How confident does the history of science allow us to be about our current well-tested scientific theories, and why? The scientific realist thinks we are well within our rights to believe our best-tested theories, or some aspects of them, are approximately true.2 Ambitious arguments have been made to this effect, such as that over historical time our scientific theories are converging to the truth, that the retention of concepts and claims is evidence for this, and that there can be no (...) other serious explanation of the success of science than that its theories are approximately true. There is appeal in each of these ideas, but making such strong claims has tended to be hazardous, leaving us open to charges that many typical episodes in the history of science just do not fit the model. (See, e.g., Laudan 1981.) Arguing for a realist attitude via general claims – properties ascribed to sets of theories, trends we see in progressions of theories, and claimed links between general properties like success and truth that apply or fail to apply to any theory regardless of its content – is like arguing for or via a theory of science, which brings with it the obligation to defend that theory. I think a realist attitude toward particular scientific theories for which we have evidence can be maintained rationally without such a theory, even in the face of the pessimistic induction over the history of science. The starting point at which questions arise as to what we have a right to believe about our theories is one where we have theories and evidence for them, and we are involved in the activity of apportioning our belief in each particular theory or hypothesis in accord with the strength of the particular evidence.3 The devil’s advocate sees our innocence and tries his best to sow seeds of doubt. If our starting point is as I say, though, the innocent believer in particular theories does not have to play offense and propose sweeping views about science in general, but only to respond to the skeptic’s challenges; the burden of initial argument is on the skeptic.. (shrink)

This paper defends the naïve thesis that the method of experiment has per se an epistemic superiority over the method of computer simulation, a view that has been rejected by some philosophers writing about simulation, and whose grounds have been hard to pin down by its defenders. I further argue that this superiority does not come from the experiment’s object being materially similar to the target in the world that the investigator is trying to learn about, as both sides of (...) dispute over the epistemic superiority thesis have assumed. The superiority depends on features of the question and on a property of natural kinds that has been mistaken for material similarity. Seeing this requires holding other things equal in the comparison of the two methods, thereby exposing that, under the conditions that will be specified, the simulation is necessarily epistemically one step behind the corresponding experiment. Practical constraints like feasibility and morality mean that scientists do not often face an other-things- equal comparison when they choose between experiment and simulation. Nevertheless, I argue, awareness of this superiority and of the general distinction between experiment and simulation is important for maintaining motivation to seek answers to new questions. (shrink)

When we get evidence that tells us our belief-forming mechanisms may not be reliable this presents a thorny set of questions about whether and how to revise our original belief. This article analyzes aspects of the problem and a variety of approaches to its solution.

I develop a general framework with a rationality constraint that shows how coherently to represent and deal with second-order information about one's own judgmental reliability. It is a rejection of and generalization away from the typical Bayesian requirements of unconditional judgmental self-respect and perfect knowledge of one's own beliefs, and is defended by appeal to the Principal Principle. This yields consequences about maintaining unity of the self, about symmetries and asymmetries between the first- and third-person, and a principled way of (...) knowing when to stop second-guessing oneself. Peer disagreement is treated as a special case where one doubts oneself because of news that an intellectual equal disagrees. This framework, and variants of it, imply that the typically stated belief that an equally reliably peer disagrees is incoherent, and thus that pure rationality constraints without further substantive information cannot give an answer as to what to do. The framework also shows that treating both ourselves and others as thermometers in the disagreement situation does not imply the Equal Weight view. (shrink)

Many who think that naked statistical evidence alone is inadequate for a trial verdict think that use of probability is the problem, and something other than probability – knowledge, full belief, causal relations – is the solution. I argue that the issue of whether naked statistical evidence is weak can be formulated within the probabilistic idiom, as the question whether likelihoods or only posterior probabilities should be taken into account in our judgment of a case. This question also identifies a (...) major difference between the Process Reliabilist and Probabilistic Tracking views of knowledge and other concepts. Though both are externalist, and probabilistic, epistemic theories, Tracking does and Process Reliabilism does not put conditions on likelihoods. So Tracking implies that a naked statistic is not adequate evidence about an individual, and does not yield knowledge, whereas the Reliabilist thinks it gives justified belief and knowledge. Not only does the Tracking view imply that naked statistical evidence is insufficient for a verdict, but it gives us resources to explain why, in terms of risk and the special conditions of a trial. (shrink)

This paper argues that if knowledge is defined in terms of probabilistic tracking then the benefits of epistemic closure follow without the addition of a closure clause. (This updates my definition of knowledge in Tracking Truth 2005.) An important condition on this result is found in "Closure Failure and Scientific Inquiry" (2017).

