ISSA Proceedings 2006 – The Ad Verecundiam Fallacy And Appeals To Expert Testimony

logo  20061. Introduction.
Much recent work in epistemology focuses on the role of testimony in generating warranted beliefs. One of the main views about this topic is known as non-reductivism. This view involves the idea that warrant by appeal to (expert) testimony does not involve inductive reasons that support belief in the reliability of the source in question. Tyler Burge’s particular version of this view (Burge 1993) is based on an a priori principle that all testimony is (at least defeasibly) probative, even to the degree that will sometimes qualify those sorts of beliefs as knowledge. Moreover, on Burge’s view, the kind of warrant that this principle imparts on certain beliefs is externalist in nature and so in forming warranted beliefs on the basis of testimony it is neither necessary that a believer know that the source of that testimony is reliable nor is it necessary that the believer know of and/or understand the (a priori) principle that Burge claims is sufficient often to warrant those beliefs. Finally, such warrant is supposed to be a priori in nature. In this paper it is argued that Burge’s view fails to provide resources sufficient to make an adequate distinction between fallacious ad verecundiam appeals to authority and legitimate appeals to authority and so Burge’s epistemology of testimony is deficient in this respect.

2. Ad verecundiam arguments.
The standard approach to the informal fallacies is to treat them as sorts of deficient arguments, most often as deficient deductive arguments. In line with this idea, ad verecundiam arguments have been most often understood to have the following sort of form (A1):

P1. A states that p is true.
P2. A is not an expert with respect to p.
C1. p is true.

Similarly, legitimate appeals to expert testimony are supposed to have the following form (A2)[i]:

P3. A states that p is true.
P4. A is an expert with respect to p.
C2. p is true.

Ad verecumdiam arguments would then seem to be best defined as invalid deductive inferences that involve appeals to inappropriate authority or appeals to the testimony of non-experts. What is epistemically important about A2-type inferences is that, when valid, they are justification preserving, and so if one is justified in accepting the premises then one is justified in accepting the conclusion. Recognition of this point highlights the error involved in A1-type inferences. Such inferences are bad because the premises do not support the conclusion such that were one justified in accepting the premises, then one would be justified in accepting the conclusion.
Of course there has been considerable debate about the adequacy of A1 as the proper analysis of appeals to inappropriate authority on a number of fronts, the two most important of which are (1) whether such arguments should be understood to be deductive in nature and (2) whether appeals to expert testimony are arguments at all (see Walton 1997, ch. 4). The first of these issues will not be addressed here, as it is largely tangential to the main point of this paper. However the second issue, the issue of whether or not fallacious and legitimate appeals to non-expert testimony are arguments is an important issue in the context of the general epistemic significance of appeals to authority. More specifically, the issue of whether appeals to expert testimony are arguments at all is an important for both the evaluation of the adequacy of certain epistemological approaches to testimony and for the evaluation of the adequacy of A1 and A2 as the standard analyses of the logical and epistemic features of appeals to authority. In any case, we can begin by noting a few things about the nature of quasi-formal treatments of fallacies that will ultimately be relevant to this discussion.
Recall that from the more or less standard perspective of informal logic A1 is supposed to be a formal, or more properly a quasi-formal, analysis of canonical examples of everyday inappropriate appeals to expert testimony that we often see exhibited in advertising and elsewhere in our epistemic exchanges. As such, A1 is supposed to represent an important and generic logical cum epistemic reconstruction of a kind of case of reasoning that fails to adequately justify belief in the conclusion. Again, this is supposed to be because the premises do not entail the conclusion. In this case this it is because P1 and P2 are supposed to fail to be relevant to the truth of C1. This is why ad verecundiam arguments are supposed to be a species of the fallacies of relevance and so are typically reconstructed as a kind of deficient argument.

