Source
Argyris (2002), LIT-745, read in NOTE-561. His 1977 HBR article, LIT-739, is unread.
What was actually shown
Argyris (LIT-745) states the theory and illustrates it with one workshop.
- The distinction. Learning is detecting and correcting error. Single-loop correction keeps the governing values; double-loop correction changes them.
- The mechanism. People’s espoused theories vary, while the theory they act on, Model I, does not. Model I aims at unilateral control, winning and suppressing negative feelings while appearing rational. It produces evaluations and attributions that are never illustrated or tested. Because it is skilled, it is unaware of itself. Organizational defensive routines (mixed messages, made undiscussable, the undiscussability made undiscussable) protect people from embarrassment and from discovering its causes. Individual and organizational routines sustain each other.
- The illustration. Thirty-four CEOs, asked to help a role-played COO who admitted his errors, gave abstract advice. They attributed his failure to learn to his closedness without testing that, and blamed him as he had blamed his board.
The theory exposes itself to refutation: Argyris says a theory-in-use is a design that should admit no exceptions, so one would disconfirm it. No independent test is reported.
What this does not say
- Not that single-loop learning is bad. Model I may be preferred for routine, single-loop work. The claim concerns embarrassing, non-routine problems.
- Not that structure fixes it. The claim is the reverse: egalitarian structures drift back to control unless individual theories-in-use change, and vice versa.
- Not independently established. Universality rests on the author’s earlier studies. The case is interpreted by the facilitator who ran it, and the quasi-experiment is proposed, not run.
Connections
- Longino (THEORY-151). Her norm of uptake requires criticism to change belief. Argyris describes how criticism fails to be taken up: it is crafted so that it cannot be tested, and its rejection is attributed to the other’s closedness.
- Drift (THEORY-144). Rasmussen’s defences erode unseen because breaches have no visible effect. Argyris adds a social route: the inconsistency is visible and made undiscussable. No relation is declared; they are different mechanisms with a similar result.
- The owner’s essay. §8.3’s recursive self-governance is double-loop learning, and §8.5’s worry that recursive revision can insulate commitments has a named mechanism here.