Text Word Count: 2,963

Abstract Word Count:

Bill. Rottschaefer,

Lewis and Clark College,

Abstract

In The Book of Evidence Peter Achinstein attempts to show that philosophy of science is important for science. He proposes to do so by providing an account of a central feature of scientific practice, the use of evidence to establish hypotheses. His aim is to provide an analytic understanding of evidential inference that is scientifically adequate. He does so by providing precising, clarifying and supplementary definitions of the concept of evidence. I focus on the latter, examining his interpretation of evidential probability as epistemic and objective. I argue that his understanding of the objectivity of the evidential relation fails to be of help to scientists. This is so because the relation, as he defines it, holds quite independently of the inferential processes that might be used by scientists. Consequently, it fails to show how inferences based on evidence are justified. Because the evidential relation is independent of inferential processes, its constitutive epistemic norms and ideals have no root in scientists in-built or learned inferential epistemic capacities. The problems with Achinstein’s analytic approach to issues in philosophy of science adds further weight to the view that a scientific naturalistic account of scientists’ epistemic practices would be more helpful to scientists.

(Word Count: 199 words)

In his The Book of Evidence Peter Achinstein sets the context and motivation for his demanding technical account of evidence with a charming story. Achinstein tells us that he was motivated to write his book by the challenge of a former Dean of his college, "a scientist with high intelligence but low boiling point" who in a faculty meeting remarked, "Peter, you have never made a contribution of interest to scientists." Achinstein charitably decided that the Dean’s remark concerned philosophers of science generally and not himself personally.

How might a philosopher of science contribute to scientific work? Consider the case of Hertz and Thomson, the one upon which Achinstein focuses. Hertz and Thomson knew a lot about cathode rays and charged or neutral particles. Hertz and Thomson also knew something about knowing about cathode rays and charged particles; in particular about evidentially based inferential practices that supported their claims. As an epistemologist of science, Achinstein vigorously affirms philosophy of science’s contribution to helping scientists improve their knowledge of how they know about their subject matter, in particular their knowledge of evidential reasoning.

I shall argue that Achinstein’s version of the objectivity of the epistemic relation does not improve a scientist’s knowledge about her knowledge. I have two interconnected reasons for this critique. These concern Achinstein’s neglect of the processes by means of which epistemic agents reliably acquire their evidential beliefs and make evidential inferences and his failure to specify consistently the epistemic agents to which the norms of evidential inference he develops apply. In this paper, I shall develop the first of these critiques. In conclusion, I suggest that a naturalistic epistemology of science does a better job of addressing these subject-related factors of objective epistemic probability than does Achinstein’s implicitly analytic approach.

Using Thomson’s discovery of cathode rays as a paradigm case for understanding scientific evidence and the inferential relation, Achinstein finds that scientists have four distinct, but legitimate understandings of evidence: veridical, potential, epistemic situation and subjective. He seeks to clarify and make more precise the scientists’ implicit or explicit interpretations, rather than reforming or replacing them. Specifically, he clarifies that when scientists claim that an alleged finding (e) is evidence for a hypothesis (h), they mean that it provides a good reason for h being true. Using probability theory, Achinstein then proceeds to make mathematically precise what is meant by "e providing a good reason for believing that h." He imposes a high probability condition requiring that e must make the probability of h being true greater than a half. Thus he rejects the view that e need only increase the probability of the truth of h relative to its competitors. He also lays down an explanatory relevance condition. This demands that the probability that there is an explanatory connection between the hypothesis and the evidence, given the truth of both is greater than a half. Though these criteria are controversial, it is arguable that they are implicit in scientific usage and represent important aspects of evidence and the evidential relationship. Each of the different concepts of evidence builds on the basic intuition that as evidence e provides a good reason to believe b^. In addition, each requires the high probability and explanatory relevance conditions. Veridical, potential and epistemic situation evidence all require that e and b (background beliefs) be true. Subjective evidence requires merely the belief that they be true. Veridical evidence requires in addition that h be true.

