ARTICLE IN PRESS

Early Testimonial Learning: Monitoring Speech Acts and Speakers Elizabeth Stephens*, Sarah Suarez*, Melissa Koenig*,1 *Institute of Child Development, University of Minnesota, 51 E. River Pkwy, Minneapolis, MN 55455 1 Corresponding author: e-mail address: [email protected]

Contents 1. Epistemic Vigilance: Can Testimonial Learning Exist Without It? 2. Children's Evaluations of Speaker Messages 2.1 The Developmental Precursors to Coherence-Checking 2.2 Children's Treatment of Labeling Errors 2.3 Children's Treatment of Grammatical Errors 2.4 Children's Treatment of Inconsistent, Illogical, and Improbable Statements 2.5 Children's Treatment of Factual and Episodic Errors 2.6 When Children Encounter Message Conflicts: Other Observations 2.7 Interim Conclusion 3. Children's Evaluations of Speakers 3.1 Natural Pedagogy 3.2 Core Dimensions of Speakers 3.3 Negativity Bias 4. Concluding Thoughts Acknowledgment References

152 154 154 156 159 160 161 162 165 166 168 170 174 177 178 178

Abstract Testimony provides children with a rich source of knowledge about the world and the people in it. However, testimony is not guaranteed to be veridical, and speakers vary greatly in both knowledge and intent. In this chapter, we argue that children encounter two primary types of conflicts when learning from speakers: conflicts of knowledge and conflicts of interest. We review recent research on children's selective trust in testimony and propose two distinct mechanisms supporting early epistemic vigilance in response to the conflicts associated with speakers. The first section of the chapter focuses on the mechanism of coherence checking, which occurs during the process of message comprehension and facilitates children's comparison of information communicated through testimony to their prior knowledge, alerting them to inaccurate, inconsistent, irrational, and implausible messages. The second section focuses on source-monitoring processes. When children lack relevant prior knowledge with which to evaluate Advances in Child Development and Behavior ISSN 0065-2407 http://dx.doi.org/10.1016/bs.acdb.2014.11.004

#

2015 Elsevier Inc. All rights reserved.

151

ARTICLE IN PRESS 152

Elizabeth Stephens et al.

testimonial messages, they monitor speakers themselves for evidence of competence and morality, attending to cues such as confidence, consensus, access to information, prosocial and antisocial behavior, and group membership.

Much of early learning depends on others, and the social transmission of information presents us with rich opportunities to learn from and about others. Socially transmitted information is often communicated through others’ testimony (i.e., those statements that present their contents as true). Although all listeners of testimony, including children, are interested in acquiring reliable information, the competence and intentions of speakers are more variable, and may not align with listeners’ interests (Faulkner, 2011; Williams, 2000). Here, we argue that all listeners encounter two primary types of conflict in speakers: conflicts of knowledge and conflicts of interest. To avoid the risks engendered by these conflicts, listeners must exercise some degree of epistemic vigilance (Sperber et al., 2010). In this chapter, we focus on two sets of cognitive mechanisms that support vigilance towards the conflicts presented by speakers. As we discuss in the first half of the chapter, children protect themselves against conflicts of information through the processes of message comprehension: some initial “coherence-checking” occurs whenever children interpret a new message, alerting them to conflicts between the message and their already established knowledge and beliefs. In the second half of the chapter, we discuss a separate set of mechanisms of vigilance engaged through source monitoring: when lacking relevant prior knowledge, children actively monitor speakers themselves for indicators of incompetence or conflicts of interest.

1. EPISTEMIC VIGILANCE: CAN TESTIMONIAL LEARNING EXIST WITHOUT IT? As argued by Sperber et al. (2010), the stable existence of communication depends upon benefits for senders and receivers alike (Sperber & Wilson, 1986). That is, communication serves the purposes of speakers who engage in conversations to influence listeners, and listeners who are dependent on communicated information. However, given the asymmetric goals of speakers and listeners, and the resulting epistemic vulnerability of listeners, most agree that true testimonial knowledge depends on the presence or practice of epistemic vigilance. For example, in classical epistemology, reductionists have argued that beliefs acquired via testimony do not

ARTICLE IN PRESS Early Testimonial Learning

153

automatically qualify as knowledge in and of themselves. Rather, testimonial beliefs must be justified by other true beliefs that have been acquired by testimony, perception, inference, or memory (Adler, 2002; Coady, 1992; Faulkner, 2003; Fricker, 1995; Hume, 1748/1995; van Cleve, 2006). Reductionists further argue that knowledge from others’ testimony is made possible by the faculties of the mind that support our experiences: our ability to monitor the speaker’s beliefs and intentions, our attention to prior correspondences between messages and facts, and our licensed inferences about future correspondences. In contrast, antireductionists, such as Reid (1764/1983), view our trust in testimonial information as intrinsically justified, similar to our trust in information gained from perception, inference, or memory (Coady, 1992; Reid, 1764/1983). Modern antireductionist perspectives argue that listeners are entitled to accept testimony unless there is clear reason not to do so, such as evidence of conflictual message contents or suspicious sources (Davidson, 1984; Lewis, 1969). Both views hold the minimal requirement that candidates for testimonial knowledge must come to their own conclusions about speaker trustworthiness (McMyler, 2007). Reductionists require listeners to entertain positive reasons for their beliefs, whereas antireductionists require counterfactual sensitivity to negative evidence against testimony (Goldberg & Henderson, 2006). In line with these ideas, we are interested in the various ways in which testimonial learning is supported by processes of epistemic vigilance (Cole, Harris, & Koenig, 2012; Sperber et al., 2010; Woolley & Ghossainy, 2013), and what such processes involve. In other words, we are trying to capture what may be involved in the “species of reasoning” that Hume argued was so common, useful and necessary to human life (Hume, 1748/1995). For Hume, “the maxim by which we commonly conduct ourselves in our reasonings is. . .that where there is an opposition of arguments, we ought to give preference to such as are founded on the greatest number of past observations” (p. 124). This sage advice depends on evaluating our observations, but what kinds of observations should we evaluate? And how should they inform our judgments of whether to accept a given claim? As we review here, our ability to critically evaluate testimonial messages given what we know is part of what allows us to accept communicated information as trustworthy. It also allows us to mistrust the source when appropriate, either on moral or intellectual grounds. When the risks of trusting information are low—for example, when communicated information is of little relevance or import to us—perhaps vigilance is as well. However, when the risks of accepting misinformation are significant and the issue is one that we care about, perhaps vigilance is heightened.

ARTICLE IN PRESS 154

Elizabeth Stephens et al.

What, exactly, are these risks? Sperber et al. (2010) argued that “the major problem posed by communicated information has to do not with the competence of others, but with their interests and their honesty (p. 361),” and, further, that the existence of dishonesty is what makes epistemic vigilance indispensable in communicative contexts. Although we agree that the problem of deliberate misinformation characterizes an important risk, and that the consequences of accepting such information are potentially grave, we disagree that this is the central risk encountered when learning from testimony. Instead, the central, and likely first, risk encountered by testimonial learners is that of well-meaning but less-than-fully knowledgeable informants. Speakers vary greatly in their competence, yet are free to share the information they desire, to whomever they wish, and in whatever way they choose. Even within the context of our most intimate relationships (e.g., with partners, parents, children, close friends), where trust is high and conflicts of interest are relatively minimal, the risk of misinformation based on poorly informed sources remains. Testimony is only as reliable as the beliefs of the speakers communicating it. Because of such risks, epistemic defense mechanisms are necessary even early in life, at the start of language comprehension. As reviewed in the following section, children rely on one such defense mechanism, referred to by some as “coherence-checking,” to compare incoming information to existing knowledge and beliefs (Mercier & Sperber, 2009, 2011).

2. CHILDREN'S EVALUATIONS OF SPEAKER MESSAGES 2.1. The Developmental Precursors to CoherenceChecking Before infants can detect problems or conflicts in testimonial messages, are they completely vulnerable to the risks of misinformation? Recent studies point to early protective mechanisms that likely precede children’s detection of problematic messages. Work on early social referencing indicates that infants invest more attention in, and adjust their behavior more in accordance with, individuals who respond to their environment in a coherent manner. Findings on infant gaze-following behaviors indicate that referential expectations about speaker gestures and speech develop early in life, as does the expectation that these cues will be interrelated or co-referential. By 6 months, infants selectively follow the gaze of an agent who shared direct eye contact with them or spoke to them in infant-directed speech (Senju & Csibra, 2008). By 12 months, infants follow the gaze of novel entities that

ARTICLE IN PRESS Early Testimonial Learning

155

feature a face, show evidence of contingent behavior, or both ( Johnson, Slaughter, & Carey, 1998). When behaviors such as gaze and deictic gestures co-occur with a speaker’s act of labeling, Gliga and Csibra (2009) demonstrate that 13-month-olds expect such actions to be co-referential and to converge on a single object. How do infants react to speakers who violate these referential expectations? Tummeltshammer, Wu, and Kirkham (2013) exposed 8-month-old infants to one face that consistently gazed in the direction of target objects in varying corners of a square screen, and a second face that inconsistently gazed in the direction of the objects. At test, infants were presented with one of the original faces gazing at a corner of an otherwise empty screen. They found that infants were more likely to follow the gaze of the previously consistent gazer. Thus, it appears that by 8-months of age, infants invest more of their attention in individuals who respond contingently to their environment than individuals who fail to do so. Tummeltshammer, Wu, Sobel, and Kirkham (2014) extended these findings with a similar set of studies that compared infants’ reactions to faces or arrows that varied in reliability. Infants consistently searched in boxes containing animal animations that were either reliably or unreliably cued by faces or arrows. When presented with novel boxes, only those exposed to reliable faces followed the cues. Interesting questions remain for future research regarding the types of characteristics (animacy, presence of eyes, face, contingent behavior) that elicit infants’ expectations of future reliability. There is also evidence that infants have expectations about emotional cues and take them into account when engaging in gaze-following behaviors: Chow, Poulin-Dubois, and Lewis (2008) found that 14-month old infants preferred to follow the gaze of a looker who had expressed happiness looking inside a container with a toy inside, rather than one who expressed happiness at an empty container. A second experiment found that infants did not generalize their knowledge of looker unreliability to another, “naı¨ve” looker, indicating that they adjusted their gaze-following behavior in accordance with their experience with a particular individual. Based on this constellation of findings, it seems that contingent gaze might be one of the earliest behaviors used by learners to infer a model’s general competence. Specifically, infants’ existing expectations about the coherent use of gaze and other related referential and emotional cues seem to guide their attention, and perhaps even their learning. Many interesting questions remain. For example, it is unclear whether infants might put more stock in certain cues than others. Furthermore, whether infants’ treatment of

