Every time you glance at a face or a scene, your brain subtly skews what you see based on what you just saw a moment before. This project tackles a fundamental gap in vision science: researchers cannot agree on what “serial dependence” (attraction to the recent past) and “negative aftereffect” (repulsion from it) actually are, because no clear, operational definitions exist. Without these definitions, experiments using different visual stimuli produce conflicting results, and no unifying theory can emerge. The researchers will run a battery of behavioural experiments to map the precise temporal, spatial, featural, and attentional boundaries of these biases. They will then build a biologically plausible computational model that explains both attractive and repulsive effects through a single underlying grouping mechanism. This is fundamental science. It will not produce a market-ready tool or app. But establishing when and why our visual system distorts reality is a necessary step toward mitigating those distortions in high-stakes settings—for example, reducing diagnostic errors when radiologists interpret sequential medical images, or improving consistency in quality-control inspections. Deeper understanding of how perception bends under the weight of the immediate past could eventually reshape how we design visual tasks where accuracy matters.
View original technical description
Extensive research in vision science has consistently demonstrated that we are continuously influenced by our past experiences, revealing a variety of repulsive and attractive serial effects. These effects manifest in our actions, perceptions, decisions, and memory of various features or objects, leading to biases either away from the recent past (referred to as negative aftereffects) or toward it (referred to as serial dependence). Despite their ubiquity and importance, the field of vision science is nowhere close to achieve a consensus on the underlying mechanisms, neural loci, and functional purposes of these different types of serial effects. We believe a big part of the problem in the current debates lies in the fundamental lack of mutual understanding regarding what "Serial Dependence" and "Negative Aftereffect" are and what they are not. There are currently no operational definitions for these two attractive and repulsive biases, and research is fragmented into a variety of visual stimuli and experimental paradigms. Our primary goal is to establish the basic properties of attractive and repulsive serial effects, particularly in the domains of temporal, spatial, featural, and attentional tunings (Aim 1). To achieve this, we will propose a battery of behavioral experiments designed to precisely define attractive and repulsive forms of serial effects in an analytical and comprehensive manner across different stimuli and paradigms. We will then leverage the knowledge gained to inform the development of a biologically plausible unifying computational model (Aim 2). This model posits that an underlying grouping mechanism occurring over time can account for both attractive and repulsive serial effects. This research project will define the temporal, spatial, featural and attentional boundaries within which serial effects shape our visual experience. The gathered knowledge will inform the development of future models and theoretical frameworks. In the medium to long term, establishing the conditions under which serial effects occur and which bias we should predict will serve as a crucial milestone in the development of strategies aimed at mitigating these detrimental biases in applied contexts, such as medical image perception.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know