Research on discrimination-based transitive inference (TI) has demonstrated a widespread capacity for relational inference in people and non-human animals. In this domain individuals learn to choose the reinforced item in a set of interrelated discriminations (e.g., A–/B+; B–/C+) and are tested for transitive inference on novel, non-adjacent pairs (e.g., A vs. C). Existing models suggest that transitive responding can be supported by associative learning mechanisms, but they fail to account for evidence that knowledge about the hierarchical nature of the task, whether instructed or discovered during training, has a dramatic influence on learning. I present a model which formalizes TI as the estimation of items’ positions along a latent dimension and tracks learners’ uncertainty about the mapping between item position and feedback. The model naturally accounts for standard effects in TI, while going beyond associative models in explaining the effects of knowledge and rich feedback on relational inference.

Self-directed learning confers a number of advantages relativeto passive observation, including the ability to test hypothe-ses rather than learn from data generated by the environment.However, it remains unclear to what extent self-directed learn-ing is constrained by basic cognitive processes and how thoselimits are related to the structure of the to-be-learned material.The present study examined how hypothesis generation af-fects the success of self-directed learning of categorical rules.Two experiments manipulated the hypothesis generation pro-cess and assessed its impact on the ability to learn 1D and 2Drules. Performance was strongly influenced by whether thestimulus representation facilitated the generation of hypothe-ses consistent with the target rule. Broadly speaking, the find-ings suggest that the opportunity to actively gather informa-tion is not enough to guarantee successful learning, and thatthe efficacy of self-directed learning closely depends on howhypothesis generation is shaped by the structure of the learn-ing environment.

Decision making research typically focuses on choices between predetermined sets of options. In many real-world de-cisions, however, individuals must generate potential courses of action themselves. Individual differences in processesinvolved in option generation therefore influence which actions are considered. We examined the role of one such fac-tor: the propensity to take risks. We hypothesized that risk-taking propensity would be related to the generation of morerisky actions associated with uncertain or unfavorable outcomes. Participants generated options in ill-structured situationsand rated the perceived risk associated with each option. As predicted, higher risk-taking propensity was associated withincreased generation of risky options that could lead to unfavorable outcomes. The riskiness of generated options wasalso related to affective state, consistent with prior evidence of emotional influences on risky decision making. The find-ings suggest that both real-life risk-taking and risky option generation arise from common cognitive processes involved inresponding to uncertainty.