Modeling the effect of dynamic contingencies on anticipatory eye movements

In a simple visual tracking task, we manipulated across blocks the probability for the target to move in one direction (Right) or another (Left). Three probability conditions were implemented (50%, 75% and 90% of rightward trials) and the pseudo-random sequence of trials was the same for all subjects (N=18). The probability bias for target direction resulted in a robust anticipatory smooth pursuit (aSP) toward the expected direction and we found a linear effect of probability on aSP velocity across the three probability conditions. As Maus and collaborators (2015) recently proposed a model of aSP-velocity as a function of the mean target speed across the recent trial-history, we implemented an agent that generates aSP in function of the binned target direction of recent trials. We challenged this model by comparing its predictions to the observed aSP changes associated to specific trial-sequences (tested across many subjects). Moreover, in a Bayesian model framework, we analyzed long-term versus short-term effect of trial history on prior actualization (as probed by aSP). As a further development of our agent sensitive to dynamic contingencies of the environment, we aim to extend our model to mimic the observed sensitivity of aSP to reward contingencies (Damasse et al. 2015).

Meeting abstract presented at ECVP 2016

reference


All material (c) L. Perrinet. Please check the copyright notice.


This work was supported by ANR project ANR-13-APPR-0008 "ANR R.E.M.".
ANR logo


TagYear16 TagPublicationsProceedings TagAnrRem

welcome: please sign in