AND

In making a new judgment, a person can access relevant past judgments and/ or process the stimulus information underlying these judgments. The present study is concerned with (i) the conditions under which past judgments are used instead of stimulus information and (ii) the effects that repeated past judgments have on the representation of the stimulus information in memory. Three ex-periments were conducted using an impression formation task. Subjects were presented with trait descriptions of hypothetical individuals and were asked to make one, three, or five impression judgments on the basis of each description. It was hypothesized that a new impression will tend to be based on an old one, rather than on trait information, as a function of the ease with which the new impression can be inferred from the old. Ease of inference, in turn, will depend on the similarity of the two impressions and the availability of the past impression. The results of the three studies are consistent with this hypothesis: When past judgments were available, (Expermints 1 and 3), the extent to which they, rather than the trait information, determined the present judgment increased with th