Sustainable nutrition, environment, and agriculture, without consumer knowledge (SNEAK)

Lead Research Organisation: University of Bristol


In the UK, food consumed out of the home accounts for a significant proportion of the impact of diet on health and the environment. For example, 42% of workers eat at a canteen and 7 million school lunches are served daily. In response, we will deliver a simple tool that; a) generates a 15-30% reduction in both the carbon footprint of meals and their sugar, fat, and salt content, b) can be implemented without compromising food acceptability and without consumers even being aware that changes have been made, and c) will be ready for immediate application at a city-wide level and beyond.

We recognise the bold nature of these claims. However, they are grounded on our modelling of food choices in a real-world context - a university catered hall of residence. Our approach exploits a simple, yet previously overlooked, principle. In any canteen setting where menu options rotate on a fixed-term basis (e.g., menu options A, B, and C are available on Monday, options D, E, and F on Tuesday, and so on), consumers eat only one meal per day. As such, the longer-term (weekly/yearly) nutritional and environmental performance of an establishment will depend on the combination of options that happen to be served on the same day. Our findings confirm that marked improvements in diet (respectively, 21%, 28%, and 27% reductions in salt, sugar, and fat) can be achieved merely by reorganising menu options in a way that increases within-day competition between undesirable meals. In practical terms this is a multidimensional problem (salt, sugar, fat, and carbon footprint must be jointly minimised) that is possible to address using well-established techniques in computational mathematics.

To achieve these ambitious targets, this project brings together a unique combination of expertise in behavioural psychology, agricultural/environmental modelling (integrating social and natural sciences), and commercial catering.

With this 'action-focused research,' we will demonstrate direct application in a university hall of residence (actual effects on diet and carbon footprint will be measured). Building on this, we will produce a co-designed online platform for non-experts to transform other catering services. To deliver this impact, we will demonstrate the real-world benefits of our approach by collecting canteen recipe data from schools across Bristol. We will then partner with an exceptional advisory team (Bristol City Council and Bristol Food Network) to develop a strategy for city-wide rollout in schools. Importantly, we will also consult with partners from the Born in Bradford Study, who have expertise in dietary interventions for children in multi-ethnic and socially deprived areas.

We will also broaden the application of our methods to a commercial food outlet. Recognising the potential of this idea, the University of Bristol has agreed to support the project by developing the UK's first 'Consumer Lab' - a public-facing facility in which lunchtime food offerings can be experimentally manipulated. This is unique, because it combines ecological validity (actual purchases are made) with the opportunity to manipulate menu offerings on any given day. Here, we will monitor the diet quality and carbon footprint of purchases, and then show how both can be improved. Again, to develop practical next steps for application, we have co-designed a detailed plan for consulting with local food outlets (e.g., cafés and takeaways).

Finally, in addition to factoring in ways to mitigate risk, we have built-in opportunities to capitalise on 'high risk (high gain) endeavours.' Specifically, because financial reward is a strong motivator, and because rapid and wide-reaching impact is needed, we plan to show how our computational approach can return significant health and environmental benefits, alongside a reduction in food costs in schools and care homes, etc, and even an increase in profit in outlets such as cafés and takeaways.


10 25 50