Talk title: Measuring sense of agency

Abstract

It is not well understood how people derive a sense of agency (SoA) over their actions. Empirical approaches have employed both explicit and implicit measures to study SoA. The most widely used implicit measure is known as Temporal Binding (TB). TB is defined as the perceived subjective compression of the time interval between a voluntary action and its associated outcome. Though TB has been replicated numerous times and is widely considered to be an implicit measure of SoA in both the theoretical and applied literature, there is still some debate regarding the underlying mechanism for the effect. Here I will highlight several issues with interpreting TB as a measure of SoA. First, empirical evidence from the growing literature suggests a dissociation between explicit measure (i.e., verbal reports) of the sense of agency and the TB effect, thereby questioning the theoretical relationship between TB and SoA. Second, the TB effect is usually observed at the aggregate level, however we reanalyzed 10 TB data sets and found a consistent pattern of variability in the empirical signature of the TB effect at the individual level. This is not just an issue of measurement noise. We implemented a Bayes factor mixed method modeling approach which simulated participant true effects (accounting for sampling noise) from two temporal binding data sets and confirmed that some individuals have true effects in the opposite of the theoretically predicted direction. Third, the theoretically predicted direction of TB is not consistently present across time intervals. In fact, the data shows a reliable regression to the mean effect (meaning repulsion for shorter intervals and binding for longer intervals). Given that the temporal binding task is as memory task we implemented a Bayesian rational memory model to simulate data from two TB data sets. These results provide some evidence to suggest memory is an alternative, or at the very least partial, explanation for the observed pattern of results at the aggregate level in temporal binding tasks. I will discuss implications of these results in the context of TB as well as for our understanding of SoA more broadly.