Natural Supervenience Base For Normative Epistemic Properties example essay topic

1. INTRODUCTION At the most general level, epistemological naturalism can be characterised as the view that epistemology, including its normative parts, should be reconciled with, and even draw upon, science. This, of course, is vague, at best. Thus stated, epistemological naturalism does not qualify as a position, and hardly even as a programme, and it certainly does not permit any more detailed discussion. Part of what I intend in the present essay is to state and seperate various views falling under the broad heading of epistemological naturalism. This will serve as a basis for a discussion of certain objections made to the most controversial part of the naturalistic programme in epistemology.

The plan of the essay is as follows. For the purpose of later discussion, I need to indicate very roughly what I take the content of certain epistemic concepts to be. In particular, I shall indicate what I mean by the property of justifiedness and the term 'epistemic norm'. This is what I do in section 2. In section 3, I seperate two broad questions that naturalistic theories address, what I call the location problem and the justification problem. The location problem concerns how to find an acceptable place for what appears to be normative epistemological properties.

The justification problem, on the other hand, concerns how to justify our ways of forming beliefs. I shall argue that epistemological naturalism becomes controversial (and exciting) primarily when seen as a response to the justification problem, and when taking the form of normative naturalism. Roughly, normative naturalism is the view that we can justify our ways of forming beliefs (our epistemic norms) by appealing to empirical facts (in a broad sense) about humans and the rest of nature, where our knowledge of these facts is not in turn thought to rest on some a priori foundation. Normative naturalism is the position which will remain at the center of discussion in the rest of the essay. In section 4, I review the main arguments for epistemological naturalism, and discuss to what extent they actually support normative naturalism.

I conclude that the many arguments normally brought forward in support of epistemological naturalism are surprisingly weak reasons for accepting normative naturalism. The best reason on offer stems from a denial of both scepticism and the possibility of providing an a priori justification of our epistemic norms, but not even that is a strong reason. I shall, however, argue that we need not be that concerned about offering reasons for normative naturalism. The crucial issue is whether the position is coherent at all.

What I have in mind here is primarily the question if normative naturalism leads to a vicious circularity, as one might suspect of a position which holds that we can display the cogency of our epistemic norms by appealing to empirical facts, which we have obtained through reliance on those very norms. I discuss this circularity objection, as I shall call it, in section 5, where I also turn Quine's and Kitcher's attempt to reply to that charge. In section 6, I present in brief outline what I take to be the best answer to the circularity charge. 2. EPISTEMIC NORMS AND OTHER EPISTEMIC CONCEPTS The general issue in the naturalization of epistemology is to provide an understanding of the relation between an empirical description of our ways of forming beliefs and the normative parts of epistemology. To pave the way for a discussion of normative naturalism, it will prove useful with a rough statement of what I take the content of certain crucial epistemic notions to be.

One central epistemic notion is the property of justifiedness (or the concept of justifiedness). We might assume that the property of justifiedness is relational, such that a belief P is justified for S in context C. When a belief is justified, the subject holding the belief is in some sense permitted in doing so. Many writers, of course, also take justifiedness of a belief, together with truth, to be sufficient for knowledge, but some disagree. For the purpose of the present discussion we can by and large remain neutral on this further issue. What is important, however, is that the concept of justifiedness is normative in the sense that saying that a belief is or is not justified essentially involves an evaluation of a belief and an agent in a particular context. Now, while the property of justifiedness is perhaps the most central, at least in post-get tier epistemology, it is actually easier, I believe, to understand the most interesting parts of the naturalistic programme in epistemology if we also introduce the concept of an epistemic norm.

I take an epistemic norm to be, roughly, a rule or directive telling us (in some sense) what to do if we want to achieve a certain epistemic goal. Rules such as those involved in induction, or inferences to best explanations are (vague) epistemic norms. So are rules or directives telling us to rely on sense perception under certain familiar circumstances, and for certain familiar purposes. This of course only serves to give a very preliminary idea of the concept of an epistemic norm, as I intend that notion to be understood. We need to make a rough distinction between epistemic norms which are successful, and epistemic norms which are not successful. This I take to be the distinction between epistemic directives which actually take those using or relying upon them to where they want to go, epistemically speaking, and directives which do not achieve this.

That is, epistemic norms might be more or less successful relative to some epistemic goal. Truth, obviously, is one epistemic goal, but it need not be the only one. Other goals might be that our beliefs or theories display simplicity, explanatory power, testability, or coherence of various sorts. For the purpose of the present discussion I shall assume that truth is one important epistemic goal of ours. In that case, success fulness of an epistemic norm comes down to some sort of reliability. I shall not here try to state or defend any detailed account of reliability, but simply take this notion for granted.

Epistemic norms are normative at least in the sense that if a subject accepts a certain epistemic goal, there is, other things being equal, a sort of epistemically correct, or right, or good behaviour to following the rule, and a sort of epistemic wrongdoing in not following it. Obviously, part of what we want from epistemology, is identification of reliable epistemic norms. There are at least two reasons why this task is crucial. Suppose we accept an externalist theory of justification, according to which a belief is justified only if it arises from a reliable belief forming practice.

In that case, in order to identify which beliefs are justified, we need to know which ways of forming beliefs are reliable. Or, more generally, we need to know which epistemic norms are reliable. The second reason goes deeper, however. Suppose we accept some form of internalist theory of justification, which does not require 'external' reliability for justifiedness. In that case, we can decide if some belief is justified, without assuming anything about the reliability of the way in which it was formed. In general, we can decide if our beliefs are justified without assuming that our epistemic norms are reliable.

Now, even if we accept such a theory of justification, it still seems to be a crucial question whether or not the epistemic norms we actually rely upon, or could rely upon, are in fact reliable. This point is nicely illustrated by the the defence of induction once offered by P.F. Strawson. Roughly, Strawson's claim was that inductive reasoning is epistemically rational simply because this is part of our concept of epistemic rationality. In our present terminology, this could be put as the claim that beliefs formed on the basis of induction count as justified, since this is simply part of our concept of justifiedness. This theory is internalist in the sense that it does not require 'external' reliability for justifiedness. Hence, one might be justified in inductively formed beliefs, even when one reasons inductively in an environment where induction is not reliable at all.

