Human cortical representation of oral temperature
Steve Guesta, Fabian Grabenhorstb, Greg Essickc, Yasheng Chend, Mike Younga,
Francis McGlonee, Ivan de Araujof, Edmund T. Rollsb,⁎
aCenter for Neurosensory Disorders, School of Dentistry, University of North Carolina, Chapel Hill, NC, United States
bUniversity of Oxford, Department of Experimental Psychology, Oxford OX1 3UD, UK
cDepartment of Prosthodontics and Center for Neurosensory Disorders, University of North Carolina, Chapel Hill, NC, United States
dPostdoctoral Researcher, Department of Radiology, University of North Carolina, Chapel Hill, NC, United States
eCognitive Neuroscience, Department of Neurological Sciences, School of Medicine, Liverpool University, UK
fDepartment of Neuroscience, Duke University, Durham, NC, United States
Received 3 April 2007; received in revised form 29 June 2007; accepted 3 July 2007
The temperature of foods and fluids is a major factor that determines their pleasantness and acceptability. Studies of nonhuman primates have
shown that many neurons in cortical taste areas receive and process not only chemosensory inputs, but oral thermosensory (temperature) inputs as
well. We investigated whether changes in oral temperature activate these areas in humans, or middle or posterior insular cortex, the areas most
frequently identified for the encoding of temperature information from the human hand. In the fMRI study we identified areas of activation in
response to innocuous, temperature-controlled (cooled and warmed, 5, 20 and 50 °C) liquid introduced into the mouth. The oral temperature
stimuli activated the insular taste cortex (identified by glucose taste stimuli), a part of the somatosensory cortex, the orbitofrontal cortex, the
anterior cingulate cortex, and the ventral striatum. Brain regions where activations correlated with the pleasantness ratings of the oral temperature
stimuli included the orbitofrontal cortex and pregenual cingulate cortex. We conclude that a network of taste- and reward-responsive regions of the
human brain is also activated by intra-oral thermal stimulation, and that the pleasant subjective states elicited by oral thermal stimuli are correlated
with the activations in the orbitofrontal cortex and pregenual cingulate cortex. Thus the pleasantness of oral temperature is represented in brain
regions shown in previous studies to represent the pleasantness of the taste and flavour of food. Bringing together these different oral
representations in the same brain regions may enable particular combinations to influence the pleasantness of foods.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Oral temperature; Flavour; Taste; fMRI; Orbitofrontal cortex; Insula; Anterior cingulate cortex
Neurophysiological studies in macaques [1–7] and func-
tional neuroimaging studies in humans [8–12] have demon-
strated that liquid oral stimuli, including pure water, activate the
anterior insula/frontal opercular cortex (primary taste cortex),
and the medial orbitofrontal cortex (secondary taste cortex).
Activity in the primary taste cortex encodes the identity and
intensity of oral stimuli, whereas activity in the orbitofrontal
cortex reflects the hedonic value of oral stimuli as shown by
reductions in responses when foods are ingested to satiety
[2,3,11,13–15]. An unanticipated observation from studies of
these cortical taste areas is that many (macaque) neurons receive
and process not only chemosensory inputs, but somatosensory
(texture/viscosity) and thermosensory (temperature) inputs as
well [16,17]. Moreover, some neurons respond only to specific
temperatures (e.g., cool vs. warm) suggesting a mechanism for
encoding oral temperature independently of taste and texture
Physiology & Behavior 92 (2007) 975–984
⁎Corresponding author. University of Oxford, Department of Experimental
Psychology, South Parks Road, Oxford OX1 3UD, England, UK. Tel.: +44 1865
271348; fax: +44 1865 310447.
E-mail address: Edmund.Rolls@psy.ox.ac.uk (E.T. Rolls).
URL: http://www.cns.ox.ac.uk (E.T. Rolls).
0031-9384/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
In contrast to the primate neurophysiology used as a model
of taste processing in humans, no neuroimaging study to date
has investigated whether changes in oral temperature modulate
the activity of these areas in humans, or the activity of middle or
posterior insular cortex, the areas most frequently identified for
the encoding of temperature information from the human hand
[18–21]. To this end, we conducted an fMRI study to identify
areas of activation in response to temperature-controlled
(cooled and warmed) liquid introduced into the mouth. The
subjects also provided psychophysical ratings of the subjective
temperature of the stimuli as well as their pleasantness to enable
investigation of how subjective oral temperature and pleasant-
ness are represented in the brain. Features of this investigation
are that a special Peltier device was designed and built to enable
the delivery of liquids at different temperatures into the mouth
in the fMRI environment ; that the temperature stimuli
consisted of tasteless liquid so that activation related to
temperature and not taste could be imaged; and that a sweet
(glucose) taste stimulus was included in the stimulus set to
enable identification of taste cortical areas, so that we could
determine whether temperature-responsive cortical areas over-
lapped or were distinct from taste-responsive cortical areas.