There is a widespread view that in order to be rational we must mostly know what we believe. In the probabilistic tradition this is defended by arguments that a person who failed to have this knowledge would be vulnerable to sure loss, or probabilistically incoherent. I argue that even gross failure to know one's own beliefs need not expose one to sure loss, and does not if we follow a generalization of the standard bridge principle between first-order and second-order beliefs. (...) This makes it possible for a subject to use probabilistic decision theory to manage in a rational way cases of potential failure of this self-knowledge, as we find in implicit bias. Through such cases I argue that it is possible for uncertainty about what our beliefs are to be not only rationally permissible but advantageous. (shrink)

It is received wisdom that the skeptic has a devastating line of argument in the following. You probably think, he says, that you know that you have hands. But if you knew that you had hands, then you would also know that you were not a brain in a vat, a brain suspended in fluid with electrodes feeding you perfectly coordinated impressions that are generated by a supercomputer, of a world that looks and moves just like this one. You would (...) know you weren’t in this state if you knew you had hands, since having hands implies you are no brain in a vat. You obviously don’t know you’re not a brain in a vat, though—you have no evidence that would distinguish that state from the normal one you think you’re in. Therefore, by modus tollens, you don’t know you have hands. At least, the skeptic has a devastating argument, it is thought, if we grant him closure of knowledge under known implication, which many of us are inclined to do: roughly, if you know p, and you know that p implies q, then you know q. (shrink)

In the aftermath of Gettier’s examples, knowledge came to be thought of as what you would have if in addition to a true belief and your favorite epistemic goody, such as justifiedness, you also were ungettiered, and the theory of knowledge was frequently equated, especially by its detractors, with the project of pinning down that extra bit. It would follow that knowledge contributes something distinctive that makes it indispensable in our pantheon of epistemic concepts only if avoiding gettierization has a (...) value that can be explained without presupposing the value of knowledge. Tracking-type knowledge has a value that no other logically possible conditions on true belief does. As an Evolutionarily Stable Strategy it preserves appropriate belief states through time and changing circumstances. If we characterize gettierization through the concept of relevance matching, then we see that avoiding gettierization has a value independent of that of knowledge, namely, understanding, and that it is unnecessary to add a clause to the tracking conditions to make them suppress gettierization directly, though fallibly. The bright line of value is between gettierization avoidance and understanding on the one hand and knowledge on the other, and so should be the bright line defining concepts. The concept of relevance matching is key to a definition of what it is to understand why p is true, as opposed merely to knowing that p is true. Perfect tracking implies perfect relevance matching, so knowledge and understanding are intimately connected but understanding also requires that one own states that accomplish the relevance matching rather than achieving it vicariously. The theory of understanding based on relevance matching implies that understanding requires appreciation of not only p but its connections to other matters, and explains how it is possible to know that p is true without understanding why. The view implies that understanding is literally simulation, and is suggestive about understanding other minds. (shrink)

Abstract: Knowledge requires more than mere true belief, and we also tend to think it is more valuable. I explain the added value that knowledge contributes if its extra ingredient beyond true belief is tracking . I show that the tracking conditions are the unique conditions on knowledge that achieve for those who fulfill them a strict Nash Equilibrium and an Evolutionarily Stable Strategy in what I call the True Belief Game. The added value of these properties, intuitively, includes preparedness (...) and an expectation of survival advantage. On this view knowledge is valuable not because knowledge persists but because it makes the bearer more likely to maintain an appropriate belief state—possibly nonbelief—through time and changing circumstances. When Socrates concluded that knowledge of the road to Larissa was no more valuable than true belief for the purpose of getting to Larissa, he did not take into account that one might want to be prepared for a possible meeting with a misleading sophist along the way, or for the possibility of road work. (shrink)

In the last three decades several cosmological principles and styles of reasoning termed 'anthropic' have been introduced into physics research and popular accounts of the universe and human beings' place in it. I discuss the circumstances of 'fine tuning' that have motivated this development, and what is common among the principles. I examine the two primary principles, and find a sharp difference between these 'Weak' and 'Strong' varieties: contrary to the view of the progenitors that all anthropic principles represent a (...) departure from Copernicanism in cosmology, the Weak Anthropic Principle is an instance of Copernicanism. It has close affinities with the step of Copernicus that Immanuel Kant took himself to be imitating in the 'critical' turn that gave rise to the Critique of Pure Reason. I conclude that the fact that a way of going about natural science mentions human beings is not sufficient reason to think that it is a subjective approach; in fact, it may need to mention human beings in order to be objective. (shrink)

