Making horses drink: the missing behavioural science in innovation policy
Examining why and how behavioural science should be integrated into policy-making
Innovation policy is often consumed with the grandiose – space exploration, epochal infrastructure, billion-pound investments. We are quick to diagnose, to propose the panacea: fifteen-minute cities to fix gender disparities in the workplace, increased social infrastructure to reduce loneliness, capped fares to recover London Underground use post-COVID. Yet, while undeniably crucial for driving faltering economic growth, these policies can often ignore the behaviours of the very populations they are designed to serve. It is not enough to build ‘community hubs’ to decrease social fragmentation, and expect local residents to show up. It is not enough to design elaborate ‘development opportunities’ – even if free! – to promote entrepreneurship, and assume that young professionals will attend. It is not enough to lead a horse to water; policies must also make them drink.
Indeed, this is the key takeaway of behavioural science, which places human behaviour squarely at the centre of policy-making. The discipline has, in recent years, been unfairly maligned: labelled simultaneously a neoliberal ploy to avoid systemic solutions and a paternalistic power-grabbing tool to manipulate populations into following public health interventions or paying their taxes on time. Still, behavioural science need not be an ideological battleground. At its core, it simply advocates for redressing a recurring gap in policy-making, which forgets that the status quo typically carries psychological benefits, and that providing infrastructure, no matter how effective it may seem, cannot on its own overcome a deeply entrenched human aversion to change.
To examine how policies can more effectively integrate behavioural science, we can consider two timely examples: (1) addressing community fragmentation due to lack of social infrastructure and (2) increasing artificial intelligence (AI) uptake to increase workplace productivity.
Reforming Social Infrastructure
Even before COVID shuttered city centres and community programmes, social cohesion was eroding. Though the declining number of social groups, a contributing factor to loneliness, has been noted since as early as the 1970s, the last two decades have seen a particularly substantial decline in socialising: in the US, for instance, teenagers see one another in person 45% less frequently than the previous generation. Meeting in cyberspace, while often perceived as equally social, may impair the development of empathy and emotional intelligence, especially if not complemented with offline interaction. Moreover, loneliness is not only individually concerning: the subsequent fragmentation of the social network has been shown to lead to decreased community trust and reduced social support in times of crisis, particularly for vulnerable groups.
In response, urban planners and policymakers have pointed to social infrastructure as the solution. By acting as a ‘neutral ground’, where individuals of different ages, backgrounds and interests can meet, such spaces enable social connection. Examples vary, but typically include both public facilities, like parks and libraries, and private businesses, such as cafes, bowling alleys, malls, and community centres.
Indeed, calls for such measures are reasonable and fairly well-evidenced – but proposed solutions often lack an awareness of psychological barriers. Even at university, where student societies offer free, useful, highly-catered events in public facilities more or less on demand, turnout is often remarkably low. Similarly, despite frequent calls for a wide range of parenting support services, when such services are delivered without considering behavioural science, uptake can be unexpectedly limited. Is it therefore any surprise that, when individuals are distracted with family life and exhausted from work, they are even less likely to look for, and make use of, existing opportunities – not just childcare facilities, but book clubs, gardening groups, community spaces, and bars?
From the outset, behavioural science frameworks like COM-B – which allow policymakers to diagnose existing barriers preventing a target behaviour from occurring – should be used to guide proposals. Policymakers must ask themselves: are individuals physically and psychologically capable of engaging with social infrastructure? Spaces should, for instance, provide equitable access for groups most disproportionately affected by loneliness, such as the elderly and those with chronic illnesses. Moreover, planned projects must take effort to address societal characteristics: in Western contexts, where individuals are typically more individualistic in their social orientations, the benefits of engaging with the community are not psychologically modelled by adults as frequently, and have no clear narrative surrounding their benefits. Consequently, seeing little personal gain from attending local events, target populations may be harder to access than policymakers anticipate. Campaigns to motivate engagement must therefore meet existing preferences; rather than advertising attendance as indicative of ‘local pride’, a community centre should emphasise reduced childcare costs or opportunities for networking.
Messaging should also take care to be memorable and consistent: viewing the same slogan repeatedly tends to increase trust in its message, so having campaigns both on and offline, through different mediums, can ensure that the message is seen, remembered, and acted upon. Messages should ideally be actionable, advertising the event happening at THIS TIME on THIS DAY; campaigns must aim to make it easy to integrate participation into a daily routine.
Policymakers should also recall that, when we do change our preferences, it is often because such behaviour has already been modelled – by someone we admire, or by the rest of our peer group. Thus, to create a sense of social expectation and encourage new social norms, initiatives should enlist the support of individuals with high local involvement. Convincing a handful of religious leaders, PTA organisers, and councillors to spread the word to their networks — who already hold their opinion in high regard – is far likelier to encourage community engagement than an onslaught of leaflets otherwise might.
