1. TESTIMONIAL INJUSTICE AND THE CASE OF CHILDREN

In Epistemic Injustice (Reference Fricker2007) Miranda Fricker introduced the eponymous concept of a specific kind of harm done to a person in their capacity as a knower. Her main focus is on testimonial injustice, which is the denigration of a person's credibility.Footnote ¹ When pervasive and systematic, testimonial injustice is usually based in a negative identity-prejudicial stereotype, such as racism or sexism. Thus, knowledge claims by blacks and women are often accorded less weight than those by whites and males, and may even be outright ignored or disbelieved.

Fricker focuses on testimonial injustice resulting from racial and gender-based prejudices. But other prejudices could have the same effect – such as those concerning class, sexual orientation, or ethnicity. Michael Burroughs and Deborah Tollefsen (Reference Burroughs and Tollefsen2016) have recently proposed that there is widespread testimonial injustice against children.Footnote ²

That women and blacks face testimonial injustice is evident. It is not so evident that children do. We do, of course, regard children as being, in general, less reliable knowers than adults. But it will be said that that is because children are, in general, less reliable knowers than adults. So there looks to be a clear contrast with the case of women and blacks. The belief that women and blacks are inferior to men and whites in rationality, morals, competence, or the like has no evidential basis. It is a mere prejudice. By contrast, the belief that children are inferior to adults in many respects (though certainly not all) would appear to be well-supported. It is not a prejudice, one might say, but a plain fact. In particular, one might think it a plain fact that children are generally epistemically inferior to adults.

A hypothetical (but not unrealistic) case should make the contrast clear. You are in an unfamiliar city and need to find your hotel. You know it is in walking distance, but are unsure of the direction (and your smartphone is not working). You decide to ask for help. The two closest people are a woman just exiting a store 10 feet away, and a man inside the same store. If you head for the man inside the store, simply because he is a man, and even though the woman is closer to you, then you commit an epistemic injustice. You assume that the woman is less epistemically competent than the man. Mutatis mutandis if the person exiting the store is black, the one inside is white, and you approach the latter simply because they are white. In both cases you deploy a negative identity-prejudicial stereotype.

But now imagine that the person exiting the store is a child of about 10, and the one inside is a middle-aged adult. It is hard to accept that you would do the child an epistemic injustice if you went inside to ask the adult. For surely you are reasonable to think it more likely that the adult knows the way to your hotel than the child. Of course, this does not mean that you would never ask a child for directions; only that, if given the choice between a child and an adult, it will often be perfectly reasonable to approach the adult, simply because they are an adult.Footnote ³

I take it that many or most of us would, all else being equal, approach the adult. There is, then, a stereotype that children are epistemically inferior to adults. But this alone doesn't constitute an injustice. With Fricker (Reference Fricker2007: 30), I use the word ‘stereotype’ in a neutral sense. As she says, it is not only acceptable but necessary for us to use stereotypes to gauge trustworthiness. We cannot get to know every person we interact with well enough to assess their trustworthiness based on their actual characteristics. So we often draw on stereotypes. These will sometimes mislead. For example, the person who won't look you in the eye may not be lying, but just shy. However, no injustice is done in such a case, as there is no epistemic culpability for the error (Fricker Reference Fricker2007: 33). There is culpability only if one lowers one's estimation of a person's credibility based on a stereotype which runs counter to readily available evidence.Footnote ⁴

Assigning low credibility to a child, then, is an epistemic injustice only if the stereotype that children are less epistemically competent than adults is plainly inaccurate (or, perhaps, if deploying the stereotype is inappropriate for some particular reason: see fn. 4). But on the face of it, it appears that the stereotype is not inaccurate. For surely, on average, adults simply know more than children. Putting it slightly less bluntly, most children lack much of the ordinary knowledge of the world that most adults possess. This is for the simple reason that such knowledge is acquired only through experience; and the average child, almost by definition, has less experience than the average adult.

