Do children learn the integers by induction?

According to one theory about how children learn the meaning of the words for the positive integers, they first learn that "one," "two," and "three" stand for appropriately sized sets. They then conclude by inductive inference that the next numeral in the count sequence denotes the size of sets containing one more object than the size denoted by the preceding numeral. We have previously argued, however, that the conclusion of this Induction does not distinguish the standard meaning of the integers from nonstandard meanings in which, for example, "ten" could mean set sizes of 10, 20, 30,... elements. Margolis and Laurence [Margolis, E., & Laurence, S. (2008). How to learn the natural numbers: Inductive inference and the acquisition of number concepts. Cognition, 106, 924-939] believe that our argument depends on attributing to children "radically indeterminate" concepts. We show, first, that our conclusion is compatible with perfectly determinate meanings for "one" through "three." Second, although the inductive inference is indeed indeterminate - which is why it is consistent with nonstandard meanings - making it determinate presupposes the constraints that the inference is supposed to produce.     

Giving the boot to the bootstrap: how not to learn the natural numbers

According to one theory about how children learn the concept of natural numbers, they first determine that "one", "two", and "three" denote the size of sets containing the relevant number of items. They then make the following inductive inference (the Bootstrap): The next number word in the counting series denotes the size of the sets you get by adding one more object to the sets denoted by the previous number word. For example, if "three" refers to the size of sets containing three items, then "four" (the next word after "three") must refer to the size of sets containing three plus one items. We argue, however, that the Bootstrap cannot pick out the natural number sequence from other nonequivalent sequences and thus cannot convey to children the concept of the natural numbers. This is not just a result of the usual difficulties with induction but is specific to the Bootstrap. In order to work properly, the Bootstrap must somehow restrict the concept of "next number" in a way that conforms to the structure of the natural numbers. But with these restrictions, the Bootstrap is unnecessary.     

Why is number word learning hard? Evidence from bilingual learners

Young children typically take between 18 months and 2 years to learn the meanings of number words. In the present study, we investigated this developmental trajectory in bilingual preschoolers to examine the relative contributions of two factors in number word learning: (1) the construction of numerical concepts, and (2) the mapping of language specific words onto these concepts. We found that children learn the meanings of small number words (i.e., one, two, and three) independently in each language, indicating that observed delays in learning these words are attributable to difficulties in mapping words to concepts. In contrast, children generally learned to accurately count larger sets (i.e., five or greater) simultaneously in their two languages, suggesting that the difficulty in learning to count is not tied to a specific language. We also replicated previous studies that found that children learn the counting procedure before they learn its logic – i.e., that for any natural number, n, the successor of n in the count list denotes the cardinality n + 1. Consistent with past studies, we found that children’s knowledge of successors is first acquired incrementally. In bilinguals, we found that this knowledge exhibits item-specific transfer between languages, suggesting that the logic of the positive integers may not be stored in a language-specific format. We conclude that delays in learning the meanings of small number words are mainly due to language-specific processes of mapping words to concepts, whereas the logic and procedures of counting appear to be learned in a format that is independent of a particular language and thus transfers rapidly from one language to the other in development.     

Are number gestures easier than number words for preschoolers?

Some researchers have argued that children's earliest symbols are based on their sensorimotor experience and that arbitrary symbol-referent mapping poses a challenge for them. If so, exposure to iconic symbols (such as one-finger-for-one-object manual gestures) might help children in a difficult domain such as number. We assessed 44 preschoolers’ processing of number gestures and number words on two tasks: give-a-number and how-many. Children were generally more accurate when mapping number words than number gestures onto a number of toys, suggesting that the iconicity of number gestures does not help children map symbols to numbers. We argue that children learn number gestures as arbitrary symbols and do not take advantage of their iconicity.     

Re-visiting the competence/performance debate in the acquisition of the counting principles

Advocates of the "continuity hypothesis" have argued that innate non-verbal counting principles guide the acquisition of the verbal count list (Gelman & Galistel, 1978). Some studies have supported this hypothesis, but others have suggested that the counting principles must be constructed anew by each child. Defenders of the continuity hypothesis have argued that the studies that failed to support it obscured children's understanding of counting by making excessive demands on their fragile counting skills. We evaluated this claim by testing two-, three-, and four-year-olds both on "easy" tasks that have supported continuity and "hard" tasks that have argued against it. A few noteworthy exceptions notwithstanding, children who failed to show that they understood counting on the hard tasks also failed on the easy tasks. Therefore, our results are consistent with a growing body of evidence that shows that the count list as a representation of the positive integers transcends pre-verbal representations of number.     

