The Ladder of Inference

The Ladder of Inference is a model developed by Argyris & Schon* to explain how we all make inferences and reason about what is happening to us in the world. It’s value as a tool is not that it is necessarily correct, but that it is helpful in guiding understanding of how we work, and can help change behaviour in order to have more productive conversations in future.

Description of the rungs on the Ladder of Inference

The ladder stands upon the constant stream of data that we experience in the world. There is too much for us to pay attention to, so we choose to select some of the directly observable data that we choose to select. It is useful to think of directly observable data as what a video camera would see and hear.  Our beliefs, values and experiences often act like filters that determine what we are aware of (think of the experience of buying a new car or phone and becoming highly-sensitised to noticing them)

The next rung of the ladder is to describe the data selected in your own words, applying meaning and labels. The meaning may be based on our personal or cultural experiences. This step is where we ask ourselves “when they say or do something what does it mean to me?”

The next steps on the ladder relate to assumptions (take for granted that something is true without verifying it). We also make inferences by coming to conclusions about what we do not know on the basis of things we do know.

The third rung is where we explain what we’ve selected and described. This is where we create a causal explanation for what is going on. We start to infer the reasons why the other person has said or done something. Words like “she’s doing this because or “the reason he did that is …” are signs that we are at this step.

The next rung is where we evaluate how we judge the other person’s behaviour relative to our beliefs and values. We might focus on how effective someone else’s behaviour or whether it is is positive or negative. We evaluate the behaviour we see, but often also the other person’s intent (which is very difficult for us to be certain about).

The final rung on the ladder involves us deciding whether or how to respond and propose action. In ineffective meetings you may see two people trading suggestions about the solution to a current problem without understanding the motives, reasons or benefits behind the suggestion. This is often referred to as “duelling ladders” or “point-counterpoint conversations”

Ladder of Inference for Blog

Ways you can use the ladder of inference to increase your effectiveness

The ladder of inference is useful at focussing on several factors which can inhibit our effectiveness:

  • We are often not aware of how quickly we make assumptions and inferences because we make them so quickly and effortlessly. Whilst it’s essential for us to make assumptions it can be useful to slow this process down, become more aware of how we are operating and act with this awareness (by being open to the fact we can jump to conclusions and inviting other’s to help by pointing this out)
  • As we are not aware of the assumptions and inferences we are making, they can feel like facts to us, which means we don’t treat them as hypothesis and instead act on them as if they are true.

The Ladder of Inference can be helpful in multiple ways.

Reflect on misunderstandings
Often when we are in disagreement with someone it is because of a difference in the meaning or assumptions that we have arrived at. However, we often fail to see that we have applied different meanings or assumptions – we believe that what we’ve experienced is the only possible experience! Slowing down the inferential steps we have taken up the ladder can highlight the source of disagreement in a way that may make it possible to make more productive progress.

Being clearer in our communication
When we are high on our ladder it increases the chance of misunderstanding and confusion. Being clearer about our own reasoning (by making public each rung on the ladder) reduces the chances of misunderstanding and confusion. When combined with an invitation for another person to inquire into our reasoning, it also makes it possible for us to find out errors in our thinking, which can lead to learning.

Being clearer on future actions we’d like, from ourselves or others
It suggests more effective ways of proposing action in future. We can assert how we’d like others to think or act in future by being more concrete (what we’d like to see them say or do) rather than abstract (“be stronger”). For example, we might say “I’m unsure what you mean when you said I had a ‘knee-jerk’ reaction. Could you help me by recalling what I said or did that made you think I reacted like that?”

Improve the quality and effectiveness of our own reasoning
It helps us reflect on the quality and effectiveness of our own thinking. A common experience is to find that we are high on our ladder, with an evaluation and judgement about someone’s behaviour or intent (in which case it is referred to as an attribution) without realising that we have low-quality input data. Further, high level inferences are often not possible to test publicly. For example, if I think that you’re incompetent, I wont be able to test it by saying “Gary, are you an idiot?”. It’s unlikely that Gary could provide an answer that we’d trust to this question. “Walking down the ladder of inference” helps us identify what it is about Gary’s behaviour that might lead him to this view in way that can be more easily tested (“Gary, when I asked if you were aware of the safety rules you said ‘yes’, but I see now that you are not wearing  a safety hat. This leads me to think that you’re acting in ways that don’t make sense to me").

It can help produce actions that are less likely to lead to defensiveness in others
It suggests ways of acting that are less likely to increase defensiveness in others.  Not only are lower-level inferences easier to publicly test, but they are also less likely to lead to defensiveness in others.  Sharing a high level evaluation is likely to make people defensive (“I think you are an ineffective communicator”), whereas using the ladder of inference and starting from the lower rungs (combined with testing) allows you illustrate your reasoning in a way that allows the other person to understand and challenge any inconsistencies in your logic (“I noticed that people were focussing on other things and appeared distracted while you spoke. I don’t think they were listening to you. If your goal was to communicate with them, then I didn’t think your behaviour was effective at achieving it”)

* The Ladder of Inference is based on the work of Argyris & Schon (1974, 1994), Argyris, Putnam and Smith (1985), Peter Senge in the 5th Discipline and Roger Schwarz (2002)


5 responses to “The Ladder of Inference”

  1. Carlo Pecchia Avatar

    Really interesting… Working with people we should understand more and more stuff related to human beings rather than (mainly) on technical stuff (software). Other IMHO underexplored field is Behavioural Economic and how it’s related to humans…

    1. benjaminmitchell Avatar

      Yes, behavioural economics is great too! As are cognitive behaviour therapy, Kegan & Lahey’s Immunity to Change, Evolutionary Psychology etc. It’s a great time to be alive!

  2. Nicolay Worren Avatar

    thanks for writing a useful and original blog combining action science with other approaches such as Lean and agile development. One comment regarding the Ladder of Inference, most of your examples suggest ways of moving downwards on the ladder, yet I assume a facilitator should also sometimes prod people to move the other way – for example, to infer general criteria from specific observations.

  3. […] it’s a future, current or past project being reviewed. It’s loosely related to the ladder of inference and Maslow’s hierarchy of needs. Evgeny Rimonov was a professor of computer science who […]

  4. […] the Work? Conversations for Double Loop Changes in Mindset One important takeaway here was the Ladder of Inference. The five rungs of this ladder are: select, describe, explain, evaluate, propose action. A […]

Leave a Reply

Your email address will not be published. Required fields are marked *