Friam

A question for the emergentists among you

Classic

List

Threaded

50 messages Options

123

glen e. p. ropella-2

Re: A question for the emergentists among you

Thus spake Owen Densmore circa 10/10/2009 11:47 AM:
> To FRIAM: how would you answer this question by Dennett: "Are centers of
> gravity in your ontology?" .. i.e. are they "real", do they "exist"?

My answer is: "Yes, centers of gravity are real." But I qualify it with
"as real as anything else we _use_ as the basis for action." Everything
we _do_ is real and any thing that effects and affects that _doing_ is real.

So, because we use centers of gravity to, say, build bridges, centers of
gravity are real.

However, because emergent properties are totally useless (except as
sorcerous rhetorical babble), they are not real.

When/if someone answers Robert's question, i.e. shows us a practical
_use_ for the label, then it still won't be real; but it'll be much
closer. You actually -- act-ually, same root word as "active" and
"actor" -- you actually have to use some thing for that thing to be
actual/real. A merely hypothetical claim that some thing _could_ be
used is inadequate. Centers of gravity are actually used; they effect
and affect actions (act-ions). To be clear about my stance, nothing
just is. Reality (if we have to use the concept) consists entirely of
actions, processes, verbs. There need be no nouns. Hence, unless a
hypothetical noun participates directly in a verb, we're free to ignore
it because it doesn't matter. It is inactive and, hence, unreal.

Centers of gravity are useful and used. Hence, they exist.

--
glen e. p. ropella, 971-222-9095, http://agent-based-modeling.com

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Gary Schiltz-4

Re: A question for the emergentists among you

In reply to this post by Russ Abbott

Of course one of the many problems with (and perhaps benefits of) human languages is that they are incredibly imprecise and flexible. Obviously Russ A has at least a slightly different definition of science than that of Robert. We could debate the merits of each definition in our own particular world views, but that would take us down the same rat hole that we've witnessed lately with regards to emergence. According to my trusty OS X dictionary, science is "the intellectual and practical activity encompassing the systematic study of the structure and behavior of the physical and natural world through observation and experiment" (pretty much matches what I've usually seen as its definition). I've always considered science to be a way of approaching problem solving, i.e. "the scientific method" encompassed by the above definition. The method can be used regardless of one's motive of using it, e.g. greater understanding of some phenomenon in order to solve some "practical" problem, or greater understanding for the simple motivation of wanting to increase the extent of "human understanding."

;; Gary

On Oct 11, 2009, at 12:09 PM, Russ Abbott wrote:

By definition science isn't applied. Whether or not new scientific results have application is a different question. [...]

-- Russ A

On Sun, Oct 11, 2009 at 7:19 AM, Robert Holmes <[hidden email]> wrote:
Merely an expression of a personal preference: if "there is no point" is true, it tells me that emergence is and can only ever be pure science. As a practitioner, I prefer my science applied -- R

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Roger Critchlow-2

Re: A question for the emergentists among you

In reply to this post by Robert Holmes

From my perspective, which is probably a minority, your question makes very little sense.

The basic conditions for "emergence" were laid down by Mill in 1843, http://www.gutenberg.org/files/27942/27942-h/27942-h.html#toc53, and there's not much to it: when you combine some things, the properties of the whole are an obvious combination of the properties of the parts; when you combine other things, the contrary. Mill didn't name it as emergence, that came later. He wasn't the first to identify the conditions, either, but that's where our seminar studies started.

All of the authors we've been reading agree to Mill's definition of emergence. They all recognize the appropriateness of the label. They all recognize the same category of phenomena as deserving the label. In short, every schoolboy knows emergence when he sees it.

So, your question places in the hypothetical future something which factually happened at least 166 years ago.

What the authors disagree about is the significance of the category. Some want it to be simply an aspect of our ignorance, remedied by progress. Some want it to be the transcendence of material causation, amen. Some want it to be the nature of reality, russian dolls of causation nested inside other russian dolls.

So, your question doesn't even acknowledge the issues that are under debate.

