Related Experiment Video
Updated: Aug 7, 2025
13:00
Measuring Attention and Visual Processing Speed by Model-based Analysis of Temporal-order Judgments
Published on: January 23, 2017
10.0K
A Statistical Foundation for Derived Attention
Samuel Paskewitz1, Matt Jones2
1Department of Psychiatry, Children's Hospital, Anschutz Medical Campus, University of Colorado Denver.
Journal of Mathematical Psychology
|March 13, 2023
Summary
A new Bayesian model of derived attention explains how organisms focus on important cues. This model advances understanding of learning and attention, predicting inattention after backward blocking.
Area of Science:
- Cognitive Psychology
- Computational Neuroscience
- Machine Learning
Background:
- Derived attention theory posits organisms attend to cues with strong associations.
- Previous models, like Rescorla-Wagner, explain learned predictiveness and inattention to blocked cues.
Purpose of the Study:
- Introduce a novel Bayesian derived attention model.
- Provide a normative, statistical explanation for derived attention.
- Explain a wider array of attentional phenomena.
Main Methods:
- Combine Bayesian linear regression with approximate Bayesian learning.
- Simultaneously estimate cue-outcome associations and prior variance.
- Model assumes cue-outcome associations share the same prior variance, representing inherent cue importance.
Main Results:
- The Bayesian model explains learned predictiveness, inattention to blocked cues, and value-based salience.
- It also accounts for retrospective revaluation.
- Novel prediction: inattention following backward blocking.
Conclusions:
- The Bayesian derived attention model offers a more comprehensive explanation of attentional phenomena.
- It provides statistical insights into how uncertainty and predictiveness influence attention.
- Further development may clarify the interplay between uncertainty and attention.