Achinstein also proposes a supplementary analysis of evidence and the evidential relation, maintaining that the probability claims embodied in these definitions refer to epistemic probabilities because they deal with epistemic objects, i. e., evidence, hypothesis and their relationships, and that these probabilities are objective, that is, independent of epistemic subjects. Arguably this supplementary analysis is the most important of the potential contributions that his analyses might make to scientists concerning their knowledge of evidential inference and evidence. Supplemental analyses might be a centrally important sort of contribution that philosophy of science makes to the sciences. Precising and clarifying analyses merely refine what the scientist already knows about her knowing about the subject matter of her discipline. On the other hand, if successful, supplemental analyses add to and improve the scientist’s knowledge about her knowledge.

Achinstein distinguishes between subjective and objective conceptions of the evidential relation. Subjective conceptions of the evidential relation make it a relation that involves an epistemic subject. What the evidence for a hypothesis is depends upon some epistemic subject. Evidence is what it is thought to be. On the other hand objective conceptions understand the evidential relation as independent of epistemic subjects. Evidence is what it is. Closely connected with this dimension is that of the instantiation of the evidential relation. Achinstein has three official options for instantiation. First, on the subjective end of the scale, evidence is relativized to the belief of an individual or a group. Second, there is relativization to any one in a given epistemic situation, a circumstance that is characterized by knowledge and/or beliefs in the context of which e provides or does not provide justification for holding h. Both of these relationships are epistemic, but only the latter is objective, since epistemic situation evidence does not depend upon the knowledge or beliefs of any particular epistemic subject (s). Nor does it require that any epistemic subject be in it. Yet this objective relationship could be manifested in epistemic situations, circumstances that an epistemic subject could be in. Achinstein often describes the third option, one further along the objective side of the scale negatively as neither being of the subjective sort nor concerning an epistemic situation. Here the relation is considered objectively, independently of any particular instantiation in a person, group or epistemic situation. This is the evidential relationship itself.

We are now ready to examine the adequacy of Achinstein’s interpretation of the objectivity of evidence and the evidential relation.

Consider would-be investigator Willie, a contemporary of Thomson. He argues along with Thomson that the deflection of the particles in the cathode ray tube provides evidence for the hypothesis that the particles constituting that ray are charged. Assume that the evidence is, as Achinstein maintains, veridical evidence and that Willie is in Thomson’s epistemic situation. Willie believes that the particles are charged. Willie, as anyone else in Thomson’s epistemic situation, would seem to be justified in believing that the particles are charged. Moreover, ex hypothesi, Willie’s belief is true. Thus Willie appears to know that the particles are charged. However, unlike Thomson, Willie is making his observations in German labs which though they have been reporting deflections, these deflections, unbeknownst to the German investigators, are usually due to the faulty detection mechanisms manufactured in Willie’s Leipzig plant. Nevertheless, on the occasion of his current observation, the detection mechanism is working correctly. There is genuine deflection. Thus, it appears to be a matter of luck that Willie rightly judges in conformity with Achinstein’s explicated notion of scientific evidence that deflection is occurring in the German lab.

Given this scenario, there are legitimate doubts about whether Willie, or anyone in Willie’s epistemic situation, knows that the particles are charged on the basis of the evidence that they are deflected. This is so, despite the fact that all of Achinstein’s conditions for the possession of potential and veridical evidence are filled and even though Willie appears to be in Thomson’s epistemic situation. Willie does not even have a justified belief that that the particles are charged even though he possesses veridical evidence that they are. Something more than Willie’s internal epistemic state, instantiating as it is does the requirements for veridical evidence, is required, if he is to have justification for his belief. Epistemologists of science need to consider the external circumstances of the epistemic situation in which Willie has acquired his belief that the particles are being deflected be taken into account. These circumstances are replete with faulty detection mechanisms. In such circumstances, Willie cannot reliably discern deflection.