ARTICLE IN PRESS 156

Elizabeth Stephens et al.

incommensurate referential cues stems from their representations of agents, and the causal links between these various cueing behaviors, remains an important empirical question. As 6- to 9-month-old infants use these skills to develop a working understanding of a few common nouns (Bergelson & Swingley, 2012), they place themselves in a position to evaluate certain claims made by speakers that include those nouns. In the following subsections, we focus on infants’ and young children’s applications of their prior knowledge (i.e., “coherence checking”) by reviewing existing research on children’s treatment and evaluations of speaker messages. We focus on children’s treatment of messages containing labeling errors, grammatical errors, inconsistency, among other types of errors, and illustrate how children’s growing knowledge base and logical understanding is employed in their speaker evaluations.

2.2. Children's Treatment of Labeling Errors Studies of very young children’s detection of and reactions to overt labeling errors point to the existence of some “automatic” processes of vigilance by 16 months of age. Early in language acquisition, infants detect messages that contradict what they know and explicitly reject and deny such messages. Koenig and Echols (2003) found that 16-month-old infants looked significantly longer at a speaker when she labeled familiar objects incorrectly than when she did so correctly. In addition to looking longer at false labeling events, many infants made explicit attempts to correct the inaccurate labeler. In work by Gliga and Csibra (2009), 13-month-old infants watched while a speaker directed their gaze to, pointed at, and labeled an object hidden behind a screen. When the screen was removed, and the previously hidden object revealed, infants looked longer when the object failed to correspond to the comments and referential actions of the speaker. Thus, well before they have accumulated a large vocabulary, infants appreciate the referential nature of pointing and naming, and demonstrate behaviors to indicate they expect such signals to convey accurate and relevant information about objects. When do infants apply information about speaker accuracy to learning contexts? Koenig and Woodward (2010) found that 24-month-olds treated inaccuracy as a feature of a particular individual and formed only short-lived and fragile word-object representations when presented with novel labels from a previously inaccurate labeler. Similar effects are seen in preschoolers who fail to retain novel word-object links from speakers who declare their

ARTICLE IN PRESS Early Testimonial Learning

157

ignorance (Sabbagh & Shafman, 2009). In combination with the heightened attention infants give to false labelers (Gliga & Csibra, 2009; Koenig & Echols, 2003), children’s poor memory for the messages that incompetent and ignorant speakers provide raises important questions concerning children’s memory for problematic speakers themselves. Perhaps young children have better memories for the identities of incompetent speakers than for competent ones. Consistent with this possibility, Corriveau and Harris (2009b) found that 3- and 4-year-olds maintained their avoidance of a previously inaccurate labeler after a week’s delay, suggesting that they can form enduring memories of speakers based on their labeling performance. These findings suggest that the detection of inaccurate messages might lead to enhanced memory for incompetent sources and poor memory for the information they present. Children’s early employment of coherence-checking goes beyond the rejection or negation of messages containing labeling errors. It also influences children’s learning decisions. For example, Koenig, Cle´ment, and Harris (2004) introduced 3- and 4-year-old children to two labelers, one who consistently labeled familiar objects correctly, and another who labeled these objects incorrectly. The speakers then presented conflicting novel labels in reference to unfamiliar objects. The authors found that children who were able to monitor the speakers’ labeling accuracy selectively learned the novel labels provided by the more reliable informant. Koenig and Harris (2005) found that 3- and 4-year-olds also preferred a previously accurate labeler over one who expressed ignorance. Quine and Ullian (1970) argue that our ability to reason about the processes contributing to a speaker’s falsehood underlies our “temerity” in treating a statement as false. Such reasoning can also be seen in young children, whose selective learning is not only influenced by the presence of testimonial errors, but also by the magnitude of the errors. For example, Kondrad and Jaswal (2012) found that 4- and 5-year-olds preferred to learn new labels from an inaccurate informant whose labeling errors were “closer” to a correct answer than from one whose answers were bizarre (e.g., referring to a partially obscured picture of a comb as a “brush” vs. as a “thunderstorm”). Relatedly, Einav and Robinson (2010) demonstrated that 4- to 7-year-olds’ sensitivity to error magnitude, as evidenced by a selective preference to learn from a speaker who made “smaller” errors, depended on their age and the error type: for quantifiable numerical errors (i.e., an incorrect number of dots on a card), children of all ages displayed sensitivity to error magnitude. However, only the 6- and 7-year-olds displayed such

ARTICLE IN PRESS 158

Elizabeth Stephens et al.

sensitivity to semantic errors varying in magnitude (e.g., calling a lion “a tiger” vs. “a mouse”). These findings indicate that for label learning, children’s semantic knowledge serves as a reference against which incoming information is compared. They also indicate that children have ideas about what counts as a more or less plausible semantic error. Coherence checking is not just a “violation-detection” device that treats all violations equally— instead, it is informed by and sensitive to a growing body of semantic, factual, and psychological knowledge. Children’s growing knowledge base includes a developing understanding of agents and minds. For example, Koenig and Echols (2003) found that false labeling was treated differently by 16-month-olds depending on whether it came from a speaker looking at the objects, from a speaker looking away from the objects, or from a machine. Infants manifested their surprise by looking longest towards the incorrect labeler who had visual access to the objects and the correct labeler who did not have visual access to the objects. Nurmsoo and Robinson (2009a)’s findings also indicate how children’s developing social knowledge and labeling knowledge interact. Three- to 5-year-olds avoided accepting new information from a previously inaccurate labeler, even though the speaker had only erred while blindfolded. Preschool-aged children might be confused about how to adjust their learning in response to a speaker who violates their expectations about co-reference by labeling objects without gazing at them. Researchers have also explored how linguistic information interacts with statements that violate or cohere with children’s factual knowledge. Noveck, Ho, and Sera (1996) examined how grammatical information interacts with perceptual knowledge, examining the influence of epistemic modals of varying force on children’s judgments about object location. Building on Hirst and Weil’s (1982) findings that 5-year-olds detect differences in force between modals like must, should, and may, they presented children with true and false statements with varying modal forms. They found that 5-, 7-, and 9-year-olds who were presented with true and false statements about hidden objects’ locations did not endorse false statements, even when they were more forceful. Evidently, social learners rely on their own prior factual knowledge and access to information to a great extent, even in the face of linguistic information that could sway their learning decisions. Together, these findings point to children’s ability to monitor, evaluate, and selectively learn from labelers who differ in accuracy. Most significantly, they illustrate the critical role that children’s existing knowledge of labels and

ARTICLE IN PRESS Early Testimonial Learning

159

their referents plays in their ability to engage in conflict-detection, which protects them from accepting misinformation. So far, we have reviewed research on children’s treatment of speakers who provide erroneous labels for highly familiar objects and concepts, which are often overt categorical errors representing violations of a fundamental form of knowledge shared by all speakers of a language. How do children treat errors that are not as obvious or easily detected? Are they more forgiving of other kinds of errors?

2.3. Children's Treatment of Grammatical Errors Research on children’s learning from speakers who vary in grammatical accuracy gives us further insight into the role of children’s prior knowledge and experience in epistemic vigilance. Jaswal, McKercher, and VanderBorght (2008), for instance, found that children endorsed novel labels from a previously accurate labeler, but did not endorse her irregular plural or past tense forms of novel nouns and verbs; instead, they endorsed regular plural or past tense forms of the novel nouns and verbs, even when they were provided by an unreliable source. In an additional experiment, informant reliability was established by a speaker’s consistent use of correct or incorrect morphology. The same endorsement pattern was found: children endorsed novel, regular morphology provided by a formerly unreliable morphologist over the irregular morphology provided by the formerly reliable morphologist. Corriveau, Pickard, and Harris (2011) argue that this pattern of results underscores the importance of children’s own experience with morphology. That is, children likely adopted the morphological forms of a previously inaccurate source because they weight their own knowledge of morphological regularities more heavily than a handful of instances of a source’s inaccuracy. To explore this possibility, they presented 4-year-olds with two speakers of differing labeling and grammatical accuracy who offered novel words of equally probable morphology. In this scenario, children selectively endorsed the accurate labeler and grammarian when learning both novel labels and novel morphology. Thus, children’s selective learning from previously reliable informants extends to different domains of language, so long as the informants’ claims are probable in light of children’s prior linguistic experience. A set of experiments by Sobel and Macris (2013) further illustrate how cues about speaker reliability interact with information about linguistic regularity. Four-year-olds were presented with two lexically accurate speakers who differed in their appropriate use of subject–verb agreement. When

ARTICLE IN PRESS 160

Elizabeth Stephens et al.

asked to learn object labels, older (but not younger) 4-year-olds relied on the syntactically accurate speaker to learn new labels for novel objects. Interestingly, when they observed speakers who were syntactically accurate but differed in lexical accuracy, all 4-year-olds used the speakers’ lexical accuracy to guide how they learned novel lexical information and novel irregular plurals. However, they did not rely on the lexically accurate speaker to learn irregular past tense forms. So, depending on the fidelity of the linguistic domain, and the kinds of exceptions that occur, reliability information alone may not always be enough. The processes of monitoring regularity information encoded in language and monitoring speaker accuracy interact.

