Explicitly predicting outcomes enhances learning of expectancy-violating information

Abstract

Predictive coding models suggest that the brain constantly makes predictions about what will happen next based on past experiences. Learning is triggered by surprising events, i.e., a prediction error. Does it benefit learning when these predictions are made deliberately, so that an individual explicitly commits to an outcome before experiencing it? Across two experiments, we tested whether generating an explicit prediction before seeing numerical facts boosts learning of expectancy-violating information relative to doing so post hoc. Across both experiments, predicting boosted memory for highly unexpected outcomes, leading to a U-shaped relation between expectedness and memory. In the post hoc condition, memory performance decreased with increased unexpectedness. Pupillary data of Experiment 2 further indicated that the pupillary surprise response to highly expectancy-violating outcomes predicted successful learning of these outcomes. Together, these findings suggest that generating an explicit prediction increases learners' stakes in the outcome, which particularly benefits learning of those outcomes that are different than expected.

Keywords: Active learning; Prediction error; Pupillometry; Surprise; Violation of expectation.

Fig. 1

Schematic overview of the study and test phase of Experiment 1. The prediction and postdiction condition were performed in different blocks. In the study phase, participants predicted the numerical fact before seeing the correct number (prediction condition, 45 trials; left panel) or after seeing the correct number (postdiction condition, 45 trials; right panel). Each study phase was followed by a test phase, in which participants were asked to recall the correct number for each of the presented facts (order of facts was pseudorandomly reshuffled). White print against a gray background was used to reduce luminance contrasts

Fig. 2

Relation between expectancy-violation (i.e., difference between expected and observed outcome; centered at the person mean) and memory accuracy, separately for the prediction (left panel) and postdiction (right panel) condition. The light grey lines show the best-fitting regression lines for each participant, the bold black line shows the best-fitting regression line at the group level. Memory performance followed a U-shaped function in the prediction condition: accuracy was highest for both highly expected and highly unexpected outcomes. By contrast, in the postdiction condition, memory performance linearly decreased the more unexpected the outcomes were

Fig. 3

(a) Relation between expectancy-violation (i.e., difference between expected and observed outcome; centered at the person mean) and memory accuracy in the prediction condition. The light grey lines show the best-fitting regression lines for each participant, the bold black line shows the best-fitting regression line at the group level. (b) Interaction between expectancy-violation and memory, separately for lower (left panel) and higher (right panel) than individual mean confidence rating. (c) Time course of mean pupil dilation across a study trial as a function of memory (=recall success 0/1) and the degree of expectancy-violation/absolute difference between expected and observed outcomes (collapsed into four bins): 0 = no expectancy violation, 1–2 = small expectancy-violation, 3–5 = medium expectancy-violation, 6–8 = high expectancy-violation). Red vertical lines correspond to the analysis window used for measuring the pupillary surprise response (i.e., “result phase”)