Precisely this is, I believe, the reason why one should not be satisfied with Strawson's defence of induction. It is simply not enough to note that inductive reasoning for conceptual reasons counts as epistemically rational, or that inductively formed beliefs count as justified. Although Strawson's conceptual point might be very reasonable, one cannot escape the crucial question if induction really is reliable under some specified set of circumstances. Or, more generally, a vital question in epistemology is if our epistemic norms are in fact reliable. This question remains central, whether or not we accept an externalist theory of justification. Thus, one task in epistemology is to justify our epistemic norms.

My brief discussion of the property of justifiedness and the notion of an epistemic norm served to illustrate two points. First, I have roughly indicated the sort of normativity that relates to epistemic concepts. Second, I have argued that a central task in epistemology is to justify our epistemic norms. Return now to epistemological naturalism. Various forms of naturalism address the two very different questions arising from this. The first is this: how can normative properties have a place in a naturalistic world, apparently devoid of normativity?

This is what I shall call the location problem. The second is this: how can we provide a justification of our epistemic norms? This is what I shall call the justification problem. 3. KINDS OF EPISTEMOLOGICAL NATURALISM I shall now state and comment briefly on various forms of epistemological naturalism. One family of views address the location problem, the question of how to reconcile what appears to be normative epistemological properties with a natural world seemingly devoid of normativity, or normative properties.

One view is Epistemic non-cognitivism. Despite appearances, there are no normative epistemic properties. And our attributions of such properties are to be understood as expressions of attitudes of approval or dissent, and are not to be taken as having a propositional content. Epistemic non-cognitivism would provide a simple answer to the location problem, and it is, I believe, fairly plausible.

As is readily seen, it is fully compatible with there being in principle a naturalistic description of the properties which we are evaluating positively in our epistemic evaluations, and it might be a worthwhile research programme to provide these descriptions. But we need not be epistemic non-cognitivists to remain a recognisable sort of epistemological naturalists. A general and widely accepted view on the relation between epistemic normativity and the natural world is this: Epistemic supervenience For any normative epistemic property E, and any agent x, if x has E, there is a natural property N, such that x has N, and necessarily, whatever agent has N also has E. This view has been proposed by such writers as James van Cleve, Ernest Sosa, Alvin Goldman, and Jaegwon Kim. Unlike epistemic non-cognitivism, epistemic supervenience acknowledges the existence of normative epistemological properties, and it holds that our ascriptions of normative epistemic properties are not purely non-cognitive.

The normative parts of these ascriptions are descriptive - they aim to describe normative epistemic properties. Subscription to epistemic supervenience does not commit to the view that the supervenience base for any particular instantiation of a normative epistemic property can be described in any great detail. Nor does it imply that we can state simple and general rules or principles which indicate the natural supervenience base for normative epistemic properties. For these reasons, epistemic supervenience does not imply that normative epistemic properties can be reduced to natural properties. But one can imagine proponents of a form of reductionism: Epistemic reductionism Normative epistemic properties are reducible to natural properties. By 'reduction' and it cognates is here meant a translation of sentences containing normative epistemic expression to sentences containing only terms referring to natural properties in their place, without remainder.

Alternatively, one might by 'reduction' mean the establishment of various property-identi es. Reductionism of many different sorts have fallen into disrepute recently. One reaction to this is to hold that the non-r educability of normative epistemic terms (or properties) indicate that they do not refer to existing properties (or exist). This is to accept Epistemic eliminativism Since normative epistemic properties cannot be reduced to normative properties, they should be eliminated. It is widely acknowledged that the move from the failure of reduction to elimination is premature. Reduction of normative epistemic terms might not be a viable option, even in principle, but we need not for that reason eliminate normative epistemic terms, since we can settle for non-cognitivism or supervenience.

I shall here simply assume that we can leave aside reduction and elimination. If epistemic non-cognitivism or epistemic supervenience were all the proponents of naturalised epistemology had in mind, the view would hardly be controversial. At most, there would be an issue about which of these to accept. But obviously, naturalised epistemology is controversial and covers controversial views to which I now turn. The controversial parts of (or versions of) epistemological naturalism concern the justification problem, the problem of how to show that our epistemic norms are reliable. There are many different naturalistic stances which one might take to the justification problem, and I shall now seperate and comment on the most important of these.

Let me start by quoting a famous passage from Quine's influential 'Epistemology Naturalized': 'Epistemology, or something like it, simply falls into place as a chapter of psychology and hence of natural science. It studies a natural phenomenon, viz., a physical human subject. This human subject is accorded a certain experimentally controlled input - certain patterns of irradiation in assorted frequencies, for instance - and in the fullness of time the subject delivers as output a description of the three-dimensional external world and its history. The relation between the meager input and the torrential output is a relation that we are promted to study for somewhat the same reasons that always promted epistemology; namely, in order to see how evidence relates to theory, and in what ways one's theory of nature transcends any available evidence. ' This appears to be the following recommendation: Quine's eliminative naturalism Epistemology, as it has traditionally been conceived, ought to be abandoned in favour of a scientific description of how our cognitive processes actually work. As has been noted many times, it is difficult to see how Quine's eliminative naturalism could be a plausible successor discipline to traditional epistemology.

One ambition in epistemology has been to decide if our everyday and scientific beliefs about the world are justified. It is obviously very hard to understand how Quine's eliminative naturalism could be a new way of achieving that same goal. Another goal in epistemology has been to deliver an assessment of our epistemic norms, i.e. to address the justification problem. Again, eliminative naturalism could hardly be an improved means of arriving at that end; it merely aims at a description of how we in fact proceed when forming beliefs. For these reasons, we might understand Quine's eliminative naturalism as a sceptical answer to the justification problem. All we can do is to describe how we actually arrive at out beliefs, scientia and ordinary.

