9723466 Lisman The long-term goal of this project is to understand the network and cellular basis of short-term memory, the type of memory used in remembering brief lists of items such as a phone number. Psychophysical tests on humans have shown that there is a limit of about 7 items that can be stored in short-term memory. The proposed research will extend previous theoretical work showing how a single brain network can keep 7 memories active at the same time using a multiplexing scheme. This scheme is closely tied to physiological observations on brain oscillations; multiplexing is achieved by a clocking mechanism, which is apparent in brain recordings as nested oscillations in which about 7 gamma (20-80Hz) oscillations occur within a single theta (4-10Hz) cycle. Memories become serially active in sequential gamma cycles. The psychophysical analysis of human memory provides a rich quantitative data set. The goal of the proposed work is to a develop a more detailed physiological model that correctly predicts this data. The first specific aim is to provide a specific model of how memories are scanned during a recognition memory test. Computer simulation methods will be used to model brain networks in a physiologically plausible way. The second aim is to understand a surprising constancy observed by Cavanagh: the time required to scan a memory and the number of memories that can be held both vary with the complexity of the memory; however the product of the two appears to be a universal constant. The third aim is to understand whether there are fundamental network constraints that limit how fast gamma oscillations can be; if they were faster, more memories could be held, but there may be reasons why gamma cannot be faster. Preliminary results related to the first two aims indicate that theta oscillation frequency may decrease as the number of stored memories increases. Furthermore, gamma frequency may decrease as the complexity of stored items increases. These are important findings because they provide a way of relating easily manipulatable aspects of short term memory to brain signals and because they are very testable. As the fourth goal of this proposal, collaborative efforts will be initiated to test these predictions.
