Reconstructive remembering of the scientific literature.

In this paper we investigate the role of reconstructive memory in citation errors that occur in the scientific literature. We focus on the case of de Groot's (1946) studies of the memory for chess positions by chess experts. Previous work has shown that this research is very often cited incorrectly. In Experiment 1 we show that free recall of this work by research psychologists replicates most of the errors found in the published literature. Experiment 2 shows that undergraduates reading a correct account of the de Groot study also make the same set of errors in recall. We interpret these findings as showing that consistent errors in secondary accounts of experimental findings are frequently reconstructive memory errors due to source confusion and schema-based processes. Analysis of a number of other examples of scientific literature that have been frequently cited incorrectly add additional support to the reconstructive account. We conclude that scientists should be aware of the tendency of reconstructive memory errors to cause violations of the scientific norm of accurate reporting of the scientific literature.