So let us suppose that Willie has been apprised of the faults of his Leipzig instruments and has spent much time with Thomson. Thomson’s instruments are fine ones in the best of the British tradition for mechanical and experimental science. Willie now judges that the particles are deflected on the basis of observations regularly made in Thomson’s lab. So now his judgments that the particles constituting the rays in the cathode tube are charged are in conformity with Achinstein’s account of evidence and, as applied to his epistemic situation, seem to produce justified true beliefs in him and anyone else in that situation. However, unbeknownst to his English colleagues, Willie has never overcome his Aristotelian training. So, in fact, the process in terms of which he judges that the particles are charged on the basis of appropriately acquired evidence that they are deflected, though in conformity with the requirements for veridical evidence, are those described by Aristotle for the skilled investigator, that is, intuitive induction. Aristotelian Willie sees by means of intellectual insight that it is the nature of charged particles in circumstances of the sort instantiated by cathode rays that they be deflected. The account of evidential inference that Achinstein offers seems to allow us to say that anyone one in Willie’s epistemic situation is justified in her belief that the particles are charged. Thus, it seems that Willie knows -- in the sense of having a justified true belief -- that the particles in the ray are charged on the basis of the evidence that they deflected. However an examination of the processes by which Willie makes his inference raises doubts about whether Willie does in fact know what he claims to know, even though all the Achinsteinian conditions for such knowledge are filled.

Achinstein considers objective epistemic probabilities to be normative. Sometimes norms are fulfilled and at other times they are not. Thomson and colleagues, having come to a point where these norms are in fact instantiated, can claim to be in norm fulfilling states. I contend that Willie can make a similar claim, even though the processes by which he comes to be in this state seem to be defective. Willie’s Aristotelian methodology allows him to conclude that the p (h/e) > 1/2. Indeed, it allows him to claim that it is near unity. Moreover, on Aristotelian criteria, the deflection in a magnetic field or electrical field is an essential property of a charged particle. As such the probability that there is an explanatory connection between h and e/ given h and e is unity, not just greater than 1/2. That is, given both h and e, there is a maximal explanatory connection between h (that the particles are charged) and e (that they are deflected, being, as they are, in an electro-magnetic field). Nevertheless, though Willie seems to know h and consequently to be justified in claiming h, he accomplishes this through a faulty process.

Though conforming to the Achinsteinian requirement’s on evidence, Willie is employing a discredited, Aristotelian method of evidential inference, even though it conforms to the Achinsteinian requirements concerning evidence. Thus, even though we concede that the Achinstein’s norms are correct and Willie’s method allows him to be in conformity with them, Willie has a problem. He is not using a proper method of evidential inference. He is not justified in claiming that the particles are charged because it was only an accident that his inference, based as it is on Aristotelian intuitive inductive procedures, fulfills Achinsteinian standards. We can fix things, if we consider something not required by Achinstein, the reliability of the processes by which Achinsteinian norms are achieved.

Suppose Willie undergoes retraining with Petra, a renowned, brilliant, inspiring and born-before her-time Achinsteinian. Not only does he continue to accept Achinsteinian criteria concerning high probability and explanatory relevance; but he is thoroughly apprised of the contingency of the relationship between evidence and hypothesis. Willie is enthusiastic about the inferential process that results in achieving the Achinsteinian criteria, though he remains unclear about that process’s nature. However, let us suppose that by hook or crook Willie now uses those processes in the inferences by means of which he correctly affirms that he has evidence that the particles in Thomson’s experiments are charged. And, of course, following Achinstein, anyone in Willie’s epistemic situation, is as Willie, justified in her belief that the particles are charged.

