Prioritization sharpens working memories but does not protect them from distraction

Abstract

Perceptual distraction distorts visual working memory representations. Previous research has shown that memory responses are systematically biased toward visual distractors that are similar to the memoranda. However, it remains unclear whether the prioritization of one working memory representation over another reduces the impact of perceptual distractors. In five behavioral experiments, we used different forms of retrospective cues (indicating the likelihood of testing each item and/or the reward for responding correctly to each item) to manipulate the prioritization of items in working memory before visual distraction. We examined the effects of distraction with nonparametric analyses and a novel distractor intrusion model. We found that memory responses were more precise (lower absolute response errors and stronger memory signals) for items that were prioritized. However, these prioritized items were not immune to distraction, and their memory responses were biased toward the visual distractors to the same degree as were unprioritized items. Our findings demonstrate that the benefits associated with prioritization in working memory do not include protection from distraction biases. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Fig. 1

Hypothesized models of perceptual distraction effect. a). Protection hypothesis: prioritized representations have high memory fidelity (sharp not fuzzy) and are protected from visual distraction (vertical not tilted). b) Vulnerability hypothesis: prioritized representations have high memory fidelity but are vulnerable to visual distraction. c) Null hypothesis: prioritized representations have high memory fidelity, but are similarly vulnerable to visual distraction as unprioritized representations.

Fig. 2

Illustrations of experimental procedures for exp 1–5.

Fig. 3.

a Psychological scaling function estimated from the control experiment. b. Illustration of the signal component of the TCC model. The transformation from signal function to response distribution is based on the signal detection rule. In a given trial, the face that has the strongest familiarity signal (green circle) is selected as the target face based on the distractor intrusion model.

Fig. 4

Absolute memory errors from distraction trials in all 5 experiments. Error bars indicate 95% confidence interval. Violin plots show individual mean data distributions.

Fig. 5

Memory biases from distraction trials in Exp 1 and each of the other experiments. Responses were biased toward distraction faces across tested targets and experiments. Error bars indicate 95% confidence intervals. Violin plots show individual mean data distributions.

Fig. 6

Retro-cue benefits were computed as the difference between prioritized and unprioritized items. Consistent retro-cue benefits following relevance cues in absolute memory errors, but no retro-cue benefits in memory bias measurements. Error bars indicate 95% confidence interval. Violin plots show individual mean data distributions.

Fig. 7.

a). Recovered parameters were positively correlated with simulated parameters across target-distractor distances. b). No systemic biases were observed in recovered parameters. c). Exp 4 is on the top row, and Exp 5 is on the bottom row. Line plots represent model predictions based on fitted parameters (MAP). Bar plots represent group-level error distributions. Model predictions fit well with the empirical data.

Fig. 8.

a). Target memory strength and distractor strength from experiment 4 & 5. Cueing led to strengthened memory signals for prioritized faces compared to both unprioritized faces and the neutral condition in both experiments. b). Active distractions led to stronger distractor signal strength in experiment 5 compared to passive distractions in experiment 4. However, distractor signal strength was indistinguishable for prioritized, unprioritized, and neutral conditions across experiments.