The Ladder of InferenceHow Argyris's Warning Became a Communication Technique

In 2012 TED-Ed released a five-minute animated lesson called Rethinking Thinking. It opens in a car park. Someone takes the space you were waiting for, and in the seconds that follow you decide what kind of person they are, what they think of you, and what you will do about it. The lesson then draws a ladder with seven rungs, climbing from the facts of the scene to the beliefs you have formed and the action you take, and asks the viewer to notice how fast the climb happens. The teaching notes suggest students draw the ladder themselves, and ask a friend to say “you’re climbing” when they see it happening. The model is credited to Chris Argyris, a professor at Harvard Business School, who died the following year.

The ladder of inference is now one of the most widely taught tools in organisational life. It appears in leadership programmes, in coaching, in conflict training and in classrooms. Its lesson, in the form most people meet it, is that we jump to conclusions and should climb back down to the facts before we act. That is a useful lesson. It is also close to the opposite of what Argyris built the ladder to show.

Where the Ladder of Inference Came From

Argyris spent his career on a question he thought management could not stop evading: why organisations full of intelligent people are so bad at learning. His answer, worked out with Donald Schön across the 1970s, began with a distinction between what people say they do and what they do. Everybody carries an espoused theory, the account of their reasoning they would give if asked, and a theory-in-use, the reasoning that can be reconstructed from how they behave. The two are usually different, and most people are unaware of the gap in their own case while being sharply aware of it in everyone else’s.

The theory-in-use, Argyris found, was nearly always the same. Across cultures and industries and at every level of seniority, when people felt threatened or embarrassed they reasoned from four governing values: stay in unilateral control, maximise winning and minimise losing, suppress negative feelings, and appear as rational as possible. He called this Model I. It is efficient, it is universal, and it makes learning impossible, because learning requires the kind of exposure that Model I exists to prevent. The alternative, Model II, asks people to produce valid information, to choose freely and with full knowledge, and to state their reasoning in a form others can test. Almost nobody, Argyris found, could do this without years of practice, however sincerely they espoused it. He was candid that many of his own interventions failed.

The ladder came out of that practice. Argyris and his colleagues used it through the 1980s as a way of mapping how a person gets from what they observed to what they concluded, so that the route could be laid out in a room and tested. In its original form it had three rungs: the directly observable data, the words said and the things done, which a camera would record; the culturally shared meaning of that data, which most people present would agree on; and the meaning the individual added, which was theirs alone. The point of the exercise was to find the lowest rung on which two people in disagreement could still agree, and to climb from there together, testing each inference as it was made. The ladder was a device for making private reasoning public.

Peter Senge’s Fifth Discipline took Argyris’s thinking into the mainstream in 1990, and the Fieldbook that followed in 1994 drew the version that has travelled ever since: seven rungs, from observable data up through selected data, added meanings, assumptions, conclusions and beliefs to action, with a loop running back from beliefs to selection, because what we already believe decides what we notice. It is a good diagram. It is also where the tool began to come loose from its purpose.

How the Ladder of Inference Became a Communication Technique

Read the explainers and you find a consistent shape. The ladder is introduced as a description of how people jump to conclusions. The reader is encouraged to locate themselves on it, to climb back down by asking whether their assumptions are valid and their data properly selected, and then to explain their reasoning to a colleague. One of the most-read guides calls it a way of getting back to the facts. In training catalogues it now sits beside lists of cognitive biases, one more flaw in individual thinking with one more corrective. The frame is personal and remedial. Somebody has climbed too fast, and the remedy is for that somebody to climb down.

What has been dropped is the second half of Argyris’s argument, which was about what happens when the climb is made in company. The ladder was never meant to be descended alone. Its use was to lay a person’s reasoning on the table, rung by rung, so that others could say where they stopped agreeing. That use depends on something the explainers leave out because it is not a skill. It depends on whether the organisation permits it.

Argyris’s name for what usually prevents it was organisational defensive routines. In an article for the Harvard Business Review in 1986, titled Skilled Incompetence, he described executives who were highly practised at behaviour that produced the opposite of what they intended, and in Overcoming Organizational Defenses in 1990 he set out the routine’s structure. A message is sent that contains an inconsistency: take risks, and be careful. Everyone acts as if the message were consistent. The inconsistency is made undiscussable. And the undiscussability is itself made undiscussable, so that anyone who raises it has broken a rule nobody will admit exists.

Set the ladder beside that routine and the problem shows. The seven-rung version asks the individual to examine their inferences. The defensive routine ensures that certain inferences, above all the ones everyone has drawn about what the organisation is doing as opposed to what it says, can never be examined aloud. A manager can climb down her own ladder as conscientiously as she likes. When she reaches the rung where the observable data is a leadership team that announces a value and then removes the person who acted on it, there is no colleague to whom she can safely explain her reasoning. The technique has been taught. The condition for using it has been withdrawn. And the person who then declines to climb down in public can be told, with a straight face, that they have jumped to conclusions.

The Lens has met this move before. As The Learning Organisation traces, Senge’s disciplines were made into a programme by removing the part that implicated the organisation’s own structure. The ladder underwent the same reduction. Argyris had built a tool for examining the collective theory-in-use, the reasoning the organisation would never espouse but acts on daily. What survived was a tool for examining the individual’s thinking in private, which is to say a tool that assigns the whole problem of inference to the person lowest in the hierarchy and leaves the organisation’s inferences where they were.

