Skepticism and information more

1 SKEPTICISM AND INFORMATION Eric T. Kerr & Duncan Pritchard University of Edinburgh ABSTRACT. Philosophers of information, according to Luciano Floridi (2010, 32), study how information should be “adequately created, processed, managed, and used.” A small number of epistemologists have employed the concept of information as a cornerstone of their theoretical framework. How this concept can be used to make sense of seemingly intractable epistemological problems, however, has not been widely explored. This paper examines Fred Dretske’s information-based epistemology, in particular his response to radical epistemological skepticism. We discuss the relationship between information, evidence and knowledge in relation to the problem of skepticism and the options available to an information-based epistemology for dealing with it. KEYWORDS: Scepticism. Epistemology; Evidence; Information; Knowledge; Philosophers of information, according to Luciano Floridi (2010, 32), study how information should be “adequately created, processed, managed, and used.” It is unlikely that we can do this without linking that study to the epistemic purposes of creating, processing, managing, and using information. Doing so, we claim with Floridi, requires attention to the epistemic value of information. In particular, our interest in information has a number of purposes, one of which is to acquire knowledge.1 If I visit the doctor to be told that I am suffering from Creutzfeldt– 2 Jakob disease he will likely bombard me with information, either through leaflets and documents or through informing me verbally. Our joint purpose is that I will know more about the disease (so as to better arm myself against its effects as well as to make me more comfortable with my condition). I will be informed that CJD is a rare but fatal brain disorder; it affects about one person in every one million people per year worldwide; symptoms typically occur at about age 60; about 90% of patients die within 1 year; and so on. We are rarely occupied with collecting information for information’s sake. I want this information because it is relevant to knowledge I wish to acquire about CJD. The information is epistemically valuable to me in the situation I am in. As has been noted elsewhere (e.g., Himma 2007), we are overloaded with information in the modern age. In this paper we examine these paths from information to knowledge and how constricting the range of relevant information is critical to information management. With the development of fairly recent technology, information has become a ubiquitous cultural buzz-word: the Information Age; information overload; the Information Superhighway; freedom of information; information technology; information science; and so on. Information and knowledge appear together frequently both in popular writing and scientific disciplines either as conflated terms for the same phenomena or related terms in some way involved in practices of inquiry, discovery, knowledge acquisition, and so on. The job of relating these concepts more precisely has tended to be undertaken by various academic disciplines that take information as a key theoretical concept. Although it is other disciplines such as information technology, knowledge management and library science that have devoted sustained analysis to information, such growing cultural awareness of information has provoked some philosophers to comment on its societal, epistemological, ontological, or axiological significance and sometimes to use it as a component in their philosophical work.2 3 Two philosophers of particular note in this regard are Fred Dretske and Floridi. Both have developed technically complex epistemologies with information playing a central role. Dretske connects information to knowledge via an ordinary dictionary definition of the former: “[By information] I mean nothing very technical or abstract. In fact, I mean pretty much what (I think) we all mean in talking of some event, signal or structure carrying (or embodying) information about another state of affairs. A message (i.e., some event, stimulus or signal) carries information about X to the extent to which one could learn (come to know) something about X from the message.” (Dretske 1983, 10) By relating information to knowledge in this way, Dretske’s information-based epistemology becomes allied to the relevant alternatives theory of knowledge (or ‘RAT’, for short) that he puts forward. According to this view, possessing knowledge depends on an agent’s capacity to rule out a certain range of alternatives which varies according to what kind of alternatives are relevant. The notion of relevancy in play here has been notoriously difficult to pin down (Floridi 2010, 300-24; Shope 2002, 37). Duncan Pritchard states the RAT view as applied to perceptual knowledge as follows: S has perceptual knowledge that p only if S can discriminate the target object at issue in p from the objects at issue in relevant alternative (not-p) propositions, where a relevant alternative is an alternative that obtains in a near-by possible world. (Pritchard 2010, 3) According