I received an inquiry about parental traits and subsequent transference effects.
I still suspect imprinting on our parents as a foundation for later mate selection (Oedipus be cursed in several senses!); it may be that we also do better with therapists who share key traits with our parents.
The other trigger for "transference" might be my old obsession of Linkage Disequilibrium. That is, a physical trait is linked also with a preference for that trait. The Widowbird example is that males have long tails. If the male has a longer tail than usual, you know both his father had a long tail and that is mother likely had a preference for men with long tails. Daughters carry the preference and the gene for a longer tail, even if it is not expressed. This "linkage" accounts for sudden increases in secondary sexual characteristics. The rapid change in phenotypic frequency is the reason for the word, "disequilibrium."
The implications are kind of fun. Even if adopted, LD suggests that we will opt for mates (and therapists?) who have traits similar to our biological parents. Imprinting suggests that adopted kids will prefer mates similar to whomever reared them.
Given that some couples have a domineering type mated to an enabler, it would be interesting to see if children of enabler mothers and domineering fathers are more apt to choose an enabler mate when older. (Assuming that more time is spent with mom!) Likewise, for therapist choice or for other types of bonding between people.
Clearly, we meet people with whom there is an instant rapport. Some of the time, these people repel us later, with the same features that first attracted us. A thrifty person becomes a tightwad; a good sense of humor evolves into clowning. (Beck wrote about this phenomenon well.)
I would love to work on the imprinting vs linkage explanations. One teenaged male commented that "Poison Ivy" was far more attractive than "Batgirl" but couldn't account for it. His mother's dark red hair certainly had no influence on him! It would be fun to use computer composits or photos of movie stars and compare "rated attractiveness" with the features of natural or adopted parents. It wouldn't allow for temperamental variables (which I suspect are the real power factors) but would give data about physical characteristics and preferences more quickly than waiting for people to grow up and marry. Another choice might be that of filming the same actresses but with different scripts so that temperamental contrasts are evident. GSR and Likert ratings might show some interesting patterns.