Another indispensable tool of the scientist, but one that is rarely mentioned, is the principle of parsimony, also known as Occam's Razor. It is the principle of accepting, all else being equal, the explanation that is simplest. This approach hardly needs a theoretical justification; we only have to think up a few everyday scenarios to see that it makes sense, and that everybody unconsciously uses it all the time.
Imagine, for example, that you cannot find your wallet, and after some searching around you finally locate it on the path directly in front of your house/flat. We may now formulate two hypotheses: either (a) you dropped it when you fished for the key to unlock the front door, or (b) you dropped it when you fished for the key to unlock the front door, it was picked up by a dog, carried down the road, snatched away by a seagull, dropped over a nearby park, picked up by a pedestrian who figured out it was yours, and then carried back to your door and deposited in front of it.
The thing is, with the tools at your disposal the two hypotheses are indistinguishable as far as the expected observations are concerned; neither can be rejected based on the data. Still, you would probably laugh hypothesis (b) out of the room if it was presented to you, precisely because it is indistinguishable from (a) except for the addition of implausible extra steps. That is Occam's Razor. What the scientist does in cases like these is, of course, precisely the same. The simpler explanation is to be preferred, superfluous and especially unevidenced entities and events are to be rejected.
There will probably be some self-satisfied philosophers who will ramble about how there must be a deductive, philosophical justification of parsimony, but I am satisfied to accept it as pragmatically as induction: it works, that is all the justification I need. And this pragmatism is acceptable because these tools are not held to the same standard as deduction, as something that must logically always lead to correct conclusions. Instead, they are heuristics, and as such it is sufficient that they work most of the time. Problem solved.
When discussing astrophysics and cosmology, the relevant physicists often have to point out that parsimony really is not quite as straightforward as counting things and taking the option with the smallest number of things. The classic example is the multiverse, which is sometimes rejected as unparsimonious compared with only one universe because, well, it assumes the existence of more universes, and of universes that we cannot see at that. However, what has to be compared is not the number of universes but instead the complexity of the rules that are assumed to govern the universe and its development. And if the simplest assumptions about those rules that fit the data are such that they would lead to the creation of multiple universes, then that is in fact the most parsimonious solution.
In my own area the application of parsimony is luckily much more straightforward than in physics. It basically boils down to accepting the smallest number of events necessary to explain the evolution or biogeographic history of the study group. To use a pretty silly example, we would assume that a fossil hominid from 5 million years ago is genealogically connected to us through as few morphological intermediates as possible, but we would not assume that it first evolved into a cauliflower and then into us. The latter is less parsimonious (quite obviously so), and the logic for its rejection is pretty much the same as in the everyday example from before.
The uses we have for parsimony in my area are nonetheless varied. They include:
- Parsimony analysis of morphological or molecular data to infer phylogenetic relationships between biological species or gene copies.
- Parsimony approaches to species tree analysis of gene trees to infer phylogenetic relationships between biological species.
- Matrix Representation Parsimony supertrees (Ragan 1992).
- Reconstruction of ancestral character states along a phylogenetic tree.
- Reconstruction of ancestral areas of distribution in biogeography.
- Parsimony Analysis of Endemicity (Rosen & Smith 1988) and its variants.
In the next few weeks I will discuss these various applications of parsimony one by one.