Last updated on 22 Jun 2018
Semifinalist for the 3 Quarks Daily Philosophy Prize 2011
In my last post, I asked whether there was a foundation for my view that species are extra-theoretical phenomena. I have done some further reading, especially Michela Massimi’s book Kant and Philosophy of Science Today, which I will have to buy. I had thought Kant was a dead issue in philosophy of science; of course I was wrong.
Massimi asks whether we can presume that there are “ready-made” phenomena that have to be “saved” (Hacking has a gloss on the phrase “saving the phenomena”; the Latin is salve, which he makes cognate to the English “solve”, thus to save the phenomena is to give an explanation for it). This, she thinks, is the widespread or commonplace view – that science explains phenomena observed without theory, by theory. Kant and those who follow him (including, I think, the whole modern philosophy of science since Mach and Duhem), hold that we construct phenomena from data, a view pushed by Bogen and Woodward (1988).
Add to the constructivist view that observation is theory-dependent – that we cannot even get data without prior theory – and we have a conundrum: in order to explain phenomena by theory we need theories which call forth the phenomena. Hence we do not ever actually discover phenomena in need of theoretical explanation; we make them. There are no “ready-made” phenomena.
If this doesn’t seem counterintuitive to you, then I think you haven’t read much history of science. The phenomena of planets was observed without any theory of celestial spheres; indeed the latter were hypothesised to explain this (non-theoretical) phenomenon. Likewise, the various reactions of naturally occurring chemicals was observed before we had either the Aristotelian or the alchemical theories of elements, or before we had Daltonian theory, all of which were attempts to explain these naturally occurring phenomena.
Granted, in physics and modern chemistry a good many of the phenomena are things that are made in experimental, controlled, settings, as Hacking insists (we intervene in the naturally occurring order to isolate causal influences). But there doesn’t seem to have been a space for unintervened phenomena, unconstructed by prior theory, in the philosophy of science. Why is that?
In part it is because of Kantian views, that we have access only to constructed phenomena and that ready-made phenomena might be indicative of confidence that we could see the Dingen-an-sich directly. This naive realism is not at all accepted by philosophers of science. A claim for the existence of natural phenomena unmediated by theory suggests that sort of realism, and so it is rejected. But does it imply naive realism? I do not think so. At best it implies that there are phenomena in the world we do not construct but can see (hear, smell, etc.) without there being theory that we base our perceptions upon. However, it is a long step from that to the claim that the phenomena are representations of the way the world is directly. Observing a chemical reaction (soda and lemon juice, for instance) is not something one can only do based on a theory of chemistry. It is enough that we have a commonsense experience of the world such that such a reaction is marked out as interesting and different.
But a response that one often encounters from philosophers is that in even observing these differences we rely upon theory: the “theory” that is embedded in our perceptual dispositions. Sometimes this is called psychologism, but in this case it is usually referred to as “evolutionary epistemology”. These are the dispositions that we have to explore a “quality space” as Quine called it. The Kantian analytic a prioria are the evolutionary synthetic a posterioria; to use a phrase of Konrad Lorenz’.
This is a markedly deflated notion of “theory”, so poor as to mock the meaning of “theory” in science. If just being able to detect motion in a visual field is a theory, everything is theory and we need not attend to theory-dependence as such. But now we must still deal with scientific theory differently, and the phenomena that it explains. Let us call scientific theory Theory with the capital, and leave the rest to psychological and sociological dispositions. Is observation Theory-dependent? Are phenomena Theoretical? I still say they need not be, although if they are constructed through intervention they may be.
Consider a case in which a scientist has an instrument that assays some property, like temperature. On the Bogen-Wooward view, the phenomena of the temperature of a heated piece of metal is the pattern of several measurements of that piece using the thermocouple (or whatever the tool is). Unless you have a pattern in the data, there is no phenomenon. Since the functioning of the thermocouple is explained by Theory (of, say, relative metal expansion rates), the observation is Theoretical. But any metal worker knows that metal heats up when put in flame, and they will have a pretty good idea of what you need to do in order to get a metal to just below the melting point, by the whiteness of the glow they give off (under uncontrolled lighting, too!); this is how swords were made for centuries. That is not Theoretical. It is just a ready-made phenomenon.
