How I try to think, and how I try not to

[haggholm]

As someone who values the scientific method and philosophy of logical inferences and dedictions, empirical observation, Occam’s Razor, and so forth with a rather long list of highly valued principles; as someone who thinks that few approaches to finding truth can compete with experiments in physics, or double (or triple!) blinded medical experiments (when evaluated as Science Based Medicine [SBM] rather than Evidence Based Medicine [EBM], with the former’s greater understanding and acknowledgement of prior probabilities, etc.)—

—As a naturalist, materialist skeptic, in other words, one of the most irritating and most pernicious cognitive traps is that of scientific-sounding rationalisation, that isn’t science at all. I daresay it’s something several people I have talked to (or do talk to) would accuse me of, had they happened to express themselves using my exact vocabulary (though it isn’t their entire beef). I do acknowledge that it’s real; I do not claim to be immune to it; I do my best to guard myself against it. …But what am I actually talking about? What is the basis for my worldview, what are the common errors and how do I try to guard against them; what are the inherent weaknesses, and how to I justify adhering to this mode of thinking in spite of them?

What science is

Science is a process of finding the truth¹ by

1. Making observations (gathering data);
2. Constructing a model (forming a hypothesis);
3. Making predictions based on the model;
4. Verifying the predictions, and throwing out the model if it’s wrong.

Of course that’s a very rough generalisation, but that’s the basic idea: You need data to construct a model; you need a model to generate predictions; and if you don’t have any verified predictions (which means they must be falsifiable! —it’s not verified unless you leave room for failure if the model’s wrong), it isn’t science.

Scientific explanations aren’t like that, though. A scientific explanation of an already-observed phenomenon is not science; it’s just based on it. It cannot be science: In order for it to be science, I need to construct a model and check that it’s a good model. We don’t always do this when we explain something scientifically. Instead, a scientific explanation is an explanation of some phenomenon or occurrence based on what science has shown is feasible. If you show me an example of purported levitation, or a UFO sighting, or similar, and I explain (very reasonably and probably correctly) that there are natural explanations for what I’ve seen, I’m just observing that it fits the current scientific models.

Post hoc rationalisation

The real problem, here, enters the picture when we try to construct models and treat them as scientific models relying completely on post hoc data—and the difference can be a subtle one. The problem is that it is impossible, without generating and testing predictions, to know whether the model is actually correct, or merely happens to apply to the data at hand—which may be woefully incomplete, subject to natural selection bias or (more sinisterly) to cherrypicking.

This is a problem not just in discussions where people attempt to sound scientific, but also historically with science itself—very notably before the modern scientific method was formulated. People who don’t trust the scientific way of finding the truth are often fond of pointing out egregious pseudosciences that were, in their own time, respected and considered as scientific as anything else. Whether the balance of humours, the theory of miasms, the notion of the luminiferous æther, phrenology, or any other such now-discredited concept (I almost said homeopathy, but it never really did have that sort of credibility), they once did serve as models to explain all the data compiled into their construction—but all of them were wrong, and were able to survive because they were not used to generate and test falsifiable predictions. …Survive for a while, that is: By now all of the above are discredited because we do know better; and sometimes, to their credit, it is the proponents of ideas that discover and publish the fact that the ideas are wrong; as with Michelson and Morley and their famous æther experiment.

Let me be very clear and explicit in saying that if you cannot generate and test predictions, it is never science in the true sense, however reasonable your explanation. (Note that prediction refers to a prediction of what we will find—in historical and palæontological sciences, the events happened long ago, but we can still make predictions about what new data will discover; hence that famous prediction in palæontology that we will never find a Cambrian rabbit fossil.) This is why interpretations of quantum physics are metaphysical rather than physical, since they do not make different predictions. (On a not unrelated observation, I despise the term string theory—it is string hypothesis until predictions are tested!)

