Abstract
Laboratory research on Visual Short-term Memory (VSTM) often uses static stimuli, which is a great simplification of ever-changing stimuli in natural vision. It is unclear whether VSTM for dynamically changing information shows similar characteristics, such as limited storage capacity, as static stimuli. The present study assessed VSTM for continuously rotating stimuli using a delayed estimation task. In Experiment 1, participants memorized the orientation of a clock hand that continuously rotated clockwise or counterclockwise for 150 milliseconds under verbal suppression. Memory set sizes (1 versus 2 in Experiment 1A; 1 versus 3 in Experiment 1B) and the directions of rotation were randomly mixed across trials. After a 1000-ms retention interval, participants estimated the orientation of a randomly probed memory item, as accurately as possible, by clicking on the clock face. The distributions of orientation estimation errors were fit with the Zhang & Luck (2008) discrete slot model with an additional parameter (μ) representing the central tendency of the error distribution using maximum likelihood estimation and hierarchical Bayesian method. As set size increased, the precision of recalled memory representation decreased with a small increase in guessing rate, replicating some previous findings using static orientation stimulus. In addition, there were systematic shifts in memory representations that were in the directions of the clock hand rotation, indicating a lag between VSTM encoding and the onset of the stimulus. More interestingly, the shifts in μ were comparable across set sizes, providing preliminary support for parallel consolidation of VSTM representations. Experiment 2 replicated the shifts in μ using colors that continuously rotated in a circular color space. Additional control experiments ruled out some alternative accounts based on overall response biases and lack of sensitivity in the present method. Together, these results have provided some insights into VSTM for dynamic stimuli.
Meeting abstract presented at VSS 2017