Subjects were recruited to participate in a single experimen-
tal session during which they sampled fluids of different
temperatures introduced into their mouths. Because of the
difficulty of testing subjects with a bulky Peltier controlled
device situated in the magnet, considerable training of the
subjects to accustom them to the experiments was provided, and
the number of subjects was limited to five. As described below,
we performed a group random effects statistical analysis on the
data to take into account the within as well as between subjects
variability. We note that although with this number of subjects
the results cannot necessarily be generalized from this group of
subjects in which significant effects were found to the whole
population , the effects described are statistically significant
within this group of subjects, as described below. We further
strengthened the statistical analyses by performing conjunction
analyses, as shown below, which, indicating convergent
activation in some brain areas of oral temperature and taste
inputs, provided further evidence that the representations of oral
stimuli described in this paper are reliable. Further evidence for
the reliability of the findings is that correlations between the
psychophysical ratings of temperature and pleasantness
obtained for every stimulus delivery, and the fMRI BOLD
signals, were found in the same brain regions where contrast
analyses showed effects of temperature, as described below. We
further note that a sixth experiment was performed, and
although not included in the group analyses because the subject
was one included in the group, similar results were obtained as
in the other five experiments as described below, providing
further evidence on the reliability of the experimental
procedures and results. In particular, in this sixth experiment,
activations to the oral temperature stimuli were confirmed to be
present in the orbitofrontal cortex, ventral striatum, and anterior
insula (Pb0.05 svc in all cases). No restrictions were placed on
subjects' consumption of food and fluids prior to participation
in the study. The study received ethical and safety approval
from the School of Dentistry Institutional Review Board at the
University of Chapel Hill.
2.2. Oral temperature stimuli and experimental design
Using fMRI, the BOLD (blood oxygenation-level dependent)
of temperature-controlled liquids into the mouth. Three tempera-
tures were selected to span the range commonly encountered
during ingestion of liquids: 5 °C (cold drink), 20 °C (tap water),
and 50 °C (warm tea). The liquid was an almosttasteless solution
Fig. 1. Schematic of trial sequence. The time line depicts the first half of a single trial. The second half of the trial was identical to the first, except that the liquid
delivered was always the tasteless control solution (artificial saliva) at 20 °C (room temperature). A single complete trial lasted 1 min 6 s.
976S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
KCl and 2.5 mM NaHCO3, dissolved in water. This ‘artificial
saliva’ was employed instead of pure water to minimize the
activation of cortical taste areas, which are activated by water in
the mouth . In addition, a 1 M glucose solution (dextrose in
the artificial saliva) at 20 °C was included as a test stimulus to
differentiate cortical areas encoding temperature from those
The stimuli were delivered in a single event design in a
pseudo-randomized permuted sequence, with ten repetitions of
each stimulus. Specifically, the order of stimuli was such that
the most extreme temperatures (i.e., 5 and 50 °C) were not
presented on consecutive trials. This constraint was placed on
stimulus presentation so that the liquids would have sufficient
time to reach their selected temperatures, given the limitations
of the temperature-control system. Each test temperature
stimulus (5, 20 or 50 °C) or the glucose stimulus (20 °C) was
followed with artificial saliva at 20 °C (the control temperature
stimulus and a rinse as well as one of the four test stimuli)
during each trial as shown schematically in Fig. 1.
2.3. Subject instructions
During each trial, the test and control stimuli were delivered
at a constant rate over a 4 s period. The subject was instructed to
use the tongue to distribute the liquid around the mouth during
the delivery period. For 2 s immediately following delivery of
each stimulus, the subject was instructed to hold the liquid in the
mouth and not to swallow until cued to do so. After swallowing
each liquid stimulus, the subject rated its perceived temperature
and pleasantness. Using modifications of the labeled magnitude
scale (LMS) described by Green et al. , the temperature
rating scale consisted of vertically stacked, back-to-back scales
for warm vs cool (i.e., a ‘butterfly scale’), and also for
pleasantness vs unpleasantness (see Fig. 2).
2.4. Apparatus for oral temperature stimulus delivery and
psychophysical data collection
Stimulus delivery and psychophysical data collection were
controlled by a PC computer located in the scanner room. The
two stimulus liquids (artificial saliva and the glucose solution)
were held in individual, disposable plastic syringes. Each liquid
was forced out of the syringe into a capillary tube (1.59 mm
internal diameter) by a separate syringe pump (Type 33,
Harvard Apparatus, Holliston, MA) secured 15 ft from the
magnet. Each tube terminated in a non-magnetic liquid heating/
cooling chamber, 2.6 ml in volume, near the subject's mouth.
The chambers were bonded to a Peltier wafer which was
electronically controlled via temperature-feedback circuitry,
enabling the liquids in the chambers to be heated or cooled
according to the experimenter's specifications. The liquid
heating/cooling device was custom-designed and built for use
in the fMRI environment (Ollie Monbureau, UNC Chapel Hill,
School of Dentistry) .
