Predicting serial position effects and judgment errors in retrospective evaluations from memory recall

Abstract

When forming global impressions in retrospect, the first, the last, and the most outstanding experience often have a lasting impact on the final evaluation of an event, as most prominently captured in the peak-end rule. Such serial position effects in impression formation provide indirect evidence that individuals reconstruct their evaluations by retrieving previous experiences from memory, instead of updating their impression online. Yet, latest work sheds doubt on the ability to predict global evaluations from memory retrieval on the individual level. In three experiments, we aim to quantify how much variability in retrospective evaluations can be attributed to memory retrieval by relating serial position effects in retrospective averaging judgments to serial recall curves from memory. The experiments revealed serial position effects in memory recall and corresponding, but less consistent effects in averaging judgments, demonstrating that individuals better recalled and more heavily weighted the first and last item. For long sequences, memory recall permitted to predict individuals’ averaging error to a moderate to strong degree, even if individuals were unaware of number recall as a potential averaging strategy (Experiment 2). Yet, shorter sequences fail to evidence the same relationship, possibly because individuals attempt to apply more optimal averaging strategies (Experiment 3). We discuss retrieval patterns as markers for distinct evaluation strategies.