2. BURROUGHS AND TOLLEFSEN ON CHILDREN AND TESTIMONY

Why, then, do Burroughs and Tollefsen think that there is “widespread epistemic prejudice against children” (Reference Burroughs and Tollefsen2016: 360)? The answer is that they focus not on children's knowledge of the world and its ways, but on children's ability to convey the knowledge that they do have. They make this clear in a paragraph which I quote in full:

Thus Burroughs and Tollefsen will grant, I take it, that one commits no epistemic injustice in asking directions of an adult over a child, just as in consulting an adult over a child about the stock market. It is reasonable to believe that a child is less likely than an adult to be competent on these topics. Burroughs and Tollefsen's focus is on cases where the topic is one in which the child is (or at least may be) competent. They claim that a significant amount of epistemic injustice is committed against children even in these cases.

In support of this claim, Burroughs and Tollefsen focus primarily on the context of formal legal testimony. They contend that there is solid empirical evidence for two propositions concerning this context. The first is that most adults see children as unreliable witnesses. For example, they cite one study in which

Burroughs and Tollefsen then argue that these attitudes on the part of adults are mistaken. Their second proposition concerning the forensic context is that children can in fact be reliable witnesses – if questioned briefly, in accessible language, and without leading questions, on a topic in which they are competent. Under such conditions, Burroughs and Tollefsen suggest, children perform comparably to adults: “children (by the age of six) are as accurate as adults in recalling events and no more suggestible than adults when those memories are questioned in appropriate ways” (Reference Burroughs and Tollefsen2016: 366). In general, they conclude, “Children can be competent testifiers and have not been shown to be more suggestible than adults in testimonial interactions” (Reference Burroughs and Tollefsen2016: 369).

Therefore, contend Burroughs and Tollefsen, children are subject to inaccurate epistemic stereotyping in the forensic context. They then argue that this is in turn evidence that children are generally subject to pervasive epistemic prejudice. The forensic context “is by no means the only context in which children are subject to epistemic appraisal. Negative prejudicial stereotypes at play in the forensic context can also be prevalent in the context of everyday adult-child exchanges” (Reference Burroughs and Tollefsen2016: 369).

So Burroughs and Tollefsen's overall argument is that the mismatch that exists in the forensic context between children's actual abilities, and adults’ perceptions of those abilities, also exists in everyday contexts. The argument can be presented in two parts, which are then drawn together to form the conclusion. The first part is:

1. (1) In the forensic context, most adults see children as significantly less credible than adults.

Therefore,

1. (2) In the everyday context, most adults see children as significantly less credible than adults.

One could worry about the inference from (1) to (2). Perhaps in the gravity of the courtroom we entertain doubts about children's epistemic abilities that we do not entertain in other contexts. On the whole, though, I do not find this concern compelling; and in any case I already agree with (2). So I shall restrict my critical remarks to the argument's second part:

1. (3) In fact, in the forensic context, if questioned appropriately on a topic in which they are competent, children can be more or less as reliable in their testimony as adults.

Therefore,

1. (4) In fact, in the everyday context, if questioned appropriately on a topic in which they are competent, children can be more or less as reliable in their testimony as adults.

The argument then concludes that given the mismatch between children's perceived credibility as expressed in (2), and their actual epistemic reliability as expressed in (4), it follows that

1. (5) There is widespread epistemic prejudice against children.

Two preliminary remarks are in order concerning my formulation of Burroughs and Tollefsen's argument.

Firstly, it is actually not always clear that Burroughs and Tollefsen mean to claim that children's epistemic reliability is equal to that of adults – even if only in cases where the topic is one in which the child is competent.Footnote ⁵ At other times they claim simply that children “receive less epistemic credibility than they deserve” (Reference Burroughs and Tollefsen2016: 359, my emphasis), which allows that children might still deserve less credibility than adults. So do (3) and (4) capture their position accurately?

