My personal take on it is this: The PRDM9 gene is basically sensitive to the genetic factors that most affect the stress levels of individual organisms. It is reasonable to conclude that any individual that is relatively more successful than its peers will be relatively less stressed than its peers.
Evolution has harnessed this as a resource; stress induces mutations. What this means is that successful individuals do not suffer much mutation and their geno/phenotype remains stable. Those who are stressed undergo mutation; in this way the correct evolutionary choices for the environment represent a signal, and the poor choices are basically cancelled out with noise, but with the proviso that should any new mutation yield a relative advantage, that mutation will immediately be discovered by the genome because of sudden drop in stress.
From this perspective, it is easy to see the genome as a problem solving algorithm that is distributed in the species and which literally runs its calculation cycles once per generation.
And from a coder's perspective, the genome is recorded in something like 38MB of data. More than enough space to put such a process into the code, as it were.
It's not survival of the fittest individual, but survival of the fittest genome. The way our brains and neurons are arranged is not the first iteration of information processing, but the second. The first one lies in our DNA, and ultimately operates by many of the same rules. Knowing this, we can use anthropic reasoning to make many educated guesses as to what we should expect to find.
Actually, seems I'm not the only one thinking along these lines. That is encouraging.
http://www.evolutionnews.org/2016/09/gloves_off_--_r103168.html
http://phys.org/news/2009-08-evolutionarily-mechanism-genes.html