2.4. Children's Treatment of Inconsistent, Illogical, and Improbable Statements When do children treat logically inconsistent statements as false? Previous research has indicated that the ability to detect logically inconsistent message content is evident by age 6. Unsurprisingly, there is age-related variability in children’s inconsistency-detecting competence, as well as variability based on the form that inconsistent statements take. Braine and Rumain (1981) found that most 5- and 6-year-olds were able to detect inconsistencies between statements offered by two puppets about the hidden contents of a box. Similarly, Ruffman (1999) found that when children were presented with two pairs of statements, where one featured a contradiction (e.g., “It’s very yummy.”/“It’s very yucky.”), and the other did not (e.g., “It’s very yummy.”/“It’s green.”), 6-year-olds successfully identified the pair of statements that did not “make sense.” Morris and Hasson (2010) found that children demonstrated the ability to determine whether two states were at odds (e.g., “There is a sticker in the box.” and “The box is empty.”) before they could detect statements’ syntax-based inconsistencies (e.g., “There is a sticker in the box and there is not a sticker in the box.”). However, even for 7- to 8-year-olds, syntax-based inconsistencies were more difficult than nonsyntax-based inconsistencies. Recent work from our laboratory suggests that children as young as four exhibit more skill than previously thought in making explicit judgments about logical consistency when presented with semantic or conceptual inconsistencies, such as “Today I saw a ball that was the smallest ball ever and it was the biggest ball ever at the same time” (Doebel, Koenig, & Rowell, 2011). While 4-year-olds detected the inconsistent statements and reported that they “did not make sense,” only 5-year-olds preferred to learn from an informant who previously made consistent statements,

ARTICLE IN PRESS Early Testimonial Learning

161

indicating that detecting inconsistency was not sufficient for making selective learning choices based on that information. This may reflect a failure on the part of 4-year-olds to connect inconsistency with irrationality or unreliability. Additionally, work on children’s treatment of messages on improbable and impossible events yields insight into the role of children’s prior knowledge in monitoring the acceptability of statements (for a review, see Woolley & Ghossainy, 2013). Shtulman and Carey (2007), for example, found that while adults easily distinguished between improbable (e.g., a person drinking onion juice) and impossible (e.g., a person eating lighting) scenarios, children between the ages of 4 and 8 showed strong inclinations toward skepticism. That is, they were likely to judge improbable events as impossible and did not show adult-like response patterns until age 8. In recent work by Corriveau, Chen, and Harris (2014), 5- and 6-year-old children were asked to make judgments about the reality status of protagonists in realistic, religious, and fantastical stories. Children from secular backgrounds were more likely than those from religious backgrounds to deny that the protagonist in religious stories was a real person. This supports the possibility that children’s skepticism towards improbable and impossible scenarios is in part tied to their lack of exposure and experience with speakers who discuss and endorse miraculous possibilities.

2.5. Children's Treatment of Factual and Episodic Errors So far we have reviewed research on children’s treatment of fundamental violations of word-object referents, statements containing grammatical mistakes and irregularities, and statements indicating inconsistency or logical deficiencies. Here, we examine the role of prior knowledge in coherence-checking and epistemic vigilance from the literature on children’s treatment of factual and episodic errors. Although we have reviewed much evidence indicating that toddlers and preschoolers detect apparent conflicts, there are circumstances under which they trust testimony that conflicts with their knowledge. Ganea, Koenig, and Millet (2011) found that 30-month-olds persisted in accepting claims from an unreliable informant about the location of a hidden toy; it was not until 36 months of age that children adjusted their willingness to update their beliefs based upon the past reliability of the speaker. Jaswal (2010) reported on other factors that bolster children’s developing abilities to engage in epistemic vigilance. When 30-month-olds were presented with

ARTICLE IN PRESS 162

Elizabeth Stephens et al.

testimony that conflicted with their naı¨ve beliefs or recent observations about the location of a treat, they were less trusting when they simultaneously saw evidence of its incorrectness or had convergent evidence confirming their initial beliefs. Therefore, if toddlers’ confidence in their existing beliefs can be reinforced or directly confirmed, they are more vigilant against false testimony. Cle´ment, Koenig, and Harris (2004) investigated children’s selective learning from testimony about perceptual information and reported two key findings. First, when asked to predict what a previously accurate source would say about a new object, both age groups predicted a correct report. However, when asked to predict what a previously inaccurate source would say, 4-year olds predicted future errors whereas 3-year-olds were less sure and more willing to predict a correct report. Second, there were limits to children’s trust in previously reliable informants when firsthand perceptual information was available. When object color was visible to children, both 3- and 4-year-olds did not endorse a previously reliable speaker’s incorrect testimony about its appearance. Fitneva, Lam, and Dunfield (2013) also found that 6-year-old children flexibly adjusted their dependence on other speakers (“experts” in this case) depending on their own access to information. That is, children relied on their own access to objects to learn about the visible properties of unfamiliar animals, but deferred to experts when learning about the animals’ hidden properties. Thus, even in early childhood, children monitor the nature of a claim, the kind of evidence required to support it, the gravity of speakers’ errors and the processes that explain them. Children’s testimonial learning decisions reflect sensitivity to the many considerations that go into our evaluations of people’s statements. As Quine and Ullian put it: “Acceptability depends, as always, on a weighing of the total evidence” (p. 63).

2.6. When Children Encounter Message Conflicts: Other Observations In this chapter, we have discussed the various ways in which children treat messages that conflict with their existing knowledge, focusing on the types of errors they encounter. In this section, we raise more general questions about conflict-detection across different domains of knowledge. First, we compare differences in children’s treatment of different kinds of message conflicts. Second, we discuss the apparent developmental gap between the ability to engage in conflict-detection and the emergence of competent

ARTICLE IN PRESS Early Testimonial Learning

163

selective learning. Finally, we consider the scope of epistemic vigilance once a conflict has been detected. Why does children’s treatment of discordant testimony—ranging from acceptance to explicit rejection and avoidance—seem to vary depending on the type of information communicated? For example, Jaswal and colleagues have shown that preschoolers continue to trust an adult who provides inaccurate information about the location of a highly desirable object ( Jaswal, 2010; Jaswal, Croft, Setia, & Cole, 2010). In contrast, many have shown that preschoolers avoid inaccurate informants: For example, they prefer previously accurate over inaccurate informants when learning new object labels ( Jaswal & Neely, 2006; Koenig & Harris, 2005; Koenig et al., 2004), and they even block learning from speakers who have mislabeled familiar objects (Koenig & Woodward, 2010). In response to such discrepancies, Koenig and Stephens (2014) have argued for the importance of certain conceptual or content-driven differences in testimony, such as key differences between “transient-episodic” claims (i.e., assertions of facts tied to a specific time and place) and “semantic–conceptual” claims (i.e., assertions about generalizable, conventional, scientific, or conceptual knowledge). They point to evidence that children are more vigilant in response to violations of commonly held, semantic, or cultural knowledge than to violations of transient, episodic knowledge, and show that children’s responses to episodic errors are less robust. This may be because episodic violations are less unusual, more readily explained by plausible processes, or nonrecurrent (or some combination of these). Further evidence for this comes from Stephens and Koenig (under review), who found that preschoolers initially exposed to differentially accurate semantic testimony were more likely to selectively learn from a previously accurate informant than children who were exposed to differentially accurate episodic testimony. These findings speak against a homogeneous treatment of testimony, and furthermore, point to children’s understanding that certain kinds of errors are more informative about a source’s competence than others. Another key observation about epistemic vigilance can be made from our review of children’s treatment of messages conflicting with their own knowledge: the ability to make explicit judgments about a speaker’s knowledge does not necessarily coincide with appropriately selective learning choices. For example, as described by Koenig and Echols (2003), 16-month-olds were able to detect and even correct mistaken labels for familiar objects. However, studies on 24-month-olds’ label learning indicate that their selective learning in response to these errors is still developing.

ARTICLE IN PRESS 164

Elizabeth Stephens et al.

Although, Koenig and Woodward found evidence for selective learning and memory in 24-month-olds, Krogh-Jespersen and Echols (2012) showed that children were willing to learn novel labels from both previously accurate and inaccurate speakers (for evidence of this delay between conflict detection and responsive learning beyond label-learning, see Cle´ment, Koenig, and Harris, 2004; Doebel et al., 2011). These conflicts suggest that children’s heightened monitoring of speakers based on the veracity of their statements does not automatically lead to avoidance during subsequent learning. Several factors might underlie children’s ability to modulate their learning in response to detected errors. First, selective learning might require cognitive skills that go above and beyond simply detecting message conflicts. A vast literature on children’s developing inhibitory control, for example, indicates that there is a major developmental shift in children’s ability to inhibit a prepotent response between ages 3 and 5 (see Carlson, 2005; Carlson, Moses, & Hix, 1998). Children with poor inhibitory control might be failing to suppress their tendency towards accepting speakers’ testimony, even testimony that conflicts with their own prior knowledge ( Jaswal et al., 2014). Another explanation is that children’s general understanding of the reasons underlying unreliability or states of knowledge is still developing. For example, work by Robinson and colleagues shows that 3- and 4-year-olds take into account an agent’s access to evidence when deciding whether to accept their claims (Robinson & Whitcombe, 2003) and will even excuse mistakes if they are due to limitations in perceptual access (Nurmsoo & Robinson, 2009b). However, blindfolded informants present new challenges to children, and they seem befuddled when asked to choose between two very odd speakers who differ in their epistemic sins: one who wears a blindfold when naming objects (and gets them wrong) and one who names them inaccurately with no blindfold (Nurmsoo & Robinson, 2009a). Children might rightfully check the coherence of such claims, and they withhold their trust until further explanations for this aberrant behavior are found. Children’s general cognitive abilities, as well as their more domain-specific knowledge about human agents are likely at play, and much work remains to shed more light on such factors that come together to support children’s learning decisions. Importantly, it could be that enhanced monitoring provoked by false utterances results in a more general sensitivity to the learning context— including social and nonsocial aspects of the environment. Kidd, Palmeri, and Aslin (2013), found that children’s ability to delay gratification to receive

ARTICLE IN PRESS Early Testimonial Learning

165

greater awards was affected by the reliability of the experimenter and perhaps of the testing situation. Children were originally told that they could play with some old, used materials or wait until the researcher brought to the room more exciting, brand-new materials. Half of those who waited were then given the disappointing news that the new materials (brand new crayons, stickers) were unavailable after all. When later tested, these children were more reluctant to delay receiving a reward than were those provided with the promised materials. Remaining questions include whether children in the unreliable condition primarily blamed their disappointments on the experimenter (who indeed apologized), or whether they associated more general aspects of the context with their thwarted expectations (e.g., the room, the experimental set-up, the conversational style used in experiments).