My candid answer is that I am not entirely sure. It is not clear exactly how much credibility the authors think children deserve. However, I shall assume that they do indeed mean to claim that there is epistemic equality between children and adults in conditions when the child is questioned appropriately on a topic in which they are competent. One reason to assume this is that Burroughs and Tollefsen explicitly say in two places (quoted above) that as forensic testifiers under those conditions, children perform as well as adults.Footnote ⁶ A second reason is that it is no easy matter to identify a level of credibility that is in between the level that children typically receive, and the level that the typical adult receives; and since Burroughs and Tollefsen make no attempt at such an identification, it is natural to assume that they default to the claim that, under the specified conditions, children deserve the same level as the typical adult.Footnote ⁷

My second preliminary remark concerns age. A 5-year-old is very different from a 10-year-old, and both are very different from a 15-year-old. And the point is not limited to children: a 25-year-old is very different from a 40-year-old, who are both very different from a 60-year-old. My formulation of Burroughs and Tollefsen's argument reduces all of this variation to the two gross categories children and adults. Is this a safe simplification to make?

The short answer, I think, is: No. But in making the simplification, I am following Burroughs and Tollefsen themselves. They explicitly address the issue of age variability only once, in a footnote on their first page, where they say:

I assume that by “early or middle childhood” they mean 2 or 3 years up to about 10 or 12. This is a worryingly wide range, especially since many studies indicate significant changes in children's testimonial abilities within that range (see §3 below). It is hazardous, then, to generalize about the epistemic trustworthiness of children across this entire age range. Burroughs and Tollefsen do so, and as I explain in §3, I think that this creates some problems for their argument.

So much for preliminaries. Here is where I am going. I agree with Burroughs and Tollefsen that there is epistemic prejudice against children. I also agree with them that we (adults) are obligated to take steps to ameliorate that prejudice, much as Fricker argues that we are obligated in the case of epistemic prejudices concerning race and gender. But I argue that Burroughs and Tollefsen exaggerate the severity and extent of the epistemic prejudice against children, by exaggerating children's epistemic reliability. I contend that the negative stereotype of children's testimonial competence is largely accurate, and hence that we are justified (ceteris paribus) in evaluating their testimony as less reliable than that of adults.

This does not absolve adults of the obligation to listen to children – and indeed more than we usually do. So Burroughs and Tollefsen are right to urge that adults should try harder to be active listeners and responsible hearers for children. But I shall argue that we must also be aware of our own limitations in this regard, and consider whether the context – and crucially, our own behavior – is likely to elicit reliable testimony from the child. I shall argue that the need for this awareness is actually a consequence of Burroughs and Tollefsen's own recommendation that we adopt a relational conception of child testimonial agency. They do not recognize the full implications of adopting such a conception.

In terms of the argument presented above, my concern is that (4) is too narrow and qualified to support the much more general (5). In §§4–5, I focus on the first qualification in (4): ‘if questioned appropriately’. In §§6–7, I turn to the second qualification: ‘on a topic in which they are competent’. I begin, though, by pointing out that even (3) is not fully true – which will show that (4) is even narrower than it already appears.

3. CHILDREN'S TESTIMONY IN THE FORENSIC CONTEXT

Burroughs and Tollefsen spend relatively little time defending (3), or even explaining it. Their most explicit comment on the matter is to say that there is

This is a problematic assertion. The four cited articles provide little support for it, and a broader review of the literature reveals a more complex picture.

In making this assertion, Burroughs and Tollefsen are most clearly relying on the cited article by Kim Oates. Their assertion is a close paraphrase of Oates' claim that “one of the most robust findings in research on children's memory is that children from 6 years of age and up are as accurate as adults in recalling events and no more suggestible than adults when those memories are questioned” (Oates Reference Oates2007: 844). One problem with this reliance on Oates is that the findings are not as robust as he says. Another is that while Oates rightly places an age qualification on his claim, Burroughs and Tollefsen pay insufficient attention to that qualification.Footnote ⁸

Oates says that there is robust evidence that by age 6, children are no more suggestible than adults. This is not true. While there is not much research on the testimony of children above preschool age,Footnote ⁹ a number of studies show that “suggestibility generally declines over the school years but that even adolescents can be significantly more suggestible than adults” (Warren and Marsil Reference Warren and Marsil2002: 128). In fact, even the single study that Oates himself cites does not support his claim: the researchers actually found that “both 6- and 3-year-old children were more suggestible than adults” (Goodman and Reed Reference Goodman and Reed1986: 328).

