Re: [xmca] moral life of babies

[Martin Packer <packer@duq.edu>]

On May 7, 2010, at 9:26 AM, mike cole wrote:

> If we are going to get keep into this, the work of Jean Mandler seems to
> require some kind of consideration. She quite explicitly critiques the
> "sensori-motor first"
> idea in Piaget's version of it which seems a least similar to Jay's
> formuation.

I had a couple of hours to spare today, and to avoid thinking about the implications of the indications in my last classes of the semester that none of my students have learned anything at all, I immersed myself in a few of Jean Mandler's texts. My conclusions, for what they are worth, follow...

Mandler is interested in the kind of knowledge that infants must have to be able to learn language, in particular. She argues that Piagetian sensorimotor (SM) schemas are not a sufficient basis for the acquisition of language; she asserts that Piaget too saw this. The "gap" in his theory is that he did not specify in any detail the new kinds of representations which he proposed become possible at the end of the SM period. Also, since speech begins well before the end of the SM stage, psychologists have needed to try to explain early speech in purely SM terms, with little success. 

Mandler is a typical information-processing psychologist. She takes for granted that *all* knowledge has the character of a representation. So it is not a question for her whether SM schemas are representations; she simply assumes that they are. The question for Mandler is whether they are *the right kind* of representation, and the problem she claims to find in Piaget's analysis is that SM schemas are "the wrong kind" for the acquisition of language and for the development of the other higher cognitive functions. This 'representationalist' assumption becomes very clear in the opening sentence of the abstract of one of her articles: "It is agreed that infants require pre-existing conceptual meanings to learn language…." (Mandler, 1994, p. 63). Agreed by whom, one wonders. Not be LSV, for one. Mandler's theory follows closely the general lines of Fodor's representational theory of mind, a central claim of which is that there is an innate, representational language of thought. 

Why are SM schemas inadequate, "not the right sort of representation for learning language"? (p. 367 of Mandler's chapter in Language and Space By Paul Bloom, Mary A. Peterson, Merrill F. Garrett, 1999, MIT Press). The problem is that they are "procedural," and so cannot serve a semiotic function, as Piaget himself explained. They are at best indexical, enabling the infant to recognize familiar objects and for one component of an event to signal what is to come next. Mandler argues that "a sensorimotor schema does not allow independent access to its parts for purposes of denotation or to enable the baby to think independently of the activation of the schema itself" (367). Such claims are a little hard to parse. What would thinking be 'independently' of a scheme's activation? Surely for Piaget this *is* what thinking is for an infant? The point seems to be that Mandler wants to describes the difference between SM schemes and later knowledge in terms of a distinction between procedural knowledge and declarative knowledge. The former, she asserts, is inaccessible to consciousness, while the latter is explicit knowledge that is accessible to consciousness. Conceptual thinking, she proposes, requires conscious awareness. 

Furthermore, since SM schemas "structure perception and control action" this implies, for Mandler, that they have "continuously varying" and "rapidly shifting" parameters, which in turn implies that they cannot "be mapped into a discrete propositional system." Putting a spoon in a bowl, for example, calls for "an intricate sequence of movements." This kind of knowledge is quite distinct from conceptual knowledge, she argues, because "the conceptual system greatly simplifies it, forming a summary of the event that constitutes its meaning." 

Mandler's point seems to be that once language is acquired, it will traffic in representations such as "one object containing another," rather than the procedural knowledge of how to *put* one object inside another. Clearly here she is privileging the propositional (declarative) aspect of speech, and downplaying its  practical, pragmatic character. This is, to put it lightly, a weakness in her approach.

Mandler seems to be a member of the camp that attributes everything to the child once they start to speak. LSV's argument against Stern applies equally to Mandler. Stern, LSV wrote, was proposing an "intellectualist" theory that was fundamentally anti-genetic: ""This is the fundamental error of any intellectualist theory. In its attempt to explain, it begins with what needs explanation" (T&L, p. 94). In the same way, Mandler proposes that "the underlying concepts needed  to learn grammatical categories are ntions such as 'actionality,' ' objecthood,' 'agent,' 'location,' and 'possession,' as though these are abstract concepts that a child will have available the instant they are able to talk grammatically, rather than being notions implicit in the language system that the child only gradually figures out.

