We should consider in each case what Question it is that is proposed, and what answer to it would, in the instance before us, be the most opposite or contrasted to the one to be examined. E.G. “You will find this doctrine in Bacon” may be contrasted, either with “You will find in Bacon a different doctrine,” or with, “You will find this doctrine in a different author.” Richard Whately, Elements of Logic, (II. iv. i).
To summarise the last post: When scientists investigate phenomena, they are implicitly or explicitly approaching the topic at issue with a set of contrasting categories, which are specified by the research question. And that set of categories determines what the phenomenon is. This determinism closely resembles another claim often made in the philosophy of science and language: that we construct all our categories – that no category is really realistic. This very popular view is often associated with the theory-dependence of observation thesis (which I have discussed in The Nature of Classification, chapter 6). Here, though, I am arguing that observation need not be theory dependent, but it is, and must be, expectation dependent.
The category antirealism claim is widespread. It often relies on ideas that scientific theories are all-encompassing worldviews (such as “paradigms”) that determine what we see, because observations underdetermine explanations and hypotheses. The Starting Problem, though, indicates that we never approach observation naively, but that we always have dispositions, either biological or sociocultural or a mixture of both, to observe some things and not others, out of the infinity of things we might observe. Science is a process of refining observations based on a process of refining our expectations (hence the reference to Bayesian logic before). In short, nobody starts knowing nothing. Science can bootstrap from folk-science because folk-science itself is already bootstrapped from dispositions.
Natural categories are thus not naive. At the same time, though, they are seen by experienced observers. The key term here is “experienced”: no observer, whether a scientist or an enthusiast, begins their observations without prior experience being passed on by those who went before and taught that observer. Such teaching may be theoretical, but it may also be ostensive (“like that”), or pragmatic (“you can do this with these”), and it will be taught through trial and correction by the prior expert.
To return to “species”, experienced observers of species are those who have at their disposal a range of tools:
- Practical training by prior experts
- Knowledge of related species (and thus expectations about what will be seen, so that novel features are highlighted)
- Whatever theoretical prior information there may be
- Cognitive ancillary tools, such as what counts as a good explanation in that field.
All these go to make up what taxonomists call “good species”. The prior expectations are based on prototypical models of species (in that group of organisms) so that what fits that model is regarded as a properly describable species; boundary cases are problems to be resolved. This implies, as I have said before, that species are not a theoretical category. That is, there is no unique set of biological groups to which the term “species” applies. It is not a natural category in general. But it can be a natural category for a group of organisms; say, a group of birds, cats or plants. If, that is, there are shared ways (modalities, as I called it in 2004) of being a species in that group. This is, of course, an empirical matter. We have to identify species as phenomena that share certain modalities for that group, and be able to identify the exceptions (as in the whiptail lizard asexual species).
This was a long diversion into a simple topic. I wondered if species could be accounted for as phenomena. From that I asked what a phenomenon is. I rejected the a priori claim that natural categories of science are necessarily derived from theories, which has long been the default opinion in philosophy of science. And I argued that phenomena are highlighted as things that stand out from our existing expectations of the domain in question, using the contrastive account of explanations (but I did not give any view about what explanation is, because that varies by domain and field). Why don’t I offer an account of explanation?
In part, as I noted, explanations satisfy local criteria for a field or domain. Consider mountains. How we explain a given mountain depends on the evidence available (evidence of stratigraphy, isotope ratios, etc.), and theoretical postulates like Steno’s principles of superposition, original horizontality, continuity, and crosscutting. Moreover, we appeal to causal processes like vulcanism, erosion, uplift, subduction, and so on, some of which are more theoretical than observed. There are, as in most philosophical debates, especially in the philosophy of science, monists and pluralists on explanation. Given that in my view (not argued here) there are no unitary, unique and universal modes of explanation in any field of science, for now, it is enough to say that explanation sets up the things that need to be explained, but so too do observation, experience and economic motivations.
Taxa (that is, categories of the natural world) are not therefore de jure objects of theory, and hence, they are not vulnerable to the antirealist attack. But neither are they ready-made, either in the mind or the “external” world. They are constructed categories that are held hostage to empirical data, to experience of the world, and to (as Hacking put it) intervention in that world. And that is enough, I think.