ISSN 1393-614X Minerva - An Internet Journal of Philosophy Vol. 11 2007

The Empirical Basis to Skepticism



Robert G. Hudson


Broadly speaking, there are two different ways in which one might defend skepticism – an a priori way and an empirical way. My first task in this paper is to defend the view that the preferred way to defend skepticism is empirical. My second task is to explain why this approach actually makes sense. I accomplish this latter task by responding to various criticisms one might advance against the possibility of empirically defending skepticism. In service of this response, I distinguish between two different kinds of hallucination, ‘metaphysical’ and ‘ordinary’, and seek to clarify the notion of a ‘presupposition’.



In general terms, there are two ways which might be used to defend the thesis of skepticism: empirical and non-empirical ways. Non-empirical approaches usually take the following form, which I call the ‘consistency approach’: skepticism is justified, one might argue, because its assertion is consistent – it is logically possible to be a skeptic. In crude terms, the skeptical argument is this: we could be wildly hallucinating right now, there could be a malicious evil demon with god-like powers intent on deceiving us every step of the way, we could be wrong in all ways imaginable, therefore, we lack knowledge of the world. How effective is this argument? Not very: the bare logical possibility of such a scenario is not a very enticing philosophy. Lots of circumstances are logically possible — that I will live forever, that elephants will start growing wings, and so on — most of which I prudently ignore. Thus, as a means of justifying skepticism, I do not think the consistency approach takes us too far. It is a priori, logically possible that we lack knowledge about world, but why think this, in any case?


Because of the apparent paucity of purely a priori, logical defenses of skepticism, I plan here to defend the view that to support skepticism we need to consider an empirical approach. My sense is that an empirical approach has been the traditional way to defend skepticism in the modern era. For instance, one of the main Cartesian supports for skepticism, the dream argument, is an empirical argument: it says that, since there are no (empirical) markers by which we can distinguish dreams from reality, it follows that we do not know we are experiencing reality, after all. Similarly, the Humean argument for skepticism about the existence of objective causation is an empirical argument. It says that, in determining whether event A causes event B, we need to be able to determine whether every instance of A is followed by an instance of B; but, since in the usual case we lack access to such a (mammoth) resource of data, we should be skeptical about the existence of a causal link between A and B. Yet another approach to recognizing the empirical basis to skepticism is suggested by Quine (1975). Quine notes that empirical science renders skeptical hypotheses reasonable, given that the scientific image of the world is markedly different from the intuitively trustworthy, manifest image of the world – tables appear solid but they are actually filled with space, objects looked coloured but colour is only a product of our minds, and so on. That is, we learn through empirical inquiry, inquiry supporting scientific theorizing, that our usual way of viewing the world is in many respects erroneous; and, having lost our intuitively most trusted, commonsensical picture of the world, we are thus prone to skepticism.


Overall, how significant is the empirical character of these arguments? Well, consider the result if, miraculously, there were empirical markers distinguishing dreams from reality – for instance, suppose there were certain distinguishing sights or sounds exhibited only by dream experiences. The dream argument, by this very empirical fact, would lack support. And there is no reason a priori why this couldn’t occur. The crux of the dream argument is that we learn through experience that dreams are fundamentally indistinguishable from reality; but, for all we know, things might have been otherwise and the dream argument might have been groundless. Analogous comments apply with respect to Humean skepticism. For all we know, it could have been the case that we have access to all the possible instances of a correlation and so could thereby empirically discern the necessity of a causal link. If this were true, we could effectively defeat Humean skepticism since there could be no further appeal to the uncertainty of future events. Finally, imagine a case where an empirically-grounded scientific advance leads scientists to re-affirm the common sense view of the world – scientists, let’s say, conclude that apparently solid objects are solid, coloured objects are coloured, and so on. In such a situation, Quine’s argument that scientific advance leads to a skeptical view of the world would be turned back.


