22 - Models of Animal Learning and Their Relations to Human Learning

Summary

This chapter reviews a contingency learning against the background of recent formal models of animal learning. It reviews a very substantial amount of research including not only human causal and predictive learning but also category learning and multiple-cue probability learning. The development of theoretical models of predictive learning has been stimulated to an enormous extent by demonstrations that cues compete with each other to gain control over behavior (so-called cue interaction effects). In causal learning scenarios, the cue and outcome are provided, via the instructions, with particular causal roles. In most cases, then, the cues are not only potentially predictive of the outcome but also cause it. Despite the challenging nature of the evidence against an associative perspective as a unique account of human predictive learning, there is also evidence that the influence of causal knowledge or rule learning is not necessarily pervasive.