Is "memory-scanning" time in the Sternberg paradigm reflected in the latency of event-related potentials?

The time taken to scan short term memory for a target (probe) digit in the Sternberg paradigm is thought to be reflected in the latency of a major positive wave in the associated event-related potentials. In the present study we have recorded and analysed reaction time and event-related potentials to a digit probe identification task in 37 healthy subjects. Using methods similar to those of earlier studies, we have confirmed the previously reported relationship between memory set size and the apparent latency of the major positive wave. However, analysis of the responses of individual subjects showed that increasing set size had no consistent effects on this wave. One-third of the subjects showed no latency change with increasing set size. In the other subjects, possible latency changes were invariably associated with wave form changes, suggesting that impression of latency shifts may arise from a comparison of non-analogous waves. We suggest that the most significant effect of increasing set size, in the majority of subjects, is a negative amplitude shift which overlaps and distorts a variable section of the major positive wave. In these subjects, an apparent shift in the latency of the major positive wave could be attributed to a combination of attenuation of earlier contributions and relative preservation of later subpeaks, with the result that the dominant positive waves at different levels of memory load are not analogous. By contrast, reaction time increased with set size in all subjects, irrespective of the presence or absence of associated wave form changes. Whereas the reaction time changes with increasing memory load in our study support the original concept of memory scanning, we found no consistent relationship between the latency of event-related potentials generated by this digit probe identification task and memory load. While the presence or absence of a latency shift in some subjects may be open to interpretation, our findings do not support the hypothesis that the latency of the major positive waves is an index of the time involved in memory scanning.