The introduction of the liquid stimuli into the mouth was
achieved using a customized, silicone-rubber snorkel mouth-
piece, fitted to each subject's dentition using fast-setting dental
impression material (Regisil 2×, Dentsply International, York,
PA). A pair of capillary tubes (0.762 mm internal diameter)
joined the outlet of the two liquid storage chambers to a Y-
the delivery tube, from chamber to tongue tip, was 17.5 cm,
corresponding to a volume between tongue tip and the chamber
of 0.08 ml. This small volume represents a ‘dead space’ between
the temperature-controlled liquid and the subject's mouth.Thus,
ambient temperature. This volume represented just 5.3% of the
total volume for the stimulus. A third capillary tube ran directly
device. This tube supplied the control stimulus, i.e., artificial
saliva at 20 °C (ambient room temperature), that was delivered
during the second half of every trial (Fig. 1).
Fig. 2. Response scales used to collect subjectiveratings of stimulustemperature
and stimulus pleasantness/unpleasantness.
977 S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
Instruction cues and response scales were shown to subjects on
responses were obtained from subjects using a button box which
stimulus delivery and psychophysical data collection.
2.5. fMRI data collection
Functional MRI was performed using a Siemens 3T Allegra
head-only scanner (Siemens Medical Systems, Erlangen,
Germany). A double echo EPI sequence was employed. The
first echo was a gradient echo and the second echo was an
asymmetric spin echo with a time offset of 15 ms. The
asymmetric spin echo reduced the signal loss in the frontal
region caused by the susceptibility artifacts, while preserving
the signal alteration induced by the functional BOLD contrast.
The images were collected in 25 coronal slices of 5 mm
thickness to cover the frontal region of the brain. The image
resolution within plane was 3 mm in both directions with field
of view of 192 mm. TR was 3 s. TEs for the two echoes were
28 ms and 68 ms, respectively.
2.6. Psychophysical data analysis
Psychophysical data were analysed via SAS statistical
analysis software using procedure MIXED for a mixed-model
ANOVA and related approaches. The data for both of the
psychophysical ratings (i.e., subjective temperature and
pleasantness) ranged from −1 (the coolest or most unpleasant
rating available) to +1 (the warmest or most pleasant rating
available). The values were analysed without transformation.
Of interest was the extent to which different physical
temperatures resulted in different levels of perceived temper-
ature and pleasantness.
2.7. fMRI data analysis
The imaging data were analyzed using SPM5 (Wellcome
Department of Imaging Neuroscience, University of London).
Activations and correlations produced by temperature and taste (activations
labeled T were in the insular taste cortex)
Conjunction (50–20, 5–20)
Positive correlation with temperature
Negative correlation with temperature −44, −36, 56
Negative correlation with pleasantness 64, −22, 42
−50, −34, 30
−52, −32, 28
−48, −28, 42
−56, −24, 32
−44, −28, 56
Glucose (T)40, 8, −6
−42, −6, −10
50, 24, 4
52, −6, 4
−36, 20, 0
40, 14, 6
44, 0, 2
36, −4, 8
40, 2, −18
−36, −8, 16
Conjunction (50−rinse, 5−rinse) (T)
with pleasantness (T)
38, 10, −6 3.07 0.000 (SVC)
Glucose6, 32, 26
12, 30, −20
2, 42, −6
0, 50, 14
−2, 16, 44
−6, 20, −4
−6, 38, 10
4, 34, 32
2, 44, 16
0, 16, 44
2, 44, 8
0, 44, 8
Conjunction (50−rinse, 5−rinse)
Negative correlation with temperature
Positive correlation with pleasantness
Glucose24, 18, −26
10, 28, −12
−28, 58, −12
2, 42, −6
10, 36, −20
26, 20, −14
44, 38, −6
−36, 28, −8
30, 22, −26
−48, 28, −10
Positive correlation with temperature
Negative correlation with temperature 8, 30, −24
Positive correlation with pleasantness 8, 28, −24
Negative correlation with pleasantness 42, 58, −4
54 42 −43.74
−14, 14, 2
14, 12, −6
12, 12, −2
−16, 10, 4
8, 16, −2
−18, 14, 2
Table 1 (continued)
Conjunction (50−rinse, 5−rinse)
Conjunction (50–20, 5–20)
14, 10, −4
16, 14, −4
Conjunction (50−rinse, 5−rinse)
Positive correlation with temperature
52, 0, 28
46, 4 46
60, 0, 34
32, −8, 42
52, −2, 40
46, 0, 42
64, −10, 203.910.000 (C)
978S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
Pre-processing of the data used SPM5 for realignment, re-
slicing with generalized interpolation, normalization to the MNI
coordinate system (Montreal Neurological Institute), and spatial
smoothing with a 8 mm full width at half maximum isotropic
Gaussian kernel and global scaling. A high-pass filter with a
cut-off period of 128 s was applied.
A general linear model was then applied to the time course of
activation where stimulus onsets were modeled as single
impulse response functions and then convolved with the
canonical hemodynamic response function, HRF and a duration
parameter of 4 s. Time derivatives were included in the basis
functions set. Following smoothness estimation, linear contrasts
of parameter estimates were defined to test the specific effects of
each condition with each individual dataset.
