Genomic imprinting refers to a surprising discovery in the 1980s that in mammals (and flowering plants, such as corn) some genes in an individual have their pattern of expression affected by the sex of the parent that contributed the gene. Typically, there are paternally active genes, in which the copy from the father is active while the copy from the mother is inactive, and conversely, there are maternally active genes, in which the paternal copy is silenced. This difference does not reside in the DNA but is some kind of extragenetic piece of information (involving, e.g., how methylated the genes are). I was aware of this new discovery by the late 1980s and felt intuitively that it must have something to do with intragenomic conflict, a topic I was then beginning to work on. Intragenomic conflict refers to situations in which the genes within an individual are being selected in different directions (due, e.g., to differing degrees of relatedness to others, e.g., Y chromosomes vs. mtDNA vs. autosomes). But I could not see a solution. Then in late 1991 Jon Seger sent me a two-page paper from Cell written by David Haig and Chris Graham that solved the problem in one stroke and provided a very striking fact in its favor. Haig and Graham (1991) pointed out that genomic imprinting had immediate and strong effects on degrees of relatedness.
An unimprinted gene, that is, a gene whose activity was identical no matter which parent donated it, was a gene without information on parental origin. Thus, when calculating degree of relatedness of offspring to parent for any given gene, one begins by saying there is a1/2 chance it came from the mother and a1/2 chance it came from the father. But an imprinted gene had exact information on where it came from: a paternally active gene in an individual came from the father; its silent homolog must have come from the mother. In modeling the effects of the active allele, we see that if it acts