Abstract
It has been shown that humans have the ability to take into account the consequences of hand movements in an optimal fashion. Here we studied saccadic eye movements in a rapid fixation task that provided explicit rewards and penalties for the saccade endpoint. Subjects were instructed to make a saccade towards a color coded target region as quickly as possible while trying to avoid a nearby penalty region. Saccades into the target won points; saccades into the penalty could yield a loss. Subjects' task was to win as high a score as possible. Subjects' scores were compared to optimal scores which were calculated taking into account each observer's movement endpoint variability, similar to a recent model of optimal movement planning under risk. 3 paradigms (‘gap’, ‘no gap’ and ‘overlap’) which lead to different saccadic latencies were used. Results show that scores in the ‘gap’ experiment are poor in trials with the shortest latencies (120 ms – 140 ms) but improve significantly with longer latencies. In the ‘no gap’ experiment, latencies were on average 20 ms longer, in the ‘overlap’ experiment on average 55 ms longer than in the ‘gap’ experiment. However, scores at a given latency level remained largely unaffected by the experimental paradigm. This shows that planning of the saccade does not interfere with other processes that occur before the eye starts moving, e.g. disengaging fixation. We conclude that, with additional time of about 40 ms – 100 ms for processing, saccades can be planned and executed optimally in some conditions. This is in line with the idea that processing of reward related information - which is believed to be encoded in area LIP - takes additional time. This time does not seem to be available in trials with very short latencies.
Supported by Deutsche Forschungsgemeinschaft (Emmy-Noether-Programm, Grant TR 528/1-3) to MS and JT.