B08

card_id: B08

input: An organization's observed outcomes/errors relative to its goals, and the (often tacit) "governing variables"/assumptions that its members' action-strategies are designed to satisfy.

operation: Detect a mismatch between intended and actual outcomes (error). In single-loop learning, correct the action strategy while holding goals/governing variables fixed (the thermostat analogy: turning heat on/off without questioning the 68-degree setpoint). In double-loop learning, additionally question and, where warranted, revise the governing variables/underlying policies themselves (the thermostat asking whether it should be set to 68 degrees at all) — a genuine assumption-reopening trigger fires when errors recur or cannot be resolved within the existing frame.

output: Either a corrected action within an unchanged policy frame (single-loop) or a revised policy/goal/governing-variable set (double-loop), plus, per the paper's case material, the surfacing of previously undiscussable organizational norms that were blocking the reassessment.

scope_conditions: Organizational settings where errors are being actively hidden or filtered as they move up a management hierarchy, and where "Model I" defensive behavioral norms (win, don't upset others, suppress negative feelings, define the situation unilaterally) are operative.

failure_conditions: The paper's central case (a >$100 million product failure known internally six years before cancellation) shows double-loop learning failing to trigger because of nested self-reinforcing norms ("do not confront policy," protected by "do not confront the norm against confronting policy"), producing a "double bind" for employees who must choose between hiding and revealing errors.

predicted_effect: Organizations that cannot double-loop learn accumulate hidden, compounding errors; asserted causally from the paper's case narrative, not from a controlled study.

measurement: No quantitative instrument verified in the sections read; the paper references a scenario-exercise method (23 paired cases contrasting what people would say vs. what they would communicate) without including its detailed results in the excerpt read.

implementation_test: Given a recurring error, ask whether the response only adjusts action within existing goals (single-loop) or also revisits the governing variables/policy that produced the situation (double-loop); the organization fails the test if avoidance norms prevent the second question from being asked at all.

decision_rights: Not centrally assigned; the paper frames double-loop learning as something enabled at the level of interpersonal/organizational norms (a "Model II" alternative to Model I), not a single decision-maker's authority.

uncertainty_handling: No formal uncertainty mechanism; the governing variables themselves are the thing reopened under evidence of persistent error — the closest analogue in this source to a prospective reassessment trigger.

verification: Not a formal verification procedure; the paper relies on comparative case analysis (its Product X case and the referenced 23-case scenario exercise) to demonstrate the single-loop/double-loop distinction.

provenance: OL01 pp.115-118.

implementation_cost: Not quantified.

limitations: Single-source card (operational_coverage=partial); no independent second source verified. The fuller original treatment (Argyris & Schön, 1978, Organizational Learning: A Theory of Action Perspective) was not retrieved or read in this pass.

statement_basis: direct: ["thermostat single-loop/double-loop analogy", "four Model I governing variables", "Product X case narrative"] inference: ["'prospective trigger' framing in the operation field is the curator's characterization of the source's own error-driven reassessment mechanism, not the source's own terminology"]