2. Reductionism/anti-reductionism and defeaters

The crux of the disagreement between reductionism and anti-reductionism precisely pertains to the criteria on the part of the hearer required for prima facie justification.Footnote ³ According to reductionism, beliefs founded on testimony are prima facie justified when the hearer has positive reasons for accepting the testimony, such as a track record of the testifier's past statements being accurate.Footnote ⁴

The intuitive motivation for reductionism is that, at least, belief in the testimony of someone who maliciously tries to deceive the hearer or who is not a reliable epistemic subject in the first place is not likely to be justified if one believes it without any other basis. Reductionism denies the prima facie justification in such cases by (R).

In contrast, anti-reductionism denies (R), namely, it does not require any positive reasons for justification.

Anti-reductionism allows for a wide range of prima facie justifications without demanding the hearer to have positive reasons. However, the adoption of anti-reductionism does not mean that ultima facie justification can also hold independently of the conditions on the hearer side. For ultima facie justification, the hearer must not have any negative reasons, i.e., defeaters, for not accepting the testimony. Anti-reductionism, as well as reductionism, accepts the following necessary condition for ultima facie justification.Footnote ⁵

Defeaters are possible or actual beliefs that a belief (hereinafter called a target belief) is false (or something that clearly implies that it is false) or that the target belief is not properly based, e.g., formed in an unreliable way, not based on adequate evidence, or not properly related to the evidence. The former is called a rebutting defeater, and the latter an undercutting defeater. For example, suppose I buy lunch at the only sandwich shop near my workplace, and on my way back to the office, I encounter a co-worker who tells me, “I went to the sandwich shop, but it was closed temporarily.” At this point, I have a rebutting defeater for the content of my colleague's testimony. In contrast, when we consider the case where, instead of my visiting the shop, I know that my colleague has a habit of making fun of me with nonsensical lies, I have an undercutting defeater. In either case, if I ignore the existence of defeaters and believe the content of my colleague's testimony, then the belief must not be justified. The rebutting/undercutting distinction is a categorization in terms of mechanism: whether the target belief is unjustified by supporting the denial of the target belief or by calling into question the basis of the target belief.

In contrast, the psychological/normative distinction is in terms of the status of the defeating beliefs themselves. Psychological defeaters are beliefs that the subject actually holds that rebut or undermine the target belief. Psychological defeaters function by being believed by the subject and are thought to lose justification of the target belief even if it is not true in itself and even if it does not have a positive epistemic status such as knowledge, justification, or warrant. (We discuss in Section 4 what is an appropriate example of a psychological defeater.)

In turn, normative defeaters function by virtue of the fact that the subject in some sense ought to believe. It is not necessary that the subject actually believes in order for normative defeaters to cause the target belief to lose its justification. For example, I am invited to my friend's house, and he tells me, “There is no mouse because I always keep my house clean.” which I believe. But then I see a gray object about seven inches long moving quickly in the corner of the room. But, hating mice, I cannot believe the horror of a mouse being in the room I am in, so I try not to think about the gray object. At this time, I do not believe that there is a mouse, and I continue to believe my friend's testimony that there is no mouse, which does not seem justified. This is a case of a normative defeater.

3. Child objection against reductionism and its variant

Anti-reductionism rejects (R), reductionism, and thus does not require the hearer's positive reasons for prima facie justification. However, since it accepts the no-defeater condition, it does require the absence of negative reasons for ultima facie justification. According to Lackey (Reference Lackey2008: Chap. 7), this makes anti-reductionism vulnerable to a variant of the child objection.

The child objection is originally directed at reductionism.Footnote ⁷

Since (c) and (d) conflict, the child objection concludes that (a), namely reductionism, should be abandoned. At the same time, the proponents of the child objection argue that children can satisfy the no-defeater condition because they do not have negative reasons, i.e., defeaters, and therefore anti-reductionism can admit the testimonial justification of children (Audi Reference Audi1997: 415).

In response, Lackey argues that what is necessary for justification is not satisfying the no-defeater condition in such a way. According to her, satisfactions of the condition that prohibits something in the form of “only if X does not φ, ……” can be characterized into two types: trivial and substantive satisfaction. Moreover, she contends that it is substantive satisfaction of the no-defeater condition that is required for justification (Lackey Reference Lackey2008: 7.2).