Oates is justified, however, in drawing a line at age 6 (even if it does not, as he suggests, mark a transition to adult-level performance). As he goes on to say, children under 6 “are more susceptible to highly leading and suggestive questions” (Oates Reference Oates2007: 845) than older children. This difference in suggestibility between early and middle childhood is well-established. In a review, Klemfuss and Ceci state that “[i]n most cases, young children are more susceptible to suggestion than are older children and adults. This has been a consistent finding since early in the 20th century” (Reference Klemfuss and Ceci2012: 274). So it is unfortunate that Burroughs and Tollefsen include Oates' age restriction only parenthetically. It is even more unfortunate that a few pages later they omit it entirely, when they conclude that “[c]hildren can be competent testifiers and have not been shown to be more suggestible than adults in testimonial interactions” (Burroughs and Tollefsen Reference Burroughs and Tollefsen2016: 369). The first half of this claim is true (though see below for discussion of how we should think about children's testimonial competence), but its second half is literally false, as preschoolers have been shown to be more suggestible than adults. So as I noted in §2, lack of attention to the variable of age is a weakness in Burroughs and Tollefsen's argument.

Finally, consider Oates' claim (echoed by Burroughs and Tollefsen) that “children from 6 years of age and up are as accurate as adults in recalling events” (Oates Reference Oates2007: 844) – by which he means free recall in response to open-ended, non-suggestive questions. There is indeed evidence for this (including from the study Oates cites, Goodman and Reed Reference Goodman and Reed1986). However, since (as I said above) there is not all that much research into the memory performance of children above preschool age, the evidence cannot be called ‘robust’.

4. THE CHALLENGE OF ELICITING TESTIMONY FROM CHILDREN

Where does all this leave us regarding (3)? Is it true that in the forensic context, child testifiers can be reliable if questioned appropriately? Well, yes, it is; but in such a weak and qualified sense that it supports only a very weak and qualified claim about children's everyday testimonial reliability.

The forensic psychology literature makes clear how difficult it is to interview children without tainting their testimony (Poole et al. Reference Poole, Dickinson and Brubacher2014). Even though professional guidelines now promote open-ended questions, it has been found that interviewers often either disregard the guidelines in favor of a more intuitive style which involves more closed and suggestive questions; or they inaccurately monitor their own efforts at following the guidelines and thus believe they are conducting a non-suggestive interview even when they are not (Brown and Lamb Reference Brown and Lamb2015). Applied researchers at the National Institute of Child Health and Human Development (NICHD) have therefore developed a much more prescriptive protocol which gives interviewers very detailed instructions on what to say and when to say it (NICHD 2014).

My point here is that while one can interview children in a way that enables them to deliver reliable testimony, doing so is not easy. The interviewer must be trained, and the interview itself must be conducted slowly and carefully. (The NICHD protocol has four introductory phases before even reaching the substantive phase of questioning.) Otherwise the resulting testimony cannot be relied upon – especially if the child is younger than 6.

Indeed, even a textbook interview which follows the NICHD protocol is still no guarantee of accurate testimony. For the child may have been influenced before the interview began (Ceci et al. Reference Ceci, Kulkovsky, Klemfuss, Sweeney and Bruck2007: 316–18; Zajac et al. Reference Zajac, Garry, London, Goodyear-Smith and Hayne2013: 612–13). Most strikingly, a series of studies summarized in Principe and Schindewolf (Reference Principe and Schindewolf2012) shows how heavily young children's reports can be tainted by conversations with peers. In a typical study (Principe et al. Reference Principe, Kanaya, Ceci and Singh2006), preschoolers saw a magic show in which a magician tried but failed to pull a rabbit out of a hat. The researchers then set up a false rumor that the rabbit had escaped and was loose in the school. In a later interview, almost all children reported that there had indeed been a rabbit loose in the school, even in response to open-ended questions. Furthermore, about half of them claimed to have actually seen the rabbit.

The particular danger of this effect is that the interviewer often cannot know what suggestions a child has been exposed to in the past which might be influencing what they are saying now.

Adults are not immune to such influences, of course. But the evidence is that they are less subject to them than are young children (Klemfuss and Ceci Reference Klemfuss and Ceci2012; see previous section). Older children will presumably be more resistant than younger children, but there is evidence that even adolescents remain more suggestible than adults (Warren and Marsil Reference Warren and Marsil2002; again, see previous section).

