Attribute Theories of Memory

Abstract

Abstract Attribute theories propose that a cognitive system constructs representations with constituent elements that reflect the attributes, features, properties, or characteristics of the world and the things in it. These theories are flexible; attributes may be used to describe tangible physical things in one’s environment but also intangible things such as plans and concepts. In the domain of memory, attribute theories describe how representations of one’s experience are stored, retrieved, and manipulated by the cognitive system. The mathematical language of linear algebra is central to these theories. Vector operations are used to define various measures of representational distance and similarity. Matrices are used to define associative structures such as synaptic weights in neural network models and sets of memories in instance theory models. This chapter examines some foundational attribute-based approaches to memory theory, including early attempts to define psychological similarity, and the influential stimulus-sampling theory. It reviews important empirical phenomena involving memory attributes, including proactive interference, encoding specificity, and source memory. Finally, the chapter considers how attribute-based approaches have influenced modern cognitive neuroscientific investigations and theory.