Abstract
In everyday life, we must extract information from the environment in order to generate predictions about future events and adjust behavior accordingly. Spatial and temporal aspects of objects in our environment can be particularly useful. Therefore, it is important to understand how spatial and temporal predictability can be learned in order to optimize behavior. Our study used a continuous performance task where a target (a red square) appeared in one of four possible locations, after one of four possible inter-trial-intervals (ITIs). Participants pressed a button as soon as they detected the target. The target location on trial n was associated with a particular location (spatial predictability) and ITI (temporal predictability) for the target on trial n + 1. Each target location was associated with a particular location and ITI (randomly determined for each participant before each block) for the following target appearance (e.g., if a target appeared in location a on trial n, there was a certain probability that it would appear in location b on trial n + 1). Across blocks of trials, we independently manipulated the degree of spatial and temporal predictability (low-25%, medium-65%, or high-90%). When target locations were highly predictive of the location of the next target, there was a significant reduction in reaction times (RT). However, high temporal predictability did not affect RT. Furthermore, there was no interaction between spatial and temporal predictability. Therefore, while participants were able to learn and use spatial information to predict where the next target would appear, they were not able to use this information to predict when the next target would appear.
Meeting abstract presented at VSS 2014