Predicting immediate ordered recall of lists with products of probabilities

Immediate ordered recall performance is often reported in terms of recall of individual items or of lists. Schweickert, Chen and Poirier (Int. J. Psychol. 34 (1999) 447) proposed that the probability of recalling a list approximately equals the product of the conditional probabilities that each item is recalled, given its immediate predecessor is recalled. This Product of Adjacent Conditionals gave a close lower bound to the probability of recalling a list, although it is not exactly right. An alternative is that dependencies are expressed through probabilities conditioned on retrieval of higher level units such as chunks. An example is a key assumption of Anderson and Matessa (Psychol. Rev. 104 (1997) 728). This alternative is rejected here. Overall, the Product of Adjacent Conditionals formula and the Anderson and Matessa theory predicted recall of a list about equally well.