seeing, mapping, and mirror neurons

[Brian MacWhinney]

Dear Info-CHILDES,

  The discussion of sensorimotor input from Lynn, Edy, Liz, and others
has  correctly interwoven several issues.  If you go back to Piaget's
analyses in "Origins of Intelligence", you see a child who is much
involved in forming primary and secondary circular reactions between
sensory and motor schemes.  It would be easy to take this
characterization of Laurent or Jacqueline as somehow the only road that
the child could traverse.  But I doubt that Piaget saw things that way.
Rather, his emphasis was on the flexibility of mental structures.

  But can the child rely on only sensation to construct the world?  I
doubt that Piaget would have considered that adequate for the
construction of the physical universe.  Even if we cannot act on
objects ourselves directly, we can easily observe the actions of others
on objects.  Rizzolatti's work shows that support for this level of
intersubjective mapping is fundamental to cognition.  Of course, there
is work from Meltzoff, Kuhl, Anisfeld, and others pointing to the
presence of mechanisms to support imitation.

   I have been thinking that these various pieces of evidence point to
rich neuronal support for the mapping of our body image onto the body
images of others.   My initial ideas on this are at:
http://psyling.psy.cmu.edu/brian/papers/bod.htm

  How might all this impact language learning?  First, I guess most of
us can agree that the learning of verb argument structures is a core
component of the larger task of language learning.  Second, there are
many theories that would further argue that argument structure is a
partial reflex of verb semantics.  But then, how does the child learn
the semantics of verbs?  One possibility is that this occurs through
the formation of embodied representations for verbs.  The NTL group at
the ICSI at Berkeley (Feldman, Lakoff, Chang, Bailey, Narayanan) has
been developing this approach for modeling the learning of the
semantics of verbs like "push" and "shove".  But if a child is born
without limbs or without the control of limbs, how could they learn
such verbs?  The perspective-taking account would hold that the child
would match and map their own body image to that of other actors.  By
observing the pushing and shoving done by others, they would then remap
to their own potentialities.  Blind children would experience the
pushing and shoving directly and would have a more difficult time
mapping to others, but would be able to do so through imagine
projection of their body image.

  Lynn correctly notes that the Rizzolatti et al. mirror neurons do not
prove that there is a mechanism for projection of our internal body
schematic.  However, Rizzolatti's results do show that monkeys (and
presumably humans too) customarily process the actions of others
through the same neural hardware that process self action.  Rizzolatti
further shows that some of these neurons respond specifically to basic
actions with some firing for grasping and others for pinching, for
example.  It is fine to then call these "grasp" or "pinch" neurons, but
the point is that these neurons fire equally for self and other.  So
there is a neuronal juncture where the two maps coarticulate.

  There are many detailed issues in child language acquisition that
could potentially be illuminated by this approach.  Consider the
child's representation of the verb "pick up", as in "pick me up".
Rather than mapping this verb's meaning initially to the perspective of
the person who lifts up, the child may view the action from the
"ergative" perspective of the child being lifted.  Eventually, both
perspectives are merged, thereby supporting a fuller set of syntactic
options.

  I'm curious whether others have considered the implications of
perspective-mapping for language acquisition.  It would seem that Nancy
Budwig's work on the replacement of subject pronoun "I" by "my" for
actions in which the child is directly involved might be a further case
of this sort, for example.

--Brian MacWhinney