David S. Olton
Matthew L. Shapiro
Stewart H. Hulse
The Johns Hopkins University
Memories differ in the extent to which the information to be remembered is associated with the context in which it was learned. Some information is strongly linked with this temporal/personal context, while other information is not. This distinction has proved useful in many different fields: predators searching for prey in natural habitats ( Olton, Handelmann, & Walker, 1981), animals learning discrimination tasks in the laboratory ( Honig, 1978), memory processing in humans ( Tulving, 1972), and the amnesic syndrome following brain damage in both animals ( Olton, Becker, & Handelmann, 1979; Olton, Becker, & Handelmann, 1980) and humans ( Rozin, 1976; Schacter & Tulving, 1983; Squire & Cohen, 1983; Winocur, 1982).
Several different theoretical frameworks and sets of terminology have been introduced to investigate the importance of temporal associations in memory. The one most relevant to the procedures used here is that of Honig ( 1978); working memory associates an event with its temporal/personal context, while reference memory processes information independently of its temporal/personal context.
The Delayed Matching-To-Sample task is one example of a task requiring working memory. At the beginning of each DMTS trial, a sample stimulus is presented and then removed. After a delay interval, this sample stimulus is presented together with other stimuli, and the animal must choose the original sample to got a reward. Because the sample stimulus varies from trial to trial, the animal must remember the specific sample presented at the beginning of each particular trial. Working memory is the process responsible for associating the sample stimulus with the trial in which it is presented.
In contrast, reference memory is the process responsible for general rules and procedures which are useful for all trials of a task. Every procedure has some reference memory components, including one like delayed matching-to-sample, which also has a working memory component The reference components of delayed matching-to-sample include the rules of the task that remain the some from trial to trial such as: don't respond during the time out, make an initiating response at the beginning of the trial to turn on the sample stimulus, wait during the delay, get food from the food hopper when the feeder operates, etc. Thus, information in reference memory can effectively guide behavior which is not dependent on any particular trial, but is based upon rules, procedures, and