Equivocation

As with any profession or culture, science is full of very technical jargon. Many of the most awful abuses of science I have come across have been straightforward instances of equivocation. The most infamous is, of course, the creationist’s claim that evolution is just a theory, apparently unaware that an explanatory framework in science can have no higher title. (Some would have it that if evolution were proven, it would be the law of evolution, but that’s just not true: Explanatory frameworks are never elevated to laws, no matter how solid. To quote Stephen Jay Gould,

Evolution is a theory. It is also a fact. And facts and theories are different things, not rungs in a hierarchy of increasing certainty. Facts are the world’s data. Theories are structures of ideas that explain and interpret facts. Facts do not go away when scientists debate rival theories to explain them. Einstein’s theory of gravitation replaced Newton’s, but apples did not suspend themselves in mid-air, pending the outcome. And humans evolved from ape-like ancestors whether they did so by Darwin's proposed mechanism or by some other yet to be discovered.

Another favourite term to pervert and subvert is that of energy, which quantum quacks gleefully use as though I feel full of energy were related by an equation to E=mc²—when in fact energy in physics is a strictly and technically defined term, related to the colloquial word by etymology and analogy, but no more. Force suffers similar abuses.

Now, I’ve used pretty obvious examples, but equivocation can actually be pretty difficult to detect when used skilfully, and it can lead to genuine misunderstanding. The equivocater’s trade lies in

1. equivocating using such language that the equivocation is not obvious (i.e. it looks as though the technical sense might apply on both sides of the equation); and/or
2. couching the equivocation in a discourse sufficiently technical that the reader (or listener) just isn’t qualified to tell them apart.

The latter is particularly pernicious because the same word may mean different things in different contexts, or have different definitions in the jargons of different fields. Apart from education (which sounds like such a good idea, but really, you can’t be well educated in everything) and listening very carefully, about the only way I know to avoid this sort of thing is to make sure that anyone who claims that two things are mathematically related can actually show the form of the equation.

Non sequiturs, red herrings, and other fishy things

Another problem, less subtle than true equivocation and more easily countered, but nonetheless a cause for vigilance and concern, enters the picture when the model is (or may be) complete, but the relevance hasn’t been established. One example I see very frequently indeed is in the context of the martial arts message boards I frequent, where every so often someone will attempt to validate his (usually arcane) form of martial arts as having the superior form of punching by using not fight records (showing that his stylists can beat up other stylists) but physical equations—attempting to show, for instance, that this method really does maximise the momentum transfer of a punch. Maybe so; maybe not. The problem here is that it was never shown that maximising the momentum transfer is really what makes a punch most powerful; the next person with the next martial art will instead present some half-arsed equations showing that he can maximise force—or impulse, or velocity, accoleration, jerk, pressure, or other physical property. It may be some time before anyone steps back from trying to disprove the equations to realise that it’s a non sequitur.

In short, one must not fall victim to the belief that just because someone backs up a claim with a scientific-sounding argument, that argument necessarily supports the claim. It is a requirement of a valid logical argument that the conclusion necessarily follows from the premises—but this has to be established; it cannot simply be assumed.

In defence of Occam’s Razor

Occam’s Razor is the philosophical principle that out of any set of explanations, the simplest is always to be preferred:

Numquam ponenda est pluralitas sine necessitate.
(Plurality must never be posited without necessity.)

Frustra fit per plura quod potest fieri per pauciora.
(It is futile to do with more things that which can be done with fewer.)

William of Ockham

My own favourite formulation runs somewhat as follows:

Given two explanations of a phenomenon, all else being equal, choose the one that requires the fewest assumptions.

This always looks like a suspect—why should we prefer the simpler explanation? Aren’t explanations of things sometimes genuinely and necessarily complex? How on Earth can you justify basing a depiction of reality on something so vague and susceptible to error?

The answer to this is twofold. First, note my inclusion of all else being equal—of course if we have two explanations for a set of phenomena, both of which generate testable predictions, and one of them produces significantly better fits to newly observed data, that’s a pretty good case to support this one even if it’s more complex. Occam’s Razor isn’t to be used as an excuse to throw out superior models.

