Learning in a Dynamic Decision Making Task: The Recognition Process

The Instance-Based Learning Theory (IBLT) suggests Recognition, the ability to discriminate among familiar classes of objects, is the first determinant of decision-making in complex tasks (Gonzalez, Lerch, & Lebiere, Conditionally accepted). According to IBLT, the main object of knowledge is an instance (described by the situation-decision-utility information). This paper investigates the acquisition of Recognition skills by ana lyzing the instances generated with extensive practice in a dynamic decision making task. We present three results. First, we examined the instances generated by decision makers using similarity measures provided by Latent Problem Solving Analysis (LPSA). LPSA analyses demonstrate that, decision makers make choices in situations increasingly more similar to past conditions as they practice in the task. Second, using multiple regression analysis and cluster analysis we identified groups of decisions that share similar features within each group. We found that the features for the most significant clusters vary as people practice in the task. Decision makers learn when to intervene most effectively as they practice in the dynamic task. Third, we investigated the effects of humans’ cognitive abilities on learning and instance similarity. We found significant correlations between general cognitive abilities and performance, but not between abilities and similarity.