On not being Derren Brown
As behavioural scientists we have many skills but mind-reading automatic mental processes is surely not one of them: provocative thinking on how to sensibly apply heuristics and nudges
Derren Brown is an illusionist well known for his apparent mind read-reading acts which he says is the result of “magic, suggestion, psychology, misdirection, and showmanship”. Is the way in which the behavioural science industry use Nudge in danger of offering a Derren Brown style act, claiming to have access to the range of mechanisms that sit in our non-conscious, shaping our behaviour?
The reason to ask this question is that at the heart of much of the discussion of behavioural science is the notion of automaticity, the way in which we seek to make decisions as intuitively as possible. To do this we operate in the realms of ‘bounded rationality’ using ‘rules of thumb’ (heuristics) to make decisions. We do this because of the complexity of situations and our inability to process all the data available that would be required to calculate the utility of every alternative action.
Our intuitive behaviours are considered to be underpinned by mental shortcuts, or rules of thumb, known as heuristics. People use heuristics to simplify decisions – especially when the topic is unfamiliar – by turning a difficult, complex question into an easier one. These can give ‘good enough’ answers (especially where some information is incomplete or unknown) and can help people to make a decision where otherwise they might have avoided doing so. Examples of heuristics might be ‘using common sense’ (buying items when they are in the sales) or using a ‘rule of thumb’ (I only buy this item when it’s on special offer).
There has been understandable enthusiasm for the use of Nudge to leverage this. Nudge is the way in which changes are made to the choice architecture, informed by these heuristics, that alters peoples’ behaviour in a predictable way without forbidding any options or significantly changing their economic incentives. To count as a mere nudge, the intervention must be easy and cheap to avoid, so nudges are not mandates.
While heuristics originated from the academic field, where they were used to help inform how the mind works as we are all aware they have moved into the policy and marketing environments. This has meant that, equipped with an understanding of the heuristics that shape decision-making, numerous policy makers and marketers have tried to change choice architecture to make it easier for people to overcome inertia and perform the target behaviour.
There is much debate about the value of nudges to change behaviour, with doubt cast about the degree to which they facilitate change and in which situations: see Lin Y et al (2017), Osman M et al (2020). It has been widely discussed many of today’s behaviour change challenges are more substantial than can be delivered by ‘nudges’ alone.
When to use
Our view is that nudges offer a means for reinforcing existing behaviour and for facilitating incremental changes. So we may find nudges go ‘with the grain’ of human behaviour to encourage people to keep buying a brand of chocolate cake by, for example, showing images of other people buying and enjoying it (using a social norms ‘nudge’). And we may nudge someone to consider buying a ginger chocolate cake (i.e. a slight variant to something that people already want or do) but we are less likely to be successful in ‘nudging’ people to buy a healthy chocolate cake option – this requires a much more substantial change and as such is unlikely to be guided by intuitive processes alone.
But even if we are considering more substantial behaviour changes, there are no real downsides to using nudges as one of a number of sources of intervention guidance. The problem comes the other way round, when the role of ‘Nudges’ is overstated. In the excitement of the possibilities of Nudge, there is a danger that the boundaries and conditions of the effectiveness of the approach are not understood. As such, we can all too easily assume that Nudge can be a panacea to a wide range of long-standing complex issues.
Of course, if we are so inclined when we are carrying around a hammer, then we can see everything as a nail. As philosopher Philip Goff points out, any theory can be stretched to explain everything: we can’t prove with 100 percent certainty that we’re not in the Matrix being fed a virtual reality by evil computers. He goes onto say “We are not looking for certainty but rather a sense that approaches are reasonably supported by the evidence”.
A more balanced perspective is surely to acknowledge the case for nudge but we are not limited by looking to nudge as only option .
The pragmatic practitioner
We have set out that we need to take care not to overstate the role of heuristics and biases and their associated nudges. But with all this in mind, what are the practitioner considerations for making good use of this body of work? There are some very pragmatic considerations that are frequently confused.
Selecting the right heuristic or bias: It seems there is a never-ending list of different possible heuristics and biases: how can you understand which is most useful to characterise behaviour that you are seeing? It is a little like trying to choose between the different sorts of theories to explain behaviour: trying to decide which is the most appropriate one is a task that even the most experienced of academics or practitioners can struggle with.
