Combes on McCraw on the Nature of Epistemic Trust: A Rejoinder, Benjamin W. McCraw

Author Information: Benjamin W. McCraw, University of South Carolina Upstate,

McCraw, Benjamin W. “Combes on McCraw on the Nature of Epistemic Trust: A Rejoinder.” Social Epistemology Review and Reply Collective 5, no. 8 (2016): 28-31.

The PDF of the article gives specific page numbers. Shortlink:

Please refer to:


Image credit: Marius Brede, via flickr

My genuine thanks to Richard Combes for continuing his thoughtful analysis of my views on epistemic trust. In this really short reply, let me offer a quick re-rejoinder to a few of his latest comments.

Combes on Trust-In and Trust-That

First, let’s get clear on Combes’ view. He claims that “one epistemically trusts S if and only if one has certain beliefs about S’s thick reliability” (2016, 8) where ‘thick reliability’ refers to the state where “one has consciously tracked S’s past history, judged that S enjoys some perhaps unique expertise, and therefore should depend on s’s testimony…” (8). That is, H trusts S just in case H believes that:

(a) H has tracked S’s history with respect to the accuracy of S’s utterances,
(b) S’s track record is reliable and
(c) H should depend on S’s future assertions. 

I assume that the dependence in (c) simply means that H will believe what S says in the future. Otherwise, there will be an element of non-propositional belief in Combes’ analysis; much to his chagrin, I take it. However, I reject this analysis: my account of epistemic trust includes elements (e.g. depending on S, being confident in S, and taking S to communicate) that are not necessarily identical to having propositional beliefs about S.[1] I want to urge two ways I think Combes’ identity thesis finds itself in trouble.

Worry #1

It’s not all that clear just why one should accept the identity thesis on Combes’ view. I can see, I think, two possible routes Combes takes to get there. One may come to this identity thesis by “introspective data” (2016, 8); i.e. to reflect on one’s own instances of trusting to see what they involve. Also, one may think that the identity thesis is true because “one cannot take another to be authoritative and epistemically well-positioned in a vacuum, without a reason” (9). Unfortunately, I don’t see either route as all that attractive. “Introspective data” doesn’t strike me as a very promising method of conceptual analysis unless wedded to some views about both the reliability and person-independent universality of one’s introspection.

It’s quite plausible, on my own introspection at least, to think that one’s introspections needn’t hold across a range of many/all rational agents and they don’t certainly seem to be infallible sources for analysis (we just might have bad internal data). And, while I agree that trust must be based on reasons, I’m not sure that implies the identity thesis. Perhaps all reasons for trust are propositions belief but perhaps not, too. We agree that trust doesn’t occur ex nihilo, but I’m not convinced the grounds/reasons for trust must be only propositional beliefs. Accordingly, this is why my analysis uses the broader language of “taking” someone to communicate something as the basis of trust.[2] It seems possible that such “takings” need not be the same as having a propositional belief. At least, Combes should give us an argument to think that any possible reason or ‘taking’ must be the same as a propositional belief about the trustee. Summing up my points, what Combes needs, I’ve argued here, is (1) some argument(s) that introspective data will be the sort of thing that gets one a good analysis of trust (or any important concept, really) and/or (2) some argument that one’s reasons for trust must be propositional belief(s).

Worry #2

I think Combes’ identity thesis is susceptible to counterexample. H Hadf;kjasdfl;sdjl;sdfsdfjkl;asdflI argue in Worry #1 that Combes’ hasn’t really established his account of epistemic trust, but this point goes further to suggest the identity thesis is false. And the reason why is that one can satisfy (a)-(c) above and still fail to trust in someone. Consider his Oprah Winfrey case: suppose both Combes and I trust her, and I’m a reliable repeater of Winfrey’s claims. Yet this won’t suffice for trust in me, on his own view: I “may be reliable insofar as [I] likewise accept her allegedly accurate testimony, and as her mouthpiece report this to the general public, but it is Winfrey [Combes] immediately trust[s], not [me]” (8). In this case, Combes knows that I am Winfrey’s ex hypothesi reliable mouthpiece, judges that I’m reliable, and should believe what I say in the future (assuming I continue to parrot Winfrey’s perfectly truth-apt utterances). It seems, then, that I satisfy all of Combes’ conditions for trust above and yet he, correctly to my lights, denies this as a case of trust in me. If so, then epistemic trust isn’t equivalent to (a)-(c) above.

Objects of Trust

I want to clear up a point of confusion about the structure of trust. Combes suggests that I claim that “an action (e.g. my wife picking up milk from the store) is an object of trust”: a view Combes thinks is “incoherent, both on logical and lexicographical grounds” (7). I certainly don’t think that one can trust an action. On my view, trust-in always has a tripartite structure: H trusts in S for some X. (I think this holds for all cases of trust: not just the epistemic variety.) What I suggest is that an action can occupy the “X” above but not the H or S. Trusters and trustees are always agents/persons or something that we can reasonably take to be an agent (e.g. a GPS). It is incoherent to trust in an action but, I submit, it’s not incoherent to trust in a person to commit an action. It’s only in this sense that I think something like an action can fit into the structure of trust.