Let me be clear: it remains true that even young children can testify as accurately as adults. But my point is that for that to happen, careful preparation is needed, and many things must go right which the interviewer cannot necessarily control. So the forensic situation is less rosy than Burroughs and Tollefsen imply. Even with the best of efforts, a child's testimony is less likely to be accurate than an adult's.

And this is all still just in the forensic context, which has notable advantages over most ordinary contexts when it comes to eliciting testimony.Footnote ¹⁰ For typically, given the seriousness of the issues involved and the costs of coming to erroneous conclusions, modern forensic investigations will proceed slowly and carefully. There is therefore considerable time available in which to consider how to solicit a child's testimony – to the extent of consulting experts, or even bringing experts in to carry it out. By contrast, in ordinary contexts we do not have such time available; or even if we do, the topic will not be important enough to warrant spending that much time on it. Nor will we have access to people with training in the best techniques for soliciting testimony from children. We simply have to muddle through with the limited time and ability that we possess.

Burroughs and Tollefsen are aware of this issue. Indeed they argue that adults need to become (in Fricker's phrase) ‘responsible hearers’ for children, and in particular that we need to adopt “a relational conception of child testimonial agency that includes the adult and her ability to create conditions for the child to be heard” (Burroughs and Tollefsen Reference Burroughs and Tollefsen2016: 370, original emphasis). I fully agree. But I think that Burroughs and Tollefsen miss the full implications of such a conception.

5. THE FULL IMPLICATIONS OF A RELATIONAL CONCEPTION OF CHILD TESTIMONIAL AGENCY

Burroughs and Tollefsen urge us to adopt a relational conception of child testimonial agency, and to put it into practice through the technique of active listening, so as to ameliorate testimonial injustice against children. I concur. However, they are not fully clear about what this conception entails, or about how adopting it would affect our testimonial interactions with children. I think that they therefore risk exaggerating the extent of that effect. To explain my concern, I must make two distinctions.

The first distinction is between two kinds of effects that adopting a relational conception of child testimonial agency might have. One kind would be practical, causing us to adjust our behavior towards children in testimonial contexts so as to support their ability to express their knowledge. A second kind would be evaluative, causing us to adjust our appraisals of children's credibility, and thus the degree to which we believe what they say.

The second distinction is between a child's inherent capacity for accurate testimony, and the actual quality of her testimony in a given instance. This distinction is crucial precisely because of the point that Burroughs and Tollefsen are making: that children's testimonial performance is significantly influenced by context, so even when a child has the capacity to testify reliably on a particular topic, she may very well not actually do so if the context – especially the behavior of her interlocutors – is not suitably supportive of that underlying capacity.

When we are considering how far to believe what a child tells us – when we are evaluating her testimony – we are concerned with the actual quality of the child's testimony, not with her capacity to testify reliably. For only the former takes into account the testimonial context. It would be unwise to appraise a child's credibility based on her inherent capacity for testimonial accuracy when the actual conditions are likely to lead her to be inaccurate.

Burroughs and Tollefsen thus do not give the relational conception of child testimonial agency its full application. They seem to assume that a child's credibility should still be evaluated based on a non-relational factor: her inherent capacity for testimonial accuracy. It is true that in this respect children are the equal of adults, as Burroughs and Tollefsen at least sometimes say. Children are capable of testifying as accurately as adults. This is shown by their ability to testify reliably in a supportive forensic context, as Burroughs and Tollefsen emphasize. But this is only a reason to evaluate children's testimonial abilities in that context as comparable to the abilities of adults. It is not reason to extend that same evaluation to contexts that are less likely to produce reliable responses. If a child's testimonial agency, and thus her credibility, depends on the testimonial context, then we should raise our appraisal of her credibility only if the context is suitably supportive of her testimonial agency. If it is not, then we should not raise our appraisal, and we might even lower it. The context crucially includes one's own behavior as the child's interlocutor. What is not brought out by Burroughs and Tollefsen, then, is that adopting a relational conception of child testimonial agency does not entail uniformly raising our estimation of children's credibility, let alone according them the same credibility as adults. Rather, it entails that we must evaluate the child's credibility in part by evaluating whether our own behavior towards the child is likely to elicit accurate testimony.

