Around the World with Paul Eccleson
Applied Behavioural Science Globally
It’s that beautiful time to dive into the newest instalment of our Around the World series! Here, we present accomplished experts in behavioral science, ready to impart their wisdom and personal insights to you, our dedicated readers.
This month, we are highlighting Paul Eccleson, winner of the ‘Nudge’ award for his work in the financial sector, with a long history of executive Risk management.
Without further ado, let’s delve into our discussion with Paul!
Paul, what led you to the field of behavioral science, and does it still play an important role in your life?
I don’t consider that I ever chose to join the field of behavioural science, really. I feel that the field joined me, not the other way around.
I qualified in the 1980s with a degree in Psychology and the science simply changed its name at some point without me noticing. I originally studied Psychology because I was interested in the computer modelling of the human mind. It may come as a surprise to your readers that the field of Artificial Intelligence was actively being researched in the 70s and 80s. After my postgrad, I joined the small UK AI community building expert systems, logic-based programming and machine learning for Hewlett-Packard’s European Research Centre. My career moved from technology development to the business application of technology in security, risk management and compliance. My Exec career saw me take board-level roles in those specialisms, but I never stopped practicing my Psychology.
I see Behavioural Science as key to a career in risk management and compliance (or pretty much any career, but I’ll come back to that). Corporations don’t take risks, people do. Corporations don’t strive for compliance, people do. Despite the undoubted technical and legal requirements of a risk and compliance role, your success is going to hinge upon being able to influence the behaviour of others. What is creating a “risk management” culture other than a large-scale exercise in setting social norms? How can you understand the decision making of a Board without also understanding social conformance, group identity, groupthink and human bias?
“Corporations don’t take risks, people do. Corporations don’t strive for compliance, people do.”
I now teach Behavioural Science as a Fellow of the International Compliance Association. My postgraduate students are the rising stars of risk and compliance across the globe, and my masterclasses focus on delivering BehSci insights to a new generation of senior professionals.
How do you stay up-to-date on research and developments — any specific resources or strategies?
This field moves very quickly. There are many new papers and books produced each year that both advance and challenge what we know about human behaviour. As well as publications such as your own, I follow the podcasts, posts, and websites of some thought leaders that I trust most:
- Richard Shotton whose books “The Choice Factory” and “The Illusion of Choice” are packed with such solidly researched, yet eminently practical, applications of BehSci to business.
- Rory Sutherland, whose content and training courses pretty much established the field of BehSci in marketing.
- Christian Hunt, whose “Human Risk” podcast reliably delivers new insights in a broad range of domains. He is also one of the few practitioners to specifically address risk and compliance professionals. His “Humanizing Rules” book is a gem.
- The Behavioural Insights Team, whose podcasts, insight articles, and working papers consistently demonstrate the application of BehSci in a range of very important areas around the world.
In addition, I follow my curiosity in areas such as health and safety, human error and statistics. For a light hearted, yet super insightful, analysis of the numbers that we encounter every day, the UK’s radio podcast “More or Less” is always worth a listen. Its presenter, Tim Harford, manages to make statistics both entertaining and understandable. A regular on the programme, Prof David Spieglehalter, is also a great source of thoughtful analysis about statistics that describe our behaviours. In particular, he carefully explains why straight men claim to have had twice as many sexual partners, on average, as straight women, which in a closed society with a roughly equal number of men and women is mathematically impossible.
Given the majority of WEIRD literature, can you tell us about a time you encountered surprising findings in your work?
An organisation I was working with wanted to understand the attitude of their staff towards a major, multi-million dollar, office move. As such, it had included premises-related questions in two questionnaire-based surveys. One, a survey specifically about the move, reported that the staff found the existing office “depressing, outdated and uninspiring”. Another, a more general mental health survey, found the existing office “open, airy and supportive”. Confused, the exec team wanted to understand how two apparently contradictory findings could exist.
It seemed that two things were causing the paradox:
- Context and priming. BehSci research clearly shows that people can be “primed” to give different responses depending upon what they are told about the situation being faced. Describing a milkshake as “healthy and nutritious” makes people feel it is doing them good. Describe the same milkshake as an “indulgent, creamy, treat” and people feel it’s a delicious, if somewhat unhealthy, dessert. More than that, their physiology responds in line with the primed expectations. In the premises-specific survey, there were a large set of questions regarding the quality, or otherwise, of the environment and if the staff would prefer new offices. In the mental health survey, the focus of the questions was on the support and wrap around care available to staff. The first survey seemed to be priming negative responses regarding the current environment, the second positive response regarding the mental health support available.
- Emotional language. If people are shown a video of two cars colliding, their judgement of the content of the video is coloured by the language used in questions about it. “How fast were the cars travelling when the smashed into one another” will lead to judgements of high speeds and tangled metal (even when there was no such damage present). Swap “smashed” for “bumped” and speeds are judged lower and there’s no damage reported. The tone of the questions in the premises survey — aimed at identifying improvements — had generated more negative responses.
How do you design Behavioral interventions to be culturally sensitive? Have cultural differences impacted your work in behavioural science?