to this rendering of the RAT view, our capacity to possess perceptual knowledge is heavily affected by our environment. Pritchard (2009, 5) makes the distinction between epistemically friendly and unfriendly environments. Most of the time, it will be very easy for us to make the necessary discriminations between, for example, hands and stubs, or canaries and crows. But epistemology is replete with thought experiments which arrange our environment such that it will not be so easy. For instance, the ‘barn façade’ case describes one such 4 environment where we no longer have the easy capacity to discriminate between barns and other things which may be in the environment (on account of how many of the items in this environment which look like barns are in fact barn façades). On this rendering of the RAT view, then, one consequently fails to know that the object before one is a barn.3 It is because of the possibility of deceptive environments like this that Dretske denies that information alone could ever answer a skeptical doubt. The argument for this is as follows: I have many defeasible reasons for thinking that I am writing these words in Edinburgh, Scotland just now (memory, testimony, observation, etc.). This gives me an informational basis for believing that I am writing these words in Edinburgh. However, I do not have an informational basis for believing that I am not a brain-in-a-vat (BIV) on Alpha Centauri who is being fed the illusion that he is writing these words in Edinburgh, Scotland. Even if the standards for knowledge are very low, and even if I know that were I in Edinburgh then I would not be a BIV on Alpha Centauri, this would not give me an informational basis for denying the skeptical hypothesis. The reason for this is my inability to discriminate between the scenario in which I am in Edinburgh and the skeptical BIV scenario in which I am on Alpha Centauri. Accordingly, argues Dretske, it follows that I receive exactly the same information in either scenario, and hence that I can have no informational basis to reject the alternative skeptical scenario. In general, Dretske argues that no signal can carry the information that a skeptical hypothesis—an hypothesis explicitly designed such that it is indiscriminable from normal circumstances, and yet involves a high degree of error—is false. In his Knowledge and the Flow of Information, for example, he writes: “No signal can rule out all possibilities if possibilities are identified with what is consistently imaginable. No signal, for instance, can eliminate the possibility that it 5 was generated, not by the normal means, but by some freak cosmic accident, by a deceptive demon, or by supernatural intervention.” (Dretske 1981, 130) And, later: “This is true of all indicators, all sources of information. That is why there is nothing in the world […] that indicates that there is a material world.” (Dretske 2005b, 22) So on Dretske’s view I can have an informational basis for believing that I am in Edinburgh but I can have no informational basis for believing that I am not a BIV on Alpha Centauri (a sceptical hypothesis which entails that I am not in Edinburgh), even whilst I know that if I am a BIV on Alpha Centauri then I am not in Edinburgh. It is for this reason that Dretske denies epistemic closure.4 In its crudest form, epistemic closure is the principle that if an agent knows one proposition, and knows that it entails a second proposition, then that agent also knows the second proposition. So, for example, if one knows that one is presently in Edinburgh, and one knows that this entails that one is not a BIV on Alpha Centauri, then one knows that one is not a BIV on Alpha Centauri. Although this principle has broad intuitive support, Dretske rejects it.5 But why is it that on Dretske’s view I can acquire knowledge about a proposition but not about a proposition which I know full well is entailed by it? Dretske is led into this position through two closely related commitments: (i) that perceptual information is never relevant to skeptical hypotheses, and (ii) that information is essentially non-factive evidence. We noted the first commitment above. Since, ex hypothesi, agents cannot discriminate between normal scenarios and skeptical alternatives, so it follows, according to Dretske, that agents lack an informational basis for dismissing skeptical alternatives. The second commitment becomes clear once we reflect that if information could be factive evidence for what it is evidence for—if, that is, it could entail the truth of what it is evidence for—then it would follow that the information we have to support our beliefs in normal circumstances 6 might well suffice to entail the denial of the target skeptical scenario. Clearly, however, Dretske does not think that we ever have evidence of this sort, and hence a non-factive view of the evidence provided by information is clearly implicit here. In order to more closely examine these commitments, consider the following local skeptical hypothesis, which we will call ‘Zebra’: Zebra Fred is at the zoo. If he perceives what he takes to be a zebra, Fred can have no informational