Hedonic and practicality beliefs form a behaviour-proximal core associated with red and processed meat intake: An...

Red and processed meat (RPM) consumption remains widespread in high-income populations despite consistent health and sustainability recommendations to reduce intake. Although attitudinal measures reliably predict meat-related intentions and intake, prior work has relied predominantly on aggregate affective and cognitive composites, which conceal how the evaluative system is organised: which evaluations are most structurally embedded, which bridge distinct clusters of meaning, and which sit closest to behaviour. We addressed this by applying an item-level Causal Attitude Network approach in a preregistered cross-sectional survey of 1554 UK adults. A regularised partial-correlation network (EBICglasso) was estimated on 21 evaluative items and a past-week RPM intake sum score; we examined community structure, node centrality, between-community connectivity, and the intake node's neighbourhood, complemented by hierarchical regressions on affective and cognitive composites. Walktrap community detection identified four communities (Hedonics, Morality, Benefits/Vitality, Practicality/Feasibility), with the intake node located within Practicality/Feasibility alongside planning, health, and sustainability-related content. The most central evaluative reactions reflected sensory reward and moral evaluation (e.g., pleasantness, tastiness, moral acceptability). Intake showed its strongest conditional association with planning, with smaller links to strength, pleasantness, sustainability, comfort, and pleasure. Between-community connectivity was concentrated in a few pathways, most notably Hedonics-Practicality/Feasibility and Practicality/Feasibility-Benefits/Vitality. Sex-stratified networks were highly similar despite higher intake among men. In regressions, the composites raised explained variance in intake from R² = .05 (demographics) to R² = .39, with a stronger association for the affective (β = .45) than the cognitive composite (β = .18). The architecture suggests, as hypothesis-generating rather than causal claims, that information-focused appeals about long-term consequences may have limited leverage and that intervention targets more plausibly lie in planning, feasibility, hedonic reliance, and benefit/vitality reframing.