Parsing the recognition memory components of the WMS-III face memory subtest: normative data and clinical findings in dementia groups.

The WMS-III face memory subtest was developed as a quick, reliable, measure of non-verbal recognition memory. While the face memory subtest has demonstrated clinical sensitivity, the test has been criticized for low correlation with other WMS-III visual memory subtests and for failing to differentiate performance between clinical groups. One possible reason for these findings may be due to the impact of response bias associated with recognition memory tests. Four studies were conducted to evaluate the utility of applying signal detection measures to the face memory subtests. The first two studies used the WMS-III standardization data set to determine age and education effects and to present normative and reliability data for hits, false positives, discriminability and response bias. The third study tested the hypothesis that using response components and signal detection measures would enhance the correlation between face memory and the other WMS-III visual memory subtests. The fourth study compared performance of patients with Alzheimer's disease, Huntington's disease, Korsakoff's syndrome and demographically matched controls on the new face memory scores. The new measures did not have higher correlation with other WMS-III visual memory measures than the standard scoring of the test. Analysis of the clinical samples indicated that the discriminability index best differentiated patients from controls. The response components, particularly delayed false positives, differentiated performance among the clinical groups. Normative and reliability data are presented.