Anticipating uncertain events: estimates of probability driving anticipatory eye movements

Montagnini, Anna; Souto, David; Masson, Guillaume S · 2009 · Crossref

DOI: 10.1186/1471-2202-10-s1-p20

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates how internal estimates of probability influence anticipatory smooth pursuit eye movements (aSPEM) in humans. Sensorimotor behavior is often guided by expectations of future events, which are likely derived from internal probability estimations. While expectancy is known to affect reaction times and visuomotor gain, this research specifically examines how the relative probability of different target motion types drives anticipatory motor responses. The authors aim to determine whether aSPEM reflects a continuous Bayesian accumulation of probabilistic information or a discrete state-based expectation. The researchers collected high-resolution eye movement recordings from four human subjects instructed to track a moving target. Two experimental conditions manipulated the probability ($p$) of either target direction (Right or Left) or speed (High or Low), with $p$ varying across blocks at values of 0, 0.1, 0.25, 0.5, 0.75, 0.9, and 1. The other variable remained constant. The analysis focused on aSPEM velocity as a function of recent trial history and the long-term block probability bias. The authors compared their empirical data against predictions from various models, including a simple leaky-integrator model and a Finite State Markov Model. The results demonstrated that a leaky-integrator model, which accounts for suboptimal information accumulation across trials, significantly explained fluctuations in aSPEM based on recent trial history. Regarding the global characteristics of aSPEM across different probability biases, the study found three key patterns: (1) a monotonic, nearly linear dependence of mean aSPEM on probability $p$; (2) a nonlinear, quadratic dependence of aSPEM variance on $p$; and (3) a unimodal distribution of anticipatory movements that shifted progressively between the distributions observed in deterministic conditions ($p=0$ and $p=1$). To explain these findings, the authors propose a model based on a continuous internal representation of $p$ and Bayesian accumulation of probabilistic information. This model assumes an independent motor component of variance proportional to the mean absolute aSPEM velocity and suggests that aSPEM distribution parameters are tuned to minimize a cost function proportional to quadratic retinal slip. The proposed continuous Bayesian model successfully captured the main properties of the experimental data and significantly outperformed a model based on discrete expectancy states (the Finite State Markov Model). These findings suggest that the brain utilizes a continuous, probabilistic representation of event likelihood to drive anticipatory motor control, optimizing performance by minimizing retinal error rather than switching between discrete expectation states. This provides insight into the neural mechanisms underlying how experience and probability estimates shape sensorimotor preparation.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• anticipation