In The Origin of Concepts, Susan Carey argues that object representations are exclusively nonperceptual (Carey Reference Carey2009). I believe that this view is mistaken. There are object perceptions – what I call “perceptual attributives as of objects” – at least in the visual and tactile perceptual modalities of a wide range of animals. There are, of course, also nonperceptual object representations. Some of these might be termed “concepts.” But perceptual object representations are not concepts. Carey's fine book describes important aspects of object representation. However, there is a need to evaluate which of the phenomena that she discusses are postperceptual and which have origins already in perceptual systems.
Carey claims, “representations of object cannot be stated in the vocabulary of perception” (Carey Reference Carey2009, p. 63). She cites Piaget and Quine, with approval, as predecessors in denying that object representations are perceptual. Piaget is credited with the claim that object representations are nonperceptual because they are multimodal (Carey Reference Carey2009, p. 34) (Piaget, Reference Piaget1954). Piaget does not show that all object representations are multimodal, only that some are. (There is evidence that visual object representations do not require intermodal experience.) Visual–perceptual object representations can underlie multimodal object representations.
Carey associates Quine (Quine Reference Quine1960) with this further argument:
If perceptual representations are limited to what currently experienced entities look like…objects cannot be represented as individuals that persist through time, independently of the observer….As Quine pointed out, a perceptual vocabulary does not include fundamental quantificational devices. The child could not represent a given object as the same one as one seen earlier, for sensory representations do not provide criteria for numerical identity. (Carey Reference Carey2009, p. 34; see also pp. 40, 63, 94–6)
Carey adds, seemingly in her own voice:
The criteria for individuation and numerical identity for ordinary objects go beyond perceptual primitives….Although perceptual primitives can specify a currently perceived, bounded entity and its current path of motion, they do not specify that the entity continues to exist when we lose perceptual contact with it. (Carey Reference Carey2009, p. 36)
It is a mistake to require of a system that has object representations that it have quantificational devices, or representations of criteria for numerical identity, or specifications of continuity under loss of perceptual contact (using that “vocabulary”). Nor need object perception represent particulars as persisting, or as independent of an observer, or as unperceived, in order to perceptually represent something as an object or body. A perception as of objects need not represent persistence, observers, perceptual contact, or independence from observers. Such requirements confuse principles according to which perception operates with representations that occur in perceptual object representation.
It is enough that a perceptual system operate under principles that require of the perceiver a capacity to track objects perceptually – where tracking requires coordination of perception with perceptual memory and perceptual anticipation. Therefore, Carey's remarks about quantification, criteria, and the inability of perceptions to specify continuity under loss of perceptual contact, are irrelevant to whether perceptual representations can represent entities as objects or bodies. (I discuss the points of the last two paragraphs in Burge Reference Burge2010, Ch. 7, 9, and 10, especially the section “Body Representation as Originating in Perception.” There I criticize arguments of Carey's colleague, Elizabeth Spelke, that no object representations are perceptual.)
No serious empirical account of perception takes its representations to depend for their content entirely on what happens at a given moment. Interrelations between perceptual systems, perceptual memory, and perceptual anticipation are common even in the simplest perceptual systems. Exercises of perceptual shape constancy or color constancy over time require coordination of current perception with perceptual memory and perceptual anticipation. Such coordinative capacities are present in insects and other relatively simple animals that may not have object perception, and plausibly lack “concepts.” Perception of motion and change are standard topics of vision research.
Object or body perception is constitutively dependent on coordination between the perceptual system and perceptual anticipations of persistence over time; commonly in motion and commonly behind barriers. Therefore, to be a perception as of a body, a representation must be associated with a tendency to perceptually anticipate certain types of continuity. A perceptual representation can present something as looking like a body in current experience, as long as the “look” is associated with perceptual anticipations of certain types of continuity.
If Carey and Quine assume that perceptual representation excludes connection to perceptual anticipation and perceptual memory in the individuation of perceptual representation, they have a mistaken view of perception. Apart from this assumption, and apart from the error of requiring that the perceiver must have quantification, or criteria, in order to perceive something as an object or body, the quoted arguments have no force against the view that perceptual representations include object (body) perceptions.
There is substantial evidence that perceptual body representations occur in the visual systems of many mammals and some birds. Anticipations of continuities that are relevant to perceiving entities as bodies are associated with very early vision. The anticipations are not matters of conception or prediction. (Peterson Reference Peterson and Goldstein2001; Wexler & Held Reference Wexler and Held2005.) Steps have been taken toward localizing the physical basis for object determination in areas of the human brain specialized for vision. (Grill-Spector et al. Reference Grill-Spector, Kourtzi and Kanwisher2001; Kourtzi et al. Reference Kourtzi, Erb, Grodd and Bülthoff2003; Nielsen et al. Reference Nielsen, Logothetis and Rainer2006.) The idea that body representation does not occur in visual (or tactile) perception simply does not accord with research in perceptual psychology. Carey gives no good reason to reject this research.
A deficient view of perception therefore underlies Carey's account of “conceptual” object representation. It is not clear which phenomena that she discusses are postperceptual, perceptual, or both. Most of her characterizations of object “concepts” apply equally to object perceptions (Carey Reference Carey2009, pp. 67–68). The key to the distinction must lie in better characterization of “central inferential processes,” including richer distinctions between the ways postperceptual and perceptual object representations relate to other processes and representations. The deficiency in drawing the perceptual/post-perceptual distinction may hamper Carey's accounts of the other representations of attributes that she discusses: representations of quantity, agency, and cause. Unquestionably, human children and higher animals do have postperceptual representations of these three attributes, as well as the attribute of objecthood. A more accurate view of perceptual representation is needed to delineate origins of postperceptual representation.
