Skip to content

Chapter 10 — Mental Models

Core Thesis

Good ideas fail to reach practice not from weak will but because they collide with tacit mental models — deeply held internal images that shape both what we do and, more insidiously, what we're even able to see. Chris Argyris's distinction is the spine of the chapter: people behave consistently with their theories-in-use, not their espoused theories — what they actually do, not what they say they believe.

Key Episode

Detroit's Big Three operated for decades on an unexamined assumption set (per consultant Ian Mitroff): styling matters more than quality, the US market is isolated from the world, workers don't affect productivity. The assumptions had been true — for a limited period — and were mistaken for permanent fact; German and Japanese automakers "educated" American consumers otherwise and took the US market from near-zero to 38% share by 1986. Royal Dutch/Shell went the opposite direction: planner Pierre Wack, foreseeing OPEC-style discontinuity in 1972, discovered that handing managers accurate forecasts changed nothing — "our scenarios would be like water on a stone" — until Group Planning redesigned scenarios to force managers to see which assumptions their own "trouble-free future" required, then walked them toward a new mental model. Shell went from "Ugly Sister," weakest of the seven oil majors in 1970, to arguably strongest by 1979 — the only major that responded to the 1973 embargo by slowing refinery investment and decentralizing rather than centralizing control.

The Mechanism

Two reflection skills anchor the discipline. Leaps of abstraction: the mind moves from concrete observation to sweeping generalization almost instantly and then treats the generalization as fact — "Laura doesn't care about people" (built from: rarely praises, stares off, cuts people off) becomes unquestionable, and nobody ever asks Laura, who turns out to have an undisclosed hearing impairment. The left-hand column: writing out an actual dialogue on the right of a page and what you were really thinking but didn't say on the left exposes the private, untested assumptions ("he lacks confidence... he lacks initiative") steering a conversation nowhere — Senge's own worked example with a colleague named Bill shows both people talking around a failed presentation instead of naming it.

Balancing inquiry and advocacy is the corrective: pure advocacy between two people escalates exactly like an arms race (Ch. 5) — each side's forceful case pushes the other to argue more forcefully, positions harden, nothing is learned. Reciprocal inquiry — stating your view and the data/reasoning behind it, then actively inviting others to find the gaps — replaces "winning the argument" with "finding the best argument."

Institutional Practice

Shell institutionalized mental-model work through scenario planning treated explicitly as "planning as learning" (Arie de Geus): the value of a plan isn't its accuracy but the institutional learning generated by making it. Hanover Insurance built a network of "internal boards" — two to four senior managers advising local general managers, explicitly not to control decisions but to test and extend the GM's own mental model — governed by a ten-point credo whose second principle warns against imposing "the authorized mental model": the goal is not congruency, though congruency often results when the process works. Hanover's three-day "Merit, Openness, and Localness" seminar, built with Argyris and Lee Bolman, videotapes managers role-playing — "it is often hilarious to see how much our own behavior deviates from what we say we do."

The Shift

From treating mental models as either invisible or as unchangeable "facts" to treating them as testable, examinable, improvable working assumptions. Senge's closing metaphor: systems thinking without mental models is the DC-3's engine without wing flaps (Ch. 1) — necessary but insufficient. Mental models expose the assumptions; systems thinking restructures them once exposed.

Key Terms

  • Mental model — deeply held, often tacit internal image that shapes perception and action
  • Espoused theory vs. theory-in-use (Argyris) — what we say we believe vs. what our actions actually imply
  • Leap of abstraction — untested jump from observed particulars to treated-as-fact generalization
  • Left-hand column — technique exposing unspoken assumptions steering a real conversation
  • Balancing inquiry and advocacy — stating your reasoning and data while genuinely probing others' reasoning, to find the best argument rather than win

Connections