Prediction accuracy dynamically modulates prediction-based false memory

Recently, in a novel paradigm, Bulatova and Fukuda (Cognition, 255, Article 106013, 2025) showed that making an arbitrary prediction about a future event was sufficient to induce robust false memories of the event even when it never took place. Here, we tested whether this prediction-based false memory reflects the adaptability of our predictive processing mechanism. More precisely, we hypothesized that the higher the likelihood of prediction confirmation, the stronger the prediction-based false memory would be. In Experiment 1, we manipulated the prediction accuracy during encoding (30% and 70%) between participants and found that prediction-based false memory was indeed greater when the prediction accuracy was higher. Furthermore, in Experiment 2, we demonstrated that this modulatory effect persisted even when prediction accuracy varied across encoding blocks within the same experiment, thus highlighting the dynamic nature of this adaptive process. Lastly, we established that this dynamic updating was not explained by explicit awareness of prediction accuracy manipulation alone. Taken together, our findings suggest that prediction-based false memory is dynamically updated based on the reliability of predictions, reflecting the adaptability of the predictive processing mechanism.