Epistemic Justification.  By Richard Swinburne. (Oxford: Clarendon Press, 2001.  Pp. viii + 262.  Price ???)   [Forthcoming in The Philosophical Quarterly]

Readers of Swinburne’s rewarding book will get a glimpse from the inside of how a sophisticated doxastic foundationalist understands epistemic justification.  What, you may ask, is doxastic foundationalism?  Well, generic foundationalism, as Swinburne understands it, is the view that some justified beliefs are basic, i.e., not grounded on or based on other beliefs.  Some versions of foundationalism allow that basic beliefs can be justified in virtue of their being based on experiences.  But doxastic foundationalism holds that all justified basic beliefs are “ones we are justified in holding without their needing other beliefs or other mental states as grounds, in the sense that intrinsically or merely in virtue of our having them, they are probably true” (p. 26).

     I say above that this book gives us a glimpse, rather than a defense, of a sophisticated version of doxastic foundationalism.  It’s not that Swinburne gives us nothing in defense of his position.  But the tone of the book is expository and conciliatory rather than polemical.  His purpose seems to be to explain his epistemological views, not to take on all those who disagree with him, providing knockdown objections to their positions while defending his own against criticisms.  He does take issue, in some ways, with coherentist positions and with foundationalist positions (especially externalist ones) contrary to his own.  But he believes that competing accounts of justification and knowledge are, for the most part, analyzing different concepts related to different uses of the words ‘justified’ and ‘knowledge’ (p. 2).  Hence, there is not, he thinks, much genuine disagreement between internalist and externalist foundationalists.  Even so, he does address the question (central in chapters 6 through 8) of which sorts of justification and knowledge are worth having.  According to Swinburne, most of the various sorts of justification and knowledge (externalist and internalist) are good things to have (p. 190).  But only objective internalist justification and the sort of knowledge that requires it are intrinsically valuable (pp. 163-64,185-86, 219-20).  If the justification and knowledge are of an externalist sort, then, says Swinburne, justified true belief and knowledge are no more worth having than mere true belief (pp. 185, 219).

     The book divides into two main parts.  The first three fifths of the book (chapters 1 through 5) are devoted to explaining the concepts to be used in the second part of the book where he compares various theories of justification and knowledge.  The most important of these concepts is probability, which takes up about half of that first part of the book (all of chapters 3 and 4). Following Ian Hacking and others, Swinburne notes that since the seventeenth century, two main kinds of probability have been identified: probability as “a feature of the physical world”, which Swinburne calls ‘externalist probability’, and “probability on evidence that something was the case in the physical world”, which he calls ‘inductive probability’ (pp. 61, 71). He goes on to distinguish three varieties of externalist probability (actual statistical, hypothetical statistical and physical) and three varieties of inductive probability (logical, epistemic and subjective).  Chapter 4 (on the correct criteria of logical probability) is the longest in the book.  Here he pays special attention to the two a priori criteria for determining the logical probability of one proposition on another, namely, scope and simplicity (the narrower the scope of the proposition and the simpler it is, the more likely it is to be true).  In the second and fifth chapters Swinburne focuses on belief and basicality respectively.  According to him, our beliefs have narrow, not wide, content and we have infallible access to them.  In discussing basicality, he offers accounts of the basing relation and of what counts as our total available evidence.

      Having completed this preliminary work, Swinburne turns, in the second part of the book, to the task of comparing and evaluating various internalist and externalist accounts of justification and knowledge.  In chapter 6 he discusses synchronic justification (the sort of justification that has to do only with the subject’s response to her current evidence), in chapter 7 the focus is diachronic justification (the sort that is determined in part by the adequacy of the investigation that produces the subject’s current evidence) and in chapter 8 it is knowledge.  As I noted earlier, his main thesis in these chapters is that only internalist sorts of justification and knowledge are intrinsically valuable.

