Cognitive state fluctuations impact learning in different contexts
We are constantly learning from the world around us. How do changes in our cognitive and attentional states impact this process? I will describe two projects examining relationships between internal state fluctuations and an automatic, fundamental process of learning—statistical learning, and a noisy, dynamic form of learning—adaptive learning. In the first project, we examined the consequences of sustained attention fluctuations for statistical learning. Participants completed a continuous performance task with shape stimuli online. Unbeknownst to participants, we manipulated what they saw in real time by inserting visual regularities (a sequence of three regular shapes) into the task trial stream when their response times suggested that they were in especially high or low attentional states. Demonstrating that attentional state impacts statistical learning, we observed greater evidence for learning of the regular sequence encountered in the high vs. the low attentional state. In project two, we reanalyzed an openly available fMRI dataset collected as participants performed an adaptive learning task in which they learned to make accurate predictions about the locations of a fallen object in an noisy and dynamically changing environment. Individual differences in a brain network signature of sustained attention predicted individual learning style, with individuals with network signatures of stronger attention showing a learning style more like that of a normative model. In addition, trial-to-trial fluctuations in a distinct network signature of working memory predicted learning performance, such that trials on which participants showed a network signature of stronger working memory were followed by closer alignment between human and model predictions on the next trial. Together, these studies reveal consequences of sustained attention and working memory fluctuations for learning in different contexts.