P. D. Magnus

Within philosophy of science, debates about realism often turn on whether posited entities exist or whether scientific claims are true. Natural kinds tend to be investigated by philosophers of language or metaphysicians, for whom semantic or ontological considerations can overshadow scientific ones. Since science crucially involves dividing the world up into categories of things, however, issues concerning classification ought to be central for philosophy of science. Muhammad Ali Khalidi's book fills that gap, and I commend it to readers with an interest in scientific taxonomy and natural kinds. He works through general issues to craft a useful philosophical conception and uses the account to think through a wide range of specific examples. Although there are differences in the details, that one-sentence summary of Khalidi's book could just as well describe my own recent monograph on natural kinds.

There is considerable disagreement about the epistemic value of novel predictive success, i.e. when a scientist predicts an unexpected phenomenon, experiments are conducted, and the prediction proves to be accurate. We survey the field on this question, noting both fully articulated views such as weak and strong predictivism, and more nascent views, such as pluralist reasons for the instrumental value of prediction. By examining the various reasons offered for the value of prediction across a range of inferential contexts , we can see that neither weak nor strong predictivism captures all of the reasons for valuing prediction available. A third path is presented, Pluralist Instrumental Predictivism; PIP for short

If two theory formulations are merely different expressions of the same theory, then any problem of choosing between them cannot be due to the underdetermination of theories by data. So one might suspect that we need to be able to tell distinct theories from mere alternate formulations before we can say anything substantive about underdetermination, that we need to solve the problem of identical rivals before addressing the problem of underdetermination. Here I consider two possible solutions: Quine proposes that we call two theories identical if they are equivalent under a reconstrual of predicates, but this would mishandle important cases. Another proposal is to defer to the particular judgements of actual scientists. Consideration of an historical episodethe alleged equivalence of wave and matrix mechanicsshows that this second proposal also fails. Nevertheless, I suggest, the original suspicion is wrong; there are ways to enquire into underdetermination without having solved the problem of identical rivals.

On-line, just as off-line, there are ways of assessing the credibility of information sources. The Internet, although it arguably makes for nothing wholly new in this regard, complicates the ordinary task of assessing credibility. In the first section, I consider a specific example and argue that Internet content providers have no clear interest in resolving these comlications. In the second, I consider four general ways that we might assess credibility and explore how they apply to life online. Finally, I argue that even careful information gathering may pose subtle pitfalls but that some of these may be mitigated on-line.

According to the standard narrative, natural kind is a technical notion that was introduced by John Stuart Mill in the 1840s and the recent craze for natural kinds, launched by Putnam and Kripke, is a continuation of that tradition. I argue that the standard narrative is mistaken. The Millian tradition of kinds was not particularly influential in the 20th-century, and the Putnam-Kripke revolution did not clearly engage with even the remnants that were left of it. The presently active tradition of natural kinds is less than half a century old. Recognizing this might help us better appreciate both Mill and natural kinds.

Debates about the underdetermination of theory by data often turn on specific examples. Cases invoked often enough become familiar, even well worn. Since Helen Longino’s discussion of the case, the connection between prenatal hormone levels and gender-linked childhood behaviour has become one of these stock examples. However, as I argue here, the case is not genuinely underdetermined. We can easily imagine a possible experiment to decide the question. The fact that we would not perform this experiment is a moral, rather than epistemic, point. Finally, I suggest that the ”underdetermination’ of the case may be inessential for Longino to establish her central claim about it.

The accepted narrative treats John Stuart Mill's Kinds as the historical prototype for our natural kinds, but Mill actually employs two separate notions: Kinds and natural groups. Considering these, along with the accounts of Mill's 19th-century interlocutors, forces us to recognize two distinct questions. First, what marks a natural kind as worthy of inclusion in taxonomy? Second, what exists in the world that makes a category meet that criterion? Mill's two notions offer separate answers to the two questions: natural groups for taxonomy, and Kinds for ontology. This distinction is ignored in many contemporary debates about natural kinds and is obscured by the standard narrative which treats our natural kinds just as a development of Mill's Kinds.

The underdetermination of theory by evidence is supposed to be a reason to rethink science. It is not. Many authors claim that underdetermination has momentous consequences for the status of scientific claims, but such claims are hidden in an umbra of obscurity and a penumbra of equivocation. So many various phenomena pass for `underdetermination' that it's tempting to think that it is no unified phenomenon at all, so I begin by providing a framework within which all these worries can be seen as species of one genus: A claim of underdetermination involves (at least implicitly) a set of rival theories, a standard of responsible judgment, and a scope of circumstances in which responsible choice between the rivals is impossible. Within this framework, I show that one variety of underdetermination motivated modern scepticism and thus is a familiar problem at the heart of epistemology. I survey arguments that infer from underdetermination to some reëvaluation of science: top-down arguments infer a priori from the ubiquity of underdetermination to some conclusion about science; bottom-up arguments infer from specific instances of underdetermination, to the claim that underdetermination is widespread, and then to some conclusion about science. The top-down arguments either fail to deliver underdetermination of any great significance or (as with modern scepticism) deliver some well-worn epistemic concern. The bottom-up arguments must rely on cases. I consider several promising cases and find them to either be so specialized that they cannot underwrite conclusions about science in general or not be underdetermined at all. Neither top-down nor bottom-up arguments can motivate any deep reconsideration of science.