Re: OOL I - Manifesto and metatheory

> From: "Perplexed in Peoria" <jimmenegay@xxxxxxxxxxxxx>
> All that is required for kin selection to operate is that the actor
> must socially interact with relatives more frequently than he does
> with the general (breeding and competitive) population.

Yes, but that can't be obtained unless the kin-altruistic individual
can somehow either discriminate kin from enemy or happen to be in an
environment where most nearby individual are kin so helping anyone
nearby is good enough discrimination. Note that assuming anyone nearby
is kin can be swamped by alien squatters, so a strategy based on that
assumption might not be stable. I'm quite sure that freefloating
chemical replicators are unable to achieve *any* way of interreacting
only or predominately with kin.

Furthermore, whereas kin altruism in animals involves direct
interaction, kin altruism in simple chemical replicators would usually
involve interacting with some *other* chemical which indirectly
benefits the kin. So your criterion wouldn't be valid. For example, an
altruistic replicator within a micro-ecosystem might attack invading
chemicals to protect other replicators of the same micro-ecosystem, or
might produce food that other replicators eat, or might produce lipids
or DNA strands to enlarge or strengthen the cell membrane to better
protect other replicators within it. I see no way, except group
selection within a micro-ecosystem, that simple chemical replicators
would have selection pressure to develop any kind of altruism.

> > (2) Tit-for-tat game theory, whereby individuals keep track of who is
> > being nice and who is being nasty, and cooperate only with others who
> > have been nice. This doesn't work unless individuals can recognize
> > specific others and distinguish them and keep track of which have been
> > nice and which haven't.
> Which doesn't necessarily require nervous systems. Plants, for
> example, could conceivably play tit-for-tat with other nearby plants.

I agree. Given inability to move around quickly, longterm connections
between nearby plants could be established, whereby all communication
along a given connection is virtually assured of being with the same
<other> as it was all previous times, hence tit-for-tat can be played
along each such connection (communication channel). And such
tit-for-tat can be done by simple chemcal methods performed locally,
not requiring any communication with other parts of the plant.

> If the sending of trustworthy signals is classified as "cooperation",
> then there are certainly ways of having an ESS that don't fit into
> the above three.

If there is a cost imposed to the altruistic sender, and benefit only
to the passive receiver, then what other mechanism would somehow reward
the sender for making that expenditure? With group selection, the
sender prevents the whole micro-ecosystem from dying, hence achieving
benefit eventually. With kin selection, copies benefit. With
tit-for-tat, there is direct benefit during the next iteration of the
game. What other delayed-reward system do you claim could exist? Please
tell!

> I said nothing about tit-for-tat, nothing about cooperation, and, in
> fact nothing about communication at all other than the fact that
> group selection is not necessary to explain it.

And I said I know of only three ways (groupSel, kinSel, titForTat) that
allow cooperation/altruism (*) to be an ESS, and you seem to say
there's yet another, and I'd like to know what you have in mind.

* ("cooperation" in the usual English sense requires both parties to
"cooperate" in the tit-for-tat single-action sense. It's that latter
sense that I am talking about here. Why should an organism be nice by
cooperating unless there's some mechanism for rewarding it for the
expense it incurred by being nice?)

(Above composed Tuesday, below started Wednesday.)

> However, since you raise the subject, let me say that I think that
> reciprocity (of which tit-for-tat is one model) is probably of
> central importance in any OOL scenario in which there is symbiosis
> (leading to union) between two species of replicators.

Carefully distinguishing this kind of repeated-interaction
game-theoretic reciprocity from group selection of trapped-together
co-evolving replicators, I don't think the kinds of simple molecules
we're proposing as first just-barely-life, nor for quite a while down
the evolutionary road from there, can possibly engage in any form of
game-theoretic reciprocity whereby they are nice to neighbors only if
those specific neighbors have been nice to them already/currently. Even
plant-like being nice to long stable touching-partners can't work with
simple chemical replicators. They're orders of magnitude too simple to
be capable of such learned behaviour.

