The same mistakes recur too reliably for learning to be doing much work

The argument

The sceptical case begins from a pattern rather than a theory. Opponents argue that if people learned from their mistakes, the record would look different than it does. Individuals return to relationships of the kind that previously harmed them, to spending that previously left them short, to work habits that previously made them ill, and to arguments that previously ended badly. Institutions repeat the failures their own inquiries documented. Generations re-enact errors that were described in detail by the generation before. A capacity which produced reliable revision would not leave this much repetition to explain. Opponents argue that the mechanisms working against learning are better understood than the mechanism supposed to support it. People attribute bad outcomes to circumstances and good ones to themselves, which protects the account they hold of their own judgement at the cost of the lesson. They remember having predicted outcomes they did not predict, which erases the error before it can be examined. They treat a decision that turned out badly as having been reasonable at the time, which is sometimes true and always available. Each of these operates automatically and in the direction of preserving the person's existing view of themselves. The claim is not that nobody ever changes. Opponents accept that people revise in narrow technical domains where feedback is immediate and impersonal, and that some people, after very large losses, change substantially. But they argue that these are the exceptions that mark the boundary: learning happens where the ego is not engaged, or where the cost was so severe that defence was no longer possible. In the ordinary run of mistakes, which are the ones the question concerns, what people mainly acquire is a better account of why the mistake was not really theirs.

Premises

Counter-arguments

Proponents reply that the argument counts only the failures. Repetition is visible and salient, while learning is invisible precisely because it works: the mistakes a person does not make again leave no trace, so the record the argument appeals to is systematically biased toward error. They also argue that the defensive tendencies described are real but partial, and that they coexist with revision rather than preventing it, as anyone who has quietly changed a habit after an embarrassment can attest. On the institutional point, proponents argue that the unit of analysis has shifted: an organisation can repeat an error while every individual in it has learned, because the people who learned are not the people who decide.

Rejecting the premises

[Rejecting P1] Repetition is visible while successful learning leaves no trace, so a record of repeated errors is biased toward failure by construction. [Rejecting P2] These tendencies are real but partial, and they coexist with revision rather than preventing it. [Rejecting P3] Treating the clear cases of learning as exceptions is a choice of description rather than a finding; they are equally well read as the mechanism working. [Rejecting C] Institutions repeating errors is compatible with every individual having learned, since those who learned are often not those who decide.