I may seem to be offering adults a permanent excuse for not trusting the word of children. For if an adult believes that children are not credible, she may not even try to be an active listener for a child, which makes it less likely that the child's testimony will be accurate – making the adult's initial belief effectively self-fulfilling. However, this does not offer the adult an excuse for that initial distrust. Precisely because the adult's approach to the exchange is so influential, the adult has a prima facie obligation to do what she can to make the exchange a productive one. She may still subsequently decide not to credit what the child has said, or to accord it less weight than she would to the same testimony by an adult. But the adult should not allow her prejudgment of the child's credibility to cause her not to even try to hear what he might have to say. To allow her prejudgment to have this effect surely would be to commit a testimonial injustice.

To this extent, then, I agree with Burroughs and Tollefsen. We should be aware of our power to shape the quality of children's testimony. Where possible, adults should endeavor to engage with children in a way that supports the child's testimonial capacities.

I disagree, however, about how far our obligations extend. It matters a lot here exactly what is meant by ‘active listening’, through which (say Burroughs and Tollefsen) the relational conception of child testimonial agency is put into practice. What is involved in being an active listener for a child? How difficult or onerous is it?

Burroughs and Tollefsen do not say much to explain how they conceive of active listening. They contrast it with merely hearing what a speaker is saying, writing that

From this, one might guess that active listening involves little more than taking such steps as using accessible language and a non-intimidating tone, and asking open-ended questions. But this impression is dispelled by the examples of active listening that Burroughs and Tollefsen go on to give. Based on these examples, active listening is not so simple or easy. The examples are all of people who study or work with children professionally: child psychiatrist Robert Coles (Reference Coles1986, Reference Coles2003), anthropologist Myra Bluebond-Langner (Reference Bluebond-Langner1978), and some general cases of judges, lawyers, police officers, social workers, and psychologists in forensic contexts. Regarding the latter cases, I explained in §4 that eliciting reliable forensic testimony from a child takes significant time and skill. It is now considered a job for experts, or at least for those who are following instructions from experts. Regarding Coles and Bluebond-Langner, both of them engage with children via activities such as “drawing, painting, playing games, etc.” (Burroughs and Tollefsen Reference Burroughs and Tollefsen2016: 371). While these activities are not necessarily difficult, they do require considerable time and materials, which often may not be available to a person who wishes to access a child's knowledge.

Active listening, then, if it is to be significantly effective, requires a significant investment of time and energy. I do not mean that non-professionals cannot learn to use it. I agree that any adult can improve their communication with children if they try to base that communication on “the foundation of [the children's] own interests, concerns, and everyday experiences” (Burroughs and Tollefsen Reference Burroughs and Tollefsen2016: 371), as Coles and Bluebond-Langner do. In particular, parents, teachers, and others who regularly interact with children should try to learn and adopt these methods. But they are too work-intensive to be deployed in every testimonial encounter with a child, even for someone who is practiced in them. Often the topic will not be of sufficient importance to warrant such efforts; or there will simply not be time; or the adult will not know the child well enough, or have the relevant skills; or there will be ways of gaining the desired information other than asking the child (as in my hotel directions example from §1).

It is much more feasible, of course, to adopt the less time-consuming aspects of active listening: accessible language, open-ended questions, and so on. But these will presumably be less effective than the methods of Coles and Bluebond-Langner, or the interview techniques exemplified by the NICHD protocol. So again, my point remains: adults should be aware of the nature and limitations of their approach to a testimonial exchange with a child, and should take these into account when evaluating the child's credibility.

Let us return, then, to Burroughs and Tollefsen's inference from (3) to (4) – that is, from the claim that children can be as reliable as adults in forensic contexts if questioned appropriately on topics in which they are competent, to the claim that the same is true in everyday contexts. My purpose in this section has been to emphasize how weak (4) is. The key word is ‘can’. Children can testify as reliably as adults – they have the capacity to do so. But that capacity will often not be realized in everyday contexts, because the eliciting adult will often lack the skill, the time, the motivation, or the rapport with the child to engage in adequately active listening.

