This thesis addresses how firms learn to innovate by using external information and knowledge in the face of limited absorptive capacity. Innovation is an interactive process that requires firms to reach beyond their boundary in search of external knowledge. However, very little is known about this process in developing countries. Using data from a developing country, Nigeria, I provide fresh evidence to the effect that interactive learning generally helps firm-level innovation. However, the positive relationship is contingent upon a firm's current level of innovativeness. As firms become more innovative, it seems less useful for them to rely on a broad spectrum of external sources. Going beyond the limits of the empirical analyses, a new agent based model (ABM) is developed to shed more light on the relationship between interactive learning, absorptive capacity and innovation. Following the notion that networks are a basic infrastructure for knowledge diffusion, the ABM analyses the role of absorptive capacity in the evolution and benefits of innovation networks under different knowledge regimes. Social capital is de-emphasised and firms ally purely for knowledge sharing. Absorptive capacity, which is itself influenced by cognitive distance, drives partner selection. Besides eliciting empirically observed network properties, the model simulation offers novel insight on the coevolutionary process between networks, industrial dynamics and absorptive capacity. In the early stages of an industry when knowledge is highly tacit, firms benefit more from centralised network positions; the reverse is the case as the industry becomes more mature. And firms’ absorptive capacity indeed influences the rate and direction of network evolution. Networks emerge or change depending on firms’ partnership decisions, which are shaped by their absorptive capacity and that of the potential partners. I put forward some implications for policy and practice based on the empirical and simulation results.