Target Fidelity, but not Inducer Fidelity, Modulates Repulsive Serial Dependence of Visual Working Memory Representations

Yildirim, Bugay; Boduroglu, Aysecan · 2025 · Crossref

DOI: 10.21203/rs.3.rs-7835332/v1

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates how uncertainty in visual working memory representations modulates serial dependence biases, specifically challenging the prevailing Bayesian accounts that predict attractive biases under noise. While prior research suggests that serial dependence optimizes perceptual decisions by attracting current estimates toward previous stimuli (inducers) when the current target is noisy, less is known about repulsive biases or the role of inducer uncertainty. The authors aimed to determine whether target and inducer fidelity modulate repulsive serial dependence in a task designed to encourage the individuation of serially presented items. The researchers employed a continuous-measure attentional blink paradigm where participants reproduced the orientations of three consecutive targets (T1, T2, T3). The lag between T1 and T2 was manipulated to be either 3 (within the attentional blink window, inducing high noise/low fidelity for T2) or 7 (outside the window, ensuring high fidelity). T3 always followed T2 at a fixed lag of 7. This design allowed the authors to test the impact of representational noise on T2 (as a target) and on T3 (where T2 served as the inducer). Orientation differences between targets were large to encourage repulsive biases, which optimize performance by separating similar representations. The results demonstrated that reduced attentional resources significantly increased the repulsive serial bias in T2 estimates. When T2 fidelity was low due to the attentional blink (Lag 3), estimates were repelled more strongly from T1 compared to when T2 fidelity was high (Lag 7). Conversely, the fidelity of the inducer (T2) did not modulate the repulsive bias in T3 estimates; individual data analyses showed no significant difference in T3 bias based on T2’s noise level. Furthermore, exploratory linear mixed-effects models revealed that repulsive biases extended beyond immediate history: T3 estimates were repelled by T1, and T2 estimates were repelled by the subsequent T3. This latter finding indicates that repulsive effects can emerge during working memory maintenance, influenced by stimuli encoded after the target. These findings challenge standard Bayesian models of serial dependence, which predict that noisy inducers should reduce bias magnitude and that working memory processes are exclusively attractive. Instead, the results support an adaptive error-reduction framework where repulsive biases help individuate noisy representations. The study concludes that visual working memory employs adaptive, goal-driven dynamics to minimize error, which may not align with strict Bayesian optimality but rather with efficient coding or adaptation mechanisms that prioritize the separation of concurrent representations.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• working memory