ocular dominance, input, and language

[Gary Marcus]

Dear Ping, Dear Brian, cc: Info-childes,

1. Ping, you haven't said what would NOT count as interactive brain
development.

2. As I said Thursday, "There's no guarantee of course that the mechanisms
that shape the ocular
dominance columns in an enucleated ferret have anything to do with the
mechanisms that shape the brain circuitry for language or cognition". I
only raised the example because you brought up Nelson's discussion of brain
development, and Nelson himself brought up visual deprivation experiments.
Since his article did a good job of indicating one of the two major
developmental options, but not the other, I thought it would be useful to
illustrate the other.  Clearly, such studies must be interpreted with care
when they are brought to bear on questions about language development.

Still, let me add that I am hardly the first person to consider  visual
development in the context of wondering whether, say, there might be an
innate language acquisition device.  Since animal models and the like make
questions about the neural basis of visual development far more tractable
than questions about the neural basis of linguistic development, other
researchers have frequently considered visual development  as a source of
insight into questions about the developmental origins of language; one
prominent recent example that mentioned ocular dominance columns and
mechanisms that might form them was Elman's et al'a (1996, MIT Press)
Rethinking Innateness.

At that time, the facts that they cited may have seemed to accord with a
more experientially-driven view of visual development, but, as I indicated,
many recent studies (summarized in Hubener and Bonhoeffer's "Eyes wide
shut" article in this month's Nature Neuroscience) now point to at least
some of the basic organization being constructed prior to experience.  It
would, in my view, be a mistake to stop seriously considering visual
development  as a source of insight into questions about language simply
because the facts about visual development no longer seem to fit so nicely
with an experientially-driven view of fundamental brain organization.

3. Although our data from seven-month-olds (Marcus, G.F., Vijayan, S.,
Bandi Rao, S., & Vishton, P.M. (1999). Rule learning in 7-month-old
infants. Science, 283, 77-80)  are consistent with the possibility that the
underlying generalization mechanism is innate, I would not claim that I
have proven that it is.

4. I have discussed the Seidenberg and Elman model elsewhere. For pointers,
and a full list of the 20 or so papers in the professional literature
that  have discussed our infant results, see below. (I will include a full
treatment of these papers, and of the neural developmental issues we have
been discussing, in my forthcoming book _The Algebraic Mind: Integrating
Connectionism and Cognitive Science_, due out in Fall 2000. For excerpts
from an earlier draft and discussion by Elman and others, see my web page.)

5. Brian, I checked with Larry Katz to make sure that he agreed with my
interpretation of his results (""Learning must play some role
in  fine-tuning things, but the basic structure seems to be available prior
to  (womb-external) experience." ). He did in fact agree, writing "What
we've shown, I think, is that ... competition [driven by womb-external
visual experience]  is not  required to initially form the structure, but
can, once it's formed, alter  the fine structure. You've captured that
exactly in your written quote."

Best wishes,
Gary

p.s. Although I have been enjoying this discussion, the end of the semester
is nigh, so I may not be able to post again for a while.

Department of Psychology
New York University
6 Washington Place
NY, NY 10012

o: 212-998-3551
fax: 212-995-4866
e-mail: gary.marcus at nyu.edu

web: http://www.psych.nyu.edu/gary/



                                 Articles that have discussed the Marcus et
al (1999) results

Altmann, G.T.M., & Dienes, Z. (1999). Rule learning by seven-month-old
infants and neural networks. Science, 284, 875a.

Berent, I. (1999). Infant rule-learning and the obligatory contour
principle: Submitted manuscript, Florida Atlantic University.

Christiansen, M.H., & Curtin, S.L. (1999). The power of statistical
learning: No need for algebraic rules. In M. Hahn & S.C. Stoness (Eds.),
Proceedings of the Twenty first Annual Conference of the Cognitive Science
Society (pp. 114-119). Mahwah, NJ: Erlbaum.

Christiansen, M.H., & Curtin, S.L. (1999). Transfer of Learning: Rule
acquisition or statistical learning? Trends in Cognitive Sciences, 3, 289-290.

Dominey, P.F., & Ramus, F. (in press). Neural network processing of natural
language: I. Sensitivity to Serial, Temporal and Abstract Structure of
language in the Infant. Language and Cognitive Processes.

Eimas, P. (1999). Do infants learn grammar with algebra or statistics?
Science, 284, 435-436.

Gasser, M., & Colunga, E. (1999). Babies, Variables, and Connectionist
Networks. In M. Hahn & S.C. Stoness (Eds.), Proceedings of the Twenty first
Annual Conference of the Cognitive Science Society (pp. 794). Mahwah, NJ:
Erlbaum.

Kuehne, S.E., Gentner, D., & Forbus, K.D. (1999). Modeling rule learning by
seven-month-old infants: A symbolic approach. Manuscript in preparation:
Northwestern University.

Marcus, G.F. (1999a). Connectionism: with or without rules? Response to
J.L. McClelland and D.C. Plaut. Trends in Cognitive Sciences,, 3, 168-170.

Marcus, G.F. (1999b). Do infants learn grammar with algebra or statistics?
Response to Seidenberg & Elman, Eimas, and Negishi. Science, 284, 436-437.

Marcus, G.F. (1999c). Reply to Christiansen & Curtin. Trends in Cognitive
Sciences, 3, 290-291.

Marcus, G.F. (1999d). Reply to Seidenberg & Elman. Trends in Cognitive
Sciences, 3, 289.

Marcus, G.F. (1999e). Rule learning by seven-month-old infants and neural
networks: Response to Altmann and Dienes. Science, 284, 875a.

Marcus, G.F., Vijayan, S., Bandi Rao, S., & Vishton, P.M. (1999). Rule
learning in 7-month-old infants. Science, 283, 77-80.

McClelland, J.L., & Plaut, D.C. (1999). Does generalization in infant
learning implicate abstract algebra-like rules? Trends in Cognitive
Sciences,, 3, 166-168.

Negishi, M. (1999). Do infants learn grammar with algebra or statistics?
Science, 284, 435.

Seidenberg, M.S., & Elman, J.L. (1999a). Do infants learn grammar with
algebra or statistics? Science, 284, 435-436.

Seidenberg, M.S., & Elman, J.L. (1999b). Networks are not 'hidden rules'.
Trends in Cognitive Sciences, 3, 288-289.

Shastri, L. (1999). Infants learning algebraic rules. Science, 285, 1673-1674.

Shastri, L., & Chang, S. (1999). A connectionist recreation of
rule-learning in infants. Submitted manuscript: University of California,
Berkeley.

Shultz, T.R. (1999). Rule learning by habituation can be simulated in
neural networks. In M. Hahn & S.C. Stoness (Eds.), Proceedings of the
Twenty first Annual Conference of the Cognitive Science Society (pp.
665-670). Mahwah, NJ: Erlbaum.