Human feelings: why are some more aware than others? A.D. (Bud) Craig |

Update
TRENDS in Cognitive Sciences
Vol.8 No.6 June 2004
| Research Focus
Human feelings:
why are some more aware than others?
A.D. (Bud) Craig
Atkinson Research Laboratory, Barrow Neurological Institute, 350 West Thomas Rd, Phoenix, AZ 85013, USA
A recent article reports that human perception of heartbeat timing is mediated by right (non-dominant)
anterior insular cortex, and that the activity and the size
of this region is directly correlated with individuals’
subjective awareness of inner body feelings and emotionality. These results support the somatic-marker
hypothesis of consciousness (a modern successor to
the James–Lange theory of emotion) and the neuroanatomical concept that human awareness is based on
a phylogenetically distinct interoceptive pathway.
Critchley, Dolan and colleagues have intensively pursued
the identification of forebrain regions involved in the
neural representation of emotion, primarily using functional imaging [1 –3]. Their recent contribution to this
endeavor is an elegant study that matches individuals’
subjective perception of their heartbeat and psychometric
measures of their interoceptive awareness and emotionality with a region of the cerebral cortex that appears to be
anatomically unique to humans [4]. Their findings solidly
confirm that right anterior insula (rAI) is important for
explicit subjective awareness and, significantly, offer a
substantive anatomical explanation as to why some
individuals are more aware of their feelings than others.
Their work sets the stage for intimate structural analyses
of the very essence of human feelings.
Interoceptive feelings
We all recognize ‘gut feelings’ that help guide our behavior.
The James – Lange theory of emotion proposed that
afferent feedback from muscles and viscera, driven by
evolutionarily shaped autonomic commands that accompany each emotion, provide the brain with a sensory
image, or ‘feeling’, that characterizes the active motivational state. Thus, you feel scared in part because your
heart races and your pupils dilate as the growling bear
approaches you. Conversely, it is said that quadraplegics
with sensory feedback limited to the face have blunted
affect [5]; and interestingly, stimulation of the vagus nerve
can relieve depression [6]. If an emotion is defined as a
motivation and a concurrent characteristic feeling [7],
then identifying the part of the brain that provides the
image of that subjective feeling (if it can indeed be
pinpointed) can show us how we are aware of our selves.
Of course, visceral sensation is notoriously vague,
and many studies have shown huge variability in
subjective awareness of inner feelings, classically
Corresponding author: A.D. (Bud) Craig ([email protected]).
www.sciencedirect.com
referred to as ‘interoception’ [8,9]. Intravenous injections of agents that modulate visceral activity (adrenalin, vasopressin, lactate, insulin, etc.) elicited varying
results. However, the heartbeat timing task used by
Critchley et al.. [4] is a fairly reliable test of
interoception, in terms of reproducibility and covariance with other measures of emotional self-awareness.
In this test, a signal is delivered either in synchrony
with each QRS wave (or oximeter pulse) or after a fixed
delay. Subjects are asked to report simply whether the
signal occurs in time with their heartbeat or not. With
an average pulse rate of about one per second, a signal
that is 500 ms late seems distinctly disparate, if one is
aware of one’s pulse at all.
A broader concept of interoception as ‘how you feel’
The concept of ‘interoception’ was classically restricted to
visceral sensations, but recent neuroanatomical and
neurophysiological results indicate that sensations related
to the ongoing physiological condition of all organs of the
body – muscles, joints, teeth, and skin as well as the
viscera – are processed together [10]. The incoming
messages are carried in small-diameter primary afferent
fibers to lamina I in the spinal dorsal horn and to the
solitary nucleus in the medulla (see Figure 1). These
pathways represent, respectively, the afferent inputs for
the sympathetic and the parasympathetic halves of the
autonomic nervous system. They provide the sensory
inputs to hierarchically integrative homeostatic mechanisms in the brainstem and hypothalamus that maintain
the health of the body. In primates and especially humans,
a phylogenetically unique thalamo-cortical extension of
these pathways to the insular cortex provides a direct
representation of homeostatic afferent activity that
engenders the distinct bodily feelings with which we are
all so familiar: pain, temperature, itch, muscle burn,
visceral sensations, hunger, thirst, taste, and even sensual
touch.