In my view, then, (4) is too weak to support (5), the claim that epistemic prejudice against children is widespread. This inference neglects the role of context, and in particular the role of the adult who is eliciting the child's testimony. In many everyday cases, I contend, an adult will not be culpable for failing to be an active listener for a child, and therefore the failure will not constitute an epistemic injustice (cf. Fricker Reference Fricker2007: 21–2). It will often simply not be reasonable, in everyday circumstances, to expect an adult to engage in the active listening that is necessary to make it likely that a child's testimony will be as trustworthy as an adult's. Hence it will often be perfectly reasonable – not a result of epistemic prejudice – for an adult to accord less credibility to a child's testimony than to an adult's.

6. THE SIGNIFICANCE OF CHILDREN'S DEARTH OF GENERAL KNOWLEDGE

So far I have argued that the qualifying phrase ‘if questioned appropriately’ in (4) constitutes a major limitation on its ability to support (5), because appropriate questioning cannot reasonably be expected in everyday contexts in the way that it can in forensic contexts. I now turn to the other qualification in (4): ‘a topic in which they are competent’.

Recall that Burroughs and Tollefsen's key claim is that “[t]he important question is whether, given a relevant domain of competence, children can be reliable testifiers” (Reference Burroughs and Tollefsen2016: 368, my emphasis).Footnote ¹¹ Their view is that even though children will be ignorant on many subjects, they can still offer accurate information about subjects in which they are competent. Burroughs and Tollefsen thus assume that a child can be a good source of testimony on particular topics even though they have a limited general knowledge. I suggest that this picture is too simple. Competence in a particular topic often depends crucially on general knowledge, so children's ability to testify even on topics with which they have direct familiarity is also limited.

To illustrate the point, I shall offer a literary example – following Fricker's lead, and also matching Burroughs and Tollefsen's focus on legal testimony.

In Ian McEwan's (Reference McEwan2001) novel Atonement, 13-year-old Briony Tallis witnesses a covert act of sexual intercourse between her older sister Cecilia and Robbie Turner, the son of the Tallis family's housekeeper. Briony misconstrues the act as rape. Then later, when it transpires that her cousin Lola was raped by an unknown assailant, Briony wrongly accuses Robbie. Largely based on her testimony, he is convicted and imprisoned. Briony misinterpreted Robbie's sexual behavior towards Cecilia (including one instance preceding the sexual act itself) as aggressive behavior. She made this interpretation because of her age and inexperience in sexual matters. Even though she did witness the sexual act (and one other instance of sexual tension between Cecilia and Robbie), she perceived it as something very different from what it was. As a result, an innocent person was imprisoned.

Burroughs and Tollefsen might reply that sexual matters are not a domain of competence for Briony, so that we would not expect her to be able to testify reliably about them. But this would miss my point. Briony is indeed not competent in sexual matters. However, she did witness the sexual act between Cecilia and Robbie, so Burroughs and Tollefsen must allow that this particular act is a domain of competence for her. The example is similar to cases they themselves refer to: “children are … likely to have access to information that adults do not in a variety of domains (for example, in cases of abuse) and their first person accounts of their lives and their interactions with others are sometimes all we have to go on” (Reference Burroughs and Tollefsen2016: 367–8). If an abused child is competent regarding the particular acts which constituted their abuse, then surely Briony is competent regarding Cecilia and Robbie's sexual act. Of course, she does not understand that act. But the same is true of the abused child – which is why one must be extraordinarily careful in eliciting testimony about such events.

So is Briony competent regarding Cecilia and Robbie's sexual act? If Burroughs and Tollefsen agree that she is, then their view that children are just as reliable as adults in testifying about matters in which they are competent is undermined, for Briony is clearly not reliable – as the resulting miscarriage of justice makes clear. On the other hand, if they say that Briony is not competent, they must explain why she is not. And the only explanation I can see is that she lacks general knowledge of sexual matters, which makes her unable to correctly interpret what she has seen. But if Burroughs and Tollefsen concede this, then their view is significantly trivialized: for a dearth of general knowledge is the norm in children, so very few of them will count as competent even concerning events which they have witnessed for themselves.

