Single-loop learning corrects errors within a fixed frame. Double-loop learning questions the frame itself. The distinction comes from Chris Argyris and Donald Schön’s work on organizational learning (Argyris & Schön, 1978).
A thermostat is single-loop. It detects that the temperature has drifted from the set point and activates the heater. The correction works because the goal — maintain 68°F — is assumed. The thermostat never asks whether 68°F is the right temperature for a room that has changed its use, its occupants, or its season. It regulates within a frame it cannot examine.
Double-loop learning operates on the frame itself. An organization practicing double-loop learning does not just ask “are we meeting our targets?” but “are these the right targets?” Not just “is this process working?” but “should we be doing this process at all?” The first question stays inside the governing variables; the second exposes them to revision.
Argyris observed that most professionals are effective single-loop learners and poor double-loop learners (Argyris, 1991). They have been trained to solve problems within accepted frames and have rarely been required to question those frames. When the frame itself is the source of the problem — when the categories have outlived the situation they were built for — single-loop competence becomes a liability. The system corrects with increasing precision toward a goal that no longer serves.
The connection to cybernetic feedback is direct. Negative feedback is single-loop: it reduces deviation from a reference state. But negative feedback cannot change the reference state. That requires a different kind of operation — one that takes the system’s own governing variables as objects of adjustment rather than fixed parameters.
Second-order cybernetics names a related shift: from observing a system to observing the process of observation itself. Argyris and Schön are working at the organizational level — how institutions learn or fail to learn — but the structural parallel holds. A system that can only correct within fixed parameters is operationally alive and structurally static. A system that can revise its own parameters has a capacity that the first system lacks, regardless of how efficiently it operates within its frame.
Related terms
- Cybernetic feedback — the mechanism single-loop learning enacts
- Second-order cybernetics — the reflexive turn that makes frames visible
- Requisite variety — the constraint on how much corrective capacity a system needs
- Autopoiesis — self-producing systems that maintain organization while replacing structure