Reading focus
Syllabus: How Do We Learn What Predicts What?
Separate the predictiveness, uncertainty, and value principles, and compare evidence from different attention measures. Follow the distinction between goal-directed selection and a learned bias that persists despite current goals.
Why you’re reading it. Broaden the fixed learning-rate picture into an account of how experience changes processing of cues themselves.
A question to consider. When a reliable predictor suddenly becomes unreliable, what would predictiveness-based and uncertainty-based accounts lead you to look for, and at what point in training?
Why this reading
Learning changes which events receive attention, and attention can shape later learning. This review separates predictiveness, uncertainty, and outcome value instead of treating every attentional bias as evidence for the same process.
Before you read
Associability is a cue’s capacity to acquire a new association. Predictiveness concerns how informative it is about an outcome; value concerns the outcome’s significance. Controlled attention follows current goals, whereas an automatic bias can persist despite them. Gaze, response speed, and learning rate measure different aspects of this problem.
Read for
- Compare the evidence for attending to reliable predictors with the evidence for attending to cues whose consequences remain uncertain.
- Ask how experiments separate a cue’s predictiveness from the value of the outcome it predicts.
- Examine the difference between a bias that helps the current task and one that persists when a previously useful cue becomes irrelevant.
Consider afterwards
- A brightly colored symbol reliably predicts a large reward. How could an experiment distinguish attention to its reliability from attention to the reward’s value?
Editorial context
This is an integrative theoretical review, not a quantitative estimate of one universal effect. Its account finds stronger human evidence for learned predictiveness than for uncertainty-driven attention and explicitly contrasts that balance with animal findings. Its proposed integration remains open to exceptions. The review’s discussion of clinical applications does not by itself establish treatment effectiveness.