Whose Inferences Get Tested: The Ladder of Inference and Power

The most important question about any ladder is who is being asked to climb down it. In practice the answer is consistent. The ladder is taught to graduates and new managers, used by coaches on their clients, and brought into disputes by whoever is facilitating. It is seldom applied upwards. When a senior leader draws a conclusion from thin data, the conclusion is called judgement, and judgement is what they are paid for. When a junior person does the same thing it is called jumping to conclusions, and there is a module for it.

The tool has also learned to work against people. A coach asks a client what evidence they have for the belief that their manager is sidelining them, and the client, who has watched it happen for a year, finds that watching does not count. An HR process treats the complainant’s account of a pattern as inference and the organisation’s account of its intentions as data. Both uses are faithful to the diagram. Both locate the error in the person who made the reading rather than in what was read, and both are most effective against readings that happen to be correct. The ladder, deployed this way, becomes a method for returning accurate perceptions to their owners with a request for more evidence.

Argyris saw this and said it in a way that made him unpopular. The people least able to use the ladder, he found, were the most successful professionals, because they had rarely failed and had never learned to examine their own reasoning under threat. In Teaching Smart People How to Learn, published in 1991, he described consultants who could diagnose a client’s defensive reasoning with precision and fell into what he called a doom loop of despair the moment the same reasoning was pointed out in themselves. Expertise, in his account, does not make people better at climbing down. It makes the climb more threatening, because there is more to lose, so it is avoided with more skill.

That is where the tool meets power. Defensive routines cluster around whatever the organisation cannot afford to examine, and what it cannot afford to examine is usually a decision someone senior has made. The ladder as a communication technique asks everyone to check their assumptions. The ladder as Argyris used it would require the assumptions behind the restructure, the strategy and the definition of success to be laid out rung by rung in a room where people could say where they stopped agreeing. One of these is a workshop. The other is a political act, and the organisation knows the difference even when the facilitator does not.

There is a further difficulty, and it concerns the bottom rung. Every version of the ladder rests on observable data, the facts as a camera would record them, and treats the selection of those facts as the first and most innocent of the inferential steps. Karl Weick spent a career showing that it is nothing of the kind. As Sensemaking explores, people do not encounter raw data and then interpret it. They are selecting as they look, and the selection is governed by what they expect to find, what their position lets them see, and what the people around them have agreed to notice. The Fieldbook’s reflexive loop gestures at this. What it does not say is that the loop runs through the organisation as well as the individual. The data that reaches a chief executive has been selected on its way up by everyone who understood what she wanted to hear, so that by the time she stands on her bottom rung and looks at the facts, the facts have already climbed several rungs of other people’s ladders. Argyris made the same point from another direction in 1994, in an article called Good Communication That Blocks Learning: the surveys and focus groups through which leaders believe they are gathering data mostly gather single-loop information, and teach employees that their role is to criticise management rather than examine their own part in what is wrong.

What the Ladder of Inference Leaves Out

The ladder is a diagram of one person reasoning. Its subject is a single mind moving from observation to action, and its moral is that the mind should move more carefully. What the diagram cannot show is that inference is also something groups do together, and that groups maintain their inferences by means the individual has little access to.

Argyris’s left-hand column exercise, which he used alongside the ladder, asked people to write out a difficult conversation in two columns: what was said on the right, what they were thinking and did not say on the left. The left-hand column is where the organisation’s real theory-in-use lives. It is full of inferences everyone has made and nobody will state: that the consultation is for show, that the new value is cover for the old behaviour, that the leader does not want to know. These are not errors. Many are accurate, and the people who hold them are not jumping to conclusions. They are reading the organisation correctly and keeping the reading to themselves, because to say it aloud would be to name the undiscussable, and the price of that is belonging. The left-hand columns, taken together, are the culture in its truest form, the shared meaning the organisation lives by and will never write down. Everything in the values statement is the right-hand column.

That reverses the usual lesson. The explainers warn that we climb too fast and act on beliefs we have not tested. The more common organisational condition is a surplus of accurate inference with nowhere to go: people have climbed carefully, on good evidence, and arrived at conclusions they are not permitted to act on or say. An organisation can hold this condition for years. Everyone’s left-hand column says the same thing, everyone assumes they are alone in thinking it, and the public conversation proceeds on an espoused theory that nobody in the room believes. Argyris called the result self-sealing. It is also an exact description of what unknowing looks like when it is built from things that are known.

The ladder cannot be blamed for this. It was designed to open these systems, and in Argyris’s hands, over months with the same executive team, it sometimes did. What it cannot do is open them from below, one person at a time, in a culture that has decided certain rungs are off limits. Teaching the ladder in that culture produces a recognisable outcome. The organisation acquires a shared language for checking assumptions, uses it on the assumptions it is comfortable checking, and credits itself with openness. The AI Transition Diagnostic asks, in its section on unknowing, what the most common response is when someone says they do not know how a change will affect their role, and offers among the options that people have stopped asking. That is what a sealed system sounds like. It is fluent, and everyone in it has an excellent command of the ladder.

Argyris’s own view was that the gap between espoused theory and theory-in-use could be closed only by people willing to have their reasoning tested in public and to be shown wrong, and that very few people, and almost no organisations, were willing. He did not think the ladder was a solution. He thought it was a way of seeing the problem clearly enough to be embarrassed by it, which was, for him, where learning began. The tool that bears his name has been made comfortable. It no longer embarrasses anyone. Whether that counts as success depends on what you think the ladder was for, and on which rung of it the people in your organisation have agreed, without ever discussing it, never to stand.