Secondly, it is an unfortunate fact that for any phenomenon, you can generate an infinite number of explanations by making them increasingly complex. A simple example is found in every statistics class, where we find that anytime we try to fit a polynomial curve to a set of data points, we can always achieve an equally good or better fit by increasing the order of the polynomial (making our explanation more complex). In fact, more complex explanations are here usually better fits to the data, because the data have measuring errors and so don’t fit exactly to the predicted curve of the right explanation, whereas a very high-order (complex) polynomial can jitter up and down to match every blip in the data.

It should be pretty clear that we can’t get very far if we attempt to maintain an infinite number of explanations for everything we try to explain. Instead, we choose the simplest explanation that works and resort to more complicated ones only if it turns out that the predictions we get from our simpler models aren’t up to snuff. Of course, this process is empirical—we may find out only in time that our model has shortcomings. Unfortunately, it’s the only way we have to proceed; it would be nice if we had some more absolute way of finding out truth, but we don’t. (This is why it is often said of science, properly done, that it tends to approach the truth asymptotically—it can never reach an unassailable, absolute position of Truth, but every subsequent model, because it has to account for all the data that the old model did explain, as well as the data that we’ve found the old model fails to explain, is a better model than the last.)

A more colloquial sort of defence of Occam’s Razor is to observe that it is the principle whereby we explain the world in terms of things we know to be possible, rather than positing arbitrarily things that we don’t know have even a chance of being true. In an example that I freely acknowledge to be something of a reductio ad absurdam, if I leave some chocolate on the table next to a child, turn around for a minute, then return to find the chocolate gone, and the child claims that he did not eat it, but rather that the chocolate was teleported away by aliens for scientific study, I would be an idiot to believe him—not because it is absolutely impossible, but because one explanation (he’s lying) rests on observably possible things (children sometimes eat chocolate; children sometimes lie), whereas the other postulates something (the existence of aliens with teleportation technology) for which there is no evidence.

Jumping from that insultingly trivial example to something that actually is a matter of debate, some people claim that the human mind cannot be the product of the brain alone, but must also rely on something called a soul, or other immaterial and scientifically undetectable entity that is not a product of material causes. Here, the naturalistic explanation is a lot less obvious—we cannot at present show how the mind results from the brain. Some would also argue that the alternative explanation (there is a soul) is less absurd than the alien hypothesis (though I would disagree).

The difference, however, is purely quantitative. We know that brains and minds are very intimately connected—we can observe the mental and psychological effects of damage to the material brain, artificial or natural alterations of brain chemistry, et cetera; we can scan activity in the physical brain correlated with thoughts and emotions; we can view the material development of the brain from embryo to developed organ. The assumption in the naturalist model is just this: That in a complicated neural network consisting of a hundred billion neurones, with feedforward and feedback loops, and modulated and assisted by glial tissue and chemical catalysts in the form of neurochemicals; in such a network that has, furthermore, evolved by fairly well-understood principles of natural and sexual selection, sufficient complexity has arisen to explain the minds that we now experience.

The soul hypothesis may look simpler because it can be summarised more briefly, but its assumptions are actually huge. It postulates the existance of something for which we have no evidence whatsoever—the materialistic hypothesis is complex, but it rests on established facts. Furthermore, it postulates that this intangible soul—which no scientific instrument has been able to detect—is yet able to exert causal effects on the material brain, since it is very clearly established that it is in the brain that our motor impulses originate. (René Descartes thought that the soul operated on the brain through the pineal gland.)

It also suffers the difficulty that if we be allowed this one, completely unfounded assumption (there exists a soul), it is difficult to see why we should disallow any unfounded assumption as a valid rival, so long as it is not falsifiable; for instance, I might equally well claim that your brain is run by a computer program (written by aliens, or the NSA, or the Illuminati); that you are mind controlled by the White Mice… These sound more ridiculous to us, but that is a cultural artefact. The Soul Hypothesis and the White Mice Hypothesis have an equal basis in evidence.

I daresay that William of Ockham, a Franciscan friar, would disagree with my conclusion; but while I think that his beliefs were most likely pretty absurd, I nevertheless think that his principle of parsimony is a necessary part of a rational worldview.

