Visuospatial coding in primate prefrontal neurons revealed by oculomotor paradigms.

1. Visual responses and their relationship to delay-period activity were studied by recording single neuron activity from the prefrontal cortex of rhesus monkeys while they performed an oculomotor delayed-response (ODR) and a visual probe (VP) task. In the ODR task, the monkey was required to maintain fixation of a central spot of light throughout the cue (0.5 s) and delay (3 s) periods and then make a saccadic eye movement to one of four or eight locations where the visual cue had been presented. In the VP task, the same visual stimuli that were used in the ODR task were presented for 0.5 s, but no response was required. The VP task was thus employed to test the passive visual response and, by comparison with cue-elicited activity in the ODR task, to examine the degree of behavioral enhancement present in prefrontal visual activity. 2. Among 434 neurons recorded from the prefrontal cortex within and surrounding the principal sulcus (PS), 261 had task-related activity during at least one phase of the ODR task, and 74 of these had phasic visual responses to the onset of the visual cues with a median latency of 116 ms. The visual responses of 69 neurons were excitatory, and 5 neurons were inhibited. Five of the neurons with excitatory visual responses also responded transiently after the offset of the cue. 3. Visual responses were classified as directional for 71 PS neurons (96%) in that excitatory or inhibitory responses occurred only for location of cues in a restricted portion of the visual field. Only 3 PS neurons were omnidirectional, i.e., responded equivalently to cues in all locations tested. 4. The best direction and tuning specificity of all PS neurons with directional visual responses were estimated from parameters yielding the best fit to a Gaussian-shaped tuning function. The best direction for the majority (71%) of neurons was toward the visual field contralateral to the hemisphere where the neuron was located. The remaining neurons had their best directions in the ipsilateral field (18%) or along the vertical meridian (11%). 5. The specificity of directional tuning for PS visual responses was quite variable, ranging from neurons that responded only to one of the eight cue locations to neurons that responded to all eight, but in a clearly graded fashion. The standard deviation parameter of the Gaussian curve indexed the breadth of directional tuning of each neuron; its median value was 37 degrees.(ABSTRACT TRUNCATED AT 400 WORDS)