Does reductionism - the notion that testimonial evidence simply reduces to facts derived from memory and experience - succeed in explaining how we can know things on the basis of testimony?
I shall consider two problems in reductionism that
Jennifer Lackey highlights to show that reductionism does not succeed in explaining how we can know things on the basis of testimony.
Parameters
To ‘know things’ I take to mean that we have a true belief that is justified in some way. That justification is what reductionism and its alternatives look to provide. For brevity’s sake I will concentrate on one-to-one testimony, although there are many types of testimony (what Lackey calls ‘epistemic heterogeneity’, 2006, p.441). I shall focus on a global reductionist account:
... a hearer must have non-testimonially based positive reasons for believing that testimony is generally reliable. (p.440)
Testimony
We are told things as children and adults that we rarely investigate. Our parents tell us our name is x and we were born in y. The rule we follow is something like:
If the speaker S asserts that p to the hearer H, then, under normal conditions, it is correct for H to accept (believe) S's assertion, unless H has special reason to object. (Adler, 2015)
Jonathan Adler calls this the default rule (DR) and it plausibly describes our everyday behaviour. But the question is: does simply telling someone something, rather than, for example, showing them, confer knowledge?
Reductionism
Reductionism is commonly identified with
David Hume (1711-1776). In
Of Miracles he writes:
[O]ur assurance in any argument of this kind is derived from no other principle than our observation of the veracity of human testimony, and of the usual conformity of facts to the reports of witnesses.
..and:
Were not the memory tenacious to a certain degree; had not men commonly an inclination to truth and a principle of probity; were they not sensible to shame, when detected in a falsehood: Were not these, I say, discovered by experience to be qualities, inherent in human nature, we should never repose the least confidence in human testimony. (Hume, 1777, p.174)
Hume is claiming that a reasonable person ‘reduces’ testimonial evidence to facts derived from their memory and experience – it is not justifiably knowledge otherwise.
Contrast this with the non-reductionist view from Hume’s contemporary
Thomas Reid (1710-1796). He claims there is a principle of veracity (PV), that people tend to tell the truth, and a corresponding principle of credulity, that people tend to believe what they are told. The principle of veracity stems from the connection between thoughts and language; the very purpose of language is to communicate the truth of one’s thoughts, and while some may lie occasionally, even liars tell the truth more than they lie. From this principle we can reasonably assume the truth of testimony in and of itself, unless we are given reason to doubt it.
Reid’s PV doesn’t seem very different from Hume’s appeal to people’s truthfulness (‘had not men commonly an inclination to truth and a principle of probity’, p.174). But to establish knowledge from testimony
a priori Reid needs a justification that is
a priori; that is, justified without appealing to evidence, such as the notion that 1+1=2. Hume justifies the truth of testimony
a posteriori, by appealing to experience.
Problems with Reductionism
Reid makes some trenchant criticisms of reductionism. In a reductionist world he notes that ‘[s]uch distrust and incredulity would deprive us of the greatest benefits of society’ (Reid, 1764, p.177). But if believing testimony is beneficial to us as social animals it does not follow that it is necessarily knowledge-imparting. A parent could tell their children that there are dragons in the wood across the busy main road nearby, to deter them from crossing the road. This would benefit the children by preventing them from risking their lives on the road, but it does not impart any knowledge to them. The alternative, truthful, method of telling the children of the perils of the road might be less effective. This breaks the
a priori connection between thoughts and language to which Reid appeals.
But if we must find reasons in our background knowledge, from perception, memory and inference, for justifiably believing testimony, we need to establish the general rule of testimony’s reliability, and, as Lackey points out (p.440), two issues arise:
1.
It’s not clear we can be confident that we have received enough reports to establish the general rule that testimony is reliable (TR).
2.
It’s not clear we have sufficient access to the facts of the world to judge a testimony’s truthfulness (TT).
Consider the first instance of testimony a child receives from her parent – say, her mother introduces a man and says, this is your uncle; the child believes it (what else can she do?).
On TR, she has only one record in her testimonial database, so she has nothing to judge her mother’s testimonial reliability by.
On TT, charitably she will know a few facts about the world; teddy bears are warm and fluffy, spoons are cold and hard, for example. But no child can investigate her uncle’s provenance, so, per reductionism, on TR and TT grounds, that the man is her uncle is not knowledge. Nonetheless this belief will be stored as a fact.
Later, the mother introduces another child as her uncle’s son, calling him the child’s cousin. By now, perhaps, the child has a larger testimonial database showing 90% reliability for her mother’s testimony, so the TR issue is somewhat mitigated. But, still, the child only has data for her mother and perhaps a few close family and friends, which can hardly be projected to establish testimony’s general reliability. One might at this stage appeal to a local reductionist approach, that ‘the justification of each particular report or instance of testimony reduces to the justification of instances of sense perception, memory, and inductive inference’ (Lackey, p.440). But this introduces problems of chains of testimony that are insoluble, I think, without an appeal to testimony’s general reliability.
On TT, the child might observe facts about her cousin – that he lives with her uncle, for example – that could give her good empirical evidence that he is her cousin. If we ignore the TR problem above, reductionism then suggests that she knows who her cousin is. But this fact itself is based on a background ‘fact’ that is not knowledge, per reductionism – that her cousin’s father is her uncle. Basing facts on non-facts looks fatal to reductionism as a coherent account of knowledge acquisition.
A reductionist might counter that by extrapolation from our limited datasets we can be justified in our beliefs on both TR and TT grounds; maybe we could confirm background beliefs retrospectively as experience increases. But we are then left with the problem of keeping track of our beliefs and their status. I’m not aware anyone does this; I’m really only aware of background beliefs, not background confirmed facts and unconfirmed ‘facts’.
