nucleation

LIT-745: Double-Loop Learning, Teaching, and Research

2 min read

Properties4
StatusActive — read 2026-10-05 (NOTE-561): in full, from a Word file of the published issue posted by Action Design, a practice founded by Argyris's co-authors. The owner chose to use it, since whether they may post it could not be confirmed. Worth reading as Argyris's compact late statement of the theory of action. Single-loop learning corrects errors within governing values; double-loop learning changes the values. The prevalent theory-in-use, Model I, keeps reasoning tacit and produces organizational defensive routines that are self-sealing. The empirical base is his own workshop transcripts, and the quasi-experimental design is proposed, not run.
Tagssocial-science
Cited as “paper” byNOTE-561 — Double-Loop Learning, Teaching, and Research
Cited as “source” byTHEORY-145 — Organizations that espouse revising their governing values characteristically fail to, because the theory-in-use people actually act on keeps premises tacit and tests self-sealing, and the resulting defensive routines make the inconsistency, and its undiscussability, undiscussable; so errors are corrected within the governing values (single-loop) while the values themselves go unexamined

Graph View

Chris Argyris (2002), Academy of Management Learning & Education 1(2):206–218 — DOI-10.5465/amle.2002.8509400

Key takeaways

  • The definitions. “Learning may be defined as the detection and correction of error. Single-loop learning occurs when errors are corrected without altering the underlying governing values.” “Double-loop learning occurs when errors are corrected by changing the governing values and then the actions.” The example is a thermostat that asks why it is set to measure temperature.
  • Espoused theory versus theory-in-use. Espoused theories vary widely. Theories-in-use, Argyris says, do not. The prevalent one, Model I, has four governing variables: be in unilateral control, strive to win and not lose, suppress negative feelings, act rationally. It requires defensive reasoning, with premises kept tacit and tests that are “self-serving and self-sealing”.
  • Skilled incompetence. Theories-in-use are learned early and run skilfully, so people are “unaware and unaware that they [are] unaware”.
  • Organizational defensive routines. These are practices that prevent embarrassment and also prevent discovering its causes. Their logic: send an inconsistent message, act as if it is not inconsistent, make that undiscussable, and make the undiscussability undiscussable. They and individuals’ Model I sustain each other, so neither can be changed alone.
  • Model II. Its governing values are valid information, informed choice, and vigilant monitoring of implementation. It works by “productive reasoning”: premises and inferences explicit, conclusions testable by logic independent of the actor’s.
  • The case. Thirty-four CEOs advising a role-played failed COO (“Andy”) reproduced the abstract, untestable advice and outward blame they diagnosed in him. They blamed Andy as Andy blamed his board.
  • A proposed test. Part III proposes hypotheses, scoring categories, time series and comparison groups that would turn the workshop into a quasi-experiment. The paper does not report running one.

Standing in the record

Filed on 2026-10-05 at the owner’s request, in place of the unread 1977 HBR article (LIT-739), for the owner’s essay on organizational will. The essay’s §8.3 separates changing policies from reconsidering “the standards through which policies acquire authority”. That is single- versus double-loop learning, and the essay should name it. The essay also says recursive revision does not guarantee normative adequacy. Argyris supplies a mechanism for why organizations that espouse revisability fail at it: defensive routines make the governing values undiscussable. The account is THEORY-145.

Backlinks