In this paper, my goal is to affirm the legitimacy of such an empirical approach to adjudicating skeptical hypotheses, and to this end I plan to address various objections that might be advanced against such a strategy. The first objection concerns the problem of vicious circularity. In particular, if empirical data succeed at justifying the skeptical hypothesis, then how can such data be said to be adequately justified, given that the skeptical hypothesis is affirmed? For if skepticism is true, the empirical data supporting the skeptical hypothesis are undermined, which in turn undermines the support provided for the skeptical hypothesis. And from here we move in a circle because, if skepticism lacks support, we restore the justifiedness of its empirical support and so again justify skepticism. So, is such vicious circularity a serious problem for the empirical justification of skepticism?


The first point to make here is that such a problem arises even if we provide an a priori justification for skepticism. For instance, if one argues that skepticism is justified because it is a consistent position, then of course the question arises whether we are justified in believing that skepticism is a consistent position to begin with. And again, if skepticism is true, we lose our justification for the claim that it is consistent, and so thereby lose our justification for skepticism. Once again, we move in a vicious circle. So, can we have any form of justification for skepticism, if skepticism has the effect of denuding its own justificatory support?


I think roughly the same answer to this question works for both a priorists and empiricists, but I shall only examine the empiricist’s line. The issue turns on how we view ‘empirical proof’, and here I take a methodological view: empirical proof involves the invocation of empirical claims in support of a hypothesis, where empirical claims are defined as claims prompted by and describing observational circumstances. This definition of an empirical claim, though accurate, is loose and only suggestive. Nonetheless, it contains a core truth, which is that an empirical claim need not be true (though it should not be known to be false) in order for it to play a role in justification. Many empirical claims in science, for example, have dubious truth values though they still stand as empirical claims; consider any of the more speculative observational claims made in contemporary physics and chemistry. Scientists often realize that their observational claims are dubious, but they recognize that there is little else at hand to use as an informational resource. Thus, as a methodology, they advocate the use of empirical sources on the basis of which they adjudicate the quality of their hypotheses, while attempting to ameliorate the prospect of error by taking great care in ensuring the veracity of observational claims through the use of various technical stratagems (e.g., calibrating apparatus, removing background information, replicating data, and so on). Nevertheless, it would be impractical for them to make the truth of observational claims a requirement of empirical justifiedness. Doing so would make empirical justifiedness a near impossibility, and would obstruct the path of science.


Accordingly, my suggestion is that we adopt such a methodological perspective in defending skepticism from the vicious circularity problem. For, in science, what ensues if a hypothesis is supported on the basis of empirical data, which hypothesis then casts doubt on the veridicality of the data? We saw these sorts of cases above in our discussion of Quine, where we noted that the scientific image of the world differs markedly from the manifest image. Thus, for example, physicists who experimentally research the atomic nature of matter (according to which atomic objects are largely space, despite the appearance of solidity, and are colourless, despite their apparent colour) use as empirical data records generated by solid and coloured machines transcribed into solid and coloured books. So, what follows from this? Here, for a scientist, what follows is not that we should plead paradox. For scientists reckon that there is an informational conduit of sorts stemming from the true state of the world to our observational states, and that it is up to us to interpret these states in a way that is consistent with our favoured theoretical perspective. In practice this might involve introducing the appropriate neuro- and socio-psychological theories to account for the genesis of the manifest image; this is one way to restore consistency between the manifest and scientific images. My suggestion, in parallel, is that we take a similar attitude towards the skeptical hypothesis and its susceptibility to vicious circularity. If we have empirical evidence for this hypothesis, we should not become bewildered if this evidence seems undercut by skepticism. For it is up to the skeptic to then interpret the evidence in such a way as to restore its coherence with the skeptical outlook. For instance, a critic of Cartesian skepticism might object that the claim, “There are no (empirical) markers by which we can distinguish dreams from reality”, the claim on which the Cartesian skeptic bases her skepticism, is itself doubtful from a skeptical perspective. Thus, Cartesian skepticism seems to confute itself. But here the skeptic can reply that, even if there are markers distinguishing dreams from reality (which would be true if the above claim were false), we are unaware of them, and so we are no further along in establishing our claim to knowledge. Notice that the skeptic’s reply in this case is an empirical argument – it rests on our lack of awareness of empirical markers distinguishing dreams from reality, leaving aside their existence. This is, I submit, how our disputes in epistemology should go, as they always go in science. We ask, what further empirical phenomena might bear on the issue at hand? In this way our enquiries are advanced as opposed to having our philosophic predilections further entrenched.


