February 5, 2021

“Look her in the eyes and tell her you never break the rules.” You’ll doubtless have seen or heard manifestations of the UK Government’s most recent advertising campaign. Close-up images of Covid patients wearing oxygen masks and a pleading gaze are begging you to stick to the guidance and stay at home. Clearly, our leaders are betting that this emotional appeal will change people’s behaviour for the better. It’s not the first time they’ve dabbled in behavioural science. Or have tried to, anyway.

Behavioural science — a catch-all term for a strand of psychology, with micro-economics, sociology and other disciplines mixed in — has been unusually prominent during the pandemic, and particularly so in the UK. For a while at the start of 2020, it was touted as the answer to all our prayers. “Nudges” to change people’s behaviours were suddenly the order of the day. “Behavioural science can”, we were assured, “save lives”; without it, “our attempts to defeat the virus will fail”. On the BMJ’s blog, UCL psychologist Susan Michie and her colleagues wrote a post entitled: “Behavioural science must be at the heart of the public health response to Covid-19”. Boris Johnson himself was directly advised by “behavioural psychologists”, and specifically referenced them at a press conference.

At the end of March, just after lockdown began, I wrote for UnHerd about the bad predictions and faulty logic of psychologists (and other people writing about risk). I argued that low-quality evidence was being relied on to support arguments that downplayed the danger of the disease. But the story of our Government’s engagement with behavioural science during this pandemic extends far beyond that. Now that a whole year has passed since the UK recorded its first Covid cases, it’s a good time to look back at what the much-lauded discipline contributed to our Covid response.

Behavioural science had great press for a few weeks, but it all fell apart pretty fast. As case numbers and deaths in the UK increased throughout March, many began to wonder why we weren’t following the example of other countries and locking down. The blame, according to some, lay at the feet of the behavioural scientists. Apparently, they’d advised the Government that the population would suffer “behavioural fatigue” — that after just a few weeks of strict lockdown measures, we’d get bored and frustrated, and start bending the rules in an attempt to get back to normal life.

Chief Medical Officer Chris Whitty, Deputy Chief Medical Officer Jenny Harries, Chief Scientific Adviser Patrick Vallance, and others all made this argument explicitly in the media. It would be counter-productive to go too hard too soon, they said. Such a huge change to people’s lives wouldn’t be sustainable, and people might make a collective break for freedom just as the virus was most prevalent.

But behavioural scientists, whose discipline was supposedly the basis of the Government’s policy, were outraged by this narrative. Michie, a member of the Government’s SAGE subcommittee on behavioural science, SPI-B, told the Guardian that the concept of “fatigue” had never been used in any of their reports (which was true). Indeed, she said, “[fatigue] is just not a concept that exists in behavioural science”. Warwick behavioural scientist Nick Chater agreed, writing that behavioural fatigue “has no basis in science”. Another member of SPI-B, Robert West, made much the same argument. An Open Letter from UK psychologists and behavioural scientists urged the Government to reveal the evidence base for the fatigue concept (I was among nearly 700 signatories).

As it turns out, there was evidence in the scientific literature for the idea of “behavioural fatigue” all along. The neuropsychologist Vaughan Bell, as far back as late March, reviewed a selection of papers (all published pre-2020) that discussed the fatigue idea in the context of people’s rule-compliance in past disease outbreaks, though they didn’t necessarily use the specific word “fatigue”. As Bell noted, some found evidence for it, some against — and the studies certainly weren’t all top-quality. But the idea spread by behavioural scientists before the first lockdown — that “behavioural fatigue” was a concept entirely divorced from the scientific literature — was flat wrong.

So far, so embarrassing: the psychologists either weren’t aware of the relevant evidence, or wanted to shift the blame for the delayed lockdown away from behavioural science (even though they’d never advocated for it). Whitty, Harries and Vallance, for their part, never pointed to any of the science they were relying on to make their claims. Confusion reigned. The lockdown was delayed on the basis of some unknown evidence. Behavioural science seemed to have failed its first and biggest test of the pandemic.

Suggested reading
The Brits' love-hate relationship with lockdown

By Dan Hitchens

To make things worse, the “fatigue” debate was essentially irrelevant to what the Government should have been doing in the face of the virus. Regardless of whether people’s compliance with the rules would decay over time, it would still have made sense to lock down earlier, because the virus spreads exponentially. Stopping it from taking hold early in an outbreak (a policy that worked in countries like New Zealand and Taiwan) means that the population — even if they start to break more rules as time goes on — will be at far less risk of the disease. To put it another way: despite the hype, the behavioural science should have been very much secondary, something to worry about only if a government fails to properly contain the spread of infection.

