Crime is often portrayed as an anti-social act, yet many offences are in fact the product of coordinated social behaviour. This raises two key questions: how does co-offending emerge, and how does collaboration between offenders influence their criminal repertoire? This talk presents a dissertation that addresses these questions by combining a synthesis of co-offending patterns with computational modelling of the mechanisms underlying them. First, drawing on a systematic review and meta-analysis, the study provides a comprehensive overview of co-offending across demographic groups and offence types, highlighting variation in prevalence, group structure, and partnership stability. Second, the talk introduces agent-based models to examine how social and environmental processes shape co-offending networks and learning dynamics. A homophily-based model shows that similarity-driven partner selection promotes stable partnerships but may limit the diversity of skills acquired. A complementary foraging model demonstrates how environmental constraints and access to partners influence both collaboration patterns and opportunities for learning. Together, these findings conceptualize co-offending as a social learning process shaped by network structure, partner selection, and environmental constraints. By integrating descriptive evidence with computational modelling, this work contributes to a more nuanced understanding of how criminal collaborations emerge and facilitate the spread of criminal skills.

Please note that this is an internal talk. If you want to participate, please write an email to .