The present experiments were designed to examine the hypothesis that the mammalian brain contains anatomically distinct memory systems. Rats with bilateral lesions of caudate nucleus or fimbria-fornix and a control group were tested postoperatively on 1 of 2 versions of the radial maze task. In a standard win-shift version, each of the 8 arms of the maze was baited once, and the number of errors (revisits) in the first 8 choices of each trial was recorded. Fimbria-fornix rats were impaired in choice accuracy, while caudate animals were unimpaired relative to controls. Different groups of rats with similar lesions were tested on a newly developed win-stay version of the radial maze, in which the location of 4 randomly selected baited arms was signaled by a light at the entrance to each arm, and which required rats to revisit arms in which reinforcement had been previously acquired within a trial. Rats with fimbria-fornix lesions were superior to controls in choice accuracy on the win-stay radial maze task, while caudate animals were impaired relative to controls. The results demonstrate a double dissociation of the mnemonic functions of the hippocampus and caudate nucleus. Some implications of the presence of 2 memory systems in the mammalian brain are discussed.

No abstract

No abstract

No abstract

This paper argues that the hippocampus is involved in mnemonic processes by virtue of being necessary to the execution of conditional operations occurring during the retrieval of acquired information from memory. A series of experiments indicating that animals suffering from hippocampal dysfunction are incapable of behaving conditionally in learning situations is described. Other effects of hippocampal dysfunction are then explained in terms of disruption of conditional retrieval processes. The paper continues by defining retrieval as the selection of one piece of information from among many possible candidates. It is argued that such a process is of little utility in basic associative theories, as associative processes result in one predominant memory for any given stimulus. Cognitive theories, in contrast, have held that many things may be known about a given stimulus and thus selection among memories is mandatory. The process of conditional retrieval is then defined as the result of interaction between a categorical operator, representing global factors, and a local operator, representing local considerations. Interaction between these two types of operators is shown to be necessary for constructing dimensional representation, including maps. To hold that the hippocampus is involved in the construction of dimensional representations in general affords an opportunity to begin relating cognitive processes to physiological knowledge and to answer some of the broad conflicts between associative and cognitive theories of learning. It is further suggested that studies of hippocampal development may shed light on cognitive development.Studies on the function of brain structures in producing behavior must be done within some sort of psychological framework; otherwise the results cannot be interpreted. The capability of the framework in interpreting the results is, in turn, an indicator of the validity of the framework. Thus, studies of the behavioral function of structures deal not only with neurological mechanisms, but also with psychological processes. Studies of the role of the hippocampus in producing behavior have something quite important to say about the nature of learning and memory and, very possibly, about cognitive development. Memory