A list-length constraint on incidental item-to-item associations.

We investigated the possibility that item-to-item associations form between items concurrently included in a capacity-limited region of working memory, but not outside of that region. Many studies indicate a central capacity limit of three to five items (e.g., Cowan Neuropsychologia 49:1401-1406, 2001). Participants received lists of three, six, or nine words along with an orienting task, selecting the most interesting word from each list. Consistent with expectations, a subsequent, unexpected test showed that memory of whether two words came from the same list or not was superior for three-word lists, as compared with six- and nine-word lists, which did not differ. This effect occurred even though the separation between the list positions of the two probe words was controlled across list lengths. The study demonstrates a source of implicit learning that depends upon a limited-capacity working memory faculty, a finding that should inspire further research on the function of working memory in long-term learning.