Anticipation of future events improves the ability to estimate elapsed time

An accurate estimate of elapsed time is essential for anticipating the timing of future events. Here, we show that the ability to estimate elapsed time on a reaction time (RT) task improved with training during which human participants learned to anticipate the onset of a "Go" signal. In each trial, a warning signal preceded the Go signal by a temporal interval (i.e., foreperiod). The duration of the foreperiod was randomly drawn from a rectangular distribution (1-2 s). Participants were required to initiate a response immediately after the Go signal and performed the task for 480 trials/day for 12 days. Anticipation should have been governed by the probability that the Go signal would occur (hazard rate), which increased for longer foreperiods. Indeed, RTs decreased for longer foreperiods and were inversely related to the hazard rate. The pattern of RT decrease was well explained by the subjective hazard rate, which was formalized based on the assumption that the uncertainty of estimates of elapsed time scales with time (Weber's law). Notably, RTs demonstrated a more linear decrease as a function of foreperiod in LATE compared with EARLY training sessions. This involved a decrease in the Weber fraction used in the subjective hazard rate. The results indicate that the uncertainty associated with estimating elapsed time was reduced as participants learned and used the hazard rate to anticipate the onset of the Go signal. This finding suggests that the ability to estimate elapsed time improves with training on behavioral tasks that implicitly engage timing mechanisms.