How similar should collaborators be in inter-organizational learning: Optimal cognitive proximity and knowledge complexity.

Effective knowledge exchange is critical for innovation in knowledge-intensive sectors, yet the relationship between cognitive proximity and collaborative innovation remains underexplored in complex knowledge environments. While prior studies confirm a non-linear, inverted U-shaped relationship where moderate similarity balances absorptive capacity and novelty, how this relationship changes with increasing knowledge complexity is poorly understood. This study investigates how knowledge complexity moderates the cognitive proximity-innovation relationship in the U.S. biotechnology sector using a comprehensive dataset of over 650,000 patents from over 57,000 unique assignee organizations spanning 1976 to 2024. We distinguish between two dimensions of collaborative innovation: collaboration volume and collaboration quality. Cognitive proximity was measured through CPC-based Jaccard similarity, while knowledge complexity was operationalized using a structural complexity framework based on knowledge combination networks. Negative binomial regressions reveal that knowledge complexity moderates the proximity-innovation relationship differently across the two dimensions. For collaboration volume, higher complexity shifts the optimal proximity point rightward and flattens the curve while maintaining the inverted U-shape. For collaboration quality, complexity produces a leftward shift and, remarkably, leads to a complete breakdown of the curvilinear relationship at high complexity levels, where the relationship becomes nearly horizontal. These contrasting patterns indicate that cognitive proximity operates through fundamentally different mechanisms in partner selection versus innovation realization. While absorptive capacity considerations dominate partnership decisions even under high complexity, novelty-generating mechanisms become disconnected from proximity effects as complexity rises. The findings refine proximity theory by demonstrating that knowledge complexity serves as a structural moderator and offer actionable insights for partner selection strategies in advanced innovation ecosystems.