Purchase this article with an account.
Hrag Pailian, Justin Halberda; Individual differences in VWM capacity assessed by the flicker task. Journal of Vision 2011;11(11):1256. doi: 10.1167/11.11.1256.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Here we demonstrate that the flicker change detection paradigm provides a more stable estimate of individual differences in visual working memory (VWM) capacity than the more typical “one shot” paradigm. Recent excitement has surrounded the drive to understand individual differences in VWM capacity, as these differences may be a locus for broader impacts throughout cognition. In our study, participants completed 240 or 480 trials of either a one shot or a flicker task, during which they were presented with displays containing 2,4, or 8 colored squares. The one shot condition used standard parameters from published papers and capacity was estimated using Cowan's K. In the flicker condition, participants saw these same displays with the memory and change displays alternating until the changing stimulus was found and capacity was estimated using Rensink's RT-based estimate. Capacity estimates were generated for each set size and compared. Reliability of capacity estimates were assessed via split-half correlation and revealed significantly higher reliability for flicker than for one shot. Dramatically, after 10 minutes of testing (the amount used in many published papers with the one shot task), split-half correlation for one shot was non-significant (R2 = .01), suggesting little to no reliability in its estimate of individual differences in VWM capacity, while split-half correlation for flicker was significant (R2 = .63). When the number of trials in the task was doubled, one shot began to produce significant split-half correlations. A likely source of the difference between these tasks is that the flicker task requires the subject to deploy VWM capacity many times during a single trial allowing for a more stable estimate of VWM capacity. Concerns about search behavior in the flicker task remain (e.g., is it random with or without replacement or something in between) and further comparison of these two paradigms seems warranted.
This PDF is available to Subscribers Only