Given this general understanding, the more or less informal logical analysis of evidential appeals that is part and parcel of quasi-formal logic is simply a kind of epistemological reconstruction aimed at explicating the basic logical structure of garden-variety attempts to justify certain beliefs by appeal to reasons and simple logical rules. At this point, in order to make things a bit more clear and concrete, it will be useful to examine some detailed cases. So let us consider the following two wholly typical sorts of appeals to testimony where the bracketed information describes relevant contextual factors:

(E1) Gary Neville says that Amanita phalloides is deadly. [Gary Neville is a famous Manchester United defender and a member of the English national team].[ii]

(E2) Gary Lincoff says that Amanita phalloides is deadly. [Gary Lincoff is the author of Toxic and Hallucinogenic Mushroom Poisoning: A Handbook For Physicians and Mushroom Hunters. In that book it is explained that Amanita phalloides, the death cap, is deadly because it causes cyclopeptide poisoning which is characterized by the following gruesome pathology:

(i) A long latent period of up to 1 day between the ingestion of the mushrooms prior to the onset of the first symptoms;
(ii) The occurrence of diarrhea, abdominal cramps, nausea and vomiting;
(iii) A 1 day period of remission of the symptoms noted in (ii), followed by
(iv) possible liver and kidney failure, and consequent death] (1977).

Notice that E1 is typically supposed to exhibit an A1-type structure and that E2 is supposed to exhibit an A2-type structure (Copi and Cohen 2004, p. 42-44). Moreover, there is supposed to be something importantly different about the epistemological situations described respectively in E1 and E2 that is revealed by the context of each utterance and so sanctions the difference in the logical reconstructions of these examples. So if we substitute the relevant detains of E1 and E2 into A1 and A2 respectively, then we get AE1 and AE2:

P’1. Gary Neville states that Amanita phalloides is deadly.
P’2. Gary Neville is not an expert with respect to Amanita phalloides.
C’1. Amanita phalloides is deadly.

P’3. Gary Lincoff states that Amanita phalloides is deadly.
P’4. Gary Lincoff is an expert with respect to Amanita phalloides.
C’2. Amanita phalloides is deadly.

Let us then focus our attention primarily on E1, A1 and AE1, while remembering that E2, A2 and AE2 are also important for the general critical points that will be raised later.

The first crucial question that arises here then is whether or not A1 is really a correct representation of the relevant epistemological and logical features of E1 and thereby whether AE1 adequately captures the essential logical features of E1. Notice that this is a question that depends on much more than the mere choice of a system of logical notation and it involves much more than the debate about which particular logical reconstruction is correct once a formalism is accepted, appearances perhaps to the contrary. More specifically, answering this question adequately depends on determining what is actually occurring in the epistemological situation described in E1and what we are clearly dealing with in such cases is the attempted testimonial transmission of information. As a result, disputes about the correctness of any particular logical analysis of ad verecundiam arguments, while important, must be regarded as secondary to our resolving the issue of the epistemological nature of testimony. The former issue is just the matter of the logical reconstruction of the latter sort of occurrence. Unfortunately, this is not at all comforting because the epistemological debate about the nature of testimony is itself a topic about which there is considerable disagreement, but nevertheless this modest conclusion is surely true. Assessing the correctness of any particular reconstruction of ad verecudiam argumentation requires resolving the epistemic problem of testimonial warrant, because ad verecundiam arguments are nothing more than failed appeals to testimony.

3. The Epistemology of Testimony.
There are currently two basic approaches to the nature of testimonial evidence: reductivism and non-reductivism. On the one hand, non-reductivism is just the view, derived from Reid, that testimony is a basic source of justification in the sense that it can generate justification and that such justification does not depend on knowledge of the frequency of veracity of testimony, or on any other empirical facts. The independence condition is crucial for those who accept this view because if testimony did require such additional knowledge, then it would be dependent on induction and thus would ipso facto not be a basic, justification-generating, source. On the other hand, reductivism is the view that testimonial justification requires knowledge of the frequency of the veracity of testimony and so on this view the justificatory status of testimony is parasitic on the justificatory status of induction. In the context of this paper we will be concerned only with non-reductivism and its implications for our analyses of ad verecundiam arguments. More exactly, we will be concerned here with the specific and influential version of non-reductivism about testimonial warrant defended by Tyler Burge (1993) and how this view fares with respect to the standard interpretation of those arguments sketched out above.

3.1 Burge’s View of Testimonial Warrant.
As has been noted here already, Burge defends a version of non-reductivism concerning testimony. As a result, for one to be warranted in accepting some item of testimony does not require one to know that the source of that testimony is reliable. Hence, on this view one can be warranted in accepting testimony without any knowledge of the degree of expertise on the matter in question possessed by the utterer of that testimony. This is largely a result of Burge’s more general claim that epistemology has too long been concerned with an overly intellectualized concept of justification (2003, 503-505). The sort of hyper-intellectualism that Burge identifies and objects to is just the sort of account of warrant that involves the possession of reasons that are mentally accessible to the knower in question and which exhibit inferential structure. Burge finds such views to be wildly implausible when applied to a number of areas of epistemic interest including both perception and, more importantly, testimony. Nevertheless, Burge does hold that such sources provide us with warranted beliefs.

