Certainty and Attention in Theory Protection: A Refined Eye-Tracking Study of Selective Belief Updating under Ambiguity (Study 2)

Read full text Summarise with VIRAT
Authors: Henning Romberg, David Torrents‐Rodas, Joshua Calder-Travis, Tania M. Lincoln
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This study investigates how belief certainty and attention contribute to selective learning patterns in belief updating. One such pattern has been described as Theory Protection (Spicer et al., 2020), referring to a mechanism by which beliefs are maintained despite disconfirming evidence. The paradigm builds on associative learning research showing that when two cues jointly precede an outcome (e.g., XY+), individuals may attribute unexpected outcomes to an ambiguous cue (X) rather than revising a well-established belief about a predictive cue (Y). This process mirrors mechanisms proposed to underlie the persistence of delusional beliefs, where disconfirming experiences are reinterpreted in ways that preserve overarching belief systems. The present study constitutes a preregistered refined follow-up to a prior implementation of the paradigm (Study 1; OSF link: osf.io/em5sh/). Building on insights from that study, the current design implements targeted procedural refinements to optimize contingency learning and enhance measurement sensitivity while retaining the core theoretical framework and task logic. In a within-subject design, participants complete a refined version of a modified allergist task (Uengoer et al., 2013) in which the causal status and ambiguity of cues are systematically manipulated. During critical compound learning trials (XY+), both cues generate prediction errors of differing magnitudes based on prior certainty. Trial-by-trial belief and certainty ratings capture subjective updating. Eye-tracking is used to test hypotheses about attentional dynamics derived from Mackintosh’s (1975) attentional model and Active Inference accounts of belief updating. Specifically, attention is predicted to shift from the previously predictive cue (Y) toward the ambiguous cue (X) during compound learning. From an active inference perspective, attentional allocation can be understood as a form of action that prioritizes information expected to be most informative given current beliefs. In causally ambiguous learning contexts, reallocating attention toward the ambiguous cue may represent an adaptive strategy to maintain a coherent explanatory model when confronted with disconfirming evidence. Behavioral analyses assess how cue certainty influences Theory Protection under optimized task conditions, and eye-tracking analyses quantify the predicted attentional shift using preregistered dwell-time indices. Together, these measures provide a confirmatory test of the cognitive and attentional mechanisms underlying belief maintenance in ambiguous learning contexts. All preregistration sections below specify the corresponding hypotheses, variables, design, and analyses.

Fields of Study & Subjects

Decision-Making and Behavioral Economics Neural and Behavioral Psychology Studies Cognitive Science and Education Research

Publication details

Publication date
1 January 2026
Language
EN
Document type
Preprint