Voxel values for each contrast resulted in a statistical
parametric map of the corresponding t statistic, which was then
transformed into the unit normal distribution (SPM Z). The
statistical parametric maps from each individual dataset were
then entered into second-level, random effects analyses
accounting for both scan-to-scan and subject-to-subject vari-
ability. More precisely, the sets of individual statistical maps
corresponding to a specific effect of interest were entered in a
multiple regression model as implemented in SPM5, and the
corresponding group effects were assessed by applying linear
contrasts to the (second-level) parameter estimates generating a
t-statistics map for each group effect of interest. The above
allowed us to perform conjunction analyses at the second
(group) level to test for example whether taste and temperature
activated the same voxels. (This was performed by running the
group analysis for each effect separately, and then, at the group
level, selecting the conjunction not global selection, in order to
test the true conjunction as implemented in SPM5, not the
global null hypothesis.) The correlation analyses of the fMRI
BOLD signal with given parameters of interest (i.e., physical
temperature of the stimuli, subjective ratings of stimulus
temperature, and ratings of stimulus pleasantness) were
performed at the second-level through applying one-sample t-
tests to the first-level t-maps resulting from performing linear
parametric modulation as implemented in SPM5.
Reported P values for each cluster as implemented in SPM
based on this group analysis are fully corrected (C in Table 1)
for the number of comparisons (resels) in the entire volume
Fig. 3. A. Subjective ratings of the temperatures of the stimulus liquid at
different temperatures (bars) and the 20 °C control liquid (broken line). The
tasteless liquidshadthe ionic constituents of saliva, andthe glucosesolutionwas
dissolved in the tasteless solution. Error bars show+1 SE. B. Subjective ratings
of the pleasantness of the test (bars) and control (broken line) oral stimuli made
during the scanning. Error bars show+1 SE.
Fig. 4. a. A region of primary somatosensory cortex shown in a parasaggittal section activated by the conjunction of the hot (50–20) and the cold (5–20) stimuli at
[−56, −24, 32]. The bar is calibrated to show the t value. b. Insula. Activations produced in the insula in the 5−rinse condition at [44, 0, 2]. Activation is also shown
above it in the premotor cortex.
979S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
[“whole-brain” multiple comparisons, ]. We supplement
these by describing further activations corresponding to clusters
of voxels significant when corrected for the number of
comparisons made within each region [small volume correction,
(SVC) applied with a sphere of 8 mm chosen to be greater than
or equal to the spatial smoothing kernel, ], in order to
provide an indication of effects appearing in further brain areas
such as the insular, anterior cingulate, orbitofrontal and
somatosensory cortices, and the ventral striatum shown to be
of interest because of activations found in prior studies of oral
representations in the brain [10–12,14,15,24,27,28].
3.1. Psychophysical results
Fig. 3A illustrates the mean ratings of subjective temper-
ature made by the subjects for the control and test solutions,
across all four of the experimental conditions. A mixed-model
ANOVA indicated that the temperature ratings varied accord-
ing to the actual temperature of the solution (F3,183=501.69,
Pb0.0001). Pairwise tests, corrected for multiple comparisons,
demonstrated that solutions at different temperatures were rated
Ratings of the pleasantness of the stimulus liquids are shown
in Fig. 3B. Subsequent analyses indicated that the pleasantness
of the test solutions varied (F3,183=29.61, Pb0.0001), with
post-hoc tests showing that the 50 °C solution was found less
pleasant than all other solutions (i.e., 5 °C and 20 °C tasteless
solution, and 20 °C glucose solution), which did not differ from
each other in their rated pleasantness.
3.2. Functional neuroimaging results
Table 1 shows the MNI coordinates, Z values, and the
significance levels of the effects described next.
3.2.1. Somatosensory cortex
Activations of somatosensory cortex areas by oral temper-
ature were found in a number of mutually consistent contrasts.
For example, the contrast hot ‘50−rinse’ showed activation in
part of the primary somatosensory region, as did ‘50–20’, and
so did cold ‘5−rinse’ and ‘5−20’, as shown in Table 1. The
activations produced by all these stimuli overlapped, and were
in a region that was centered bilaterally near y=−30 and z=34.
These findings are supported by a conjunction analysis for areas
activated by both hot and cold. An example of the conjunction
analysis is shown in Fig. 4a, which indicates a region of primary
somatosensory cortex activated by the conjunction (50–20 in
conjunction with 5–20). Interestingly, this area was not
activated by glucose-rinse, indicating that this is a somatosen-
sory cortical area not involved in taste processing, though the
new evidence we present here shows that it is involved in oral
temperature processing. (During the rinse period shown in
Fig. 1 the tasteless control solution was delivered as a rinse, and
this is therefore a useful control period used in some of the
As expected, anterior regions of the insula, at for example
[40, 8, −6] and [−42, −6, −10] were activated by the taste of
glucose (see Table 1). This is in a region shown in previous
studies to be activated by taste stimuli, and is probably the
primary taste cortex [8,11,24]. In macaques, some neurons in
the primary taste cortex are known to be activated by oral
temperature stimuli . Nearby parts of the insula were
activated by the temperature stimuli. For example, activations
in this anterior insula region were found for the contrasts hot
‘50−rinse’ and cold ‘5−rinse’, and cold–hot ‘5–50’ (see
Table 1). An example of activations produced in the insula in
the ‘5−rinse’ condition at [44, 0, 2] is shown in Fig. 4b, and this
is part of the primary taste cortex. The same diagram shows
activation described later in the premotor cortex. The
activations for the oral temperature conditions were in a region
of the insula extending from y=20 to y=−6, and z=6 to
z=−18. A conjunction analysis between temperature hot
‘50−rinse’ and cold ‘5−rinse’ revealed a significant effect
centered at [40, 2, −18]. Overall, the region of the insula
activated by oral temperature included at least part of the
insular region that is the primary taste cortex.
