Behavioural Science is diverse in its practice. As a practitioner of behavioural science since over a decade, the way I see it, BI is an extension of UX or any design thinking and HCD applications. A simple case in point I had long back written on how: https://medium.com/behscfst-in-practice/design-thinking-behavioural-sciences-for-innovation-change-and-impact-49ab8f2974f
Yes BI is behaviour-centred. Some use existing behavioural models to understand those behaviours. However no model can be applied to all contexts. And so some have developed their own processes to breakdown behaviours.
And most of these self built processes can be summarised into one: context drives behaviours. So most of the time, BI is about better understanding context and looking at ways to intervene in the context. Usability, user experience, pain points / satisfaction, all are part of the context that are driving decisions and behaviour. So behaviour centred interventions do not move away from improving usability / user experience or reducing pain points. Having said that, it is my perception that the field is still learning the ropes of usability testing through a lens of behavioural science.
Because it is behaviour-centred and behaviours are driven by contexts, not all projects require behavioural personas to be built. Building personas is based on project need. Especially with digital products serving communities and users, personalisation is impossible without knowing user personas / user segments. The difference is, personas built with BIs are anchored to context appraisors rather than user profiles.
I would argue that applying behavioural science in any problem solving process would render the approach inherently empathetic as the focus is on behaviours and their whys. Empathy is inevitable. Also, a researcher / strategist empathising is only as good as how they resist their own prejudices not find a playground in the project.
RCTs are the well-known research approaches. However there are many BI practice that use mixed research or only qual social research or a variation of design research. Among BS most practitioners, their research methodologies are rooted in evidence about human decision making.
Finally on prototyping, yes BI proposes interventions, but is it even possible to implement without prototyping and testing? Whether one implements low, medium or high fidelity prototypes is up to the project team. For example I once read that Instagram always implements only high-fidelity prototypes. Because it helps understand the user behaviours better. But that does not mean there it is not a prototype. In a way it all semantics: prototype or intervention, one cannot gauge outcomes without implementing and testing in whatever fidelity.
