What is Bounded Rationality?

Every executive team believes it weighs the options. Almost none has the information, the attention or the time to do so, and Herbert Simon spent a career explaining why that is not a character flaw. It is the human condition, and pretending otherwise is how organisations end up designing decision processes for a species that does not exist.

Rationality has limits, and Simon named them

Classical economics assumed a decision-maker who knows every option, can compute every consequence and has a settled preference for ranking them. Simon, writing from inside real organisations in Administrative Behavior (1947) and the 1955 paper A Behavioral Model of Rational Choice, pointed out that no such person has ever held a job. Real people decide with partial information, under deadline, while attending to a dozen other things. Their rationality is real but bounded: bounded by what they know, bounded by what they can hold in mind, bounded by the clock. The Nobel committee honoured the insight in 1978, and the decades of behavioural research since have only deepened it.

The point is often misread as pessimism about people. It is the opposite. Simon was rescuing decision-makers from an impossible standard, and moving the burden to where it belongs: the environment in which they decide.

Satisficing, not optimising

Simon gave the coping strategy a name: satisficing, a blend of satisfy and suffice. A bounded decision-maker does not survey the whole option space and select the maximum. They search until they find the first option that clears their threshold of acceptability, and they take it. When the search is expensive and the deadline is near, the threshold quietly drops.

This is not laziness; it is arithmetic. Optimising requires the full set of options and the capacity to compare them, and neither exists in a live business. The relevant question is never whether your leadership team satisfices. It does. The question is what determines the threshold, and what happens to the options nobody had time to find.

One concrete example

Clearly illustrative, with no customer implied. A services firm of a few hundred people has a renewal to price by Friday. The account lead knows the client, the delivery lead knows the team, and the finance partner has last quarter’s numbers. Between them they can assemble perhaps a third of the relevant evidence before the deadline: the rest sits in systems nobody has time to query, in the heads of people not in the room, or nowhere at all. Three options get sketched. The first one that looks defensible to everyone present is chosen, the meeting ends, and the option nobody sketched, the phased renewal that would have fitted the client’s budget cycle, is never seen at all. Nobody behaved badly. Every person in that room was intelligent, experienced and diligent. The structure decided which evidence arrived and how far the search ran, and the structure was a calendar invite.

Design for the mind you actually employ

The lesson executives most often draw from decision research is that people need training in critical thinking. Simon’s lesson was structural, and harder-edged: an organisation is a decision-making structure, and it either compensates for bounded rationality or it amplifies it. The answer to bounded rationality is not smarter people; it is better decision structure. Structure determines what evidence is in view when the choice is made, how wide the option search runs before the deadline closes it, and whether the threshold of acceptability is explicit or just a feeling in the room.

Done well, structure extends the bounds. Systems can hold more options than a meeting can, search wider than an analyst can, and remember longer than a team can. The human keeps the judgement; the structure carries the load that was never a fair thing to ask of a person.

Bounded rationality and the ONX vocabulary

Bounded rationality is the founding assumption of decision intelligence: if people optimised naturally, none of its machinery would be needed. The vocabulary maps onto Simon’s limits directly. Limited information is answered by evidence that carries its quality state, so a decision-maker can see at a glance what is measured, modelled, inferred, stated or unmeasured, instead of treating everything in the deck as equally solid. Limited attention is answered by a Decision Room that narrows the argument to the constraints that actually bind, rather than spreading it across everything that could be discussed. And limited time is answered by letting the system hold and re-price the option space as facts change, so the search does not end just because the meeting did. The human still decides, exactly as Simon would have insisted. The structure just stops pretending the human is someone else.