9319549 MURNANE The overall goal of this research is to understand how context information affects recognition memory. This will provide a deeper understanding of the ways in which the learning environment and the environment in which remembering takes place interact and will be useful in developing optimal learning environments and in providing aid to the memory impaired. Many current theories of memory divide information into two broad classes. The information one is trying to remember is called item or target information. Other information that is stored in memory with the target information is called context information. Context information may include information about the physical environment in which the target information was learned (environmental context), stray thoughts that occurred while the target information was learned (semantic context), or the learner's mood while the target information was stored in memory (emotional context). Context information plays a critical role in many explanations of how people can retrieve individual episodes or events from among the many similar events stored in memory. However, studies that use recognition to measure memory performance often fail to find that changes in context information have an effect, raising serious questions about whether context information in fact plays the critical role assigned to it by current theories. Murnane will address the role played by context information in recognition through the continued development and testing of his own context model, a formal model of context and recognition. The context model is a generalization of several computational memory theories that share the characteristic that recognition decisions are based on information derived from a large set of items stored in memory rather than on information derived only from the memory representation of the test item. Murnane will test his model against two dominant hypotheses about why manipulations of context information sometimes fail to produce predicted results, the mental reinstatement hypothesis and the outshining hypothesis. The basic idea of the mental reinstatement hypothesis is that subjects imagine, or mentally reinstate, the learning context when tested in a different context, and then use this mentally reinstated context information in making recognition decisions. Mental reinstatement is assumed to be just as effective as physically reinstating the context (e.g., by placing the subject back in the learning context during the recognition test). Murnane's context model makes different predictions from the mental reinstatement hypothesis; these will be tested in a series of experiments that will (a) clarify the conditions under which mental reinstatement is effective, (b) determine whether mental reinstatement is as effective as physical reinstatement, and (c) determine whether mental reinstatement is equally effective with different types of context information. The second popular hypothesis that attempts to explain the conflicting recognition results, the outshining hypothesis, is that item information is such a strong memory cue relative to the context information that it masks, or "outshines" any effects caused by changes in context information. Preliminary development of the context model has produced predictions that are contradictory to the outshining hypothesis and preliminary studies support the context model. Several experiments will provide a critical test of the two hypotheses. In addition, other predictions from the context model will be tested, the model will be extended to cover different types of context information, a version of the model that is capable of making quantitative predictions will be constructed, and the model will be extended to account for a broader set of empirical data than covered by current models. For example, the model predicts that the effects of changes in the relative strengths of context and item information on recognition differ depending on whether item or c ontext information is strengthened. Several studies will address how changes in each type of information affect recognition. ***
