ocular dominance and input

[Gary Marcus]

Dear Brian (cc: Ping, and info-CHILDES),

In my view, you are confusing the question of activity-dependence with the
question of learning. In my discussion of Hubener and Bonhoeffer (and
implicitly Crowley & Katz, Nature Neuroscience, 2, 1125-1130),  I wrote
that the "basic structure [of some aspects of V1] seems to be available
prior to (womb-external) experience." You seem to challenge my
interpretation, by saying that it is open question whether
activity-dependence plays an important role in the formation of (the
relevant aspects of) V1.  But while I agree that  it is for now unclear
whether activity-dependence plays an important role in the formation of
(the relevant aspects of) V1, I don't think that lack of clarity has
anything to do with my question of womb-external learning. It could well
turn out that activity plays an important role; but if the
relevant  activity is internally generated (as it would have to be in a
darkened womb or in Crowley  and Katz's ferrets, whose retinas were
surgically removed), we could not attribute the relevant aspects of
development to learning.

As Katz put it in earlier paper with Shatz, in which he and Shatz discussed
internally-generated waves:

>  visual experience alone cannot account for many features of visual
> system development. In nonhuman primates, for example, ocular dominance
> columns in layer 4 begin to form in utero and are fully formed by birth.
> Thus, although visual experience can modify existing columns, initial
> formation of the stripes is independent of visual experience. Other
> features of cortical functional architecture, such as orientation tuning
> and orientation columns, are also present before any visual experience...
> [Katz, L.C., & Shatz, C.J. (1996). Synaptic activity and the construction
> of cortical circuits. Science, 274, 1133-1138].

Now, you may not care about the question of "learning vs. maturation", but
I do. For me, it is interesting to ask whether particular aspects of the
machinery that supports language and cognition arise prior to experience --
I thought that that was how we got into this discussion in the first place.
For me, then it is interesting that the ocular dominance columns can
develop in a darkened womb or in an emulated ferret; I wrote in because I
found Ping's letter (and  Nelson's review) to be unbalanced, emphasizing
the role of (womb-external) experience without making it clear that such
experience is not always required.

I find the question of whether brain development is "interactive" far less
interesting, for I honestly don't see how the answer could be no. If the
development of the stomach or liver is interactive, surely the development
of the brain must be too. Crowely and Katz's work is grist for Nelson's
mill because everything is. In mammalian development, practically
everything is interactive -- cells always communicate with one another,
with any given cell's fate in part dependent on its neighbors. Let us not
confuse interactivity with learning, though -- some interactions are a
consequence of learning, others not. A more nuanced view is the one
championed early on in  Elman, et 's recent book Rethinking Innateness
(1996, MIT Press); as they put (page 22; see also their table 1.2, page 23):

>the term innate refers to changes that arise of interactions that occur
>within the organism itself during ontogeny. That is, interactions between
>the genes and their molecular and cellular environments without recourse
>to information from outside the organism.

Crowely and Katz's work could be taken as showing that the formation of
ocular dominance columns is innate in exactly this sense.

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.  But
it is important to realize that nature's tool kit includes both mechanisms
for sculpting microcircuitry on the basis of experience, and mechanisms for
sculpting microcircuitry in the absence of experience.

Best,
Gary

p.s. The network simulations that you mention go towards showing one way in
which you could in principle use patterns of neural activity to shape the
ocular dominance columns, but they do not prove that such mechanisms are
*actually used*. As Hubener and Bonhoeffer noted "Crowely and Katz have now
cast doubt on this view" -- in this way, Crowely and Katz's work  serves as
a potent reminder that there's more than one way to equip a ferret.




----------------------------------------------------------------------------
---------------------------------

At 05:43 PM 12/2/99 -0500, Brian MacWhinney wrote:
>Dear Gary, Ping, and Info-CHILDES,
>
>   The Hubener and Bonhoeffer article from Nature Neuroscience that you cite
>is the commentary article that interprets the Crowley and Katz article that
>appears later in the issue.  The Crowley and Katz works does indeed show that
>the ocular dominance columns can form in an almost normal way without visual
>input.  Gary says that this work shows that "Learning must play some role in
>fine-tuning things, but the basic structure seems to be available prior to
>(womb-external) experience."  I would not interpret either Crowley and Katz
>or Hubener or Bonhoeffer as saying this.  This is not a debate about
>learning.  It was about whether the wiring of V1 is triggered by mere
>connectivity or by connectivity plus activation.  In both scenarios, the
>foromation of ocular dominance columns is not the responsibility of genetic
>mechanisms, but of the topography of connections between V1 and the LGN.
>
>   Hubener and Bonhoeffer state that, "Because so many experiments
>demonstrated that visual experience can influence the layout of cortical
>maps, it became widely accepted that vision might actually be instrumental in
>their establishment." (p. 1043)   What this gloss leaves out is the role of
>the interesting neural network simulations of Miller, Stryker and their
>students which show that the emergence of the ocular dominance columns can
>arise through competition between the two ocular pathways for projection to
>uncommitted neural territory in V1.
>
>   So the real question is not about learning vs maturation, but about whether
>competition is grounded on connection (Crowley and Katz) or requires
>connection and activation (Stryker & Harris and others).  It seems to me that
>the Miller and Stryker analysis goes through just fine even with the Hubener
>and Bonhoeffer result.  However, it is certainly interesting to know that
>mere connectivity alone is enough to organize ocular dominance columns.
>Eventually, this work could tell us something about the molecular basis of
>competition for connectivity.
>
>   So, like Ping, I take this Crowley and Katz work as a further elaboration
>on Chuck Nelson's point.
>
>   Now, for the real question, what does all of this have to do with language
>development?
>
>--Brian MacWhinney