mercator would have delighted in the “modern phrenology,” where detailed maps of motor, sensory, and other functions are plotted over the surface of the human cerebral cortex. Moreover, within each region dealing with a sensory function there are finer maps that code for the spatial origin of the sensory stimulus. Thus there is an orderly map of visual space within the primary visual cortex and an orderly map of the body surface in the primary somatosensory cortex (the familiar sensory “homunculus”). As is well known, these maps devote disproportionately more cortical surface area to regions with the greatest density of sensory input (the fovea for vision, the face and hands for touch). These cortical representations of sensory modalities tell us the nature and strength of what is happening to us, but it is generally understood that these inputs need to be organized in maps to tell us where in our body this is happening. These maps apply not only to the external body surface but extend to include (with admittedly coarser spatial resolution) the internal aspects of the body (esophagus, stomach, airways, bladder, etc.). Although we now have quite a good knowledge of how the cerebral cortex is arranged to make sense of signals from the external world (exteroception), more recent findings are expanding our views on how the cortex gets to know about the body itself (interoception). A key player in this appears to be the insula (1, 5).
The insula (the island of Reil) is an anatomical extension of the surface of the parietal lobe of the cortex after it folds in via the Sylvian fissure, but it is phylogenetically older. Both the parietal cortex and the opercular cortex, which lies between it and the insula, contain somatosensory maps. These include the primary and secondary somatosensory cortices, which are now recognized as comprising several submaps of different sensory modalities. A paper by Hua et al. (8) in this issue of the American Journal of Physiology-Regulatory, Integrative and Comparative Physiology now provides clear evidence for a topographic representation of nonnoxious cooling of the skin, on the basis of functional magnetic resonance imaging (fMRI). Interestingly, this map is located in the dorsal posterior insula (dpIns) rather than in the parietal cortex, and it maps the upper vs. lower parts of the body in an anterior-posterior direction. This confirms and extends a previous discovery from the same laboratory that temperature sensation maps to the insula (3).
The literature on the functions represented in the insula tells several stories and the insula itself probably comprises several sensory maps. First, there is the gustatory cortex, where taste sensations from the contralateral surface of the tongue are mapped onto a region over the rostral insular and opercular cortices (10). Second, there is the visceral sensory cortex, in which the alimentary tract is represented, with at least some topographic detail, over a significant portion of the insula. In locally anesthetized but conscious subjects, electrical stimulation there commonly produced visceral sensations such as nausea or “a sinking feeling in the abdomen” (9). Other parts of this region also include the cortical representation of visceral afferents such as arterial baroreceptors, whose influence presumably does not directly reach consciousness (10). For regions other than the uppermost parts of the alimentary tract, this viscerosensory cortical area may take the form more of an interoceptive “modality map” than of an accurate spatial map of the stimulus origin (10). Third, and most complex, there is a region of the right anterior insula that is considered to provide a representation of the internal state of the body (“how do you feel?”) (1, 5). It responds to a wide variety of stimuli and bodily states that relate to homeostasis and the sense of well-being. Examples are thirst, air hunger, temperature, the urge to micturate and the sense of muscular effort (1). There is also evidence that this region is involved in the perception of the body's own autonomic responses, such as awareness of the heartbeat (5). This region may be a primate specialty, is best developed in humans, and is largely lateralized to the nondominant hemisphere (1).
A fourth sensory representation in the insula is the main focus of the current paper by Hua et al (8). This region occupies the dorsal posterior region of the insula (dpIns). It receives temperature (and pain) information from the contralateral side of the body. The critical new information provided by the human fMRI study of Hua et al. is that the dpIns representation of temperature is topographically organized, providing the substrate for localization, as well as discrimination of innocuous temperature (8). As proof of this point, Hua et al. showed that cooling the neck caused activation in an area significantly anterior to that activated by cooling the hand (8). In combination with previous work (3), this assembles a convincing case for the dpIns being a primary cortical sensory area for temperature sensation. This is a satisfying result because it accounts for the clinical finding that parietal cortex lesions may profoundly disable tactile discrimination from the affected side of the body, yet leave temperature discrimination and localization intact.
Two other brain regions that were activated by innocuous cooling of the hand deserve comment. The right anterior insula was activated in both the present and previous imaging studies of cooling (3, 7, 8). This may be interpreted as reflecting the interoceptive significance of the cool stimulus (see comments above). A further brain region activated by hand cooling in this study was the dorsal medial cortex, in a locus somewhat above the anterior cingulate cortex. This was suggested to reflect homeostatic motivation associated with thermoregulation but at a level insufficient to activate the cingulate cortex itself (8). Interestingly, warming and cooling a much wider body surface area has been reported by others to cause extensive activation of the anterior cingulate cortex (7).
The general significance of these findings is that they place the cortical receiving area for innocuous temperature signals in the insula, which is considered to be part of the limbic cortex. This and other findings have been integrated into a general scheme of the brain processing “interoceptive” afferent signals as homeostatic emotions (1, 2, 6). The relevant afferent signals include those that are processed by lamina 1 of the spinal dorsal horn and ascend via the spinothalamic tract (including temperature, pain, itch, and other skin sensations, as well as the sense of muscular effort), and also visceral afferent signals that ascend via the nucleus tractus solitarii and parabrachial complex. Additionally, “visceral” sensations of other origin such as air hunger, thirst, and a full bladder contribute (1). In this scheme, the insula represents the limbic sensory cortex, while the motivations and actions that are generated by homeostatic emotions are processed by the “limbic motor cortex,” including the anterior cingulate gyrus (4). These pathways thus provide the neural substrates for higher-level homeostatic reflexes where emotional responses complement and refine the more automatic neuroendocrine and autonomic reflexes integrated by subcortical regions.
- Copyright © 2005 the American Physiological Society