Diet-related greenhouse gas emissions contribute substantially to climate change, yet little is known about how consumers represent the carbon footprints (CF) of close substitutes that compete side by side on supermarket shelves. We address this gap by introducing the Food Twin Method, which captures realistic substitution decisions by pairing higher- and lower-CF alternatives within three supermarket categories: ready-to-eat meals, dairy products, and fruits and vegetables. In a between-subjects survey (N = 2,069), participants estimated the CF of 18 items presented either in a structured, shelf-like format grouped by category and Food Twins or in an unstructured random order. Participants generally identified which product had the higher CF, indicating reliance on broad categorical cues. However, judgments were strongly compressed: participants showed limited sensitivity to the magnitude of differences, frequently overestimated low-CF items, and flattened large real gaps between substitutes, including within-category extremes. Structured presentation shifted responses but did not improve discrimination accuracy; it reduced misordering mainly by increasing equal judgments rather than correct discrimination, consistent with greater averaging. These patterns align with the notion that food choices occur in a wicked learning environment where salient but weak cues dominate and feedback about emissions is scarce. By focusing on within-category substitution, the Food Twin Method reveals perceptual blind spots that cross-category designs can miss, including failures to detect high-CF “free riders” and low-CF “hidden gems.” The findings inform carbon-label design and broader choice-architecture interventions aimed at improving sensitivity to emission differences in realistic food environments.

Collective Appetite Project