Measuring infant memory

Existing memory measures L/I, L/B, L/(L + I) and U(L + B) are defined in terms of baseline response rate, B, response rate immediately after acquisition, I, and response rate at a long-term retention test, L. These measures an compared to a new measure S = (L - B)/(I - B). S, the fraction of acquired kicking rate remaining at the end of the retention interval, is preferred to the existing measures. S graphically portrays the forgetting function. Point estimates and confidence intervals can be used to describe the proportion retained and to test hypotheses about the memory level. This is an improvement over the ordinal tests performed with the existing measures. ANOVAs can be performed on S values to compare retention over days and between treatment groups. S estimates provide the basic data for fitting candidate forgetting laws to the obtained data. S appears to be a virtually unbiased estimator of the true proportion retained and has sensitivity comparable to that of the best of the existing measures. (C) 1996 Academic Press, Inc.