These feelings represent ‘the material me’, and so this
broader concept of interoception converges with the socalled somatic-marker hypothesis of consciousness proposed by Damasio [5]. In this proposal, the afferent sensory
representation of the homeostatic condition of the body is
the basis for the mental representation of the sentient self.
Recursive meta-representations of homeostatic feelings
allow the brain to distinguish the inner world from the
outer world. Most strikingly, degrees of conscious
awareness are related to successive upgrades in the
Update
240
TRENDS in Cognitive Sciences
Right
anterior
insular
ACC
Dorsal
posterior
insula
Thalamus
MDvc
VMb
VMpo
Mid brain
Pons
Medulla
NTS
Lamina 1
TRENDS in Cognitive Sciences
Figure 1. A summary diagram indicating the parallel ascending pathways for
small-diameter afferent activity that originate in lamina I and the nucleus of the
solitary tract (NTS). These pathways are relayed from the thalamus (by way of
VMb and VMpo, the basal and the posterior parts of the ventral medial nucleus) to
the dorsal posterior insular cortex, which engenders the interoceptive representation of the physiological condition of the body. This region in humans is rerepresented in the right (non-dominant) anterior insula, which is important for
subjective awareness of feelings and emotions. (ACC, anterior cingulate cortex;
MDvc, ventro-caudal portion of the medial dorsal nucleus)
self-representational maps. That is, individual differences
in emotional awareness are predicted to be directly related
to differences in the capacity for interoceptive feelings. It is
from this perspective that the results of Critchley et al. [4]
have deep significance.
A revealing experimental design
Critchley et al. designed a Latin square using a synchronized or desynchronized heartbeat tone signal, in contrast
with a series of ten similarly timed tones that either did or
did not include an oddball tone (of matching perceptual
difficulty). This enabled ANOVA comparisons of both the
main effects of attending to the heartbeat or the auditory
cue, as well as comparisons of the interactions between
these variables and the timing (synchronized or desynchronized). The subjects recorded their ‘yes/no’ judgments
of hearbeat synchronicity with the same finger movements
during a relatively short scanning session (15 min), which
minimized attention lapses. Interoceptive attention illuminated several forebrain regions, including anterior
cingulate (limbic motor cortex), lateral sensorimotor
www.sciencedirect.com
Vol.8 No.6 June 2004
cortex (a target of visceral afferent activity [11,12]),
supplementary motor cortex (involved in manual
responses), and bilateral insular cortices. This pattern
supports the general view that a network of inter-related
forebrain regions is involved in interoceptive attention and
emotional feelings [9]. The interaction between desynchronized timing and interoceptive attention also highlighted several regions (including the precuneus) but,
strikingly, revealed strong activation mainly in one of the
sites illuminated by interoceptive attention: rAI. This
indicated a crucial role for rAI in explicit subjective
awareness of the mismatched heartbeat signal, confirming
previous data showing that a progression of activity to rAI
and orbitofrontal cortex is essential for discriminative
subjective judgments of interoceptive feelings [10,13].
Their next steps were profoundly revealing. Critchley
et al. discovered that activity in rAI was strongly
correlated with individual performance accuracy on the
heartbeat detection task (didactically normalized by
individual performance accuracy in the auditory task).
Furthermore, individual anxiety and negative affect
(albeit not positive affect; cf. [14]) measured during the
task showed a similar correlation with rAI activity, making
a firm connection with subjective emotional feeling. Next,
using MRI morphometry they found that rAI and the
adjacent orbitofrontal region were the only cortical sites
for which physical size correlated with individual interoceptive awareness. Finally, in a separate sample, they
found that rAI was the only site for which size correlated
with self-rated bodily awareness. Thus, their data indicate
that both the activation and the size of rAI are uniquely
correlated with the subjective awareness of internal
feelings of individual humans.