Just as a person who never lies because he lacks the ability to do so cannot claim any moral superiority, a person who lacks the ability to consider defeaters and hence does not have defeaters cannot have superior epistemic status. Therefore, it is not the mere trivial satisfaction but the substantive satisfaction of the no-defeater condition that is necessary for justification. In addition, according to Lackey, reasons are reasons, whether positive or negative, and thus as long as children are not capable of considering positive reasons, as the proponents of the child objection to reductionism assume, they should also be incapable of considering negative reasons (Lackey Reference Lackey2008: 199). Even if children satisfy the no-defeater condition, it is in the sense of trivial satisfaction. It follows from the above that we can construct a variant of the child objection.Footnote ⁸

Insofar as the absence of a defeater is necessary for ultima facie justification, even anti-reductionism cannot avoid the unacceptable consequence that children do not have any testimonially justified beliefs. Therefore, an appeal to children's abilities makes no difference to the pros and cons of reductionism and anti-reductionism. This is Lackey's response to the child objection against reductionism.

Nevertheless, while reductionism is challenged by both the original child objection and its variant, anti-reductionism need to only address the variant. Although there are several possible paths for anti-reductionists to deal with the variant, here we consider the following two:Footnote ⁹

These claims, or their combination, may allow anti-reductionism to deal with the variant of the child objection. Our principal focus is to address the success or failure of the latter claim through psychological studies of selective trust. In advance of this, in the next section, we narrow the extent of the types of defeaters that children are required to consider through the examination of the former claim.

4. Abolishing the no-defeater condition partially

Recent literature on defeaters has indicated the potential for certain alleged types of defeaters does not actually inhibit the justification of the target belief. This section specifically introduces arguments that challenge psychological defeat.Footnote ¹⁰ If these arguments hold true, then the no-psychological-defeater condition becomes unnecessary and the conditions for ultima facie justification should be somewhat less stringent.

According to Lackey (Reference Lackey2008: 44–45), psychological defeaters impede the justification of the target belief, independent of its own epistemic status, e.g., justification. Nevertheless, Lackey's example, wherein Miranda, who has a certain belief regarding the Civil War based on a textbook, comes to believe that the writer of the textbook was an unqualified scholar or a deceiver, lacks any information about how Miranda's beliefs about the author were acquired or based.Footnote ¹¹ Despite the fact that there can naturally be psychological and normative defeaters simultaneously, if the primary feature of psychological defeaters is that they can function without justification, we should consider scenarios where the defeater clearly lacks justification, such as when her belief stems from or is based on a strong dislike for the author or a malicious rumor by an individual who is at odds with the author.

It is questionable, however, that the justification of the target belief is prevented in a case with such additional details. If one acquires the beliefs about the author but still maintains the original beliefs about the Civil War, then Miranda is indeed irrational in the sense that she possesses an inconsistent set of beliefs. However, whether irrationality immediately renders the target beliefs unjustified is a matter of debate. Firstly, particularly for the externalist understanding of knowledge and justification, whether what is required is reliability, sensitivity, or safety, the idea that such an epistemic status should be understood in relation to truth is widely accepted. Even if one believes in the author's disqualification solely because she dislikes him, there is nothing to support the truth of that belief, and thus the mere belief should be irrelevant to the likelihood of the textbook's content being true.Footnote ¹²

Although the above assumes externalism, Graham and Lyons (Reference Graham, Lyons, Brown and Simion2021: 48–50) argue that even if one adopts a plausible version of internalism, the idea of the psychological defeater is still unacceptable. According to them, what is expected of a subject in an irrational state is to relinquish one of the inconsistent beliefs, not necessarily the target belief. If the evidence the subject has as a whole supports the target belief, or if the preponderance of the set of beliefs is consistent with the target belief, for instance, in the cases where Miranda knows that the author is a specialist in American history or she has encountered similar statements in numerous other textbooks and scholarly works, it is rather the beliefs that constitute psychological defeaters that should be discarded.Footnote ¹³

If it were necessary to assess the ability of children to consider psychological defeaters, that would be a formidable task. To demonstrate that the children who pass a test are capable of considering psychological defeaters rather than normative defeaters, one would need to convince them of something on clearly flawed grounds, confirm that they indeed believe it, and then see how the beliefs are reflected in their doxastic state regarding the target belief.Footnote ^14, Footnote ¹⁵ Nevertheless, if the aforementioned arguments hold, this concern becomes irrelevant. The alleged psychological defeaters do not prevent the target belief from being justified, and the criterion for ultima facie justification does not necessitate the absence of psychological defeaters. The no-defeater condition would thus be revised as follows.

This condition is weaker than the original. Thus, the argument for rejecting premise (b) of the variant of the child objection can focus on demonstrating that children have the capacity to consider normative defeaters.

