Mathematical models of human memory
Misha Tsodyks, Institute for Advanced Study/Weizmann Institute
I will present my recent work on mathematical modeling of human memory, as studied with random lists of words and other items. I will argue that memory recall of random lists is governed by the universal deterministic process resulting in the analytical relation between the number of items in memory and the distribution of the number of items that can be successfully recalled. These results are well supported by the experiments on human subjects. The retention of items in memory on the other hand is not universal and differs for different types of items being remembered, in particular retention of words and sketches is different even when sketches are made to only carry information about an object being drawn. I will discuss the putative reasons for these observations and introduce the phenomenological model predicting retention curves, i.e. the probability to retain an item in memory as a function of time elapsed since it was presented.