Experimental manipulation of working memory model parameters: an exercise in construct validity.

As parametric cognitive models become more commonly used to measure individual differences, the construct validity of the interpretation of individual model parameters needs to be well established. The validity of the interpretation of 2 parameters of a formal model of the Continuous Recognition Memory Test (CRMT) was investigated in 2 experiments. The 1st study found that manipulating the percentage of trials on the CRMT for which degraded pseudowords were presented altered the model's stimulus encoding parameter but not the working memory displacement parameter. The 2nd experiment showed that manipulating the number of syllables forming a pseudoword altered the model's working memory displacement parameter for each syllable added to the pseudoword. Findings from both experiments supported the construct representation of the model parameters, supporting the construct validity of the model's use to interpret CRMT performance. Combining parametric models with the manipulation of factors that theory predicts are related to model parameters provides an approach to construct validation that bridges experimental and individual difference methods of studying human cognition.