In The Origin of Concepts, Susan Carey argues that object representations are exclusively nonperceptual (Carey Reference Carey2009). I believe that this view is mistaken. There are object perceptions – what I call “perceptual attributives as of objects” – at least in the visual and tactile perceptual modalities of a wide range of animals. There are, of course, also nonperceptual object representations. Some of these might be termed “concepts.” But perceptual object representations are not concepts. Carey's fine book describes important aspects of object representation. However, there is a need to evaluate which of the phenomena that she discusses are postperceptual and which have origins already in perceptual systems.
Carey claims, “representations of object cannot be stated in the vocabulary of perception” (Carey Reference Carey2009, p. 63). She cites Piaget and Quine, with approval, as predecessors in denying that object representations are perceptual. Piaget is credited with the claim that object representations are nonperceptual because they are multimodal (Carey Reference Carey2009, p. 34) (Piaget, Reference Piaget1954). Piaget does not show that all object representations are multimodal, only that some are. (There is evidence that visual object representations do not require intermodal experience.) Visual–perceptual object representations can underlie multimodal object representations.
Carey associates Quine (Quine Reference Quine1960) with this further argument:
If perceptual representations are limited to what currently experienced entities look like…objects cannot be represented as individuals that persist through time, independently of the observer….As Quine pointed out, a perceptual vocabulary does not include fundamental quantificational devices. The child could not represent a given object as the same one as one seen earlier, for sensory representations do not provide criteria for numerical identity. (Carey Reference Carey2009, p. 34; see also pp. 40, 63, 94–6)
Carey adds, seemingly in her own voice:
The criteria for individuation and numerical identity for ordinary objects go beyond perceptual primitives….Although perceptual primitives can specify a currently perceived, bounded entity and its current path of motion, they do not specify that the entity continues to exist when we lose perceptual contact with it. (Carey Reference Carey2009, p. 36)
It is a mistake to require of a system that has object representations that it have quantificational devices, or representations of criteria for numerical identity, or specifications of continuity under loss of perceptual contact (using that “vocabulary”). Nor need object perception represent particulars as persisting, or as independent of an observer, or as unperceived, in order to perceptually represent something as an object or body. A perception as of objects need not represent persistence, observers, perceptual contact, or independence from observers. Such requirements confuse principles according to which perception operates with representations that occur in perceptual object representation.
It is enough that a perceptual system operate under principles that require of the perceiver a capacity to track objects perceptually – where tracking requires coordination of perception with perceptual memory and perceptual anticipation. Therefore, Carey's remarks about quantification, criteria, and the inability of perceptions to specify continuity under loss of perceptual contact, are irrelevant to whether perceptual representations can represent entities as objects or bodies. (I discuss the points of the last two paragraphs in Burge Reference Burge2010, Ch. 7, 9, and 10, especially the section “Body Representation as Originating in Perception.” There I criticize arguments of Carey's colleague, Elizabeth Spelke, that no object representations are perceptual.)
No serious empirical account of perception takes its representations to depend for their content entirely on what happens at a given moment. Interrelations between perceptual systems, perceptual memory, and perceptual anticipation are common even in the simplest perceptual systems. Exercises of perceptual shape constancy or color constancy over time require coordination of current perception with perceptual memory and perceptual anticipation. Such coordinative capacities are present in insects and other relatively simple animals that may not have object perception, and plausibly lack “concepts.” Perception of motion and change are standard topics of vision research.
Object or body perception is constitutively dependent on coordination between the perceptual system and perceptual anticipations of persistence over time; commonly in motion and commonly behind barriers. Therefore, to be a perception as of a body, a representation must be associated with a tendency to perceptually anticipate certain types of continuity. A perceptual representation can present something as looking like a body in current experience, as long as the “look” is associated with perceptual anticipations of certain types of continuity.
If Carey and Quine assume that perceptual representation excludes connection to perceptual anticipation and perceptual memory in the individuation of perceptual representation, they have a mistaken view of perception. Apart from this assumption, and apart from the error of requiring that the perceiver must have quantification, or criteria, in order to perceive something as an object or body, the quoted arguments have no force against the view that perceptual representations include object (body) perceptions.
There is substantial evidence that perceptual body representations occur in the visual systems of many mammals and some birds. Anticipations of continuities that are relevant to perceiving entities as bodies are associated with very early vision. The anticipations are not matters of conception or prediction. (Peterson Reference Peterson and Goldstein2001; Wexler & Held Reference Wexler and Held2005.) Steps have been taken toward localizing the physical basis for object determination in areas of the human brain specialized for vision. (Grill-Spector et al. Reference Grill-Spector, Kourtzi and Kanwisher2001; Kourtzi et al. Reference Kourtzi, Erb, Grodd and Bülthoff2003; Nielsen et al. Reference Nielsen, Logothetis and Rainer2006.) The idea that body representation does not occur in visual (or tactile) perception simply does not accord with research in perceptual psychology. Carey gives no good reason to reject this research.
A deficient view of perception therefore underlies Carey's account of “conceptual” object representation. It is not clear which phenomena that she discusses are postperceptual, perceptual, or both. Most of her characterizations of object “concepts” apply equally to object perceptions (Carey Reference Carey2009, pp. 67–68). The key to the distinction must lie in better characterization of “central inferential processes,” including richer distinctions between the ways postperceptual and perceptual object representations relate to other processes and representations. The deficiency in drawing the perceptual/post-perceptual distinction may hamper Carey's accounts of the other representations of attributes that she discusses: representations of quantity, agency, and cause. Unquestionably, human children and higher animals do have postperceptual representations of these three attributes, as well as the attribute of objecthood. A more accurate view of perceptual representation is needed to delineate origins of postperceptual representation.