This paper focuses on utilization of multiagent systems and cognitive architectures for analysis of financial market dynamics. Sloman's H-Cogaff cognitive architecture is utilized as a blue-print to design trader architecture (Sloman 2002). The trader architecture combines a Markov-based bias mechanism and an XCS classifier based learning mechanism. An agent-based artificial stock market is designed to observe trader behavior and market price dynamics. Initial results reveal that the bias mechanism affects price dynamics and results in deviation of the stock price from the efficient market price.