In searching for solutions, organizational decision-makers often exhibit local search, limiting their long-term adaptability in complex environments. Existing explanations attribute local search to either cognitive (e.g., absorptive capacity) or social constraints (e.g., conflicts) that prevent agents from pursuing distant alternatives. However, recent experimental studies find that individuals may still engage in local search even without such constraints. Here, we introduce “learning through generalization” as a mechanism that can produce local search behaviors independently. Using computational modeling, we demonstrate that locality in search behaviors can endogenously emerge through generalization, especially when generalization breadth—i.e., the extent to which agents generalize experience from one alternative to others—is neither too narrow nor too broad. Ironically, our result shows that the intermediate generalization breadth, while producing local search, can actually help agents reach the global peak. Our model contributes to a nuanced view of local search and highlights the intertwined relationship between learning and search in adaptation.