Preschool children master the logic of number word meanings

Although children take over a year to learn the meanings of the first three number words, they eventually master the logic of counting and the meanings of all the words in their count list. Here, we ask whether children's knowledge applies to number words beyond those they have mastered: Does a child who can only count to 20 infer that number words above 'twenty' refer to exact cardinal values? Three experiments provide evidence for this understanding in preschool children. Before beginning formal education or gaining counting skill, children possess a productive symbolic system for representing number.     

Learning the Natural Numbers as a Child

How do we get out knowledge of the natural numbers? Various philosophical accounts exist, but there has been comparatively little attention to psychological data on how the learning process actually takes place. I work through the psychological literature on number acquisition with the aim of characterising the acquisition stages in formal terms. In doing so, I argue that we need a combination of current neologicist accounts and accounts such as that of Parsons. In particular, I argue that we learn the initial segment of the natural numbers on the basis of the Fregean definitions, but do not learn the natural number structure as a whole on the basis of Hume's principle. Therefore, we need to account for some of the consistency of our number concepts with the Dedekind‐Peano axioms in other terms.     

Does learning to count involve a semantic induction?

We tested the hypothesis that, when children learn to correctly count sets, they make a semantic induction about the meanings of their number words. We tested the logical understanding of number words in 84 children that were classified as "cardinal-principle knowers" by the criteria set forth by Wynn (1992). Results show that these children often do not know (1) which of two numbers in their count list denotes a greater quantity, and (2) that the difference between successive numbers in their count list is 1. Among counters, these abilities are predicted by the highest number to which they can count and their ability to estimate set sizes. Also, children's knowledge of the principles appears to be initially item-specific rather than general to all number words, and is most robust for very small numbers (e.g., 5) compared to larger numbers (e.g., 25), even among children who can count much higher (e.g., above 30). In light of these findings, we conclude that there is little evidence to support the hypothesis that becoming a cardinal-principle knower involves a semantic induction over all items in a child's count list.     

Rips的“证明心理学理论”述评

该文对Rips提出的“证明心理学理论”做了综合述评。这一理论主要包含三方面的内容:对推理过程与人类记忆相互关系的解释;(2)根据逻辑学中的“自然推理规则”进行修正后用于解释人类推理过程的推理规则;(3)关于如何控制推理过程的论述。Rips认为他于1983年设计并实施的以“自然推理系统”所含各推理规则为实验材料的实验结果支持该理论的基本观点。     

Domain-Specific Principles Affect Learning and Transfer in Children

In this paper I discuss the curious lack of contact between developmental psychologists studying the principles of early learning and those concentrating an later learning in children, where predispositions to learn certain types of concepts are less readily discussed. Instead, there is tacit agreement that learning and transfer mechanisms are content-independent and age-dependent. I argue here that one cannot study learning and transfer in a vacuum and that children's ability to learn is intimately dependent on what they are required to learn and the context in which they must learn it. Specifically, I argue that children learn and transfer readily, even in traditional laboratory settings, if they are required to extend their knowledge about causal mechanisms that they already understand. This point is illustrated in a series of studies with children from 1 to 3 years of age learning about simple mechanisms of physical causality (pushing-pulling, wetting, cutting, etc.). In addition, I document children's difficulty learning about causally impossible events, such as pulling with strings that do not appear to make contact with the object they are pulling. Even young children transfer on the basis of deep structural principles rather than perceptual features when they have access to the requisite domain-specific knowledge. I argue that a search for causal explanations is the basis of broad understanding, of wide patterns of generalization, and of flexible transfer and creative inferential projections—in sum, the essential elements of meaningful learning.     

Children’s mapping between symbolic and nonsymbolic representations of number

When children learn to count and acquire a symbolic system for representing numbers, they map these symbols onto a preexisting system involving approximate nonsymbolic representations of quantity. Little is known about this mapping process, how it develops, and its role in the performance of formal mathematics. Using a novel task to assess children’s mapping ability, we show that children can map in both directions between symbolic and nonsymbolic numerical representations and that this ability develops between 6 and 8 years of age. Moreover, we reveal that children’s mapping ability is related to their achievement on tests of school mathematics over and above the variance accounted for by standard symbolic and nonsymbolic numerical tasks. These findings support the proposal that underlying nonsymbolic representations play a role in children’s mathematical development.     