2.7. Interim Conclusion Through the course of language development, children become increasingly able to exploit their growing knowledge base to minimize the risks of misinformation signaled by testimonial messages. Use of the selective trust paradigm clarifies that children recognize inaccurate messages, identify the person responsible and avoid learning from inaccurate sources in the future. Furthermore, children’s treatment of unreliable testimony gives us insight not only into the mechanisms involved in their selective social learning, but also into the heart of what they know. We have reviewed extensive empirical evidence that children’s existing knowledge of word-object referents, syntactic and morphological regularities, logical possibilities, and factual and episodic information empowers them to optimize their learning and minimize risk via a coherence-checking or conflict-detection mechanism. Furthermore, we have described, and have raised questions about, the nature of these mechanisms and how they might develop, how they might be applied in different domains of knowledge, and how they might support epistemic vigilance. However, children often lack existing content knowledge to critically evaluate speakers’ claims. How do children protect themselves in these instances? In addition to the cognitive processes involved in coherencechecking, which allow children to detect conflicts of information or beliefs in testimonial messages, children must be equipped with cognitive mechanisms that support the monitoring of the speaker. Such mechanisms monitor for variations in speaker competence, warmth, group membership, and use

ARTICLE IN PRESS 166

Elizabeth Stephens et al.

of pedagogical cues. In the following section, we outline candidate cognitive processes involved in evaluating, monitoring, and remembering speakers, review current frameworks of early social learning that emphasize children’s treatment of source information, and review recent findings from the selective trust literature that elucidate the characteristics of sources that young children are most attentive to and likely to take into account when making selective learning decisions.

3. CHILDREN'S EVALUATIONS OF SPEAKERS Many of the child’s most relied upon informants are those that they interact with regularly: siblings, parents, friends, and perhaps extended family and grandparents. These informants are the sources of much of what children come to know, as well as central sources of information regarding when the truth is likely to be disguised or withheld, and when such decisions are based in ignorance, etiquette, fear, or deviousness. These central sources also, of course, misjudge, misconstrue, misinterpret, and misremember. For these reasons, it will be useful for children to keep track of speakerspecific information, remember the sources of one’s information, and engage in source monitoring processes. Although children have generally been characterized as poor monitors of source information relative to adults, they tend to perform best at the types of tasks most relevant to successful social learning: tasks requiring them to distinguish between self- and other-generated information or to distinguish between multiple dissimilar agents (see Roberts, 2002). For example, 4-year-olds performed as well as adults when determining which of two distinct voices had spoken particular words (Lindsay, Johnson, & Kwon, 1991), and even 3-year-olds performed at above-chance levels when determining whether they learned about the contents of a drawer by looking, being told, or making an inference from a clue (Gopnik & Graf, 1988). Recent research from our laboratory further suggests that preschoolers’ external source monitoring performance might be specifically enhanced when discriminating between multiple human speakers relative to other types of sources. Preschoolers demonstrated enhanced source monitoring performance when they learned about the locations of a hidden treat from various speakers (e.g., “A girl in a green shirt, blue shirt, or gray shirt.”) as opposed to from various sensory modalities (e.g., “Did she tell you the name, show you a picture, or make the sound?”) (Stephens, Corriveau, & Koenig, 2013). There are several reasons why children perform particularly well on these types of source monitoring tasks.

ARTICLE IN PRESS Early Testimonial Learning

167

According to the Source Monitoring Framework ( Johnson, Hashtroudi, & Lindsay, 1993), source attributions are based on qualitative characteristics of memory. Distinguishing between external sources (e.g., my mom vs. my friend) is likely easier for young children than distinguishing between internal sources (e.g., my thinking about writing a note vs. actually writing a note) because they encode more qualitative characteristics of external events that can serve as informative source cues (e.g., the voices of different individuals, what they were wearing, what they were looking or pointing at) and facilitate their successful source attributions and potentially, social learning decisions. Second, young children likely rely on source memory processes to make selective learning decisions. Whitcombe and Robinson explored implicit source memory processes in 3- and 4-year-olds and found that while most preschoolers consistently accepted information from a better-informed source (Whitcombe & Robinson, 2000) and updated their beliefs on the basis of that source’s testimony (Robinson & Whitcombe, 2003), many failed to explicitly report source information. Similarly, Haigh and Robinson (2009) showed that preschoolers often performed poorly when asked “how they knew” information about a hidden toy, but nonetheless correctly recalled several source-related details about the learning event, including whether they touched or saw the toy and whether they were better informed than another source. Thus, children might remember and utilize source information during selective learning tasks before they can even explicitly report source information. Young children do sometimes demonstrate explicit source memory, however, and show particularly good performance when their focus is directed towards external sources, as opposed to themselves, during encoding. Crawley, Newcombe, and Bingman (2010), for instance, found that 4- and 6-year-old children more accurately recalled which of two speakers had spoken several statements when they were instructed to focus on the emotions or features of the speakers as they delivered the statements. In typical selective learning tasks, the incompetence or immorality of speakers likely attracts children’s attention to the source, resulting in similarly enhanced source memory in these tasks. Third, successfully monitoring and responding to problematic speakers in selective learning tasks likely relies on executive function skills. Drummy and Newcombe (2002) showed that 4-year-olds’ source memory performance was associated with their performance on an executive function task, even after controlling for intelligence. Similarly, Rajan, Cuevas, and Bell (2014) showed that executive function predicted 4- and

ARTICLE IN PRESS 168

Elizabeth Stephens et al.

6-year-olds’ source memory performance over and above language ability and age. In concert with work reviewed earlier, these findings suggest that executive function may play a role in monitoring and remembering speakerspecific information as well as in rejecting the testimony presented by incompetent and immoral speakers, once detected. Finally, young children might rely on low-level facial processing capacities to make social evaluations and learning decisions. Cogsdill, Todorov, Spelke, and Banaji (2014) found that 3- to 6-year-old children consistently inferred character traits such as trustworthiness, competence and dominance based on the features of faces and did so with substantial consensus. Interestingly, the attributions of the younger children were similar to those of adults, and by 5 years of age, children’s attributions were as consistent as adults’. Children also consistently made social evaluations of faces, judging them as “nice” or “mean” in accordance with how trustworthy, competent, and dominant they appeared. Further evidence suggests that children’s perceptual judgments of facial trustworthiness influence their actual trust behavior. Ewing, Caulfield, Read, and Rhodes (2014) found that 5- and 10-year-old children selectively trusted individuals perceived as trustworthy in an economic game. These rapid judgments of individuals based on scant visual experience have real consequences for children’s evaluations of and interactions with speakers. In summary, to avoid accepting misinformation when unable to evaluate speaker messages, children likely turn their attention to characteristics of speakers themselves, recruiting cognitive processes such as source monitoring, source memory, executive function, and face processing. In what follows, we review current frameworks of social learning that emphasize specific aspects of children’s attention to source characteristics. Specifically, we discuss natural pedagogy, core dimensions of speakers, and the negativity bias, highlighting their varying perspectives on the characteristics of sources that matter most to children.

3.1. Natural Pedagogy According to the natural pedagogy perspective of early learning, human communication evolved to promote the transfer of generalizable knowledge between individuals (Csibra & Gergely, 2009). From this perspective, experts are particularly motivated to share generalizable information with novices, and novices are innately prepared to receive generalizable information from others. Consistent with these ideas, adults and older children often

ARTICLE IN PRESS Early Testimonial Learning

169

tailor their presentation of information to facilitate generic learning in pedagogical contexts (e.g., Gelman, Ware, Manczack, & Graham, 2013), and even very young infants appear sensitive to adults’ ostensive pedagogical cues such as direct eye contact (Farroni, Csibra, Simion, & Johnson., 2002) and referential cues such as pointing and labeling (Gliga & Csibra, 2009). Proponents of natural pedagogy argue that young children are not only biased to attend to these ostensive–referential signals, but are also biased to accept and generalize information communicated in ostensive–referential contexts (Csibra & Gergely, 2009). Thus, from this perspective, children are predisposed to trust testimony in pedagogical contexts and should demonstrate low monitoring and high acceptance and generalization of novel information in response to the pedagogical signals that speakers communicate. Empirical evidence indicates that children are indeed inclined to accept novel testimony from unknown speakers in pedagogical contexts. Substantial research on word learning demonstrates that infants and very young children readily learn and generalize adults’ labels for novel objects (e.g., Baldwin, 1993). Likewise, studies of categorization have shown that older infants and preschoolers will accept and make inferences based on an unfamiliar adult’s testimony about the classification of objects, even when it directly conflicts with their own perceptual judgments ( Jaswal, 2004; Jaswal & Markman, 2007). Moreover, children’s willingness to accept testimony about unexpected classifications is not restricted to the preschool years, nor is it restricted to a single culture. Chan and Tardif (2013) found that American and Chinese kindergarteners and second-graders tended to categorize ambiguous exemplars according to a teacher’s inaccurate testimony, even when the teacher was no longer present. According to the natural pedagogy perspective, adults are likely to provide accurate information, and children are entitled to trust it when it is delivered in a pedagogical context. Therefore, a critical task in early development is learning to discriminate between information communicated in pedagogical contexts and information communicated in other contexts such as pretense, in which adults often intentionally provide inaccurate testimony to children. Recent research indicates that, in contrast with their behavior in pedagogical contexts, children do not automatically accept and generalize novel information communicated by adults in pretend scenarios. For example, Sutherland and Friedman (2013) found that 3- and 4-year-olds consider the plausibility of novel claims made in pretend scenarios and refuse to generalize implausible claims or claims that conflict with their prior knowledge to “real life” scenarios.

