The dominant model of organisational culture change rests on an assumption so widely shared it is rarely examined: that change proceeds from the inside out. Change what people believe and value, and behaviour will follow.
It is a reasonable assumption. It is also, the evidence suggests, wrong — or at least wrong about the direction of causation.
What the evidence actually shows
Daryl Bem's self-perception theory, developed in the late 1960s, proposed something counterintuitive: that people often infer their attitudes from their behaviour rather than the reverse. We discover what we believe by observing what we do. The internal state follows the external act.
Festinger's cognitive dissonance research demonstrated something complementary: that behaviour inconsistent with existing beliefs produces psychological pressure to revise those beliefs. Acting differently, over time and in the right conditions, produces thinking differently.
William James, writing a century earlier, arrived at the same insight through philosophical rather than experimental means: acting as if something were true is often the most effective route to believing it. The pragmatist reversal — action first, conviction follows — has been confirmed by subsequent research repeatedly enough to be taken seriously as a design principle.
Tell people what the culture should be, and you change what they say. Design the conditions in which different behaviour is easier, and you eventually change what they believe. Only one of these produces culture change.
The organisational application
If individual belief change follows behaviour change, the implication for culture work is significant. The intervention target should be behaviour — and specifically, the environmental conditions that make desired behaviour the path of least resistance — rather than attitude or value.
This reframing has a name in behavioural economics: choice architecture. Thaler and Sunstein demonstrated that the most effective way to change behaviour is not to change minds but to change the environment in which choices are made — to make the desired behaviour easier, the undesired behaviour harder, without restricting options or issuing instructions.
In the culture change context, the question becomes: what in the current environment is making the existing behaviour the natural default? What cues trigger it? What rewards reinforce it? What would need to change in the environment for the desired behaviour to become the easier option?
These questions produce specific, environmental, testable answers. "We need to be more courageous" does not. And specific answers are what sustained behaviour change requires.
What this means for how you evaluate change
The behaviour-first inversion has a further implication for how culture change is evaluated. If the standard is belief change — measured by survey — culture change will always appear slower and less complete than it is, because beliefs lag behind behaviour. The survey measures the wrong variable at the wrong moment.
If the standard is behavioural change — measured by the specific practices identified in the design conversation and tracked across sessions — progress is visible earlier, more concretely, and in the terms that most directly matter for organisational performance.
The team that has agreed, in specific behavioural terms, to three things they will do differently in their next project meeting has made a commitment that their beliefs will eventually catch up with. The team that has completed a values alignment programme and updated their survey scores has made a statement that their behaviour may or may not follow.
Both teams might describe their experience as a culture conversation. Only one of them has started the process that produces culture change. The difference is which comes first: the behaviour or the belief.
Start with the behaviour. The belief will follow.