Nourishing sustainability innovation: Scientific trajectories in industrial protein research

This paper investigates how distinct scientific trajectories inform sustainability-oriented innovation in the protein production sector. Focusing on USPTO patents in subclass A23J (1990–2015), we apply a two-step text analysis that links topic modeling of academic publications to the classification of patent texts. This method identifies four trajectories and evaluates their influence on follow-on innovation, with particular attention to citations from sustainability-related patents. The results reveal substantial heterogeneity: some trajectories, such as those centered on cellular and bioprocessing mechanisms, consistently generate sustainability spillovers, while others, such as protein chemistry, exhibit limited diffusion. These findings contribute to a view of science as a cognitive map that structures industrial research by making more explicit the underlying causal mechanisms that govern each trajectory, thereby shaping which domains are more likely to yield impactful and sustainability-aligned innovations. • Scientific trajectories in protein innovation vary in how they generate spillovers, especially for SDG-related ones. • A two-step text method links academic scientific language to patents, uncovering four distinct trajectories beyond technology classes. • Science acts as a cognitive map that structures industrial research by clarifying the causal mechanisms underlying each trajectory. • Science’s influence on sustainability innovation depends on each trajectory’s epistemic structure, challenging uniform impact assumptions.