Self Interested Ontologists,
Re Doug’s categories of interest and
rewards, as applied to the scenario I sketched in John’s bacterial world
of Use Case 1, I think the following cumbersome binding expressions work as a
very first approximation, at least for discussion purposes. Someone who
knows CycL might be able to formulate these better:
Agents: Strepta, Chlamy, Bacilla, Amoebi;
Actions: (sends-message Strepta),
(receives-message
Chlamy),
(splits
Chlamy
(list Bacilla Amoebi)),
(flails-cilia
Bacilla),
(pushes-with-gravity
Bacilla),
(moves-further-into-cavity
Amoebi),
(pops
Bacilla),
(is-coated-protectively
Amoebi)
Beneficiaries:
(highest-benes
Amoebi),
(middle-benes
Chlamy),
(middle-benes
Bacilla),
(lowest-benes
Strepta)
Malficiaries:
(highest-mals
Strepta),
(middle-mals
Chlamy),
(middle-mals
Bacilla)
Clearly Amoebi is the most benefacted by
the actions of all four bacteria, but since the four share most of their genes
anyway, all four receive benes in some amount, if you consider that the
interest of all is to propagate their genes. But Amoebi’s genes
aren’t actually propagated by the actions in Use Case 1. Her
benefit and interest is entirely in surviving the noxious toothpaste, perhaps
to propagate at some future date, but perhaps not. That is,
Interests:
(propagate-genes
Strepta),
(propagate-genes
Chlamy),
(propagate-genes
Bacilla),
(survive-event
Amoebi)
By rewriting these approximations into
some more CycL format, perhaps we can specify the Situation-Action-Reward
spectrum for all the players in the Use Case 1 so that Doug’s ontology of
self interest can be more aligned with that use case.
Comments, suggestions, critiques, curses?
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
From: Rich Cooper
[mailto:rich@xxxxxxxxxxxxxxxxxxxxxx]
Sent: Friday, August 12, 2011 6:31
PM
To: '[ontolog-forum]
'
Subject: Self Interest Ontology -
Bacteria Use Case 1
Self Interested Ontologists,
Let’s consider a use case for the
bacterial film example. I will propose one, and if there are comments,
please feel free to add your $0.02 or to correct mine as appropriate.
USE CASE 1.
A bacterial film covers teeth. One
bacterium, Strepta, senses a chemical gradient she associates with problems to
come. So Strepta sends a chemical message M to the film at large.
Further away in the film, Chlamy
identifies the message, which she interprets as “Watch it; there is
dangerous antibacterial toothpaste in the vicinity.”
Chlamy senses the direction of the message,
and quickly forms a waist in the plane of the message direction, then splits at
right angles in the cross product, splitting into two daughter cells Bacilla
and Amoebi, while Chlamy ceases to exist as a unit, having split.
Bacilla is on top, and her weight plus her
flailing cilia push Amoebi one micron down into the cavity. When the
noxious chemical (was it really toothpaste?) touches Bacilla, she pops,
spreading proteins, fats and carbohydrates which coat Amoebi in her cavity,
effectively protecting Amoebi from the noxious chemical gradient.
Strepta may share very few genes (self
interest objects) with Chlamy, but Chlamy’s offspring have very accurate
copies of Bacilla’s genes, so both have mutually high self
interest. By splitting, Chlamy preserves her genes. Bacilla
preserves her genes, which are faithful copies of Chlamy’s, by protecting
her twin sister Amoebi with her (Bacilla’s) own life.
What is in it for Strepta? She may
be millions of generations distal from Amoebi, the ultimate beneficiary of
Strepta’s message. So the gradient of the film should somehow
represent the contribution of Strepta’s gene pool to Amoebi, which is
what gives Strepta (through their shared ancestral forebears) genetic reasons
to send her message to distal parts of the film.
Should there be a limit to the distality
with which Strepta uses her cellular resources to send the message far and
wide? It should be related to the likelihood of Strepta’s genes
being preserved as compared to the likelihood that her resources consumed to
send the message are wasted, if the chemical turns out not to be toothpaste but
simple custard.
Wouldn’t statistical decision theory
hold in this use case 1; wouldn’t the likelihood of resource loss be
approximately equal to the likelihood that Amoebi, with genes that are equal to
Strepta’s, is saved from annihilation to continue propagating those genes
into a future film? Wouldn’t the sensors and effectors used by
Strepta evolve to be quantitatively related to the survival benefits enjoyed by
her sacrificed materials?
Comments welcome,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2