2018/05/31 by Christopher W. Lynn, Ari E. Kahn, Lynn, Christopher W. +5 · 3 citations
Computer Science · Neuroscience · Psychology · #Biological Physics (physics.bio-ph) #FOS: Biological sciences #FOS: Physical sciences #Mental Health Research Topics #Neural Networks and Applications #Neural dynamics and brain function #Neurons and Cognition (q-bio.NC) #Physics and Society (physics.soc-ph)
paper · pdf · doi:10.48550/arxiv.1805.12491
openalex publication_date 2018/05/31 · openalex created_date 2023/03/18 · openalex updated_date 2026/07/28
Humans are adept at uncovering abstract associations in the world around\nthem, yet the underlying mechanisms remain poorly understood. Intuitively,\nlearning the higher-order structure of statistical relationships should involve\ncomplex mental processes. Here we propose an alternative perspective: that\nhigher-order associations instead arise from natural errors in learning and\nmemory. Combining ideas from information theory and reinforcement learning, we\nderive a maximum entropy (or minimum complexity) model of people's internal\nrepresentations of the transitions between stimuli. Importantly, our model (i)\naffords a concise analytic form, (ii) qualitatively explains the effects of\ntransition network structure on human expectations, and (iii) quantitatively\npredicts human reaction times in probabilistic sequential motor tasks.\nTogether, these results suggest that mental errors influence our abstract\nrepresentations of the world in significant and predictable ways, with direct\nimplications for the study and design of optimally learnable information\nsources.\n