By examining both reductionism and credulism I will argue that if we have any justification to trust testimony, it is a reductionist one.
In the first section I will describe reductionism and differentiate between two variants of this view. In the second section I will describe credulism and its irrationality (assuming that our goal is to have as many justified true beliefs and as few false ones as possible) and defend reductionism. Finally, in the third section I will try to argue that a rational employment of credulism ultimately leads to reductionism.
According to reductionism, which can be traced back to David Hume, testimony is not a source of justification by itself but is reducible to, and not merely mechanically dependent on, more basic epistemic sources, such as perception, memory and inference. In other words, if we have any justification to trust the testimony of others, it is on more basic non-testimonial grounds. According to Hume, those grounds are mostly of inductive nature because that what’s being told is true is not something we can infer through a priori reasoning, but is something we’ve been accustomed to through experience.
For the sake of accuracy I will differentiate between two versions of reductionism – global and local.
1.1. Global reductionism
Lackey (2010) provides a description:
According to global reductionism, the justification of testimony as a source of belief reduces to the justification for sense perception, memory, and inductive inference. Thus, in order to be justified in accepting the testimony of speakers, hearers must possess non-testimonial based positive reasons for believing that testimony in general is reliable. (p. 319)
The big problem is that global reductionism is not particularly useful. To see that let’s suppose that we were exposed to enough occurrences of testimony to reliably conclude that testimony in general is reliable. This aggregate conclusion, however, tells us nothing about how justified we are in trusting a particular testimony or a testimony within a particular context. In other words, the fact that testimony in general is reliable doesn’t seem as a good justification to believe a particular instance of testimony; in fact, if we do so without taking any other considerations, we are relying purely on luck.
For instance, if we were in an unfamiliar city and we asked for directions, we won’t be justified in trusting the testimony of the locals merely on the aggregate conclusion alone without, for example, acknowledging that in such contexts people are generally very reliable, because we can easily imagine a world where testimony in general is reliable, but people in the particular context are not, so whether we get true belief about the directions would be down to luck. Hence, whether testimony in general is reliable is irrelevant to whether it’s reliable in particular contexts and thus not useful.
Conversely, if the aggregate conclusion shows that testimony in general is not reliable, we would miss out on many true beliefs by simply refusing to trust any testimony, which again doesn’t seem rational.
1.2. Local reductionism
Local reductionism is far more flexible and avoids those problems. Again I will resort to Lackey (2010) for the description:
According to local reductionism, which is the more widely accepted of the two versions, the justification for each instance of testimony reduces to the justification for instances of sense perception, memory, and inductive inference. So, in order to be justified in accepting the testimony of speakers, hearers must have non-testimonial based positive reasons for accepting the particular report in question. (p. 319)
This allows our justification and belief in testimony to be context and instance sensitive, i.e. I might be justified to believe in testimony in one context but not in another, or believe in one testimony but not in another even if they are in the same context.
For example, in our asking-for-directions scenario, I might still think that in this context people are generally reliable, but that won’t be sufficient justification for trusting the testimony of a person who seems to suffer from dementia.
From the next section onward I will use the terms reductionism and local reductionism interchangeably.
2. Credulism, its problems, and defending reductionism
Credulism can be traced back to Thomas Reid, who claims that humans possess a built in propensity to speak the truth (the principle of veracity) and a complementary propensity to confide in the veracity of others (the principle of credulity).
Credulism, contrary to reductionism, claims that testimony is an autonomous source of epistemic authority, i.e. testimony is a source of justification in itself, hence it’s not reducible to other epistemic sources even if it is mechanically dependent on them. By this view we are justified in trusting testimony by default, without any independent grounds, unless there are special reasons for doubt.
This, however, seems like wishful thinking and making virtue out of necessity.(Pritchard, 2014). Also, it is not clear what counts as special grounds for doubt – need it be direct and specific to the particular instance of testimony (e.g. the testifier seems incompetent) or could it be a more general principle regarding particular contexts (e.g. business presentations promising quick riches), or even а more general principle about testimonials as a whole(e.g. living by the maxim ‘don’t trust in testimonies without having independent grounds in their support’).
One example credulists often use to support their case and attack reductionism is that it seems that very young children don’t have any non-testimonial grounds to believe the testimony of their parents, yet they trust them and learn a whole lot this way.This example, as I see it, only shows that the justification someone has to believe in credulity is reductional – it relies on repeated observations of the human nature (babies seem to trust in testimony without non-testimonial justification, humans seem to possess a propensity to speak the truth, etc).Also it doesn’t show that testimony is a justification in itself, but only that children perceive it as such, because that’s how we are supposedly ‘programmed’ to perceive it. What I’m trying to say is that the ultimate ground here is our perception, which is non-testimonial.
This was an argument just in terms of how we justify testimonial claims, and not should we justify them this way. Even if it isn’t so evident in the first type of question, in the second type credulity definitely loses, because it leaves us to the dangers of gullibility. It just doesn’t seem rational if our goal is to gather as many justified true and as few false beliefs as possible to trust testimony just by virtue of itself. It’s not uncommon for people to testify falsely if they have something to gain from it; other people are just pathological liars. In either case we need a relevant independent ground. I’ll argue for a weaker version of reductionism – one that allows the independent grounds to be testimonial, as long as in the end we have relevant non-testimonial grounds for the testimonial grounds. Let me illustrate this:
- Peter claims x and I believe him, because Sue said he is a reliable testifier.
- I trust Sue’s testimony on inductive grounds – every time Sue said someone was a reliable testifier, their testimony turned out to be true.
2 is a sufficient non-testimonial justification for believing that Sue is a reliable testifier and that Peter is a reliable testifier. This form of reasoning can be applied to as many levels of testimonial grounds as needed.
3. Rational credulism only if reductionism
One possible way to rationally use credulism would be on a very large scale. Let’s say there were a computer that collects gazillions of testimonies and stores them in with the value ‘true’ next to each (conceivably, the computer, similar to a baby, has no special grounds to doubt any of them). Invariably at some point the database will filled with a lot of, if not directly counterpositive, at least incompatible testimonies. However, they can’t be simultaneously ‘true’, so we need some way to decide which are ‘true’ and which to set as ‘false’. However, we can’t do this by holding that testimony is a source of justified belief by itself, because that way all of the testimonies in our database will have equal claim to be ‘true’, but they can’t be. We can’t rely on further testimonial grounds to justify the particular testimony, because there will be other testimonies which conflict with those too. The only solution in this case is to rely on the relevant non-testimonial grounds for each testimony, which gets us to employ the usage of reductionism.
I’ve shown that if we have any justification in believing others (both in the senses of how we justify belief in others’ testimony and should we justify it that way from epistemological standpoint) ultimately relies on non-testimonial grounds, i.e. in doing so we are employing reductionism.
Lackey, J. (2010). Testimonial Knowledge.The Routledge Companion to Epistemology.316-325
Pritchard, D. (2014). Testimony and Memory.What Is This Thing Called Knowledge?.80-86