EVALUATING THE FRAGMENTATION HYPOTHESIS - THE ANALYSIS OF ERRORS IN CUED-RECALL

This paper investigates whether cued recall of multidimensional stimul i is all-or-none, as predicted by the Fragmentation Hypothesis (Jones, 1976); or probabilistic, as is commonly assumed in models of associat ive memory. To test this, composite stimuli were cued repeatedly, by e ach of their attributes in turn, to see whether the patterns of recall were consistent with all-or-none fragments. This test also requires a model to account for the inconsistent patterns of recall which are to be expected as a result of correct guessing. Of necessity, therefore, this paper also investigates the nature of guessing to enable the tes t of all-or-none recall. Two experiments are reported, both conditions of the same design and using the same pictorial stimuli. The first an alysis looks at errors and seeks to discover interactions within seque nces of responses. There are two stages involved. First, the most like ly sources of each answer are indentified and classified according to a number of different categories of interest. Second, there is a stati stical evaluation of the frequency with which these different categori es occur. This analysis reveals: (a) the systematic recall of previous errors; and (b) that guesses comprising a pair of elements from the s ame (incorrect) stimulus occur more frequently than is expected by cha nce. Both processes have a systematic effect upon the pattern of corre ct guesses which is not predicted by the models of guessing commonly u sed. A model of cued recall is presented which combines the Fragmentat ion Hypothesis (including the assumption of all-or-none recall) with a model of guessing which hypothesises that a proportion of guesses act also as implicit cues for recall. This ''memory checking'' model of g uessing is shown to predict the observed processes in guessing well. O verall, the combined model shows a satisfactory fit to the data, provi ding support for the all-or-none assertion. However, it is shown that a small proportion of inconsistent patterns of recall cannot be explai ned by the guessing model, and a low level of recall failure and forge tting is proposed. The recall of fragments is therefore closely approx imated by all-or-none recall, but cannot be exactly so.