Fig. 5. Orbitofrontal cortex. a. A conjunction of temperature and glucose (50−rinse, 5−rinse, and glucose-rinse) activated the orbitofrontal cortex at [−46, 29, −10].
This activation extends up to and is continuous with activations in the anterior insula. b. Positive correlation of the BOLD signal with the subjective ratings of
temperature of the temperature series in the caudal orbitofrontal cortex [−30, 24, −16]. A positive correlation reflected a higher BOLD signal with increasing
temperature. The activation extended forwards as far y=31.
980S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
3.2.3. Orbitofrontal cortex
Hot, cold, hot–cold, and cold–hot contrasts showed
activation in the orbitofrontal cortex, as shown in Table 1. As
expected from previous studies [11,12], glucose taste activated
the orbitofrontal cortex. Of particular interest was that a
conjunction of temperature and glucose (50−rinse, 5−rinse,
and glucose-rinse) showed activation in the orbitofrontal cortex.
The region where this particular conjunction showed a
significant result was in the lateral orbitofrontal cortex [−46,
29, −10] as shown in Fig. 5a, and can be clearly shown
extending up to and continuous with activations in the anterior
3.2.4. Cingulate cortex
Activations by oral temperature and glucose taste were found
in nearby parts of the anterior cingulate cortex. The contrasts hot
‘50−rinse’, cold ‘5−rinse’ and hot–cold ‘50–5’ each showed
activation of the pregenual cingulate (e.g. 5−rinse shown in
Fig. 6a at [0, 50, 14]) and a more dorsal area of the anterior
cingulate cortex (e.g. 5−rinse shown in Fig. 6a at [−2, 16, 44]).
This result was supported by the conjunction analyses between
hot ‘50−rinse’ and cold ‘5−rinse’ (see Table 1, pregenual at [2,
44, 16] and dorsal cingulate at [0, 16, 44]). A medial
orbitofrontal cortex/subgenual part of the cingulate was
activated by hot oral temperature ‘50−rinse’ [2, 42, −6] and
glucose taste [12, 30, −20], but not by cold ‘5−rinse’.
3.2.5. Ventral striatum
Another region for which activations were found for oral
temperature contrasts is the ventral striatum. Activations were
found for hot ‘50−rinse’ and ‘50–20’, cold ‘5−rinse’, and for a
conjunction between hot and cold (see Table 1). Fig. 6b shows a
conjunction between 50–20 and 5–20, with bilateral activation
([16, 14, −4] on the right). No significant activation in the
ventral striatum was found for the glucose taste condition in this
group of subjects.
3.2.6. Premotor cortex
Activations in the premotor cortex were found in overlapping
areas for taste and oral temperature contrasts. Glucose taste
activated the premotor cortex at [52, 0, 28] while hot ‘50−rinse’
and cold ‘5−rinse’ activated nearby areas, as shown in Table 1.
This result is supported by a conjunction analysis between
50−rinse and 5−rinse, which showed an activated area at [52,
−2, 40], as well as by a conjunction analysis which included
the oral temperature stimuli (50–rinse, 5–rinse) and the
glucose taste stimulus, which showed an activated area at [46,
0, 42]. An example of the activation in the premotor cortex by
oral temperature (contrast 5–rinse) (centered at [60, 0, 34]) is
included in Fig. 4b.
3.2.7. Correlation analysis of fMRI data with subjective
For these correlation analyses, the three temperature stimuli
5, 20 and 50 °C were used. Positive correlations of the BOLD
signal with this temperature series were found in the lateral
orbitofrontal cortex (area 47/12 l as defined by Price ), as
illustrated in Fig. 5b, and in the premotor cortex, as shown in
Table 1. (The positive correlation in the lateral orbitofrontal
cortex may be related to the fact that the higher temperatures
were rated as less pleasant, as shown in Fig. 3.) Negative
correlations (i.e. greater activations for colder temperatures)
were found in the medial orbitofrontal cortex, in the pregenual
cingulate cortex, and in the somatosensory cortex, as shown in
Table 1. (The negative correlation in the medial orbitofrontal
and pregenual cingulate cortex may be related to the fact that the
lower temperatures were rated as more pleasant, as shown in
Fig. 3.) The correlation analyses not only provide confirmation
of the evidence described above based on contrasts of different
temperatures, but also show how the subjective feeling of
temperature is correlated with activations in these different
3.2.8. Correlation analysis of fMRI data with hedonic ratings
For these correlation analyses, the three temperature stimuli
5, 20 and 50 °C were also used. Positive correlations of the
BOLD signal with the pleasantness ratings of this temperature
series was found in the medial orbitofrontal cortex, for example
at [8, 28, −24] (see Table 1). Positive correlations with
pleasantness were not evident in the primary taste cortex and in
Fig. 6. a. Anterior cingulate cortex. An activation by cold (5–rinse) is shown in the pregenual cingulate cortex at [0, 50, 14] (marked by crosshairs), together with
activationin a more dorsal area of the anterior cingulate cortex producedby the same coldstimulusat [−2, 16, 44].b. Ventral striatum. A conjunctionbetweenhot (50–
20) and cold (5–20) produced bilateral activation (e.g. [16, 14, −4] on the right).