Bootstrapping in a language of thought: a formal model of numerical concept learning

In acquiring number words, children exhibit a qualitative leap in which they transition from understanding a few number words, to possessing a rich system of interrelated numerical concepts. We present a computational framework for understanding this inductive leap as the consequence of statistical inference over a sufficiently powerful representational system. We provide an implemented model that is powerful enough to learn number word meanings and other related conceptual systems from naturalistic data. The model shows that bootstrapping can be made computationally and philosophically well-founded as a theory of number learning. Our approach demonstrates how learners may combine core cognitive operations to build sophisticated representations during the course of development, and how this process explains observed developmental patterns in number word learning.     

Biased belief in the Bayesian brain: A deeper look at the evidence

A recent critique of hierarchical Bayesian models of delusion argues that, contrary to a key assumption of these models, belief formation in the healthy (i.e., neurotypical) mind is manifestly non-Bayesian. Here we provide a deeper examination of the empirical evidence underlying this critique. We argue that this evidence does not convincingly refute the assumption that belief formation in the neurotypical mind approximates Bayesian inference. Our argument rests on two key points. First, evidence that purports to reveal the most damning violation of Bayesian updating in human belief formation is counterweighted by substantial evidence that indicates such violations are the rare exception—not a common occurrence. Second, the remaining evidence does not demonstrate convincing violations of Bayesian inference in human belief updating; primarily because this evidence derives from study designs that produce results that are not obviously inconsistent with Bayesian principles.     

To infinity and beyond: Children generalize the successor function to all possible numbers years after learning to count

Recent accounts of number word learning posit that when children learn to accurately count sets (i.e., become “cardinal principle” or “CP” knowers), they have a conceptual insight about how the count list implements the successor function – i.e., that every natural number n has a successor defined as n + 1 (Carey, 2004, 2009; Sarnecka & Carey, 2008). However, recent studies suggest that knowledge of the successor function emerges sometime after children learn to accurately count, though it remains unknown when this occurs, and what causes this developmental transition. We tested knowledge of the successor function in 100 children aged 4 through 7 and asked how age and counting ability are related to: (1) children’s ability to infer the successors of all numbers in their count list and (2) knowledge that all numbers have a successor. We found that children do not acquire these two facets of the successor function until they are about 5½ or 6 years of age – roughly 2 years after they learn to accurately count sets and become CP-knowers. These findings show that acquisition of the successor function is highly protracted, providing the strongest evidence yet that it cannot drive the cardinal principle induction. We suggest that counting experience, as well as knowledge of recursive counting structures, may instead drive the learning of the successor function.     

Knowledge, expectations, and inductive reasoning within conceptual hierarchies

Previous research (e.g. Cognition 64 (1997) 73) suggests that the privileged level for inductive inference in a folk biological conceptual hierarchy does not correspond to the “basic” level (i.e. the level at which concepts are both informative and distinct). To further explore inductive inference within conceptual hierarchies, we examine relations between knowledge of concepts at different hierarchical levels, expectations about conceptual coherence, and inductive inference. In Experiments 1 and 2, 5- and 8-year-olds and adults listed features of living kind (Experiments 1 and 2) and artifact (Experiment 2) concepts at different hierarchical levels (e.g. plant, tree, oak, desert oak), and also rated the strength of generalizations to the same concepts. For living kinds, the level that showed a relative advantage on these two tasks differed; the greatest increase in features listed tended to occur at the life-form level (e.g. tree), whereas the greatest increase in inductive strength tended to occur at the folk-generic level (e.g. oak). Knowledge and induction also showed different developmental trajectories. For artifact concepts, the levels at which the greatest gains in knowledge and induction occurred were more varied, and corresponded more closely across tasks. In Experiment 3, adults reported beliefs about within-category similarity for concepts at different levels of animal, plant and artifact hierarchies, and rated inductive strength as before. For living kind concepts, expectations about category coherence predicted patterns of inductions; knowledge did not. For artifact concepts, both knowledge and expectations predicted patterns of induction. Results suggest that beliefs about conceptual coherence play an important role in guiding inductive inference, that this role may be largely independent of specific knowledge of concepts, and that such beliefs are especially important in reasoning about living kinds.     

