Dear John and Ravi,
Please see below,
Sincerely,
Rich
Cooper,
Rich Cooper,
Chief Technology Officer,
MetaSemantics Corporation
MetaSemantics AT EnglishLogicKernel DOT com
( 9 4 9 ) 5 2 5-5 7 1 2
http://www.EnglishLogicKernel.com
-----Original Message-----
Subject: Re: [ontolog-forum] Fruit fly emotions mimic human emotions - ontology
discovery possible?
Rich,
Thanks for the references:
>
http://www.caltech.edu/news/do-fruit-flies-have-emotions-46769
>
> Here is a TEDx talk by the
same professor (David Anderson):
>
>
http://tedxcaltech.caltech.edu/content/david-anderson
One point that Anderson makes
confirms a claim that I have often
repeated: there is a huge
amount of diversity or heterogeneity in the way different parts of the brain
works
RC: And plasticity - extreme
damage, such as hemispherectomies, done when patients are young enough, don't
even stop the development of very normal behaviors, all handled in the
remaining half of the brain. Even older stroke patients often recover
some of the lost capabilities through relearning them.
Anderson noted that the same
chemical -- dopamine in this case -- had very different effects in different
parts of the brain. A pill that increases or decreases dopamine levels in
all parts of the brain would be a very crude tool.
RC: I wasn't planning to use it,
in that way or in any other way. I am interested in the emotions, their
interrelationships, and math models of how they work in a library of situations.
I haven't gotten into chemistry enough to play with dopamine; I don't even like
to cook.
RC
> Would it be possible to
automate an evoked response that demonstrates
> each emotional state
designated by the professor as a "component"?
> If so, would it then be
possible to write an ontology discovery
> program that explores that
space using a buncha fruit flies crossed
> with a buncha experimental
situations?
Note what he said: if you
consider dopamine level as a feature or component, it's going to have different
effects on different circuits.
Any theory based on a weighted
sum of inputs (as many neural networks and related methods do) is going to be a
very gross oversimplification.
It might give useful results for
some purposes, but not for others.
RC: I don't plan to diddle it
with dopamine. I am looking for an algorithm that could, with sizeable
numbers of fruit flies, and sizeable numbers of situations experimentally simulated
to the flies, elicit the ontology of the fruit fly's response CLASS TYPEs through
observing the behavior of the fruit flies. Those TYPEs make up the
ontology, along with a little inferential wisdom, TBD.
For anybody who wants to know
more about the brains of insects and other arthropods, I recommend the
following book:
Strausfeld,
Nicholas James (2012) Arthropod Brains: Evolution,
Functional
Elegance, and Historical Significance, Cambrdge, MA:
Harvard
University Press.
The price is $68.50 at Harvard
U. P., Amazon, or Barnes & Noble.
But it's beautifully illustrated
-- it's like a technical coffee table book.
I did a bit of googling and
found a recent discovery by Strausfeld:
>
http://www.livescience.com/23862-oldest-arthropod-brain-complex.html
One of his colleagues had found
a fossil of an arthropod that was 520 million years old. While examining
it, Strausfeld could see "the dark brown silhouette of preserved brain
nestled in the arthropod's
skull":
> This complex, insectlike
brain suggests that rather than insects
> arising from simple
branchiopods, today's arthropods descend from a
> complex-brained ancestor.
Branchiopods would later have shed some of
> this complexity, Strausfeld
said, while other crustaceans and insects
> kept it. In fact, he said,
the brain may have evolved to segment into
> three parts very early on;
mammals, including humans, have a
> forebrain, midbrain and
hindbrain, suggesting a common organization.
If the overall structure has
been preserved for over 500 million years, it must be important.
RC: Agreed. We don't yet understand the brain in
part because we don't yet understand the role that emotion plays in
individuals, or how they use emotions to create the drive to work in longer
plans, more significant accomplishments, than the day to day behaviors, without
such longer term plans.
Any claim that the brain is a
homogenous lump with all neurons working in the same ways to do the same things
must be oversimplified. There may be some similar low-level mechanisms,
but they perform different functions in different parts of the structure.
Bottom line: Don't expect
a "unified theory" based on a simple combination of features or
components.
John
RC: But do use a simple
framework of combinations of the common components to explore the emotion space.
It provides an enumerator for identifying components individually, and for grouping
them in functionally known, named clusters. That provides the brain a
source of experiences to map out the context space in which all problems spread
out. That is the idea behind algebra, after all. It should also
work for emotions. After all, they both can be very frustrating.
But there is always a reason behind emotions, even when it seems to be
misplaced.
RC: To enumerate the complexity
of each response by each fly in each situation, each response can be
analytically partitioned into which components of each emotion type are present.
Add that component's count within the partition so that a scalable sample can
be formed.
RC: That information should
help inform observers about how to classify each response, without inessential
and unnecessary anthropopathy.
-Rich