Primate frontal eye fields. III. Maintenance of a spatially accurate saccade signal.

1. We studied the activity of single neurons in the monkey frontal eye fields during oculomotor tasks designed to assess the activity of these neurons when there was a dissonance between the spatial location of a target and its position on the retina. 2. Neurons with presaccadic activity were first studied to determine their receptive or movement fields and to classify them as visual, visuomovement, or movement cells with the use of the criteria described previously (Bruce and Goldberg 1985). The neurons were then studied by the use of double-step tasks that dissociated the retinal coordinates of visual targets from the dimensions of saccadic eye movements necessary to acquire those targets. These tasks required that the monkeys make two successive saccades to follow two sequentially flashed targets. Because the second target disappeared before the first saccade occurred, the dimensions of the second saccade could not be based solely on the retinal coordinates of the target but also depended on the dimensions of the first saccade. We used two versions of the double-step task. In one version neither target appeared in the cell's receptive or movement field, but the second eye movement was the optimum amplitude and direction for the cell (right-EM/wrong-RF task). In the other the second stimulus appeared in the cell's receptive field, but neither eye movement was appropriate for the cell (wrong-EM/right-RF task). 3. Most frontal-eye-field cells discharged in the right-EM/wrong-RF version of the double-step task. Their discharge began after the first saccade and continued until the second saccade was made. They usually discharged even on occasional trials in which the monkey failed to make the second saccade. They discharged much less, or not at all, in the wrong-EM/right-RF version of the double-step paradigm. Thus most presaccadic cells in the frontal eye fields were tuned to the dimensions of saccadic eye movements rather than to the coordinates of retinal stimulation. 4. Eleven movement cells (including 1 which also had independent postsaccadic activity for saccades opposite its presaccadic movement field) were studied, and all had significant activity in the right-EM/wrong-RF task. 5. Almost all (28/32) visuomovement cells, including 12 with independent postsaccadic activity, discharged in the right-EM/wrong-RF task. None of the four that failed had independent postsaccadic activity. 6. The majority (26/40) of visual cells were responsive in the right-EM/wrong-RF task.(ABSTRACT TRUNCATED AT 400 WORDS)