Single neuron studies provide one foundation for understanding many facets of multisensory integration. These studies have used a variety of criteria for identifying and quantifying multisensory integration. While a number of techniques have been used, there lacks an explicit discussion of the assumptions, criteria, and analytical methods traditionally used to define the principles of multisensory integration. This was not problematic when the field was small, but with rapid growth a number of alternative techniques and models have been introduced, each with its own criteria and sets of implicit assumptions to define and characterize what is thought to be the same phenomenon. The potential for misconception prompted this reexamination of traditional approaches in order to clarify their underlying assumptions and analytic techniques. The objective is to review the traditional quantitative methods advanced in the study of single-neuron physiology in order to appreciate the process of multisensory integration and its impact.

Multisensory neurons and their ability to integrate multisensory cues develop gradually in the midbrain [i.e., superior colliculus (SC)]. To examine the possibility that early sensory experiences might play a critical role in these maturational processes, animals were raised in the absence of visual cues. As adults, the SC of these animals were found to contain many multisensory neurons, the large majority of which were visually responsive. Although these neurons responded robustly to each of their cross-modal inputs when presented individually, they were incapable of synthesizing this information. These observations suggest that visual experiences are critical for the SC to develop the ability to integrate multisensory information and lead to the prediction that, in the absence of such experience, animals will be compromised in their sensitivity to cross-modal events.

Many neurons in the superior colliculus (SC) integrate sensory information from multiple modalities, giving rise to significant response enhancements. Although enhanced multisensory responses have been shown to depend on the spatial and temporal relationships of the stimuli as well as on their relative effectiveness, these factors alone do not appear sufficient to account for the substantial heterogeneity in the magnitude of the multisensory products that have been observed. Toward this end, the present experiments have revealed that there are substantial differences in the operations used by different multisensory SC neurons to integrate their cross-modal inputs, suggesting that intrinsic differences in these neurons may also play an important deterministic role in multisensory integration. In addition, the integrative operation employed by a given neuron was found to be well correlated with the neuron's dynamic range. In total, four categories of SC neurons were identified based on how their multisensory responses changed relative to the predicted addition of the two unisensory inputs as stimulus effectiveness was altered. Despite the presence of these categories, a general rule was that the most robust multisensory enhancements were seen with combinations of the least effective unisensory stimuli. Together, these results provide a better quantitative picture of the integrative operations performed by multisensory SC neurons and suggest mechanistic differences in the way in which these neurons synthesize cross-modal information.

A growing number of brain imaging studies are being undertaken in order to better understand the contributions of multisensory processes to human behavior and perception. Many of these studies are designed on the basis of the physiological findings from single neurons in animal models, which have shown that multisensory neurons have the capacity for integrating their different sensory inputs and give rise to a product that differs significantly from either of the unisensory responses. At certain points these multisensory interactions can be superadditive, resulting in a neural response that exceeds the sum of the unisensory responses. Because of the difficulties inherent in interpreting the results of imaging large neuronal populations, superadditivity has been put forth as a stringent criterion for identifying potential sites of multisensory integration. In the present manuscript we discuss issues related to using the superadditive model in human brain imaging studies, focusing on population responses to multisensory stimuli and the relationship between single neuron measures and functional brain imaging measures. We suggest that the results of brain imaging studies be interpreted with caution in regards to multisensory integration. Future directions for imaging multisensory integration are discussed in light of the ideas presented.