Burroughs and Tollefsen assume that a child who has witnessed or experienced a certain event will possess knowledge of that event, and that we who wish to access that knowledge merely need to help the child find the words to express it. But this picture presupposes an overly neat (and empiricist) distinction between experience and knowledge. Ironically, it is a picture which is challenged by Fricker's discussion of hermeneutical injustice (Ch. 7). As she explains, a person can lack knowledge of an event precisely because they lack the concept with which to frame that knowledge. Crucially, it is not that the person possesses the knowledge but cannot express it. Rather, their lack of the pertinent concept means that they do not even possess the knowledge. Fricker's main example is of the situation of women prior to the recognition of sexual harassment as a kind of behavior. Women who had been subjected to sexual harassment could not express what had happened to them, because in a key sense they did not know what had happened. They knew, of course, about the individual acts (e.g. loaded remarks, unwanted staring or touching), but they did not know what the acts constituted. Something like this is often the case, I suggest, for children – though not because of a hermeneutical injustice, but simply because the child is too young to have accumulated the experience (or discourse with others) that is necessary to acquire a concept. And I do not refer just to cases involving sexual behavior. The dearth of general knowledge can affect knowledge of any topic or domain. Consider, for example, a young child who has been to the supermarket many times, but does not yet understand money. On a particular day, let us imagine, his father has a heated exchange with a check-out employee about an apparent discrepancy in the price of an item. The child witnesses the entire exchange. He may even understand many of the words that were used. However, he might be a very unreliable witness concerning what actually happened. For even though he saw the exchange, he did not know what it was about, and so his account of it might be radically incomplete, or even radically wrong (as in the case of Briony).Footnote ¹²

Of course, one might try to question the child very carefully so as to separate his interpretation of what he saw from what he actually saw (Robbie Turner no doubt wished that Briony had been questioned in this way). But even if this is possible – and I am not sure whether it is – it reinforces my earlier point that eliciting information from children is a task which requires great care and perhaps some training. The forensic context might allow for this, but most everyday circumstances will not.

7. IGNORANCE OF CHILDREN'S DOMAINS OF COMPETENCE

I would like to make one more point regarding Burroughs and Tollefsen's “important question” as to “whether, given a relevant domain of competence, children can be reliable testifiers” (Reference Burroughs and Tollefsen2016: 368). It is that this question is unrealistically narrow when applied to everyday circumstances. While a forensic context will typically allow the time and resources for a careful estimation of whether the child actually is competent in the domain in question, in ordinary contexts such scrutiny is usually not possible. And this fact naturally will be, and ought to be, taken into account in our estimation of a child's epistemic reliability.

In everyday life, we often simply do not know what our interlocutor's epistemic competences are – which topics they are familiar with and which they are not. So when we evaluate someone's credibility, a significant factor is our estimation of how they will respond if the domain of inquiry is one about which they know little or nothing. A credible, trustworthy interlocutor will not only be able to accurately convey the knowledge that they do have, but will also possess the self-awareness to recognize when they know little or nothing, and will respond accordingly. For example, if their knowledge of a certain event is only second-hand, they will acknowledge this fact, rather than simply asserting that they know what happened. Or if their understanding of a topic is shaky, they will be cautious or hesitant. Or if they estimate themselves to know nothing about the topic, they will simply say so.

In short, the evaluation of a person's credibility is not limited to whether they can deliver information on topics in which they are competent. It also includes whether they are mindful of the limits of their competence.

We have seen that young children, in particular, tend to be weak in this respect. The studies by Principe and colleagues (Principe and Schindewolf Reference Principe and Schindewolf2012) show, for example, that young children are liable to report a rumored event as if they themselves had experienced it – as with the magician's escaped rabbit. So in a situation where we ourselves do not know whether there was an escaped rabbit, and we are trying to find out whether there was, if we assume that the child is in a good position to know, we may be misled by their testimony.

Thus even if, in fact, the topic is something of which the child really does have direct experience, and even if the child really is testifying with total reliability about that topic, it does not follow that we should assign high credibility to their testimony – for we ourselves may not know whether the child has direct experience, or is testifying reliably.