In order to avoid this sort of over-inellectualization, the general basis of Burge’s view of the sort of warrant that is involved in testimony (and perception) is externalist in nature. Moreover, in the process of rejecting reductionism Burge also reveals that the sort of warrant that is involved in testimonial acceptance is a priori in nature. For Burge this means that the kind of warrant that is involved in testimony does not in any epistemic way depend on sense experience or perception (1993, p. 466-467). We are supposed often to be warranted in accepting testimony based on the mere satisfaction of the acceptance principle (SAP), the simple version of which states that:
A person is entitled to accept as true something that is presented as true and that is intelligible to him, unless there are stronger reasons not to do so (1993, p. 467).[iii]

Burge refers to the sort of warrant derived from principles of this sort as entitlement in order to distinguish this sort of warrant from the traditional reason-based sorts of justification characteristic of traditional hyper-intellectualized epistemology.

So based on SAP we are essentially entitled to accept as true anything uttered to us that is intelligible unless we possess some relevant defeater that pertains to that testimony or to the source of that testimony. As Burge would have it principles of this sort are simply norms the fulfillment of which has positive epistemic status because they are conceptually related to truth in some important manner (2003, p. 506-507). In the case of testimony this condition is understood to be satisfied because intelligibility is supposed to be conceptually related to rationality and thereby to truth in accordance with a kind of a principle of charity as it applies to rational discourse (1993). In other words, it is rational, and thus warrant-generating, to accept as true any intelligible testimony because truth-telling is the normative default position we should adopt with respect to intelligible interlocutors in the absence of any known defeaters. In line with this Burge formulates a more complex version of the acceptance principle (CAP) as follows:
A person is entitled to accept a proposition that is presented as true and that is intelligible to him, unless there are stronger reasons not to do so, because it is prima facie preserved (received) from a rational source, or resources for reason; reliance on rational resources or resources for reason ¾ is, other things equal, necessary to the function of reason (1993, p. 469).

Intelligibility is thus supposed to be a prima facie and a priori indication of rationality, and hence of truth.

Testimonial warrant is then supposed to be a priori because satisfaction of CAP requires no epistemic reference to sensations or perceptual states at all (1993, p. 472). Notice also that in rejecting internalism Burge is also committed to the view that we are entitled to accept many items of testimony even if we have no knowledge of CAP. He explains,
The Acceptance Principle is not a premise in an argument applied by the recipients of information. It is a description of a norm that indicates that recipients are sometimes entitled to accept information from other immediately without argument (1993, p. 476).
So the ultimate view that Burge defends is that we are warranted in accepting testimony without any argument at all, without any explicit or implicit reasons, because the acceptance of intelligible testimony in the absence of defeaters is always warranted. It is always warranted because it is the satisfaction of an a priori norm of reason that one need not even be aware of in order for this sort of a priori entitlement to obtain (1993, p. 467), often to the degree that it even constitutes knowledge (1993, p. 485).

4. Burge and Ad Verecundiam Arguments.
Given Burge’s account of testimonial warrant what then can we say about the distinction between E1 and E2 and their logical reconstruction as AE1 and AE2? First, note that absent the crucial contextual factors included as bracketed information in those cases, we could make no epistemic distinction between Gary Neville’s utterance and Gary Lincoff’s utterance on Burge’s view. Both Garys simply utter tokens of the sentence “Amanita phalloides is deadly.” As a result, one would be equally warranted in accepting the intelligible proposition that Amanita phalloides is deadly as true whichever person it came from. This is simply because in both cases one would be satisfying CAP, and so one would be a priori justified in believing that Amanita phalloides is deadly, irrespective of the contextual factors at work in E1 and E2. Now, of course, one might later learn that Gary Neville is no expert with respect to the matter of mycological toxicity and so that justification would be defeated by one’s doxastic possession of that reason. This just indicates that Burge accepts the view that a priori knowledge is defeasible when one is in doxastic possession of information in the form of an explicit reason that undercuts some item of testimony, and while the view that there is defeasible a priori knowledge is an awkward one Burge is not alone in accepting this view (2003, p. 506-507, Kitcher 1983). In any case, in the absence of such explicit defeating knowledge, the hearer is justified to the same degree in accepting that proposition when that information comes from Gary Neville as it would be had it come from Gary Lincoff. This is so because that justification derives exclusively from satisfying CAP and none of it is derived from making any inference, good, bad or otherwise.
Burge’s view is curious in this respect as in being an externalist about CAP itself he is perfectly happy to accept that one’s warrants for one’s beliefs need not be a function of doxastic items that one is aware of. So why should we not just apply the principle of charity here and simply accept that one’s justification for accepting that Amanita phalloides is deadly is a function of the, often unknown, contextual information in the context of an utterance that plays a role in an explicit or even implicit inference that we are making even though we might be unaware of that we are making such an inference? This would work equally well. The logical reconstructions of our epistemic circumstances need not only include what we are self-reflectively aware of and Burge himself accepts this general point in endorsing externalism with respect to CAP. So there seems to be no special reason why we should not simply retain this more traditional position, and there are also some good additional reasons why doing so is useful. Specifically, it allows us to make the important epistemic distinction between E1-type and E2-type situations even when we are not aware of the importantly different contexts (especially defeaters), and it preserves the standard logical model of appeals to testimony that allows for us to distinguish ad verecundiam and legitimate appeals to authority from a purely logical point of view.

