Food environments shape food acquisition, diet-related health outcomes, and food securi-ty, yet the relationship between individuals and food environments remains difficult to capture through approaches focused on isolated variables. Agent-based models (ABMs) offer a useful way to address this complexity by representing heterogeneous actors, their decision rules, and the environments in which they interact. This paper presents an ongo-ing scoping review of ABMs applied to food environments, with a focus on how food access is represented and how interactions between individuals and food environments are modeled. A Web of Science search identified 193 records; after document-type filter-ing and deduplication, 176 records were screened, resulting in a final corpus of 32 stud-ies. Preliminary findings show that this literature is relatively recent, interdisciplinary, and fragmented. Most studies are geographically explicit, empirically informed, and policy relevant. Accessibility is most often represented through economic and spatial dimen-sions, while temporal and social or cultural dimensions appear less frequently. Likewise, although many models include heterogeneous consumers, relatively few explicitly incor-porate peer-network effects, social norms, or adaptive learning. These findings suggest that current ABM applications to food environments remain more developed in represent-ing material and spatial aspects of access than socially embedded processes.

We model UK meat consumption with a modular agent-based model to explore how diets change under different theoretical assumptions and implementations, focusing on the theory of reasoned action, and habit theory. The model is data-driven using national survey data, and economic statistics. We validate the model against meat consumption data from the OECD. The model makes a novel contribution to understanding sustainable dietary change by utilising a modular design which provides a customisable platform to test intervention scenarios under explicit assumptions.

Agent-based models (ABMs) are increasingly used to study
dietary behaviors and test interventions aimed at improving diet quality.
However, many existing models do not explicitly evaluate the mechanisms
needed to reproduce real world data and assess interventions.
This makes it difficult to judge the validity and usefulness of the ABM
for its stated purpose. Based on previous literature and a conceptual
map, we identified three key sets of mechanisms, namely personal, social,
and spatial, which translate to submodels in the ABM. We systematically
compare model variants to evaluate which combination of these
submodels best reproduces observed data on fruit, vegetable, and sugary
drink consumption in Amsterdam. Our proof-of-concept results show
that including multiple submodels more closely approximates observed
food shopping patterns than the personal submodel alone. These findings
suggest that the selection of mechanisms for ABMs should be carefully
evaluated.

An increasing body of literature underscores the adverse health outcomes associated with ultra-processed food consumption. These foods have been linked to type 2 diabetes, heart disease, and increased mortality. Public health authorities worldwide have employed various strategies to raise awareness among consumers. One such strategy is the use of front-of-package labels. Front-of-package warning labels aim to discourage excessive consumption of sugary drinks and ultra-processed foods – both categories being energy-dense and nutrient-poor. Our proposal focuses on inspecting the dynamics of consumption dynamics when consumers are exposed to prod-ucts containing front-of-package warning labels. We aim to identify the po-tential impact of different factors that influence food consumption dynamics. To that end, we collected data that helped us to quantify consumer preferences and factors affecting purchasing decisions. We built an (agent-based) computational model that captures consumption dynamics based on belief functions calibrated according to mathematical formulations of partial evi-dence. The model explores how consumers make purchasing decisions when faced with warning labels on products containing excess sugar, sodium, and saturated fats.

Food systems in sub-Saharan African cities are increasingly pressured by rapid urban sprawl and socio-economic changes. While land conversion from cropland to built-up area is limiting the opportunity for urban agriculture, food demand is rising because of population growth and changing diets. Meanwhile, socio-economic segregation – often associated with urbanization - can hinder access to food. For the case study of Kampala (Uganda), we spatiotemporally model the land-food system using an agent-based approach. Based on 747 household surveys, we recalibrated the Agent-based model of Social Segregation and Urban Expansion (ASSURE) by Vermeiren et al. (2016) and included food system dynamics to assess future trajectories (2020–2040) of Kampala's dependency on urban agriculture. While food that is both produced and consumed within the city is often considered the most resilient food source in times of crisis, we show that it is particularly this source that is threatened. Overall, about 10 % of the urban and peri-urban agricultural land in Kampala is projected to disappear by 2040. This may lead to decreased levels of food security and dietary diversity, particularly for households that strongly rely on urban agriculture. Information on the links between urban growth and local food provision is essential for planners who aim to develop strategies for more secure, just and sustainable African urban food systems.