Levels of Processing in Human Memory

Abstract

Throughout my professional career as a cognitive psychologist, I have been interested in the topics of memory, attention, perception, and thinking – how best to characterize them, how they relate to each other, and how they change over a person's lifespan. In everyday life, these mental activities are usually considered to be rather different from each other – remembering meeting someone a month ago seems different from seeing the person in front of you – and this separation is often echoed in psychology textbooks, in which perception, memory, and decision-making are treated in different chapters. Much of the current thinking in cognitive psychology has reacted against this commonsense view, however, suggesting instead that these areas of study are better regarded as closely related and interacting aspects of one general processing system. This latter position was one starting point for the formulation of the levels of processing (LOP) framework for memory research proposed by Robert Lockhart and myself in 1972. The LOP article with Lockhart, plus a later empirical article with Endel Tulving in 1975, are my most-cited pieces of published research, and may therefore be regarded as my best-known scientific contributions to cognitive psychology. Additionally, the general ideas in which the LOP framework was embedded – for example, that remembering should be regarded as an activity of mind rather than a collection of structural “memory traces” waiting to be revived – have always been central to my thinking about memory. Thus, the LOP paper and its spinoffs have been the starting point for much of the work that my lab has produced over the years. In the 1960s, ideas about learning and memory were changing from the belief that the formation of associations between two mental events was the crucial element, to concepts derived from information-processing theories. From this latter point of view, the brain/mind was regarded as a highly sophisticated computer, processing sensory information from the environment, performing computations on that information, and finally translating the products into relevant actions. To accomplish these operations efficiently, the proposed system needed a variety of memory stores, holding information of different qualitative types either temporarily, while it was processed, or relatively permanently, in the case of learned knowledge.