Food waste as a complex social system: How computational social science can help

Food waste isn’t just the result of bad habits or clunky logistics. It emerges from a web of social norms, motivations, and constraints that interact in ways that are hard to predict. Traditional research can measure correlations and test single interventions, but it often misses how behaviors ripple through groups, places, and time.

Computational social science (CSS) changes that equation. By building models that reproduce patterns of human behavior, researchers can test hypotheses and explore “what if?” scenarios in silico before moving to the real world. It’s an approach long used in physics and engineering, and it’s increasingly central to tackling complex societal problems.

From methods to impact: CMSS at NORCE

The Center for Modeling Social Systems (CMSS) at NORCE is one of Europe’s leading teams applying CSS to public-good challenges. The group combines multiple modeling paradigms to match the problem at hand:

• Agent-based modeling to simulate individual decision-making and social influence
• Microsimulation to represent diverse households or firms at scale
• System dynamics to capture feedback loops and accumulations
• Participatory modeling to integrate stakeholder knowledge into design and validation

With rigorous calibration and validation, these models can represent complex systems credibly—surfacing insights that simpler methods rarely reveal.

The social complexity of food waste

Food waste is a climate, resource, and food-security problem. But cutting waste isn’t only a technical exercise; it hinges on social patterns, including:

• Injunctive norms and values (e.g., thrift, indulgence, sustainability)
• Descriptive norms (what people perceive others to do)
• Opportunities and constraints (time pressure, plate size, availability)
• Abilities and skills (meal planning, storage, self-control)

Because these factors interact, surveys or isolated experiments can only go so far. That’s why CMSS contributed simulation models to the Horizon Europe CHORIZO Project, which aims to reduce food waste across the supply chain.

Model 1: Hotel breakfast buffets

Working with a European hotel chain, CMSS built an agent-based model to simulate buffet behavior. Each guest is an “agent” with attributes such as gender, travel purpose (business/leisure), and attitudes toward food. Decisions around portioning and waste are shaped by injunctive norms (hunger, indulgence, sustainability awareness), opportunities (plate size, time available), abilities (self-control), and social influence.

After verification and validation, the team ran 36 scenarios with 150 simulation runs each—varying plate size, guest mix, and messaging strategies (positive prompts vs. provocative cues). The pattern-level takeaways were clear:

• Smaller plates and subtle prompts to “take what you’ll eat” tended to reduce waste without hurting guest satisfaction.
• Provocative messaging could backfire among some segments, increasing waste or shifting it across courses.
• Leisure-heavy mornings differed from business-heavy ones; time pressure and norms influenced piling behavior.
• Combining design changes (plate size, placement) with social cues outperformed single levers.

Running such experiments in a live hotel would be costly and risky. The simulation enabled safe, data-grounded exploration of strategies before any real-world rollout.

Model 2: Household waste in Belgium and Spain

CMSS also developed a microsimulation of household food management calibrated with data from Belgium and Spain. Each household varies in size, income, diet, shopping frequency, storage practices, and attitudes. The model tracks purchasing, consumption, and discard behaviors across thousands of unique households.

Across 24 scenarios with 100 runs each, two insights stood out:

• Small changes add up: modest shifts—like better meal planning, freezer use, and clearer date-label understanding—produced meaningful population-level waste reductions.
• Targeted interventions work: focusing on high-waste segments (e.g., busy families with irregular schedules) delivered outsized gains compared with broad, generic campaigns.

In short, incremental behavioral improvements, strategically targeted, can drive system-wide impacts.

Why CSS matters for policy and practice

• It captures dynamics. Feedback loops, peer effects, and context-specific trade-offs become explicit.
• It de-risks experimentation. Hundreds of scenarios can be tested virtually before live pilots.
• It reveals heterogeneity. What works for business travelers may not work for leisure guests; households differ, too.
• It integrates evidence. Field data, expert knowledge, and stakeholder insights can be combined in one framework.
• It guides design. Models point to high-leverage combinations—nudges plus design tweaks, or targeted supports for specific groups.

Beyond food waste: a broader portfolio

CMSS’s multidisciplinary team is known for participatory modeling, involving stakeholders from the start so models reflect on-the-ground realities and decision constraints. The group works with governments, industry, and international organizations on topics such as green transport, climate mitigation, sustainable fisheries, and democratic innovation.

The common thread: using computation to build robust, transparent tools that support real-world decisions.

The path forward

Reducing food waste requires more than technical fixes; it demands understanding—and reshaping—the social systems that produce waste in the first place. The CHORIZO project shows how CSS can help: from hotel buffets to household kitchens, simulations reveal effective, scalable, and socially informed strategies that would be costly or infeasible to test at scale in vivo.

As computational methods that revolutionized physics and computer science move deeper into the social realm, the potential payoff is substantial: smarter policies, better-designed interventions, and faster learning cycles.

A better world isn’t found, it’s built.