The role of rAI in awareness
These findings strongly support the composite view of the
somatic-marker hypothesis [5] and the primate interoceptive pathway [10] that progressive meta-representations of homeostatic afferent activity in rAI engender
the emotional feelings that characterize human sentience.
The rAI is activated (selectively or in conjunction with the
anterior cingulate) during many emotions, including
anger, happiness, sadness, disgust, and lust, as well as
by music [10]. The ancillary activation of anterior cingulate cortex is consistent with its proposed role as a
behavioral agent, which interacts with rAI and subcortical
homeostatic regions to form the hierarchical network
representing ‘self ’ [1,5,10,15 – 17]. Although the necessary
observations from lesion studies in human patients are
still accumulating [18], the reports of pain asymbolia and
amusia [19] subsequent to rAI damage are consistent with
loss of subjective feelings (in contrast to anosognosia, in
which there is loss of spatial reference rather than
subjective feelings).
Notably, timing disparities between multimodal stimuli
also cause graded, selective activation of rAI [20]. This can
be accommodated if awareness depends on a capacity for
an overview of a quantal series of meta-representations of
feelings across time (from the past into the future).
Intriguingly, the phrase ‘time stood still’ is often used to
describe moments of heightened awareness. Such a
Update
TRENDS in Cognitive Sciences
hypothetical capacity in rAI would readily explain its
association with the uniquely human faculty for music
(in contrast to prosody) – the rhythmic temporal progression of emotionally laden moments. The activation of
cerebellar vermis observed by Critchley et al. might be
related to timing as well (or alternatively, to autonomic
control). They did report a relative suppression of rAI
activity during attention to the desynchronized auditory
stimulus; nevertheless, the issue of timing deserves
further examination.
The morphometric results of Critchley et al. converge
with other anatomical findings that associate rAI, orbitofrontal, and anterior cingulate cortices with human
sentience [10]. For instance, a novel cell type, the socalled spindle cell, is exclusively located in these regions of
the human brain [21]. Recent evidence indicates a
trenchant phylogenetic correlation, in that spindle cells
are most numerous in aged humans, but progressively less
numerous in children, gorillas, bonobos and chimpanzees,
and nonexistent in macaque monkeys [22]. Notably, this
phylogenetic progression also parallels the results of the
mirror test for self-awareness [23].
Evolving awareness
The most important implication of this study emerges from
the correlation of both activity and physical size of rAI with
individual subjective awareness. This implies that individual differences in subjective interoceptive awareness,
and by extension emotional depth and complexity, might
be expressed in the degree of expansion of rAI and adjacent
orbitofrontal cortices. This clearly resembles the relationship between the size of auditory and motor cortices and
the pitch and performance ability of musicians, which is
partly innate and partly the result of training [24]. A
morphological correlation also exists between alexithymia
and right anterior cingulate [25]. The rapid development of
rAI within a brief evolutionary timescale suggests that
nested interoceptive re-representations could be directly
related to the advantages of advanced social interaction.
The extent to which functional imaging and morphometry,
together with other measures of emotional significance
such as galvanic skin resistance and heart rate variability,
can be related to higher social emotions (such as guilt,
empathy, exclusion; e.g. [26]) should receive much more
intense study following the objective structural correlation
revealed by the work of Critchley et al.
Note added in proof: A group in the same laboratory
have just reported imaging and morphometric evidence
relating rAI and anterior cingulate to empathic pain, a
landmark result indicating that the neural basis for
subjective feelings also provides the means for awareness
of the feelings of others [27].
www.sciencedirect.com
Vol.8 No.6 June 2004
241
Acknowledgements
I am very grateful to several colleagues for sharing their comments on this
manuscript.
