Andre Sahakian¹ (), Surya Gayet¹, Chris L.E. Paffen¹, Stefan Van der Stigchel¹; ¹Utrecht University

Visual working memory (VWM) temporarily stores task-relevant visual information to enable interactions with the environment. VWM is typically studied in temporally rigid paradigms in which memory arrays (viewing time) and retention intervals (the delay to the probe) are determined by the experimenter, and are often kept constant. In everyday VWM use, however, there is huge variation in (1) how long we look at things to put them in memory, and (2) how much time passes before we act on the memory. The temporal variations in both viewing times and delays depend on many internal (e.g., strategy, resources) and external factors (e.g., stimulus availability, physical constraints). Here, we ask how the recall performance of VWM content develops across these two orthogonal temporal dimensions: how do memories build up over viewing time and how do they decay over increasing delay periods? We employed a copying task, in which participants were tasked to recreate an “example” arrangement of items in an adjacent empty “workspace”. We tracked their unconstrained viewing and copying behavior at the level of individual items, recording how long items were viewed, and how much time passed before they were placed. Our results show that performance monotonically increased for viewing times up to one second (per item), but plateaued afterwards. Interestingly, while inspections exceeding one second did not improve performance for short (two second) delays much, inspections beyond one second did improve performance for longer delays. Our findings suggest that usable representations are produced quickly, while longer inspections make representations more resilient against decay. Likely, the natural variations partly arise due to strategic use of time and VWM resources in everyday behavior: when information needs to be applied immediately, a quick look should suffice, but when the use will be delayed, looking for longer might be worth the time investment.

Acknowledgements: This project was supported by an ERC Consolidator Grant [grant number ERC-CoG-863732] to Stefan Van der Stigchel, and a Veni grant from Netherlands Organisation for Scientific Research (NWO) [grant number: Vl.Veni.191G.085] to Surya Gayet.