the importance of the nativist–empiricist debate: thinking about primitives without primitive...

3
The Importance of the Nativist–Empiricist Debate: Thinking About Primitives Without Primitive Thinking Barbara Landau Johns Hopkins University ABSTRACT—J. P. Spencer et al. (2009) are intent on dis- mantling the nativist–empiricist debate and they offer three main supporting arguments. It is argued here that these arguments are simply orthogonal to the authors’ main goal. More importantly, it is argued that even the most radical empiricist must embrace some set of primi- tives as the basis for learning; in some cases, rich primi- tives may actually be the learning mechanism. The nativist–empiricist debate is crucial to a healthy dialogue that allows us to think about how learning and devel- opment occur, and is therefore vital to our scientific progress. KEYWORDS—nativism; empiricism; dead reckoning; lan- guage; navigation Spencer et al. (2009) are intent on dismantling the nativist– empiricist debate, and they frame their arguments as follows. First, they argue that ‘‘development is often a nonobvious process that does not easily conform to our intuitions or rational expecta- tions’’ (p. 79). Second, they argue that the fact of evolution ‘‘does not remove the need to explain developmental process’’ (p. 80). Third, they argue that developmentalists must be held to the ‘‘highest of scientific standards’’ because the study of develop- ment is extraordinarily complex. They offer a wide range of empirical phenomena illustrating that development is complex and that human capacities change as a consequence of interac- tion between the organism and its environment. Their three points are surely correct, but they are simply orthogonal to the authors’ main plea: that developmental scien- tists no longer engage in the nativist–empiricist debate. This debate is important precisely because it helps us frame answers to questions about development. Part of the problem is Spencer et al.’s apparent misunderstand- ing of the modern nativist position. First, this position does not embrace the idea that development is simple; it is the very fact of its complexity that drives the nativist to try to understand what kinds of rich structures might be capable of supporting the mas- sive changes that occur over development. Examples abound: Modern studies of language have long sought to understand how hearing one’s native language results in learning that language despite radical differences in experience (Landau & Gleitman, 1985). Modern studies of numerical cognition seek to understand how knowledge of number changes from a system common to many species to the one shown by humans alone (Feigenson, Dehaene, & Spelke, 2004). Modern studies of the development of spatial representation seek to understand how humans (and other species) represent location, how these representations are transformed over one’s own movement, and how these transformed representations support inferences about one’s current location (Wang & Spelke, 2002). Each of these examples requires that we understand the basic building blocks of a system, and how these might change over development, as bodies grow, as minds and brains mature, as the world of interaction expands. Second, the nativist may consider the evolutionary contribu- tion as part of the story of how these building blocks—rather than others—might have become available to the organism, but this position does not entail that development itself does not matter. Our capacity for mental rotation likely embodies abstract The author would like to thank Lisa Feigenson, Justin Halberda, Steven Gross, and members of the Language and Cognition Lab and the Cognitive Development Lab, who all took the time to read the Spencer et al. paper and discuss with me the issues raised therein. Correspondence concerning this article should be addressed to Barbara Landau, Johns Hopkins University, 231 Krieger Hall, 3400 N. Charles St., Baltimore, MD 21218; e-mail: landau@cogsci. jhu.edu. ª 2009, Copyright the Author(s) Journal Compilation ª 2009, Society for Research in Child Development Volume 3, Number 2, Pages 88–90 CHILD DEVELOPMENT PERSPECTIVES

Upload: barbara-landau

Post on 21-Jul-2016

215 views

Category:

Documents


3 download

TRANSCRIPT

CHILD DEVELOPMENT PERSPECTIVES

The Importance of the Nativist–Empiricist Debate:Thinking About Primitives Without Primitive

Thinking

Barbara Landau

Johns Hopkins University

ABSTRACT—J. P. Spencer et al. (2009) are intent on dis-

mantling the nativist–empiricist debate and they offer

three main supporting arguments. It is argued here that

these arguments are simply orthogonal to the authors’

main goal. More importantly, it is argued that even the

most radical empiricist must embrace some set of primi-

tives as the basis for learning; in some cases, rich primi-

tives may actually be the learning mechanism. The

nativist–empiricist debate is crucial to a healthy dialogue

that allows us to think about how learning and devel-

opment occur, and is therefore vital to our scientific

progress.

KEYWORDS—nativism; empiricism; dead reckoning; lan-

guage; navigation

Spencer et al. (2009) are intent on dismantling the nativist–

empiricist debate, and they frame their arguments as follows.

First, they argue that ‘‘development is often a nonobvious process

that does not easily conform to our intuitions or rational expecta-

tions’’ (p. 79). Second, they argue that the fact of evolution ‘‘does

not remove the need to explain developmental process’’ (p. 80).

Third, they argue that developmentalists must be held to the

‘‘highest of scientific standards’’ because the study of develop-

The author would like to thank Lisa Feigenson, Justin Halberda,Steven Gross, and members of the Language and Cognition Lab andthe Cognitive Development Lab, who all took the time to read theSpencer et al. paper and discuss with me the issues raised therein.

