Massimo Pigliucci in Aeon:
[E.O] Wilson claims that we can engage in a process of ‘consilience’ that leads to an intellectually and aesthetically satisfactory unity of knowledge. Here is how he defines two versions of consilience: ‘To dissect a phenomenon into its elements … is consilience by reduction. To reconstitute it, and especially to predict with knowledge gained by reduction how nature assembled it in the first place, is consilience by synthesis’.
Despite the unfamiliar name, this is actually a standard approach in the natural sciences, and it goes back to Descartes. In order to understand a complex problem, we break it down into smaller chunks, get a grasp on those, and then put the whole thing back together. The strategy is called reductionism and it has been highly successful in fundamental physics, though its success has been more limited in biology and other natural sciences. The overall image that Wilson seems to have in mind is of a downward spiral wherein complex aspects of human culture — literature, for example — are understood first in terms of the social sciences (sociology, psychology), and then more mechanistically by the biological sciences (neurobiology, evolutionary biology), before finally being reduced to physics. After all, everything is made of quarks (or strings), isn’t it?
Before we can see where Wilson and his followers go wrong, we need to make a distinction between two meanings of reductionism. There is ontological reduction, which has to do with what exists, and epistemic reduction, which has to do with what we know. The first one is the idea that the bottom level of reality (say, quarks, or strings) is causally sufficient to account for everything else (atoms, cells, you and me, planets, galaxies and so forth). Epistemic reductionism, on the other hand, claims that knowledge of the bottom level is sufficient to reconstruct knowledge of everything else. It holds that we will eventually be able to derive a quantum mechanical theory of planetary motions and of the genius of Shakespeare.