ARTICLE IN PRESS 170

Elizabeth Stephens et al.

In general, consistent with a natural pedagogy perspective, when children have no information about unfamiliar speakers, have no relevant prior knowledge that bears on the message, and encounter messages in the presence of clear ostensive–referential cues, they tend to accept what they are told. However, children most often have some prior knowledge, either about some aspect of the world or the speaker who is addressing them, and much of the evidence reviewed earlier indicates that they bring that knowledge to bear when evaluating testimony. Indeed, it may be that children generally accept novel information from speakers unless they have reasons not to, and under those conditions, children reject or revise previously held testimony-based beliefs when a trusted speaker proves problematic (e.g., Scofield & Behrend, 2008). Children’s sensitivity to considerations or reasons that “defeat” a piece of testimony, along with a capacity to question or doubt the testimony, are critical to developing complete accounts of social learning. If listeners, both child and adult, enjoy some presumptive entitlement to accept testimony under overtly pedagogical conditions, they do so only because this presumption can be suspended when a piece of testimony or a speaker elicits a conflict.

3.2. Core Dimensions of Speakers What type of evidence do children monitor and take into account when evaluating potential sources of information? When faced with unfamiliar speakers, adults and children need to determine, first, whether a speaker intends to provide accurate and helpful information, and second, whether the speaker has the ability to provide accurate and helpful information. Consistent with these considerations, current conceptions of social cognition and perception argue that adults’ evaluations of others are based almost entirely on the core dimensions of moral warmth (i.e., friendliness, helpfulness) and competence (i.e., intelligence, efficacy) (Fiske, Cuddy, & Glick, 2007). Young children seem to similarly monitor and modulate their social learning based on evidence of speakers’ moral warmth and competence, and appropriately suspend trust when faced with evidence of speaker immorality or incompetence. 3.2.1 Competence As discussed in the first section of this chapter, preschoolers can often rely on prior knowledge to monitor the coherence of speakers’ messages. Incoherent messages often clearly signal speaker incompetence, and children avoid learning from speakers with histories of blatant inaccuracy or irrationality.

ARTICLE IN PRESS Early Testimonial Learning

171

However, when young children lack relevant content knowledge to engage in coherence checking, they seem to exploit a number source-specific sociopragmatic cues to infer an unfamiliar speaker’s competence. For example, children monitor speakers’ admissions of ignorance and appear to block long-term learning and generalization from ignorant speakers. Sabbagh and Shafman (2009) presented preschoolers with a confident labeler or an admittedly ignorant one. Children who were exposed to the ignorant labeler inhibited the formation of an enduring word-object association after detecting the ignorant speaker’s incompetence. Jaswal and Malone (2007) similarly found that preschoolers were less likely to take a speaker’s testimony into account when making inferences about the function of an ambiguous object when the speaker expressed uncertainty about the object’s classification. Young children understand the connection between seeing and knowing (e.g., O’Neill, 1996) and monitor speakers’ relevant access to information to evaluate their situational competence during learning (see Nurmsoo & Robinson, 2009a,b). Furthermore, preschoolers understand that a speaker with relevant knowledge ceases to be effective when not in a position to use or act on that knowledge. Kushnir, Wellman, and Gelman (2008) found that preschoolers trusted a more knowledgeable informant when deciding which of two blocks “made a toy go” only when that informant was allowed to see and select a specific block for himself. Children also take into account speakers’ more specific competencies, or expertise, as well. Even infants appear to appreciate differences in expertise. For example, 12-months-old prefer to look at an experimenter, as opposed to their mothers, when presented with strange toys in a lab setting (Stenberg, 2009), suggesting that they expect the experimenter to have more information about the lab scenario. Preschoolers also discriminate between speakers with labeling expertise and those with mechanical expertise, systematically asking a previously competent labeler for the names of novel objects but asking a previously competent “fixer” to repair malfunctioning toys (Kushnir, Vredenburgh, & Schneider, 2013). Furthermore, young children expect that an object’s creator will possess particular expertise about its intended function. Using ambiguous objects, Jaswal (2006) found that young children were more likely to infer an object’s function based on an unexpected label if the labeler claimed to be the object’s creator. Finally, young children recognize that individuals of different ages, such as children and adults, have expertise in different information domains. VanderBorght and Jaswal (2009) found that preschoolers expect adults to be more competent than

ARTICLE IN PRESS 172

Elizabeth Stephens et al.

children when communicating the nutritional value of foods, but expect children to be more competent than adults when communicating about toys. All of these epistemic considerations underscore children’s ability to engage in the kind of epistemic reasoning that testimonial learning requires. Finally, young children monitor, interpret, and exploit bystanders’ reactions to testimony as indirect indices of a speaker’s competence. Young children most often endorse a speaker whose testimony is received with nonverbal signs of approval as opposed to disapproval from bystanders, even when the bystanders are no longer present (Fusaro & Harris, 2008). Additionally, preschoolers take into account the consensus of the majority when inferring speaker competence. When asked to identify the referent of a novel label, preschoolers tend to select the object indicated by the majority over that indicated by a lone dissenter (Corriveau, Fusaro, & Harris, 2009). In summary, even when young children lack relevant prior knowledge to directly evaluate the competence of a speaker’s testimony, they monitor sources for alternative indicators of competence. By preschool, children track speakers’ admissions of ignorance, inhibiting long-term learning from ignorant speakers. They also make inferences based on their understandings of knowledge acquisition and divisions of cognitive labor about how likely speakers are to know specific types of information given their present position and specific areas of expertise. Finally, preschoolers call on precocious social cognitive skills to assess other people’s evaluations of a given speaker and retain and incorporate that information into their own selective learning decisions. Children’s sensitivity to these various ways in which speakers demonstrate their competence is impressive, and it calls for further research concerning the types of adjustments they make when their trust in competent informants is challenged. It can happen that trusted sources will lose their authority, and it would be useful to know if children’s adjustments are sweeping or calibrated, and how far children can monitor the beliefs that might require reassessment (for work in this direction, see Scofield & Behrend, 2008). 3.2.2 Moral Warmth Children show sensitivity to behavioral indicators of moral warmth beginning early in infancy (e.g., Hamlin & Wynn, 2011; Kuhlmeier, Wynn, & Bloom, 2003; Vaish, Carpenter, & Tomasello, 2009). Hamlin, Wynn, and Bloom (2007), for example, found that infants as young as 6 months of age preferred to reach for a previously helpful agent as opposed to a previously hindering agent, suggesting that even very early social evaluations

ARTICLE IN PRESS Early Testimonial Learning

173

are influenced by individuals’ past treatment of others. However, not until preschool do children take into account moral information when deciding whether to learn from speakers. Mascaro and Sperber (2009) presented 3- and 4-year-olds with two puppets, one described as “mean” and the other described as “nice.” Both age groups were more likely to trust the “nice” puppet over the “mean” puppet, indicating that children as young as 3 years of age are able to track general information about speakers’ moral warmth and use that global trait information to modulate their levels of epistemic trust. Four-year-olds, but not 3-year-olds, also appropriately rejected the claims of a puppet described as a “big liar who always tells lies,” indicating a clear developmental shift in the types of epistemic inferences drawn from specific immoral behavior. Similarly, Vanderbilt, Liu, and Heyman (2011) demonstrated that 3-year-olds readily accepted advice about a treat’s location from speakers who consistently tricked or helped others in the past, whereas 4- and 5-year-olds only accepted advice from those who consistently helped previous finders, again indicating a shift in children’s understanding of the implications of immoral behavior for social learning. Older children begin to prioritize speakers’ intentions over other information when deciding whom to accept novel information from. Liu, Vanderbilt, and Heyman (2013) demonstrated that 5- and 6-year-olds more often trusted speakers who had previously demonstrated good intentions towards a finder than bad intentions, regardless of the correctness of the information they communicated about the locations of hidden treats, underscoring the importance of moral considerations in older children’s learning decisions. When young children lack direct cues to a speaker’s moral warmth or good intentions, they exploit indirect cues to speaker morality to make selective learning decisions. One salient indirect indicator of speakers’ intentions is their group membership (e.g., Killen, Margie, & Sinno, 2006). Individuals often attribute moral warmth to their in-group, and such attributions are associated with increased helping and facilitative behaviors (see Fiske et al., 2007). Young children appear highly sensitive to group membership information when learning from speakers, perhaps inferring that members of their in-group will be particularly motivated to provide them with relevant and valuable information. For example, all else being equal, preschoolers generally prefer to accept testimony from their mothers over strangers (Corriveau, Fusaro, et al., 2009; Corriveau, Meints, & Harris, 2009; Corriveau, Harris, et al., 2009), familiar teachers over unfamiliar teachers (Corriveau & Harris, 2009a), and native-accented speakers over foreign-accented speakers (Kinzler, Corriveau, & Harris, 2011). Children’s learning is not only

ARTICLE IN PRESS 174

Elizabeth Stephens et al.

