To the Editor: Drs. Malhotra and Goldman point out some caveats that should be considered in the interpretation of our data. We appreciate these constructive comments and would like to make some additions to the discussion.

It has been shown that the number of repeats in the DRD4 gene influences clozapine and spiperone binding (1). Therefore, it is was natural to collapse the 8-repeat allele with the allele closest to it in size (i.e., the 7-repeat allele), since these alleles can be assumed to have the most similar binding properties. Drs. Malhotra and Goldman call this an "arbitrary grouping," but we think that the grouping is actually based on prior information about the properties of different alleles of the DRD4 gene.

Drs. Malhotra and Goldman also point out that we conducted a "presence or absence" analysis, despite the lack of data implicating a codominant mode of action for these alleles. As discussed in our article, the strength of our sample was that it was drawn from the population-wide, nonselected epidemiological group of the genetically isolated Finnish population. Everyone in the original sample was scored for novelty-seeking parameters (2), which enabled us to select extreme scorers for monitoring differences in allele frequencies between high and low scorers, regardless of the mode of action of the alleles. We also performed an additional analysis of the presence or absence of the different alleles. To group the individuals according to all possible genotypes would have required a significantly larger sample.

Given the original findings (Ebstein et al., 1996, and Benjamin et al., 1996), our observed association between novelty seeking and the shorter alleles of DRD4 does replicate the finding of an association, but not to the same allele as in the original reports. Therefore, we proposed that some other polymorphism in linkage disequilibrium with DRD4 might be responsible for the variation in novelty-seeking scores. We agree with Drs. Malhotra and Goldman that it would be interesting and important to study any such polymorphism.

Drs. Malhotra and Goldman interpret our finding as possible support for their finding of an association between the 7-repeat allele and low novelty-seeking scores in Finnish alcoholics. This is on the basis of the assumption that we included a higher proportion of substance abusers in our study group because of our sampling strategy. As we discussed in our article, it is certainly possible that we included in the sample a high proportion of subjects with any given behavioral trait to which extreme novelty seeking might contribute either as a constituent or a risk factor (e.g., some personality disorders and substance abuse). Therefore, the studied polymorphism might be primarily associated with any such trait and only secondarily with novelty seeking. However, since the study sample was collected from the normal population, we are reluctant to draw any conclusion regarding the results concerning alcoholism or any other specific trait. Only further studies of both normal populations and individuals with different personality disorders can reveal the influence of different DRD4 alleles on behavior.