340
MEASURING
INFANT MEMORY
Richard S. Bogart2 Department of Psychology University of Massachusetts, Amherst Amherst, Mass. 01003-7710 Researchers studying infant memory have measured the base rate of a kicking response, conditioned the response by making stimulus motion contingent upon kicking, and then studied retention as a function of various variables by testing at later times. The test periods are the Baseline, Immediate Retention Period, and Long Term Retention Test (LTRT), and the kicking rates B, I, and L are obtained, one from each period. Two measures derived from B, I, and L, the retention ratio L/I and the baseline ratio L/B are commonly used. Two other measures also have been used. These measures purportedly reflect the amount of memory strength still existing at LTRT. Various weaknesses of these measures are considered, including that they yield only statements of the form “some forgetting has occurred,” ”no forgetting has occurred, ” “there is some retention,” or “there is no retention.” Furthermore, the retention ratio L/I, purportedly revealing “the extent to which infants remember” (Rovee-Collier, 1990, p. 523), takes the value B/I when L = B. According to Rovee-Collier and Shyi (1992, p. 7 - 8), if L = B, forgetting is complete. Complete forgetting corresponds to zero memory strength. But the retention ratio L/I takes the value B/I when L = B. Thus, even when forgetting is complete, the retention ratio takes a nonzero value. This nonzero value is the proportion of I that B happens to be and is uninterpretable. This prompts the use of the baseline ratio L/B which, if greater than 1.0, indicates some memory strength remains. Neither L/I nor L/B directly addresses the question of how much memory strength remains. The proposed measure, S = (L - B)/(I - B), permits the same tests that are performed with L/I and L/B, embodies the following 4 assumptions of Rovee-Collier and Shyi (1992) and violates none of them: If the infant responds at the same high rate during the long-term test as during the immediate test, then retention is “perfect,” and the retention ratio is 1.00 (100%); the less the infant kicks during the long-term retention test relative to the immediate retention test, the lower the retention ratio, and the greater the extent of forgetting; a mean retention ratio significantly less than 1.00 confirms that a significant amount of forgetting has taken place; if infants respond at the same rate as they did prior to training (i.e., a baseline ratio of l.OO), however, they are treating the mobile as novel, and we infer that forgetting is “complete.” The proposed measure is a monotone function of memory strength, takes the value 0 at 0 memory strength and takes the value 1 at maximal memory strength. It is interpretable as the proportion of the original memory strength still remaining at LTRT. Various desirable properties of S concerning bias and precision are described including the possibility of making statements more precise than the ordinal inquality statements made with the extant measures. References Rovee-Collier, C. (1990). The “memory system” of prelinguistic infants. In A. Diamond (Ed.) Annals of the New York Academy of Sciences: vol. 608. The development and neural bases of higher cognitive functions. pp. 5 17-542. Rovee-Collier, C. and Shyi, G. C.-W. (1992). A functional and cognitive analysis of infant long-term retention. In Howe, M. L., Brainerd, C. J., and Reyna, V. F. (Eds.), Development of long-term retention. New York: Springer-Verlag. pp. 3-55.