Gustatory neural coding in the monkey cortex: the quality of sweetness.

1. We analyzed the activity of single neurons in gustatory cortex of alert cynomolgus monkeys in response to the four basic taste stimuli and to a range of chemicals, all of which are predominantly sweet to humans. 2. We recorded taste-evoked responses from a cortical area that measured 4.0 mm in its anteroposterior extent, 5.6 mm dorsoventrally and 2.2 mm mediolaterally. Taste-responsive neurons constituted 4.7% of the 3,066 neurons tested in the course of 66 recording tracks. Nongustatory cells included those responsive to mouth movement (34.1%), tongue touch (1.9%), stimulus approach (0.7%), and tongue extension (0.5%). The functions of 58.2% of the cells we isolated could not be determined. 3. The mean breadth of tuning of these cortical taste neurons was a moderate 0.59 (range 0.00-0.93). 4. There was no evidence that taste cells with similar functional attributes were clustered in the cortex, i.e., there was no apparent topographic organization of taste qualities. 5. A taste space was generated from the correlations among patterns of neural activity evoked by the stimulus array. Within the space, NaCl was most isolated from other stimuli; the profiles elicited by HCl, quinine HCl, and water were all moderately intercorrelated and were clearly distinct from the cluster of sweet stimuli. 6. The 19 sweet chemicals formed a coherent cluster centered on the simple carbohydrates (glucose, fructose, sucrose, maltose) and sorbitol. Nearest this core were calcium cyclamate, aspartame, and cran-raspberry juice. In the next concentric ring were acesulfame potassium, xylose, xylitol, sorbose, polycose, and myoinositol. Increasingly distant from the sugars were sodium saccharin, stevioside, neohesperidin DHC, L-tryptophan and monellin. 7. We compared these results with those of a human psychophysical study of sweet stimuli. Using the position of glucose as a reference, we measured the distances to all other stimuli that were common to the two studies (n = 15). The correlation between the human psychophysical data and those derived from evoked activity in the macaque cortex was +0.82. 8. The high correlation between human psychophysical and macaque electrophysiological data implies that the subtle distinctions among stimuli that are predominantly sweet are quite similar for these two species and reinforces the value of this neural model for human taste perception.