Abstract
Responses are faster when a warning signal (S1) precedes an imperative stiumlus (S2) by a foreperiod (foreperiod effect). When a foreperiod is variable, both the current and the previous foreperiods modulate reaction time. Los et al. (2014) proposed that the temporal relation between S1 and S2 on each trial is stored in a distinct memory trace, which along with earlier formed memory traces, leads to the variable foreperiod effects on subsequent trials. We modeled the data from three experiments using a Bayesian method to investigate whether the variable foreperiod effects systematically changed over time. For all three experiments, the model that included both trial- and block-level variability performed better than those not including both levels. However, only in Experiment 1 where the foreperiods were relatively long (400 ms and 1400 ms), were the parameters related to the systematic change of variable foreperiod effects significantly distinct from zero. In Experiments 2 (50 ms and 200 ms) and 3 (50 ms vs. 400ms), the evidence was unclear based on an HDI + ROPE decision rule. These results suggest different causes of the variable foreperiod effects between short and long foreperiod scenarios.