Demonstrative Induction and the Skeleton of Inference

It has been common wisdom for centuries that scientific inference cannot be deductive; if it is inference at all, it must be a distinctive kind of inductive inference. According to demonstrative theories of induction, however, important scientific inferences are not inductive in the sense of requiring ampliative inference rules at all. Rather, they are deductive inferences with sufficiently strong premises. General considerations about inferences suffice to show that there is no difference in justification between an inference construed demonstratively or ampliatively. The inductive risk may be shouldered by premises or rules, but it cannot be shirked. Demonstrative theories of induction might, nevertheless, better describe scientific practice. And there may be good methodological reasons for constructing our inferences one way rather than the other. By exploring the limits of these possible advantages, I argue that scientific inference is neither of essence deductive nor of essence inductive.     

An association between understanding cardinality and analog magnitude representations in preschoolers

The preschool years are a time of great advances in children’s numerical thinking, most notably as they master verbal counting. The present research assessed the relation between analog magnitude representations and cardinal number knowledge in preschool-aged children to ask two questions: (1) Is there a relationship between acuity in the analog magnitude system and cardinality proficiency? (2) Can evidence of the analog magnitude system be found within mappings of number words children have not successfully mastered? To address the first question, Study 1 asked three- to five-year-old children to discriminate side-by-side dot arrays with varying differences in numerical ratio, as well as to complete an assessment of cardinality. Consistent with the analog magnitude system, children became less accurate at discriminating dot arrays as the ratio between the two numbers approached one. Further, contrary to prior work with preschoolers, a significant correlation was found between cardinal number knowledge and non-symbolic numerical discrimination. Study 2 aimed to look for evidence of the analog magnitude system in mappings to the words in preschoolers’ verbal counting list. Based on a modified give-a-number task ( [Wynn, 1990] and [Wynn, 1992] ), three- to five-year-old children were asked to give quantities between 1 and 10 as many times as possible in order to assess analog magnitude variability within their developing cardinality understanding. In this task, even children who have not yet induced the cardinality principle showed signs of analog representations in their understanding of the verbal count list. Implications for the contribution of analog magnitude representations towards mastery of the verbal count list are discussed in light of the present work.     

Children's Understanding of the Natural Numbers’ Structure

When young children attempt to locate numbers along a number line, they show logarithmic (or other compressive) placement. For example, the distance between “5” and “10” is larger than the distance between “75” and “80.” This has often been explained by assuming that children have a logarithmically scaled mental representation of number (e.g., Berteletti, Lucangeli, Piazza, Dehaene, & Zorzi, 2010 ; Siegler & Opfer, 2003 ). However, several investigators have questioned this argument (e.g., Barth & Paladino, 2011 ; Cantlon, Cordes, Libertus, & Brannon, 2009 ; Cohen & Blanc‐Goldhammer, 2011 ). We show here that children prefer linear number lines over logarithmic lines when they do not have to deal with the meanings of individual numerals (i.e., number symbols, such as “5” or “80”). In Experiments 1 and 2, when 5‐ and 6‐year‐olds choose between number lines in a forced‐choice task, they prefer linear to logarithmic and exponential displays. However, this preference does not persist when Experiment 3 presents the same lines without reference to numbers, and children simply choose which line they like best. In Experiments 4 and 5, children position beads on a number line to indicate how the integers 1–100 are arranged. The bead placement of 4‐ and 5‐year‐olds is better fit by a linear than by a logarithmic model. We argue that previous results from the number‐line task may depend on strategies specific to the task.     

A probabilistic model of theory formation

Concept learning is challenging in part because the meanings of many concepts depend on their relationships to other concepts. Learning these concepts in isolation can be difficult, but we present a model that discovers entire systems of related concepts. These systems can be viewed as simple theories that specify the concepts that exist in a domain, and the laws or principles that relate these concepts. We apply our model to several real-world problems, including learning the structure of kinship systems and learning ontologies. We also compare its predictions to data collected in two behavioral experiments. Experiment 1 shows that our model helps to explain how simple theories are acquired and used for inductive inference. Experiment 2 suggests that our model provides a better account of theory discovery than a more traditional alternative that focuses on features rather than relations.     

Decision processes in verifying category membership statements: Implications for models of semantic memory

Current models of semantic memory assume that natural categories are well-defined. Specific predictions of two such models, the Smith, Shoben, and Rips (1974a) two-stage feature comparison model and the Glass and Holyoak (1974/75) ordered search model, were tested and disconfirmed in Experiment I. We propose an alternative model postulating fuzzy categories represented as sets of characteristic properties. This model, combined with a Bayesian decision process, accounts for the results of three additional experiments, as well as for the major findings in the semantic memory literature. We argue that people verify category membership statements by assessing similarity relations between concepts rather than by using information which logically specifies the truth value of the sentence. Our data also imply that natural categories are fuzzy rather than well-defined.