Bending the arrow of time with curves of errors
Many species of animals can time intervals in seconds-to-minutes range and organize their responses adaptively around temporal contingencies that predict biologically critical outcomes. Although the resultant timing behavior is typically (on average) accurate, endogenous timing uncertainty results in substantial trial-to-trial behavioral variability that conforms to the same psychophysical laws in different species. Adding to such similarities, I have been consistently observing that both humans and rodents can integrate their magnitude-representational uncertainty (including its psychophysical features) into decision-making in a nearly normative fashion and maximize the reward rate attained. These findings raise the question of whether metric errors, as the trial-based manifestations of uncertainty in magnitude representations, can be monitored by decision makers in the absence of feedback. In this presentation, I will first talk about the relationship between interval timing and decision making within the framework of optimality and their integration at the level of the underlying generative processes. I will then present evidence demonstrating that humans and mice can keep track of their timing errors. These recent findings suggest that the adaptiveness of decision-making in response to magnitude-representational uncertainty might be mediated by “metric error monitoring”.
Invité par Valérie Doyère