981S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
the somatosensory cortex. Negative correlations (i.e. greater
activations for temperatures rated as unpleasant) were found in
the primary taste cortex (for example at [38, 10, −6]), in the
lateral orbitofrontal cortex ([42, 58, −4]), and in the
somatosensory cortex (e.g. [64, −22, 42], as shown in Table
1; and in the amygdala ([30, 12, −24] Z=3.43 P=0.000 SVC).
The correlation analyses provide confirmation of the evidence
described above based on contrasts of different temperatures,
but also show that pleasantness was correlated with activations
in the medial orbitofrontal cortex, and that the subjective feeling
of unpleasantness was correlated with activations in a number of
areas, including the lateral orbitofrontal cortex (see Table 1).
Correlations with unpleasantness were also found in or near the
taste insula and in the somatosensory cortex. These could be
related to the fact that the warmer temperatures were less
pleasant as shown in Fig. 3B, and these brain areas may have
responded more to the warmer vs the cold temperature.
A major finding of this investigation is that the temperature
of what is in the mouth produces activation in the anterior part
of the insula, in a region which was shown in this study was also
activated by the taste of glucose. This region is probably the
primary taste cortex, based on its correspondence with the
anterior taste insula in macaques which receives taste inputs
directly from the taste thalamus  and contains taste neurons
[1–5,31], and on the fact that it has been shown in previous
functional neuroimaging studies in humans to be responsive to
taste stimuli [8–12]. The representation of oral temperature in
the taste insula has been established by an investigation in
which some single neurons in the macaque primary taste cortex
were found to be tuned to oral temperature, some to taste, and
some to both oral temperature and taste . The present
findings indicate that a similar situation holds in humans, in that
both oral temperature and taste activate this anterior part of the
The activations in the insula by oral temperature were mainly
at the anterior end of the insula, in a region which at least largely
overlapped with the primary taste cortex. There was no strong
evidence for oral temperature representations more posteriorly
in the insula. However, consistent with our finding of oral
temperature representations in the insula, representations of
temperature on the body surface, for example the hand, are
found in the insula, but further posteriorly [19,32,33], that is
behind the primary taste cortex. Taken together, these
investigations show that temperature is represented in the
human insula, with the oral representation in the taste insula,
where it can potentially be combined with taste representations;
and the temperature of other parts of the body is represented in
more posterior parts of the insula that have somatosensory
representations of other parts of the body.
We note that Craig has referred to the representation of the
temperature of parts of the body surface as “interoception” .
However, this is puzzling, for at least touch to the body surface
might normally be considered as exteroceptive sensing, and the
temperature of the hand does not appear to be conceptually
more interoceptive than this. However, the temperature of what
is in the mouth might be thought to be closer to true
interoceptive sensing, and perhaps the term interoceptive
sensing would apply to oral temperature sensing somewhat
better. Indeed the epithelium of much of the gastrointestinal
tract is of endodermal origin , and interoceptive might apply
to this. However, we note that embryologically the mucous
membranes of the lips, cheeks, gums, part of the floor of the
mouth, and the palate have their origin in tissue that is of
ectodermal origin , so even perhaps oral temperature
sensing has its origins in part in external sensors.
Part of the advantage of having a representation of oral
temperature (and also oral texture ) in the taste cortex may be
that neurons can then encode combinations of taste, texture and
oral temperature, which in particular combinations may have
significance, both in modern times and in evolutionary history.
These combinations may provide the basis for particular
combinations of temperature, taste, texture and odor to be
especially pleasant, and also provide a basis for sensory-specific
satiety, by allowing a reduction of firing of neurons that respond
to particular combinations . An additional part of the
evolutionary adaptive value of oral temperature sensing may be
that this could give an indication of the potentially damaging
consequences for thermal regulation of the ingestion of large
quantities of cold fluid.
In contradistinction to the anterior insular representation of
oral temperature, primary somatosensory cortex (areas 1, 2 and
3) also represents oral temperature, but not oral taste, as shown
by the findings of this investigation. Thus oral temperature (and
we surmise also body surface temperature) has a representation
that is independent of the insula, and in primary somatosensory
areas. Thus not all temperature representations are in the
“interoceptive insula” [19,32].