     In order to engage Swinburne’s book critically, it will be helpful to examine some aspect of it in greater detail.  Thus, for the remainder of this review, I will focus on something that plays a pivotal role in Swinburne’s doxastic foundationalism, namely, his account of what he calls ‘rightly basic beliefs’.  Rightly basic beliefs are basic beliefs in propositions that have a prior logical probability greater than 0.5 (p. 145).  Two factors affect the prior logical probability of a proposition: its intrinsic probability (which is its probability on a tautology, determined in the case of contingent propositions by simplicity and scope) and its self-probability (which is the probability it gets in virtue of being believed – the higher the confidence in the belief, the higher the self-probability).  For noncontingent propositions, whatever self-probability they get by virtue of our believing them is totally overwhelmed by their intrinsic probability.  Thus, no matter how strongly we believe or disbelieve them, the logical probability of noncontingent propositions is 1 or 0 (p. 147).  The intrinsic probability of contingent propositions tends to be very low.  But their self-probability plays a significant role in determining their logical probability, either lowering it even further or raising it, in some cases close to 1, as when one is very confident in some basic belief in a contingent proposition (pp. 148-49). 

     The reason Swinburne thinks that self-probability can increase the prior logical probability of contingent propositions is this. He thinks that justification is determined by logical probability; yet clearly, memory and perceptual beliefs are justified despite the fact that their contents have a low intrinsic probability; so they must get a boost in logical probability from somewhere else (p. 149).  But why does Swinburne think this boost in logical probability comes by virtue of our believing the propositions?  First, he thinks that if S believes p in the basic way, then S will (if she thinks about it) believe that her belief that p is forced upon her by the world; and “the very fact of my believing that some proposition is being forced upon me by my experience or by the deliverance of reason is as such reason for me to believe it to be true” (p. 140).  Why think that?  Because, says Swinburne, “[i]f that were not so, I would never have a good reason ... for believing anything” (p. 141).

     This account faces a number of difficulties that Swinburne doesn’t adequately address.  First, there are the problems that arise from his claim that perceptual beliefs get none of their justification from experience.  According to Swinburne, their justification comes from the logical probability they have in virtue of our holding them.  If this were right, then our experience could be completely different or absent altogether and yet the logical probability of the very same perceptual beliefs (and, therefore, any justification they get in the basic way) would remain the same so long as we maintained the same high confidence in them.  But that is quite implausible.  Second, Swinburne seems to think it is important, for justification, that our beliefs have reasons.  And in seeking to identify a reason for the basic belief p, he points to the fact that we would (if we thought about it) believe that q, where q is the proposition that the belief that p is forced upon me by the world.  But if the belief that p is justified in virtue of our having (or being able to have) a reason, namely, the belief that q, why think it is a basic belief?  Furthermore, one wants to ask if the belief that q needs to be justified itself in order to function as a justification-producing reason.  One would think that it must.  But then we are off on a regress because it too will need a reason that will itself need a further reason, and so on.  Why not conclude instead, as foundationalists typically do, that a belief can be justified even if we don’t have a reason for it?  Third, why conclude from the fact that the intrinsic probability of the contents of our memory and perceptual beliefs is low that there must be some other source of logical probability for them?  Why not instead take that fact as evidence for thinking that the justification of basic beliefs is not determined by the logical probability of their contents?  Finally, why think that Swinburne’s account of justification in terms of logical probability is an internalist account?  Even if a logically omniscient person will have privileged access to which of her basic beliefs are logically probable, most people won’t for, the simple reason that they won’t always have privileged access to which of their beliefs are in contingent propositions or to which of their beliefs in noncontingent propositions are true.

     Many more questions than these are raised by Swinburne’s doxastic foundationalism.  But anyone who, like myself, is inclined to think it very implausible that all intrinsically valuable justification is determined largely by logical probabilities can’t simply dismiss the idea without digesting and responding to Swinburne’s careful and meticulous exposition of a sophisticated version of that position.


Purdue University                                                                         MICHAEL BERGMANN