There is no way in which we can provide a non-question begging justification of our ways of forming beliefs. It is widely appreciated that Quine's explicitly stated arguments, as they appear in 'Epistemology Naturalized', are quite insufficient to justify his eliminative naturalism. I shall later return briefly to some of Quine's arguments. Before doing so, let us turn to a non-sceptical version of epistemological naturalism: Normative naturalism Epistemic norms can be justified by appeal solely to a posteriori theories about the world. This is true for epistemic norms like those involved in our use of induction and inferences to best explanations, and even for the most basic epistemic norms, such as those embodied in empiricism.

The crucial point in normative naturalism, of course, is that we have to do with a posteriori justification of even our most basic epistemic norms, whereas many have held that in so far as basic epistemic norms can be justified at all, they can be justified only a priori. Thus, the opposition to normative naturalism is, on the one hand, the non-sceptic a priori st, as one might call him, which is someone who thinks that we can on a purely a priori basis show our epistemic norms to be reliable. On the other hand, there is the sceptic, who thinks that we cannot show our basic epistemic norms to be reliable. This sort of sceptic might either say that none of our epistemic norms lends themselves to any a priori defence, either because there are not a priori justifiable beliefs at all, or because our epistemic norms, in particular, cannot be justified a priori. Another sort of sceptic might say, that while there are a priori reasons in favour of some norms regulating the formation of empirical beliefs, this merely postpones the problem.

A priori reasoning itself involves epistemic norms, which obviously cannot be justified a priori (this sceptic claims). Hence, even the a priori defence of the other epistemic norms fails. I shall later distinguish various importantly different versions of the aspirations in normative naturalism. Normative naturalism is, I believe, the really controversial view in the naturalist camp.

Almost everyone accepts some version of epistemological non-cognitivism or epistemological supervenience, and hardly anyone accepts Quine's eliminative naturalism. It should be obvious how different Quine's eliminative naturalism is from normative naturalism. In the former, there is no pretensions whatsoever of providing any sort of justification of our epistemic norms, whereas this sort of ambition is at center stage in normative naturalism. Notice also how very different epistemic non-cognitivism and epistemic supervenience are from normative naturalism. One can accept either of the two former views, and still reject normative naturalism, and no doubt many would be inclined to do so. Conversely, one could hold that normative epistemic properties are non-natural properties, which do not even supervene on natural properties, while at the same time believing that we can show our epistemic norms to be successful only by appealing to scientific theories of the world.

In general, therefore, there is no tight connection between the sort of answer one prefers concerning the location problem, and the sort of answer one prefers to the justification problem. Epistemic non-cognitivism and epistemic supervenience compete neither with non-sceptic a prior ism, normative naturalism, or with scepticism. Their oppositions are, respectively, cognitivism and some sort of view which holds that normative epistemic properties are real, but not natural, and not supervenient. Below I shall discuss the merits of various forms of normative naturalism in much more detail. In particular, I shall consider the charge that the whole endeavor falters because viciously circular.

But first, I shall comment on what seems to be two different versions of naturalised epistemology. In the movement sometimes known as 'the historical turn' in the philosophy of science, one finds the methodological claim that when devising philosophical theories about epistemic norms and other epistemic concepts philosophers of science ought to use empirical findings, in particular historical studies, about how scientists do in fact reason. The idea is, in the words of Laudan, that '... the failure of great scientists to have made choices in conformity with the recommendations issuing from any methodology of science stands as a dramatic reductio of that methodology. They [the Historicist's, as Laudan calls them] say the failure of existing methodologies to allow a reconstruction of the action of former scientists as rational decisively shows the inadequacy of those methodologies.

' Or, in the present terminology, if philosophers of science somehow think they can devise theories of reliable epistemic norms or epistemic rationality in splendid isolation from considerations of what scientists actually do in their scientific reasoning, they are utterly mistaken. On the contrary, it should be one of the fixed points in our construction of theories of epistemic rationality that at least the most eminent scientists come out a rational. We should therefore conduct empirical investigations as to how scientists are in fact reasoning, or have been reasoning. How precisely should we understand the historical turn? What can historical studies tell us about epistemic norms, or anything else of interest to epistemology? I believe that the historical turn is best understood as either a version of epistemic supervenience, or as a version of normative naturalism.

Taken as a version of epistemic supervenience, the m ethological advice in the historical turn simply comes down to the following. We are intuitively strongly inclined to say that the scientific beliefs of the best scientists are justified, that the epistemic norms they employ are successful, and that their reasoning is rational. We also think that there must be a natural supervenience basis for these epistemic properties, and empirical investigation of how scientists actually reason will be indispensable in elucidating this supervenience basis (at least if it is to be described beyond a certain level of detail). There is no doubt that this sort of investigation might be worthwhile.

But this sort of investigation only tell us more about what counts as justified belief, or as good epistemic norms, given the concepts we have. It does not as such tell us which epistemic norms are in fact reliable. The point may be put in this way. We might count some pattern of scientific reasoning as rational, and decide to elucidate this pattern in more detail. Ultimately, we can, perhaps at least, describe the natural properties that determines each individual instantiation of the pattern.

However, this is not the sort of investigation which will tell us if this pattern of reasoning is in fact truth conducive, that is, whether or not it embodies reliable epistemic norms. We might, however, understand the historical turn in a different way. We might take empirical investigation, including historical studies, of scientific reasoning to be a way of discovering which epistemic norms are in fact reliable and which are not. In the end, we might in this way provide a kind of justification of certain epistemic norms.

This way of proceeding, obviously, is an instance of normative naturalism. Empirical theories about how scientists have in fact been reasoning, and observations of the outcome, are used to corroborate epistemic norms, to show that they are in fact conducive to truth. Of course, when we understand the historical turn in this way, we can see that there is no reason in principle to limit our inquiries into how scientists in fact reason, or have been reasoning. It might turn out that epistemic norms can be evaluated for their conduciveness to specified goals in other ways, i.e. by including theories from cognitive science and our general understanding of what the world is like.