influenced by socially relevant group membership information but also by arbitrary, uninformative affiliations. Macdonald, Schug, Chase, and Barth (2013) showed that 4-year-olds did not preferentially endorse the labels provided by a previously reliable out-group member over those of an unreliable in-group member, even though group membership was randomly assigned and bore no social significance. Hetherington, Hendrickson, and Koenig (2014) explored the specific influences of group membership and moral behavior on children’s social preferences and learning decisions. They assigned 4- and 5-year-old children to arbitrary groups and found that preschoolers’ preferences for and willingness to share with an in-group member were substantially reduced when she behaved antisocially, but they still preferred to learn from the antisocial in-group member as opposed to a neutral out-group member. These findings suggest that children’s moral judgments, like their competence judgments, are not absolute nor unqualified but, rather, nuanced, responsive, and specific. Although young children monitor agents’ moral behavior and dislike immoral agents, they do not necessarily view in-group members’ immoral actions as indicative of their quality as potential informants. Instead, they maintain a preference to learn from immoral in-group members, even when they are disliked. Clearly, young children’s social evaluations, like adults,’ rely heavily on evidence of moral warmth. Monitoring of agents’ moral warmth is evident as early as infancy, and older preschoolers recognize its implications for testimonial learning situations. Although speakers’ moral behavior is significant to children very early in development, the consideration of its importance in testimonial learning situations seems to occur later in development. Shafto, Eaves, Navarro, and Perfors (2012) demonstrated that the computational model best capturing 3-year-olds’ performance on selective trust tasks makes inferences about speakers’ knowledge only and neglects to make inferences surrounding speakers’ intentions. In contrast, 4-year-olds’ performance was best captured by a model that took into account both speakers’ knowledge and intentions, suggesting a clear developmental shift in the tendency to monitor speakers’ moral behavior in testimonial learning scenarios and to inhibit learning from immoral or deceptive individuals.

3.3. Negativity Bias In general, adults weigh negative events and information more heavily than positive events or information, a “negativity bias” or “positive–negative asymmetry effect” that has been consistently and ubiquitously found in

ARTICLE IN PRESS Early Testimonial Learning

175

the domains of attention, memory, emotion, learning, and impression formation, among others (see Baumeister, Bratslavsky, Finkenauer, & Vohns, 2001). Such a bias is also evident in infants and young children (see Vaish, Grossmann, & Woodward, 2008) and has recently been proposed to participate in children’s selective learning decisions (Koenig & Doebel, 2012). Here, we provide further evidence that indicators of speaker incompetence or immorality are better monitored, evaluated, and remembered by children than are indicators of competence or morality. Fiske et al. (2007) argue that individuals generally heed positive evidence of competence and overlook instances of incompetence. However, young children seem to demonstrate a clear negativity bias when evaluating speakers’ competence. Corriveau, Meints, et al. (2009), for example, demonstrated that 4-year-olds preferentially endorsed a previously accurate labeler over a neutral speaker, and a neutral speaker over a previously inaccurate labeler. Three-year-olds, however, displayed a clear negativity bias: they showed no preference for the testimony of a previously accurate labeler over that of a neutral speaker, but preferentially endorsed a neutral speaker over a previously inaccurate labeler. This pattern suggests that the youngest age group predominantly monitored speakers for signs of incompetence. Consistent with this finding, young preschoolers penalize speakers who make very few errors. Pasquini, Corriveau, Koenig, and Harris (2007) presented children with two speakers who differed in relative accuracy, with one always making fewer labeling errors than the other. Four-year-olds preferentially endorsed novel testimony provided by the more accurate speaker in every condition. Three-year-olds, in contrast, were highly attentive to instances of inaccuracy and only preferred the more accurate speaker when she had made no errors in the past, exhibiting striking mistrust after exposure to a single testimonial error. Doubt toward incompetent speakers was also apparent in research conducted by Koenig and Jaswal (2011), who presented preschoolers with two speakers, one of whom lacked experience with dogs and made errors in labeling their breeds and one whose knowledge was unremarkable and who expressed neutral preferences for the dogs. Preschoolers avoided the incompetent source and preferentially accepted the neutral speaker’s testimony regarding not only novel dogs but also novel objects, indicating a global avoidance of the incompetent source. Often, in everyday life, children (and adults) lack access to multiple informative episodes of a stranger’s competence or incompetence, and must decide whether or not to accept their testimony based on a single encounter. Fitneva and Dunfield (2010) examined 4-year-olds, 7-year-olds, and adults’ selective

ARTICLE IN PRESS 176

Elizabeth Stephens et al.

learning decisions after observing a single instance of informants accurately or inaccurately responding to a question about a storybook scene. When requested to indicate who they wanted to ask about additional elements of the scene, 7-year-olds and adults most often selected the previously accurate informant, indicating that a single instance of inaccuracy sufficed to provoke their negative speaker evaluations. Taken together, these studies suggest that when presented with indicators of competence, preschoolers are more likely to monitor for evidence of incompetence as opposed to competence. Inaccuracies, even if few in number or restricted in content, provoke negative social evaluations, epistemic vigilance, and expectations of generalized incompetence in young children. Young children also seem disproportionately attentive to immoral behavior relative to moral behavior. Kinzler and Shutts (2008) showed 3- and 4-year-olds a series of faces described as frequently committing mean or nice actions. Both age groups showed enhanced memory for faces described as “mean” compared to those described as “nice,” as well as enhanced recollection for the specific actions the “mean” faces were said to have committed (Baltazar, Shutts, & Kinzler, 2012). This suggests a clear bias in young children to attend to and remember individuals associated with immoral as opposed to moral behavior, one that might be evolutionary adaptive (e.g., Cosmides & Tooby, 1992). As discussed previously, preschoolers take into account moral information when deciding whether to learn from speakers and prefer to learn from those described as “nice” as opposed to those described as “mean” (e.g., Mascaro & Sperber, 2009). However, it is unclear whether young children’s selective trust in a moral speaker is attributable to a specific preference for a moral speaker or vigilance against an immoral speaker. Doebel and Koenig (2013) investigated this issue by familiarizing preschoolers to either a moral speaker or an immoral speaker, each of whom was paired with a neutral speaker. Children exhibited a clear negativity bias when asked to discriminate the nicer of the two speakers: they showed enhanced performance when discriminating between immoral and neutral behavior relative to when discriminating between moral and neutral behavior. Thus, preschool-aged children appear to demonstrate a negativity bias when monitoring speakers for moral behavior, showing heightened face memory and source memory for immoral relative to moral individuals, and heightened attention to immoral behavior when making social evaluations. However, young children’s learning from speakers does not appear to be driven by negativity or positivity biases. When provided with evidence of

ARTICLE IN PRESS Early Testimonial Learning

177

speaker morality in the absence of other relevant speaker characteristics, findings suggest that, by 4 years of age, children generally prefer to learn from an individual judged to be nicer (Doebel & Koenig, 2013; Mascaro & Sperber, 2009). Yet, when children receive additional information about speakers, evidence of their good intentions and moral warmth enters into a wider network of considerations. As Hetherington and colleagues’ (2014) findings clarify, young children do not always defer to the informant judged to be nicer. Instead, they take into account group membership, and considered it to be a more relevant consideration in this learning context than the speaker’s prior antisocial behavior. Children might similarly demonstrate a learning preference for speakers with questionable intentions who nonetheless have more authority, expertise, or access to information.

4. CONCLUDING THOUGHTS To sum, we have argued here that infants and young children first encounter and detect messages that conflict with their prior knowledge, before they encounter and detect individuals who present conflicts of interest. Such message conflicts are likely central to igniting the testimonial learning processes for three reasons: First, such conflicts may serve to raise the very possibility of false language to infants (Koenig & Woodward, 2010). Second, once the possibility of false testimony is raised, our review of recent research suggests that children treat errors differently depending on the type of conflict involved. Conflicts of meaning (e.g., calling a ball “a shoe”; “Here is a brushes”) and conflicts of logic (e.g., “the ball is X and it is not X”; “1 + 1 ¼ 6”) elicit not only a quick rejection but also a ready appraisal of the speakers who produce such statements. Conflicts of transient (“The treat is in the red box.”) or episodic facts (e.g., “The ball is soft.”) elicit less rejection and an interest in the kind of access that the speaker had when making the claim (Brosseau-Liard & Birch, 2011). Further, claims that are not internally inconsistent but that violate aspects of the child’s experience and knowledge of the physical world (e.g., miraculous or impossible claims) invite different interpretations depending on the child’s prior experience (Corriveau et al., 2014). While more work is needed, research investigating different types of “message conflicts” suggests that children are interested in the kind of backing or support that a claim receives and also, that what backing suffices depends upon the claim it is to meant to support. Third and finally, these message conflicts elicit appraisals of the speaker. These conflicts

ARTICLE IN PRESS 178

Elizabeth Stephens et al.

likely set in motion processes described earlier that support those parts of the reasoning process that target the identity of the speaker, the group or part of the population that made the claim, and how they came to hold it.

ACKNOWLEDGMENT This chapter was supported by the National Institute of Health under a Ruth L. Kirschstein National Research Service Award (grant number 5T32HD007151) from the NICHD to E. Stephens and by the National Science Foundation (NSF award #1024298) to M. Koenig.

