Effects of attention on MT and MST neuronal activity during pursuit initiation

The responses of neurons in monkey extrastriate areas MT (middle temporal) and MST (medial superior temporal), and the initial metrics of saccadic and pursuit eye movements, have previously been shown to be better predicted by vector averaging or winner-take-all models depending on the stimulus conditions. To investigate the potential influences of attention on the neuronal activity, we measured the responses of single MT and MST neurons under identical stimulus conditions when one of two moving stimuli was the target for a pursuit eye movement. We found the greatest attentional modulation across neurons when two stimuli moved through the receptive field (RF) of the neuron and the stimulus motion was initiated at least 450 ms before reaching the center of the RF. These conditions were the same as those in which a winner-take-all model better predicted both the eye movements and the underlying neuronal activity. The modulation was almost always an increase of activity, and it was about equally frequent in MT and MST. A modulation of >50% was observed in approximately 41% of MT neurons and 27% of MST neurons. Responses to all directions of motion were modulated so that the direction tuning curves in the attended and unattended conditions were similar. Changes in the background activity with target selection were small and unlikely to account for the observed attentional modulation. In contrast, there was little change in the neuronal response with attention when the stimulus reached the RF center 150 ms after motion onset, which was also the condition in which the vector average model better predicted the initial eye movements and the activity of the neurons. These results are consistent with a competition model of attention in which top-down attention acts on the activity of one of two competing populations of neurons activated by the bottom-up input from peripheral stimuli. They suggest that there is a minimal separation of the populations necessary before attention can act on one population, similar to that required to produce a winner-take-all mode of behavior in pursuit initiation. The present experiments also suggest that it takes several hundred milliseconds to develop this top-down attention effect.