History of science is just one way of showing epistemic norms to be successful. Thus, what might have seemed to be fourth kind of epistemological naturalism in fact turns out to be either a version of epistemic supervenience, or of normative naturalism. Let me finally turn to P.F. Strawson, who has also defended a version of epistemological naturalism, which he finds proposed also by Hume and Wittgenstein: 'They [Hume and Wittgenstein] have in common the view that our "beliefs" in the existence of body and, to speak roughly, in the general reliability of induction are not grounded beliefs and at the same time are not open to serious doubt. They are, one might say, outside our critical and rational competence in the sense that they define, or help to define, the area in which that competence is exercised.

' One might think here that Strawson says that one is justified in making the presuppositions necessary for some inquiry - the fact that they are beyond reach of 'critical and rational competence' does not mean that they are epistemically unjustified, but rather that they are justified in some different way. But this is not Strawson's view: 'It is to be remembered that the point has been, not to offer a rational justification of the belief in external objects and other minds or of the practice of induction, but to represent skeptical arguments and rational counter-arguments as equally idle - not senseless, but idle - since what we have here are original, natural inescapable commitments which we neither choose nor could give up. ' This view is compatible with non-scepticism, of course. But its most natural motivation is the concession to the sceptic that our ordinary ways of forming beliefs, and the epistemic norms we usually rely upon, cannot be shown to be reliable. Hence, like was the case with Quine's eliminative naturalism, we can interpret Strawson's version of epistemological naturalism as accepting certain forms of scepticism with respect to the justification problem. Unlike Quine, however, Strawson does not suggest that science can provide some form of substitute for what we want in epistemology.

Rather he says that we might as well neglect the whole issue, since whatever we say about it, it will not make any practical difference. Obviously, this is very different from normative naturalism, the pretensions of which is to provide some form of justification of our epistemic norms. To conclude, let me summarise the most important points of this and the preceeding section. One family of naturalistic views concerns the nature of the relation between epistemic (and presumably normative) properties and natural (and presumably non-normative) properties.

There is a well-known array of positions available here, whose common denominator is the denial of the view that there are irreducibly non-natural epistemological properties which exist independently of natural properties. The naturalist will here presumably accept epistemic non-cognitivism or epistemic supervenience, whereas epistemic reductionism and epistemic eliminativism seem unmotivated. A quite different family of naturalistic views concerns the ways in which we can justify our epistemic norms. What these views have in common is that they hold that no a priori justification of any important segment of our epistemic norms is forthcoming.

There is a sceptic and a non-sceptic reaction to this. The sceptic reaction can be distilled from Quine, and consists of the view that we should settle for a mere description of our belief forming practices. A sceptic reaction can also be associated with Strawson (and Hume and Wittgenstein). The tenor of Strawson's view is the concession that a kind of scepticism is true, but that this doesn't really matter since our nature by and large compels us to arrive at our beliefs in certain ways, whether we can justify these ways or not. Normative naturalism, however, is a non-sceptical reaction to the the lack of a priori justification of our epistemic norms.

It claims that we can, in some sense, provide a justification of our epistemic norms by appealing to scientific theories of the world. 4. WHY NORMATIVE NATURALISM? I shall now turn to a discussion of normative naturalism. Why should we accept this view? I shall here turn to a brief review of Quine's arguments for his version of naturalised epistemology.

This might at first seem surprising, since I said above that Quine favoured eliminative naturalism, not normative naturalism. However, despite the fact that Quine explicitly subscribes to eliminative naturalism, there is also some indication that Quine's settled view is not eliminative naturalism, but normative naturalism. So let us briefly review the main arguments that Quine offers, and see to what extent they support normative naturalism. It seems clear that they do not. In 'Epistemology Naturalised' Quine offers at least these distinct lines of arguments for his brand of naturalism, whatever it is. One is the appeal to the failure of phenomenalistic reductionism, primarily as this programme was pursued by Carnap.

The sort of reductions that Carnap attempted are not to be had, according to Quine. The other main argument is the failure of the corresponding programme in epistemological foundationalist. There is, Quine surmises, no prospect of deducing science from observation, and apparently, this moves Quine to say that since deductive certainty is not to be had, we might as well turn to a scientific description of belief formation. Of course, this is far too swift.

There are many other options which Quine simply ignores, e.g. explicating justification in terms of fallible but nevertheless epistemologically basic beliefs, accepting non-deductive inferential relations, moving from foundationalist to coherent ist theories of justification, and so on. Such views have been discussed extensively in the literature, and there no reason to say more about them here. The observation that a particular programme in epistemology fails is only a weak consideration in favour of any one alternative strategy, and therefore only a weak reason in favour of normative naturalism. Even if one granted Quine not only the failure of a certain rather narrow programme in foundationalist epistemology, but the failure of any recognisable epistemological project, this alone would not be a decisive argument for normative naturalism, because there is still scepticism to be considered. What other reasons could be offered in favour of normative naturalism? Well, we have already seen one general motivation behind the urge to naturalism epistemology, viz. the location problem.

However, worries about the location problem will not provide support for normative naturalism. If we accept, for example, epistemic supervenience, this remains neutral about how to justify the reliability of our epistemic norms. It might perhaps be thought that epistemic eliminativism, the view that there are no normative epistemic properties at all, could motivate the general idea in normative naturalism. But if anyone ever thought this, it is a mistake, and even incoherent, since this train of thought somehow moves from the view that there are no normative epistemic properties to the view that our epistemic norms can be justified by appeal to scientific theories. But this seems to amount to the admission that our epistemic norms (or our beliefs about their reliability) might after all possess a normative epistemic property. Let us therefore leave being these considerations, and turn to a number of arguments for epistemological naturalism presented by Philip Kitcher in his widely read paper 'The Naturalists Return'.

