# Distributed Cognition Edwin Hutchins's challenge to cognitive science: the individual mind is the wrong unit of analysis for most real-world cognitive work. In his extended study of navigation on US Navy ships, Hutchins found that the cognitive achievement of successfully navigating a vessel through a harbour is not located in any one person's head. It is distributed across the crew, their instruments, their charts, their procedures, and the physical arrangement of the bridge. Remove the instruments and the expertise of the navigator drops dramatically. Change the crew configuration and the collective intelligence changes. The ship navigates well or poorly as a system — not as the sum of individual performances. This is distributed cognition: the idea that cognitive processes — perceiving, remembering, deciding, problem-solving — extend beyond the individual brain into the tools, artefacts, and social arrangements that the individual operates within. The unit of cognitive analysis is the functional system, which includes people, their tools, and the environment. ## Cognitive artefacts Donald Norman's related concept: cognitive artefacts are tools designed to extend or augment cognitive capacity. A checklist doesn't just remind you — it externalises memory and makes forgetting less likely. A dashboard doesn't just display data — it organises perception and makes certain patterns visible that would be impossible to hold in working memory. A shared calendar doesn't just coordinate schedules — it creates a distributed representational state that the whole group can read and update. The insight is that these artefacts are not aids to thinking — they are part of the thinking. The pilot who flies without instruments is doing a fundamentally different cognitive task than the pilot who flies with them, not just a harder version of the same task. When the artefacts change, the cognition changes. This matters for organisations: what a team can think and decide is partly a function of what artefacts they have access to. Better tools do not just make existing work faster — they make different cognitive work possible. ## Representational states Hutchins's technical contribution: cognitive work involves the propagation of representational states through a functional system. Information exists in different media — in someone's head, in a written record, in a physical display, in a shared understanding — and cognitive work is the process of moving information from one representational form to another in ways that make new action possible. The navigation team's work is a sequence of transformations: raw sensor data → plotted position → course calculation → helm order → vessel movement. Each transformation is a cognitive act, and each is performed by a different part of the system — person, instrument, or procedure. No single element does all of it; the intelligence is in the sequence and the coordination. For teams doing complex work, this analysis reveals where the cognitive load actually sits, which is rarely where people assume. The bottleneck is usually in a specific transformation — the point where information must change form and the system lacks the artefact or the protocol to do it reliably. Cognitive load analysis, in the Hutchins sense, is an audit of representational state transformations: where does information need to move, what carries it, and where does it break down? ## What this changes for teams The standard framing of team intelligence is additive: better individuals → better teams. Distributed cognition challenges this. A team with low individual expertise but excellent artefacts and coordination protocols can outperform a team with high individual expertise and poor artefacts. Hutchins's Navy crews were not selected for exceptional individual navigational intelligence; they performed extraordinarily well because the system they operated within was extraordinarily well designed. The inverse: a high-expertise individual placed in a poorly designed cognitive environment — bad tools, broken information flows, missing shared representations — will underperform relative to their capability. Much of what organisations diagnose as individual performance problems is actually distributed cognition problems: the cognitive artefacts are inadequate, the representational state transformations are broken, or the social coordination protocols have degraded. The coaching implication: before attributing performance problems to individual capability, ask what the person's cognitive environment looks like. What artefacts do they have? What information do they have access to? What shared representations does the team maintain? What transformations are they expected to make without adequate support? The answers often reveal that the system is doing work that looks like the individual's failure. ## Social coordination as cognitive infrastructure Hutchins's work on crews shows that social structure is cognitive structure. The division of roles on the bridge is not just an organisational convenience — it is the architecture of the cognitive system. Who talks to whom, in what sequence, about what, using which shared representations: all of this is the mechanism by which distributed cognition happens. This means that social breakdown is cognitive breakdown. When relationships on a team deteriorate, or when roles become unclear, or when the shared representational systems (dashboards, documents, meeting rhythms) fall into disrepair, the team's cognitive capacity degrades — even if every individual is as competent as before. The change is at the system level. The corollary: building team capability is partly about building social and artefactual infrastructure, not just developing individual skills. A team that invests in better shared representations, clearer protocols, and healthier relational dynamics is augmenting its own cognitive capacity. ## See Also - [[04_Garden/Notes/Observe/Shared Mental Models]] — shared mental models are the internal representational complement to Hutchins's external artefacts; together they constitute the team's full cognitive infrastructure - [[Socio-Technical Systems]] — the sociotechnical gradient is a distributed cognition problem: when the system of record diverges from the system of work, the artefacts stop supporting the cognitive task - [[Tacit Knowledge and Transmission]] — tacit knowledge is knowledge that has not yet been externalised into cognitive artefacts; distributed cognition explains why externalisation is so difficult and why it matters - [[Communities of Practice]] — CoPs are the social structures within which cognitive artefacts are developed, refined, and transmitted; the community's shared practice is also its shared representational system - [[Cohesion Theory]] — the positional dependencies Darwin identifies are partially distributed cognition dependencies: established combinations have developed the social coordination protocols that make the cognitive system function - [[Sensemaking]] — Weick and Hutchins are complementary: Hutchins focuses on the external artefacts and coordination protocols that support cognition; Weick focuses on the internal meaning-construction process; both are necessary - [[Theory of Constraints]] — the constraint in a system is often a cognitive bottleneck: a point where representational state transformation breaks down; finding the constraint is partly a distributed cognition audit - [[Environment as Designer]] — the lens that names the reciprocal relationship; distributed cognition is the mechanism by which the environment constitutes (not just supports) thinking - [[Environment Audit (15 min)]] — auditing the cognitive artefacts and coordination protocols in a specific environment - [[Edwin Hutchins]] — primary theorist; Cognition in the Wild (1995) is the foundational text