Does the cognitive map concept apply equally to physical and non-physical spaces, and what evidence supports that?

A cognitive map is a mental layout, not necessarily a picture of a room

A cognitive map is an internal representation of how things are arranged relative to one another, so that you can work out routes, distances, and shortcuts without having to re-explore from scratch. In its classic form it describes physical space: the hippocampal-entorhinal system is thought to encode the geometric properties of an environment [3]. The question is whether the same kind of representation also covers spaces you cannot walk through, such as a set of social relationships or a body of knowledge.

The strongest version of the claim comes from a review of the orbitofrontal cortex, a brain region central to value-based decisions. The authors argue that its seemingly complex functions may be explained by one role: providing a goal-directed cognitive map that guides adaptive, reward-based behavior in both social and non-social domains [1]. That is a proposal about how to unify many findings, not a direct demonstration that the two kinds of map are identical.

Brain and disease evidence links physical and abstract navigation

The Alzheimer's literature offers a complementary line. Spatial navigation fails early in Alzheimer's disease and mild cognitive impairment, sometimes before noticeable memory symptoms, and this is attributed to disruption of the brain's navigation system and its cognitive map [2]. Crucially, the same review reports that these deficits extend into abstract, knowledge-based domains, not just physical wayfinding [2]. Because the same condition damages both, the two abilities appear to depend on overlapping machinery, though the review synthesizes existing studies rather than testing the overlap directly.

The orbitofrontal account and the Alzheimer's account point in the same direction from different angles: one from normal value-guided behavior across social and non-social settings [1], the other from what breaks when the system is damaged [2]. Neither establishes that a social map and a spatial map are built the same way at the level of individual neurons; the orbitofrontal authors themselves note that how exactly its neurons support social function, and how social and non-social computations relate, remains unresolved [1].

Space and non-spatial features seem to be stored together

If physical space were represented independently, then something like color should not change your sense of distance. In virtual-reality experiments, participants learned the locations of houses and then judged pairwise distances and reconstructed maps. Houses that were similar in color were drawn closer together on the reconstructed maps in both experiments, and distance judgments were smaller for similar houses in the second experiment [3]. The authors conclude that this bias is difficult to reconcile with an independent representation of space, and support theories in which space is encoded together with other properties [3].

This is evidence about how a physical map is built, not proof that abstract spaces use the same map. But it removes a sharp boundary: the mechanism that encodes space appears to have a broader function than geometry alone [3], which is what you would expect if the same machinery is reused for non-physical material.

The map is not a perfect survey, and lab maps are not everyday maps

Even for physical space, the mental map is not a faithful scale drawing. People systematically deviate from true Euclidean geometry when navigating, and a model that deliberately distorts a Euclidean map to fit behavior predicted human data from a non-Euclidean maze as well as a non-metric model did [5]. The authors argue that such an embedding arises naturally from integrating local position information into a metric framework, and may be the better model of the human cognitive map [5]. In other words, a cognitive map can be metric-ish without being accurate.

The setting also matters. In one study, men outperformed women on a virtual environment called Silcton, but there was no sex difference for familiar real-world locations; within sexes, women were more accurate with real-world locations than with Silcton, while men showed the opposite pattern [4]. In a second experiment, women were more accurate with two real-world environments than with Silcton [4]. The authors caution that performance in a non-immersive virtual lab environment does not seem to indicate real-world map accuracy [4]. So findings about cognitive maps depend on whether the space is virtual or real, familiar or new, and this qualification applies to physical spaces before it is extended to abstract ones.

Sources used in this answer

Related research questions