In addition to her critique of Piaget, Mandler examines and discards the associationist account of concept formation, in which perceptional similarity of exemplars provides the basis for  generalization which will form superordinate concepts. Eimas and others believe that 6 month old infants' ability to make perceptual discriminations (between horses and zebras, for example) is the basis for the first concepts. Mandler's objection is that this approach can't explain the formation of important properties such as 'animate.' Nor can it explain the transition from perceptual categories to "more abstract or theory-laden" concepts. Mandler claims that her research shows that 7 month olds have already formed concepts such as animal vehicle, and plant. She interprets this as evidence that infants actually form 'global' categories first, and only later differentiate among various exemplars. 

Mandler's own proposal is that there exists in infancy a "preverbal propositional representational system" (373). The representations in this system have the form of "image-schemas," a type of spatial representation "of a special kind" which has an analog (continuous) character but at the same time provides "some of the desirable characteristics of propositional representations." They "form discrete meaning packages," and they can be "combined both sequentially and recursively" with one another. They provide "an excellent medium to bridge the transition from prelinguistic to linguistic representation" (373).

How is an image-schema different from a perceptual representation?  Each image-scheme "represents a meaning" (1994, p. 63). Although they are based upon perceptual information, this information is "redescribed" into "simplified spatial representations." The advantage of these, for Mandler, is that they provide the child with "an analogue-digital interface": perception is analog (continuous), while language, she asserts, is made up of 'discrete propositional forms." 

For example, an infant watching an object move will form an image-schema of its path, without regard to the kind of object. The analysis of paths that the image-schema makes possible then provides the basis for concepts such as animal (because animals move in a particular way). Such "kinetic" representations are abstract and non-perceptual, the earliest kinds of conceptual meanings. They are an intermediary between perception and language. 

But is it in fact the case that for Piaget, SM schemas are representations, a matter of "perceptual categories and motor routines," as Mandler puts it? In fact Piaget claims that it is only at the end of infancy that "the semiotic function" makes its appearance. SM schemas are not representations, they are patterns of interaction with the world the child lives in (primarily the physical rather than the social world, to be sure, but that is an argument for another day). Schemas enable an infant to achieve a practical classification of objects as, for example, 'suckables.' These practical classifications become progressively differentiated and coordinated during infancy. In Piaget's account, this process eventually enables the transition to representational knowledge. Having read several times the book "Play, Dreams, and Imitation," where Piaget tries to explain this transition I have to say I find it unconvincing. But the important here is that Piaget did not consider SM schemas to be representations. They are forms imposed on the infant's actions, actions which are intentional and consciously directed. 

A paper by Müller and Overton (1994) makes many of the same points I have  made here. They argue that "Mandler’s and allied proposals have tended to throw out the baby with the bath water. In this respect, our thesis will be that an action-based logic is a necessary prerequisite for the competence of mental representation and that this remains an indispensable feature of any coherent theory of infant cognition." They list 6 problems that image-schema theory is unable to overcome, most seriously that it too (like Piaget) fails to explain how the capacity for representation becomes possible. Since image-schemas are, by definition, inaccessible to consciousness, they cannot be used by an intentional agent to refer to some object or event. 

Finally, I think Etienne is absolutely correct to insist that we consider the social situation that is defined by the experimental settings of infant research. Much of Mandler's own work involve 'preference tasks' in which infants are seated, observing visual displays. In such situations, perception is divorced from action and the child is separated from important adults. Dubious metatheoretical assumptions are built into this situation.  

Martin

_______________________________________________
xmca mailing list
xmca@weber.ucsd.edu
http://dss.ucsd.edu/mailman/listinfo/xmca