Effective diagnosis: Given these mechanisms are non-conscious, then without conducting some form of experimental design it is very difficult to be confident about which of the many possible heuristics is in operation. That notwithstanding, is not uncommon to see cases where people have observed qualitative interviews or pored over survey data and claimed to have done just that. Much of the time this feels like an over-claim: in everyday life we all necessarily need to speculate that certain mechanisms are in operation (this is basic theory of mind) but to be able to confidently attribute what people have said to specific non-conscious information processing mechanisms starts assuming behavioural scientist must have the Derren Brown style skills that we mentioned at the outset.
Confusion between Diagnosis and Design: When people are referencing heuristics it is often unclear which role they occupy. Are we trying to use them to understand behaviour or to design ways to change (‘Nudge’) behaviour. And when they are used to change behaviour, how can we know that they are relevant? Social norms are often cited as an important Nudge (therefore used for Design) but often this is without any form of Diagnosis to suggest that it is relevant in the particulars of the behaviour change that is being sought. No wonder that the use of Nudges can be confusing as people assume that all effects are relevant in all situations. But can social norms really have the same value in selecting a chocolate cake variant as deciding to go for a run or doing our recycling. Just as a doctor not having an effect Diagnosis to prescribe medication, then a behavioural science practitioner lacking an effective Diagnosis of the target behaviours means we are simply guessing when it comes to the Design of interventions needed to facilitate positive outcomes.
Some possible solutions
What can we do here? We consider there are two related ways to tackle this. First, we need to have some form of organising principle for different sorts of heuristics that help us decide which might be helpful explanations of the behaviours we are seeing. Buster Benson’s Cognitive Bias Codex organises them according to:
Too much information
Not enough meaning
Need to act fast
What should we remember?
Essentially these offer a means to understand the type of function different heuristics play. We think this is helpful for understanding these mechanisms but it is not all that useful for a practitioner – it still describes the mechanism, not how they influence behaviour. With this in mind, we wanted a means of organising heuristics that allowed us to understand how they might be implicated in certain outcomes. As such, we propose a FUSES framework:
Fast: We use shortcuts to quickly process information.
Understanding: We create simple meaning from information
Social: We often look to social norms to gain an accurate understanding of and effectively respond to social situations.
Easy: Our memory simplifies and reduces the information we recall.
Simple: We use shortcut rules to filter out information in the environment.
Within these we reference a number of different heuristics and biases. For example, within Fast there is Sunk Cost, Appeal To Novelty Fallacy, Hyperbolic Discounting and Dunning Kruger effect. These are all ways that help us to process information in our environment in a fast manner.
How does this help us? As set out earlier, it is difficult to ‘Diagnose’ our behaviour in terms of heuristics. Our solution is to use the different FUSES functions for the Diagnosis and then use heuristics to assist in the Design of intervention strategies. So, using this example, we can assess the extent to which people are able to process information in a ‘Fast’ manner through the use of simple survey self-report (or indeed observational measures). Identifying which, if any, of these are sources of difficulty, allows us to them turn to the relevant heuristics to facilitate the design of an intervention activity.
So, if we find that people are struggling to make decisions quickly when choosing between options on an online shopping site, we can use the heuristics as the relevant Design guidance provocations, helping to inform ways in which the Design of the site can be optimised to facilitate this.
It is easy to understand the appeal of Nudges as they are cheap and simple ways of changing behaviour to help meet marketing and policy goals. There is a huge literature on Nudge approaches and a significant evidence base for their effectiveness in meeting challenges, which are well documented in a variety of locations (John, Smith & Stoker 2009; Hansen & Jespersen 2013; Wilson & Chatterton 2011).
The problem the behavioural science industry has is that if Nudge constantly takes centre stage in conversations then, not unreasonably, there will be an expectation that practitioners can deliver sparkling results. In some instances this may be possible but in others, as is hopefully clear, this is far from the case. We need a more sensible conversation about nudge that helps the wider audience for behavioural science understand that it is one, of many tools in the practitioner toolkit. We suspect this is increasingly understood. But also, as practitioners we need to think carefully about how to use these tools without making significant leaps of faith in our own mind-reading skills.