Challenges Facing Contemporary Associative Approaches to Acquired Behavior

Abstract

Despite the considerable success of contemporary associative models of learning in stimulating new behavioral research and modest success in providing direction to both neuroscience and psychotherapy, these models are confronted with at least three challenges. The first challenge is to the assumption that animals encode only one or a few summary statistics to capture what has been experienced over many training trials. This assumption is contrary to overwhelming evidence that the brain retains episodic information. The second challenge is that the learning-performance distinction has been largely ignored. Most models erroneously assume that behavior is a nearly perfect reflection of what has been encoded. The third challenge is to account for interactions between stimuli that have been presented separately (e.g., stimulus interference) as well as between stimuli that have been presented together (e.g., stimulus competition).The purpose of this review is to assess the common denominators of most contemporary associative models of learning as a group (as opposed to specific models), with an emphasis on the major challenges facing these models. By associative, I am referring to models designed to account for Pavlovian responding, although these models have often been extended quite successfully to instrumental behavior as well. The qualifier contemporary is necessary because new associative models are likely to be proposed in the future that do not subscribe to these common denominators. Miller and Escobar (2001) discussed the dangers of contrasting whole families of models rather than specific models. Here I try to minimize this danger by limiting my remarks to families of contemporary models rather than families of models including past and future instances. There will also be some discussion of the major alternatives to contemporary associative models in order to appreciate how well they address the problems confronting associative models. The paper is organized around three basic problems: the assumption that animals encode only summary statistics about prior experiences, the assumption that behavior is a nearly veridi-

An instance theory of associative learning

“…First, differently from the summative learning theories, an instance-based model acknowledges evidence that nonhuman animals have good long-term memory for the specific (Fagot & Cook, 2006;Vaughan & Greene, 1984). Second, Minerva-AL distinguishes encoding from retrieval, a point that has received rare consideration in formal learning theory (see, however, Bouton, 1993;Miller, 2006;Miller & Matzel, 1988;Stout & Miller, 2007). For example, consider how Minerva-AL explains backward blocking.…”

Connectionism and Classical Conditioning

“…To my mind, this is one of the intriguing aspects of one of the challenges to modern associative theories (Miller, 2006). It has been argued that most of these theories only provide an account of how associative strengths are modified (learning), and do not provide an account of how they are converted into responses (performance).…”

“…This theory merely states that there is a monotonic relationship between association and response (Miller, 2006). Rescorla and Wagner assumed that increases in associative strength would not result in decreases in response magnitude.…”

“…Importantly, the Rescorla-Wagner model is not a model of behavior (Miller, 2006). It describes how associative strengths change over time.…”

“…What is the formal relationship between these learning rules and the Rescorla-Wagner model? Might these different activation functions and learning rules be more related to other models that have emerged in an attempt to deal with some of the problems that the Rescorla-Wagner model has encountered (Mackintosh, 1975;Pearce & Hall, 1980)? Miller (2006 has called for renewed emphasis on reconciling the distinction between learning and performance.…”