Abstract
Can one gain testimonial knowledge from unsafe testimony? It might seem not, on the grounds that if a piece of testimony is unsafe, then any belief based on it in such a way as to make the belief genuinely testimonial is bound itself to be unsafe: the lack of safety must transmit from the testimony to the testimonial belief. If in addition we accept that knowledge requires safety, the result seems to be that one cannot gain testimonial knowledge from unsafe testimony. In a pair of recent papers, however, Sanford Goldberg has challenged this apparently plausible line of thought. Goldberg presents two examples intended to show that a testimonial belief can be safe, even if the testimony on which it is based is unsafe: the lack of safety need not transmit from the testimony to the testimonial belief. In this paper, I question whether Goldberg’s examples really do show that one can gain safe testimonial belief from unsafe testimony. The problem, I explain, is that both examples appear (for different reasons) to be open to objection. Nevertheless, I argue that although Goldberg’s examples do not establish his conclusion, the conclusion itself is true: one can gain safe testimonial belief from unsafe testimony. I base my argument on an example which differs in structure from Goldberg’s examples, and I argue that due to this difference, my example avoids the problems which Goldberg’s examples face.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Lackey (2008, ch. 3.3).
Nozick (1981, p. 179).
For further discussion of the way in which the safety theorist should conceive of bases for belief, see Williamson (2009).
Goldberg (2005, p. 302).
Lackey (2008, ch. 3.3).
I take the table above from Lackey (2008, p. 82).
To be clear, the issue here is not simply that Sonny’s belief depends for its safety on an act of credibility-monitoring. This would not in itself be a reason to classify his belief as non-testimonial, I take it, since many uncontroversially testimonial beliefs depend for their epistemic credentials on credibility-monitoring of a certain standard type, i.e. that which hearers themselves usually perform. The issue is rather that in this case, Sonny’s belief depends for its safety not merely on such standard credibility-monitoring, but also on an additional act of credibility-monitoring of a non-standard type, performed by a third party. This is why Sonny’s belief does not count as testimonial, in Lackey’s view.
Goldberg (2007, p. 322).
We can imagine that since Joan is highly attentive, there is no close possibility in which she fails to notice Ralph’s less confident tone of voice. We can also imagine that since the newspapers Ralph reads are carefully edited, there is no close possibility in which they contain a misprint.
I assume here that in at least some of the close possibilities where something other than Randy made the rustle, Randy did not happen coincidentally to be trespassing at the time.
Might one object that there is a difference between the bases of Fred’s actual and counterfactual assertions, namely that while his actual assertion is based on a rustle made by Randy, his counterfactual assertion is based on a rustle made by something else? I do not find this objection convincing. The problem is that if we count Fred’s testimony in this case as safe, given the external difference between the bases of his actual and counterfactual assertions, it seems we will have to count a very large proportion of true testimony in general as safe, since in a very large proportion of such cases, there will be some similarly external difference between the bases of the actually true and counterfactually false testimony. But this would rob the notion of safe testimony of much of its interest, since the effect would be largely to erode the distinction between safety and truth. The same point arises at the level of belief. Suppose we count Fred’s belief that Randy is trespassing as safe, on the grounds that there is an external difference between the bases of his actual and counterfactual beliefs. If so, then we will have to count a very large proportion of true beliefs in general as safe. But this would mean, among other things, that the safety principle for knowledge would have very little explanatory work to do, since very few true beliefs would fail to meet the constraint it imposes on knowledge. The general moral is that if we choose to conceive of bases to some degree externalistically (as e.g. Williamson (2009) does), we should bear in mind that there must be some limit to the sorts of external differences between bases that ought to be counted as relevant to safety. If one ignores this point one risks trivializing the notion of safety.
Some further suppositions which we can and should make about the example, to confirm that Randy’s belief is safe: i) there is no close possibility in which Fred asserts what he does, but in a quite different location, e.g. near the edge of his land; ii) there is no close possibility in which the range of Fred’s land is much less than it actually is; iii) there is no close possibility in which the carry of Fred’s voice is much greater than it actually is.
We can imagine variants of TERRITORIAL FARMER in which this is not so. For example, we can imagine that Randy infers from Fred’s assertion that Fred must know that whoever is in the bushes is trespassing; we can imagine that Randy then further infers, given his own belief that he is the one in the bushes, that he is trespassing. In such a case, Randy’s belief would not count as testimonial in the strict sense at issue, since it wouldn’t be based purely on Fred’s testimony. However, the key point for our purposes is that although we can fill out the details of TERRITORIAL FARMER in this way, we are not obliged to. The simpler case described in the main text is surely quite coherent.
References
Audi, R. (1997). The place of testimony in the fabric of knowledge and justification. American Philosophical Quarterly, 34, 405–422.
Comesaña, J. (2005). Unsafe knowledge. Synthese, 146, 395–404.
Goldberg, S. (2005). Testimonial knowledge through unsafe testimony. Analysis, 65, 302–311.
Goldberg, S. (2007). How lucky can you get? Synthese, 158, 315–327.
Graham, P. (2000). Conveying information. Synthese, 123, 356–392.
Lackey, J. (2008). Learning from words: testimony as a source of knowledge. Oxford: Oxford University Press.
Neta, R., & Rohrbaugh, G. (2004). Luminosity and the safety of knowledge. Pacific Philosophical Quarterly, 85, 396–406.
Nozick, R. (1981). Philosophical explanations. Oxford: Oxford University Press.
Pritchard, D. (2005). Epistemic luck. Oxford: Oxford University Press.
Sosa, E. (1999). How to defeat opposition to Moore. Philosophical Perspectives, 13, 141–153.
Sosa, E. (2000). Skepticism and contextualism. Philosophical Issues, 10, 1–18.
Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.
Williamson, T. (2009). Reply to Goldman. In P. Greenough & D. Pritchard (Eds.), Williamson on knowledge. Oxford: Oxford University Press.
Acknowledgments
Many thanks to the editor and to an anonymous referee for their helpful comments on an earlier draft.