2009 TheOriginofConcepts

Jump to: navigation, search

Subject Headings: A-not-B error; bootstrapping process; conceptual change; epistemology


Some chapters

Cited By


Book Overview

Only human beings have a rich conceptual repertoire with concepts like tort, entropy, Abelian group, mannerism, icon and deconstruction. How have humans constructed these concepts? And once they have been constructed by adults, how do children acquire them? While primarily focusing on the second question, in The Origin of Concepts, Susan Carey shows that the answers to both overlap substantially. Carey begins by characterizing the innate starting point for conceptual development, namely systems of core cognition. Representations of core cognition are the output of dedicated input analyzers, as with perceptual representations, but these core representations differ from perceptual representations in having more abstract contents and richer functional roles. Carey argues that the key to understanding cognitive development lies in recognizing conceptual discontinuities in which new representational systems emerge that have more expressive power than core cognition and are also incommensurate with core cognition and other earlier representational systems. Finally, Carey fleshes out Quinian bootstrapping, a learning mechanism that has been repeatedly sketched in the literature on the history and philosophy of science. She demonstrates that Quinian bootstrapping is a major mechanism in the construction of new representational resources over the course of childrens cognitive development. Carey shows how developmental cognitive science resolves aspects of long-standing philosophical debates about the existence, nature, content, and format of innate knowledge. She also shows that understanding the processes of conceptual development in children illuminates the historical process by which concepts are constructed, and transforms the way we think about philosophical problems about the nature of concepts and the relations between language and thought.

Chapter 1. Some preliminaries

Human beings, alone among animals, create rich conceptual understanding, representing concepts such as evolution, electron, cancer, infinity, galaxy… What accounts for the human capacity for conceptual representations? Any concept we may focus on has two ultimate sources — evolution and culture. Some concepts, such as object and number, arise in part through natural selection, in the course of hominid evolution or in some ancestor common to the apes, or to the primates, or even earlier. Other concepts, such as kayak fraction, and gene, are cultural constructions. Humans create complex artifacts, and religious, political, and scientific institutions, and in turn create new representational resources.

Although it is obvious that we must look both to evolutionary and historical processes to account for the origins of concepts, the problem for cognitive science is to provide a precise, explanatory, account of the origin of any particular concept in which we may be interested. When it comes to the concept of integer, or of animal, for examples, what are the relevant innate representational resources bequeathed to us by evolution? In creating such concepts, must we go beyond innately given representations? In what ways, and by what processes, do individuals and groups of individuals construct new representational resources? How is knowledge culturally constructed and maintained? These questions must be answered case by case--there is no reason, at the outset, to expect the answers for spatial knowledge, for example, to be the same as the answers for mathematical or biological knowledge or for knowledge of language or chess.

This book develops five case studies in detail: the concepts object, intentional agent, number, animal and living thing, and matter, weight and density. In the course of exploring these cases many others are touched, including artifact, contagion, and causality. But what really matters to me is not the cases (although I do admit finding each one intrinsically fascinating), but the lessons to be drawn from them. My goal in this book, as in the Nicod lectures, is to demonstrate that the disciplines of cognitive science now have the empirical and theoretical tools to turn age-old philosophical dilemmas into relatively straightforward scientific problems. I shall illustrate the progress science has made in resolving debates about the existence, nature, content and format of innate knowledge, about the thesis that conceptual resources are continuous throughout the life space, about the nature of concepts and intuitive theories, about the distinction between conceptual change and belief revision, and about controversies concerning the relations between language and thought.

1.1 Concepts and mental representations

Concepts are units of thought, the constituents of beliefs and theories, and those that interest me here are roughly the grain of single lexical items. Indeed, the meanings of words are paradigm examples of concepts. I am concerned with the mental representation of concepts; I use phrases such as “the infant’s concept animal” to mean the infant’s representation of animals. I assume representations are states of the nervous system that have content, that refer to concrete or abstract entities, to properties, to events. I do not attempt a philosophical analysis of mental representations; I will not try to say how it is that some states of the nervous system have symbolic content. Such a theory would explain how the extension of a given representation is determined, as well as providing a computational account of how that representation fulfills its particular inferential role, how it functions in thought.1 Here I merely assume that such a theory will be forthcoming. In the pages to come, I work backwards from behavioral evidence for some concept’s extension and inferential role to characterize that concept’s content and to specify something of its nature and format of representation.

There are many different types of mental representations and one challenge to cognitive science is to find the principled distinctions among them. Different types of representations may well have theoretically important differences in origins, developmental trajectories, types of conceptual roles, and relations to their extensions. Also, some theories of conceptual development posit shifts in kinds of mental representations available to children of different ages — from a perceptual similarity space to natural kind concepts (Quine, 1977), from sensori-motor to symbolic representations (Piaget, 1954), from implicit to explicit representations (Karmiloff-Smith, 1990), for examples. Such theories depend, of course, on defensible distinctions among types of mental representations.

I will join forces with the many writers who draw a distinction between perceptual representations, on the one hand, and conceptual representations, on the other. Chapter 2 examines thesis that infants begin with perceptual representations and only construct conceptual representations later in development. Differentiating the perceptual from the conceptual is difficult. There are probably many different distinctions at work here, and most are probably ends of continua rather than categorical. An intuitive characterization of perceptual representations as what things in the world look like, sound like, feel like, taste like, contrasts these with conceptual representations as what things is the world are.. Distinctive properties of perceptual representations include, first of all, that their extensions are fixed by virtue of innate, modular, sensory input analyzers. There are innate shape analyzers, phoneme detectors, color detectors, motion detectors, and so forth. That representations of red have the content red is ensured by evolution, by how color vision works. Second, perceptual representations have very little in the way of inferential role. Almost nothing else follows from the fact that something is red. Third, and related to the above two points, perceptual representations are inferentially close the output of sensori-analyzers. Consider the difference between the representation of red or loud, on the one hand, and the representation of electron or life, on the other. Although we certainly can sometimes identify electrons or living things perceptual evidence, there is a long inferential chain between a path in a cloud chamber to the presence of an electron, or from what a bacteria colony on a petri dish looks like to the fact that it contains living things.

Natural kind concepts, paradigm conceptual representations, are at the other end of the continuum, contrasting with perceptual representations in all three respects. There are no innate input analyzers for tigers or electrons, natural kind concepts have rich conceptual roles, and there is a long inferential chain between the perceptible properties of natural kinds and the content of concepts of natural kinds. According to the Kripke/Putnam (Kripke, 1972; Putnam, 1975) analyses of natural kind concepts, their extensions are fixed not by the mind but by some social process of ostensive definition and by the essential nature (a metaphysical matter, not an epistemelogical one) of the entities so dubbed. The discovery of the extension of gold or of wolf is a matter for science, not for philosophy, linguistics, or psychology. As for the psychology of natural kind concepts, they fall under the assumption of “psychological essentialism,” (Medin and Ortony, 1989). It is a fact about our mind that we assume (usually correctly, as it turns out, but it needn’t be) that individuals of a given natural kind have hidden essences which both determine their kind and their surface properties. Often we have no fleshed-out guess as to a kind’s essential properties.

A natural kind concept’s features fall along a continuum from core to periphery, a continuum determined by explanatory depth (Ahn, Kim, Lassaline & Dennis, 2000; Keil, 1989). Its core, its essence, consists of its inferentially deepest features, and for natural kinds, these are its causally deepest features. Thus, the analysis of concepts of natural kinds is deeply intertwined with the analysis of the conceptual structures that represent causal/explanatory knowledge: intuitive theories.

Some writers deny a principled distinction between perceptual representations and conceptual representations, claiming that all mental representations at root perceptual representations (e.g., Thelan, Schoner, Scheir & Smith, 2000). Others (e.g., Quine, 1977; Piaget, 1954) grant the distinction and believe that conceptual development in the first few years of life involves a transition from perceptual representations alone to a representational repetoire that contains both types. These positions are considered in Chapter 2.

This book’s first major thesis is that there is a third type of conceptual structure, called “core knowledge” by Spelke, that differs systematically from both perceptual domains of representation and from theoretical conceptual knowledge. I shall argue that core knowledge is the developmental foundation of human conceptual understanding. Like perceptual domains, the entities in core domains of knowledge are identified by modular innate perceptual input devices, but the representations are conceptual, not perceptual. Unlike perceptual representations, they have relatively rich inferential roles in thought, and there is a longer inferential chain between the output of sensori-analyzers and the content of the representations that articulate core knowledge. However the conceptual role of the concepts that articulate core knowledge is vastly less rich than that of the concepts embedded in intuitive theories, and the inferential depth between perceptual properties and the content of core knowledge is vastlyin the case of intuitive theories. Finally, knowledge acquisition in core domains is supported by innate domain-specific learning devices, whereas that in intuitive theories is not. Chapters 3 and 4 characterize core knowledge more fully and summarize evidence for human core knowledge of objects, contact causality, intentional causality, number and emotion.

Chapter 2. Innate Perceptual Representations: the Empiricist Picture

The British Empiricists’ picture of conceptual development finds articulate and ardent defenders to this day. This staying power has two explanations. First of all, the empiricists staked out an ambitious set of phenomena that a theory of concepts must be responsible for. They sought to explain how concepts refer, how people categorize, how concepts function in thought, how human knowledge is warranted, and how human knowledge is acquired. And they offered a theory that accounted for all of these phenomena in a comprehensive, integrated, and unmatched manner. Second, the theory contains important grains of truth.


Chapter 3. Core knowledge of objects

3.1. Core knowledge

Chapter 2 challenged a widely shared assumption about the ontogenetic origins of human conceptual understanding, the assumption that the initial stock of representations are limited to perceptual or sensori-motor primitives. My characterization of an alternative to the empiricist picture draws especially on the writings of Renee Baillargeon, Randy Gallistel, Rochel Gelman, Alan Leslie, and Elizabeth Spelke. These writers and I believe that human cognition, like that of all animals, begins with highly structured innate mechanisms designed to build representations with specific content. Following Spelke et al. (1992) I call these real world content domains “core domains,” and the mental structures that represent them “core knowledge.”

Core knowledge has several properties. First, core knowledge has conceptual content; it cannot be characterized in terms perceptual or sensorimotor primitives. Also, as we will see here, core knowledge is conceptual in a second sense; it includes representations that are explicit in the sense of being accessible, attended, and available to thought and to guide action. Second, core knowledge is articulated in terms of representations that are created by specialized perceptual input analyzers. As in the empiricist view of concepts, part of what determines the content of a concept in core knowledge is a causal connection between the real world entities in its domain and the concept. The causal connection between a concept in core knowledge and its referents is mediated by the operation of perceptual input analyzers that have been constructed, through natural selection, specifically for the purpose of representing certain classes of entities in the world. Third, the perceptual analysis devices that identify the entities that fall under core domains continue to operate throughout life. Core knowledge is elaborated during development, as core knowledge systems are learning devices, but it is never rendered irrelevant. It is never overturned nor lost as are earlier intuitive theories that are replaced by later, incommensurable ones. Fourth, some core knowledge (including that of objects) is shared by other animals. At least some early developing cognitive systems in humans have a long evolutionary history. Fifth, core knowledge acquisition is supported by domain specific learning devices. In most of these respects, core knowledge representations resemble perceptual representations and differ from those in almost all other domains of human conceptual knowledge, including intuitive theories.




 AuthorvolumeDate ValuetitletypejournaltitleUrldoinoteyear
2009 TheOriginofConceptsSusan CareyThe Origin of Concepts2009
AuthorSusan Carey +
titleThe Origin of Concepts +
year2009 +