First direct evidence of two stages in free recall and three corresponding estimates of working memory capacity

Abstract

I find that exactly two stages can be seen directly in sequential free recall distributions. These distributions show that the first three recalls come from the emptying of working memory, recalls 6 and above come from a second stage and the 4th and 5th recalls are mixtures of the two. A discontinuity, a rounded step function, is shown to exist in the fitted linear slope of the recall distributions as the recall shifts from the emptying of working memory (positive slope) to the second stage (negative slope). The discontinuity leads to a first estimate of the capacity of working memory at 4-4.5 items. Working memory accounts for the recency effect. The primacy effect comes from the second stage with a contribution also from working memory for short lists (the first item). The different slopes of the working memory and secondary stages, and that the two have different functional forms, accounts for the u-shaped serial position curve. The total recall is shown to be a linear combination of the content of working memory and items recalled in the second stage with 3.0-3.9 items coming from working memory, a second estimate of the capacity of working memory. A third, separate upper limit on the capacity of working memory is found (3.06 items), corresponding to the requirement that the content of working memory cannot exceed the total recall, item by item. This third limit presumably corresponds to the least chunked item. This is the best limit on the capacity of unchunked working memory.