Perceptual illusions provide clues to excitatory–inhibitory balance in migraine neocortex

The well known instances of reversing figures (e.g. the Necker cube, Figure 1) have intrigued scientists and philosophers for generations. These images provide a route to observing the brain’s intrinsic competitive operations, mechanisms probably involved in choice and decision-making at many neural levels. At lower levels in the visual pathway, this can involve rivalry between the two eyes when their inputs are conflicting in ways that cannot be reconciled (e.g. vertical lines in one eye, horizontal lines in the other); at higher cortical levels it entails rivalry between two or more perceptual organizations or interpretations as in the Necker cube, the face/vase illusion and the young lady/old woman illusion. Measurements of the rate at which these perceptual switches occur and the distribution of the intervals between switches has provided the basis for neural network models of the possible underlying brain mechanisms. While varied in detail, these models generally involve some form of reciprocal inhibition between the two organizations coupled with some component of adaptation that leads to the gradual weakening of inhibitory domination of one organization over the other (1). OPEN IN VIEWER

The relevance of these phenomena to the study of migraine is that they provide one route to examining possible intrinsic differences in the balance between excitation and inhibition in the human neocortex, an issue that has attracted considerable attention in migraine. Abnormally weak inhibition, which has been suggested to occur in migraine (2), should lead to abnormally rapid reversals. In the first attempt to examine this, using binocular rivalry, reversal rates were marginally slower than normal, ruling out any loss of inhibition at least at this level (3). Other neuromodulatory influences on this basic switching network could also affect the reversal rate. The observation that patients with clinical anxiety show more rapid than normal reversal rates has been interpreted as supporting an abnormal modulatory role for serotonin (4).

McKendrick et al. (5) have applied this experimental approach in a very ingenious fashion to ask two questions. The first is whether the trend toward slower oscillations seen in binocular rivalry (presumably occurring in the striate cortex (V1)) is also present and perhaps even stronger in perceptual tasks known to depend on extrastriate regions higher up the visual cortical processing hierarchy. Secondly, they ask whether similar effects can be seen in a second sensory system – audition. In this, their work was motivated by the observation that electrophysiological abnormalities following habituation occur not only in vision, but also in audition and somatosensation (6).

The two tasks chosen by McKendrick et al. (5) are well known in the visual and auditory fields, respectively, but have rarely been examined in the same experimental subjects. The visual task involves two sets of overlaid dots moving in different directions (upwards to the left and upwards to the right). The two percepts one can experience are either of two transparent sheets of dots, one sliding over the other, moving in different directions, or alternatively, the whole field of dots appearing to be glued together and moving as one along a new trajectory (in this case directly upwards) that is different from the trajectories of either of the dot sets. The extraction of global motion direction is a well established property of the middle temporal (MT/MST) complex (7).

The auditory task is analogous, and was introduced by Bregman to study auditory stream segregation (8). A set of alternating high and low tones (high-low-high; high-low high) are heard either as a single stream with a galloping rhythm or as two separate streams of low notes and high notes playing in parallel. In both tasks, participants track the alternations between the two organizations by pressing one of two buttons depending on which organization is currently experienced.

The results in the present paper (5) are focused largely on length of the interval to the first reversal; however, similar results are seen for mean reversal time over the entire 30 s tested. As found earlier for binocular rivalry (3), intervals between these perceptual switches were longer in migraineurs than in controls and, in this case, a correlation was found between headache frequency and interval to first reversal: those with most frequent headaches show the greatest slowing in perceptual reversal, independent of their classification as migraine with or without aura. Furthermore, the visual and auditory results show a significant positive correlation, lending stronger credence to the argument that this represents a global cortical abnormality affecting multiple systems similarly within an individual.

The authors’ preferred explanation for their findings (5) is that this reflects reduced levels of serotonin leading to lower cortical pre-activation interictally, a concept first introduced in the migraine field by Schoenen and his collaborators (6,9). While this is an appealing notion, we are far from having a detailed mechanistic understanding of the underlying processes involved. ‘Lower pre-activation’ is a term that has not been well defined in the literature. It could mean a cortex that is maintained under an enhanced level of intracortical inhibition. It could reflect a cortex with reduced glutamate stores. Or it could mean a cortex, the excitability of which has been down-regulated in some way by one of the many neuromodulators acting upon it, the role suggested for serotonin (6).

Serotonin has widespread and often apparently antagonistic effects in the nervous system. For example, in the lateral geniculate nucleus (the visual way-station in the thalamus), at least three synaptic roles for serotonin (5HT) have been described (10,11), and the picture is similar in the cortex, where location, density and type of 5HT receptors vary across cytoarchitectural areas. A recent study by Watakabe et al. (12) has provided some fascinating insight into the complementary roles played by two 5HT receptor types in cortical area V1; the modulatory effect of serotonin is activity-dependent at both receptors but in opposite directions. Findings of this sort will provide a basis on which more detailed models may be built. A caveat is that most of our knowledge of serotonin’s actions at the cortical level of the visual pathway is restricted to V1, whereas the phenomena described in the present work (5) certainly have their bases at a level beyond the primary visual and auditory cortices.

As pointed out by the authors (5), one must also remember that the neocortex is innervated by multiple modulatory systems, and serotonin need not be the critical system at play in slowing perceptual switching in the case of migraine. Histamine, for example, exerts exert widespread effects across all levels of the visual cortical hierarchy (13).

Another possibility was raised in a recent study using an ambiguous visual stimulus somewhat similar to McKendrick et al.’s (5) visual motion stimulus: perceptual switching may be mediated by a separate region of neocortex within the parietal lobe (14). Whether correlations between switching rate and parietal lobe structure will also be found to apply to auditory stimuli remains to be tested. The point is that a single cortical locus could act as the oscillator that modulates the activity of numerous cortical sensory regions through feedback pathways.

In any event, the study by McKendrick et al. (5) provides a valuable new tool through which behavioral variation in migraine may eventually be linked to electrophysiological and/or structural changes in the brain.