Greg Detre
Sunday, February 09, 2003
It
initially seems so intuitive that being brought up in a civilised, stimulating,
human setting would be as beneficial for apes� cognitive development as for humans,
but our intuition furnishes few details about which areas of cognitive
development would most benefit, or why. We might consider that simply having a
more advanced teacher would result in faster and more complex learning, just as
neural networks trained by an advanced player learn much more rapidly than
those trained against themselves. Similarly, humans may simply expose them to a
wider range of objects from which to generalise, and new ways of manipulating
them that are easier to be shown than to discover.
Unfortunately,
as Call and Tomasello point out, most evidence of advanced skills being
demonstrated in the wild is anecdotal, and debates about the extent to which
teaching, motor imitation, deception and the like have been observed still
rage. My feeling is that since such the evidence of such behaviour tend to be
extremely rare, and always contentious, it�s more likely to be accidental or at
the level of the physical stance, especially for high-level intentional
behaviour like deception.
The main conclusion
of the paper is that Boesch�s suggestion that apes display their most advanced
cognitive skills in the wild is not supported by the evidence. However, we
shouldn�t forget that Savage-Rumbaugh�s success with Kanzi may partly result
from her use of 55 acres of natural forest, and activities and topics of
discussions that are intended to mirror those of an animal in the wild.
Gopnik and
Meltzoff are interested in what the precursors and prerequisites to a
fully-fledged theory of mind might be, and how they might lay the foundations
for perceiving others as being and operating in fundamentally the same way as
ourselves.
They argue
that simply having an innate special attentiveness to the human facial pattern
is insufficient for building the link between self and other since infants
rarely see their own faces to see the parallel. Similarly, the adult-child
�conversational dances� are not exclusive to interactions with people (after
all, even a toy swinging above the cot takes turns to move after the infant has
touched it), nor is it obvious how they might form a basis from which a theory
of mind might develop.
Their
theory is interesting. They argue that the foundation for our theory of mind is
the primary perceptual judgement �here is something like me�. Going in the
opposite direction to most theories, they propose that babies are in some way
pre-wired to notice structural and temporal parallels between their own and
others� actions. Simply put, babies can imitate and notice imitation
(�cross-modal equivalences�) of themselves much earlier than previously
thought. This important observation underpins preferential attention towards
other humans, rather than the other way around.
Then they
take into account the fact the findings (e.g. Eklund et al.) that simply making
the facial gestures that correspond to certain emotions can trigger our feeling
those emotions. Babies that pay special attention to faces and have some sort
of innate ability to detect and reproduce the behaviour of adults are
effectively mirroring the emotional state of the adults they are imitating.
That this could develop over the long-term into empathy is a powerful idea.
Personally,
it seems to me that the least plausible but critical step in their hypothesis
is that infants have an innate means of mapping motor behaviour from others to
themselves, especially facial movements. Motor imitation is defined as
learning the operant task directly through the observation of the model�s
behaviour (Heyes 1994, Zentall 1996). It is usually considered the cognitively
most demanding category of social learning (for the observer), since it
requires the translation of a visual input into a matching motor output which
may involve more complex central processing than other mechanisms of social
learning (Whiten & Byrne 1988, Heyes 1988). As they point out, this perception of
cross-modal equivalences certainly cannot be learned by infants as young as
they are suggesting, so must be innate in some way. There is no a priori reason
why we can�t have an innate means of mapping from a visual impression of
others� facial movements to a motor command to make the same facial gestures.
However, it does seem like a more specific and complex mechanism than we
usually expect to find hard-wired, and it�s very difficult to imagine with
current knowledge how it might be represented genetically or develop neurally
within high-level architectural constraints. But these are areas we know little
about, so perhaps our intuitions shouldn�t be trusted here.
I also want to mention at least one alternative to their hypothesis that ought to be borne in mind. When considering the mirror test as an indicator of self-recognition in primates, Povinelli points out that even success may not indicate genuine psychological self-recognition, but rather an extended motor self-concept. That is, they do not really recognise themselves but simply learn an equivalence between their behaviour and what they see in the mirror. It seems to me that when the infants are noticing adults imitating them (either structural or temporal), they might actually be believing that they have control over the adults� bodies in the same mysterious way they can control their limbs. That is, they might be including the adults� bodies as part of their own, like a more distant arm. This isn�t wildly different from what Gopnik and Meltzoff are suggesting, but there is something less interesting and persuasive if their claim was to be that infants� theory of mind is prompted by the realisation that they can have an (indirect) causal effect on imitators� movements, as opposed to the stronger realisation that other people are other people like them.
Finally, I don�t want to say much about their thoughts on autistic children. As they acknowledge, the scant evidence so far seems to support their claims, and the contrast with Down�s syndrome is worthy of note, but a longitudinal study that included young autistic children would say much more about how imitation and theory of mind might be causally related.
They could perhaps also point to the lack of strong evidence for motor imitation in animals (e.g. primates: Byrne 1995, Whiten et al 1996. Also birds, e.g. ravens: Fritz & Kotrschal) as suggesting why animals fail to develop a real theory of mind.