TITLE:
What Is a Cognitive Map? A Critical Review of Its Conceptualization, Neural Basis, and Methods of Construction
AUTHORS:
João Miguel Alves Ferreira, Sergii Tukaiev
KEYWORDS:
Cognitive Maps, Spatial Navigation, Hippocampus, Entorhinal Cortex, Relational Representations, Cognitive Spaces, Representational Geometry, Spatial Cognition, Memory, Neuroimaging
JOURNAL NAME:
Advances in Molecular Imaging,
Vol.15 No.4,
October
12,
2026
ABSTRACT: Cognitive maps were originally proposed as internal representations of spatial environments that enable flexible navigation and adaptive behaviour beyond rigid stimulus-response associations. Since Tolman’s seminal formulation, the concept has expanded dramatically to encompass temporal sequences, conceptual knowledge, social hierarchies, and abstract relational structures. This expansion has generated substantial theoretical and empirical progress, yet it has also produced considerable conceptual ambiguity: what, precisely, qualifies a representation as a cognitive map rather than a simpler associative, schematic, or statistical structure? In this critical narrative review we examine the historical evolution of the cognitive-map construct, its proposed neural substrates (with particular emphasis on the hippocampal-entorhinal system), and the methodological approaches used to identify cognitive maps and to reconstruct their structure from behavioural, computational, and neural data. Here, ‘construction’ refers to the researcher’s methodological reconstruction of putative map-like structure from empirical data, as distinct from the organism’s own process of learning such a representation (the latter is addressed separately in Chapter 5). These approaches target relational structure, metric or topological organisation, generalisation, flexible inference, and navigation through representational space, and are contrasted with alternative frameworks that emphasise memory, prediction, or hierarchical coding. Behavioural paradigms, computational models, multivariate neuroimaging analyses, and high-resolution anatomical approaches are compared with respect to their ability to operationalise map-like representations. Building on this synthesis, we propose a framework that distinguishes the content of a map from the computational and neural mechanisms that encode it, and we discuss how multimodal integration of functional, structural, and microstructural data may refine future investigations. The review concludes by outlining principles for the empirical identification of cognitive maps and by highlighting unresolved tensions between spatial origins and domain-general claims.