Abstract
We investigate how observers plan rapid movements to touch a target when only the prior probability of target occurrence (at each spatial location) is available during the initial part of the movement.
Methods: Subjects were asked to touch visual targets (6 mm wide bars) on a computer screen placed approx. 40 cm in front of them. The targets could be at any of 5 equally-spaced fixed locations along a horizontal line across the middle of the screen. At the beginning of each trial, subjects saw a color-coded display that signaled the prior probability of the target being at each possible location on that trial. Each location was marked as low probability (LP) or high probability (HP); there was a 0.7 probability that the target would be one of the HP locations. The independent variable was the prior: the number of HP targets and their locations on the screen. If there was only one HP target, for example, then subjects knew that the target would be there on 70% of trials. The target appeared only after the finger passed through an invisible trigger plane 1/3 of the distance between the start point and the target screen. Touching the target earned a reward and missing the target or failing to reach the target screen within 700 ms incurred penalties. Subjects were instructed to earn the greatest reward possible. We used an Optotrak 3020 motion capture device to record the subject's fingertip trajectory on each trial. We analyzed the reach trajectories and distribution of movement endpoints for each possible target location and choice of prior.
Results: Endpoint variance increases for targets located further from the mean of the prior. Movement speed at the trigger plane varies with the shape of the prior distribution: more sharply-peaked distributions elicit faster reaches. We will describe a novel method of comparing human performance to optimal performance defined as maximizing expected gain.