basis for believing that what he perceives is not, in fact, a cleverly-disguised mule. In other words, the signal carrying this information does not allow him to discriminate between ‘a zebra in my perceptual field’ and ‘a cleverly disguised mule in my perceptual field’. Fred may interpret the signal as evidence that there is a zebra in front of him as a matter of habit, or perhaps relying on other evidence such as the sign on the fence or assumptions about what kinds of animals are in a zoo. However, his information is, it seems, non-factive. Just because I receive a signal such as this does not entail that there is in fact a zebra in the pen. More generally, as Dretske claims, it appears that none of the information that the subject possesses which indicates that he is perceiving a zebra is information which offers him an adequate epistemic basis on which he can dismiss the ‘cleverly disguised mule’ skeptical scenario. This way of thinking about our evidential position with regard to skeptical challenges has, however, been challenged. Ram Neta (2002; 2003), for example, has argued that the scope of your evidence is affected by context. Under this account, there is a range of contexts in which evidence (read: information) is factive. Neta argues that the skeptic only appears to succeed by restricting what counts as evidence. In normal contexts my evidence typically is factive, and it only becomes non-factive in skeptical contexts in which very demanding standards for what counts as evidence are in play. Hence, 7 in the zebra case, my evidence for believing that there is a zebra before me could well be factive in normal contexts. For example, if my evidential state in normal contexts is that of seeing that there is a zebra before me, then, since seeing that p entails p, my evidential state actually entails that there is a zebra before me, and which hence entails that I am not currently being presented with a cleverly disguised mule. Relatedly, if my evidence, in normal contexts, for believing that I have two hands is that I can see them before me, then I have evidence which entails not only that I have two hands, but also that I’m not a handless BIV on Alpha Centauri. According to Neta, however, the context can change in such a way as to restrict the scope of one’s evidence. If I were to gain evidence that cast doubt upon my belief that I have hands—for example, if I were to witness a room of BIVs—then this would make the possibility that I am a BIV a relevant alternative. This is effectively what the skeptic does: to describe such a scenario and cast doubt upon what was previously undoubted. There are two ways in which this may be done. Either the skeptic may simply suggest the possibility of a skeptical hypothesis that had previously been ignored or unexamined by the subject. This may place an onus on the subject to now eliminate that possibility in order to be correctly said to know the proposition. This intuition suggests that we cannot know a proposition until we have ruled out all relevant alternatives and that the range of relevant alternatives is determined by the conversational context. (Pritchard 2010, 19) In other words, being made aware of an alternative, however implausible or absurd, can make that alternative relevant. The second way in which the skeptic can make the alternative relevant is by actually offering evidence for thinking that a skeptical scenario has obtained. For example, consider an extension to the case of Zebra: 8 Zebra* Fred’s friend and skeptic, Frank, mentions to Fred that he once read a science-fiction story in which all the world’s zebras are replaced by hologram zebras and the real zebras are taken to a neighbouring planet. A little while later, Frank notices a pot of paint lying beside the animal and brings this to Fred’s attention by gesturing towards it. He also tells Fred that the sign on the outside of the pen appears to have been written over an older sign, suggesting that a different message was once written there. In this example, Frank initially merely presents Fred with a radical skeptical hypothesis. In the view of some epistemologists, such pronouncements can change the conversational context in which evidence requirements and relevant alternatives are set. Frank’s story may rob Fred of his knowledge that there is a zebra in the pen before him. In the subsequent details of the story, however, Frank presents Fred with perceptual information and testimonial evidence for calling into doubt Fred’s knowledge of what is in the pen. According to Neta, in these skeptical contexts Fred’s evidence is no longer factive. In particular, it is now no longer the case that one’s evidence can entail the denials of skeptical hypotheses, given that they are in play and problematising our epistemic position. So although in normal contexts my evidence that I am seeing two hands could be that I see that I have two hands, in skeptical contexts where the skeptical hypothesis is at issue my evidence can at most be that I seem to see that I have two hands, where this evidential standing clearly does not entail the target proposition. It is, on the other hand, possible