Hindsight bias and the misinformation effect: separating blended recollections from other recollection types.

Questioning the presence of "truly" blended recollections, we investigated two cognitive phenomena: hindsight bias and the misinformation effect. At first glance, both phenomena seem to result from the same interference process, whereby the subsequent encoding of conflicting information impairs the recall of earlier encoded (original) material. Experiment 1 compared both paradigms using numerical items as material: hindsight as well as misinformation subjects revealed the same mean shift in their recollection of the original values. The additional analysis of a multinomial model, however, suggested that blended recollections occurred in the hindsight condition only. The misinformation effect, on the other hand, appeared to be based on averaging across two different recollection types. Experiment 2 further investigated how the memory-trace strength influences the likelihood for blended recollections to occur. In a misinformation procedure, one group of subjects read the original information twice, another group thrice. Again, recollections were similarly shifted towards the misinformation in both groups. But the multinomial model revealed that only the second group (with a stronger memory representation of the original information) showed blended recollections. Taken together, these results suggested that: (1) a minimum memory-trace strength of the original information must be met for blended recollections to occur; and (2) hindsight bias and the misinformation effect--though superficially similar--are induced by different cognitive processes.