So, having responded to the vicious circularity problem, let us consider some further objections to the claim that skepticism is supported (if at all) empirically. To begin with, one might argue that in the absence of relevant evidence, such as dreams, hallucinations, and the like, skeptical hypotheses are worse than weakened, as I had suggested, but plainly false. For, one might argue, the skeptical hypothesis essentially is the hypothesis that what we take for perception is hallucinatory (or is illusory, or is a dream, and so on), and so in a situation where there is a complete absence of hallucinations, any skeptical hypothesis that trades in hallucinatory episodes is not even possible. Consequently, to describe the skeptical hypothesis in such a case as simply ‘weakened’ is a gross understatement. Similarly, suppose we never had experiences of dreaming. Would not Descartes’ dream argument be outright refuted? For how could all our experiences be parts of dreams, when there are no dreams to begin with? Again, as regards Humean skepticism, suppose correlations always held universally. Would not Hume’s doubt regarding causation be completely undermined, and not simply lacking empirical support? If there are no failed correlations, the hypothesis that all correlations fail makes little sense.


But these assessments betray a confusion. For simplicity, let us focus on the notion of hallucination (similar arguments would hold mutatis mutandi for dreams and failed correlations). We need to distinguish here between two kinds of hallucination, ‘ordinary’ hallucination and ‘metaphysical’ hallucination. Metaphysical hallucination is a state where all our sensory experiences are illusory in fundamental and striking ways, a state in which we are completely mistaken about the nature of the world displayed by ordinary experiences, a state deceptively generated by evil demons, vat scientists and the like. By contrast, ordinary hallucinations are those experiential states that, in the normal course of affairs, we ‘label’ as hallucinatory, states induced as we all know by the ingestion of various drugs, by excessive thirst and hunger, and so on. Given this distinction, where we are considering a circumstance where we never endured ordinary hallucinations, drawing the inference that skeptical hypotheses are false would be fallacious, for it is possible to be in a state of metaphysical hallucination while enduring no ordinary hallucination. This sort of circumstance is built in as a possibility with skeptical hypotheses. Nevertheless, what we can say here is that, if we never suffered ordinary hallucinations or other kinds of deceptive experiences such as illusions or dreams, we would not find it to be a compelling thesis that we are metaphysically hallucinating. For, never having before experienced deceptive phenomena, it would not make much sense for us to suppose that we are continually and perpetually deceived, despite the apparent normalcy of our situation. But here we would simply be stating an epistemic fact about our situation, about what there is evidence to believe, as opposed to making a metaphysical claim. That is, where we lack an empirical, evidential basis to say we are metaphysically hallucinating, there is nothing inconsistent in saying that we are metaphysically hallucinating.


Still, how can we know that we sometimes hallucinate and sometimes dream, or appreciate at all the fact that our sensory faculties sometimes operate in ways that admit the possibility of illusory impressions, unless universal skepticism is false? For, one might argue, to know that we are sometimes deceived is to know that we are usually able to perceive things correctly, that is, to know that we have a backdrop of regularly veridical experiences. But if this is correct, how can one claim that the occurrence of ordinary hallucinatory experiences — which occur only sporadically — provides support for skepticism if the occurrence of (only) sporadic hallucinations presupposes the falsity of skepticism?


