No abstract

In the literature, a memory advantage for bizarre items over common ones has been found only in a few studies, especially with materials prepared ad hoc by the experimenter and with free recall rather than cued recall tests. These results contrast with the widespread conviction that bizarreness helps recall. The present paper explores the role of some variables involved in the “bizarreness” effect: (1) It examines the typical self-generation procedure in which the subject is asked to create an interaction between a pair of nouns, as well as the case in which only one noun is given. Higher freedom in generating sentences appears to correspond to higher free recall of bizarre items. (2) It is shown that bizarre items must be distinguished from “unusual” ones, which have different effects on memory. (3) By contrasting groups instructed to use either imagery or verbal elaboration, it is shown that the bizarreness effect is linked to the use of imagery. Instructions to use imagery without the possibility of creating bizarre representations do not improve the recall of common items. (4) The classification of parts of sentences generated reveals that, under common instructions, one subject's choice of verb and noun is more likely to be shared by other subjects. This fact may explain the different effects found by previous research in cued and free recall. (5) The overestimation of the recallability of bizarre items appears less evident than in previous research, probably because subjects had direct experience of the difficulties met in generating bizarre images.

Existing literature on bizarreness effects in verbal learning mainly focuses on the common assumption that bizarre images are easier to recall than common ones. In so doing, however, researchers have obtained more negative than positive results. Further, among the few investigations that found this effect three used the same procedure and the same material in which a bizarre relationship between subject and object was achieved by substituting human beings for animals and vice versa. It was observed that a clear bizarreness effect may still be observed if inanimate sentences are constructed in accordance with the same principle of distorting typical relations (Expt 1). This effect remains, albeit in a milder form, when aspects of the original procedure are manipulated, such as eliminating complementary sentences (Expt 2) or abstract sentences (Expt 3) from the lists. This seems to demonstrate that the traditional opinion that bizarre images make the mnemonic task easier is, at least in part, true. Nevertheless, if one uses the procedure used by Merry & Graham (1978) and others with semantically unpredictable sentences, the bizarreness effect was attenuated and even normal sentences sometimes achieved higher scores (Expt 4). In general, these results reveal that bizarre sentences are more likely to result in better recall as regards number of nouns remembered, whereas the opposite is true as regards number of complete sentences remembered. This is consistent with the view that recall of normal sentences is facilitated by the unitization. These four experiments carried out with adult subjects demonstrate the generality of the bizarreness effect and some of the factors which contribute to it