Focusing on Attention: An Analysis on Effectual Manipulations of Working Memory Performance

Abstract

Working memory is the centerpiece of executive reasoning and thus critical for optimal behavior. While the interaction between attention and working memory has been known for over four decades, the exact details of this relationship are still unclear. Using a delayed estimation paradigm with variable delay and spatial cues, we found that attention significantly increases performance by decreasing mean angular error, increasing precision, and decreasing guessing and swapping errors. Our experimental analyses and model results suggest that attention achieves these effects by reducing sensory noise at encoding and preventing noise accumulation during maintenance. Additionally, we analyzed the effects of delay, color values, and gain as measured by pupil diameter on performance. While all of our experimental variables affected working memory performance through at least one metric, altogether they could not explain the wide variation in performance across trials and individuals. Finally, we analyzed the effects of our experimental variables to accurately model working memory resource as either discretely or continuously allocated. Zhang & Luck report that a discrete model fits attentional and delay manipulations. We replicated the result using an experimental setup that precludes this prediction by the discrete or continuous model. Because the result occurs independently of the model’s prediction, the discrete model can’t necessitate the result. In contrast to Zhang & Luck, we found a smooth decay of working memory precision across the delay as predicted by a continuous model.