You’ll have noticed, of course, that our government did fail to properly contain the spread of infection. We didn’t lock down early enough — twice — and we’ve now had over 100,000 coronavirus deaths. So perhaps there’s more of a role for behavioural science now. But have we learned anything in the past year about how to mould human behaviour to keep people from Covid?

My behavioural scientist colleagues won’t like me for saying this, but the answer is: not really. A number of recent studies have explored how to improve interventions such as hand-washing, face-touching, mask-wearing, and vaccine-getting. (Let’s leave aside the awkward fact that the first two probably aren’t all that relevant to stopping the spread – the behavioural scientists weren’t to know that at the time.) These interventions are a smorgasbord of tricks to get people thinking differently about the virus: asking them to plan for the future; targeting their emotions; telling them how social distancing can save the economy; focusing them on reasoning rather than gut feelings; and so on — all things that might be adapted into government messaging, if they work.

But do they? Some of them, such as asking a person to write down the name of a loved-one they want to protect, have shown no effects whatsoever. Others, like focusing on moral messages, or those delivered by a respected figure, are supported by patchy evidence that shows generally small effects. Some have shown effects only in particular circumstances: one study tried a range of different interventions and found that some worked, but only for people who had just recently begun social distancing. For those who had been social distancing for longer, the interventions backfired — the participants reported going outside more and distancing less after they’d taken part in the intervention. Oops.

Even for the studies that appear more promising, one has to remember what’s actually being measured: in most cases, it’s people’s self-reported intentions to do things like wash their hands or wear masks. It’s what they tick on a questionnaire during the experiment — not necessarily what they’ll do in the real world. Of course, intentions and actions will be related to one another, but not perfectly. Researchers should be a lot more careful than the authors of the questionnaire study that appeared in September, whose very title claimed that “Empathy Promotes Physical Distancing and Wearing of Masks”. The only data they gathered concerned self-assessed motivations — not behaviours. “Intentional science” sounds somewhat less impressive than “behavioural science”, but it would, alas, be much more accurate.

Yet more evidence is appearing all the time: perhaps the behavioural scientists will crack a solid, reliable intervention soon. But the discipline’s contribution to the pandemic feels like a classic case of over-promising and under-delivering. The confident claims at the start of the pandemic — that behavioural science would play a crucial role in the Government’s toolkit to keep us all safe — simply fell flat when the actual data were examined. Had the hype been dialled back a little, behavioural science’s messy evidence base would look more like a discipline doing its best to find out what works (which is a genuinely difficult task), rather than a big disappointment.

Even if our ability to alter human behaviour using clever psychological techniques is weaker than we’d like, it is still useful for understanding human behaviour in the pandemic. To take just one example: knowing what people believe, and how those beliefs change over time, is important for governments and health systems to target their policies.

UCL’s Covid-19 Social Study is packed with interesting data on this topic. One of the overall findings is that people’s compliance with the rules (again, self-reported, so caveat emptor) has waxed and waned along with Covid cases: the recent second wave was accompanied by more people saying they follow all the rules. The main driver of people’s behaviour seems to be how many infections there are in the country (which ultimately depends on government policies like lockdown and border control); “soft” behavioural science interventions like nudges are only going to be useful at the margin.

Suggested reading
From Covid to crime: how media hype distorts risk

By Tom Chivers

To those new UK Government adverts, they seem to be an attempt to use a mixture of empathy, shame and fear to promote compliance. Although there’s plenty of evidence to show that the latter of these has worked in previous contexts, it’s hard to apply any previous research to our unprecedented situation. It certainly feels to me that more attention paid to actual Covid patients, rather than just graphs of case numbers, would jolt people out of complacency — but that’s the hunch of a behavioural scientist, rather than evidence from behavioural science.

And there’s nothing wrong in theory with a behavioural scientist giving you their opinion — if they’re clear that’s what it is. The superficially convincing sheen of “science” is too easily applied to what are often just thoughts and judgements by well-meaning, well-informed experts (or mostly well-informed: the “fatigue” debacle illustrates how even the experts can fail to be on top of their brief). Setting the behavioural scientists up as some kind of oracular group of truth-tellers who’ve figured out the intricacies of the human mind is setting them up for failure — and trivialising the very real contributions they can make.

So can I look you in the eyes and tell you that the new adverts will work? Not really. But what do I know? I’m just a behavioural scientist.

Sign up to UnHerd's new weekly email, in which Will Lloyd selects the best (and occasionally worst) writing from around the web.

Free, every Friday morning.