This study examines the behaviour of a set of heterogeneous agents whose approach to decision-making is based on individual Information Processing Rules. These reflect their attitudes to the information in their recent observations relative to their past experience. The characteristics and influence of these agents are explored in a simulated speculative market which is order-driven. Using a variant of Bayesian sequential updating, agents base their predictions on a general model that includes the random walk and incorporates some features of Technical Analysis. After an overview of the short and long-term behavioural implications of the rules as they affect individual agents, the compositional behaviour of different types is examined. These types interact within an evolutionary structure in which agents are replaced in the market according to increasingly rigorous criteria of financial success. The continuous process of renewal that eventuates indicates that increased accountability requirements, while resulting in greatly increased profits for some individuals, tend to generate an unstable market, greater risk and increased volatility.