In this paper, we investigate how behavioral contagion in terms of mimetic strategy learning within a social network would affect the asset price dynamics. The characteristics of this paper are as follows. First, traders are characterized by bounded rationality and their adaptive learning behavior is represented by the genetic programming algorithm. The use of the genetic programming algorithm allows traders to freely form forecasting strategies with a great potential of variety in functional forms, which are not predetermined but may be fundamental or technical or any mix of these two broad categories, as they need to adapt to the time-varying market environment. The evolutionary nature of the genetic programming algorithm has its merit for modeling mimetic behavior in the context of information transmission in that, other than making duplicates of an entire trading rule as if a mind-reading technique exists, strategy imitation could take place down to the level of building blocks that genetic operators work out or pieces of information that constitute a strategy and are more ready to be transmitted via word-of-mouth communication, which is more intuitive compared to the existing literature. Second, the traders are spatially heterogeneous based on their positions in social networks. Mimetic learning thus takes part in local interactions among traders that are directly tied with each other when they evolve their trading strategies according to the relative performance of their own and their neighbors'. Therefore, specifically, we aim to analyze the effect of network topologies, i.e., a regular lattice, a small-world, a random network, a fully connected network, and a preferential attachment network, on market dynamics regarding price distortion, volatility, and trading volume, as information diffuses across these different social network structures.Index Terms-Agent-based modeling, artificial stock market, genetic programming, social network.