The taste insula projects into the orbitofrontal cortex in
macaques , and corresponding to this, in the present
investigation activations by oral temperature were also found in
the orbitofrontal cortex, and in the pregenual cingulate cortex,
which receives projections from the orbitofrontal cortex and
contains taste neurons [36,37]. Activations in the medial
orbitofrontal cortex and cingulate cortex were correlated with
the pleasantness ratings of the temperature stimuli provided on
every trial of the functional imaging, consistent with much other
evidence that the hedonic value of many stimuli is represented
in these regions [27,28,38–42]. Activations in the lateral
orbitofrontal cortex were negatively correlated with the pleas-
antness ratings of the temperature stimuli provided on every
trial of the functional imaging, consistent with other evidence
that unpleasant stimuli tend to be represented in this region
[27,28,38,43,44]. However, the present study also revealed that
with the temperature stimuli used there was also a negative
correlation with pleasantness in the anterior insula, and this may
have been related to the fact that with the temperature stimuli
used, any difference in activations to different temperatures was
also potentially related to hedonics, as shown by the data in
A region to which the orbitofrontal cortex projects, the
ventral striatum, also had activations that reflected oral
982 S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
temperature, as shown in Table 1. This region in implicated in
conditioned incentive effects [27,45], and the activations by oral
temperature might be part of a system for responding to
rewarding or punishing conditioned temperature stimuli.
In addition, a part of ventral premotor cortex area 6 was
activated in this study by the oral temperature stimuli (and in
previous studies with taste/oral somatosensory stimuli including
water in the mouth ). This region is known to receive inputs
in macaques from the insular (primary) taste cortex . This
region also receives somatosensory inputs from much of the
length of the fronto-parietal opercular cortex , so may
reflect activations in these areas to oral stimuli.
In conclusion, in this fMRI study we have shown that oral
temperature is represented in the human primary taste cortex in
the anterior insula, and in regions to which it connects such as
the orbitofrontal cortex. Part of the advantage of having a
representation of oral temperature in these regions is that
neurons can then encode combinations of taste, texture and oral
temperature [7,16]. These combination-responsive neurons may
provide the basis for particular combinations of temperature,
taste, texture and odor to be especially pleasant [27,47].
 Scott TR, Yaxley S, Sienkiewicz ZJ, Rolls ET. Gustatory responses in the
frontal opercular cortex of the alert cynomolgus monkey. J Neurophysiol
 Rolls ET, Scott TR, Sienkiewicz ZJ, Yaxley S. The responsiveness of
neurones in the frontal opercular gustatory cortex of the macaque monkey is
independent of hunger. J Physiol 1988;397:1–12.
 Yaxley S, Rolls ET, Sienkiewicz ZJ. The responsiveness of neurons in the
insular gustatory cortex of the macaque monkey is independent of hunger.
Physiol Behav 1988;42:223–9.
 Yaxley S, Rolls ET, Sienkiewicz ZJ. Gustatory responses of single
neurons in the insula of the macaque monkey. J Neurophysiol
 Scott TR, Plata-Salaman CR. Taste in the monkey cortex. Physiol Behav
 Kadohisa M, Rolls ET, Verhagen JV. Neuronal representations of stimuli in
the mouth: the primate insular taste cortex, orbitofrontal cortex, and
amygdala. Chem Senses 2005;30:401–19.
 Verhagen JV, Kadohisa M, Rolls ET. The primate insular/opercular taste
cortex: neuronal representations of the viscosity, fat texture, grittiness,
temperature and taste of foods. J Neurophysiol 2004;92:1685–99.
 Small DM, Zald DH, Jones-Gotman M, Zatorre RJ, Pardo JV, Frey S, et al.
Human cortical gustatory areas: a review of functional neuroimaging data.
 Zald DH, Lee JT, Fluegel KW, Pardo JV. Aversive gustatory stimulation
activates limbic circuits in humans. Brain 1998;121:1143–54.
 O'Doherty J, Rolls ET, Francis S, Bowtell R, McGlone F. The repre-
sentation of pleasant and aversive taste in the human brain. J Neurophysiol
 de Araujo IET, Kringelbach ML, Rolls ET, Hobden P. The representation
of umami taste in the human brain. J Neurophysiol 2003;90:313–9.
 de Araujo IET, Rolls ET. The representation in the human brain of food
texture and oral fat. J Neurosci 2004;24:3086–93.
 Rolls ET, Sienkiewicz ZJ, Yaxley S. Hunger modulates the responses to
gustatory stimuli of single neurons in the caudolateral orbitofrontal cortex
of the macaque monkey. Eur J Neurosci 1989;1:53–60.
 Rolls ET. Brain mechanisms underlying flavour and appetite. Philos Trans
R Soc Lond B 2006;361:1123–36.
 Rolls ET. Sensory processing in the brain related to the control of food
intake. Proc Nutr Soc 2007;66:96–112.
 Kadohisa M, Rolls ET, Verhagen JV. Orbitofrontal cortex neuronal
representation of temperature and capsaicin in the mouth. Neuroscience
 Rolls ET, Verhagen JV, Kadohisa M. Representations of the texture of food
in the primate orbitofrontal cortex: neurons responding to viscosity,
grittiness and capsaicin. J Neurophysiol 2003;90:3711–24.
 Casey KL, Minoshima S, Morrow TJ, Koeppe RA. Comparison of human
cerebral activation pattern during cutaneous warmth, heat pain, and deep
cold pain. J Neurophysiol 1996;76(1):571–81.