to gain evidence supporting local skeptical claims. In the case of Zebra, if I were to notice a pot of paint next to the animal or its flaking ‘skin’, then this may provide an evidential or informational basis for believing that the animal is a cleverly disguised mule. If one subscribes to Dretske’s relevant alternatives theory or Neta’s contextualism, then the absence of such signals means, respectively, that either we are not required to rule out this possibility or that we are in an ordinary context in which the denials of skeptical hypotheses are 9 known. Consider O. K. Bouwsma’s (1965) adventures: when Tom peels away part of his face he receives a signal carrying the information that he is in a world made of paper (i.e., that a sceptical hypothesis—viz., that the world he perceives is not what ‘real’—is true). Of course, one could take this a stage further and ask if the perception of a paper world is also the victim of a sceptical trick but there the same test will apply. Whilst Tom is in the paper environment he has the capacity to discriminate and can come to know. Information, in these local skeptical scenarios, is relevant to what Tom knows. In Zebra, it appears, we have perceived signals that carry the information that the animal may be a painted mule. What is relevant information is constrained by sceptical or non-sceptical environments. Just as the victim of CJD does not need to know about the controversies over the aetiology of CJD (because he is a sufferer not a specialist doctor), he does not need to know the denials of sceptical hypotheses which may cast doubt upon what knowledge he possesses about CJD (because he is an epistemic agent and not an epistemologist). The upshot of this is that information not only has the function of providing a basis for knowledge but also an alternatives or contextdefining function. This gives pluralist epistemologies such as relevant alternatives theory and contextualism practical application as epistemic sorting-machines for information managers: in what contexts can we know what we want to know, what information is relevant, what information changes the contexts for knowledge, what are the epistemic limits of information? To return to our original scenario, the knowledgeable specialist is one who can inform me of relevant information about CJD and also point me in the direction of reliable information sources elsewhere (and steer me away from dodgy websites and quack medical treatments). In most cases, these sources will not be denials of skeptical hypotheses but they will be sources of 10 information which will increase the likelihood of my acquiring knowledge about my condition and how to cope with it. It would be an odd special case if local skepticism were the only epistemological problem that can be affected by informational signals in a context. Information services such as libraries, databases and internet search engines can also make use of relevant alternatives in order to organize and structure their resources and content. Here are two apparent truisms. First, that our interest as inquirers in information is often motivated by our desire to gain knowledge about something.6 Second, that we are almost always faced with limited information about the target issue. At the very least, one can always think that it would be better if one had more information about this subject matter. What falls out of these two statements? One might think that, as Aristotle claimed of knowledge (De Anima, 402a1), more information is always better than less and so we should endeavor to collect as much information as possible on the matter in question with the hope of, at some point, turning it into knowledge. Cursory reflection reveals that this is evidently false. (Himma 2007) Internet search engines are a good example. Type in a random search string and it will probably return hundreds of thousands of results. No human could sort through that amount of information and so the search engine is designed to return those results that are likely to be most beneficial to the user first. A great problem of the Information Age is our inability to keep the technology for sorting and filtering relevant information apace with the rapidly developing technology for collecting information. This is a familiar problem for anyone tasked with making use of any of the many web search engines out there. Access is almost always there, but relevancy is sporadic and limited. Thus, in order to deal with problems as they arise one needs to put constraints on what evidence and information is relevant. According to Neta, the skeptic unduly restricts evidence in certain contexts. What information 11 management effectively does is make the same judgments about appropriate restrictions. Dretske’s account is primarily an account of perceptual knowledge and information. He therefore feels entitled to conclude that, since the mere appearance of an object cannot communicate its non-skeptical status, any signal which carries information about appearance cannot answer a skeptical doubt. However, we have provided examples (such as Bouwsma’s adventures and Zebra) where perceptual