References
1 Critchley, H.D. et al. (2001) Neuroanatomical basis for first- and
second-order representations of bodily states. Nat. Neurosci. 4,
207 – 212
2 Critchley, H.D. et al. (2002) Volitional control of autonomic arousal: a
functional magnetic resonance study. Neuroimage 16, 909 – 919
3 Dolan, R.J. (2002) Emotion, cognition, and behavior. Science 298, 1191
4 Critchley, H.D. et al. (2004) Neural systems supporting interoceptive
awareness. Nat. Neurosci. 7, 189 – 195
5 Damasio, A.R. (1993) Descartes’ Error: Emotion, Reason, and the
Human Brain, Putnam, New York
6 Sackeim, H.A. et al. (2001) Vagus nerve stimulation (VNS) for
treatment-resistant depression: efficacy, side effects, and predictors
of outcome. Neuropsychopharmacology 25, 713
7 Rolls, E.T. (1999) The Brain and Emotion, Oxford University Press
8 Cannon, W.B. (1987) The James – Lange theory of emotions: a critical
examination and an alternative theory. Am. J. Psychol. 100, 567 – 586
9 Cameron, O.G. (2002) Visceral Sensory Neuroscience, Oxford University Press
10 Craig, A.D. (2002) How do you feel? Interoception: the sense of the
physiological condition of the body. Nat. Rev. Neurosci. 3, 655
11 Strigo, I.A. et al. (2003) Differentiation of visceral and cutaneous pain
in the human brain. J. Neurophysiol. 89, 3294
12 Ito, S. and Craig, A.D. (2003) Vagal input to lateral area 3a in cat
cortex. J. Neurophysiol. 90, 143– 154
13 Craig, A.D. et al. (2000) Thermosensory activation of insular cortex.
Nat. Neurosci. 3, 184– 190
14 Zautra, A.J. (2003) Emotions, Stress, and Health, Oxford University
Press, New York
15 Damasio, A. (2003) Mental self: the person within. Nature 423, 227
16 Johnson, S.C. et al. (2002) Neural correlates of self-reflection. Brain
125, 1808– 1814
17 Petrovic, P. et al. (2002) Placebo and opioid analgesia – imaging a
shared neuronal network. Science 295, 1737 – 1740
18 Bechara, A. and Naqvi, N. (2004) Listening to your heart: interoceptive
awareness as a gateway to feeling. Nat. Neurosci. 7, 102 – 103
19 Bamiou, D.E. et al. (2003) The insula (Island of Reil) and its role in
auditory processing. Literature review. Brain Res. Brain Res. Rev. 42,
143– 154
20 Bushara, K.O. et al. (2001) Neural correlates of auditory-visual
stimulus onset asynchrony detection. J. Neurosci. 21, 300 – 304
21 Nimchinsky, E.A. et al. (1999) A neuronal morphologic type unique to
humans and great apes. Proc. Natl. Acad. Sci. U. S. A. 96, 5268– 5273
22 Allman, J. et al. (2004) The spindle neurons of frontoinsular cortex
(area FI) are unique to humans and african apes. Soc. Neurosci. Abstr.
725.5
23 Macphail, E.M. (1998) The Evolution of Consciousness, Oxford
University Press
24 Gaser, C. and Schlaug, G. (2003) Brain structures differ between
musicians and non-musicians. J. Neurosci. 23, 9240– 9245
25 Gundel, H. et al. (2004) Alexithymia correlates with the size of the
right anterior cingulate. Psychosom. Med. 66, 132 – 140
26 Eisenberger, N.I. et al. (2003) Does rejection hurt? An FMRI study of
social exclusion. Science 302, 290 – 299
27 Singer, T. et al. (2004) Empathy for pain involves the affective but not
sensory components of pain. Science 303, 1157 – 1162
1364-6613/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tics.2004.04.004