The epistemology of testimony

A two-mode theory of testimony

In a number of papers I have developed an anti-justificationist, two-mode theory of testimony. Analytical philosophers, when writing about epistemology in general and testimony in particular, use the definition of knowledge as "justified true belief" to guide their enquiries. This is not the only way in which "knowledge" can be defined; there are many concepts of knowledge and I'm interested in the sort that we use on a daily basis as we go about our everyday lives. This is rarely, if ever, justified true belief.

The first mode

In the first mode of belief-acquisition we respond to testimony as if that response were governed by the defeasible acquisition rule: Accept others' assertions. This is hardwired into us. Our unthinking reaction to hearing or reading any statement is simply to accept it. We have an innate tendency to accept other people's assertions. It should be noted that this acceptance is not justified in any way whatsoever. There follow some of the theses I have argued for in my various papers.

We are recipients of testimony before we are transmitters or originators of testimony

Since assertions are linguistically formulated, before we can respond to testimony or make our own assertions we have to have mastered some key features of language. Dummett has forcibly argued, correctly in my opinion, that learning a language and acquiring testimonial knowledge necessarily go hand in hand (The Seas of Language, p. 424):

[A] disposition to accept what others tell one is central and fundamental to acquiring language at all: unless one does so, one cannot be said to understand language. As one's mastery of language deepens, one learns to curb this fundamental disposition in certain cases. One learns the various possibilities of error on the part of others; one learns also how language may be employed dishonestly.

This means that we must have acquired a certain amount of testimonial knowledge before we can pass on what others have told us or make our own assertions and that includes assertions about what we can percieve.

Our response to testimony changes over time

The acquisition rule is defeasible. We can override it if we think there is something dodgy about a statement we hear or read or the way in which it is presented. Such overriding factors are learnt. Children tend to be more credulous than adults. As people mature they become more sceptical about various sorts of assertion made by certain people.

If we do override the acquisition rule, there are several things we can do. We could also do nothing, in which case we're likely to remember for a certain amount of time that a particular person made this assertion or that we read it in a particular book, but as time passes our memory of the assertion and its associated meta-data is likely to fade. (By "meta-data" in this context I mean information about an assertion in addition to its content. This could include knowing who made the assertion and when and also the situation in which it was made. I'm not suggesting that we retain the same pieces of meta-data about every assertion we encounter.) If the statement interests us, we may decide to investigate its truth. If this is something that can be done quickly, we may do it there and then, otherwise we will wait until we have the time to do so. For more on this possibility, see below under the heading "The second mode" below.

People can respond to the same information in different ways

A theory of testimony has to be able to account for the obvious fact that diversity of opinion exists in society. For example, some philosophers, such as Dummett, believe that the theory of meaning is the foundation of philosophy, whereas others, such as Popper, hold that epistemology is the foundation of philosophy. For some, accepting one or other of these positions may well be the result of a long process of deliberation, but I suspect that there are many who agree with Dummett, for example, simply because they heard him or someone else assert this position. My theory can easily accommodate the position that people can respond differently to the same assertion. Overriding factors are learnt from experience and testimony; not everyone learns the same collection of such factors.

The second mode

We cannot avoid acquiring false beliefs through testimony

The acquisition rule is not foolproof and so we sometimes acquire false information and sometimes reject true information. Thus, it makes sense to put some time aside to reflect on our beliefs in order to try and weed out the false ones. It is also sensible to sometimes reconsider statements that we have previously rejected to see if any of them are true. When we do such things, we are operating according to the second mode of belief-acquisition.

We can only check a tiny number of our beliefs

Because checking whether or not a statement is true can be a very time-consuming procedure, the number of statements that we can thoroughly investigate in this way is quite limited. In early 2007, for example, it was widely reported in the media that the average person in Britain is caught on CCTV cameras about 300 times a day. This statistic went viral; it was everywhere. Most people simply accepted it as being true without a second thought, as you'd expect according to my account of testimony. David Aaronovitch was more wary and decided to investigate the matter further. (He came to the conclusion that this statistic was false.) This is an example of someone operating according to the second mode of belief-acquisition. He published his findings in an article in The Times entitled "The strange case of the surveillance cameras" (Tuesday, 3 March 2009, p. 26; if you subscribe to The Times you can read the article online). This example shows why we can only check a very small fraction of the assertions we encounter every day. Not only is checking time-consuming, but, usually, in the process of checking one claim we acquire many beliefs that we haven't checked. To conclude that the average Briton is not captured on CCTV cameras about 300 times a day, Aaronovitch had to accept many other statements as true. He rejected one claim, but in so doing acquired many additional beliefs that he did not check. The number of his unchecked beliefs actually increased.


  • Michael Dummett, The Seas of Language, Oxford. Oxford University Press, 1993.

© Antoni Diller (26 February 2015)