We explore the future of behavioural science and it’s possible intersection with the world of data science.
Ever walked out of a movie and found yourself replaying a certain line or scene over and over again in your head? Okay, in my case it wasn’t a movie, it was Nudgestock 2021 and it wasn’t a line but a speech by Rory Sutherland. He said
“The best ideas nowadays don’t emerge within disciplines, they emerge at the intersection between them” - Rory Sutherland
Now, granted this isn’t much of a revelation, but it’s important to understand the context behind which it was said. You see, in the world of behavioural science and economics, people working or researching in this field (including me) felt that solutions to most problems in this day and age could be solved using behavioural science (BS) and its principles. And the pièce de résistance of behaviourally induced solutions are nudges. What is a nudge you ask? It is an intervention that gently steers individuals towards a desired action without reducing any available options. To quote the authors of the theory itself “To count as a mere nudge, the intervention must be easy and cheap to avoid. Nudges are not mandates. Putting fruit at eye level counts as a nudge. Banning junk food does not.”
So coming back to #Nudgestock, Rory’s opening speech made me think a lot and I think it set the right precedent for us BS enthusiasts and specialists to stop and think, is behavioural science the answer to everything? And as it turns out it’s not. But this got me thinking about the future of BS? What does the next decade of this field look like?
Oftentimes we find ourselves racing to fix problems and most of the time, our solutions are pinned to something we have tried before or something we know that works. Don’t get me wrong, resorting to tried and tested solutions isn't necessarily a bad thing, but we need to inject some creativity into the process. For example, Sam Tatum spoke about the concept of biomimicry. It is a practice that learns from and mimics the strategies found in nature to solve human design challenges. At its very core, this field came about due to the intersection of one or more studies.
Therefore, I reckon it’s time we break down silos and expand our knowledge base. We should not limit ourselves at the intersection of psychology and economics but rather explore the connections between B.S and other fields like neuroscience, sociology, philosophy, and data science, for example.
The Perfect Match
One particular marriage I am very interested in seeing is that of behavioural science and data science. Partly because these two fields have a lot in common. Both of these fields can be viewed as different responses to a single observation; that people are predictably irrational. They both focus on optimising how decisions are to be made in the most efficient or cost effective or profitable manner.
Let’s start with predictive analytics. These algorithms have become all too common in our daily decision-making process. They now help us find our way in a new city via GPS, determine our entertainment choices in streaming services like Spotify, Netflix etc, and generally shape our online behaviour through recommendations and personalised offers on platforms like Amazon. It is used as a tool by companies too often see the bigger or clearer picture, akin to wearing spectacles for correcting for myopic vision. They help analytically abled companies to operate efficiently and profitably in emerging markets by accounting for all possible permutations and combinations of decisions that work best in these markets. Although these algorithms are designed to facilitate choice by humans, it is baffling to discover how little research into human behaviour normally goes into their design.
As a result, these algorithms are more suited to assist a rational agent than a human being. Netflix is a case in point. They have long proclaimed that their recommendation algorithm strives to help users find a program to enjoy with “minimal effort.” But despite their advancements, a 2016 study found that it takes the average Netflix user 18 minutes to choose a program to watch. From a behavioural economics point of view, we call this phenomenon choice overload.
Now imagine the possibility of combining behavioural insights with predictive algorithms. Fortunately for us, a version of this has already been tried; both for the better and for the worst.
Obama’s 2012 election is widely regarded as the first big data election. Using the combined power of behaviour science and analytics, they were able to optimise the efforts of their campaign workers. The campaign’s data scientists built, and continually tested models that could prioritise and estimate the number of voters who could be persuaded to vote for Obama. To do so they employed numerous behavioural economics principles like sunk cost fallacy, by asking people to fill out ‘commitment cards’ with the premise being people are most likely to follow through with something if they have already invested some time into it. Social proofing by way letting other people in the community know of the actions of their neighbours driving towards a decision themselves. If you are interested in learning about this, I highly recommend reading, The Victory Lab: The Secret Science of Winning Campaigns by Sasha Issenberg.
However, a few years down the line, Cambridge Analytica happened and we all know what happened there. The brazen exploitation of user data, the psychological warfare and behavioural manipulation that followed, brings us to a rather difficult impasse; what can we do to prevent misuse of behavioural data?
Short answer; oversight. There needs to be strict boundaries on how and where to use such data. I reckon at this rate of data consumption, we will soon reach an ethical junction wherein we would need to create strong ethical guardrails to keep us from straying too far into the dark and to protect the behavioural science against the corrupting influence of the rich, powerful and from those vested interests lies solely in profitability above all else. Whilst GDPR and CCPA are steps in the right direction, we would also require a set of laws drafted specifically for handling behavioural data.
In summary, I think the future of this field is bright. Studying the intersection of behavioural science and data science is only one among numerous combinations. Therefore, instead of looking at fields individually, we must start looking at the intersection of these fields itself. So yes, in conclusion I am agreeing with Rory and Nudgestock 2021. I reckon it’s time to break down silos and expand our potential solution spaces by combining BS with other fields of studies or disciplines. This is what I think the future of BS will be.
Written by Arjun Manohar