Working memory capacity is limited and much work has been done for decades on estimating its value. However, if stimuli contain redundancies, compression mechanisms probably appear which change the point of view on that capacity. By means of a behavioral study and a computational simulation, we aim at showing that working memory is not just a fixed number of items. We first present a theoretical framework in the domain of information theory in order to analyse this point of view. Then, we show the results of an experiment studying the effects of some regularities on memory recall performance, as well as a simulation using a model of chunking and two different memory models. Our results show that it is probably wrong to consider working memory capacity as a fixed number of items. It is better to express it in terms of a quantity of information.
展开▼
