Monday, February 20, 2012

A Co-operative Model for Understanding Human Behaviour

What does it mean, when the modeller and modelled become a part of the model?


Don't get tongue-tied just yet. I was just as confused as you might be when I first heard Mihnea Moldoveanu, a professor from the Rotman School of Management in Toronto, speak about his new approach to modelling human behaviour, way back in the Fall of 2011. After digesting what he had outlined, however, and reading through his accompanying book - "Inside Man: The Discipline of Modelling Human Ways of Being", I gained a much better understanding of what he was getting at. In fact, the Co-operative Model for Understanding Human Behaviour that he was proposing follows in the exact same footsteps as Lean Thinking and Design Thinking - both of which I've covered extensively. So, although it was no surprise when I saw the same principles being applied to the study of human behaviour - it was no less interesting to hear about how it worked.




What's a Model?

Let's start at the beginning. Why try to model or understand human behaviour in the first place? The answer is simple: if we can build an understanding of how people behave, then we can predict how they might behave in the future, and in turn, influence it.

Models then, in turn, help us to frame that understanding in a way that other scientists can learn from and use. They are representations of "reality" - of behaviour or thought - that can be used to take us from a set of known or observed variables (like past choices, past thoughts and impressions) to a set of predictions about the future. Simply said, models explain and predict. To help with predictions, these models contain algorithms and formulas that help produce outputs (the predicted behaviour) from inputs (past behaviour). They help us "make inferences about what we do not know on the basis of what we do know".

Although mainly used in the social sciences, like Anthropology, models for understanding human behaviour are extremely valuable to marketers. After all - isn't the entire discipline of marketing about influencing consumer behaviour?




One-way Modelling and the "Old Science"

Traditionally, models in science have predominantly been one-way. What that means is that the scientist who is developing a model about why people behave the way that they do is never truly in close contact and co-operation with the people whom they are modelling.


In his book, Moldoveanu describes the aspects of the three types of traditional scientists:

Speculators: ie. Galileo, Newton, Einstein. They generate new distinctions and introduce new concepts
that are useful in describing, manipulating, and/or creating phenomena.
Calculators: ie. Laplace, Penrose, Thorne. They use the basic schemata articulated by speculators and perform the necessary logical and computational work that takes concepts to testable models.
Experimenters: ie. Kepler, Eddington, Penzias. They create effects in the laboratory and the field that are based on distinctions articulated by the speculators and sharpened by the calculators.

He also describes the three types of "Old Science" and their approaches to modelling human behaviour:

Descriptive Science: What is the case? Informs us about how people respond to certain actions and situations.
Normative Science: What should be the case? Informs us what logically and objectively the response would be to certain actions and situations.
Prescriptive Science: What should be done, given what we know about the case? Tells us how people should respond to an action or situation given what we know or what they know.


Ultimately, because of the way in which traditional scientists build models, and the way in which "Old Science" approaches process of modelling, they end up thinking of those being modelled as an "ideal type", "statistical ensemble", or a "norm". This really just means a stereotype. In order to keep models simple, they strive to avoid understanding people as individuals, with their own desires, beliefs, and reactions to the process of being modelled. They reduce complicated human beings into ideal agents that they can then fit into their models. This can lead to, to put it bluntly, bad predictions.




Co-operative Modelling and "Ascriptive Science"

The approach that Moldoveanu is proposing, however, is much different. Calling it "Ascriptive Science", his approach is a co-operative, live exercise in modelling human ways of being, through ongoing interaction between the modeller and the person being modelled.


Here are three ways in which this model is different:

(1) Modelling human behaviour, thoughts, and feelings is an interactive process between the modeller and the modelled. In a way, we're back to the idea of Customer Co-creation: the individual whose behaviour is being modelled is available to the modeller for questioning, deliberation, debate, and even confrontation; and the modeller will often need to use this opportunity for dialogue in order to better understand the behaviour of the modelled. In fact, the ongoing feedback cycle between the modelled and the modeller can lead to the production of new behaviour and new models to investigate.

(2) It treats the modelling languages and framework that it's working from as fundamentally unfinished and the modelling enterprise as an iteratively and self-correcting activity. Once again, we're back to the idea of Kaizen - of continuous improvement based on ongoing feedback. The model is no longer static, but a living, changing, evolving organism.

(3) It changes the subject being modelled. The models emerge as an interventional device and not merely a representational one. They can question the modeller, help to define the model, and even change their own behaviour in response to it. They aren't behind a one-way mirror - they're in the same room.


This approach, although some may see it as messy, ends up freeing the modeller to be more flexible in how they approach the process of modelling behaviour. The modeller is now free to:

  • Create models that answer the specific kind of questions they’re interested in
  • Interact with the subjects of the model
  • Force them into answering certain kinds of questions
  • Induce them to interpret themselves, to elaborate, and alter their behaviour

Unlike in the "Old Science", the modeller must deal with individuals in the present - the here and now, whose here-and-now predictions matter, and matter a lot. Experimenting, interrogating, training, and calling the modelled to task are all part of the process of Co-operative Modelling. The modeller says to the person being modelled, "help me understand".


This results in a type of scientist, the "Ascriptive Scientist", that is a hybrid of the three types of traditional scientists. Ascriptive Scientists draw on the core skills of each of the Speculators, Calculators, and Experimenters, and engage in the Normative, Descriptive, and Prescriptive behavioural sciences - all at the same time. As Moldoveanu explains it, "He speculates in that he uses and refines the basic distinctions made by Normative Science to put together structured models of behaviour, which he uses calculatively to make predictions about a subject’s behaviour and thinking, which he then tests by designing real-time experiments whose results he can then use to modify or fortify his speculative models."



Real-time experiments, customer co-creation, an iterative approach to developing new solutions - sound familiar? Yes, it's amazing how much the aspects of both  Lean Thinking and Design Thinking  are permeating every facet of both our lives and different industries - even behavioural science. The key takeaway from Moldoveanu's new approach - when you're looking to "model" your consumer's behaviour (that is, to understand, predict, and influence it), be open to doing it the co-operative, or Ascriptive way. Interact with the people whom you're modelling, involve them in the process of model-building, and be open to experimenting and testing that model as it evolves.

That's what it means when the modeller and modelled become a part of the model. Still tongue-tied?

No comments:

Post a Comment