5. Defeaters and selective trust

Since the mid-2000s, there has been growing research in the field of developmental psychology examining children's selective trust, or the tendency to distinguish between others in terms of their trustworthiness as sources of information and learn from certain individuals.Footnote ¹⁶ In a typical experiment, two informants with different attributes make contradictory statements about the name or function of an object, and the children are asked whom they believe, or experimenters observe whom they turn to for answers to the questions posed. These studies demonstrate that young children, particularly those aged 3 and older, utilize epistemic characteristics such as accuracy and expertise as well as social characteristics such as age and familiarity as cues to select specific individuals to believe the testimony or to ask for the answer.Footnote ¹⁷

What has been outlined above appears to be quite different from the portrayal of children that the child objection posits.Footnote ¹⁸ Citing various studies on selective trust, Clément (Reference Clément2010) criticizes philosophers engaged in the reductionism/anti-reductionism debate on the basis that children are not as gullible as assumed.Footnote ¹⁹ Nonetheless, to determine what we can say and to what extent about the reductionism/anti-reductionism debate from these empirical studies, it will be necessary to comprehend precisely the premises of the child objection, especially (a) and (b). First, there can be a range in the extent to which it is necessary to judge that children have the capacity to consider positive and negative reasons. A relatively weak requirement might be that one can appropriately change one's treatment of testimonies when there are reasons to (or not to) believe them. This could be described as sensitivity to reasons. It is this type of ability that can be assessed in psychological research on selective trust. If the demands of the proponents of the child objection are more stringent, such as the ability to believe a particular testifier by reasoning using concepts such as “accuracy” and “reliability,” or the ability to provide reasons why one believes a particular testimony when asked why, then, it will be difficult to infer out the implications for the child objection directly from the existing research on selective trust.Footnote ²⁰ Thus, what we can examine here is limited to whether children can consider reasons in a weak sense.

Additionally, as a common point against both Lackey (Reference Lackey2008), who holds that child objection poses difficulties for anti-reductionism as well as reductionism, and Clément (Reference Clément2010), who maintains that it is not problematic for either, it is crucial to distinguish the ability to consider positive reasons and the ability to consider negative reasons. For instance, it is not self-evident whether a person who, in principle, rejects the testimony of others and believes only when there are positive reasons to do so has the capacity to consider negative reasons, and vice versa. In the following, we investigate what insight research on selective trust can impart into the ability to consider negative reasons, i.e., defeaters, along the lines of the undercutting/rebutting distinction. In the course of that, we will also obtain partial insight into the positive reasons.

5.1. Undercutting defeaters

The most typical type of undercutting defeater would be the inaccuracy of the testifier. Studies examining children's reactions to different testifier accuracies primarily compare individuals who repeat correct statements about the names or functions of things with those who repeat incorrect statements (Clément et al. Reference Clément, Koenig and Harris2004; Koenig et al. Reference Koenig, Clément and Harris2004). The results show that children tend to believe the former in these situations, but it is not clear whether they are responding to accuracy or inaccuracy, or both. Therefore, Corriveau et al. (Reference Corriveau, Meints and Harris2010) conducted an experiment with 3- and 4-year-olds in which they set up a total of three situations in which they compared an accurate testifier to a neutral testifier with no information about accuracy, an inaccurate testifier to a neutral testifier, and an accurate testifier to an inaccurate testifier. They found that both 3- and 4-year-olds are more likely to believe the testimony of a neutral testifier than that of an inaccurate testifier (Table 1). Similar results were obtained by Vanderbilt et al. (Reference Vanderbilt, Heyman and Liu2014). These suggest that children accept the testimony, at least considering the typical undercutting defeater.Footnote ²¹

Table 1. Response tendencies of 3- and 4-year-olds in three situations

At the same time, the experiment by Corriveau et al. (Reference Corriveau, Meints and Harris2010) includes a test of the ability to consider certain positive reasons. In the situation comparing accurate and neutral testifiers, only 4-year-olds showed a tendency to believe the accurate ones, while 3-year-olds were indiscriminate. This suggests that the ability to consider negative reasons, as far as accuracy is concerned, develops earlier than the ability to consider positive reasons.

