Abstract
Saccadic latencies are conventionally viewed as reflecting the accumulation of information during decision-making process but we have previously shown that latency distributions may be strongly affected by reinforcement contingencies (Madelain et al., 2007). Here, we probe the possibility to control saccadic latencies in a choice paradigm.
Six subjects made saccades toward a horizontally stepping target within 80-300ms. For each subject we constructed two classes of latencies, “short” and “long”, using the first and last quartiles (e.g. [80;151]ms and [185;300]ms respectively). We then concurrently reinforced each class in three blocked conditions (approximately 20000 saccades per subject) with different probabilities such that the relative frequencies of reinforcing “short” versus “long” latencies were 9/1, 1/9 and 1/1.
We observed modifications of latency distributions depending on the experimental conditions: distributions shifted toward the shorter or longer tail or became strongly bimodal (e.g. modes = {135; 235}) and the relative proportion of latencies matched the relative proportion of reinforcers earned from each option (slope up to 0.95).
Our results indicate that learned contingencies might considerably affect the allocation of saccades in time, and provide strong evidence of a voluntary control of saccadic latency. The functional significance of this control extends well beyond information accumulation.