Australasian Mathematical Psychology Conference 2019

When memories fade do sampling effects linger?

Keith Ransom
School of Psychology, The University of Adelaide
Amy Perfors
University of Melbourne
Danielle Navarro
University of New South Wales

When describing their Bayesian framework for generalisation Tenenbaum and Griffiths (2001) noted that many different different sampling assumptions are possible on the part of the learner. A growing body of literature has demonstrated that different assumptions may lead to qualitatively different patterns of inference. But relatively little is known about how the effects of sampling assumptions are represented mentally, or whether and how these effects might persist. We report the results of a single category generalisation experiment aimed at exploring these issues. By systematically varying both the sampling cover story and whether it is given before or after the training stimuli are presented we are able to determine whether encoding or retrieval issues drive the effect of sampling assumptions. We find an effect of sampling cover story when it presented before the training stimuli, but not after, which we interpret in favour of an encoding account. Further, when we compare the performance of people who had to rely on their own memory of training stimuli against those who did not, we find that generalising “from memory” increases willingness to extend the target category to novel items.

Tenenbaum, J. B., & Griffiths, T. L. (2001). Generalization, similarity, and Bayesian inference. Behavioral and Brain Sciences, 24(04), 629–640.