Role of arcuate frontal cortex of monkeys in smooth pursuit eye movements. II. Relation to vector averaging pursuit.

When monkeys view two targets moving in different directions and are given no cues about which to track, the initiation of smooth pursuit is a vector average of the response evoked by each target singly. In the present experiments, double-target stimuli consisted of two identical targets moving in opposite directions along the preferred axis of pursuit for the neuron under study for 200 ms, followed by the continued motion for 800 ms of one target chosen randomly. Among the neurons that showed directional modulation during pursuit, recordings revealed three groups. The majority (32/60) showed responses that were intermediate to, and statistically different from, the responses to either target presented alone. Another large group (22/60) showed activity that was statistically indistinguishable from the response to the target moving in the preferred (n = 15) or opposite (n = 7) direction of the neuron under study. The minority (6/60) showed statistically higher firing during averaging pursuit than for either target presented singly. We conclude that many pursuit-related neurons in the frontal pursuit area (FPA) carry signals related to the motor output during averaging pursuit, while others encode the motion of one target or the other. Microstimulation with 200-ms trains of pulses at 50 microA while monkeys performed the same double-target tasks biased the averaging eye velocity in the direction of evoked eye movements during fixation. The effect of stimulation was compared with the predictions of three different models that placed the site of vector averaging upstream from, at, or downstream from the sites where the FPA regulates the gain of pursuit. The data were most consistent with a site for pursuit averaging downstream from the gain control, both for double-target stimuli that presented motion in opposite directions and in orthogonal directions. Thus the recording and stimulation data suggest that the FPA is both downstream and upstream from the sites of vector averaging. We resolve this paradox by suggesting that the site of averaging is really downstream from the site of gain control, while feedback of the eye velocity command from the brain stem and/or cerebellum is responsible for the firing of FPA neurons in relation to the averaged eye velocity. We suggest that eye velocity feedback allows FPA neurons to continue firing during accurate tracking, when image motion is small, and that the persistent output from the FPA is necessary to keep the internal gain of pursuit high and permit accurate pursuit.