We are far from ignorant about some of the neural mechanisms that underlie the integration of information from different sensory modalities. The study of single neurons that receive inputs from more than a single sensory modality has produced a growing body of information about the sites in the nervous system in which cross-modal convergence takes place as well as some of the mechanisms by which information from different sensory modalities is brought together and integrated. The manner in which these multisensory neurons deal with their converging inputs and the consequences of their action are among the subjects of this article. However, before dealing with this issue, it is important to note that although we often tend to frame questions about the phenomenon of multisensory integration in terms of human experience (especially when considering its perceptual consequences), its appearance did not await the evolution of Homo sapiens. Rather, it is an ancient scheme of information processing that is likely to have been present in the presumptive primordial unicellular organism. The caveat "presumptive" is used here because no one can say with certainty what that unicellular organism was like. Nevertheless, it is reasonable to suppose that it, like its modern counterparts, had different sensory receptors tuned to the transduction of chemical, mechanical, and/or photic energy, and that each of these sensory transducers used the same signaling mechanism (an electrical current produced by the movement of ions across the cell's membrane). Consequently, the primordial unicellular organism, like its extant cousins, was a multisensory organism. Because the various currents produced by activating its different sensory receptors can readily affect one another, unicellular organisms are obligatory multisensory integrators. Simple multicellular organisms are also unable to sequester specific sensory signals because of the nature of their intercellular connections, and the segregation of at least some modality-specific information probably did not occur until the appearance of advanced invertebrates. Thus, multisensory integration, rather than sensory segregation, is likely to have been the initial form of sensory information processing.
The capacity for multisensory integration was retained in some form throughout the evolution of multicellular organisms, and there is no known beast, vertebrate or invertebrate, in which there exists a complete segregation of sensory signals on a sense-by-sense basis. Presumably, the ability to derive information from the combined action of different senses that would be unavailable from their individual action had considerable survival value and thereby engendered the maintenance and elaboration of systems capable of multisensory synthesis. However, during the evolution of complex species, an interesting duality was formed;
some regions of the nervous system became specialized for processing information based on its modality and others for processing information regardless of the modality from which it was obtained. Presumably, it is the action of the former systems that accounts for the sensory qualia referred to previously. The preeminent example of such a system is the primary projection pathway (but see Section VII for evidence that cross-modal influences are evident even in the primary projection systems). In the vertebrate visual system, for example, this involves the projections from the retina to the primary thalamic relay station (the lateral geniculate nucleus) and from there to primary visual cortex. Multisensory areas exist outside the primary projection pathways and have been found at every level of the neuraxis.
In light of the importance of multisensory information processing for normal perception and behavior, it is surprising to note that we are only beginning to understand the neural bases by which the brain accomplishes this feat. When compared to our understanding of how modality-specific sensory cues are processed, our knowledge of multisensory integration seems rudimentary. However, we do know that whenever different modality-specific inputs converge at the level of the single neuron (Fig. 2) there is the opportunity to synthesize them.
Was this article helpful?