Re. Bird Brains

["Adrian Thomas" <adrian.thomas@zoology.oxford.ac.uk>]

Irene Pepperberg gave a talk here in December about Alex, and I met her and
Alex (the parrot) at Tucson when I interviewed for an Asst, Prof. job there
in 96. 

The Alex work is really remarkable, but its a lot more complicated than Alex
being able to talk or birds being being exceptionally clever. Irene herself
is very careful about the way she talks about Alex's abilities - she is very
reticent about answering questions such as "Does Alex have a concept of
'number'?" or "What does Alex mean by 'red'?" (the latter being particularly
interesting given that whereas as we are trichromatic birds are probably at
least tetrachromatic (as well as equivalents of our 3 colour receptors they
have a UV receptor and complex oil-droplet filter systems in a large
percentage of their cone-cells, how these combine to give colour vision is
not yet known - but they ought to at least have better colour resolution,
and discrete concepts like 'red' don't necessarily make sense given that the
spectrum is continuous).

Irene is very clear (having done the experiments to demonstrate it) that it
is the demonstrator-model training technique she has used that has resulted
in Alex's abilities, other African Grey's she has trained using other
techniques do not even approach Alex's achievements. She is now extending
the work to other species and individuals, and is looking at African Grey
social structure in the wild.

What is clear from Pepperberg's work with Alex is that Alex does as well in
the sorts of cognitive-ability tests people use on other animals (such as
chimps) as anything else (CHimps, Dolphins etc.). What that means is less
clear. Obviously we would expect birds like parrots, crows and pigeons that
live in complex social groups, and have long lifespans, to benefit from high
intelligence. However birds seem to do things rather differently from us:
for example birds (pigeons, crows, chickens) are much better at tasks
involving image recognition than we are, it turns out that they do this by
using a hard-wired technique of snap-shot recognition. What they do is make
a neural record of a particular image, then when asked to pick it out they
move their heads until the image currently on the retina coincides with the
memorised image. That is why biirds bob their heads, and peer at things
through one eye with their head held still at an odd angle. In contrast our
approach seems to be to make a detailed 3D model of the world and try to
match the retinal image to the model. Theirs is a simple rule-based
technique, ours relies more on brute-force processing. 

Different strokes.

The snap-shot image recognition stuff is by Marian Dawkins, who also wrote
the best book on conciousness and cognition in animals 'Through our eyes
only?'

Cheers,
Adrian