Occam’s Razor or rationalisation?

When I get involved in discussions or debates questioning this worldview and philosophy of mine, and (this is, alas, a limited subset) when the person I am conversing with is neither stupid nor patently insane, one of the most common and most reasonable objections runs something like this (in spirit; I reformulate rather than paraphrase):

What you say generally makes sense, but because the philosophy ultimately relies on empiricism in determining what’s real and what’s not, it is—as you acknowledge—imperfect, and unless you end up clinging dogmatically to your skeptical beliefs (contrary to your own philosophy) you are bound to change your mind on things as new evidence emerges.

Well, given that, by what right do you reject (for instance) the soul concept so strongly? You may think that there is no evidence, but evidence may come up, and you certainly have no strong counterevidence. Why not just keep an open mind?

…As, of course, I should, but as the saying goes, You should keep an open mind to new ideas, but not so open that your brains fall out. It should be noted and emphasised that when I say There is no such thing as a soul, what I mean is To the best of my knowlege, and according to the best evidence and reasoning available, there is no reason to think that there exists such a thing as a soul. (Whether I am truly as open-minded as I should like to new evidence in the areas where my worldview is heavily invested is, of course, hard to say; but I’m defending the way I try to structure my thinking, not rating how well I measure up to my own ideals.)

Ultimately, I regard this as a quantitative rather than a qualitative distinction. It is not whether I am willing to credit the possibility of some improbable concept or not, but just how large or small a probability I am willing to credit it with. My own tendency and preference is to treat extremely improbable things (souls, færies, ghosts, celestial teapots, alien abductions, gods, Great Green Arkleseizures, effective reiki or homeopathic remedies, etc.) as so improbable that unless better evidence or arguments are presented, I will not bother with the possibilities of their reality at all. I will freely acknowledge that they are non-zero, in the absolute sense, but it’s such a small one (0.00…01%) that it’s not worth bothering with. This is, of course, where I often and easily come off as contentious or even contemptuous, and people whose approaches otherwise look similar to mine may differ in how small a prior probability we should assign these things. (If you want a gentler version, read the late, great Dr. Carl Sagan’s The Demon-haunted World.)

Of course, being but a human, I’m very likely to be wrong about a great many things, some of which may very well be of fundamental importance to my worldview. If we were to line up all the things I consider to be ludicrously improbable, I expect that, due to human fallibility and my own sometimes unfortunate tendency to over-swift rejection of apparently unscientific notions, I will be wrong about a number of them. But I consider the probability of my being wrong on any specific such matter to be very small. In the end it comes down to whether I would rather say Well, maybe to everything I come across, or whether I’m willing to say Yes and No and be prepared to eat my hat when the day comes when I turn out to be wrong. I’m willing to do the latter (against which occasion I cleverly do not wear a hat). To me, there’s no other sensible way to proceed. If I accept the soul hypothesis as likely enough that I shouldn’t reject it utterly out of hand, I cannot remain intellectually honest and consistent without accepting any number of other such claims—UFOs, homeopathy, reiki, unlucky black cats, astrology, and all the rest.

I would rather be forceful and intellectually honest and consistent, and occasionally be wrong, than either waffling about every incredible claim anyone cares to make, or arbitrarily accepting improbabilities based solely on their cultural or social acceptability.

---

¹ Some people hold that hunches and intuition are as good guides to truth as is scientific inquiry; to which, apart from obvious retorts (Would you rather travel on an airplane built according to scientific principles, or on one based on intuited ærodynamics?) I would like to reply that whenever a hunch is opposed to reason, I have a hunch that reason is right; so hunches cancel out and we’re left to trust to reason².

² Unless you belong to that school of thought that claims that there is no such thing as objective reality; that all reality is subjective. I have a vast dislike for such solipsism, but I suppose I can’t deny you your subjective reality; in my reality, however, things are objective and that school of thought has no value or standing.

Comments

[petter-haggholm.livejournal.com]

Now there’s a reply whose meaning is not easily deduced…

…Unless you just want cake. I can make this sort of brownie-like cake.