This brings us to the second point of criticism. What is perhaps most interesting about Burge’s view is that, in accepting his particular account of testimonial warrant, Burge is committed to the view that neither AE1 nor AE2 is a correct analysis, respectively, of E1 and E2. This is the case because he explicitly claims that the transmission of information by testimony does not involve any argumentation at all. As a result, on Burge’s view, that one’s behavior conforms to an A1-type or an A2-type structure cannot have anything to do with justification and testimony, pace the standard understanding of ad verecundiam argumentation. So, importantly, Burge must hold that E1 does not involve any fallacy at all, because it does not involve any argument whatsoever. As a result, there is no way that one can say that defective reasoning conforming to A1 is occurring in E1 on Burge’s view. Hence, if we accept that informal logic is a tool by which we reconstruct the basic logical features of everyday discourse, Burge cannot possibly offer any account of ad verecundiam argumentation, at least as it is understood as per typical textbook treatments of that fallacy. As a result, ad verecundiam arguments cannot really be adequately distinguished from legitimate appeals to authority on his view and we cannot really say that anything is epistemically wrong about formulating the belief that Amanita phalloides is deadly on the basis of Gary Neville’s testimony to that effect in the absence of a such a defeater. However, it is natural to feel that we would be being duped in such a case and hence are not really justified in those circumstances.[iv] But we cannot uphold this view on Burge’s account. Again, this is simply because on Burge’s view there essentially are no ad verecundiam arguments as they are understood in the standard sense outlined above. Burge’s view does allow one to distinguish E1 from E2 by simply noting that E1 involves a potential defeater whereas E2 does not, at least when we are in doxastic possession of the relevant contextual information, but the point still stands. Burge’s view offers no account of how to make the distinction between E1-type and E2-type cases absent this information and there is essentially nothing wrong with accepting testimony in situations like E1. More importantly, on his view the distinction between E1-type cases and E2-type cases would not in any case be a logical distinction.
These observations seem to constitute a sort of reductio ad absurdum of Burge’s view, albeit of a relatively weak sort. What we can say is that we can and should make the logical distinction between E1-type and E2-type examples and we should resist endorsing the sort of implausible gullibility that CAP would sanction by basing the epistemic status of testimony on the actual doxastic possession of defeaters (see Fricker 1994). There is something logically and epistemologically wrong about a reasonable adult accepting as justified the claim that Amanita phalloides is deadly when uttered by Gary Neville when he or she does not know who Gary Neville is (see Dawkins 1995). By the same token, absent any information about Gary Lincoff, it would also seem to be wrong to accept that Amanita phalloides is deadly when he utters a statement to that effect absent any knowledge of the relevant contextual factors concerning Lincoff’s bona fide expertise on mycological matters.

