Human perceptual judgments are consistently influenced by recent sensory experience. To examine how attentional demands modulate this relationship, the researchers investigated serial dependence across both numerosity and orientation judgments. Combining empirical behavioral testing with an ensemble of Bayesian computational models, the study evaluated how human participants judge visual stimuli when attentional resources are manipulated, establishing how divided attention alters the incorporation of recent perceptual history.
The experimental investigation evaluated participants across three distinct laboratory experiments employing a dual-task design to modulate attention. In each experiment, participants performed a primary sensory estimation task paired with an attention-demanding central task. In Experiment 1, participants estimated numerosity in dot clouds. In Experiment 2 and Experiment 3, participants estimated the orientation of Gabor patches under differing levels of stimulus contrast, with Experiment 2 presenting high-contrast Gabor patches and Experiment 3 presenting low-contrast Gabor patches. Pairing these visual estimation tasks with an attention-demanding central task allowed the investigators to modulate attention systematically and observe how attentional load alters perceptual judgments across differing visual tasks and stimulus conditions.
Across all three experiments, the researchers found stronger serial dependence under divided attention. Rather than relying solely on empirical behavioral observations, the researchers employed an ensemble of Bayesian models to disentangle the separate influences of attention, sensory evidence, and prior responses in numerosity and orientation judgments. Six specific Bayesian models were constructed to evaluate how perceptual history shapes decision-making. These six computational models differed specifically in which of the three components of prior experience were included, namely average sensory evidence, prior sensory evidence, and prior response, and in how each model expressed the influence of history on current perceptual judgments.
The Bayesian modeling demonstrated that all three factors influence perceptual judgments, operating through distinct and opposing mechanisms.
Importantly, the strength of the influence exerted on these individual components was not fixed, but depended directly on contextual variables. The researchers determined that the weight of each component varied according to the nature of the task, whether participants were performing numerosity estimation or orientation judgments. The availability of attention, regulated by the presence of the attention-demanding central task, likewise determined how strongly individual historical components influenced ongoing perceptual judgments.
These findings led the authors to conclude that perceptual decisions arise from a context-specific integration of prior experiences. Because individual components exert opposing attractive and repulsive forces, the overall magnitude and even the net sign of serial dependence, determining whether the net perceptual outcome is attractive or repulsive, reflect specific combinations of task demands, evidence quality, and attentional availability. The study emphasizes that serial dependence is a flexible, context-dependent process and highlights the crucial importance of computational modeling for disentangling the individual factors that shape human perception.
Subject of Research: The effect of divided attention on serial dependence in numerosity and orientation judgments using Bayesian integration models.
Article Title: Quantifying effects of inattention on serial dependence in numerosity and orientation judgments using a Bayesian integration approach
Article References: Schädlich, L. S., Kristjánsson, Á., Carbon, C.-C., & Pastukhov, A. (2026). Quantifying effects of inattention on serial dependence in numerosity and orientation judgments using a Bayesian integration approach. BMC Biology. https://doi.org/10.1186/s12915-026-02772-8
Image Credits: AI Generated
DOI: 10.1186/s12915-026-02772-8
Keywords: serial dependence, attention, numerosity, orientation, Bayesian models, perceptual judgments, dual-task design, BMC Biology, Quantifying, effects, inattention, serial
News Source: Drew Townsend. (November 7, 2026). Divided Attention Strengthens Serial Dependence in Visual Judgments. Scienmag.