So there remains room for ready-made phenomena, even if to perceive it one needs to be trained and experienced, without Theory. Hacking makes something like this point about microscopes and telescopes – we ascertained that they worked without a theory of optics (even though some pretty good Theories of optics had been around since the Arabs), by checking through direct observation that they were properly imaging objects, and making ampliative inferences from there.
To return to species: they began to be properly named at the fin de siecle of the herbalist medieval tradition, based largely upon experienced observation by people like Gesner, Bauhin, von Puch and others. There was no Theory to speak of that required species, apart from a need to track naturally occurring phenomena, and yet many of the species they named remain good species today. Moreover, some assay-driven splitting of species, based largely upon molecular techniques, is counter-intuitive when the groups so split (into “Operational Taxonomic Units”) are Theory-driven but run counter to broader observation of interbreeding, ecological adaptation, and so on. Theory-based instrumentalism is not triumphant by any means. [I have in mind here the DNA barcodists who rely on a short stretch of mitochondrial DNA in eukaryotes, cox-1, and similar sequences in “prokaryotes”.]
Species can be identified by experienced observers in the absence of Theory. One must be trained and corrected in one’s taxonomic apprenticeship, and much of the “knowledge” employed is tacit practice, gained, I suggest, from past experience of phenomenal distributions of closely related organisms. The sense of Theory employed is ancillary, and not crucial (if one assay starts to fail, well, as Groucho once said, we have others). The Theory-dependence of modern natural history (excluding, that is to say, physics and general chemistry) is not central to identifying the phenomena that stand in need of Theory to explain them.
The Theory of the assays, however, can range from distal and irrelevant (as the quantum theory of optics is for identifying, say, mammal species) to closely relevant and even partially overlapping the domain of observation (quantum optics may play a critical role in identifying microbial species, for example). I suggest that we think of Theory-dependence in a more nuanced way: a phenomenon is Theory-dependent iff there is no phenomenon observable unless a Theory is necessarily employed in the observation.
So particle tracks in physics are Theory-dependent. However, I do not believe most species observations are (if we restrict ourselves here to species that can be observed in the absence of specialist tools like DNA barcoding chips). The characters and properties of species of fishes, flowers, and foxes are ready-made phenomena that can be seen if one learns how even in the absence of Theory, or else we cannot account for the history of taxonomy and natural history in general. Moreover, such ready-made phenomena are a good reality check on the assay-driven phenomena that do rely to greater degrees upon Theory. Maybe I cannot see DNA clustering, but I can see traits (especially homologies) if I study the groups. If DNA clustering implies there is a phenomenon of species that we do not pick up by trained direct observation, that becomes a testable hypothesis; it is not true by assay definition. It might turn out that cox-1 is a highly variable gene in some species, or that it is shared across many species unchanged; such is the nature of evolution.
I therefore propose that we call things “phenomena” when they are patterns of observation, and index them to the theoretical domain in which they properly occur (that is, in which they are to be explained). If the observation does not rely on the explanatory Theory, they are to be regarded as Theory-independent for that domain. There will, of course, be Theory that explains why we can observe those phenomena (and it might be a distal application the the relevant domain: we see because we evolved to see objects of that kind, and evolution explains why there are objects of that kind – this is not to say that the observation is Theory-dependent). But it is not necessarily Theory of that domain.
Many objects are observed because they are indeed objects of that domain. I am not denying this. I am merely saying that we have unjustly excluded from our philosophies of science a phenomenon of observing phenomena without Theory. One learns that by observing history, without a Theory of history …
Bogen, James, and James Woodward. 1988. Saving the phenomena. The Philosophical Review 67 (3):303–352.
Hacking, Ian. 1983. Representing and intervening: introductory topics in the philosophy of natural science. Cambridge UK: Cambridge University Press. [Is it just me or is this the least introductory book ever? How many deep discussions have followed this “introduction”?]
Massimi, Michela. 2008. Kant and philosophy of science today, Royal Institute of Philosophy supplements,. Cambridge; New York: Cambridge University Press.
If you are going to say that ordinary observations are not theoretical, then why not also say that observations made under theories are not theoretical? I am not seeing an important difference. Is “theory” even well defined?
