Timing Performance Error in Rewarded and Non-Rewarded Tasks

Abstract

The literature on human and nonhuman animal interval timing disagrees about whether perceived time is a linear or power function of real time, and to what extent reward influences timing performance. Two competing computational learning and timing models, Temporal Difference (TD, Schultz, 2013) and Sometimes Competing Retrieval (SOCR, Stout & Miller, 2007) are reviewed. The present experiments investigate human interval timing error in both reward and non-reward conditions. The experiments were simulated by a computational model to identify both the function that describes the effect of interval duration on the distribution of variance (e.g., scalar or linear) and the relative predictive power of the SOCR and TD models, and the effects of reward on interval timing. Specifically, it was hypothesized that 1) timing variability is scalar, not linear, 2) that a modified SOCR model explains the data, and 3) that interval timing performance is less variable in rewarding situations than in non-rewarding situations. Timing trials involved the presentation of a reference duration; participants then produced their estimate of that duration while under cognitive load (random number generation and serial math tasks) through key presses on a computer. The results failed to support these hypotheses. However, reward produced a nonsignificant tendency towards early responding. Finally, suggestions for further research, including further computational modeling and investigation of the neural substrata of reward and timing are discussed.