What Is a Cognitive Map? Organizing Knowledge for Flexible Behavior

Abstract

It is proposed that a cognitive map encoding the relationships between entities in the world supports flexible behavior, but the majority of the neural evidence for such a system comes from studies of spatial navigation. Recent work describing neuronal parallels between spatial and non-spatial behaviors has rekindled the notion of a systematic organization of knowledge across multiple domains. We review experimental evidence and theoretical frameworks that point to principles unifying these apparently disparate functions. These principles describe how to learn and use abstract, generalizable knowledge and suggest that map-like representations observed in a spatial context may be an instance of general coding mechanisms capable of organizing knowledge of all kinds. We highlight how artificial agents endowed with such principles exhibit flexible behavior and learn map-like representations observed in the brain. Finally, we speculate on how these principles may offer insight into the extreme generalizations, abstractions, and inferences that characterize human cognition.

Episodic memory retrieval success is associated with rapid replay of episode content

“…Individual episodes of experience are important building blocks for creating a representation of the structure of the world (2). Episodic representations that support replay are likely to be important for how we successfully navigate spatial, social, and abstract environments (3,6,(39)(40)(41)(42)(43). In turn, memory closely interacts with decision making (e.g.…”

Map making: Constructing, combining, and inferring on abstract cognitive maps

“…A powerful advantage of a cognitive map of an environment or task is the ability to make inferences from sparse observations that can dramatically accelerate learning and even guide novel decisions never faced before Behrens et al, 2018;Jones et al, 2012;Stachenfeld, Botvinick, & Gershman, 2017;Tolman, 1948;Vikbladh et al, 2019), a hallmark of behavioral flexibility and a key challenge in artificial intelligence Kriete, Noelle, Cohen, & O'Reilly, 2013;Wang et al, 2018). This is in part because a cognitive map of a task space allows "shortcuts" and "novel routes" to be inferred, as in physical space.…”

Grid coding, spatial representation, and navigation: Should we assume an isomorphism?

“…They found a sixfold modulation of activity based on the angles of such trajectories in entorhinal cortex, prefrontal cortex, and temporal parietal junction. Based on these and other studies (Bellmund et al, ; Garvert et al, ; Horner et al, ), Bellmund et al () argued that such grid coding might extend more broadly into abstract or “cognitive spaces” (see also: Behrens et al, ). Such grid‐like coding also manifests during situations that involve coding temporal intervals (Kraus et al, ; Tsao et al, ), further suggesting a grid coding mechanism as a potential universal “metric” for processing information.…”

“…In addition to their presence during spatial navigation, studies have also observed such grid coding during tasks involving representation of conceptual categories, imagination, and scene processing (Bao et al, ; Bellmund, Deuker, Schroder, & Doeller, ; Constantinescu, O'reilly, & Behrens, ; Garvert, Dolan, & Behrens, ; Horner, Bisby, Zotow, Bush, & Burgess, ; Killian, Jutras, & Buffalo, ). One theoretical interpretation of such grid coding is that it forms the basis of a universal spatial metric representation that underlies many, if not all, forms of spatial and nonspatial representations (Behrens et al, ; Bellmund, Gardenfors, Moser, & Doeller, ; Bush, Barry, Manson, & Burgess, ; Hasselmo, Giocomo, Brandon, & Yoshida, ; Hawkins, Lewis, Klukas, Purdy, & Ahmad, ).…”

“…One theoretical interpretation of such grid coding is that it forms the basis of a universal spatial metric representation that underlies many, if not all, forms of spatial and nonspatial representations (Behrens et al, 2018;Bellmund, Gardenfors, Moser, & Doeller, 2018;Bush, Barry, Manson, & Burgess, 2015;Hasselmo, Giocomo, Brandon, & Yoshida, 2010; Hawkins, Lewis, Klukas, Purdy, & Ahmad, 2018).…”