@unpublished{cogprints9775, month = {September}, title = {Locating Working Memory in Word Free Recall and Estimating its Capacity at Less Than 3.1 +/- 0.2 Words}, author = {Dr. Eugen Tarnow}, year = {2014}, keywords = {recall; working memory; short term memory; memory search}, url = {http://cogprints.org/9775/}, abstract = {The evidence for limited capacity is strong yet there is a controversy as to whether this limited capacity can be seen in word free recall data. After five decades since the first word free recall experiments, I present clear direct evidence for a dual store model in word free recall. I show that there is a discontinuity in the item probability during the first recall, the recall that is the most likely to originate in working memory. This discontinuity indicates a separation between the items that are initially in working memory and the items which are retrieved from elsewhere at a later time, presumably via an association with one or more working memory items. The items that are not initially in working memory are those that are the least likely to be recalled overall. Assuming that the first recall is an accurate statistical representation of working memory, and that the probability of an item in working memory cannot exceed the overall probability of recalling that item, I arrive at an upper limit on the working memory capacity of 3.1 +/- 0.2 words.} }