Correspondence concerning this article should be addressed toBarbara Landau, Johns Hopkins University, 231 Krieger Hall,3400 N. Charles St., Baltimore, MD 21218; e-mail: [email protected].

ª 2009, Copyright the Author(s)Journal Compilation ª 2009, Society for Research in Child Development

Volume 3, Number

ment is extraordinarily complex. They offer a wide range of

empirical phenomena illustrating that development is complex

and that human capacities change as a consequence of interac-

tion between the organism and its environment.

Their three points are surely correct, but they are simply

orthogonal to the authors’ main plea: that developmental scien-

tists no longer engage in the nativist–empiricist debate. This

debate is important precisely because it helps us frame answers

to questions about development.

Part of the problem is Spencer et al.’s apparent misunderstand-

ing of the modern nativist position. First, this position does not

embrace the idea that development is simple; it is the very fact of

its complexity that drives the nativist to try to understand what

kinds of rich structures might be capable of supporting the mas-

sive changes that occur over development. Examples abound:

Modern studies of language have long sought to understand how

hearing one’s native language results in learning that language

despite radical differences in experience (Landau & Gleitman,

1985). Modern studies of numerical cognition seek to understand

how knowledge of number changes from a system common to many

species to the one shown by humans alone (Feigenson, Dehaene,

& Spelke, 2004). Modern studies of the development of spatial

representation seek to understand how humans (and other species)

represent location, how these representations are transformed over

one’s own movement, and how these transformed representations

support inferences about one’s current location (Wang & Spelke,

2002). Each of these examples requires that we understand the

basic building blocks of a system, and how these might change

over development, as bodies grow, as minds and brains mature, as

the world of interaction expands.

Second, the nativist may consider the evolutionary contribu-

tion as part of the story of how these building blocks—rather

than others—might have become available to the organism, but

this position does not entail that development itself does not

matter. Our capacity for mental rotation likely embodies abstract

2, Pages 88–90

The Importance of the Nativist–Empiricist Debate 89

geometrical principles that have evolved to reflect principles of

the world in which we have evolved (Shepard, 2001). Still, the

developmentalist wants to understand how this capacity sharpens

over development. Third, surely every developmental scien-

tist—whatever his or her theoretical persuasion—would agree

that we should be held to the highest scientific standards: Our

theories should be coherent and as well articulated as possible,

our empirical tests should be designed to test key principles of

the theories, and our interpretation of empirical results should

extend and enhance those theories.

Spencer et al. particularly worry about positing primitives for

developmental systems. They seem to believe that because their

position emphasizes change, they do not have to specify their

own building blocks. But they do—every theory must start with

primitives. If these are not specified, then we cannot specify the

mechanisms of change, as those changes will operate on the

primitive building blocks. The examples that Spencer et al. offer

to support their arguments are meant to emphasize the detailed

nature of change, but they do not offer theories of either what the

building blocks are, or of how change happens. To do so, they

would need to first specify what the basic building blocks of their

theory should be: the elements of sensations, as proposed by

classical empiricists such as Locke and Hume? Distributed pat-

terns of retinal input, or more abstract structures available to

vision, such as reference systems? Conceptual units such as

nouns and verbs, as assumed by many modern developmental

psycholinguists? After specifying the building blocks, they would

have to explain how experience and interaction with the world

change those building blocks and their organization—either by

growing them, reorganizing them, or eliminating them. One can-

not ignore building blocks just because there is change to

explain. Rather, one cannot explain change unless one also spec-

ifies the nature of the building blocks.

In fact, one of the principal domains of study explored by

Spencer et al. is a perfect example of how a theory of an innate

mechanism can help explain learning. This is the domain of

navigation and the mechanism of dead reckoning. Spencer

et al. worry about the proposal that dead reckoning is an innate

capacity whose appearance does not rely on prior relevant

experience. They object that some of the evidence

offered—navigation of alpine geese along novel and lengthy

routes—is not convincing because the geese were not tested at

birth. They further object that infants’ dead reckoning emerges

‘‘gradually after the onset of independent locomotion.’’ If they

mean that dead reckoning becomes more precise as the infant

moves around the world, then yes, but it is the presence of the

dead reckoning mechanism that would allow the infant to cali-

brate over development at all.

It is instructive to read the discussion by Gallistel, Brown,

Carey, Gelman, and Keil (1991) on the nature of the dead-

reckoning mechanism. The most widely cited example of

dead reckoning is the case of the foraging ant, who leaves

its nest and navigates over long distances to find food. After

Child Development Perspectives, V

finding food, it turns and runs in a direct line to the home

nest without benefit of odor trails or the sight or scent of

the nest. To do this, the ant must compute the rough direc-

tion and distance of its nest at every moment along its out-

ward-bound path. There is powerful evidence for this. If the

ant is displaced in midsearch, it will take the course that

would have been correct if it had not been displaced, show-

ing that the computation of distance and direction from the

nest is guiding its (now erroneous) path. The mechanism

underlying the ant’s behavior is path integration—the inte-

gration of direction and distance over time—so that the ant

has a running record of its current position relative to home.

