Abstract
Response latency generally increases as a function of the number of stimulus-response (S-R) alternatives, a relationship described by Hick's law. However, previously we have demonstrated that visually guided saccades (prosaccades) show no increase in latency as the number of S-R alternatives increases in logarithmic steps from 1–8. In contrast, we found that the latencies of saccades directed 180° from the target (antisaccades) rise in accordance with Hick's law. We wondered whether the ability to make responses equally fast regardless of the number of S-R alternatives is unique to prosaccades, or whether such performance can be achieved in other types of saccades through learning. In this experiment, we trained 4 naive observers to make antisaccades for a minimum of 8 sessions. We collected 48 trials for each of the four S-R uncertainty levels (target-set sizes of 1,2,4,8) in every session. In the control condition, we likewise measured response latencies for prosaccades and saccades directed 90° counterclockwise from the target at the start and end of the study. We calculated the slope of the response latencies across S-R uncertainty levels for each session. While the slopes for the control conditions remained constant for the pre- and post-test sessions, the antisaccade latency slopes decreased and approached zero over the course of training. We conclude that prosaccades are not the only type of saccades that be can made with no increase in latency irrespective of S-R uncertainty, as antisaccades can approach and possibly equal this level of performance with sufficient training. This finding suggests plasticity of the stimulus-response pathways involved in the response selection of antisaccades.