Last updated: 2026-10-09

U
Undergraduate level

Can an Organisation Know Something No One in It Knows?

A ship's navigation team can fix the vessel's position to within a few hundred metres, continuously, in poor visibility, without any single person on the bridge holding the calculation in their head. One officer reads a bearing off a landmark. Another plots it on the chart. A third cross-checks it against dead reckoning and calls out the discrepancy. The position that results is a real, accurate piece of knowledge — and it exists nowhere except in the handoffs between people, instruments, and a marked-up piece of paper. Edwin Hutchins spent years watching exactly this happen aboard US Navy ships, and the conclusion he drew wasn't that the individual officers were unusually skilled at mental arithmetic. It was that the cognitive work of fixing a position was being done by a system — people, charts, and procedures together — that had properties none of its human members had on their own1.the model lives in the handoffs, not the heads

This series asks a recurring question about individual minds: how does a bounded agent construct a useful model of a world it cannot completely know? The organisational version of that question is the one this page opens with, and it's less obviously answerable than it looks. Is an organisation merely a convenient label for a collection of people who happen to coordinate well, or can it meaningfully be said to know something that no individual member knows?

A Ship That Knows Where It Is, and No One Who Does FoundationalKnowledge that endures for decades — core principles

Hutchins' own answer is direct: cultural activity systems have cognitive properties of their own, properties that differ from the cognitive properties of the individual people taking part in them1. The navigation team is his central case, but the same shape of claim turns up anywhere a group's output depends on a division of cognitive labour that no participant could do alone. An air traffic control room tracks dozens of aircraft at once across several controllers, paper strips, and radar displays, each controller holding a partial, overlapping picture that the room as a whole somehow keeps consistent. A hospital ward knows a patient's full clinical picture only as a composite of what the attending physician, the nurses on three different shifts, the pharmacist, and the chart between them each hold — no single person on that list has the complete picture at any one time, and the ward still gets the dosing right far more often than it gets it wrong. A military command structure takes reports from scattered units with incomplete visibility of the battlefield and produces a single operational picture that no individual unit's report, read alone, would support.

Three Senses of "The Organisation Knows" FoundationalKnowledge that endures for decades — core principles

The organisation knows X" is doing at least three different jobs, and keeping them apart matters before asking whether any of them is true. The weakest sense is distributed storage: different people each hold a different piece of the full picture, the way a hospital ward's physician, nurses, and pharmacist each know something the others don't. Nothing here is surprising — it just says a filing system has more than one drawer. A stronger sense is distributed processing: the computation itself, not just the facts going into it, is spread across people, instruments, and procedures, so that no one step, and no one person, performs the whole operation. The ship's position fix is this sense exactly — the calculation is the handoff between the bearing-taker, the plotter, and the dead-reckoning check, and asking "which person computed the position?" has no good answer because the computing was the coordination. The strongest sense is emergent coordination: a piece of knowledge that exists only in the pattern of interaction itself and cannot be recovered from any subset of the participants examined separately, the way an air traffic control room's up-to-date picture of the sky vanishes the moment you try to locate it in any one controller's head rather than in the room's running exchange of strips, call-outs, and radar sweeps.

Is This a Metaphor, or a Real Claim? FoundationalKnowledge that endures for decades — core principles

Granting that distributed storage and distributed processing both happen leaves the harder question open: does any of this add up to an organisation genuinely knowing something, or is "the organisation knows" always eliminable, in principle, into a longer sentence about individuals and their coordination with nothing left over? This series' sister page on the distributed mind argues a closely related point about a single artificial system built from many component processes: multiplying identical agents is not the same as multiplying perspectives, and productive incompleteness — where no one part contains a full account of the whole, yet the whole still functions — is the normal condition of a sufficiently complex system, not evidence that nothing real is going on underneath. The same distinction applies one level up. An organisation whose members hold genuinely different, complementary models of the situation is closer to the strong sense of organisational knowledge than one where every member holds an identical, redundant copy of the same information — redundancy protects against single points of failure, but it doesn't produce anything a single competent member couldn't produce alone, which is exactly the caution this series' own distributed-mind argument raises about adding more of the same kind of part rather than a genuinely different one.

Strongest Objection: An Organisation Is Not a Brain FoundationalKnowledge that endures for decades — core principles

The analogy has an obvious limit worth taking seriously rather than talking around. A brain has something no organisation obviously has: a comparatively fast, richly interconnected substrate that integrates signals from every specialised part into something like a shared workspace, the architecture this series' sister page on Global Workspace Theory describes for individual cognition. An organisation's "integration" runs through meetings, memos, shared documents, and institutional memory — slower by orders of magnitude, lossier, and interruptible in ways a working nervous system typically isn't. Calling a hospital ward's composite picture of a patient a form of "knowing" risks smuggling in an assumption of unity the evidence doesn't support: a ward is not plausibly having an experience of knowing anything, whatever is true of the people working in it.the sister page unpacks this neural architecture

That objection narrows the claim without eliminating it. Hutchins' own navigation-team case already shows how a slower, artefact-mediated substrate can do real integrative work without anything like a biological nervous system underneath it: the chart is the shared workspace, the verbal call-out is the broadcast, and the "memory" that makes the position fix durable from one watch to the next lives in the ship's log rather than in any one officer's recall. An organisation's version of a global workspace, where one exists at all, is built out of artefacts and procedures rather than neurons — slower, more fragile, and far easier to break by removing the wrong document or the wrong meeting than a brain is by removing one neuron, but not therefore nothing.

Provisional Conclusion FoundationalKnowledge that endures for decades — core principles

An organisation can genuinely know things no individual member knows, at least in the distributed-processing and emergent-coordination senses distinguished above — the navigation team's position fix and the air traffic control room's current picture of the sky are real phenomena, not figures of speech, and eliminating them into "many individuals, well coordinated" loses the fact that the coordination itself is where the knowing happens. Whether that licenses calling the organisation a mind, rather than a system that does some of what a mind does without the unity a mind seems to have, is a further step this page leaves open on purpose — the same caution this series applies to a single artificial system's claim on cognition applies here too, one level up.

Questions for Further Thought

  • If a navigation team's position fix exists only in the handoffs between its members, who — if anyone — is the one doing the knowing?
  • Does an organisation need anything resembling a global workspace to count as knowing something collectively, or is artefact-mediated coordination (a shared chart, a shared document) already enough?
  • Would redundancy — several members holding identical information — ever count as organisational knowledge in the strong sense, or does the strong sense require genuine difference between what each member holds?

Further Reading

  • Hutchins, E. (1995). Cognition in the Wild. MIT Press.

References

  1. Hutchins, E. (1995). Cognition in the Wild. MIT Press. ↩ ↩