Quantifying the value of carbon label information in food choice using drift diffusion modelling

The use of carbon labels as an intervention to increase more sustainable food consumption has seen many mixed results, with some studies showing that consumers do not utilise the carbon labels in their decisions. To address the mixed results in the literature, we present a novel and in-depth evaluation of how carbon labels work by quantifying the importance of carbon label information relative to taste preferences in food decisions via a computational modelling approach. Participants (n = 48) were presented with multiple trials of two sandwiches alongside their carbon labels. Participants' choice and response time were recorded whilst visual attention was tracked with an eye-tracking device. The Multi-attribute Attentional Drift Diffusion Model (maaDDM) was fitted to data through Bayesian STAN modelling in R. The analysis revealed that carbon labels were used to a moderate extent similar to individual taste preference in choosing sandwiches, but the extent of use varied as a function of participant's perception of the negative impact of GHG emissions (the more negative perception, the greater use of carbon labels). We further explore the insights gained from maaDDM on participant's information sampling behaviour, and discuss the implications for policies to identify a critical valuation threshold of carbon labels.