In the three simple two-allele systems described above, the biochemical expression of the alleles in the heterozygote is exactly intermediate between the two homozygotes.  However, in each of the cases the resultant pattern of phenotypic expression is different. The variant alleles are in consequence described as dominant or recessive, depending on the phenotype of the heterozygote relative to that of the two homozygotes.  The level of enzyme activity relative to the ++ "wildtype", or the nature of the phenotype (in these examples, 'darker' or 'lighter'), or any perceptions about "normal" versus "abnormal" are absolutely irrelevant.

    Biochemical expression in diploids is most simply modeled as the additive result of two alleles, each of which produces 50% of the total enzyme activity. The standard phenotype of a homozygous genotype is thus the result of the expression of two fully functional alleles that produce 50% + 50% = 100% of the standard activity. Then, a heterozygous genotype with a functional allele and a non-functional allele produces 50% + 0% = 50% of the standard activity. The phenotype of the heterozygote will then depend on the degree of haplosufficiency of the single functional allele. That is, does a single functional allele provides sufficient enzyme activity such that a standard phenotype is obtained? As it happens, for most gene loci this is the case: the functional allele can then be described as "dominant" to the non-functional allele. Though the non-standard allele contributes reduced or no activity, the standard allele still contributes its standard amount, such that net activity in the heterozygote is unchanged. Thus is most cases, the functional allele can be described as "dominant" to the non-functional allele. In those cases where a single functional allele does not provide sufficient enzyme activity to produce the phenotype when it is homozygous, that allele is described as haploinsufficient. The standard alleles can then be described as 'recessive" to the non-functional allele, which is therefore dominant.

    Other circumstances influence the phenotypic expression of heterozygous genotypes. "Up-regulation" of gene expression in a heterozygote may compensate for the non-functional allele by increasing transcription of the alternative allele, such that the amount of enzyme produced approaches that of the homozygous genotype. Contrariwise, the presence of a defective protein may interfere with the activity of the standard protein, e.g., by competitive binding of substrate such that the standard enzyme cannot convert it to product [this happens in Tribbles]. In other cases, presence of a defective protein itself produces a dominant phenotypic effect, notably accumulation of protein plaques in the tri-nucleotide repeat diseases. Finally, recall that mutations in the promoter region may either increase or decrease gene transcription, without producing detectable allele variation in the protein-coding region.

    On this understanding, "dominant" and "recessive" are simply short-hand terms to describe the interactions of alleles underlying production of a molecular or other phenotype. The fundamental limitation of Classical Genetics, in the absence of an understanding of the functional operation of Genes as DNA sequences, was to treat dominance relationships (dominant, recessive, semi-dominant, co-dominant, etc.) as intrinsic properties of genes rather than as epiphenomena involving the creative use of upper- and lower-case letters.

 [Homework: What do I mean by that last phrase?]