Traitwell's Future of Genomics Series: Honest Signalling
We're continuing our discussion about the future of genomics.
We’re continuing our analysis of the future of genomics. For those who are curious, please visit Traitwell.com. Be sure to check out our free apps.
The potential offered by genetic information to assist choice of friends, mates and even employees has been dramatically underexploited so far, offered more as a half-baked curiosity (recall ‘computer dating’) than as a practical proposition with a strong basis.
Attractiveness itself has a genetic basis, and friend/mate pairs do not display a random collection of traits. Notice that this includes similarity as well as dissimilarity which may be desirable for some traits, as a complement or as a division of labour. Both kinds of aggregation can be applied here: over individuals (services which act as filters to accelerate searching for desirable matches), and over traits (basing a choice on many traits will raise the chances that a better choice was made than, say, random guessing).
The search for suitable friends, mates and employees is an special case of the ‘honest signalling’ problem, important to the study of human and animal behaviour. Those to be matched both have a vested interest in using so-called honest signals to do this. These are signals that are very hard to fake. One way to do this is to impose a steep cost for showing the signal. Sometimes this is a monetary cost, displayed through the possession of many years of education in not particularly useful subjects, or in conspicuous consumption, deliberately wasting resources—many argue that the former is an example of the latter. This works because the signal is associated with abilities that are valuable to the receiver of the signal: the ability to provide for the future, the underlying abilities needed to even get the required money in the first place (including higher intelligence and stable personality traits, better than average health and numerous other correlates). Money is hard to come by. It is nearly impossible to get away with pretending to have it when you have to actually spend it, the more wastefully the better. Here IOUs are not accepted.
There are tremendous inefficiencies in existing honest signals. There is the waste inherent in the conspicuous consumption itself. There is the waste of the time required to acquire wealth (or the years required to get advanced degrees from an expensive premium institution) through at least early adulthood. But there is also the loss of good candidates because of the imperfect association between actual accumulated wealth and ability. A cheaper alternative in time and resources that remains very hard to fake improves efficiency. Genes can be used as honest impossible-to-fake signals. Genome sequencing reveals them. Aggregation makes them reliable.
First suppose that someone has, say, a mate, but is uncertain about how to proceed with the relationship, less or more seriously. Suppose also that reproduction is a goal (there are others). A service—let’s call it CheckMate—which analyzes the sequenced genomes of both parties can provide valuable, some might say priceless, advice about straightforward conditions caused by genes, which may be recessive and very hard to detect otherwise. Assume also that both parties want to know the truth, to remove concerns about privacy. Genetic counselling is of course already available for doing this internationally, but it is typically triggered by known family histories, where knowledge is uncertain and fragmentary, and requires expensive and time-consuming professional engagement, the results of which vary in accuracy. (We will consider screening of already-conceived embryos later.)
Democratizing access to this process, extending to a much broader audience, removing as much friction and latency as possible, has great potential for improving outcomes. Here the choice about what conditions to consider, and how to value them, positively or negatively, can be left entirely up to users of the service, empowering them. Formal genetic counselling may be recommended as a follow-up choice.
More complex polygenic traits are in scope here, including behavioral ones, once again leveraging aggregation over many such traits to raise reliability. One may elect to screen for, say, schizophrenia risk, psychopathy risk or other mental pathologies. For Alzheimer’s risk, or for intelligence potential. One may also elect to check for potential inbreeding using runs of homozygosity scores as discussed above, or other general signs of genetic quality (e.g. relative frequency of minor alleles). This is particularly relevant among populations with high rates of clannishness and extensive cousin marriage. Voluntary use of such services will be incentivized by governments or cultural bodies, to subsidize or eliminate the cost, because of the society-wide benefits, cumulative over time.
There is a distinction here between negative selection (things one does not want, say sickle-cell anaemia) and positive selection (things one does want, say intelligence). But this direction can be left up to the user: who can anticipate what people want? They may search for dissimilarity benefits (known as heterosis) by outbreeding very widely, or they may think too-distant outbreeding undesirable for whatever reason.
Another way this may be generalized is to consider pair-bonding traits in the mates, without regard to reproduction, as predictors of (say) individual health, marriage stability, or other outcomes. The reasoning is analogous.
Suppose next that someone does not have a candidate mate, but seeks one, with the prospect of reproduction or lifelong companionship, and wishes to optimize the chances of success, again according to criteria that they are free to define and value themselves. Ensure that they are able to obtain a filtered list of candidate mates from a large pool who have offered themselves for consideration, taking advantage of the inexpensive honest genetic signals they can send. Now this is a filtering process, aggregating over many individuals and over many traits at the same time, thus doubly reliable. Having obtained a shortlist, the user may elect to proceed further using their own discretion, or other tests. The value of the filtering step will be apparent to the composite process. This need not be limited to sexual friendship either, but also generalizes to ordinary friendship (conceptualize a service that people voluntarily sign up in search of compatible friends and upload their own DNA for matching).
Matching may also be cross-species. Pet owners may be matched to pets with desirable behavioural traits. It will be possible to go much further than breed selection in the near future.
Within each breed there is considerable variation. Suppose one wants to select a pet with the low level of aggression, to reduce the danger to small children, but still select from a mainstream breed like a German Shepherd. Polygenic scores for aggression will accomplish this sort of filtering. Reducing the likelihood of hip dysplasia at the same time, and any number of similar disorders, aggregates over traits. It is highly likely that homologies (or just as well, contrasts) will be found between the genetic traits of pets and those of their owners, facilitating matching. (Many of the genes shared by different species are homologous, because natural selection recycles parts relentlessly.) Aggressive humans commonly like aggressive pets for instance. Genes which make humans aggressive will probably be implicated in pets too.
We highly recommend the book, How to Tame a Fox (and Build a Dog): Visionary Scientists and a Siberian Tale of Jump-Started Evolution, for those so inclined.