Abstract
Errors in a running memory task are analysed. Participants were presented with a variable-length list of items and were asked to report the last four items. It has been proposed (CitationMorris & Jones, 1990) that this task requires two mechanisms: the temporal storage of the target set by the articulatory loop and its updating by the central executive. Two implicit assumptions in this proposal are (a) the preservation of serial order, and (b) participants' capacity to discard earlier items from the target subset as list presentation is running, and new items are appended. Order preservation within the updated target list and the inhibition of the outdated list items should imply a relatively higher rate of location errors for items from the medial positions of the target list and a lower rate of intrusion errors from the outdated and inhibited items from the pretarget positions. Contrary to these expectations, for both consonants (Experiment 1) and words (Experiment 2) we found recency effects and a relatively high rate of intrusions from the final pretarget positions, most of them from the very last. Similar effects were apparent with the embedded four-item lists for catch trials. These results are clearly at odds with the presumed updating by the central executive.
Acknowledgments
This research was conducted while the first author was receiving Grant DGES PB97-0072 and M. Rosa Elosúa was receiving Grant DGICYT PB94-1573, both from the Ministerio de Educación y Cultura of Spain. M. Teresa Lechuga was supported by a UNED predoctoral grant under the supervision of R. Marcos Ruiz. We thank Mike Page, Geoff Ward, and an anonymous reviewer for helpful comments on an earlier version of this paper. We also thank Mike Epps for his help in the English version and students who voluntarily participated in this research.
Notes
When we include all the list lengths in the analyses a second-order interaction of list length by pretarget position by response position emerges, F(12, 528) = 2.36, MSE = 0.83, p < .006. While position −2 error rates for seven-letter lists are basically at the same level in any of the response positions, for the six-letter lists there was an increase in these error rates in the first response position. Note that in this case position −2 is also the first letter of the presented list (see ).
There were four experimenters. In order to check experimenter reliability, 8 participants were selected at random (2 per experimenter), and the number of mismatches between the tape-recorded response and that written by the experimenter was found. The analysis of the results showed that (a) the experimenters committed very few errors (2 out of 100, 2 out of 100, 3 out of 100, and 3 out of 100, respectively), and (b) the errors did not coincide.
The two-syllable names are for F, H, J, L, M, N, N˜, R, S, V, X, and Z.