Less discussed than Hume’s skepticism about what grounds there could be for projecting empirical hypotheses is his concern with a skeptical regress that he thought threatened to extinguish any belief when we reflect that our reasoning is not perfect. The root of the problem is the fact that a reflection about our reasoning is itself a piece of reasoning. If each reflection is negative and undermining, does that not give us a diminution of our original belief to nothing? It requires (...) much attention to detail, we argue, to determine whether or not there is a skeptical problem in this neighborhood. Consider that if we subsequently doubt a doubt we had about our reasoning, that would suggest a restoration of some of the force of our original belief. We would then have instead of runaway skepticism an alternating sequence of pieces of skeptical reasoning that cancel each others’ effects on our justification in the original proposition, at least to some degree. We will argue that the outcome of the sequence of reflections Hume is imagining depends on information about a given case that is not known a priori. We conclude this from the fact that under three precise, explanatory, and viable contemporary reconstructions of what this kind of reasoning about reasoning could be like and how it has the potential to affect our original beliefs, a belief-extinguishing regress is not automatic or necessary. (shrink)

The transferability problem—whether the results of an experiment will transfer to a treatment population—affects not only Randomized Controlled Trials but any type of study. The problem for any given type of study can also, potentially, be addressed to some degree through many different types of study. The transferability problem for a given RCT can be investigated further through another RCT, but the variables to use in the further experiment must be discovered. This suggests we could do better on the epistemological (...) problem of transferability by promoting, in the repeated process of formulating public health guidelines, feedback loops of information from the implementation setting back to researchers who are defining new studies. (shrink)

There is much disagreement about how extensive a role theoretical mind-reading, behavior-reading, and simulation each have and need to have in our knowing and understanding other minds, and how each method is implemented in the brain, but less discussion of the epistemological question what it is about the products of these methods that makes them count as knowledge or understanding. This question has become especially salient recently as some have the intuition that mirror neurons can bring understanding of another's action (...) despite involving no higher-order processing, whereas most epistemologists writing about understanding think that it requires reflective access to one's grounds, which is closer to the intuitions of other commenters on mirror neurons. I offer a definition of what it is that makes something understanding that is compelling independently of the context of cognition of other minds, and use it to show two things: 1) that theoretical mind-reading and simulation bring understanding in virtue of the same epistemic feature, and 2) why the kind of motor representation without propositional attitudes that is done by mirror neurons is sufficient for action understanding. I further suggest that more attention should be paid to the potential disadvantages of a simulative method of knowing. Though it can be more efficient in some cases, it can also bring vulnerability, wear and tear on one's personal equipment, and unintended mimicry. (shrink)

Deduction is important to scientific inquiry because it can extend knowledge efficiently, bypassing the need to investigate everything directly. The existence of closure failure—where one knows the premises and that the premises imply the conclusion but nevertheless does not know the conclusion—is a problem because it threatens this usage. It means that we cannot trust deduction for gaining new knowledge unless we can identify such cases ahead of time so as to avoid them. For philosophically engineered examples we have “inner (...) alarm bells” to detect closure failure, but in scientific investigation we would want to use deduction for extension of our knowledge to matters we don’t already know that we couldn’t know. Through a quantitative treatment of how fast probabilistic sensitivity is lost over steps of deduction, I identify a condition that guarantees that the growth of potential error will be gradual; thus, dramatic closure failure is avoided. Whether the condition is fulfilled is often obvious, but sometimes it requires substantive investigation. I illustrate that not only safe deduction but the discovery of dramatic closure failures can lead to scientific advances. (shrink)

Health anxiety is, among other things, a response to a universal epistemological problem about whether changes in one’s body indicate serious illness, a problem that grows more challenging to the individual with age and with every advance in medical science, detection, and treatment. There is growing evidence that dysfunctional metacognitive beliefs – beliefs about thinking – are the driving factor, with dysfunctional substantive beliefs about the probability of illness a side‐effect, and that Metacognitive Therapy (MCT) is more effective than Cognitive (...) Behavioral Therapy (CBT). However, hypochondria is distinct from other forms of anxiety, I argue, in ways that make some reality‐checking techniques of CBT and MCT of limited usefulness. I propose a Re‐Calibration Technique (RCT) that complements these therapies by focusing on a metacognitive belief that has not been studied: the patient’s presumption of his own personal reliability in judging symptoms, an assumption exposed every time he disagrees with a doctor. I propose a technique whereby a patient keeps a long‐term register of every episode of alarm about symptoms and its resolution, possibly years later. When healthcare‐seeking impulses arise the patient then uses his own track record to re‐calibrate his confidence that medical attention is needed. The new technique allows one to improve self‐judgment about whether one has an illness or not by improving self‐knowledge of one’s own reliability. (shrink)