REFERENCES Adler, J. (2002). Belief’s own ethics. Cambridge, MA: MIT Press. Baldwin, D. A. (1993). Infants’ ability to consult the speaker for clues to word reference. Journal of Child Language, 20, 395–418. Baltazar, N. C., Shutts, K., & Kinzler, K. D. (2012). Children show heightened memory for threatening social actions. Journal of Experimental Child Psychology, 112, 102–110. Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohns, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5, 323–370. Bergelson, E., & Swingley, D. (2012). At 6–9 months, human infants know the meanings of many common nouns. Proceedings of the National Academy of Sciences of the United States of America, 109(9), 3253–3258. Braine, M. D., & Rumain, B. (1981). Development of comprehension of “or”: Evidence for a sequence of competencies. Journal of Experimental Child Psychology, 31(1), 46–70. Brosseau-Liard, P., & Birch, S. A. (2011). Epistemic states and traits: Preschoolers appreciate the differential informativeness of situation-specific and person-specific cues to knowledge. Child Development, 82, 1788–1796. Carlson, S. M. (2005). Developmentally sensitive measures of executive function in preschool children. Developmental Neuropsychology, 28(2), 595–616. Carlson, S. M., Moses, L. J., & Hix, H. R. (1998). The role of inhibitory processes in young children’s difficulties with deception and false belief. Child Development, 69(3), 672–691. Chan, C. C., & Tardif, T. (2013). Knowing better: The role of prior knowledge and culture in trust in testimony. Developmental Psychology, 49, 591–601. Chow, V., Poulin-Dubois, D., & Lewis, J. (2008). To see or not to see: Infants prefer to follow the gaze of a reliable looker. Developmental Science, 11(5), 761–770. Cle´ment, F., Koenig, M., & Harris, P. (2004). The ontogenesis of trust. Mind & Language, 19(4), 360–379. Coady, C. A. J. (1992). Testimony: A philosophical study. Oxford: Clarendon Press. Cogsdill, E. J., Todorov, A. T., Spelke, E. S., & Banaji, M. R. (2014). Inferring character from faces: A developmental study. Psychological Science, 25, 1132–1139. Cole, C. A., Harris, P. L., & Koenig, M. A. (2012). Entitled to trust? Philosophical frameworks and evidence from children. Analyse & Kritik, 34(2), 195–216. Corriveau, K. H., Chen, E. E., & Harris, P. L. (2014). Judgments about fact and fiction by children from religious and nonreligious backgrounds. Cognitive Science, 1–30. Corriveau, K. H., Fusaro, M., & Harris, P. L. (2009). Going with the flow: Preschoolers prefer nondissenters as informants. Psychological Science, 20, 372–377. Corriveau, K., & Harris, P. L. (2009a). Choosing your informant: Weighing familiarity with recent accuracy. Developmental Science, 12, 426–437. Corriveau, K., & Harris, P. L. (2009b). Preschoolers continue to trust a more accurate informant 1 week after exposure to accuracy information. Developmental Science, 12(1), 188–193.

ARTICLE IN PRESS Early Testimonial Learning

179

Corriveau, K. H., Harris, P. L., Meins, E., Fernyhough, C., Arnott, B., Elliot, L., et al. (2009). Young children’s trust in their mother’s claims: Longitudinal links with attachment security in infancy. Child Development, 80, 750–761. Corriveau, K. H., Meints, K., & Harris, P. L. (2009). Early tracking of informant accuracy and inaccuracy. British Journal of Developmental Psychology, 27, 331–342. Corriveau, K. H., Pickard, K., & Harris, P. L. (2011). Preschoolers trust particular informants when learning new names and new morphological forms. British Journal of Developmental Psychology, 29(1), 46–63. Cosmides, L., & Tooby, J. (1992). Cognitive adaptations for social exchange. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture (pp. 163–228). New York, NY: Oxford University Press. Crawley, S. L., Newcombe, N. S., & Bingman, H. (2010). How focus at encoding affects children’s source monitoring. Journal of Experimental Child Psychology, 105, 273–285. Csibra, G., & Gergely, G. (2009). Natural pedagogy. Trends in Cognitive Sciences, 13, 148–153. Davidson, D. (1984). Truth and interpretation. New York, NY: Clarendon. Doebel, S., & Koenig, M. A. (2013). Children’s use of moral behavior in selective trust: Discrimination versus learning. Developmental Psychology, 49, 462–469. Doebel, S., Koenig, M. A., & Rowell, S. (2011). Young children detect logical inconsistency. In Poster presented at the 2011 biennial meeting of the society for research in child development, Montreal, QC. Drummy, A. B., & Newcombe, N. S. (2002). Developmental changes in source memory. Developmental Science, 5, 502–513. Einav, S., & Robinson, E. J. (2010). Children’s sensitivity to error magnitude when evaluating informants. Cognitive Development, 25(3), 218–232. Ewing, L., Caulfield, F., Read, A., & Rhodes, G. (2014). Perceived trustworthiness of faces drives trust behavior in children. Developmental Science. http://dx.doi.org/10.1111/ desc.12218. Farroni, T., Csibra, G., Simion, F., & Johnson, M. H. (2002). Eye contact detection in humans from birth. Proceedings of the National Academy of Sciences of the United States of America, 99(14), 9602–9605. Faulkner, P. (2003). The epistemic role of trust. In R. Falcone, S. Barber, L. Korba, & M. Singh (Eds.), Trust, reputation, and security: Theories and practice (pp. 30–38). Berlin: Springer. Faulkner, P. (2011). Knowledge on trust. Oxford: Oxford University Press. Fiske, S. T., Cuddy, A. J., & Glick, P. (2007). University dimensions of social cognition: Warmth and competence. Trends in Cognitive Sciences, 11, 77–83. Fitneva, S. A., & Dunfield, K. A. (2010). Selective information seeking after a single encounter. Developmental Psychology, 46, 1380–1384. Fitneva, S. A., Lam, N. H., & Dunfield, K. A. (2013). The development of children’s information gathering: To look or to ask? Developmental Psychology, 49(3), 533–542. Fricker, E. (1995). Critical notice: Telling and trusting: Reductionism and anti-reductionism in the epistemology of testimony. Mind, 104, 393–411. Fusaro, M., & Harris, P. L. (2008). Children assess informant reliability using bystanders’ non-verbal cues. Developmental Science, 11, 771–777. Ganea, P., Koenig, M. A., & Millet, K. (2011). Changing your mind about things unseen: Toddlers’ sensitivity to prior reliability. Journal of Experimental Psychology, 109, 445–453. Gelman, S. A., Ware, E. A., Manczack, E. M., & Graham, S. A. (2013). Children’s sensitivity to the knowledge expressed in pedagogical and nonpedagogical contexts. Developmental Psychology, 49, 491–504. Gliga, T., & Csibra, G. (2009). One-year-old infants appreciate the referential nature of deictic gestures and words. Psychological Science, 20(3), 347–353.

ARTICLE IN PRESS 180

Elizabeth Stephens et al.

Goldberg, S., & Henderson, D. (2006). Monitoring and anti-reductionism in the epistemology of testimony. Philosophy and Phenomenological Research, 72(3), 600–617. Gopnik, A., & Graf, P. (1988). Knowing how you know: Young children’s ability to identify and remember the source of their beliefs. Child Development, 59, 1366–1371. Haigh, S. N., & Robinson, E. J. (2009). What children know about the source of their knowledge without reporting it as the source. European Journal of Developmental Psychology, 6, 318–336. Hamlin, J. K., & Wynn, K. (2011). Young infants prefer prosocial to antisocial others. Cognitive Development, 26, 30–39. Hamlin, J. K., Wynn, K., & Bloom, P. (2007). Social evaluation by preverbal infants. Nature, 450, 557–559. Hetherington, C., Hendrickson, C., & Koenig, M. (2014). Reducing an in-group bias in preschool children: The impact of moral behavior. Developmental Science, 17, 1042–1049. http://dx.doi.org/10.1111/desc.12192. Hirst, W., & Weil, J. (1982). Acquisition of epistemic and deontic meaning of modals. Journal of Child Language, 9(3), 659–666. Hume, D. (1748/1995). Of miracles. Section X, An inquiry concerning human understanding (pp. 117–141). Upper Saddle River, NJ: Prentice Hall Publishing. Jaswal, V. K. (2004). Don’t believe everything you hear: Preschoolers’ sensitivity to speaker intent in category induction. Child Development, 75, 1871–1885. Jaswal, V. K. (2006). Preschoolers’ favor the creator’s label when reasoning about an artifact’s function. Cognition, 99, B83–B92. Jaswal, V. K. (2010). Believing what you’re told: Young children’s trust in unexpected testimony about the physical world. Cognitive Psychology, 61(3), 248–272. Jaswal, V. K., Croft, A. C., Setia, A. R., & Cole, C. A. (2010). Young children have a specific, highly robust bias to trust testimony. Psychological Science, 21, 1541–1547. Jaswal, V. K., & Malone, L. S. (2007). Turning believers into skeptics: 3-year-olds’ sensitivity to cues to speaker credibility. Journal of Cognition and Development, 8, 263–283. Jaswal, V. K., & Markman, E. M. (2007). Looks aren’t everything: 24-month-olds’ willingness to accept unexpected labels. Journal of Cognition and Development, 8, 93–111. Jaswal, V. K., McKercher, D. A., & VanderBorght, M. (2008). Limitations on reliability: Regularity rules in the English plural and past tense. Child Development, 79(3), 750–760. Jaswal, V. K., & Neely, L. A. (2006). Adults don’t always know best: Preschoolers use past reliability over age when learning new words. Psychological Science, 17, 757–758. Jaswal, V. K., Perez-Edgar, K., Kondrad, R. L., Palmquist, C. M., Cole, C. A., & Cole, C. E. (2014). Can’t stop believing: Inhibitory control and resistance to misleading testimony. Developmental Science, 17, 965–976. http://dx.doi.org/10.1111/desc.12187. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3–28. Johnson, S., Slaughter, V., & Carey, S. (1998). Whose gaze will infants follow? The elicitation of gaze-following in 12-month-olds. Developmental Science, 1(2), 233–238. Kidd, C., Palmeri, H., & Aslin, R. N. (2013). Rational snacking: Young children’s decisionmaking on the marshmallow task is moderated by beliefs about environmental reliability. Cognition, 126(1), 109–114. Killen, M., Margie, N. G., & Sinno, S. (2006). Morality in the context of intergroup relationships. In M. Killen, & J. Smetana (Eds.), Handbook of moral development (pp. 155–183). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Kinzler, K. D., Corriveau, K., & Harris, P. L. (2011). Children’s selective trust in nativeaccented speakers. Developmental Science, 14, 106–111. Kinzler, K. D., & Shutts, K. (2008). Memory for “mean” over “nice”: The influence of threat on children’s face memory. Cognition, 107, 775–783.

