Masson M E, Caldwell J I, Whittlesea B W
Department of Psychology, University of Victoria, British Columbia, Canada.
J Exp Psychol Learn Mem Cogn. 2000 Jul;26(4):1005-22. doi: 10.1037//0278-7393.26.4.1005.
A reconstructive account of memory is presented to explain the finding that report of a word (C2) appearing in a rapidly presented list is reduced when it is orthographically similar to an earlier word (C1) in the list. By this account, the effect arises when the list is reconstructed from memory, not at the time of list presentation as proposed by accounts based on failure of encoding or tokenization. The reconstructive account is supported by a series of experiments that show a retroactive effect in which report of C1 is enhanced by similarity to C2; a nonword C1 can either interfere with or enhance report of C2, depending on how accurately C1 is encoded; manipulation of reconstructive processes can eliminate or enhance the effect of orthographic similarity; and a bidirectional trade-off in the report of an orthographically similar C1-C2 pair, whereby report of one member compromises report of the other.