One of the most common criticisms one hears of the idea of granting a legitimate role for social values in theory choice in science is that it just doesn’t make sense to regard social preferences as relevant to the truth or to the way things are. “What is at issue,” wrote Susan Haack, is “whether it is possible to derive an ‘is’ from an ‘ought.’ ” One can see that this is not possible, she concludes, “as soon as one expresses (...) it plainly: that propositions about what states of affairs are desirable or deplorable could be evidence that things are, or are not, so” (Haack 1993a, 35, emphasis in original). The purpose of this chapter is not to determine whether this widespread view is correct, but rather to show that even if we grant it (which I do), we may still consistently believe that social values have a legitimate role in theory choice in science. I will defend this conclusion by outlining a view about social values and theory choice that is available to a Constructive Empiricist anti-realist, but not to a realist. (shrink)

Why should we make our beliefs consistent or, more generally, probabilistically coherent? That it will prevent sure losses in betting and that it will maximize one’s chances of having accurate beliefs are popular answers. However, these justifications are self-centered, focused on the consequences of our coherence for ourselves. I argue that incoherence has consequences for others because it is liable to mislead others, to false beliefs about one’s beliefs and false expectations about one’s behavior. I argue that the moral obligation (...) of truthfulness thus constrains us to either conform to the logic our audience assumes we use, educate them in a new logic, or give notice that one will do neither. This does not show that probabilistic coherence is uniquely suited to making truthful communication possible, but I argue that classical probabilistic coherence is superior to other logics for maximizing efficiency in communication. (shrink)

This develops a framework for second-order conditionalization on statements about one's own epistemic reliability. It is the generalization of the framework of "Second-Guessing" (2009) to the case where the subject is uncertain about her reliability. See also "Epistemic Self-Doubt" (2017).

Over the centuries since the modern scientific revolution that started with Copernicus, Galileo, Kepler, and Newton, two things have changed that have required reorientation of our assumptions and re-education of our reflexes. First, we have learned that even the very best science is fallible; eminently successful theories investigated and supported through the best methods, and by the best evidence available, might be not just incomplete but wrong. That is, it is possible to have a justified belief that is false.

It is widely accepted that in fallible reasoning potential error necessarily increases with every additional step, whether inferences or premises, because it grows in the same way that the probability of a lengthening conjunction shrinks. As it stands, this is disappointing but, I will argue, not out of keeping with our experience. However, consulting an expert, proof-checking, constructing gap-free proofs, and gathering more evidence for a given conclusion also add more steps, and we think these actions have the potential to (...) improve our reliability or justifiedness. Thus, the received wisdom about the growth of error implies a skepticism about the possibility of improving our reliability and level of justification through effort. Paradoxically, and even more implausibly, taking steps to decrease your potential error necessarily increases it. I will argue that the self-help steps listed here are of a distinctive type, involving composition rather than conjunction. Error grows differently over composition than over conjunction, I argue, and this dissolves the apparent paradox. (shrink)

On analogy with testimony, I define a notion of a scientific theory’s lacking or having candor, in a testing situation, according to whether the theory under test is probabilistically relevant to the processes in the test procedures, and thereby to the reliability of test outcomes. I argue that this property identifies what is distinctive about those theories that Karl Popper denounced as exhibiting “reinforced dogmatism” through their self-protective behavior (e.g., psychoanalysis, Hegelianism, Marxism). I explore whether lack of candor interferes with (...) the testing of theories, and conclude that (1) our default attitude toward theories that lack candor in a given test should be suspicion, but (2) the circumstance that a theory lacks candor in a testing situation does not preclude obtaining independent evidence for the auxiliary assumptions to which the theory is probabilistically relevant, and thereby eliminating the problem that lack of candor creates. Thus, Popper was right to think that lack of candor is a bad thing, but wrong to conclude that candor is a criterion of the scientificity of a theory. Seeing this requires recognition of some differences between intuitive relevance and probabilistic relevance, and proper appreciation of the notion of screening off and of the fact that probabilistic relevance is not transitive. (shrink)

Disclaimer: This work on marginal science has not been funded by the National Science Foundation.

This is a very short textbook on probabilistic reasoning, expected utility decision-making, cognitive biases, and self-correction, especially in application to medical examples. It also includes a chapter on concepts of health.