Children's Pragmatic Inferences as a Route for Learning About the World

Children use disagreement to infer what happened

In a rapidly changing and diverse world, the ability to reason about conflicting perspectives is critical for effective communication, collaboration, and critical thinking. The current pre-registered experiments with children ages 7 to 11 years investigated the developmental foundations of this ability through a novel social reasoning paradigm and a computational approach. In the inference task, children were asked to figure out what happened based on whether two speakers agreed or disagreed in their interpretation. In the prediction task, children were provided information about what happened and asked to predict whether two speakers will agree or disagree. Together, these experiments assessed children's understanding that disagreement often results from ambiguity about what happened, and that ambiguity about what happened is often predictive of disagreement. Experiment 1 (N = 52) showed that children are more likely to infer that an ambiguous utterance occurred after learning that people disagreed (versus agreed) about what happened and found that these inferences become stronger with age. Experiment 2 (N = 110) similarly found age-related change in children's inferences and also showed that children could reason in the forward direction, predicting that an ambiguous utterance would lead to disagreement. A computational model indicated that although children's ability to predict when disagreements might arise may be critical for making the reverse inferences, it did not fully account for age-related change.

Learning the Meanings of Function Words From Grounded Language Using a Visual Question Answering Model

Interpreting a seemingly simple function word like "or," "behind," or "more" can require logical, numerical, and relational reasoning. How are such words learned by children? Prior acquisition theories have often relied on positing a foundation of innate knowledge. Yet recent neural-network-based visual question answering models apparently can learn to use function words as part of answering questions about complex visual scenes. In this paper, we study what these models learn about function words, in the hope of better understanding how the meanings of these words can be learned by both models and children. We show that recurrent models trained on visually grounded language learn gradient semantics for function words requiring spatial and numerical reasoning. Furthermore, we find that these models can learn the meanings of logical connectives and and or without any prior knowledge of logical reasoning as well as early evidence that they are sensitive to alternative expressions when interpreting language. Finally, we show that word learning difficulty is dependent on the frequency of models' input. Our findings offer proof-of-concept evidence that it is possible to learn the nuanced interpretations of function words in a visually grounded context by using non-symbolic general statistical learning algorithms, without any prior knowledge of linguistic meaning.

Using contrastive inferences to learn about new words and categories

In the face of unfamiliar language or objects, description is one cue people can use to learn about both. Beyond narrowing potential referents to those that match a descriptor (e.g., "tall"), people could infer that a described object is one that contrasts with other relevant objects of the same type (e.g., "the tall cup" contrasts with another, shorter cup). This contrast may be in relation to other objects present in the environment (this cup is tall among present cups) or to the referent's category (this cup is tall for a cup in general). In three experiments, we investigate whether people use such contrastive inferences from description to learn new word-referent mappings and learn about new categories' feature distributions. People use contrastive inferences to guide their referent choice, though size - and not color - adjectives prompt them to consistently choose the contrastive target over alternatives (Experiment 1). People also use color and size description to infer that a novel object is atypical of its category (Experiments 2 and 3): utterances like "the blue toma" prompt people to infer that tomas are less likely to be blue in general. However, these two inferences do not trade off substantially: people infer a described referent is atypical even when the descriptor was necessary to establish reference. We model these experiments in the Rational Speech Act (RSA) framework and find that it predicts both of these inferences. Overall, people are able to use contrastive inferences from description to resolve reference and make inferences about a novel object's category, letting them learn more about new things than literal meaning alone allows.

