Systemic Failure Archive and Research

Organisations as Cognitive Systems

5 min read

An organisation is not only a collection of people who think; it is itself a cognitive system. It perceives, represents, remembers, infers, decides and learns — not in any single head, but through the coordinated activity of people, artefacts, routines, records, information systems and, above all, communication. Treating the organisation as the unit of cognition is the frame that runs through this site: most of the failures catalogued here are not, at bottom, individual mistakes but breakdowns in an organisation’s collective ability to know its own situation and act on what it knows.

The individual-mind analogy is a working tool, not a metaphysical claim. What it buys is a vocabulary. We can ask what an organisation attends to, how it encodes the world, what it remembers and forgets, how it draws inferences, and whether it can update — and we can then locate a failure precisely in one of those functions rather than in a person.

An intellectual lineage

The idea has deep roots that converge from several directions. Simon, and March and Simon, treated the organisation as a decision-making and information-processing system operating under bounded rationality, a line continued in Cyert and March’s behavioural theory of the firm. Hutchins’ distributed cognition took the further step of relocating cognition itself into the sociotechnical system — spread across people and artefacts, so that the proper unit of analysis is the crew-and-instruments, not the individual navigator. Weick’s work on sensemaking describes how organisations enact and interpret their environments, and how that interpretive work can collapse under stress. The cybernetic tradition — Ashby’s law of requisite variety, Beer’s Viable System Model — casts the organisation as a self-regulating system that must generate enough internal variety to match the variety of its environment. Luhmann’s systems theory goes furthest, treating organisations as autopoietic systems that reproduce themselves through decisions and communications rather than through their members. And Nelson and Winter’s routines-as-organisational-memory supplies a frankly biological framing, in which routines function like genes.

These traditions differ in almost everything else, but they share one move: they make the organisation, not the person, the cognising agent.

The cognitive functions — and how each fails

The value of the frame is that organisational cognition can be decomposed into functions, and each has a characteristic failure mode visible in this archive.

  • Attention and perception — what the system can and cannot sense. Blind spots are structural, not personal. In Mid Staffordshire the warning signals existed across the system but there was no locus at which they converged into recognition.
  • Representation — how the world is encoded internally. A model can harden into a fixed point defended against evidence, as in the Horizon scandal, where a false representation of the system’s own reliability was protected for two decades.
  • Memory — routines, records and information systems as organisational and transactive memory. An organisation can remember the wrong thing, or fail to retrieve what it holds: knowledge present somewhere in the system that never reaches the decision.
  • Inference and sensemaking — the interpretive work of turning signal into meaning, which under pressure can collapse entirely (Weick).
  • Decision — bounded and satisficing; in Luhmann’s frame the decision is the very element through which the organisation reproduces itself, which helps explain why organisations defend prior decisions so fiercely, even against their own later interest.
  • Learning — single- versus double-loop (Argyris and Schön). The deepest failures are failures to revise the governing model, not merely the action; defensive routines block the double loop. Normalisation of deviance is the clearest example of a learning function running in reverse.

Why the frame matters for failure

Treating the organisation as a cognitive system relocates the diagnosis from “who erred” to “what in the system’s architecture prevented it from knowing.” Requisite variety exceeded; a representation defended past its evidence; signal that existed but never converged; a memory encoding the wrong routine. It also resists the reflex to fix failure by removing individuals while leaving intact the cognitive architecture that produced the error — which is why so many of the cases here recur in new institutions with new people.

The autonomous-systems turn

The frame becomes sharpest at its frontier: what happens when parts of organisational cognition are delegated to AI and machine-learning systems. When perception, inference and even decision are handed to computational components, the architecture changes — representations become opaque, the locus of decision blurs, and the organisation may lose the capacity to notice when the delegated cognition is wrong. This is where automation bias stops being an individual disposition and becomes a property of the whole system, and it is the central governance question raised by autonomous decision-making in organisations.

In this archive

Key sources

  • Simon, H. A. (1947). Administrative Behavior. Macmillan.
  • March, J. G., & Simon, H. A. (1958). Organizations. Wiley.
  • Cyert, R. M., & March, J. G. (1963). A Behavioral Theory of the Firm. Prentice-Hall.
  • Hutchins, E. (1995). Cognition in the Wild. MIT Press.
  • Weick, K. E. (1995). Sensemaking in Organizations. Sage.
  • Ashby, W. R. (1956). An Introduction to Cybernetics. Chapman & Hall.
  • Beer, S. (1979). The Heart of Enterprise. Wiley.
  • Luhmann, N. (2018). Organization and Decision. Cambridge University Press.
  • Nelson, R. R., & Winter, S. G. (1982). An Evolutionary Theory of Economic Change. Harvard University Press.
  • Argyris, C., & Schön, D. A. (1978). Organizational Learning: A Theory of Action Perspective. Addison-Wesley.

Graph View

Backlinks