The purpose of this study was to gather evidence relative to the proposition that matching (recognition) performance for A-B pairs following an unlearning paradigm cannot be used to infer associative loss. The alternative was to assume that matching performance is based on frequency information which is independent of associative information. The A-D and A-Br paradigms were used, and the expected difference in matching performance was found. This difference was correlated with a difference in the precision of frequency information for A-B pairs as would be expected if frequency mediated matching performance. (Author)