieli, J. D. E. (2000). The frontopolar cortex and human cognition:
Evidence for a rostrocaudal hierarchical organization within the human
prefrontal cortex. Psychobiology, 28(2), 168–186.
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., Wiens, S., Rotshtein, P., Oehman, A., & Dolan, R. J. (2004). Neural
systems supporting interoceptive
189–195.
Cunningham, W., Johnson, M., Gatenby, J., Gore, J., & Banaji, M. (2003). Neural
components of social evaluation. Journal of Personality and Social Psychology,
85(4), 639–649.
Cunningham, W., Raye, C. L., & Johnson, M. K. (2004). Implicit and explicit
evaluation: fMRI correlates of valence, emotional intensity, and control in the
processing of attitudes. Journal of Cognitive Neuroscience, 16(10), 1717–1729.
de Araujo, I. E., & Rolls, E. T. (2004). Representation in the human brain of food
texture and oral fat. Journal of Neuroscience, 24(12), 3086.
deWit,S.,&Dickinson,A.(2009).Associativetheoriesofgoal-directedbehaviour:Acase
for animalñhuman translational models. Psychological Research, 73(4), 463–476.
Elliott, R., Dolan, R. J., & Frith, C. D. (2000). Dissociable functions in the medial and
lateral orbitofrontal cortex: Evidence from human neuroimaging studies.
Cerebral Cortex, 10(3), 308–317.
Ferguson, M., & Bargh, J. (2004). Liking is for doing: The effects of goal pursuit on
automatic evaluation. Journal of Personality and Social Psychology, 87, 557–572.
Friston, K., Frith, C., Liddle, P., & Frackowiak, R. (1991). Comparing functional (PET)
images: The assessment of significant change. 1. Journal of Cerebral Blood Flow
and Metabolism, 11, 10.
Garavan, H., Pendergrass, J. C., Ross, T. J., Stein, E. A., & Risinger, R. C. (2001).
Amygdala response to both positively and negatively valenced stimuli.
NeuroReport, 12(12), 2779–2783.
awareness.
NatureNeuroscience,7(2),
Gilbert, S. J., Spengler, S., Simons, J. S., Steele, J. D., Lawrie, S. M., Frith, C.
D., et al. (2006). Functional specialization within rostral prefrontal cortex
(Area10):Ameta-analysis.
Journal
932–948.
Goel, V., & Dolan, R. J. (2003). Reciprocal neural response within lateral and ventral
medial prefrontal cortex during hot and cold reasoning. NeuroImage, 20(4),
2314–2321.
Hinton, E. C., Parkinson, J. A., Holland, A. J., Arana, F. S., Roberts, A. C., & Owen, A. M.
(2004). Neural contributions to the motivational control of appetite in humans.
European Journal of Neuroscience, 20(5), 1411–1418.
Holmes, A., & Friston, K. (1998). Generalizability, random effects, and population
inference. NeuroImage, 7, 1.
Jacobsen, T., Schubotz, R., Hoefel, L., & Cramon, D. (2006). Brain correlates of
aesthetic judgment of beauty. NeuroImage, 29(1), 276–285.
Kahneman, D. (2003). A perspective on judgment and choice: Mapping bounded
rationality. American Psychologist, 58(9), 697–720.
Killgore, W. D., & Yurgelun-Todd, D. A. (2006). Affect modulates appetite-related
brain activity to images of food. International Journal of Eating Disorders, 39(5),
357–363.
Knutson,B.,Fong, G., Adams, C.,
Dissociationofreward anticipation
fMRI. NeuroReport, 12, 5.
Kringelbach, M. L., O’Doherty, J., Rolls, E. T., & Andrews, C. (2003). Activation of the
human orbitofrontal cortex to a liquid food stimulus is correlated with its
subjective pleasantness. Cerebral Cortex, 13(10), 1064–1071.
Maddock, R., Garrett, A., & Buonocore, M. (2003). Posterior cingulate cortex
activation by emotional words: fMRI evidence from a valence decision task.
Human Brain Mapping, 18(1), 30–41.
Mitchell, J. P., Heatherton, T. F., & Macrae, C. N. (2002). Distinct neural systems
subserve person and object knowledge. Proceedings of the National Academy of
Sciences, 99(23), 15238.
Montague, P. R., King-Casas, B., & Cohen, J. D. (2006). Imaging valuation models in
human choice. Annual Reviews Neuroscience, 29, 417–448.
O’Doherty, J., Kringelbach, M. L., Rolls, E. T., Hornak, J., & Andrews, C. (2001).
Abstract reward and punishment representations in the human orbitofrontal
cortex. Nature Neuroscience, 4, 95–102.
Paulus, M. P., & Frank, L. R. (2003). Ventromedial prefrontal cortex activation is
critical for preference judgments. NeuroReport, 14(10), 1311–1315.
Piech, R., Lewis, J., Parkinson, C., Owen, A., Roberts, A., Downing, P., et al. (2009).
Neural correlates of appetite and hunger-related evaluative judgments. PLoS
ONE, 4(8), e6581. doi:10.1371/journal.pone.0006581.
Plassmann, H., O’Doherty, J., Shiv, B., & Rangel, A. (2008). Marketing actions can
modulate neural representations of experienced pleasantness. Proceedings of the
National Academy of Sciences, 105(3), 5.
Ramnani, N., & Owen, A. M. (2004). Anterior prefrontal cortex: Insights into function
from anatomy and neuroimaging. Nature Reviews Neuroscience, 5(3), 184–194.
Rolls, E. T. (2000). The orbitofrontal cortex and reward. Cerebral Cortex, 10(3),
284–294.
Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological
Bulletin, 119(1), 3–22.
Small, D. M., & Prescott, J. (2005). Odor/taste integration and the perception of
flavor. Experimental Brain Research, 166(3), 345–357.
Small, D. M., Zatorre, R. J., Dagher, A., Evans, A. C., & Jones-Gotman, M. (2001).
Changes in brain activity related to eating chocolate: From pleasure to aversion.
Brain, 124(9), 1720.
Steele, J. D., & Lawrie, S. M. (2004). Segregation of cognitive and emotional function
in the prefrontal cortex: A stereotactic meta-analysis. NeuroImage, 21(3),
868–875.
Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic axis of the human brain.
New York: Thieme.
Völlm, B. A., Taylor, A. N. W., Richardson, P., Corcoran, R., Stirling, J., McKie, S., et al.
(2006). Neuronal correlates of theory of mind and empathy: A functional
magnetic resonance imaging study in a nonverbal task. NeuroImage, 29(1),
90–98.
Winston, J. S., Strange, B. A., O’Doherty, J., & Dolan, R. J. (2002). Automatic and
intentional brain responses during evaluation of trustworthiness of faces.
Nature Neuroscience, 5(3), 277–283.
of Cognitive Neuroscience,18(6),
Varner,
and
J.,
outcome
& Hommer,
with
D.(2001).
event-related
R.M. Piech et al./Brain and Cognition xxx (2009) xxx–xxx
7
ARTICLE IN PRESS
Please cite this article in press as: Piech, R. M., et al. Neural correlates of affective influence on choice. Brain and Cognition (2009), doi:10.1016/
j.bandc.2009.09.012