In Remembrance of William K. Estes
National Medal of Science recipient William K. Estes passed away on August 17th at the age of 92. His long and productive career encompassed the science of learning and memory from behaviorism to cognitive science, with seminal contributions to both. He was also an active member of APS and was the founding editor of the journal Psychological Science.
Estes (born June 17, 1919) began his graduate studies under the tutelage of B. F. Skinner during the early 1940s. Together, Estes and Skinner developed a conditioning paradigm, called conditioned suppression, which represented a new technique for studying learned fear (Estes & Skinner, 1941). Being able to measure rats’ freezing behavior in response to a tone that predicted an upcoming shock allowed Estes to quantify trial-by-trial changes in the learned response. Within a few years, this paradigm became one of the most widely used techniques for studying animal conditioning, and it is still in use today.
As soon as he completed his PhD, Estes was called into military service. When the war ended, Estes returned to the United States and to the study of psychology. Much to Skinner’s dismay, Estes soon began to stray from his mentor’s strict behaviorism. He started to use mathematics to describe mental events that could only be inferred indirectly from behavioral data, an approach quite unacceptable to behaviorists.
Estes built on Hull’s mathematical modeling approach to develop new methods for interpreting a wide variety of learning behaviors (Estes, 1950). In contrast to Hull and the learning theory of the time, Estes suggested that what seems to be a single stimulus, such as a yellow light, is really a collection of many different possible elements of yellowness, only a random subset of which are noticed (or “sampled,” in Estes’s terminology) on any given training trial. Only those elements sampled on the current trial are associated with the outcome or reward. Over time, after many such random samples, most elements become associated with the correct response. At this point, any presentation of the light activates a random sample of elements, most of which are already linked with the response.
Estes’s idea, which he termed stimulus sampling theory, gave a much better account than other theories (such as Hull’s) of the randomness seen in both animal and human learning, and it helped to explain why even highly trained individuals don’t always make the same response perfectly every time: on any given trial, it’s always possible that (through sheer randomness) a subset of elements will be activated that are not yet linked to the response. In other applications, Estes showed how stimulus sampling theory explains how animals generalize their learning from one stimulus (e.g., a yellow light) to other, physically similar stimuli (e.g., an orange light), as Pavlov had demonstrated back in the 1920s.
Estes’s work marked the resurgence of mathematical methods in psychology, reviving the spirit of Hull’s earlier efforts. Estes and his colleagues established a new subdiscipline of psychology, mathematical psychology, which uses mathematical equations to describe the laws of learning and memory. From his early work in animal conditioning, through his founding role in mathematical psychology, to his later contributions to cognitive psychology, W. K. Estes continued to be a vigorous proponent of mathematical models to inform our understanding of learning and memory.
– Mark Gluck
Professor of Neuroscience,
Center for Molecular & Behavioral Neuroscience
Rutgers University – Newark
Co-Director, Memory Disorders Project at Rutgers-Newark