It is a mechanism by which the ant can learn where it is

at the end of its route relative to the home nest, and can

use that representation to produce the right path to get

home each time. This mechanism can sensibly be considered

a building block for the ant: an innate mechanism that can

be used by the ant to learn where he is moment-to-moment,

by measuring distance and direction on the ant’s very first

foray.

Importantly, because this is a learning mechanism, it can be

calibrated, as recently reported by Wittlinger, Wehner, and

Wolf (2006): Desert ants whose legs are artificially lengthened

(hence take longer strides) overestimate distances, whereas

those whose legs are shortened (taking shorter strides) underes-

timate distances. Thus, the mechanism is capable of change

depending on leg length. Equally important, when these altered

ants now start out from their nest on a new journey, they travel

to the source of food and return to their home nest with their

usual accuracy—they immediately adjust to their new leg

length and alter their travel vector accordingly. Thus, the mech-

anism expects change. The application to human infants is clear:

The path integration mechanism is a starting point—a building

block—that enables learning of one’s current location. It can

be calibrated with growth and experience (longer leg length,

walking faster, learning to run, etc.). Learning one’s current

location at time t relative to other locations can support repre-

sentations of a unified cognitive map that permits inferences

about new routes. Our understanding of the dead reckoning

mechanism is still incomplete, but the evidence suggests that it

is a hard-wired part of the ant’s nervous system as well as that

of human infants. The mechanism is a building block that both

embraces and explains learning because it IS the mechanism of

learning.

Spencer et al.’s objections to the idea that dead reckoning is

an innate mechanism seem to hinge on their idea that experience

counts. Well, experience does count. Just not the way they envi-

sion: Experience is what feeds the mechanism to produce repre-

sentations of current location, and experience is what tunes the

mechanism to keep it robust over development. But experience

does not cause the mechanism.

This conclusion will likely not sit well with Spencer et al., for

reasons aptly put by Gallistel et al. (1991, p. 27):

olume 3, Number 2, Pages 88–90

90 Barbara Landau

Here (with dead reckoning) we have a spectacular example of a

dedicated specialized learning mechanism that makes possible the

moment-to-moment acquisition of positional information. People’s

reaction to the claim that this is a learning mechanism is also reve-

latory of the (we believe erroneous) conceptual framework within

which most of us think about learning . . . when one draws the con-

clusion that . . . the learning mechanism is the mechanism that

integrates velocity with respect to time, they balk . . . the way out

of this paradox is to recognize that learning is, generally speaking,

the acquisition of particular mechanisms dedicated to the acquisi-

tion of particular kinds of representations—in this case, a represen-

tation of the animal’s momentary spatial position.

Even the strongest empiricist position—say, that of Locke or

Hume—requires a starting point to explain development. If

Spencer et al. want to suggest that there are no starting points

at all, they will find themselves quite disadvantaged in trying

to develop a coherent theory of any domain. If they want to

argue that the building blocks offered by nativists are the

wrong ones, then fair enough. In that case, they can propose

different building blocks and accompanying mechanisms, and

see how well these account in detail for a given set of empiri-

cal phenomena, thereby providing exciting new contributions to

the ongoing debate. But none of this requires that we abandon

the nativist–empiricist dialogue. Rather, it embraces that

dialogue, and invites us to think deeply about the origins and

Child Development Perspectives, V

nature of human understanding—issues pondered throughout

history by the greatest minds in philosophy, psychology, and

cognitive science.

REFERENCES

Feigenson, L., Dehaene, S., & Spelke, E. S. (2004). Core systems ofnumber. Trends in Cognitive Sciences, 8(7), 307–314.

Gallistel, C. R., Brown, A., Carey, S., Gelman, R., & Keil, F. (1991)Lessons from animal learning for the study of cognitivedevelopment. In S. Carey & R. Gelman (Eds.), The epigenesis ofmind: Essays on biology and cognition (pp. 3–36). Hillsdale, NJ:Erlbaum.

Landau, B., & Gleitman, L. R. (1985) Language and experience:Evidence from the blind child. Cambridge, MA: Harvard UniversityPress.

Shepard, R. (2001). Perceptual-cognitive universals as reflections of theworld. Behavioral and Brain Sciences, 24, 581–601.

Spencer, J. P., Blumberg, M. S., McMurray, B., Robinson, S. R.,Samuelson, L. K., & Tomblin, J. B. (2009). Short arms and talkingeggs: Why we should no longer abide the nativist–empiricistdebate. Child Development Perspectives, 3, 79–87.

Wang, R., & Spelke, E. (2002). Human spatial representation: Insightsfrom animals. Trends in Cognitive Science, 6(9), 376–382.

Wittlinger, M., Wehner, R., & Wolf, H. (2006). The ant odometer:Stepping on stilts and stumps. Science, 312, 1965–1967.

olume 3, Number 2, Pages 88–90