Is behavioural science using the wrong model?

A recent blog challenges the underlying theories that shape the discipline: the author has a point

“We don’t have a hundred biases, we have the wrong model.” 

So said Jason Collins in a recent blog, perhaps somewhat provocatively likening the use of biases as akin to the activity of ancient astronomers who were required to compile an exhaustive number of deviations to retain the broken model of the universe revolving around the earth. Collins challenge is whether the model at the heart of behavioural science is similarly broken.

Applied behavioural science has taken off in the last decade or so and is now rapidly developing into a healthy and growing industry spanning both the private and public sectors.  And yet, even some of the most vocal advocates of the use of biases and nudge are now starting to propose a more progressive understanding of human behaviours and ways to change it. Are we at a tipping point where the fundamentals of the industry are now being renegotiated?

It certainly feels that the time is right for Frontline to take a plunge into conceptual waters as we question the underlying foundations of behavioural science and set out why this is not simply an arcane theoretical discussion but has very practical consequences for people working in the field. 

The backdrop

The behavioural science industry has traditionally oriented itself (albeit loosely) around the area of behavioural economics, relying on the work of cognitive psychology to identify ‘biases’ in human behaviour and seek to use nudges as solutions to help deliver positive outcomes.  Collins suggests that the underlying theoretical basis for behavioural scientists is the rational-actor model of economics in which people make decisions based on their preferences and the constraints that they face.  Behavioural economics offers the many deviations from this rational-actor model but in doing so, de-facto retains it at the heart of its discipline. In this way, biases are the equivalent of the deviations that astronomers used to protect what was increasingly apparent, that the underlying model was simply wrong.

Not everyone agrees with this of course. While as he points out Richard Thaler takes the  view that it is possible the rational-actor model is as good as it gets, others may challenge whether the rational-actor model informs their practice when it can feel the day job is exactly the opposite of that, a debunking of the rational choice model.  Instead, some are surely more aligned with Behaviourism as the theoretical underpinning, referencing a body of work which discounts the importance of internal states in favour of the notion that behaviours are acquired through the rewards and punishments meted out by environmental stimuli.  Explanations based on ‘habit’ or choice architecture have in common principles with Behaviourism, relating to the primacy of mental automaticity and the importance of environment shaping our behaviour.

Of course not everyone will agree with this analysis and others may well be refer to other conceptual underpinnings but these seem, to us at least, the most significant ones, even if these are not always explicitly acknowledged by the practitioner community.

Arguably, this ‘behavioural economics’ approach has become the ‘traditional’ (if we can use that terms for a young profession) toolkit for much of behavioural science.  But there is no reason why this should in fact define the discipline. Perhaps the time has come to make a progressive leap for two main reasons: one is more conceptual in nature, the other more empirical.  There is also a consideration of the degree to which the traditional models underpinning the profession actually helps the practitioner in tackling problems.  We set these out in turn now.

The conceptual case

The conceptual argument can be illustrated by the work that has been done on inattentional blindness.  The traditional school of behavioural science would consider that our apparent inability to see things is a form of deficit, resulting in irrational judgements and decisions: people are operating in a manner that is too much for our limited processing capacity, so we use mental shortcuts to assimilate the world. 

An example of this is inattentional blindness due to ‘priming’, meaning when we are unknowingly prompted to be sensitive to particular stimuli in the environment. In what was heralded at the time as a ground-breaking piece of work, psychologist John Bargh, found that people primed with words conventionally related to age in the United States such as ‘bingo’, ‘wrinkle’, ‘Florida’ walked more slowly than the control group as they left his lab, as if they were older. Priming allows us to navigate the environment in a good enough way, without having to expend an inappropriate degree of deliberative energy.

This has fed through into popular imagination, such as the film ‘Focus’ where Will Smith’s character spent the day “priming” the victim to subconsciously recognise and choose the number 55 by having it represented all around him. It appears on the lapel of the hotel door man, in light fixtures, on a poster in the elevator, on shirts people on the street are wearing, and on a Rolling Stones track playing in the background. So, the film suggests, by the time the victim actually needs to choose a number, the choice has already effectively been made.

So what is the conceptual challenge to this?  We can unpack this by looking at the classic Simon and Chabris study where they show a person in a gorilla suit walking in a film sequence who is very frequently missed because the participants in the study as they have been ‘primed’ to count the number of basketballs passes. Kahneman uses this as an example of how we can be ‘blind to the obvious’.

But there is a different conceptual argument that offers an alternative explanation for the very same behaviour: Teppo Felin suggests we can think of the ‘primes’ as the equivalent to questions or theories that direct our awareness to those features of our environment among the huge array of things that we could choose to look at. In that experiment we could be asked to look at all manner of things from the hair colour of the actors to the gender and ethnic composition. Any of these are clear but only if you are looking for them and not something else. The fact that we miss some is not a function of blindness or bias, but an entirely sensible and successful process given what we set out to do. These reflect the way we naturally direct our perception and awareness: priming is a way of describing this act but attributing a very different explanation.

The empirical case

Not only is the case for priming conceptually questioned but the research data does not firmly support it.  Bargh’s original studies on priming have been subject to a deal of discussion and controversy concerning the failure to replicate his original studies.   In a similar vein, loss aversion, one of the pillars of much traditional behavioural science has run into significant difficulties.

The discussion concerning replication will clearly continue to run on but suffice to say, given the core set of effects used to understand behaviour are looking shaky, then aligned with the points that Collins makes, it perhaps feels a little uncomfortable place to be placing this model at the front and centre of a behavioural science practice.

