RESPONSES OF CELLS IN THE TAIL OF THE CAUDATE-NUCLEUS DURING VISUAL-DISCRIMINATION LEARNING

1. The tail of the caudate nucleus and adjacent ventral putamen (ventr ocaudal neostriatum) are major projection sites of the extra-striate v isual cortex. Visual information is then relayed, directly or indirect ly, to a variety of structures with motor functions. To test for a rol e of the ventrocaudal neostriatum in stimulus-response association lea rning, or habit formation, neuronal responses were recorded while monk eys performed a visual discrimination task. Additional data were colle cted from cells in cortical area TF, which serve as a comparison and c ontrol for the caudate data. 2. Two monkeys were trained to perform an asymmetrically reinforced go-no go visual discrimination. The stimuli were complex colored patterns, randomly assigned to be either positiv e or negative. The monkey was rewarded with juice for releasing a bar when a positive stimulus was presented, whereas a negative stimulus si gnaled that no reward was available and that the monkey should withhol d its response. Neuronal responses were recorded both while the monkey performed the task with previously learned stimuli and while it learn ed the task with new stimuli. In some cases, responses were recorded d uring reversal learning. 3. There was no evidence that cells in the ve ntrocaudal neostriatum were influenced by the reward contingencies of the task. Cells did not fire preferentially to the onset of either pos itive or negative stimuli; neither did cells fire in response to the r eward itself or in association with the motor response of the monkey. Only visual responses were apparent. 4. The visual properties of cells in these structures resembled those of cells in some of the cortical areas projecting to them. Most cells responded selectively to differen t visual stimuli. The degree of stimulus selectivity was assessed with discriminant analysis and was found to be quantitatively similar to t hat of inferior temporal cells tested with similar stimuli. Likewise, like inferior temporal cells, many cells in the ventrocaudal neostriat um had large, bilateral receptive fields. Some cells had ''doughnut''- shaped receptive fields, with stronger responses in the periphery of b oth visual fields than at the fovea, similar to the fields of some cel ls in the superior temporal polysensory area. Although the absence of task-specific responses argues that ventrocaudal neostriatal cells are not themselves the mediators of visual learning in the task employed, their cortical-like visual properties suggest that they might relay v isual information important for visuomotor plasticity in other structu res. 5. Although there was no evidence of neuronal responses related s pecifically to the learning of stimulus-response associations in the t ask, there was a tendency for novel stimuli to elicit stronger neurona l responses than familiar ones. This demonstrates that some form of le arning is reflected in the neuronal responses of these neostriatal cel ls. 6. Cells recorded in cortical area TF were also visually responsiv e, and many responded selectively to different visual stimuli. In cont rast to the ventrocaudal neostriatum, however, responses of a subpopul ation of cells in area TF did appear to be influenced by the reward co ntingencies of the task. That is, some cells responded either preferen tially, or even exclusively, to the positive stimuli in the task, a pr eference that developed within a few trials as the animal learned to d iscriminate new pairs. These neuronal responses to positive stimuli we re not directly related to the motor response of the animal, because t hey were not present on error trials when the animal responded to the negative stimulus; nor were they directly related to the delivery of t he reward, because they were time locked to the onset of the visual st imulus rather than to the reward. The results suggest that some of the mechanisms underlying the learning of stimulus-response associations may be located in the cortex itself.