Final Report Summary - NEUROINT (How the brain codes the past to predict the future)
Our brains continuously monitor features of our environment to make predictions about what is likely to happen in the near future. This process merges statistical learning, assessment of uncertainty and prediction. Our research project used psychological and neurobiological tools to examine principles that govern this process. In particular, we strongly relied on functional neuroimaging, a non-invasive method for identifying the magnitude of activity in brain areas during task performance. Our core research question can be summarized as follows: is there a core, general-purpose brain system that codes information in the recent past in order to make predictions about the near future?
Over multiple studies, it emerged that the answer to the question we posed is a resounding NO. We found that different brain systems tracked the regularity of sensory inputs, depending on the modality of the input (e.g. auditory, visual) but also depending on very subtle ways by which regularity was manipulated. Furthermore, contrary to our initial expectations, brain regions that tracked regularity did not necessarily show a strong relation between their activity and input regularity; interestingly, some brain systems tracked regularity in an inverse-U manner, responding more weakly to highly regular and highly random inputs, and more strongly to inputs with mid-levels of regularity. Our findings also suggested that assuming that the very basic assumption -- that there are fixed brain systems that code for regularity -- may be a wrong conceptual starting point. Instead, we found that the way the brain organizes into synchronized networks is itself very strongly impacted by the regularity in the environment. Finally, we have found it is crucial to consider individual differences in accounting for how people capitalize on regularities to predict the future, because individuals with higher working memory capacity benefit more from those and use this knowledge to anticipate future events. To summarize, our results show that the brain’s responses to regularities is very strongly context dependent – regularities in different inputs are tracked in different systems, and activity in these systems, coupled with the person’s own cognitive ability determine how these regularities are used to predict the future.
Over multiple studies, it emerged that the answer to the question we posed is a resounding NO. We found that different brain systems tracked the regularity of sensory inputs, depending on the modality of the input (e.g. auditory, visual) but also depending on very subtle ways by which regularity was manipulated. Furthermore, contrary to our initial expectations, brain regions that tracked regularity did not necessarily show a strong relation between their activity and input regularity; interestingly, some brain systems tracked regularity in an inverse-U manner, responding more weakly to highly regular and highly random inputs, and more strongly to inputs with mid-levels of regularity. Our findings also suggested that assuming that the very basic assumption -- that there are fixed brain systems that code for regularity -- may be a wrong conceptual starting point. Instead, we found that the way the brain organizes into synchronized networks is itself very strongly impacted by the regularity in the environment. Finally, we have found it is crucial to consider individual differences in accounting for how people capitalize on regularities to predict the future, because individuals with higher working memory capacity benefit more from those and use this knowledge to anticipate future events. To summarize, our results show that the brain’s responses to regularities is very strongly context dependent – regularities in different inputs are tracked in different systems, and activity in these systems, coupled with the person’s own cognitive ability determine how these regularities are used to predict the future.