A couple of anti-reductionist suggestions point to some more issues that an enlightened reductionist account should address.
Testimony to be trusted?
Paul Faulkner draws a distinction between practical testimony, such as the ‘dragons’ example, and epistemological testimony. Echoing Reid’s ‘distrust’ objection, he says that reductionism ignores ‘...the practical dimension of testimony. It misses out on the reasons that trust provides’ (The Open University, 2014, 2:23). Faulkner’s ‘assurance view’ (ibid, 4:22) suggests that by trusting testifiers we can take what they say as knowledge. There does seem to be a trust component to testimony; when that trust is broken, we take it very seriously. One of the ten commandments is not to lie; if journalists are found out telling falsehoods it can end their career; and likewise for scientists who falsify evidence in scientific papers.
But this doesn’t seem to help at all in the ‘dragons’ testimony case; children nearly always trust their mother’s testimony and they are rarely let down. But the mother’s testimony is split between the practical and the epistemological, so there has to be an account that distinguishes the knowledge-imparting from the pragmatic, and trust doesn’t seem to provide it. The children are right to trust their mother, but not because she is imparting knowledge. From a reductionist viewpoint, there can be plentiful evidence available to trust someone, but how could the trust be established in the first place given the problems of TR and TT above?
Testimony as a practice?
Alan Millar says:
...telling is a move in a practice. The practice may be conceived as that of informing through telling, but it should be understood that the practice embraces both informing through telling, understanding acts of telling, and adopting a stance towards what one is being told. (Millar, 2010, p.177-178)
So the testimony must be what Millar calls ‘felicitous’ (p.178). Testimony can be deliberately deceptive, in which case it is not felicitous. This allows us to distinguish the ‘dragons’ testimony from knowledge-imparting testimony; the mother is perhaps engaging in the practice of ‘safeguarding through telling’ rather than ‘informing through telling’.
Millar’s anti-reductionist account appeals to a perceptual-recognition account of knowledge acquisition that stands apart from a perceptual evidence account. He writes:
The crucial point though is that we account for the acquisition of knowledge in these cases in terms of the exercise of an ability to recognize a phenomenon as having a certain significance. It is the ability that is in the driving seat and its possession does not turn on independent support for any generalization that informs it. (p.187)
Generalisations form a large part of our knowledge acquisition skills, but Millar suggests that we can acquire knowledge by recognition; so we can recognise from tracks on a path that deer have been there. While a certain amount of the background knowledge to this judgement is plainly observational, there is, he claims, a recognitional skill that has been learnt that cannot be reduced to perception, memory and inference – ‘...we should also take seriously the idea that our knowledge that p from someone's telling us that p is recognitional as well’ (ibid).
This seems plausible, but would mean that those who haven’t acquired certain recognitional skills are simply incapable of acquiring knowledge. And it’s clear that many children, and I daresay adults, might suffer in this regard. Children who believe that dragons live in the woods nearby and that Santa delivers their Christmas presents are clearly underdeveloped in the perceptual-recognition stakes. But then how are any of their testimonial beliefs knowledge? Millar recognises this problem when he writes:
If early learning is to be conceived as the acquisition of knowledge through being told, as in the straightforward cases, then the knowledge will not meet the conditions I have laid down. (p.192)
And he offers an approach to the acquisition of such knowledge which is reductionist. He says that knowing that Hobart is the capital of Tasmania ‘consists in an ability to recall a publicly available, known fact, which has been gained from repeated encounters with reliable sources of information’ (p.192).
Millar’s non-reductionist suggestion addresses something like Adler’s DR; how we are correct in normal circumstances to accept someone’s testimony, absent reasons not to believe it. But he still posits reductionism for early learning, which leaves his account vulnerable to the same problem for background beliefs of reductionism simpliciter; facts relying on non-facts.
Conclusion
While the two alternatives to a reductionist explanation discussed here don’t work, they highlight some issues that reductionism misses. The assurance view recognises the importance of trust in testimony; perhaps a reductionist account of trust could be worked up to fully integrate this into reductionism. Millar’s approach observes that we learn skills that give us knowledge that resists reduction, and applies this principle to testimony. So, on the issues discussed here, since testimony has not been successfully excluded from a rigorous account of knowledge acquisition, I don’t think reductionism so far succeeds in explaining how we can know things on the basis of testimony.
Bibliography:
Adler, J. (2015) ‘Epistemological Problems of Testimony’, The Stanford Encyclopedia of Philosophy (Summer 2015 Edition), Edward N. Zalta (ed.)
Hume, D. (1777) ‘Hume on testimony and experience’ in Price, C. and Chimisso, C. (eds) (2014) Knowledge and Reason (A333 Book 5) , Milton Keynes, The Open University.
Lackey, J. (2006) ‘Knowing from testimony’, Philosophy Compass, vol. 1, no. 5, pp. 432–48.
Millar, A. (2010) ‘Knowing from being told’, in Haddock, A., Millar, A. and Pritchard, D. (eds) Social Epistemology, Oxford, Oxford University Press.
Price, C. and Chimisso, C. (2014) Knowledge and Reason (A333 Book 5) , Milton Keynes, The Open University.
Reid, T. (1764) ‘Reid on veracity and credulity’ in Price, C. and Chimisso, C. (eds) (2014) Knowledge and Reason (A333 Book 5) , Milton Keynes, The Open University.
The Open University (2014) ‘Faulkner on testimony (Part 2)’ [Audio clip], A333: Key questions in philosophy.
