Sunday, March 02, 2008

Being wrong for all the right reasons.

Just a bit of a ramble with some totally unresearched ideas here. Well, not totally unresearched, but impressions and the like, but without explicit references.

Working with fossils can be as frustrating as it is rewarding. It's a lot of fun, but each fossil only tells you so little. In fact, a fossil might be said to really tell you nothing. By themselves, fossils are dry bones, unanimated. Everything we know about fossils depends on our interpretations of them. This seems a little nihilistic, not to mention ripe for creationist misquotation. I'll warn now that there is nothing of use to the creationist here. I'm not talking about the big picture about evolution. I'm talking about the finer points. The crossed t's and dotted i's. Creationists might claim that there are no transitional fossils and try to use my words as a way of showing that even I, the palaeontology graduate student, thinks there's a problem. However, I'm not only predicating this post on the reality that there are many transitional fossils, but that our human frailties might even be preventing us from recognizing more transitional fossils than we have. I believe that we can cut the story of early vertebrate evolution even more finely than most palaeontologists are willing to admit.

When studying the fossil record, we need to recognize patterns. Humans recognizing patterns is tricky business -- we're pattern-seeking animals. We see shapes in the clouds or in random scattering of pebbles in a riverbed. We are often very prone to error, as our common sense thinking can fail us when more arcane matters are in question. We have confirmation bias, where we easily remember the confirming instances for our pet hypothesis but ignore, discard, reject, or rationalise any contradictory observations.

But this is why we have science. The idealised scientist aims to eliminate bias, tries to suspend wishful thinking, and (most importantly) challenges conventional wisdom and common sense thinking. The greatest discoveries in science were not the confirmations of things we already believed, but the revelation the startling facts that were totally inaccessible to our naked perceptions--often demonstrating how wrong we actually were.

There is apparently a world that exists independently of our ability to perceive it. And so, there are necessarily truths about the world that we may not be comfortable with. The point is that science can only tell us something new if it doesn't exist to support those comforting narratives we tell ourselves about how the world works. It is an "unnatural way of thinking" as the embryologist Lewis Wolpert put it:
[T]o do science it is necessary to be rigorous and to break out of many of the modes of thought imposed by the natural thinking associated with ‘common sense’. p. xiii-xiv. Wolpert, L. 1993. The Unnatural Nature of Science. Faber & Faber, Ltd. London
What Wolpert is saying is that science is almost like an affront to a very sacred sense of understanding the world: common sense. It forces us to think in ways that sometimes feel counterproductive, uncomfortable, and even revealing conclusions that do not look like they make sense. The conclusions might be very difficult (or even impossible) to understand.

Palaeontology has, for a long time, been a discipline of narratives. Stories, of whatever sort, tied to fossils in order to explain the patterns observed. The tradition has often been one of very elegant, if not fanciful, speculation that has, in some way, been tied to peculiar observations about fossils. No one can count the speculative hypotheses on dinosaur behaviour, for instance. Some have proven more testable than others, of course. The stories of evolutionary relationships between fossil species, as well as between fossil and living species, were at one time unverfiable just-so stories. At least this was the case in terms of their expression as narratives. We understood two things to be related because they bore homologous structures, but we also knew two things to be homologous because they were borne in related creatures. The cladistic revolution changed that and allowed us to express homology in terms of the nested distribution of similarities. It be came an explicit way of uncovering the patterns in our observations about fossils.

Palaeontology is still experiencing its growing pains in becoming a mature, accountable, rigorous science. The tools, like cladistics and related methods, are there and so is the ambition to use them. However, the steps towards Wolpert's vision of a science are not complete.

Cladistics, when treated with care and in an honest attempt to eliminate your bias, can be very helpful. However, it can just as easily be used to come up with a tree that makes you feel comfortable. It's just an algorithm. You can shove whatever you want in to make whatever you want happen. The point is what the algorithm does, and the logic behind choosing to apply it.

I won't go into the details of it, because I don't think that's what this rant is about. The point is that it acts like a filter for our observations. We can record apparent similarities between a bunch of fossil, or living, or fossil and living things and interpret, in our own minds, what that means. Or, we can subject it to a particular type of analysis that might not give us what our brain tells us it should be. What it will (hopefully) give us is a result of the kind we want, based on logical principles that we have worked out beforehand. These arguments are, themselves, worked out based on some general principles.

Methods like cladistics might not give us the right result. Methods will, at one time or another and however frequently or infrequently fail. Usually for some reason (the method is bad, we used it improperly, or just pure randomness).

The point, interestingly, isn't even about the right answer. It's actually about the wrong answer. We can and will be wrong when trying to explain the world. But we can be wrong for two reasons: the right reasons and the wrong reasons. It's like anything else, you can have all the calculations and the protocols and experimental controls right, and still Nature can throw you a curveball. Or it can be impossible to collect all the neccessary data, or simply impossible to know how much to collect. Nevertheless, we have to try. This can lead to us being wrong, but we will be wrong at no fault of our own. Alternatively, we can be too caught up with getting a result -- especially the result we want. This leads to us being wrong for all the wrong reasons.

So, for those of us working with fossils, I think we can be wrong for the righ reasons and wrong for the wrong reasons. I think honest attempts to challenge received wisdom, to upset old taxonomies, and question the authority of old is a good thing. It may lead us to stronger hypotheses, if not simply to a more honest evaluation of our data. It might be that we simply don't have the evidence to say all the things we're saying. We might, for the time being, only have the evidence for a very coarse picture of interrelationships of some fossil organisms. But the more we strive for a result, and the more we strive for a result that makes us feel comfortable, the farther our thinking get from Wolpert's description of true scientific thinking.

No comments: