In her book, Carey (Reference Carey2009) says that the elements of core cognition include some very basic concepts, such as the concept object (p. 41) and the concept agent (p. 186). She also says that the representations of core cognition have an “iconic” format (pp. 68, 458). This appears to be a contradiction, because conceptual representations, even as Carey understands them, cannot be iconic. I vote for a resolution in favor of iconic format.
Carey says that concepts are “units of thought, the constituents of beliefs and theories” (p. 5) and that “words express concepts” (p. 247). Moreover, she frequently interprets infants as using their concepts as predicates, in thinking of something as an object or as an agent (pp. 41, 187, 268). Thus, concepts, for Carey, appear to be components of the sorts of thoughts we can express in sentences. If an infant possesses the concept object, then it can think a thought that we can express in words thus: “That is an object.” That does not mean that only creatures that have language possess concepts, or that all concepts are word-like symbols, which Carey would deny (p. 68).
Iconic representations are mental images. Iconic representations do not represent only sensory qualities such as color and shape, as Carey well knows (p. 135). Mental images can represent a sequence of events among objects. Our understanding of objects may consist largely in our being disposed to imagine that events will unfold in certain ways rather than others (p. 460). If I imagine a ball disappearing behind a screen and the screen coming down to reveal that very ball, I represent the ball as persisting through time and as hidden behind an occluder; in doing so, I utilize knowledge of object permanence. In imagining a hand striking a box behind an occluder, I may exercise an understanding of causation.
Iconic representations are never concepts. Iconic representations of a red ball and a blue ball do not represent what the two balls have in common, as the concept ball does. A concept, such as dog, has an argument place. By substituting a representation of a particular object into that place, we can form a whole thought, such as that Fido is a dog. An iconic representation does not have an argument place. There could be a kind of thinking in pictures in which a mental image of a collie did the work of the concept dog in forming the sorts of thoughts we express in English with the word “dog.” But in that case, the mental image of a collie would cease to be an iconic representation of a particular collie. There are iconic representations of objects, but no iconic representation is the concept object.
Carey frequently infers that a representation is conceptual on the grounds that it cannot be defined in terms of sensory and spatiotemporal qualities and has an “inferential role” (pp. 97, 115, 171, 449). Infants' representations of objects are not merely representations of statistical relations between sensory qualities (p. 34), and they are intermodal (p. 39). In saying that a representation has an inferential role, she does not mean that it is governed by rules of inference defined over strings of symbols (p. 104). She means that it guides expectations (p. 61) and is integrated into several domains of cognition (p. 95). The concept object has an inferential role inasmuch as “objects are represented as solid entities in spatial and causal relations with each other” (p. 103).
But these considerations do not persuasively argue that the representations of core cognition can also be used as predicates. In expecting that there will be two objects behind the screen, not one, the child may be simply imagining two objects. There is no need to suppose that the child judges of each that it is an object and distinct from the other. An imagistic representation of the arrangement behind the screen may be a consequence of the infant's understanding of the way objects behave; it is not merely a synthesis of sensory qualities. It can integrate contributions from several sensory modalities. An imagistic representation can demonstrate an understanding of causal and numerical relations. That understanding can take the form of an imagistic understanding of what one can expect to perceive in the course of observing real objects. An infant can represent an object as solid without judging that it is solid by imagining that it will behave as solid objects in fact behave.
Carey introduces yet a third kind of representation, object files (p. 70). She thinks it's useful to imagine that the infant keeps a file on each object that it perceives, in which it stores information about that object's past. But an object file is not iconic, because an iconic representation is not a thing in which information is collected. And an object file is not the concept object, which we might put to use in representing every object. Yet in Carey's thinking, the notion of object file seems to blur the distinction between concept and iconic representation (p. 459).
Carey frequently attributes to infants a kind of conceptual thought that is not in any way reducible to iconic representation. She thinks experiments show that infants classify objects as agents (p. 186). In comparing explanations of looking-time studies of infants' reactions to causal scenarios, she pits an explanation in terms of the general concept cause against an explanation in terms of generalizations “stated over perceptual features” (p. 241). In considering how infants might solve individuation problems, the only options she considers are that they conceive of objects as falling under kinds and that they conceive of them as having properties (pp. 268–70). None of the options on offer is an explanation in terms of iconic representations.
If Carey had taken on board her own claim that core cognition employs iconic representations, then very different hypotheses might have come into view. Quite generally, an infant's understanding of the world might take the form of its being disposed to imagine certain sequences of events in preference to others. It is not obvious that skills in imagination can only be explained as deriving from prior conceptualizations.
In her book, Carey (Reference Carey2009) says that the elements of core cognition include some very basic concepts, such as the concept object (p. 41) and the concept agent (p. 186). She also says that the representations of core cognition have an “iconic” format (pp. 68, 458). This appears to be a contradiction, because conceptual representations, even as Carey understands them, cannot be iconic. I vote for a resolution in favor of iconic format.
Carey says that concepts are “units of thought, the constituents of beliefs and theories” (p. 5) and that “words express concepts” (p. 247). Moreover, she frequently interprets infants as using their concepts as predicates, in thinking of something as an object or as an agent (pp. 41, 187, 268). Thus, concepts, for Carey, appear to be components of the sorts of thoughts we can express in sentences. If an infant possesses the concept object, then it can think a thought that we can express in words thus: “That is an object.” That does not mean that only creatures that have language possess concepts, or that all concepts are word-like symbols, which Carey would deny (p. 68).
Iconic representations are mental images. Iconic representations do not represent only sensory qualities such as color and shape, as Carey well knows (p. 135). Mental images can represent a sequence of events among objects. Our understanding of objects may consist largely in our being disposed to imagine that events will unfold in certain ways rather than others (p. 460). If I imagine a ball disappearing behind a screen and the screen coming down to reveal that very ball, I represent the ball as persisting through time and as hidden behind an occluder; in doing so, I utilize knowledge of object permanence. In imagining a hand striking a box behind an occluder, I may exercise an understanding of causation.
Iconic representations are never concepts. Iconic representations of a red ball and a blue ball do not represent what the two balls have in common, as the concept ball does. A concept, such as dog, has an argument place. By substituting a representation of a particular object into that place, we can form a whole thought, such as that Fido is a dog. An iconic representation does not have an argument place. There could be a kind of thinking in pictures in which a mental image of a collie did the work of the concept dog in forming the sorts of thoughts we express in English with the word “dog.” But in that case, the mental image of a collie would cease to be an iconic representation of a particular collie. There are iconic representations of objects, but no iconic representation is the concept object.
Carey frequently infers that a representation is conceptual on the grounds that it cannot be defined in terms of sensory and spatiotemporal qualities and has an “inferential role” (pp. 97, 115, 171, 449). Infants' representations of objects are not merely representations of statistical relations between sensory qualities (p. 34), and they are intermodal (p. 39). In saying that a representation has an inferential role, she does not mean that it is governed by rules of inference defined over strings of symbols (p. 104). She means that it guides expectations (p. 61) and is integrated into several domains of cognition (p. 95). The concept object has an inferential role inasmuch as “objects are represented as solid entities in spatial and causal relations with each other” (p. 103).
But these considerations do not persuasively argue that the representations of core cognition can also be used as predicates. In expecting that there will be two objects behind the screen, not one, the child may be simply imagining two objects. There is no need to suppose that the child judges of each that it is an object and distinct from the other. An imagistic representation of the arrangement behind the screen may be a consequence of the infant's understanding of the way objects behave; it is not merely a synthesis of sensory qualities. It can integrate contributions from several sensory modalities. An imagistic representation can demonstrate an understanding of causal and numerical relations. That understanding can take the form of an imagistic understanding of what one can expect to perceive in the course of observing real objects. An infant can represent an object as solid without judging that it is solid by imagining that it will behave as solid objects in fact behave.
Carey introduces yet a third kind of representation, object files (p. 70). She thinks it's useful to imagine that the infant keeps a file on each object that it perceives, in which it stores information about that object's past. But an object file is not iconic, because an iconic representation is not a thing in which information is collected. And an object file is not the concept object, which we might put to use in representing every object. Yet in Carey's thinking, the notion of object file seems to blur the distinction between concept and iconic representation (p. 459).
Carey frequently attributes to infants a kind of conceptual thought that is not in any way reducible to iconic representation. She thinks experiments show that infants classify objects as agents (p. 186). In comparing explanations of looking-time studies of infants' reactions to causal scenarios, she pits an explanation in terms of the general concept cause against an explanation in terms of generalizations “stated over perceptual features” (p. 241). In considering how infants might solve individuation problems, the only options she considers are that they conceive of objects as falling under kinds and that they conceive of them as having properties (pp. 268–70). None of the options on offer is an explanation in terms of iconic representations.
If Carey had taken on board her own claim that core cognition employs iconic representations, then very different hypotheses might have come into view. Quite generally, an infant's understanding of the world might take the form of its being disposed to imagine certain sequences of events in preference to others. It is not obvious that skills in imagination can only be explained as deriving from prior conceptualizations.