A lot of work on neural computation just looks at understanding how animals process information coming from a single sense, but of course our ears don't shut off when we look at something, and it's well-established on a behavioral level that you can get more of a response with a noise and a flash than with either alone. Here is a careful paper looking at the neural basis of such multisensory integration in cats.
They start with individual spike-trains, and use those to build post-stimulus time histograms for the spiking rate, but mostly they on the number of spikes evoked in the first 50 ms after the stimulus, which assumes a kind of rate coding. (No attempt at working out a temporal code is made here.) They do, however, do a pretty thorough job of analyzing the rates, including a nice null model of additivity (using the rastergrams), and corrections for spontaneous activity. The abstract does a pretty good job of summarizing the results of this analysis. They use them to suggest a simple and attract model, but dash those hopes a few paragraphs later.
That most multisensory interactions were consistent with simple summation of modality-specific influences suggests that a very basic linear model of the SC might account for many of the current observations. For example, the fact that superadditive interactions were common only in response to combinations of very weak modality-specific stimuli suggests that such superadditivity reflects temporal summation of the [post-synaptic potentials] consequent to near-threshold activity on the auditory and visual input channels. Subadditive interactions were relatively uncommon; however, their correspondence with combinations of the most effective stimuli could likewise be the simple consequence of approaching an intrinsic firing frequency limit of the SC neuron. In principle, a qualitatively similar relationship between computational mode and stimulus efficacy could be produced by an integrate-and-fire model that includes threshold and saturating nonlinearities. [p. 6506]
Alas, as they say, such a nice near-linear model has two fatal flaws. First, it doesn't account for the variability among neurons, and the fact that "few displayed the transition in its entirety from superadditivity to subadditivity". Second, the SC gets input from the cortex as well as from the sense organs, and while blocking cortical input leaves modality-specific responses alone, it eliminates multi-modal enhancement, which is not compatible with the near-linear model.
As you may have guessed, there is no feline cuteness to the paper at all, though I am happy to report that "[a]t the end of an experiment, injection of anesthetics and paralytics was terminated, and animal [sic] was allowed to recover normal respiration and locomotion before being returned to its home cage" (p. 6500).
Posted at July 22, 2005 13:32 | permanent link