The practitioner case

Setting all that aside, a practicing behavioural scientist needs a set of tools that offer tools which can ‘diagnose’ the mechanisms underlying behaviour alongside a toolbox of possible solutions.  The challenge is that if automaticity is given primacy then diagnostic tools have problems  and the solutions are, to put it mildly, challenged.  Furthermore, as Collins sets out, the lack of a theoretical framework means there are few guiding principles to help shape practice:

Suppose you are studying a person deciding on their retirement savings plans. You want to help them make a better decision (assuming you can define it). So which biases could lead them to err? Will they be loss averse? Present biased? Regret averse? Ambiguity averse? Overconfident? Will they neglect the base rate? Are they hungry? From a predictive point of view, you have a range of countervailing biases that you need to disentangle. From a diagnostic point of view, you have an explanation no matter what decision they make. And if you can explain everything, you explain nothing.”

One means of addressing this has been to use more holistic frameworks to behaviour, such as COM-B or ISM which have gained widespread traction in the public sector at least. Arguably the adoption of broader ‘behaviour change’ frameworks such as COM-B were a reaction to the traditional approaches in the industry.  They do a good job of very pragmatically pulling together different bodies of work to represent a range of theories that can be used to diagnose behaviour ‘in the round’.  As a practitioner tool we certainly find these frameworks are indeed very helpful as a ‘first principles approach’ to understanding behaviour and then offering guidance on finding ways to shape outcomes.

But of course while these frameworks pull in different theories, their formulation is in itself atheoretical. COM-B is based on a forum of leading psychologists making pragmatic considerations of the elements that should be integrated rather than the result of wider overarching theorising.  No harm in this at all (we often do the same) but we cannot ignore that there are inevitably a wide range of judgements that need to take place in any applied work.

As such theory remain important. If activity is not guided by an explicit use of theory then there is a danger of falling back on unexamined (implied) theories of human behaviour which are no less influential simply for not being made explicit. 

So if the challenge that Collin’s has highlighted has merit, what are we to do? Should we now be searching for a new ‘grand theory’?

The quest for a progressive behavioural science theory

Collins suggests that we are now at a place where:

Rather than experiments that allow us to distinguish between competing theories, we have experiments searching for effects.

Should we therefore be doing the work to develop a new underpinning for the industry? A grand ‘theory of everything’ that can be used to define the way the industry understands behaviour and drives development of measurement tools and intervention toolkits? 

Whilst this might seem tempting, this may be in danger of making the mistake of assuming that human science is akin to the physical sciences, that one theoretical model can explain observations. We suggest that this is simply not viable, things are way too messy. Philosopher of science Mary Midgely might have suggested that we can answer this question by thinking about the way we use an atlas: if we have a question about population we look at one page in an atlas, a question about topography a different one, political geography another and so on.  We do not expect one map to answer all questions for us.

This sounds remarkably close to practitioner good practice:  instead of selecting a theoretical framework and seeking to prove or disprove that theory, the whole process is switched.  What we do is to place the question in front of us and then ask what theory (page in the atlas) is most helpful to us, and then draw on the relevant body of knowledge and measurement tools. Of course understanding what body of theory relates to which class of question is a big question (and defining the class of question is not without controversy.)

In practice what can happen all too often is that practitioners do in fact put the theory first – albeit this is implied rather than made explicit.  Someone who places behavioural economics at the front of their practice is de facto giving primacy to theoretical traditions relating to automaticity. And of course this is perfectly respectable as long as all parties understand it is one of many possible ways to look at the question. 

The environment we live in now is much more pluralistic than in the past:  there has been a decline in a single agreed understanding of the world as represented by data presented by a small group of experts.  As such, understanding broad schools of thought relating to human behaviour is surely the more progressive position to be taking. 

Any individual may move between these schools but perhaps more likely people coalesce around particular schools that align with their world views and skill sets. One school, for example, may be more aligned with socio-cultural models of behaviour, others around computing, political based explanations or geographical.  The reality is that the question that needs answering will then determine which of these is most useful to answer it.

Conclusion

That science is always a work in progress and things are not fixed is well understood in the community. How science progresses is the subject of intense philosophical debate with Karl Popper proposing that science progresses by a process of falsification; theories for which predictions conflict with experimental outcomes are discarded, and science then progresses as a process of elimination. The challenge here is that cultures can develop around particular perspectives which, as we have seen from the publication bias in psychology, can mean progression through falsification is hard to rely on.

Thomas Kuhn famously considered science as consisting of periods of ‘normal science’ in which experiment and theory are performed within a particular paradigm, with scientists steadfastly holding on to theories despite anomalies. Then very occasionally, the reigning paradigm is overturned. But even when this sort of paradigm shift occurs, it is not based on reason alone but a shift in the dominant worldview that is held.

With this in mind then we need to pay attention to people such as political theorist Paolo Gerbaudo who speculates that we are witnessing a moment of global transition of ideas, aligning with historical cycles in ideologies that take place every fifty years or so.   And alongside Gerbaudo let us also consider people such as Kenneth Gergen who would argue that social sciences need to reflect these shifts, not least because what was shaping behaviour decades ago is not what is shaping behaviour today.

Radical thinking perhaps. but we surely live in a time where radical thinking is needed and these need to translate through into practitioner work. The turbulent time we live in makes a case for applied work providing real direction for the social sciences: human challenges are played out in front of us whether related to COVID, climate change, cost of living or technology disruption:  all of these are requiring practical steps that are placing practitioners at the very heart of thinking, creativity and advancement of science, in order to find solutions that work.