Drinking-related differences in the memory organization of alcohol expectancies.

Increased understanding of how risk-related variables influence later alcohol use and alcoholism requires a shift from identifying correlated variables to developing models of mediational processes. Memory operation, with alcohol expectancies as critical content, is one such model. To explicate this model, similarity judgment, which is well established for discerning memory organization, was used. Euclidean distance-based algorithms, including multidimensional scaling and hierarchical clustering, were then used to empirically specify memory structures and processes as a function of individual differences in drinking. The resulting semantic networks support previous findings and show for the first time that heavy drinkers' networks may be more tightly configured than those of light drinkers. These findings bring researchers closer to a computational model of the psychopharmacological process that governs drinking.