TY - JOUR TI - Memory and prediction in a changing environment DO - https://doi.org/doi:10.7282/T3B56MTD PY - 2016 AB - This study aims to understand the effect of changes in the environment on memory and prediction. Previous work has suggested that when recalling events from memory, people use the entire distribution of information they have seen, and not a more recent set. However this work has only tested situations in which the distribution changes slowly, not those in which people encounter larger, more distinct changes in the stimulus distribution. Using an established experimental paradigm in categorical perception, I test memory and prediction in order to quantify performance changes with delineated temporal changes to the distribution of stimuli. I assess whether people use multiple belief distributions for decision making regarding the likely occurrence of future stimuli. I use a Bayesian approach to model how people update their beliefs, and the influence of changing beliefs on memory. This framework assumes that beliefs are a weighted combination of a prior belief and new evidence. Specifically, I will fit three models and discuss the qualitative fit to data. I will conclude with the use of a mixture model—which combines information from an overall distribution and a current distribution—and find that it provides a good explanation of participant behavior in this paradigm. KW - Psychology KW - Memory KW - Decision making KW - Bayesian statistical decision theory LA - eng ER -