5.3. Implications for the reductionism/anti-reductionism debate

Setting aside some of the caveats that are discussed later, these results have the following implications for reductionism/anti-reductionism. First, anti-reductionists might say that the variant of the child objection is not persuasive because, although 2-year-olds are still in the process of developing their abilities, children at least 4 years old have shown the ability to consider both types of defeaters. Reductionists, on the other hand, might think that while the ability to consider negative reasons may indeed develop somewhat earlier than the ability to consider positive reasons, the difference is almost nonexistent, or if there is, it is only about one year and not significant (Table 2). What is suggested here as the conclusion is that the appeal to children's ability to consider reasons does not make as decisive a difference in the superiority of reductionism/anti-reductionism as was initially thought. This conclusion itself is no different from that of Lackey (Reference Lackey2008) and Clément (Reference Clément2010). However, the grounds for it lie between those provided by them. That is, children are not fully capable of considering reasons but are not totally helpless: they are in the process of developing this capacity. The period in which they have the capacity to consider negative reasons, which both reductionism and anti-reductionism require, but not positive ones, which only reductionism requires, is not very long, if at all. Thus, as far as children are concerned, the difference between reductionism and anti-reductionism is at most whether or not children in this short period of time are capable of testimonial justification.

Table 2. Ability of 2- to 4-years-olds to consider negative and positive reasons and reductionist/anti-reductionist admission or denial of their testimonial justification

However, there are several concerns in drawing conclusions about children's ability to consider reasons from the above experiments. First, these situation settings are not typical in selective trust research, and in particular, there are not many studies, such as Corriveau et al. (Reference Corriveau, Meints and Harris2010), that distinguish between the ability to consider positive reasons for trusting the testifier and the ability to consider undercutting defeaters that indicate the testifier's unreliability. In addition, the situations set up in the experiment represent only a small fraction of the cases of undercutting and rebutting defeaters, and it cannot be directly concluded that some children have the ability to consider defeaters in general based on the fact that they pass the above tasks.

In addition, whether the settings of the experiments adequately capture normative defeaters depends on how they should be understood. Normative defeaters are thought to be beliefs that epistemic subjects should have, but when one should have such beliefs is debatable. One idea is that it is when the subject actually has good evidence to support the belief. This is also the understanding that Lackey (Reference Lackey2008: 7.2) employs in constructing a variant of the child objection. In the aforementioned experiments, the children actually have evidence by hearing the testifier repeat false statements or by seeing the color of the toy with their own eyes, and in this sense, they have a normative defeaters. In contrast, recent research has presented the idea that normative defeat may hold when there is a duty to look for evidence even if one does not actually have the evidence (Goldberg Reference Goldberg2016; Lackey Reference Lackey2016: Sec. 7). For example, a mouse exterminator visits a house on business and the house owner tells him, “There is no mouse because I always keep my house clean.” The exterminator believes them without checking inside the house. He has no evidence that mice are there. However, since he is obligated to look for evidence, his belief that there are no mice does not seem justified.

If the evidence sufficient for normative defeat to hold is not limited to that actually possessed by the subject, what criteria should be used to determine the scope of evidence, and whether such an expanded understanding of normative defeaters should be allowed in the first place are still under discussion and have not been settled (cf. Nottelmann Reference Nottelmann2021). However, if normative defeaters are not limited to what the evidence in a subject's possession actually supports, then the experiments of Corriveau et al. (Reference Corriveau, Meints and Harris2010), Vanderbilt et al. (Reference Vanderbilt, Heyman and Liu2014), and Clément et al. (Reference Clément, Koenig and Harris2004) clearly do not sufficiently test the ability to consider normative defeaters. This is because what these experiments examined is how children reflect the defeaters with evidence in their doxastic states.

Psychological research may also shed light on whether children are able to consider normative defeaters in an extended sense. For example, Vanderbilt et al. (Reference Vanderbilt, Heyman and Liu2014), Ronfard et al. (Reference Ronfard, Chen and Harris2018), and Hermansen et al. (Reference Hermansen, Ronfard, Harris, Pons and Zambrana2021) suggest that preschoolers do not show a tendency to spontaneously explore on their own or seek information from another informant after hearing a testimony, even when given the opportunity to do so. However, in these settings, unlike the mouse exterminator, children are not required to determine the truth of the proposition in question with a high degree of accuracy. Thus, we cannot conclude from these studies alone that children are incapable of considering extended normative defeaters.

In light of the above, the following studies are eagerly awaited in the interest of the child objection. First, more research is needed to confirm the reproducibility of the results of the various experiments that have been referred to so far. In addition, experiments with neutral testifiers are desirable, not only for accuracy, but also for other epistemic characteristics such as expertise and confidence that can constitute defeaters or positive reasons. Finally, measuring the ability to consider normative defeaters in an extended sense would require an experiment that sets up a situation in which children have an obligation, or at least some reason, to determine the truth or falsity of the proposition in question when they are given the opportunity to explore it further on their own or to seek information from another informant.