I’ll suggest an alternative way of looking at that. The mathematician, doing Euclidean geometry, uses a theoretical procedure to construct the bisector of an angle. However the mathematician, particularly if a platonist, will insist that the bisector was already there even before he constructed it. Unfortunately, it was not visible. So the construction was done to make it visible, to discover what already existed. My suggestion is that you think of phenomena revealed by theory laden data in the same kind of way.
John – A good post, addressing a really difficult set of issues in philosophy of science. I’d appreciate a bit of clarification about how Theory is necessarily employed in the observation of some phenomena. It’s your example of particle tracks that got me wondering… It’s certainly true that seeing streaks in a cloud chamber _as_ particle tracks involves physical theory. But the streaks can be observed by people ignorant of all physical theory, even though there’s a lot of physics invoved in the construction and operation of the apparatus that generates the streaks/tracks. So, my question is whether the streak absent its interpretation as a particle track is a phenomenon in your sense?
I think even our everyday observations are theory-dependent, or at least, are not metaphysically neutral. Certain qualities like spatio-temporal contiguity etc, make us think that there is some object, a tree, that persists through time. All of these things are not givens. None of them explicable through mere correlations among sense-data or your preferred empiricist primitive. What’s not theory here? Even to pick out some patterns as more important or more coherent than others is probably resolvable into some implicit theory.
Something is spamming your commenters, on behalf of “Gravitar”, and signing as you. Or at least I assume it is spam, and you are not upset at my lack of a Gravitar image 😉
It is interesting to see how Kant’s views are always distorted. Transcedental concepts are misunderstood as “theory” which can give us some knowledge regarding perceived things even though Kant several times warned that without empirical “Anschaungen” or thinking in pure terms they mean nothing. Actually “Begriffe” are often rules that govern our reason (btw. whats the diffrence between der Verstand and der Vernunft in English? According to Hegel it is very important difference!).
So Darwinists speak about “dog” and do not know if it is our abstraction or imagination or what. The question was addressed by Kant in his “Transscedentalen Doctrin der Urtheilskraft” in Critique of pure reason more than 200 hundreds years ago.
I have made a blog post (A mathematician’s take on phenomena) that is related to your topic, and outlines the mathematical model that I am using for such questions.
Having been at Stanford I am obviously biased, but Kant is most certainly not dead in the philosophy of science though I am not enough of a historian to understand the trajectory. But I take it that at a minimum, Michael Friedman’s work such as Dynamics of Reason has been well-received. As for theory-ladenness, there is some work in perception that shows that when read in a certain way, theory-ladenness is nearly trivial. No, we didn’t need celestial spheres to observe the planets, but we must have had some conceptual framework in place or we wouldn’t have “observed” anything at all. For example, you might think that we observed some planets traveling across the sky. But of course they don’t actually do that on anything like the obvious description that anyone (even most people today) has in mind. That “observation” only makes sense in a conceptual framework where the earth is stationary, the points of light that look relevantly similar in brightness and size move in some continuous motion on a plane rather than say moving closer and further away, other points of light in the sky (stars) do not move at all, etc. This is most certainly a “theory” of the relevant kind even if it is not a full blooded scientific theory in some way that you care about. To take another example, it has been pointed out (perhaps Hull?) that to even say things like this animal has four legs requires a theory about legs and correspondence between parts of different animals, etc. and I would hypothesize that extremely basic comparative anatomy is far “less theory laden” than astronomical observations if such a comparison even makes sense.
Hence my distinction between Theory (in science) and theory (as a general account of the conceptual furniture of a percipient). If everything is theory, then the term has little purchase.
That comment of David’s regarding spinnerets in Science as a Process always bothered me. Sure you need some contrast to see a salient feature, but any five year old can see where the silk comes from on a spider. If that is theory-dependence then the notion is vapid. And identifying legs is theory-dependent? I don’t think we are talking about the same issues if that is the case (although of course a leg can be a non-leg when it is not used for walking, as the post I wrote about Aristotle and the Mayfly showed).
My claim is not that there is nothing one needs to have in order to observe, a club-footed empiricism. My claim is that phenomena can be independent of any theory of the domain being observed.
