Mapping pain in the brain
A D (Bud) Craig
Many areas of the brain are involved in the experience of pain. These areas have been thought to form a distributed pain-processing 'neuromatrix' centred on the portions of the cerebral cortex related to the sense of touch. However, a new view suggests that specific pain centres exist, which have evolved from a primitive system of the brain that controls the health of the body, or body physiology. The overlap between these areas and emotion-processing regions of the brain could explain the peculiarly human subjective qualities of pain.
When we stub our toe, it hurts but only because our brain says so. Damage-detecting sensory neurons flash a message to the spinal cord, spinal cord neurons relay the message to the brain, and the brain decides (a) damage has occurred, (b) it has been inflicted on the toe, and (c) something needs to be done (we start hobbling, raise the foot, utter an expletive). It may feel as if our toe is throbbing, but the experience is all contained within a mental projection of the condition of our toe within our brain. But exactly which parts of our brain achieve all this, and how do they manage it?
Brain mapping techniques in humans and animals have provided many insights into the brain regions that represent the sensation of pain. Results from anatomical and physiological experiments in experimental animals, and from experimental, clinical and functional imaging (fMRI, PET) studies in humans, have identified more clearly which parts of the brain and nervous system are involved. They have also led to new concepts that challenge existing views and hence this area is notable for its controversy and enthusiastic debates.
In the consensus model that has held sway for the past 40 years, pain is thought of as an aspect of the sense of touch that can become sensitized in pathological states and can affect our emotional responses. In this view, neurons originating in the deep dorsal horn of the spinal cord, which are activated by convergent input from sensory neurons responsive to many signals (touch, pressure, pinch, cold and heat from all tissues), send messages (via the thalamus) to the specialized areas of the cortex that represent touch (the somatosensory cortex). This region is held to be the core of a 'neuromatrix' distributed over the entire forebrain that deciphers pain from the pattern of activity across all of the convergent neurons.
This convergence model is thought to explain mysterious clinical phenomena, such as referred pain (pain felt remote from the site of tissue damage), allodynia (pain abnormally elicited by light touch or cooling), hyperalgesia (excessive pain felt over wide areas), and ultimately psychological aspects such as hyperpathia (abnormally strong emotional reactions to pain), which in this view are all explained in terms of sensitization, 'crossed wires' or aberrant central processing of multiple inputs. Neuronal recordings from convergent neurons at several levels seem to support this view, as does the confusingly broad array of forebrain regions implicated by some studies of pain in humans.
Further, this model has difficulty explaining how such a convergent system can represent the distinctly different feelings of pain elicited by sharp points, burning heat, biting cold, aching muscles or cramping viscera.
New experimental approaches are also producing results inconsistent with the classical model but together may be suggesting a viable alternative.
Studies in rodents offer further support, as chemical lesion of lamina I cells can selectively reduce pain-like behaviours. Other lamina I neurons uniquely encode several other distinct sensations, such as warm, cool, itch, muscle burn, and sensual touch, so it is easy to see why these sensations are all lost when the ascending pathway that is crucial for pain is lesioned. Altered lamina I responses can also readily explain pathological pain sensations, the result of inappropriate triggering of these pain-specific spinal neurons.
In addition, functional imaging studies in humans, and maps of electrical responses to laser-induced pain, indicate strong activation of the parieto-insular cortex during pain. Further, a lesion of this region can strongly reduce pain. This region is uniquely activated during cooling stimulation, muscle sensations, sensual touch and other feelings from the body as well, whereas the somatosensory cortex is not. Nevertheless, some studies report activation of the somatosensory cortex during painful stimulation, which seems to support the consensus view.
There is also strong activation during pain at a second site in the medial frontal cortex, the caudal part of the anterior cingulate, an area implicated in controlling our so-called motivational behaviours (how we act to meet our 'needs'). This activation is directly related to perceptions of the unpleasantness of pain, and is accompanied by activation in several sub-cortical sites, such as the amygdala, cerebellum and striatum. Most authors interpret these multiple sites as an interconnected network with distributed functions.
A possible synthesis
The VMpo and its parieto-insular cortical target (the interoceptive cortex) are active in response to all sensations emanating from within the body related to its health. They can thus be viewed as a internal representation of the physiologic status of the body. This fits with the intimate association of these regions and lamina I neurons with homeostasis (the maintenance of the body's physiologic status).
Further, the interoceptive cortex exists only in primates and is enormously enlarged in humans. These findings suggest an entirely new perspective: pain in humans is represented in a phylogenetically novel cortical extension of an ancient homeostatic system.
Pain can then be seen as a specific homeostatic response consisting of a distinct 'physical' sensation (represented in the parieto-insular cortex) and an 'emotional' component (represented in the anterior cingulate). This view can readily explain the effects on pain of stimulation or ablation of these particular areas, and the interactions between pain and autonomic activity and other feelings from the body (including sensual touch!).
Notably, in this view, relatively simple animals display pain-like behaviours that represent the integrated output of homeostatic control regions in the brainstem that is, behaviors based on survival needs but they do not have a cortical image of sensations from the body.
Most dramatically, a re-mapping of the interoceptive cortex in the right anterior insula of humans seems to provide a basis for emotional awareness of the 'material me'; so this new view also offers a substantive explanation for the emotional interactions of pain (e.g. psychosomatic pain) and its variability, as well as an anatomical basis for a widely held hypothesis that self-awareness (or consciousness) is based on a mental image of the homeostatic condition of the body (that is, of 'how you feel').
Testing the model
Studies of the reduction of pain by context and suggestion (the placebo effect) offer a fascinating view of the brain modulating its own activity (i.e. the effects of belief). Interactions of pain with emotion and other feelings from the body can be related directly to particular neuroanatomical structures. Improvements in high-resolution scanning will someday afford the opportunity to identify differences in the brains of individual patients.
The coming decade will certainly be an exciting period in the study of pain!
A D (Bud) Craig is Atkinson Research Scientist in the Division of Neurosurgery, Barrow Neurological Institute, Phoenix, AZ 85013, USA. E-mail: BCraig@chw.edu.
Craig A D (2002) How do you feel? Interoception: the sense of the physiological condition of the body. Nat. Rev. Neurosci. 3: 655-666.
Petrovic P, Kalso E, Petersson K M, Ingvar M (2002) Placebo and opioid analgesia - imaging a shared neuronal network. Science 295: 1737-1740.
Schlereth T, Baumgartner U, Magerl W, Stoeter P, Treede R D (2003) Left-hemisphere dominance in early nociceptive processing in the human parasylvian cortex. Neuroimage 20: 441-454.
Neuromatrix model vs. Homeostatic model
Click to enlarge