> 1. Weak mutualism (or ecological mutualism) is a relationship between
> species. For example, plants provide food to animals, who then return
> CO2 to the atmosphere and nitrogen to the soil, thus benefiting plants.
> Cycles of this sort are of little evolutionary significance because no
> individual plant or animal organism has much incentive to hold up his
> end of the bargain.

This isn't keeping your end of the bargain. This is dumping your waste
into the environment with total disregard to what life may be harmed or
killed by your toxic waste, and some other species learning how to
detoxify your waste and in some cases making profit from it. Most life
either is killed by even slight amounts of oxygen, hence needs to hide
from it deep under mud or rock, or needs to spend valuable energy
detoxifying it. From a cladistic viewpoint, only a small portion of
life can actually make use of oxygen. Yet the clade that does
photosynthesis with water as the hydrogen donor keeps dumping this
oxygen into the atmosphere without regard to the harm it does to the
majority of the clades.

Even the example you cite, of plants generally providing food to
animals (ignoring the specific cases where plants recruit animals to
carry seeds through their guts or on their fur), and animals&fungi
usually providing CO2 and inorganics useful to plants, is just another
example of dumping waste without regard to whether it helps or harms
others, without any sort of 'bargain' involved.

> 2. Strong mutualism (or 1-1 mutualism) involves an exchange of
> benefits between individuals (presumably of different species). A
> flowering plant provides nectar to the bee, and the bee reciprocates
> by transporting pollen. The reciprocity of the transaction is almost
> automatic - neither bee nor plant has any need to remember individuals
> of the other species.

The memory is done at the genomic level, with the disadvantage that the
learning process takes millions of years, but taking advantage of the
fact that the enemy's adaption process also takes millions of years. If
the insect species evolves a way to cheat by taking nectar without
delivering pollen, then those plants which are most vulnerable to this
form of cheating decline i numbers, while those plants that by chance
have a way to refuse giving nectar to all those insects, or even better
a way to tell the difference between cheaters and non-cheaters and give
nectar only to the non-cheaters, increase in numbers. One round of the
tit-for-tat game consists of one multi-million-year evolutionary period
during which a counter-strategy can evolve. A cheater can achieve an
advantage for only one game round, then gets punished during the next
game round, just like prisoner's dilemma game but with a much longer
time per round.

At evolutionary time scales, there is no such thing as an individual
organism, nor even individual genome, there are only clusters of
genomes that form species. Genomes really do recognize other (clusters
of) genomes and cooperate or non-cooperate with them selectively per
tit-for-tat strategy, in the sense of having genes that produce
phenotype characters which cooperate with phenotype characters of
non-cheater genomes but which refuse to cooperate with phenotype
characters of cheater genomes.

> However, if the species can remember each other, then there is the
> possibility of deferred strong mutualism (tit-for-tat) in which the
> reciprocity is not automatic.

What you wrote there was a bit vague. Did you have anything particular
in mind when you wrote it? Would you consider my previous paragraph,
about evolutionary-timescale tit-for-tat, as what could have been meant
by what you wrote just above here. Would you agree that whenever a
species "remembers" another species, it's probably either the kind of
evolutionary-timescale "learned behaviour" that I described above,
which couldn't occur during the very eary stages of OOL, or the normal
kind of learned behaviour that can occur only in a sufficiently
intelligent creature far beyond what we had during any of OOL
whatsoever nor even in most present life?

As an example of true learned behaviour, imagine a chimpanzee which
tastes various kinds of fruit, finds some are so bitter they have to be
spit out, some don't taste too bad but make the chimp sick, so the
chimp learns to avoid not just those individual plants with that fruit
but *all* other plants that look similar (most of which are of the same
species), and perhaps communicates such knowledge with other members of
his/her social group. However this is an individual, or a social group
that shares wisdom, learning about other species, not the species per
se learning about other species. By comparison, the
evolutionary-timescale "learned behaviour", due to mixing of genes
despite geographic barriers over such long timescales, truely is a
whole species learning about other species.