For the most part, psychological responses across the globe seem pretty constant. For example the “attitude-behaviour” gap — that people say one thing and do another — seems a very replicable finding regardless of the cultural context. Recent work by the Behavioural Insights Team has shown that posters and other interventions on the Bangladesh bus network, had increased awareness of sexual harassment amongst travellers, but done little to reduce its frequency or the likelihood of bystander intervention. That’s a common finding wherever such interventions are run. I regularly teach students from across the globe, from a large variety of cultures. In the stories that they tell me, and the BehSci challenges they are faced with, I see much more commonality than differences across cultures. As Paul Simon sang, perhaps this is the “myth of fingerprints” — we are both unique and similar at the same time.
“I see much more commonality than differences across cultures. As Paul Simon sang, perhaps this is the “myth of fingerprints” — we are both unique and similar at the same time.”
Can you describe an experience navigating conflicting priorities or goals in a behavioral intervention project and what you learned?
Whilst at DAS Legal Expenses Insurance, I led an intervention aimed at improving the understanding of our products. What is and isn’t covered by an insurance policy can be tricky to understand, and this is particularly true of unusual covers such as those for legal expenses. Although we subsequently won a retail “Nudge” award at “Nudgestock” in 2015, the findings were in tension with the UK regulatory advice about product literature. By law, each insurance product needs to be accompanied by a “clear, fair and not misleading” Key Facts document. The general approach to writing this document is to create extensive text in a clear and “understandable” format. We tested 7 different introductory documents on over 900 people and found that the traditional approach was nowhere near the best at explaining our product. The clear winner was a punchy, SMS-like, series of short phrases. Not the best English, but definitely the most impactful. In order to be “clear” we needed to break with accepted regulatory practice.
What do you think the future of behavioural science will look like in the next decade? What challenges do you predict, and how should they be addressed?
I believe that the replication crisis, discrediting of certain key researchers and the difficulties in converting theory into reliable interventions will undermine the deployment of BehSci. There are a growing number of BehSci practitioners in a variety of disciplines and the challenge for them will be to continue to experiment with interventions when BehSci is under pressure. As we discussed earlier, context is everything. Theory can guide you to what might work, but practitioners will need the resources to stage field trials to discover what interventions might be most effective. Better resourced organisations will be able to test and deploy. Less well resourced firms may decide that their money is better spent elsewhere. Additionally, BehSci interventions often result in only small changes in behaviour. A small percentage change in behaviour is better than none, and on a large population scale can have significant returns, but such marginal gains may be considered a poor return on investment to cash strapped organisations.
“BehSci interventions often result in only small changes in behaviour. A small percentage change in behaviour is better than none, and on a large population scale can have significant returns.”
How do you see the relationship between human researchers and artificial intelligence evolving? What challenges or opportunities do you anticipate?
One of the opportunities for researchers is the detection of undiscovered relationships in large data sets. For example, one analysis has suggested that AI could extract “Unknowable Knowledge” from the data held in a museum’s collections database. Such databases may hold links, cross-references, correlations and storylines that a pattern discovery algorithm could map, but that would be impossible to find for a human researcher. As they put it, “Forget the AI scaremongering, let’s use this technology to discover new things about ourselves”.
I also see opportunities for AI in the elimination of some human biases. The well-known “four eyes principle” is a common behavioural control implemented in organisations, but can be flawed. Fingerprints that do not match can be corroborated by a reviewer if the checker understands that the original investigator had identified a positive match. The quality assurance judgement is skewed if the fingerprints have already been judged to match. It seems possible to me that, where such human bias puts an important process in jeopardy, AI could be deployed to annul that bias. The same may be true in data analysis for BehSci research. AI could play a role in analysing raw data, testing conclusions and finding unusual patterns in reported results.
What would you like to share with Habit Weekly readers about your work?
Whatever area we work in, a key risk for any organisation comes from its humans. It is common to blame failures in organisations on weaknesses in our staff. Human Risk mutates into a blame culture and results in punitive retribution for miscreants. The words we use as BehSci researchers makes this worse: human “bias” and “human error” are the most common diagnoses when we look for the root causes of problems.
I don’t see it this way at all. Our unique talent as human beings is the ability to convert “cognitive” tasks into compiled, automated, fast responses based on learnt rules and experience. Our “lazy” brains try to push calorie hungry cognition out to less onerous ways of solving problems. Our skill and wisdom is derived from our learnt experience, and compiled into fast routines to make our lives less exhausting. This is what makes us “experts” — we can navigate our complicated world without us being overwhelmed by the millions of stimuli we are receiving. On those rare occasions when our learnt rules and compiled behaviours aren’t appropriate, applying a close approximation is what we do as “experts”.
But sometimes that approximation isn’t good enough.
Our human response is to perform our tasks as instinctively as possible — compiling the sequence of actions as much as we can, using our experience of “quick fire” rules and getting the job done. Seen in this light, “human error” isn’t erroneous at all — it’s the natural consequence of us being skilled experts. Humans do not set out to mess up, make poor judgements, or fail. If they fall prey to their own biases and failures of experience, this is just the downside of them being experts. Give such people an understanding and supportive response. We expect them to be experts, we should live with the downside.
“ Seen in this light, “human error” isn’t erroneous at all — it’s the natural consequence of us being skilled experts. We expect them to be experts, we should live with the downside.”
Big thanks go out to Paul for the detailed insights he shared with us this month and his brilliant work applying behavioural science to the financial world. You can connect with him on LinkedIn here to explore more of his work, achievements, and posts!