Still not certain that a hard distinction between Theory and theory exists. No doubt all observation is (T/t)eory dependent in some way, but sophisticated observations require Theory not to mention sophisticated instrumentation, and just every day observation requires theory and conceptual furniture that’s not nearly as sophisticated in its logical structure. Does it really matter that much whether phenomena are the theoretical terms in some other primitive theory that are being translated into some other theory rather than a different domain? Also, how do I individuate domains? My primitive 5 year old theory about spider silk does not encompass or quantify over the lineage of silk producing animals, the complex genetic relationships that create the molecular machinery to produce silk, or the ecosystemic relationships that make silk production a boon to spider survival.
I missed the article on Aristotle. Nice!
Their were of course only three things Aristotle did not understand; according to Irish folk tradition.
Intinn Mana (a womens mind), Lionadh agus traghadh na taoide (the coming and going of the tide) and most importantly Saithar na mBeach (the work of the Bees)
In order to understand the work of the bee he erected a glass box and put a hive in it, to find out how they made honey. But when he came back to it the next day the bees had plastered the glass box with wax and he could not see inside.
He was so vexed that he kicked and broke the box and the bees flew out and stung him and blinded him.
After that he went away travelling as people had no respect for him, as he was blind and because he never found out how the bees made honey.
That must have been Aristotle O’Shaunessy, not the Greek, because
1. Aristotle of Stagira never went blind
2. Glass was not so common or workable into sheets that a classical Greek could have constructed a glass hive
3 The Greek Aristotle was too patient an observer to have ever lost his temper like that.
I will stipulate that you can make one observation of a phenomenon without using theory. That is OK. You can’t abduct theory from one observation anyways.
But observe two instances of the same phenomenon, and you can start building theory. Unfortunately, someone like me is going to ask what theory you used to decide that your observations were two instances of the same phenomenon.
It seems to me that any observation not laden with theory is also not laden with significance. Nice posting, though.
Whoops. I missed your reply to Velasco. That answers me pretty well too. Oh, well. Move along. Nothing to see here.
That is a good question, implicit in the notion of a “domain”. For example, the domain of biology has certain criteria and constraints (zoologists tend to identify things that move, for example, under their own steam). So, anything that meets these criteria will be salient as phenomena. But that doesn’t make these things objects of a theory of the domain.
So species are things that resemble parents through generation, bracketed by our knowledge of other things in that domain. But the theories of biology, if they simply lack species as objects, are not thereby defining species into existence. They are ready-made in just the sense that we are forced to identify them if we apply the same criteria we do elsewhere.
No domain is approached tabula rasa, of course. To restrict theory-independence to anything that is tabula rasa is to set the bar so high everything is theoretical; and as I said, if everything is theoretical we don’t need to distinguish theory from Theory and we don’t need to explain scientific practice. That is one position, but I find it rather absurd.
One may be an empiricist without being a Lockean empiricist.
I just want to pick again on “species are things that resemble their parents through generation, bracketed by our knowledge of other things in that domain.” I just don’t find that a useful definition of “species”, and have said so before. I’ve asked for clarification before too. If that’s a version of the lineage species concept, it isn’t clear. If it isn’t, what is it?
A digression, I know. But worth exploring.
And as I have said before John, that isn’t a definition. It’s a phenomenal concept – something we observe which calls for an explanation in terms of various theories such as genetics, phylogenetics, ecology, development and so on. My point is that we noticed and named species long before there even were such theories, and what is even odder, they often remain good species now under our theoretical explanations.
The lineage, cluster, ecological etc., conceptions of the concept species define what [those kinds of] species are; the generative concept is simply an observation that things resemble their parents by generation (whereas mineral and chemical species do not because they don’t generate).
You are overinterpreting here. The modern conceptions of the concept explain the concept and the phenomena it denotes, that’s all. But it is simply projection to insist that the concept must always and only mean some preferred definition of the theoretical explanation. In history we call it Whiggism.
I’m afraid that was largely opaque to me. What you seem to be talking about is inheritance, not species. Nothing about inheritance, per se, would seem to imply that there would be species.
Now, what we observe are clusters of similar organisms, populations minimally but consistently different from other such clusters in the same area. Might call those species, I suppose.
Nothing about any theory of science implies there must be species (rather than populations, lineages, types, genomic clusters, etc.). That is my point.
Then once again, I appear to understand nothing about your point.
I hope you return to this subject at some point.
I find it quite straightforward and understandable if I direct it towards medieval sources.
Comments are closed.