> Note that I said that the reciprocity is "almost" automatic.
> Cheating IS possible - a flower could provide pollen, but no nectar,
> or a mutant bee could have teflon-coated hairs on the legs that do
> not transport pollen.

Oops, I already read this before getting the inspiration to write about
evolutionary-timescale tit-for-tat "learned behaviour" above. I just
want to add here that if a species does not have the evolutionary
capability to evolve a defense to cheating, it'll be so vulnerable to
such cheating, and cheaters will be unpunished for so very many rounds
of the game, that cheating will become quite profitable, and probably
drive the victim species to extinction. Nowadays I would expect that
all species are capable of evolving defense to cheaters.

> Models of this situation are VERY complex, involving Zahavi processes
> (*) as well as the possibility of group-selective and kin-selective
> mechanisms.

I never heard of "Zahavi processes", so I tried to do a Google search,
but searching for both words produced what looked like all false
matches, unless the matches referring to "Zahavi's handicap principle"
were relevant, let me know and I'll look closer, while searching for
either "Zahavi processes" or "Zahavi process" with quotes to group the
terms produce no hits whatsoever.

I don't see how group selection can be useful for this process except
in tightly-knit social groups among intelligent animals, or in
co-evolving trapped genomic units trapped together in micro-ecosystems,
except in that case the very early stages of OOL wouldn't yet have
sufficient genomic control over phenotype to be able to develop such
complex adaptoins. I see the development of altruism within such a
micro-ecosystem, such that eventually it qualifies a fullfledged
"organism" of cooperative parts instead of just an ecosystem of
independent parts, as occurring prior to any evolution of such
organism.to recognize/distinguish friend/foe among other organisms.
So I don't see any of these cases relevant for OOL theories.

> 3. Iron-clad mutualism is strong mutualism without the possibility of
> cheating by either party. This (impossibility of cheating) might arise
> in one of two ways:
> a. There are no cheating variants because there are no variants - no
> mutations. Many models of primitive replicators (including yours
> IIRC) make heritable variation in the replicator almost impossible.

But there's no reason such a mutualism would pop out by pure chance.
Without any evolutionary mechanism to produce it, and with no variation
there's no mutation hence no evolution, QED.

> b. Cheating is physically impossible. A "fair" exchange of benefits
> must take place or there is no exchange at all. For example, the
> host interacts with symbiotic mitochondria by exchange of material
> through port-antiport mechanisms. ...

This is irrelevant to OOL because such complex biochemical pathways and
exchange mechanisms didn't evolve until a long time after OOL.

> Iron-clad mutualism between two autocatalytic cycles could take place
> if the two exchange materials at an intersection between the two
> cycles.

I see no reason to expect such a situation would arise by chance even
once on any given planet, although given a gazillion different planets
all doing OOL independently it's reasonable that one or two planets
here and there at random might chance on such a situation, and no
mechanism to evolve toward it during the non-evolving just-barely-life
stage of OOL.

> An ecological mutualism between two cycles, in which the waste
> products of one are released to the "soup", from which they are
> extracted as food by the second (which then releases wastes, etc.) -
> such ecological mutualism is IMO without evolutionary significance
> for OOL.

Except in the sense that the existance of a large bulk of "waste" from
one replicator might make it more likely for a second replicator to
form, specifically a new replicator that happenes to use that
very-non-Geilstein "waste" as part of its food, agree?

> For OOL, we need iron-clad mutualism, or else we should forego the
> symbiotic union model all together, and start with a quasi-cellular
> organism.

The only kind of mutualism I see likely in semi-early OOL is when
replicators get trapped together and undergo group selection whereby
inner-cooperating micro-ecosystems survive better than infighting
micro-ecosystems, causing the quantity of inner-cooperation to increase
over time among the surviving micro-ecosystems. I see no mutualism
greater than chance/fluke to appear prior to the appearance of
trapped-together replicators, i.e. no selection pressure toward
increased (above chance/fluke levels) mutualism until group-selection
of replicators within micro-ecystems appears, and then *only*
inner-mutualism at that time. Agree or disagree?

.

Relevant Pages