With compelling narratives and the right ‘influencers’, more inventive interventions become possible: community buy-in allows exciting ideas like ‘Rebuild by Design’ competitions, which can stimulate innovative multi-purpose designs. The resultant greater trust emerging from stronger social networks enables the creation of community funds with participatory budgeting processes and may even catalyse necessary legislative changes – without NIMBY resistance. The cumulative effect of instituting these interventions will not only combat existing social fragmentation, overcoming the existing harms of isolation, but will also confer extensive economic and political benefits. With a stronger social fabric, individuals are more incentivised to engage in civic action, increasing democratic stability. Moreover, with increased pride in their local identity, residents are more emboldened to contribute to economic growth by setting up new businesses. Greater social trust also has the benefit of reversing the brain-drain effect: stronger social networks lead to younger residents being more likely to return to their community after graduating university, helping further economically ‘level up’ their community.
Building effective policy around social infrastructure is vital for reaping the ensuing reward of vibrant economic activity. The alternative – lonely, socially fragmented communities – is not only likely to place further stress on strained mental health services, but also economically damaging. Thus, to effectively create the innovation hubs that the UK so desperately needs, it is imperative that behavioural science is involved.
Increasing Artificial Intelligence (AI) Usage
Behavioural science can be equally useful for innovation in a workplace context, as well as a social one. AI’s benefits for productivity already feel like old news: whether for running data analysis, spotting errors, or brainstorming, when used to assist, rather than replace cognitive effort, artificial intelligence has proven an efficient tool. Indeed, studies have suggested that integrating usage of Generative AI technologies, like ChatGPT, into highly-skilled employees’ workflows can improve performance by up to 40%. Yet, despite frequently advertised benefits, the ‘digital drag’ – wherein workplaces have low adoption of Generative AI – remains. A recent survey of business leaders found 67% of respondents saw limited-to-zero usage of AI within their workplaces – and only 7% admitted ‘extensive use’. It is not that individuals are protective of their skills, nor that they are (rightly!) wary of data privacy concerns – most attitudes towards AI are neutral. We are just, perhaps unsurprisingly, hostile to change.
Behavioural science frameworks reveal that technology adoption rests on human factors. If governments want businesses and civil service employees to yield the promised growth of AI sooner, they would do well to focus first on how best to overcome our bias to the status quo. Disruptive technologies will undeniably face resistance, particularly from long-standing employees, who have built habits with incumbent technology systems. We must ask: what might be the benefits motivating employees, both in government and in the private sector, to stick to routine rather than experiment? How can we reduce the psychological pain of switching?
For a start, transitioning to a new system must be made simpler. Even enterprising employees are likely to feel overwhelmed by the quantity of resources available and, though generative AI tools may seem intuitive, complexities will certainly emerge during integration. Employees are used to dealing with specific details such as the proprietary systems of their organisation and the security protocols expected of them. No tool slides perfectly into a pre-existing routine, especially when workflows can differ considerably on a company basis. Thus, managers must take it upon themselves to identify what aspects of new tools are most useful for complementing existing work, without overcomplicating things by introducing needlessly fancy use-cases, and create easy-to-follow, workplace-specific guides.
Messaging is important here, too. We tend to be more sensitive to losses than equivalent gains, so managers should frame sticking to default technology as a “loss”. In the public sector, this could be focussed on collectively important goals, like increasing national productivity; in the private sector, this could emphasise the employee’s own career development. Importantly, messaging may assume that employees are intrinsically motivated; yet, if expecting workers to go out of their way and stay ‘ahead of the curve’ on relevant technological advances, employers must first create a culture of high psychological safety. This entails training managers who are willing to entertain novel, high-stakes ideas, without acting territorial about workplace hierarchies. Quelling negative beliefs by assuring employees that, in the short-term, Generative AI will be seen as a tool, rather than a potential replacement, can increase perceptions of psychological safety further. Moreover, workplaces should clearly and honestly communicate long-term contingency plans, in case the need for replacement arises: providing a guarantee of timescales and collaborating with the government on retraining schemes could be effective ways to further reduce employee anxiety.
In turn, per the principles of ‘open innovation’, certain workers will feel encouraged to experiment, increasing the likelihood of breakthroughs. Effective messaging could therefore be instrumental to innovation: in the private sector, employees may happen upon improvements which could induce positive spillovers for national economic growth. Similarly, increased efficiency in data entry and analysis could be key to addressing low productivity in the public sector.
With sufficient social support to invest time into side projects, some workers (‘lead users’) will feel more confident about approaching new technology, including Generative AI. The key to spreading AI usage beyond one or two workers is through interventions which are easy, attractive, social and timely, as suggested by the ‘EAST’ framework developed by the UK’s Behavioural Insights Team. In practice, this could mean giving lead users the opportunity to run collaborative – or competitive – internal workshops during lunch, with rewards, like an extra day of holiday or a restaurant voucher, given for attendance. Admittedly, for companies seeking prestige or governments trying to set themselves apart from the opposition, such interventions may be less concrete and bragworthy than making claims about extensive new infrastructure purchased. Yet, if an organisation wants to truly reap the benefits of the technology it purchases, it must start by considering the product in tandem with its users: employees.
By no means is behavioural science the cure-all to ensure growth. Rather, it is an ingredient, vital for ensuring the end-product performs as expected. Such an ingredient is easy enough to implement: all it requires is policymakers and employers that take the time to understand the behaviours responsible, diagnose psychological barriers, and ultimately design interventions which specifically overcome these.
