Gustatory neural coding in the monkey cortex: mixtures.
1. Psychophysicists have shown that the intensity and quality of a taste stimulus, as perceived by humans, is modified by including that stimulus in a mixture. Gustatory neurons in the primary taste cortex (anterior insula and frontal operculum) of the cynomolgus macaque are involved with the coding of stimulus intensity and quality, and so should reflect the impact of these stimulus interactions. 2. We recorded the activity of 48 neurons in primary taste cortex in response to the oral application of each of the four basic stimuli, their six possible dyads, the four triads, and the tetrad of all four. Stimuli were maintained at a constant intensity in all mixtures by increasing their concentrations as the number of components rose. 3. Glucose was the most effective basic stimulus, followed by quinine HCl, NaCl, and HCl. The mean response to dyads was suppressed by 50% from the sum of responses to the two unmixed components. The response to triads was 62% lower than the sum of responses to their three components, and activity evoked by the tetrad was suppressed by 74% from the sum of all four individual responses. Therefore there was nearly total suppression in the sense that the responses to the mixtures were approximately 1/2, 1/3, and 1/4 the sums of responses to two, three, and four components, respectively. 4. Neurons could be divided into four subtypes: those that responded best to each of the basic stimuli. All subtypes except HCl cells were about equally suppressed when their preferred stimulus was included in a mixture. HCl was a particularly ineffective stimulus, such that this subtype responded poorly and so was less susceptible to mixture suppression. 5. Taste quality, as indexed by correlation coefficients among profiles of activity, was quite predictable for dyads. If the mixture included HCl, the profile it generated correlated poorly (about +0.20) with that of HCl and rather well (about +0.60) with that of the other component. If HCl was not included, the mixture's profile correlated about +0.40 with that of each component. 6. The profile generated by the mixture of three stimuli was predictable only if one of the components was HCl. In that case, the triad elicited a profile midway between those of the other two components, i.e., the contribution of HCl was largely ignored. When HCl was not involved, or when all four basic stimuli were combined, the resulting profiles were poorly correlated with those of all basic stimuli. 7. The contribution made by each basic taste to human perception and to the macaque's neurophysiological response was compared for all mixtures. The contribution was often quite similar for human and macaque, but when differences occurred, they were typically due to lower activity from HCl cells in the macaque, a loss that was replaced mainly by larger responses from glucose neurons. 8. The magnitude of responses to mixtures in the macaque taste cortex matches well with expectations from human psychophysical studies. The presumed quality of the response to mixtures is also similar, except that HCl is less effective in monkeys and sugars more so.