Kitcher's arguments are interesting because he actually seems to suggest a view akin to normative naturalism. One of Kitcher's main arguments concerns the dependency of epistemological properties on psychological processes and properties. Perhaps it has been common in epistemology to accept a kind of apsychologism, according to which epistemic properties of a belief (such as the property of justifiedness) is dependent not on the psychological properties of the belief or the subject holding the belief, but only on the logical relations between the content of this belief and other beliefs which are held, or could be held, by the subject. Now, Kitcher suggests an objection to apsychologism, starting from the observation that whatever the logical relations between S's belief that P and her belief that Q, if S believes that Q for causally deviant reasons, then S will not be justified in her belief that Q, despite the logical relations to the content of her belief that P. This is no doubt a strong objection to apsychologism as defined here.

In fact it is so strong that one starts wondering if anyone ever really subscribed to this form of apsychologism, as distinct from inadvertently advocating it. What is more important, however, is that Kitcher's objection to apsychologism only supports epistemic supervenience. It simply shows that when we say that some belief is justified, we expect there to be some natural property, presumably one involving the causal processes of belief formation, which must be instantiated. This observation yields no support for normative naturalism; one can easily accept the sort of epistemological supervenience arising from Kitcher's observation and yet reject normative naturalism. Perhaps, therefore, Kitcher's real reason for proposing normative naturalism is scepticism about a priori reasoning in general.

We need not here be concerned with whether there is no such thing as a priori reasoning at all. We can do with the more modest observation that no one has as yet succeeded in providing persuasive a priori reasons in favour of any interesting epistemic norms, or at least for the bulk of the epistemic norms on which we no doubt rely in our everyday and scientific reasoning. Kitcher is not alone here. It is often assumed that the case for naturalised epistemology depends on the rejection of genuine a priori knowledge or justification. The more specific proposal would be that if we reject the possibility of a priori justification of epistemic norms, we should instead turn to normative naturalism. But of course, this is not enough.

If a priori justification of epistemic norms is not possible, it might that we we should accept some form of scepticism, as was pointed out above. There is, however, a much simpler case to be made for normative naturalism, which does not depend on the failure of other programmes. If normative naturalism is viable in the sense that epistemic norms can actually be shown to be reliable by appeal to a posteriori theories about the world, we need no further reasons to embark on the programme. In other words, if normative naturalism represents one way in which we can justify our epistemic norms, this is enough to make normative naturalism interesting. This is so whether or not genuine a priori justification of some of our epistemic norms, is available in addition. Therefore, all we need to do is to decide whether normative naturalism is in fact viable.

I here primarily have in mind the question if normative naturalism falters on a vicious circularity. 5. THE CIRCULARITY OBJECTION Here is Fumerton, in his critical remarks on epistemological naturalism: 'You want a solution to the problem of induction. There is potentially no difficulty for the naturalistic epistemologist.

If reliabilism is true, and if inductive inference is a conditionally reliable belief-dependent process, then we can inductively justify the reliability of inductive inference. Our inductive justification for the reliability of inductive inference might itself be reliable and if it is that will give us second-level justification. ' 'None of this, of course, will make the skeptic happy. You cannot use perception to justify the reliability of perception!

You cannot use memory to justify the reliability of memory! You cannot use induction the justify the reliability of induction! Such attempts to respond to the skeptic's concerns involve blatant, indeed pathetic, circularity. ' I take the challenge to be that we cannot in general justify our reliance upon epistemic norms by appealing to empirical information, because this procedure would be question-begging or circular. No doubt these worries about the prospects of normative naturalism are shared by many philosophers.

I shall refer to them as the circularity objection, although I do not mean to suggest that there is only one rather than a family of similar objections. I now want to consider a few common responses to these sorts of worries. Consider first Quine's famous remark in 'Epistemology Naturalised' that 'such scruples against circularity have little point once we have stopped dreaming of deducing science from observations'. This remark now appears confused.

When reading his text one cannot but think that Quine has eliminative naturalism in mind when he suggests that we should not worry about circularities. Presumably, no-one would disagree about that, but it leaves the challenge to normative naturalism unanswered. As mentioned, Quine has later insisted that his brand of naturalism does not repudiate normativity after all, and that it never did. This suggests that Quine really recommends normative naturalism rather than eliminative naturalism. But if what Quine had in mind really was normative naturalism, it is rather difficult to understand how he can dismiss the circularity objection simply like that. In conclusion, it is hard to see that Quine in 'Epistemology Naturalized' has anything to offer which can explain why the sort of circularities involved in normative naturalism does not render the whole project incoherent.

Let us then turn again to Philip Kitcher. In the passages where he comes closest to responding to the circularity objection, he says the following: 'A traditional naturalistic epistemology would include principles of the form "P is cognitively optimal", where the notion of cognitive optimality is understood to depend on generation of cognitive success in the actual world and where the defence of the principle will involve appealing to empirical information about human beings and the rest of nature. Faced with a skeptic who denies that the principle should be accepted, traditional naturalists will thus appeal to some body of empirical statements. Persistent skeptics will then inquire why these statements should themselves be accepted. If the original skeptical challenge has been cleverly posed, the attack will have been directed at a process of belief formation that is widely implicated in the proper generation of beliefs about nature (at least on the naturalist's view). In consequence, defense of the body of empirical statements will require the use of P, at which point the sceptic will complain that the question has been begged.

' This can be understood as a version of the circularity objection. Now, it is noticable that Kitcher is actually at the brink of agreeing with the sceptic as he goes on to say: 'The skeptic's demand is for synchronic reconstruction of beliefs: take the totality of things you believe, subtract this claim and everything that you cannot defend without assuming it, and now show that the claim is correct. ' ' [... ] one can produce unanswerable challenges by calling into question single claims og bodies of doctrine that are presupposed in all empirical investigations. On naturalism's own ground, there are bound to be unanswerable forms of skepticism. ' But apparently Kitcher does not think that normative naturalism amounts to scepticism after all, because in the paragraph immediately following the above quotation, he goes on to say: 'Traditional naturalists should therefore decline blanket invitations to play the game of synchronic reconstruction.

