It is well established that a stimulus-onset-asynchrony between the fixation-target offset and the saccade-target onset considerably affects saccade latencies. A gap (fixation-target disappearing before the target-saccade onset) triggers short latency saccades. An overlap (fixation-target disappearing after the target-saccade onset) triggers long latency saccades. Here, we probe the possibility to control saccadic latencies using classical conditioning by systematically pairing a gap with one saccade direction and an overlap with the other.
In classical conditioning (Pavlov, 1927), unconditional stimuli (US; e.g., food) –eliciting unconditional responses (UR; e.g., salivation)– are paired with initially neutral stimuli (NS; e.g., bell). After repeated pairing, the NS –then called conditional stimuli (CS)– come to elicit conditional responses (CR) comparable with the UR.
We first associated a saccade direction (i.e. leftward or rightward, NS1 or NS2) with either a 150ms gap (US1) or a 150ms overlap (US2). We then introduced leftward and rightward probe-trials in which there was no SOA (CS1 and CS2; 20% of trials). Once steady state was observed, we did a return-to-baseline and then reversed the direction pairing.
During baseline, we observed no difference in latencies across saccade directions. During training (6200 pairing trials on average), the gap and overlap (US) resulted in shorter and longer latencies (UR), respectively (median latencies differed by 156ms on average; all outside the 98% null hypothesis CI). For the probe-trials (700 trials), we observed considerable differences in latency distributions (CR; i.e., on average 58ms; all outside the 98% null hypothesis CI) consistent with direction pairing (CS). Interestingly, during the return-to-baseline sessions (200 trials), there was a slight maintenance of the CR (median latencies differed by 22ms).
Our results demonstrate control of saccadic latencies by saccade direction using classical conditioning. This study further establishes that learned environmental contingencies affect the temporal allocation of saccades (Vullings & Madelain, 2017).