To the Editor: We appreciate the letter by Dr. Bell and colleagues regarding our recent report of verbal working memory impairment in schizophrenia patients and their first-degree relatives. It is noteworthy that cognitive deficits in schizophrenia may be open to rehabilitation. When we reported that verbal working memory impairment may be an endophenotypic marker for the genetic diathesis for schizophrenia, we did not intend to imply that this deficit is "fixed and untreatable."

Endophenotypes are stable, heritable, endogenous characteristics that identify genetic risk for a given disorder (1). Although stability is a key component of this definition, it does not signify that such characteristics cannot be changed through targeted intervention. Just as individuals with stable patterns of behavior or attributional styles may benefit from psychotherapy, individuals with otherwise stable cognitive abilities may benefit from "neurocognitive retraining."

We agree with Dr. Bell et al. that genetic vulnerability should not be equated with irreversibility. We are all familiar with phenylketonuria, a disorder transmitted by a recessive gene, in which the genetic vulnerability for mental retardation can be prevented through environmental intervention (i.e., diet).

We were pleased to see that Dr. Bell et al. used different batteries for training and testing working memory in their participants, as their findings suggest that improvement in performance is not task specific. It would be interesting to see if the improvement in working memory generalized to less similar tasks, such as spatial working memory and self-ordered tasks, on which schizophrenia patients have also demonstrated impairment (2). Just as one might not expect training on tasks such as the Tower of London to carry over to all executive function tasks, working memory improvements may also be limited in scope.

In our report, we noted that cognitive deficits in schizophrenia are open to some criticism because factors associated with chronic mental illness (e.g., active psychotic symptoms or medication effects) could potentially color performance and limit conclusions. It may be that the significant (25%) improvement in the control condition of Dr. Bell et al. supports this assertion. For this reason, it would be useful to look at similar training in unaffected relatives of schizophrenia patients for whom the aforementioned criticisms do not apply. Differential rates of improvement in relatives and normal control subjects might be suggestive of ongoing working memory deficits.

In summary, the findings of Dr. Bell and colleagues are not at odds with ours nor do they detract from their significance. Although it is useful to consider the potential for rehabilitation of cognitive deficits in schizophrenia, with reference to endophenotypes, it would be more useful to consider rehabilitation of working memory deficits in their unaffected but at-risk relatives and, ultimately, to determine if such intervention reduced the likelihood of their developing schizophrenia.