Children's use of pragmatic inference to learn about the social world

Young children often endorse stereotypes-such as "girls are bad at math." We explore one mechanism through which these beliefs may be transmitted: via pragmatic inference. Specifically, we ask whether preschoolers and adults can learn about an unmentioned social group from what is said about another group, and if this inferential process is sensitive to the context of the utterance. Sixty-three- to five-year-old children and fifty-five adults were introduced to two novel social groups-Stripeys and Dotties-and witnessed a speaker praising abilities of one group (e.g., "the Stripeys are good at building chairs"). To examine the effect of context, we compared situations where the speaker was knowledgeable about the abilities of both groups, and had been queried about the performance of both groups (broad context), versus situations where the speaker was only knowledgeable about one group and was only asked about that group (narrow context). Both preschoolers and adults were sensitive to context: they were more likely to infer that the group not mentioned by the speaker was relatively unskilled, and were more confident about it, in the broad context condition. Our work integrates research in language development and social cognitive development and demonstrates that even young children can "read between the lines," utilizing subtle contextual cues to pick up negative evaluative messages about social groups even from statements that ostensibly do not mention them at all. HIGHLIGHTS: After hearing a speaker praise one group's skill, preschoolers and adults infer that an unmentioned group is relatively less skilled across a range of measures. These inferences are context-sensitive and are stronger when the speaker is knowledgeable of and asked about both groups' skill level. These results shed light on how children may indirectly learn negative stereotypes, especially ones that adults are unlikely to state explicitly. This work extends previous research on children's developing pragmatic ability, as well as their ability to learn about the social world from language.

Challenging the White = Neutral Framework in Psychology

In the United States, White samples are often portrayed as if their racial identities were inconsequential to their thoughts, feelings, and behaviors, and research findings derived from White samples are often portrayed as if they were generalizable to all humans. We argue that these and other practices are rooted in a "White = neutral" framework (i.e., the conceptualization of White samples as nonracial). First, we review existing data and present some new data to highlight the scope of the White = neutral framework. Second, we integrate research from across psychological science to argue that the continued use of the White = neutral framework will prevent psychology from becoming a truly objective and inclusive science for at least three reasons: (a) Research with White samples will be valued over research with samples of color, (b) norms that maintain White neutrality will remain unchallenged, and (c) the role of White identity in psychological processes will remain underspecified and underexamined. Third, we provide recommendations for how to move beyond the White = neutral framework in hopes of encouraging all psychological scientists to move toward a White ≠ neutral framework in which all samples are identified for the unique and diverse perspectives that they bring to the world.

Children Change Their Answers in Response to Neutral Follow-Up Questions by a Knowledgeable Asker

Burgeoning evidence suggests that when children observe data, they use knowledge of the demonstrator's intent to augment learning. We propose that the effects of social learning may go beyond cases where children observe data, to cases where they receive no new information at all. We present a model of how simply asking a question a second time may lead to belief revision, when the questioner is expected to know the correct answer. We provide an analysis of the CHILDES corpus to show that these neutral follow‐up questions are used in parent–child conversations. We then present three experiments investigating 4‐ and 5‐year‐old children's reactions to neutral follow‐up questions posed by ignorant or knowledgeable questioners. Children were more likely to change their answers in response to a neutral follow‐up question from a knowledgeable questioner than an ignorant one. We discuss the implications of these results in the context of common practices in legal, educational, and experimental psychological settings.

The Unintended Consequences of the Things We Say: What Generic Statements Communicate to Children About Unmentioned Categories

Adults frequently use generic language (e.g., "Boys play sports") to communicate information about social groups to children. Whereas previous research speaks to how children often interpret information about the groups described by generic statements, less is known about what generic claims may implicitly communicate about unmentioned groups (e.g., the possibility that "Boys play sports" implies that girls do not). Study 1 (287 four- to six-year-olds, 56 adults) and Study 2 (84 four- to six-year-olds) found that children as young as 4.5 years draw inferences about unmentioned categories from generic claims (but not matched specific statements)-and that the tendency to make these inferences strengthens with age. Study 3 (181 four- to seven-year-olds, 65 adults) provides evidence that pragmatic reasoning serves as a mechanism underlying these inferences. We conclude by discussing the role that generic language may play in inadvertently communicating social stereotypes to young children.

