Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Sketchpad Series 

Tracking sampling strategies during value construction to guide novel choice

Poster Session B - Sunday, March 8, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom

Ata Karagoz¹ (), Akram Bakkour¹; ¹University of Chicago

Humans often need to assign value to novel options by drawing on memory. We sometimes do this by generalizing the value of learned categories and at other times by making relational comparisons between familiar component features of novel options. It remains unclear whether people recruit categorical, relational, or both strategies when making novel choices. To investigate how value is constructed during novel choice, we present a novel task where choice options are composed of three attributes that vary along continuous dimensions (shape, angle, color), each of which are linked to underlying reward categories. After learning this initial attribute-to-value structure, participants can infer value categorically or rely on relational comparisons between the choice options on a particular dimension. To allow us to track how participants acquire and use information in their estimation of value we enforce explicit sampling of attributes using a mouse-contingent bi-resolution display paradigm, blurring attributes until participants hover over them. We predict that categorical vs relational strategies will be reflected in how participants sample attributes. Performing categorical inference should lead to sparser sampling where individuals visit each attribute of a choice option, retrieve the category value, and calculate a sum. In contrast, relational comparison should elicit denser sampling as participants alternate between the options on an attribute-by-attribute basis. Preliminary analyses indicate that participants sample more relationally when attribute distances between options are small in perceptual space, consistent with the idea that participants rely on relational judgments when categories are harder to discern.

Topic Area: THINKING: Decision making