ARTICLE IN PRESS Early Testimonial Learning

181

Koenig, M. A., Cle´ment, F., & Harris, P. L. (2004). Trust in testimony: Children’s use of true and false statements. Psychological Science, 15(10), 694–698. Koenig, M. A., & Doebel, S. (2012). Children’s understanding of unreliability: Evidence for a negativity bias. In M. Banaji & S. Gelman (Eds.), Navigating the social world: What infants, children and other species can tell us. Oxford University Press. Koenig, M. A., & Echols, C. H. (2003). Infants’ understanding of false labeling events: The referential roles of words and the speakers who use them. Cognition, 87(3), 179–208. Koenig, M. A., & Harris, P. L. (2005). Preschoolers mistrust ignorant and inaccurate speakers. Child Development, 76(6), 1261–1277. Koenig, M. A., & Jaswal, V. K. (2011). Characterizing children’s expectations about expertise and incompetence: Halo or pitchfork effects? Child Development, 82, 1634–1647. Koenig, M. A., & Stephens, E. (2014). Characterizing children’s responsiveness to cues of speaker trustworthiness: Two proposals. In S. Einav & E. Robinson (Eds.), Trust and skepticism (pp. 13–27). New York: Psychology Press. Koenig, M. A., & Woodward, A. L. (2010). Sensitivity of 24-month-olds to the prior inaccuracy of the source: Possible mechanisms. Developmental Psychology, 46(4), 815–826. Kondrad, R. L., & Jaswal, V. K. (2012). Explaining the errors away: Young children forgive understandable semantic mistakes. Cognitive Development, 27(2), 126–135. Krogh-Jespersen, S., & Echols, C. H. (2012). The influence of speaker reliability on first versus second label learning. Child Development, 83(2), 581–590. Kuhlmeier, V., Wynn, K., & Bloom, P. (2003). Attribution of dispositional states by 12-month-olds. Psychological Science, 14, 402–408. Kushnir, T., Vredenburgh, C., & Schneider, L. A. (2013). Who can help me fix this toy? The distinction between causal knowledge and word knowledge guides preschoolers’ selective requests for information. Developmental Psychology, 49, 446–453. Kushnir, T., Wellman, H. M., & Gelman, S. A. (2008). The role of preschoolers’ social understanding in evaluating the informativeness of causal interventions. Cognition, 107, 1084–1092. Lewis, D. K. (1969). Convention: A philosophical study. Cambridge, MA: Harvard University Press. Lindsay, D. S., Johnson, M. K., & Kwon, P. (1991). Developmental changes in memory source monitoring. Journal of Experimental Child Psychology, 52, 297–318. Liu, D., Vanderbilt, K. E., & Heyman, G. D. (2013). Selective trust: Children’s use of intention and outcome of past testimony. Developmental Psychology, 49, 439–445. Macdonald, K., Schug, M., Chase, E., & Barth, H. (2013). My people, right or wrong? Minimal group membership disrupts preschoolers’ selective trust. Cognitive Development, 28, 247–259. Mascaro, O., & Sperber, D. (2009). The moral, epistemic, and mindreading components of children’s vigilance towards deception. Cognition, 112, 367–380. McMyler, B. (2007). Knowing at second hand. Inquiry, 50(5), 511–540. Mercier, H., & Sperber, D. (2009). Intuitive and reflective inferences. In J. St. B. T. Evans, & K. Frankish (Eds.), In two minds: Dual processes and beyond (pp. 149–170). New York, NY: Oxford University Press. Mercier, H., & Sperber, D. (2011). Why do humans reason? Arguments for an argumentative theory. Behavioral and Brain Sciences, 34(02), 57–74. Morris, B. J., & Hasson, U. (2010). Multiple sources of competence underlying the comprehension of inconsistencies: A developmental investigation. Journal of Experimental Psychology. Learning, Memory, and Cognition, 36(2), 277. Noveck, I. A., Ho, S., & Sera, M. (1996). Children’s understanding of epistemic modals. Journal of Child Language, 23(3), 621–643.

ARTICLE IN PRESS 182

Elizabeth Stephens et al.

Nurmsoo, E., & Robinson, E. J. (2009a). Identifying unreliable informants: Do children excuse past inaccuracy? Developmental Science, 12(1), 41–47. Nurmsoo, E., & Robinson, E. J. (2009b). Children’s trust in previously inaccurate informants who were well or poorly informed: When past errors can be excused. Child Development, 80, 23–27. O’Neill, D. K. (1996). Two-year-old children’s sensitivity to a parent’s knowledge state when making requests. Child Development, 67, 659–677. Pasquini, E. S., Corriveau, K. H., Koenig, M. A., & Harris, P. L. (2007). Preschoolers monitor the relative accuracy of informants. Developmental Psychology, 43, 1216–1226. Quine, W. V., & Ullian, J. S. (1970). The web of belief. New York, NY: Random House. Rajan, V., Cuevas, K., & Bell, M. A. (2014). The contribution of executive function to source memory development in early childhood. Journal of Cognition and Development, 15, 304–324. Reid, T. (1764/1983). Section XXIV. Of the analogy between perception and the credit we give to human testimony. In R. Beanblossom & K. Lehrer (Eds.), Thomas Reid’s inquiry and essays. Indianapolis, IN: Hackett Publishing. Roberts, K. P. (2002). Children’s ability to distinguish between memories from multiple sources: Implications for quality and accuracy of eyewitness statements. Developmental Review, 22, 403–435. Robinson, E. J., & Whitcombe, E. L. (2003). Children’s suggestibility in relation to their understanding about sources of knowledge. Child Development, 74(1), 48–62. Ruffman, T. (1999). Children’s understanding of logical inconsistency. Child Development, 70(4), 872–886. Sabbagh, M. A., & Shafman, D. (2009). How children block learning from ignorant speakers. Cognition, 112, 415–422. Scofield, J., & Behrend, D. A. (2008). Learning words from reliable and unreliable speakers. Cognitive Development, 23, 278–290. Senju, A., & Csibra, G. (2008). Gaze following in human infants depends on communicative signals. Current Biology, 18(9), 668–671. Shafto, P., Eaves, B., Navarro, D. J., & Perfors, A. (2012). Epistemic trust: Modeling children’s reasoning about others’ knowledge and intent. Developmental Science, 15, 436–447. Shtulman, A., & Carey, S. (2007). Improbable or impossible? How children reason about the possibility of extraordinary events. Child Development, 78(3), 1015–1032. Sobel, D. M., & Macris, D. M. (2013). Children’s understanding of speaker reliability between lexical and syntactic knowledge. Developmental Psychology, 49(3), 523–532. Sperber, D., & Wilson, D. (1986). Precis of relevance: Communication and cognition. Behavioral and Brain Sciences, 10, 697–754. Sperber, D., Cle´ment, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, G., et al. (2010). Epistemic vigilance. Mind & Language, 25(4), 359–393. Stenberg, G. (2009). Selectivity in infant social referencing. Infancy, 14, 457–473. Stephens, E., Corriveau, K., & Koenig, M. (2013). Knowing how you know: Preschoolers show enhanced monitoring of speakers relative to other sources of information. In Poster presented at the biennial meeting of the cognitive development society, Memphis, TN. Stephens, E., & Koenig, M. (under review). The nature of the error is key: Children’s selective learning in semantic and episodic domains. Sutherland, S. L., & Friedman, O. (2013). Just pretending can be really learning: Children use pretend play as a source for acquiring generic knowledge. Developmental Psychology, 49, 1660–1668. Tummeltshammer, K. S., Wu, R. W., & Kirkham, N. Z. (2013). 8-month-olds know whose face is reliable. In M. Knauff, M. Pauen, N. Sebanz, & I. Wachsmuth (Eds.), Proceedings of the 35th annual conference of the cognitive science society. Austin, TX: Cognitive Science Society.

ARTICLE IN PRESS Early Testimonial Learning

183

Tummeltshammer, K. S., Wu, R., Sobel, D. M., & Kirkham, N. Z. (2014). Infants track the reliability of potential informants. Psychological Science, 25, 1730–1738. Vaish, A., Carpenter, M., & Tomasello, M. (2009). Sympathy through affective perspective taking and its relation to prosocial behavior in toddlers. Developmental Psychology, 45, 534–543. Vaish, A., Grossmann, T., & Woodward, A. (2008). Not all emotions are created equal: The negativity bias in social-emotional development. Psychological Bulletin, 134, 383–403. van Cleve, J. (2006). Reid on the credit of human testimony. In J. Lackey, & E. Sosa (Eds.), The epistemology of testimony (pp. 50–74). Oxford: Oxford University Press. Vanderbilt, K. E., Liu, D., & Heyman, G. D. (2011). The development of distrust. Child Development, 82, 1372–1380. VanderBorght, M., & Jaswal, V. K. (2009). Who knows best? Preschoolers sometimes prefer child informants over adult informants. Infant and Child Development, 18, 61–71. Whitcombe, E. L., & Robinson, E. J. (2000). Children’s decisions about what to believe and their ability to report the source of their belief. Cognitive Development, 15, 329–346. Williams, M. (2000). Dretske on epistemic entitlement. Philosophy and Phenomenological Research, 60, 607–612. Woolley, J. D., & Ghossainy, M. E. (2013). Revisiting the fantasy–reality distinction: Children as naı¨ve skeptics. Child Development, 84(5), 1496–1510.

Early testimonial learning: monitoring speech acts and speakers.

Testimony provides children with a rich source of knowledge about the world and the people in it. However, testimony is not guaranteed to be veridical...
290KB Sizes 0 Downloads 3 Views