 Craig AD, Chen K, Bandy D, Reiman EM. Thermosensory activation of
insular cortex. Nat Neurosci 2000;3(2):184–90.
 Davis KD, Kwan CL, Crawley AP, Mikulis DJ. Functional MRI study of
thalamic and cortical activations evoked by cutaneous heat, cold, and
tactile stimuli. J Neurophysiol 1998;80(3):1533–46.
 Maihofner C, Kaltenhauser M, Neundorfer B, Lang E. Temporo-spatial
analysis of cortical activation by phasic innocuous and noxious cold
stimuli-a magnetoencephalographic study. Pain 2002;100(3):281–90.
 Guest S, Essick G, Young M, Lee A, Phillips N, McGlone F. Oral
hydration, parotid salivation and the perceived pleasantness of small water
volumes. Physiol Behav 2006;89:724–34.
 Friston KJ, Holmes AP, Worsley KJ. How many subjects constitute a
study? Neuroimage 1999;10(1):1–5.
 de Araujo IET, Kringelbach ML, Rolls ET, McGlone F. Human cortical
responses to water in the mouth, and the effects of thirst. J Neurophysiol
 Green BG, Shaffer G, Gilmore MM. Derivation and evaluation of a
semantic scale of oral sensation magnitude with apparent ratio properties.
Chem Senses 1993;18:683–702.
 Worsley KJ, Marrett P, Neelin AC, Friston KJ, Evans AC. A unified
statistical approach for determining significant signals in images of
cerebral activation. Hum Brain Mapp 1996;4:58–73.
 Rolls ET. Emotion Explained. Oxford University Press; 2005.
 Kringelbach ML, Rolls ET. The functional neuroanatomy of the human
orbitofrontal cortex: evidence from neuroimaging and neuropsychology.
Prog Neurobiol 2004;72:341–72.
 PriceJL. Architectonicstructureof the orbitalandmedialprefrontalcortex.
In: Zald DH, Rauch SL, editors. The Orbitofrontal Cortex. Oxford: Oxford
University Press; 2006. p. 3–17.
 Pritchard TC, Hamilton RB, Morse JR, Norgren R. Projections of thalamic
gustatory and lingualareas in the monkey, Macaca fascicularis. J Comp
 Scott TR, Plata-SalamanCR, Smith VL, Giza BK. Gustatory neural coding
in the monkey cortex: stimulus intensity. J Neurophysiol 1991;65:76–86.
 Craig AD. How do you feel? Interoception: the sense of the physiological
condition of the body. Nat Rev Neurosci 2002;3(8):655–66.
 Brooks JC, Zambreanu L, Godinez A, Craig AD, Tracey I. Somatotopic
organisation of the human insula to painful heat studied with high
resolution functional imaging. Neuroimage 2005;27(1):201–9.
 Hamilton WJ, Boyd JD, Mossman HW. Human Embryology. 2 ed.
Cambridge: Heffer & Sons; 1952.
 Baylis LL, Rolls ET, Baylis GC. Afferent connections of the orbitofrontal
cortex taste area of the primate. Neuroscience 1995;64:801–12.
 Rolls ET, Verhagen JV, Gabbott PL, Kadohisa M. Taste and Oral Texture
Representations in the Primate Medial Orbitofrontal and Pregenual
Cingulate Cortices; 2007.
 Rolls ET. The anterior and midcingulate cortices and reward. In: Vogt BA,
editor. Cingulate Neurobiology & Disease. Oxford: Oxford University
 Rolls ET. The neurophysiology and functions of the orbitofrontal cortex.
In: Zald DH, Rauch SL, editors. The Orbitofrontal Cortex. Oxford: Oxford
University Press; 2006. p. 95–124.
 Rolls ET, McCabe C, Redoute J. Expected value, reward outcome, and
temporal difference error representations in a probabilistic decision task.
Cerebral Cortex; 2007, doi:10.1093/cercor/bhm097.
in cravers vs non-cravers. European Journal of Neuroscience in press.
and olfactory pathways in the human brain. Eur J Neurosci 2007;25:1855–64.
983S. Guest et al. / Physiology & Behavior 92 (2007) 975–984
 Rolls ET. Memory, Attention, and Decision-Making: A Unifying
Computational Neuroscience Approach. Oxford: Oxford University
 O'Doherty J, Kringelbach ML, Rolls ET, Hornak J, Andrews C. Abstract
reward and punishment representations in the human orbitofrontal cortex.
Nat Neurosci 2001;4:95–102.
 Rolls ET, Kringelbach ML, de Araujo IET. Different representations of
pleasant and unpleasant odors in the human brain. Eur J Neurosci
 Cardinal N, Parkinson JA, Hall J, Everitt BJ. Emotion and motivation: the
role of the amygdala, ventral striatum, and prefrontal cortex. Neurosci
Biobehav Rev 2002;26:321–52.
 Cipolloni PB, Pandya DN. Cortical connections of the frontoparietal
 Rolls BJ, Wood RJ, Rolls ET. Thirst: the initiation, maintenance, and
termination of drinking. Prog Psychobiol Physiol Psychol 1980;9:263–321.
984S. Guest et al. / Physiology & Behavior 92 (2007) 975–984