Week 5 Seminar: Structure not sensed
The internal map, and the result that suggests it was never really about space at all.
Tolman argued in 1948 that rats in mazes were not acquiring chains of associations but something map-like — a representation of layout they could use in ways their training had never rewarded. O’Keefe and Dostrovsky found cells in 1971 that fired for place. The Nobel followed, eventually, and the story looked finished: there is a spatial system, here are its cells.
Then in 2016 Constantinescu, O’Reilly and Behrens put people in a scanner and had them navigate a two-dimensional space of concepts rather than places. The hexagonal grid-like signal turned up anyway, in entorhinal cortex and ventromedial prefrontal cortex, and it was stable across sessions a week apart.
Why this changes the shape of the course#
If the same machinery organises conceptual relations and physical layout, then what we have been calling a spatial system is more plausibly a general machine for relational structure, of which space is one instance — and a conveniently visible one, which is why we found it there first.
That matters for everything after this week. It means the thing introduced here isn’t a step on the road to cognition. It’s closer to the substrate that later weeks are running on. Whittington and colleagues’ Tolman-Eichenbaum Machine, from 2020, is the model that tries to unify the two, and it is on the reading list for those who want the formal version.
The limit#
A map buys you the ability to act on parts of the world you are not currently touching. What it costs is that the map can be wrong in ways the map cannot represent. Spatial judgement distorts in specific, predictable directions, and the distortion is not available to the person doing the judging — which is the first appearance in the course of a pattern that will become the whole argument by week 12.
Reading#
The paper this seminar is built around:
- Constantinescu, O’Reilly and Behrens (2016). Organizing conceptual knowledge in humans with a gridlike code. Science.
The rest of the week’s reading:
- Tolman (1948). Cognitive maps in rats and men. Psychological Review.
- O’Keefe and Dostrovsky (1971). The hippocampus as a spatial map. Brain Research.
- Menzel and colleagues (2005). Honey bees navigate according to a map-like spatial memory. PNAS.
- Whittington and colleagues (2020). The Tolman-Eichenbaum Machine. Cell — the formal unification, for those who want it.
Grades are not attached to papers here, because a grade belongs to a claim rather than to the paper it appears in — see evidence grades. Several of these papers contain claims at two different grades.
The spec
What you bring to the session. Written so a reader can tell whether it has been met without asking you.
- you can explain what a cognitive map lets an animal do that association alone cannot
- you can describe what Constantinescu and colleagues found, and why it reframes the previous eighty years
- you can state one thing a map-holding system cannot detect about its own map
Teaching team
- Tuyet Nabarro — lecturer