Responses of primate cortical neurons to unitary and binary taste stimuli.

Abstract

1. The responses of 126 neurons in primary gustatory cortices of two rhesus monkeys were recorded during sapid stimulation of the tongue with 18 taste stimuli. Ten of these stimuli were dissolved in distilled water (DW): 1.0 M sucrose (Suc), 0.1 M and 0.03 M sodium chloride (NaCl), 0.003 M hydrochloric acid (HCl), 0.001 M quinine hydrochloride (QHCl), 0.03 M monosodium glutamate (MSG), 0.03 M polycose, 0.3 M glycine, 0.1 M proline, and 0.1 M malic acid. Seven other stimuli were dissolved in 0.03 M MSG; the last stimulus was a mixture of 1.0 M Suc and 0.03 M NaCl. 2. The average spontaneous rate (2.2 +/- 0.2 spikes/s, mean +/- SE) and response to DW (2.5 +/- 0.2) of these 126 neurons was low but within the range previously reported for neurons in primate taste cortex. Suc was the most effective stimulus for 24.1% of the neurons tested followed by NaCl (15.7%), QHCl (14.8%), HCl (11.1%), MSG (10.2%), and other miscellaneous unitary gustatory stimuli (8.3%). Binary taste mixtures were the most effective stimuli for 15.7% of the sample. The net responses (corrected for DW, in spikes/s) for Suc-best (3.3), NaCl-best (4.3), HCl-best (3.4), QHCl-best (2.3), and MSG-best (4.1) were sluggish, but comparable with that reported previously. 3. The response breadth of the 82 neurons that responded best to either Suc, NaCl, HCl, or QHCl measured with the entropy coefficient indicated a moderate response breadth for these neurons (mean = 0.79; range = 0.30-0.98). According to the response criteria adopted in this experiment (water response +/- 1.96 SD), however, 81 of these 82 neurons (98.1%) responded to only one or two of the four basic taste stimuli. The disparity between the entropy- and criterion-based measures of response derive from the nature of the two statistics. Adjustments that would make the entropy statistic less inclusive and the definition of a response according to statistical criteria less exclusive would increase their concordance. 4. Three multivariate statistics (cluster, principal axis factor, and multidimensional analysis) were used to analyze the data. Cluster analysis enabled us to divide the 82 taste neurons into groups on the basis of response similarity. Each of the four largest groups was dominated by neurons that responded best to one of the four basic taste stimuli: Suc, NaCl, QHCl, and HCl (ranked in descending order); the fifth largest cluster contained neurons that responded best to MSG. Principal axis factor analysis demonstrated that 80.8% of the total variance could be accounted for by three factors. Neurons responding best to Suc, NaCl, and QHCl each were closely associated with one of those three factors, but the loadings of the HCl-best neurons were evenly distributed across all three factors. The communality coefficient of these three factors was > 80% for the Suc-, NaCl-, HCl-, and QHCl-best neurons; the MSG-best neurons, by comparison, had very few high loadings on any of these three factors and a correspondingly low communality coefficient of 40.4%, a difference that was statistically significant from the other four groups. Thus the three factors related to Suc-, NaCl-, HCl-, and QHCl-best neurons are not relevant to MSG-best neurons. We used multidimensional analysis to arrange the neurons that responded best to Suc, NaCl, HCl, QHCl, and MSG into five loosely arranged and partially overlapping clusters. A multidimensional space based on stimulus similarity showed that MSG was as different from the four basic taste stimuli as they were from one another. 5. Mixture suppression, a common observation in human psychophysical experiments, was examined at the neurophysiological level by including binary tastants in the stimulus battery. The average response of 19 Suc-best neurons to 1.0 M Suc (4.1 spikes/s) decreased to near 0 when the solvent was changed from DW to either 0.03 M MSG or 0.03 M NaCl. Similar decrements were observed in NaCl- and MSG-best neurons tested with Suc/NaCl mixtures.