' Kitcher seems to think that this is enough to avoid the problem. There is a particular set of questions that the sceptic wants to ask, and a particular set of constraints that he imposes on the permissible answers. But we can avoid the embarrassing sceptical conclusion simply by declining to take up the challenge on those terms. Although Kitcher's stance towards the sceptic is by no means uncommon, it is nevertheless hard to be completely satisfied with what he says.

Kitcher seems to assume that the sceptical problem arises only when someone actually poses a sceptical challenge (or insists on one). Thus, he seems to think that if no one questions the body of empirical information appealed to in the defence of some epistemic norm, then all is well and good, and we can congratulate ourselves with having justified the norm. But why think that someone must actually pose the sceptical challenge? Why not say, as BonJour does on one place, that if the sceptic did not exist, epistemologists would have to invent him? Secondly, even if we accept some view of this kind, it is not clear why it is appropriate simply not to take seriously the charge, or simply decline to answer it, as Kitcher seems to suggest that we should. At least we need to be told why we do not have to worry about such questions.

Perhaps the most important point for the present discussion is this. The suspicion is that a certain justificatory project is incoherent in the sense that it purports to provide a way in which we can give some sort of positive epistemic status to our epistemic norms, but because of the inherent circularities the result will never be an increase in positive epistemic status. To illustrate: Suppose I believe that everything Mrs. Thatcher did and said was right, but I happen to believe this because Mrs. Thatcher once told me that this was the truth. Now, there seems to be a problem here, independently of whether or not anyone ever challenges my faith in Mrs. Thatcher, or my reasons for my high regards of her.

If there is a similar sort of problem in normative naturalism, there is no need for the sceptic to point out the problem. And the problem certainly does not go away simply because we decline to talk about it. In general therefore, it seems that the circularity objection to normative naturalism is not one of responding to a sceptical challenge, although the circularity objection of course might be used in mounting a sceptical challenge. Kitcher does not reply to the circularity objection in that form. As far as I am aware, he never considers the issue. 6.

RESPONDING TO CIRCULARITY I shall now outline what I take to be the best response to the circularity objection. Essentially, I shall separate various epistemic notions, and argue that some of these are perfectly compatible with the sort of circularity involved in normative naturalism, whereas others are not. Furthermore, I shall argue that although the sort of circularity involved in normative naturalism can reasonably be claimed to prevent us from reaching some of the aspirations we might have had in epistemology, this does not amount to a serious sceptical conclusion. First, we need to bring into focus again: (1) A belief's possessing the property of justifiedness For the sake of the discussion to follow, I shall now assume some form of externalist theory of the property of justifiedness. Often it is supposed that epistemological naturalism is simply tantamount to some sort of reliabilism, or a related kind of externalism, but this is a mistake. One can accept epistemic supervenience, and still hold an internalist theory of justification, e.g. a theory according to which a belief is justified only if it stands in some proper inferential relation to other beliefs that the subject holds, or might hold.

To hold that a belief's justifiedness in this way depends (inter alia) on inferential relations to other beliefs seems consistent with holding that the presence of these beliefs, and the proper inferential relationships, in turn supervenes on natural properties. There are many externalist views, but we need not worry about the details of these views here. We can simply assume, for the sake of discussion, that a belief's property of justifiedness consists of this belief's being acquired by some method which is reliable in normal circumstances. Assume, for example, that perception is reliable, that is, beliefs formed on the basis of perception (perceptual beliefs) are likely to be true (for some subject in some circumscribed set of circumstances.

According to epistemic externalism, a subject might be justified in accepting perceptual beliefs, although the subject cannot in any normal sense argue for the truth of these beliefs, or does not even explicitly believe that perception is reliable. For the following discussion it is important to note that the property of justifiedness is different from: (2) Showing that someone is justified in accepting a particular belief, i.e. explaining the features in virtue of which one is justified in accepting the belief. Showing that some belief is justified requires a theory of justification. Assuming again epistemic externalism, it also requires information about the way in which this belief was formed. Putting epistemic externalism and some factual information together, we might, for a particular belief held by a particular subject, explain the properties in terms of which this belief is justified. If epistemic supervenience is correct, this part of the explanation can, in principle at least, proceed entirely in terms of natural properties, in the sense that the properties in virtue of which a particular belief is justified are natural properties.

According to epistemic externalism, one might very well be justified in accepting some belief without knowing about or understanding the properties in virtue of which this belief is justified, that is, without being able to explain what makes one justified. There is a further distinction to keep in mind. Showing that someone is justified in accepting some belief is not the same as (3) Providing a cogent argument that this particular belief is true, or likely to be true Providing a cogent argument that some belief is true, or likely to be true, is appealing to a set of premises from which the truth or the likely truth of the target belief follows in a non-question-begging manner. Or, as I shall sometimes say, it is to provide epistemic reasons for the belief.

It is common ground that, on epistemic externalism, a subject might be justified in accepting a belief, without being in a position to provide a cogent argument that the belief is true, however weak 'being in a position' is here construed. Being justified in accepting a belief does not require epistemic reasons, or even the availability of epistemic reasons. What is more difficult to see, perhaps, is that explaining what makes a belief justified is different from providing an epistemic reason for it, i.e. that (2) and (3) are really different. It is helpful here to consider again the defence of induction Strawson once gave. As mentioned, Strawson's claim was roughly that inductive reasoning is epistemically rational as a consequence of what our concept of epistemic rationality is.

Or, in the present terminology, we might say that beliefs formed on the basis of induction count as justified, since this is so simply in virtue of our concept of justifiedness. A consequence of such a view is that if a subject reasons inductively in a heavily induction hostile environment, his beliefs might very well turn out to be false, but they will nonetheless be justified. Thus, we can, according to Strawson's theory, explain the properties in virtue of which the beliefs are justified, viz. that is has been arrived at by inductive reasoning. Obviously, however, we will not thereby have acquired a reason to think that the beliefs in question are true. Hence, there is a difference between explaining justification and providing an epistemic reason. In general, any theory of justification which denies that a belief's being justified entails that it is likely to be true, allows for a discrepancy between explain justification and providing epistemic reasons.

