Single-loop and double-loop learning

There are two ways to respond to a problem: fix the action, or question the assumption behind it. The difference, single-loop versus double-loop learning, separates tinkering from real change.

Single-loop and double-loop learning, a distinction drawn by Chris Argyris, describe two depths of organisational learning. Single-loop learning corrects actions within existing assumptions and goals; double-loop learning questions and revises the assumptions and goals themselves. The first asks are we doing things right; the second asks are we doing the right things.

Adjusting within the frame

Single-loop learning is the more common and comfortable kind. When something goes wrong, the response is to adjust behaviour to get back on track, without questioning the underlying objectives or beliefs. A thermostat exemplifies it: it detects that the room is too cold and turns up the heat, never asking whether the target temperature is right. Most organisational problem-solving is single-loop: tweaking processes, correcting errors, improving performance against goals that are taken for granted.

Questioning the frame

Double-loop learning goes deeper, turning attention back on the governing assumptions, values, and goals that shaped the action in the first place. It asks not only whether the action achieved the goal but whether the goal, the strategy, the very framing of the problem, was right. This is harder, because it requires confronting beliefs that people are invested in and that often go unexamined. A firm that keeps improving a product no one wants is doing single-loop learning well; double-loop learning would ask whether to make that product at all.

Why the deeper loop is rare

Double-loop learning is rare because it is threatening. Questioning governing assumptions exposes the possibility that the organisation, and its leaders, have been wrong about something fundamental, which triggers defensiveness. Argyris found that organisations build elaborate defensive routines that protect existing assumptions from challenge, so that genuine double-loop learning is actively, if unconsciously, resisted. The result is firms that are skilled at solving the wrong problems efficiently.

The distinction is valuable because it explains why organisations can work hard at improvement and still fail to adapt: they are doing single-loop learning when their situation demands double-loop. Lasting change usually requires the harder, more uncomfortable loop, the willingness to ask whether the assumptions everyone has been operating on are still true.