On the standard logical interpretation of ad verecundiam arguments and their epistemic significance we can account for these aspects of testimonial transmission of information. Were one to accept C’1 on the basis of P’1 in the absence of awareness of P’2 one would be unjustified in doing so for logical reasons and the same general lesson applies in the case of AE2. Absent reasons, even implicit and weak ones, to the effect that the testimony in question is good we would not be justified in accepting Lincoff’s pronouncement either. This of course amounts to a simple rejection of non-reductivism and this then seems to indicate that non-reductivism is a superior account of testimonial warrant. Even if we are not prepared to go this far, we should at least be prepared call a spade a spade and to accept that we are not justified in accepting a proposition on the basis of testimony when the testimony comes from an inappropriate source and we are not aware of this. What non-superficial or non-question-begging epistemic purpose is served by holding that we would be warranted in accepting such a belief in such circumstances?[v] Burge’s only real answer seems to be that so-called hyper-intellectualized epistemology is not realistic in its application to testimony. However, this contention is dubious as CAP is at least as unrealistic as the standard view, but it is also simply a misunderstanding of the nature of logical reconstruction.

This brings us to the crux of the issue. Essentially the problem with Burge’s view is that it is not merely the intelligibility of the message that is relevant to the justification of information received as testimony and so testimony is not as simple and basic as Burge believes. Also, it is simply not obvious that the default normative assumption in cases of testimony should be that of truth-telling. The kind of principle of charity that Burge assumes links intelligibility to truth is simply too weak to allow us to draw important distinctions in epistemic behavior involving testimony. As Richard Dawkins puts it, “[c]hildren are naturally credulous (1993, p. 32),” but, “[t]here is no charm in the near infinite gullibility of children (1993, p. 33).” This is why we think that logical analysis of our epistemic behavior is so crucial. It allows us to reveal our epistemic failings by revealing them often to be, in part, logical failings and doing so allows us to guard against these sorts of logical and epistemic deficiencies. Again quoting Dawkins, we should recognize that “[g]rowing up, in the fullest sense of the word, should include the cultivation of a healthy skepticism (1993, p. 35-36),”and that “[w]e need to replace the automatic credulity of childhood with the constructive skepticism of adult science (1993, p. 36).” Science, of course, involves reasoning and this is precisely the point where Burge’s view fails. His view of testimony denies that when we engage in information transmitting dialogue we should recognize that there is a logical norm to be upheld that is crucially related to rationality via the logical structure of our discourse in direct opposition to CAP, even if we are not reflectively aware of such logical structure. Specifically, we ought not to glorify the failures to observe this kind of healthy skepticism and to obey logical principles that are part and parcel of rational behavior that are endorsed in Burge’s simple de-intellectualized epistemology of testimony. His view amounts to nothing more than the brute acceptance of whatever others tell us in the absence of explicit defeating reasons to the contrary. Such behavior is illogical and involves the commission of a fallacy of which we are all acutely and commonly aware and which we should label as such, but this is precisely what we cannot do given Burge’s view of testimony.

[i] The dashed line in this inferential scheme is used to indicate that good inferences of this sort may be either cogent inductive arguments or valid deductive arguments that are enthymemes. In the latter case the omitted premise might be something like “What experts claim is true”.
[ii] The implicit implication here is, of course, that Gary Neville is not, in fact, an expert on mycology.
[iii] For Burge this seems to mean that merely conforming to this rule is sufficient for entitlement, as opposed to literally following the rule.
[iv] Here I am implicitly rejecting that there is anything like prima facie evidence. Evidence is what really justifies belief, not what appears to do so.
[v] One main motivation for Burge seems to be the rejection of a rather extensive skepticism that might appear to follow from our rejecting testimony as a basic source of knowledge. However this is blatantly question-begging whether or not such skepticism actually follows from the rejection of testimony as a basic epistemic resource.

Burge, T. (1993). Content Preservation. The Philosophical Review 102, 457-488.
Burge, T. (2003). Perceptual Entitlement. Philosophy and Phenomenological Research 67, 503-548.
Copi, I. And C. Cohen (2004). Essentials of Logic. Upper Saddle River: Prentice Hall.
Dawkins, R. (1995). Putting Away Childish Things. Skeptical Inquirer January/February, 31-36.
Fricker, E. (1994). Against Gullibility. In: B. K. Malital and A. Chakrabarti (Eds.), Knowing from Words (pp. 125-161), the Netherlands: Kluwer.
Kitcher, P. (1983). The Nature of Mathematical Knowledge. New York: Oxford University Press.
Lincoff, G. (1977). Toxic and Hallucinogenic Mushroom Poisoning: A Handbook For Physicians and Mushroom Hunters. New York: Van Nostrand Reinhold.
Walton, D. (1997). Appeal to Expert Opinion. University Park: The Pennsylvania University Press.