Which visual working memory model accounts best for target representation in the attentional blink?

Abstract

AbstractPeople often fail to detect the second of two targets when there is a short time interval of ∼500 msec or less between them. This phenomenon is known as the attentional blink (AB). Accumulating evidence suggests that the AB is a result of a failure to select and consolidate the second target in working memory. The current literature has assumed that the standard mixture model of visual working memory (VWM) explains representation in the AB better than resource-based VWM models. However, no existing study has systematically compared VWM models in the AB domain. Here, we present a comparison of eight widely-used VWM models in four different AB datasets from three separate laboratories. We fitted each model and computed the Bayesian information criterion (BIC) values at an individual level, across different conditions and experiments, based on which we compared the models by their average model ranks. We found that, for most experiments presented here, the standard mixture model, the slot model, and their variants do outperform the others. We nevertheless also observed that certain details, such as the stimuli or spatial arrangement of targets used in the AB task, can result in different model rankings. Our results can help researchers to select the best model for their AB studies in the future, and thereby gain a better understanding of their data.