It might be thought, however, that precisely externalist theories of justification do hold that justification entails likely truth. But the more plausible versions of epistemic externalism does not have this implication. Suppose we say that a belief is justified just if it arises from a belief forming mechanism which is reliable in a normal world. Now, suppose that using such a reliable mechanism, you form some belief, but you are not operating in a normal world.

This belief is justified, and we can explain what makes it so. But obviously, we have not thereby been given a reason to think that it is true. Hence, even on plausible versions of externalism, there is a difference between explaining justification and providing a reason to think that a belief is true. Return now to epistemic norms. As I said above, the controversial parts of naturalistic epistemology concern how to justify our acceptance of various epistemic norms (in our everyday cognitive life as well as in science). Following what was said above, we can see that we need to carefully separate various issues.

Consider first: (4) The property of being justified in relying upon some epistemic norm. Note that one might rely upon some epistemic norm without believing that this norm is reliable. Suppose, for example, that I reason by way of inference to best explanation at some particular occasion. It might very well be the case that I do not hold the further belief that such inferences or epistemic norms are reliable. I simply trust them, and thereby I might be said to presuppose that these epistemic norms are reliable. I might, we can assume, be justified in relying on some norm, and let us assume that some form of epistemic externalism accounts for this kind of justifiedness.

Put very crudely, one is justified in relying on some epistemic norm just if one's doing so results from the right sort of process, where the right sort process presumably includes some sort of ability to discriminate between reliable and unreliable epistemic norms. Consider now again the sort of circularity at issue. Does this circularity present any obstacle to being justified in relying on epistemic norms? This seems clearly not to be the case. Given the sort of epistemic externalism regarding norm acceptance just outlined, it simply depends on natural facts. Nothing akin to the circularity problem arises here.

Turn now to the different issue of (4.1) Showing or explaining that someone is justified in relying upon some epistemic norm. Assume that we have stated the relevant externalist theory of norm acceptance, and that we have access to the relevant empirical (or modal) information. We can then, it seems, explain in terms of what some subject is justified in relying upon some epistemic norm. Or is there a circularity issue which prevents this?

Suppose, for example, that we want to explain why S is justified in trusting his visual perception. Our theory of justification in conjunction with various empirical facts entails that S is justified in accepting the norm. However, it now turns out that we have to rely on visual perception to obtain the facts which forms part of the explanation of what makes S justified in relying on visual perception. Does this invalidate the explanation? It seems not. No vicious circularity seems to arise.

Perhaps someone would like to strengthen requirements on explanations by holding that in explaining matters we must appeal to justified beliefs, not just beliefs. Even this further requirement does not make the explanation viciously circular. There is nothing to prevent that we might be justified in our perceptually based beliefs that some facts obtain, where these facts in turn is part of what explains what makes S justified in relying upon visual perception. There is simply no vicious circle to be found here. Obviously, one could question the empirical facts appealed to in these explanations, or rather the truth of the beliefs in these facts.

This would be a way of challenging the success of the explanation, because false beliefs cannot explain. But this does not show that explaining justifiedness in accepting norms involves an objectionable circularity. It merely shows that this explanation, like any other, can be challenged by challenging the explanans. Let us now turn to the slightly different issue concerning (5) The property of being justified in accepting the belief that some epistemic norm is reliable. and let us also immediately consider (5.1) Showing or explaining that someone is justified in accepting the belief that some epistemic norm is reliable. With a few obvious modifications, what has just been said applies to (5) and (5.1). Schematically put the situation is this.

Suppose that S believes that epistemic norm N is reliable. Suppose further that S has come to accept this belief by way belief forming mechanism M which is in fact reliable. This implies that S's belief is justified. Suppose further that M in fact incorporates N, that is, that M somehow makes use of N. In this case, I take epistemic externalism to imply that S's belief that N is reliable is justified. The fact that S, when forming this belief, in a sense relies on N itself does nothing to prevent this. Hence, the sort of circularity objection that one might raise has no force.

Furthermore, there seems to be no problem in explaining how someone might be justified in believing that some norm N is reliable, although he has somehow relied on N itself when arriving at that very belief. Again, the circularity objection does not apply. Could S also provide a cogent argument for the truth or likely truth of his belief that N is reliable, even when S need to rely on N in the sense explained? This is the issue of (5.2) Providing a cogent argument for, or epistemic reasons for, the belief that some epistemic norm is reliable.

As I have tried to make clear, providing arguments for the truth or likely truth of a belief is different from explaining that someone is justified in accepting it. I believe that whereas explanations of justification are not vitiated by the kind of circularity at issue, arguments are. Let us now take a closer look at these issues. Suppose S defends the reliability of perception by the following argument: (1) S believes that P 1 by way of perception, and P 1 (2) S believes that P 2 by way of perception, and P 2 (3) S believes that P 3 by way of perception, and P 3... (n) S believes that Pn by way of perception, and Pn (n+1) some inductive principle Perception, as used by S, is reliable Suppose further that S proposes this argument, and that S needs to use perception in order to assert the premises of the argument.

Suppose that S's perception is in fact reliable (under the circumstances used in asserting the premises). Suppose, as before, an appropriate form of epistemic externalism, which in conjunction with S's perception being reliable implies that S's perceptual beliefs are justified. Finally, suppose that we set aside whatever worries that might be raised by the use of induction in premise (n+1), and simply assume that the argument is valid. Reasoning patterns of this form are, I shall say, epistemically circular. It is interesting to note that we here have to do with a valid argument. Furthermore, the argument is not circular in the trivial sense that the conclusion of the argument appears as one of the premises.