I think we can accept that, if we take ourselves to observe only sporadic instances of illusion, then we presuppose ourselves to have generally veridical representations of reality. But, from here, can we go further and conclude that such instances of illusion are far more common? This is the classic skeptical inference: given our occasional propensity to deception, the skeptic infers that we are prone to deception to a far greater extent than we originally suspect. And, by drawing this inference, it seems we must suspend the presupposition we made in asserting our premise, that we have generally veridical access to the world. So, is such an inference possible? In other words, if a premise presupposes a particular claim, can one derive a conclusion from this premise that confutes this presupposed claim? Clearly this depends on what we mean by ‘presupposition’. For any claim C, I take a presupposition of C to be an assumption one tends to make when asserting C, and not specifically an assumption that must be true if C is to be asserted as true. Thus, where the claim is, “we are sometimes deceived”, we tend to make this assertion under the assumption that we are usually accurate in our perceptions. However, it could be true that we are both sometimes deceived and always deceived. This is at least true where we are always deceived! Indeed, the fact that we are sometimes deceived evidentially supports the claim that we are always deceived – e.g., if I occasionally make mistakes, it is a legitimate worry that I might be making mistakes all the time. This can be true, despite the fact that in originally asserting “we are sometimes deceived” we were under the impression that we were usually getting things right.


I had earlier made the distinction between ordinary and metaphysical hallucination. Here it is tempting to suggest that, while in the context of metaphysical hallucination, ordinary hallucinations are not illusory after all, for so long as we have identified them as illusory we are in reality connected to the truth, since they are illusory. Accordingly, one might ask, how can I suggest that the occurrence of ordinary hallucinations provides support for skepticism? If skepticism is true — that is, we are metaphysically hallucinating — then in a sense ordinary hallucinations are telling us the truth: they are telling us that the world is much different from how we usually perceive it, and in fact it is much different. Thus, in a remarkable stroke of epistemic luck, we refute skepticism in that we possess a sort of direct access to our illusory state of being. In acceding to skepticism and affirming our state of metaphysical hallucination, we end up in a state of knowledge confirmed by our various, ordinary hallucinations.


But surely there is a serious flaw in this understanding of ordinary hallucination in the context of metaphysical hallucination? For when we identify a state as an ordinary hallucination we usually have in mind some naturalistic or physicalistic explanation for the event, for example, that the hallucination results from the ingestion of various drugs, from excessive thirst and hunger, and so on. We do not have the habit of explaining such hallucinations as resulting from the devious machinations of an evil demon, a vat scientist, and so on. Indeed, to provide such skeptical explanations for the occurrence of ordinary hallucinations would be quite irregular, and would de-legitimize our description of them as ‘ordinary’ hallucinations. But if ordinary hallucinations are understood in this ‘ordinary’ way, it follows that they are not accurate perceptions of a skeptical reality, one in which we are in a state of metaphysical hallucination. For if they are accurate perceptions of anything, they are of states of drug ingestion, or of extreme hunger and thirst, and the like. Yet the existence of such disturbed states is itself an illusion from the perspective of a metaphysical hallucination. So we re-affirm ordinary hallucinations as illusory representations of the world in the context of a metaphysical hallucination. As such, even assuming the truth of skepticism, the unreliable nature of ordinary hallucinations is affirmed and they are able to provide some justification for the claim that we are always subject to hallucinations, metaphysically speaking.[1]






Quine, W. V. O. (1975), “The Nature of Natural Knowledge”, in Samuel Guttenplan, ed., Mind and Language (Oxford: Oxford University Press), pp. 67-81.








[1] The author gratefully acknowledges support from the Social Sciences and Research Council of Canada, grant number 410-2006-214.






Copyright © 2007 Minerva



All rights are reserved, but fair and good faith use with full attribution may be made of this work for educational or scholarly purposes.





            Robert Hudson teaches Philosophy at the University of Saskatchewan in Saskatoon, Saskatchewan, Canada.




         Mailto: Robert Hudson


Return to Minerva (Volume 11) Main Page                                                      Go to Top of This Page