Nevertheless Willie’s enthusiasm portends a deeper and darker epistemic story. Suppose it comes out in The Epistemic National Inquirer that Willie would follow Petra no matter her epistemic commitments. Indeed, as psycho-epistemic analysis later lays bare, Willie’s past reveals multiple epistemic affairs. There was Tricia, the tarot card tart, Patsie, the numero uno numerologist and Pattie, the chancy creation science probabilist. Willie swore by all of these mentors of method at one time or another in his epistemic career. Applying each of their methods in turn during his epistemic escapades, lucky Willie performs proper evidential inferences that instantiate an adequate set of criteria concerning the evidential relation. That is, Willie performs a perfectly proper Achinsteinian evidential inference. Moreover, he swears that he has done so on the basis of dispositions to follow processes that are entirely adequate. Despite these facts his claim to justifiably hold h on the basis of e seems problematic. For in each case we have the following situation. Willie uses his Achinsteinian disposition for evidential inference to form a correct judgment about h on the basis of e. However, his second order disposition to apply his first order, entirely proper Achinsteinian disposition, appears to be only accidentally related to using his Achinsteinian disposition.

Willie’s epistemic affairs have produced second order epistemic dispositions that are unreliable in producing the type of first order disposition that reliably generates his evidential inference. We have arrived in this analysis at a point where we need a requirement of recursive reliability. We need a second order disposition that reliably generates the reliable first order disposition. The point of the example is not that accounting for Willie’s would-be epistemic success seems to depend on stopping the familiar infinite regress. Naturalistic epistemologists argue that as finite epistemic agents we must be satisfied with less ambitious epistemic goals than stopping infinite regresses. The regress to be examined is finite and will not proceed synchronically, but diachronically. Inevitably it will lead to consideration of the practices of epistemic instructors in the ontogeny and social learning history of the individual and to examination of evolutionary cognitive predecessors at the species level. More to the current point, is the conclusion that the evidential relation depends on the reliability of cognitive processes. Because it prescinds from the issue of reliable processes, Achinstein’s account of evidence does not provide an understanding for either epistemologist or scientist of the objective epistemic reality that he hopes to capture in his interpretation of the evidential relation as involving objective epistemic probability. At best, Achinstien’s account provides epistemic norms in terms of which reliable processes can be judged. But, it tells the philosopher and scientist little if anything about what processes are reliable ones. Indeed, it raises the further troubling question about how these norms can be established independently of a consideration of the kinds of cognitive processes that scientists can and do engage in. I develop this critique in a longer version of the paper.

Without consulting my scientist colleagues on the matter, I have argued that Achinstein’s analytically based supplementary analyses are of dubious assistance to scientists. Along with Achinstein, I take it that scientists are interested in having justified beliefs about the subject matter of their investigations and that they would find it helpful to have a better understanding of how and why it is that their practice of evidential inference is successful. I argue, however, that Achinstein’s analysis does not provide this kind of help. I make my case on the basis of two major points. First, I argue that Achinstein’s analysis fails to identify the processes of evidence acquisition and evidential inference to h, whether reliable or not. Thus, it provides the scientist with no indication of the processes that constitute the observations and inferences used to claim evidential justification for his hypothetical claims, let alone any indication of their reliability. This is so because Achinstein’s account of objective epistemic evidence prescinds from epistemic processes, their reliability and the sources of their reliability. It is a-psychologistic. Secondly, in the longer version of this paper, I argue that the epistemic norms embodied in Achinstein’s account of evidence do not adequately identify the epistemic agents to which they are to be applied. His abstract epistemic norms are independent of the capacities of epistemic subjects making them vulnerable to being impossible to achieve or irrelevant to scientific inferential capacities. Achinstein’s more concrete formulations, on the other hand, presuppose various capacities of epistemic agents, thus seemingly violating his conditions on objective epistemic probability.

My two critiques are not unrelated since the naturalistic epistemological approach that I advocate maintains that justification should be understood in terms of the reliability of epistemic processes of human epistemic agents whose in-built and acquired capacities set the limits for what can currently be done and aspired to. Finally, it should be remarked that the naturalistic epistemologist maintains that finding out about these processes and capacities is not a task that philosophers of science can accomplish alone. It will involve scientists who study our epistemic processes, for instance, cognitive ethologists, comparative psychologists, cognitive psychologists and scholars of the social study of science. This apparently circular process will in the end, the naturalistic epistemologist maintains, turn out to be beneficent for all parties, even to those who study cathode rays and electrons.