The Pervasive Role of Pragmatics in Early Language

Language is a fundamentally social endeavor. Pragmatics is the study of how speakers and listeners use social reasoning to go beyond the literal meanings of words to interpret language in context. In this article, we take a pragmatic perspective on language development and argue for developmental continuity between early nonverbal communication, language learning, and linguistic pragmatics. We link phenomena from these different literatures by relating them to a computational framework (the rational speech act framework), which conceptualizes communication as fundamentally inferential and grounded in social cognition. The model specifies how different information sources (linguistic utterances, social cues, common ground) are combined when making pragmatic inferences. We present evidence in favor of this inferential view and review how pragmatic reasoning supports children's learning, comprehension, and use of language.

Social Pragmatics: Preschoolers Rely on Commonsense Psychology to Resolve Referential Underspecification

Four experiments show that 4- and 5-year-olds (total N = 112) can identify the referent of underdetermined utterances through their Naïve Utility Calculus-an intuitive theory of people's behavior structured around an assumption that agents maximize utilities. In Experiments 1-2, a puppet asked for help without specifying to whom she was talking ("Can you help me?"). In Experiments 3-4, a puppet asked the child to pass an object without specifying what she wanted ("Can you pass me that one?"). Children's responses suggest that they considered cost trade-offs between the members in the interaction. These findings add to a body of work showing that reference resolution is informed by commonsense psychology from early in childhood.

Still Suspicious: The Suspicious-Coincidence Effect Revisited

Imagine hearing someone call a particular dalmatian a "dax." The meaning of the novel noun dax is ambiguous between the subordinate meaning (dalmatian) and the basic-level meaning (dog). Yet both children and adults successfully learn noun meanings at the intended level of abstraction from similar evidence. Xu and Tenenbaum (2007a) provided an explanation for this apparent puzzle: Learners assume that examples are sampled from the true underlying category (strong sampling), making cases in which there are more observed exemplars more consistent with a subordinate meaning than cases in which there are fewer exemplars (the suspicious-coincidence effect). Authors of more recent work (Spencer, Perone, Smith, & Samuelson, 2011) have questioned the relevance of this finding, however, arguing that the effect occurs only when the examples are presented to the learner simultaneously. Across a series of 12 experiments ( N = 600), we systematically manipulated several experimental parameters that varied across previous studies, and we successfully replicated the findings of both sets of authors. Taken together, our data suggest that the suspicious-coincidence effect in fact is robust to presentation timing of examples but is sensitive to another factor that varied in the Spencer et al. (2011) experiments, namely, trial order. Our work highlights the influence of pragmatics on behavior in experimental tasks.

How language shapes the cultural inheritance of categories

It is widely recognized that language plays a key role in the transmission of human culture, but relatively little is known about the mechanisms by which language simultaneously encourages both cultural stability and cultural innovation. This paper examines this issue by focusing on the use of language to transmit categories, focusing on two universal devices: labels (e.g., shark, woman) and generics (e.g., "sharks attack swimmers"; "women are nurturing"). We propose that labels and generics each assume two key principles: norms and essentialism. The normative assumption permits transmission of category information with great fidelity, whereas essentialism invites innovation by means of an open-ended, placeholder structure. Additionally, we sketch out how labels and generics aid in conceptual alignment and the progressive "looping" between categories and cultural practices. In this way, human language is a technology that enhances and expands the categorization capacities that we share with other animals.

The Ontogeny of Cultural Learning

Developmental research has the potential to address some of the critical gaps in our scientific understanding of the role played by cultural learning in ontogenetic outcomes. The goal of this special section was to gather together leading examples of research on cultural learning across a variety of social contexts and caregiving settings. Although the field of developmental psychology continues to struggle with the persistent problem of oversampling U.S. and Western European populations, we argue that the articles in this special section add to the growing evidence that children everywhere draw on a repertoire of cultural learning strategies that optimize their acquisition of the specific practices, beliefs, and values of their communities. We also identify future directions and outline best practices for the conduct of research on cultural learning.