Cognitive Map: How the Brain Represents Space

The cognitive map is one of the most important concepts in neuroscience and cognitive psychology — and one of the most misunderstood. It is not a metaphor. It is a real neural representation: a spatial encoding of the environment distributed across specific brain regions, built from dedicated cell types, and used to plan movement, remember locations, and reason about the world.

Understanding the cognitive map — what it is, how the brain builds it, and what happens when it fails — illuminates not just navigation but a surprisingly wide range of cognitive functions, including memory, decision-making, and the organisation of knowledge itself.

Tolman's Original Insight

The concept was introduced by psychologist Edward Tolman in a landmark 1948 paper, "Cognitive Maps in Rats and Men." Tolman was challenging the dominant behaviourist view that animals learn only through stimulus-response chains — that a rat in a maze learns a sequence of turns, not a spatial layout.

His experiments showed something different. Rats that had explored a maze would, when their usual route was blocked, spontaneously take shortcuts through parts of the maze they had never traversed before. This flexible rerouting was impossible if they had only learned a sequence of responses — it required an internal representation of the layout, a map that could be consulted to generate novel routes. Tolman called this internal representation the cognitive map.

This was a radical idea for 1948. The behaviourists dismissed it. Seven decades of neuroscience have confirmed it in extraordinary detail.

The Neural Architecture: Place Cells and Grid Cells

The neural basis of the cognitive map was discovered in two stages that together earned the 2014 Nobel Prize in Physiology or Medicine.

In 1971, John O'Keefe discovered place cells in the rat hippocampus — neurons that fire specifically when the animal is in a particular location in its environment. Each place cell has a receptive field (a "place field") covering a region of space, and together the population of place cells provides a complete coordinate system for the environment. When a rat moves to a new environment, the place cells reorganise — "remap" — to create a new map specific to that space.

In 2005, May-Britt and Edvard Moser discovered grid cells in the entorhinal cortex — neurons that fire in a regular hexagonal lattice across the environment, creating a metric coordinate system. Where place cells mark specific locations, grid cells provide the metric fabric — the distances and directions — that holds the map together. Grid cells are the GPS coordinate system; place cells are the pins on the map.

Additional cell types complete the system: head direction cells track orientation, firing when the animal faces a specific direction regardless of location. Border cells fire near the edges of environments. Together, these cell types give the hippocampal formation the raw data needed to know where you are, which way you're facing, and how space is structured around you.

The Human Cognitive Map

Research on the cognitive map in humans confirms that the human hippocampus and entorhinal cortex support map-like spatial codes with the same functional organisation as in rodents. Neuroimaging studies show grid-like coding patterns in the human entorhinal cortex during virtual navigation — the hexagonal firing signature of grid cells visible at the population level through fMRI.

The human cognitive map has the same three-layer architecture identified in rodents. The hippocampus and entorhinal cortex provide the spatial codes. The parahippocampal and retrosplenial cortices anchor those codes to recognisable landmarks — connecting the abstract coordinate system to the real-world features that make a specific location identifiable. The frontal lobe then uses these representations to plan routes, evaluate alternatives, and update navigation plans when circumstances change.

This architecture is why specific brain damage produces specific navigation deficits. Hippocampal damage impairs the ability to build new spatial maps of unfamiliar environments while leaving old, long-established maps partially intact. Retrosplenial cortex damage disrupts landmark recognition and the ability to orient relative to the environment. Frontal damage impairs route planning and flexible updating without necessarily destroying the spatial representations themselves.

How the Cognitive Map Is Built

The cognitive map is not downloaded — it is built incrementally through exploration. As you move through an environment, two parallel processes update the map:

Path integration — tracking position through movement. Even in complete darkness, without any landmarks, the hippocampal system can estimate current position by integrating movement signals: speed, direction, and elapsed time. This internal dead-reckoning accumulates error over time, which is why landmark corrections are needed for long-range accuracy.

Landmark anchoring — correcting and calibrating the path-integration estimate using recognisable environmental features. When you see a familiar landmark, its position in your stored map gives your current path-integration estimate a reference point that resets accumulated error. This is why environments with more distinctive landmarks support better spatial memory than featureless ones.

The quality of the cognitive map that emerges depends on how actively and how completely the environment is explored — and crucially on whether the navigation was self-directed or passive. Following someone else through a space, or following GPS instructions, produces weaker cognitive maps than finding your own routes, because the path-integration and landmark-anchoring processes are only fully engaged during active spatial reasoning.

The Cognitive Map Beyond Space

One of the most exciting developments in cognitive neuroscience is the discovery that the hippocampal cognitive map system is not confined to physical space. Research shows that the hippocampus automatically extracts relational structures from experience — not just spatial relationships but temporal sequences, conceptual associations, and social hierarchies — and represents them using the same map-like architecture.

Grid-like coding patterns have been found in the human entorhinal cortex during tasks involving abstract, non-spatial knowledge — such as navigating through a conceptual space of objects varying along two dimensions. The same neural machinery that tracks position in physical space appears to track "position" in abstract knowledge spaces.

This finding reframes the cognitive map from a navigation tool into a general-purpose relational organiser. The brain uses spatial coding as its primary means of structuring all kinds of relational knowledge — which may be why spatial thinking is such a powerful tool for understanding abstract ideas, and why people instinctively use spatial metaphors to describe non-spatial concepts ("a high-status position," "a close relationship," "a distant memory").

Cognitive Maps and Spatial Reasoning

The cognitive map is the foundation on which all spatial reasoning builds. Spatial navigation depends on it directly. But so do the other spatial abilities tested by the tools on the Spatial Reasoning hub — mental rotation requires maintaining and transforming spatial representations, the Spatial Span Test measures the working memory capacity that underlies map building, and maze navigation exercises the active planning and spatial orientation skills that the cognitive map supports.

Keeping the cognitive map sharp requires active spatial engagement — navigating environments, planning routes, reading maps, and reasoning about spatial relationships. The tools on the Spatial Reasoning hub provide structured ways to exercise these abilities. A more complete explanation of how mental maps function in everyday navigation is in the article on mental maps.