Moreover, we have assumed that S is justified in asserting each of the premises of the argument. So shouldn't we concede that S is also justified in accepting the conclusion, and is so justified by an inference from the premises of the argument? After all, isn't it generally the case that if one is justified in accepting the premises of a valid argument and infers the conclusion from the premises, then one thereby becomes justified in accepting the conclusion? These issues deserves a much more careful discussion. Here I only wish to remark that it is defensible to deny the cogency of epistemically circular arguments, that is, to deny that one can become justified in accepting a belief by way of an epistemically circular argument. In other words, one might reasonably hold that epistemically circular arguments are not cogent, despite the fact that they are not circular in any trivial sense.

No doubt critics of normative naturalism would agree, and let us assume that they are right. The question now is what the implications are of this concession for normative naturalism. First, it is implied in the previous discussion that we should separate two very different versions of normative naturalism: Normative naturalism as an explanatory project, the aim of which is to explain how we can be justified in our acceptance of, or reliance upon, various epistemic norms (or whatever we take the object of the explanation to be). Normative naturalism as a justificatory project; that is, a project the aim of which is to provide cogent arguments for accepting the proposition that certain epistemic norms are reliable. What the previous discussion indicates is that normative naturalism cannot succeed as a justificatory project, provided that we make the following further assumption: when attempting to provide a cogent argument for the reliability of our epistemic norms, epistemic circularity is unavoidable. Sooner or later, our acceptance of the premises in arguments for the reliability of epistemic norms will require that we rely upon those very norms.

Epistemic circularity is, one might say, endemic. I haven't defended this further assumption here, but others have, and I believe they are right. Does this spell the end of normative naturalism? Does it imply that normative naturalism is really a sceptical position? No, and it should now be obvious why. The concession was that normative naturalism understood as a justification project fails, since (as we assumed) epistemic circularity is endemic, but epistemically circular arguments are not cogent.

However, when conceived as an explanatory project, in the way I have tried to describe, normative naturalism survives circularity. Hence, the concessions made are fully consistent with our reliance on various epistemic norms being fully justified, and in us being justified in accepting the beliefs that these norms are reliable. Furthermore, we can, without vicious circularity, explain how we are justified in relying on various epistemic norms, and justified in believing that these norms are reliable. Perhaps someone will insist that the failure of normative naturalism as a justificatory project is a sceptical outcome. But it seems that what counts as a sceptical conclusion should depend on what our hopes were. Relative to the high hopes of providing a cogent argument for the reliability of our basic epistemic norms, it certainly is a sceptical conclusion.

But perhaps this was unnecessarily ambitious in the first place. Relative to our common sense view that a number of our beliefs are justified, among these beliefs about the reliability of our epistemic norms, the failure of the justificatory project is not a sceptical conclusion, since this is a far cry from the conclusion that we are never justified in any of our beliefs, or that we are never entitled to rely on any of the epistemic norms we in fact rely upon. The conclusion is not even that we cannot explain in virtue of what we are so justified. The conclusion is merely that there are ways of picking epistemic norms, such that our attempts to argue that these norms are reliable will be epistemically circular and hence, we assumed, not cogent. Alm eder, Robert (1990) 'On Naturalizing Epistemology', American Philosophical Quarterly, vol. 27, no. 4, 1990,263-279.

BonJour, L. (1994) 'Against Naturalised Epistemology', Midwest Studies in Philosophy, XIX, 1994,283-300. BonJour, L. (1998) In Defense of Pure Reason, Cambridge University Press 1998. Cleve, van James (1985) 'Epistemic Supervenience and the Circle of Belief', The Monist, vol. 68, 90-104. Dop pelt, Gerald (1990) 'The Naturalist Conception of Methodological Standards in Science: A Critique', Philosophy of Science 57, 1-19. Foley, Richard (1994) 'Quine and Naturalised Epistemology', Midwest Studies in Philosophy, XIX, 1994,243-260.

Fumerton, R. (1994) 'Scepticism and Naturalistic Epistemology', Midwest Studies in Philosophy, XIX, 1994,321-340. Goldman, A.I. (1994) 'Naturalistic Epistemology and Reliabilism', Midwest Studies in Philosophy, XIX, 1994,301-320. Hylton, Peter (1994) 'Quine's Naturalism', Midwest Studies in Philosophy, XIX, 1994,261-282. Kappel, K (2000 b) 'Virtues of Virtue Reliabilism', forthcoming.

Kappel, K (2000 c) 'Externalism, Scepticism, and the Goal of Epistemology', forthcoming. Kappel, K (2000d) 'Epistemic Circularity and Transmission of Justification. Part I', forthcoming. Kappel, K (2000 e) 'Epistemic Circularity and Transmission of Justification. Part II', forthcoming.

Kim, J. (1988) 'What is "Naturalised Epistemology"?', Philosophical Perspectives 2, 1988. Reprint in Naturalising Epistemology, Second Edition, ed. Kornblith, MIT Press 1994, 33-55. Kitcher, P. (1992) 'The Naturalist Return', Philosophical Review 101, 1992, 53-114. Kornblith, Hilary (1994) ' Introduction: What Is Naturalistic Epistemology?', in Naturalising Epistemology, Second Edition, ed. Kornblith, MIT Press 1994, 1-14.

Laudan, Larry (1987) 'Progress or Rationality? The Prospects for Normative Naturalism', American Philosophical Quarterly, vol. 24, no. 1, 1987, 19-31. Laudan, Larry (1990) 'Normative Naturalism', Philosophy of Science 57, 44-59. Leptin, Jarre t (1990) 'Re normalizing Epistemology', Philosophy of Science 57, 1990, 20-33. Lipton, P. (1991) Inference to the Best Explanation, Routledge 1991. Maff ie, James (1990) 'Recent Work on Naturalised Epistemology', American Philosophical Quarterly, vol. 27, no. 4, 1990,281-293.

Quine, W.V. O (1969) 'Epistemology Naturalised', in Quine, Ontological Relativity and Other Essays, Columbia University Press 1969, 69-90. Rosenberg, Alexander (1990) 'Normative Naturalism and the Role of Philosophy', Philosophy of Science 57, 1990, 34-43. Strawson, P.F. (1983) Scepticism and Naturalism: Some Varieties (1983 Woodbridge Lectures)? ?