I’m sure that Combes won’t agree here, but that disagreement is probably just a symptom of our deeper disagreement about the nature of trust vis-à-vis trust’s status as a purely propositional attitude. However, there’s a way to preserve some kind of tripartite analysis for a view like Combes endorsing the identity thesis. For him, S trusts in H just in case there’s some p about S’s reliability that S believes. Now, this formulation replaces an explicit proposition—p—with my admittedly vague placeholder—X. (I use the ‘X’ so that it can stand for propositions, actions, attitudes, etc.) However, the structural fix is easy: all one must do is stipulate that the p in the formulation above may describe an action (of either S or H). So, Combes could simply say that I trust my wife to get milk from the store if and only if I believe: “my wife will get milk” on the basis of my judgment of her past reliability. Combes and I may still disagree deeply about the nature of trust, but I think these considerations move us closer together, at any rate.

Reliability Through Thick and Thin

Combes’ (8) distinction between thin/nominal reliability and thick/robust reliable is clever, nuanced, and helpful (in my assessment) to making his trust-in/trust-that identity view more plausible. Thin reliability is simply believing S’s utterances due to S’s own credibility (as a speaker). The robust version of reliability, recall from above, requires the truster to consciously judge S to be credible and to be so on the basis of the truster’s own intellectual field work in forming this judgement. I take this to be the part of Combes’ analysis meant to capture the point that trust must be based on reasons. Here, the reasons must be explicit propositional belief(s) and consciously considered—unlike the nominal version that will not suffice for trust. Combes’ “thickened” account certainly moves him closer to my own account, but it does lead to a (potential) problem for him.

Above, I’ve argued that there are cases of non-trust that satisfy his analysis: breaking his biconditional from right to left. His move to make thick/robust reliability the operative reliability in trust, however, has the reverse problem: it seems that there are cases of trust that don’t satisfy his more robust conditions. In this case, epistemic trust isn’t equivalent to (a)-(c) from the left to right hand side.

Consider a mundane case of trusting a stranger, say, for the time or directions. If you ask a stranger, “what time is it”, and believe her words, it seems natural (intuitive, reasonable on introspection, etc.) to think that you trust her for your newly-formed belief that it is X o’clock. Yet you don’t satisfy one of Combes’ conditions; namely, you don’t believe her on the basis of having “consciously tracked [her] past history”, primarily because you are completely unaware of her past history. Now, I think this case is a clear trust in another person. Assuming so, then Combes’ definition will falsely exclude this as a genuine case of trust.

However, Combes can suggest, instead, that this really isn’t trust or it’s something more like the trust-like behavior, absent real trust, displayed by children. That response, personally, sounds implausible to me, but I’m not sure that there’s anything beyond a see-saw of intuitions (introspections?) that can help ameliorate this possible disagreement. It’s possible (probable?) that Combes’ identity thesis, trust-as-belief-that view inclines him to reject cases that this as genuine trust and my rejection of such a view might color my assessment that such mundane beliefs are grounded in trust. If so, then I’m not sure how to motivate a way past our theoretical differences in this particular context.

However, if one does agree that this scenario does provide a real instances of epistemic trust, it seems that even Combes’ robust view of reliability underpinning his analysis of trust won’t work. Cases like this, of let’s call it “easy trust”, are too easy and with too little conscious reflection and/or belief to satisfy Combes’ analysis. But, for the sake of our theoretical disagreement, I’ll leave my objection as a conditional: IF cases of easy trust are genuine trust (rather than merely apparent trust or simply trust-like behavior), then Combes’ account of trust won’t analyze them correctly.

Again, I want to genuinely thank Combes for his thoughtful and charitable engagement with my work and our previous discussion(s). And I thank the SERRC for the opportunity to keep our dialogue going. I’ll end with what, I think, is our most important agreement: the need to move from a simple analysis of epistemic trust to a further account of when such trust is either well or poorly-placed. I vigorously agree that such an account is needed, important, and almost certainly philosophically interesting. However, the analysis of X is the first place to begin the longer, richer story of when X is good/bad and that’s all I’m attempting *now*. The normative question is a topic for a later day, but it’s a day I hope comes quite soon indeed.


Combes, Richard. “McCraw on the Nature of Epistemic Trust—Part II.” Social Epistemology Review and Reply Collective 5, no. 7-10 (2016): 7-10.

McCraw, Benjamin W. “The Nature of Epistemic Trust.” Social Epistemology 29, no. 4 (2015): 413-430.

[1] For reference, my (2015, 425) argues that H places epistemic trust in S that p IFF:

(1)H believes that p;
(2)H takes S to communicate that p;
(3)H depends upon S’s (perceived) communication for H’s belief that p; and
(4)H sees S as epistemically well-placed with respect to p.

[2] See my (2015, 421) for the discussion of “taking” someone to communicate something.

Categories: Critical Replies

Tags: , , , ,

Leave a Reply