information does justify a skeptical hypothesis or a nonskeptical proposition. It would seem that Dretske is wrong to think that information is irrelevant to combat local skeptical scenarios. Agents can receive information (even if we think of information as non-factive) for dismissing such scenarios (once we do not limit their information to the bare visual scene). (Pritchard 2010) Whether Dretske is right about radical skeptical scenarios depends on whether information is ever factive. If it is always factive then Dretske has no need to deny closure. Even if information is only sometimes factive (i.e., in ordinary contexts, à la Neta) then Dretske is still wrong. Let us consider an argument that reasons (under which heading we may include perceptual evidence) are factive, which is from John McDowell (1995). Earlier in the paper, we discussed Neta’s comment that external world skepticism is not meant to cast doubt upon certain ‘inner’ reasons such as ‘that I am not having a visual experience of a white expanse before me’. McDowell argues against a tacit assumption throughout epistemology that these inner reflections can encompass factive empirical reasons. (Pritchard 2008, 10) However, McDowell does not think that no empirical reasons are factive. In the case of veridical perception, we have a kind of perceptual evidence which is not present in cases of non-veridical perception such as illusion or hallucination. McDowellian epistemological disjunctivism presents an option for Dretske which has 12 so far been left unexplored but which may undermine his case against epistemic closure, with concomitant implications for his theory of information. In brief, if perceptual evidence is (sometimes) factive, then Dretske is wrong to say that there is no perceptual evidence which can serve as evidence against skeptical hypotheses. Dretske’s view is that all perceptual evidence is defeasible when it comes to radical skeptical hypotheses. No matter how competently one receives and judges the information one is presented with, these processes never amount to something which entails the denial of the target sceptical hypothesis. The view is intuitive and persuasive but the McDowellian view offers one alternative: that there is a disjunct between cases of factive and non-factive reasons. That is, there is some reason or warrant or a kind of support missing in cases of radical skepticism that is present in so-called ‘ordinary’ cases. Dretske takes it for granted that any given knowledge claim can be subject to a skeptical rebuttal. Such rebuttals challenge the upgrading of an information-based belief (that something appears to be the case) to information-based knowledge (knowledge that something is the case). In the case of Zebra* there is information that carries the signal to Fred that what is in the pen is a painted mule. Dretske might insist that this does not undermine his thesis as these pieces of information may themselves be subject to skeptical hypotheses and are providing only non-factive evidence. However, if one follows McDowell down his disjunctivist path then it is not inevitable that Dretske takes such a position and consequently not inevitable that he is lead to reject the principle of epistemic closure. Neta presents a contextualist account of evidence or reasons in which the evidential requirements for knowledge are affected by context. Dretske closely links information to non-factive evidence but under the contextualist account there are cases of factive evidence which would provide information-based knowledge of the denials of 13 skeptical hypotheses in some cases. Additionally, McDowell provides a non-contextualist account of evidence or reasons in which there is an epistemic component present in some cases, not present in others (such as cases of hallucination or illusion—the hallmark of skeptical hypothesizing), and in which factive evidence warrants the denial of skeptical hypotheses. As a consequence, these distinctions between skeptical and ordinary contexts or between factive and non-factive evidence present alternatives to Dretske’s inference that perceptual information can never give us evidence or reasons to refute skeptical hypotheses. At the beginning of this paper we described a scenario in which a patient may seek information as a means to gaining knowledge about a medical matter. If information such as this were always susceptible to skeptical challenges then this susceptibility would be uncomfortably passed on to the knowledge claims based upon the evidence it carries. Such worries caused Dretske to abandon a key principle explaining how we reliably expand our knowledge: epistemic closure. We have presented an alternative epistemological picture here which does not have such drastic consequences. REFERENCES Bouwsma, O. K. (1965). ‘Descartes’ Evil Genius’, Meta-Meditations: Studies in Descartes, (eds.) A. Sesonske & N. Fleming, Belmont, Calif.: Wadsworth. Choo, C. W. (2002) Information Management for the Intelligent Organization: The Art of Scanning the Environment 3rd ed., Medford, New Jersey: Information Today, Inc. DeRose, K. (1995). ‘Solving the Skeptical Problem’, Philosophical Review 104, 1-52. Dretske, F. (1970). ‘Epistemic Operators’, Journal of Philosophy 67, 1007-23. —— (1971). ‘Conclusive Reasons’, Australasian Journal of Philosophy 49, 1-22. 14 —— (1981). Knowledge and the Flow of Information, Cambridge, MA: MIT Press. —— (1983). ‘The Epistemology of Belief’, Synthese 55.1, 3-19. —— (2000). ‘The Pragmatic Dimension of Knowledge’, in his Perception, Knowledge and Belief: Selected Essays, Cambridge: Cambridge University Press. —— (2005a). ‘The Case Against Closure’, Contemporary Debates in Epistemology, (eds.) E. Sosa & M. Steup, 13-26, Oxford: Blackwell. —— (2005b). ‘Is Knowledge Closed Under Known Entailment?’, Contemporary Debates in Epistemology, (eds.) E. Sosa & M. Steup, 13-26, Oxford: Blackwell. —— (2005c). ‘Reply to Hawthorne’, Contemporary Debates in Epistemology, (eds.) E. Sosa & M. Steup, 43-6, Oxford: Blackwell. —— (2006). ‘Information and Closure’, Erkenntnis 64, 409-13. Gomes, A. (Forthcoming). ‘McDowell’s Disjunctivism and Other Minds’, Inquiry. Fallis, D. (2002). ‘Introduction’, Social Epistemology and Information Science, special issue of Social Epistemology, 16.1, 1-4. —— (2004). ‘Epistemic Value Theory and Information Ethics’, Mind and Machines 14.1, 101-117. Fallis, D. & Whitcomb, D. (2009). ‘Epistemic Values and Information Management’, The Information Society 25.3, 175-189. Floridi, L. (2005). ‘Is Semantic Information Meaningful Data?’, Philosophy and Phenomenological Research 70.2, 351-370. —— (2010). The Philosophy of Information, Oxford: Oxford University Press. Goldman, A. (1976). ‘Discrimination and Perceptual Knowledge’, The Journal of Philosophy 73, 771-91. —— (1999). Knowledge in a Social World, Oxford: Oxford University Press. Harms, W. F. (1998). ‘The Use of Information Theory in Epistemology’, Philosophy of Science 65.3, 472-501. Hawthorne, J. (2005). ‘The Case for Closure’, Contemporary Debates in Epistemology, (eds.) E. Sosa & M. Steup, 26-43, Oxford: Blackwell. Himma, K. E. (2007). ‘The Concept of Information Overload: A Preliminary Step in Understanding the Nature of a Harmful Information-Related Condition’, Ethics and Information Technology 9, 259–72. Jäger, C. (2004). ‘Skepticism, Information, and Closure: Dretske’s Theory of Knowledge’, Erkenntnis 61.2-3, 187-201. McDowell, (1995). DHP to add ref Shope, R. K. (2002). ‘Conditions and Analyses of Knowing’, The Oxford Handbook of Epistemology, (ed.) P. K. Moser, 25-70, Oxford: Oxford University Press. 15 Neta, Ram. (2002). ‘S Knows that P’, Noûs 36, 663-81. —— (2003). ‘Contextualism and the Problem of the External World’, Philosophy and Phenomenological Research 66, 1-31. Nozick, R. (1981). Philosophical Explanations, Cambridge, MA: Harvard University Press. Pritchard, D. H. (2008). ‘McDowellian Neo-Mooreanism’, New Essays on Disjunctivism, (eds.) A. Haddock & F. Macpherson, 283-310, Oxford: Oxford University Press. —— (2009). ‘Wright Contra McDowell on Perceptual Knowledge and Scepticism’, Synthese 171, 467-79. —— (2010). ‘Relevant Alternatives, Perceptual Knowledge and Discrimination’, Noûs 44, 245-68. Shackel, N. (2006). ‘Shutting Dretske’s Door’, Erkenntnis 64, 393-401. NOTES 1 It should be noted that gathering, creating, processing, managing and using information is not always done for the acquisition of knowledge or other epistemic standings. Sometimes, for example, information is collected for the sake of collecting more information or for justifying policy decisions. Nevertheless, the kind of information-based inquiry we explore here is that which is pursued with the final purpose of gaining knowledge about the matter at hand. This is the kind of inquiry pursued in Dretske (1981) and Floridi (2010), among others. These scholars accordingly view information as, in their own distinctive ways, an important component of epistemology. 2 See, for example, Fallis (2004) and Goldman (1999, 161-82). 3 The barn-façade case was first put forward in print by Goldman (1976), who credits the example to Carl Ginet. 4 For Dretske’s initial rejection of epistemic closure, see Dretske (1970; 1971). See also his recent exchange with Hawthorne (Dretske 2005a; 2005c; Hawthorne 2005). 5 Although there are few philosophers these days who deny this principle, it was also famously denied by Nozick (1981), for reasons very similar to the reasons offered by Dretske. 6 For an extended discussion of the goal of information collection and dissemination see Fallis (2002). Note that even those who deny that the goal of information services is for users to acquire knowledge grant that in a large range of contexts our goal in collection and disseminating information is to acquire knowledge. For example, the information management scholar Chun Wei Choo expresses, albeit in different terms, a widely held view when he states that the primary goal of information management is to ‘harness the information resources and information capabilities of the organization in order to enable the organization to learn and adapt to its changing environment.’ (Choo 2002, xv) Later, Choo writes that the ‘transfiguration of information into knowledge is the goal of information management.’ (Choo 2002, xiv)