My discussion of dog packs was supposed to suggest that the recognition of the category is actually prehistoric. Language is filled with words for collections of X some of which aren't obvious combinations of X's, and the words often have associated verbs and adjectives that cannot be applied to the individuals. A single cow cannot stampede. Successfully hunting large grazing mammals with hand tools required understanding of individual animal behavior and of herd behavior, and it required the ability to act upon the appropriate theory, individual or herd, at the appropriate time.

Was there a survival advantage to learning the lesson that one should look for exceptions to the rule that more is simply more?

-- rec --

On Sat, Oct 10, 2009 at 8:21 PM, Robert Holmes <[hidden email]> wrote:

Wow, I post a question, go on a 6-hour hike and this is what I come back to...

I still don't feel that I've got a straight answer to my question, other than Doug's (which I suspect is the most accurate) and Russ's (which I really hope isn't true). So let me try again: once I've established that a phenomenon is emergent by using a yet-to-be developed metric (Owen's formalism) or philosophic enquiry (Nick's & other's approach) - then what?

In fact, let's not limit ourselves to the present situation (because I suspect that the current answer is simply "Nothing. Identifying emergence is an end in it's own right"). What would you like to be able to do once you'd attached the "emergent" label to a phenomenon? What's your best case, your grand vision? Imagine the best of all possible worlds and tell me: what would you want to be able to do once that "emergent" label gets attached?

-- Robert

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Russ Abbott

Re: A question for the emergentists among you

Roger, Well said.

But there is a further question. Can anything be added to your (Mill's) statement that when you combine some things (e.g., combining a bunch of cows into a herd) the result has properties that the components lack. That is, what, if anything, can one say about those phenomena that exhibit this property? Do those phenomena have anything in common?

My claim is that asking (and answering) that question is doing science -- according to Gary's definition of science as an activity. To the extent that that activity yields statements that would qualify as scientific knowledge, one then has advanced the state of our knowledge about nature, which is what science is intended to do.

So that's what I would argue one should be doing when one is asking about emergence: looking for and attempting to abstract and characterize the common elements (if any) in all emergent phenomena.

-- Russ A

On Sun, Oct 11, 2009 at 12:58 PM, Roger Critchlow <[hidden email]> wrote:

From my perspective, which is probably a minority, your question makes very little sense.

The basic conditions for "emergence" were laid down by Mill in 1843, http://www.gutenberg.org/files/27942/27942-h/27942-h.html#toc53, and there's not much to it: when you combine some things, the properties of the whole are an obvious combination of the properties of the parts; when you combine other things, the contrary. Mill didn't name it as emergence, that came later. He wasn't the first to identify the conditions, either, but that's where our seminar studies started.

All of the authors we've been reading agree to Mill's definition of emergence. They all recognize the appropriateness of the label. They all recognize the same category of phenomena as deserving the label. In short, every schoolboy knows emergence when he sees it.

So, your question places in the hypothetical future something which factually happened at least 166 years ago.

What the authors disagree about is the significance of the category. Some want it to be simply an aspect of our ignorance, remedied by progress. Some want it to be the transcendence of material causation, amen. Some want it to be the nature of reality, russian dolls of causation nested inside other russian dolls.

So, your question doesn't even acknowledge the issues that are under debate.

My discussion of dog packs was supposed to suggest that the recognition of the category is actually prehistoric. Language is filled with words for collections of X some of which aren't obvious combinations of X's, and the words often have associated verbs and adjectives that cannot be applied to the individuals. A single cow cannot stampede. Successfully hunting large grazing mammals with hand tools required understanding of individual animal behavior and of herd behavior, and it required the ability to act upon the appropriate theory, individual or herd, at the appropriate time.

Was there a survival advantage to learning the lesson that one should look for exceptions to the rule that more is simply more?

-- rec --

On Sat, Oct 10, 2009 at 8:21 PM, Robert Holmes <[hidden email]> wrote:

Wow, I post a question, go on a 6-hour hike and this is what I come back to...

I still don't feel that I've got a straight answer to my question, other than Doug's (which I suspect is the most accurate) and Russ's (which I really hope isn't true). So let me try again: once I've established that a phenomenon is emergent by using a yet-to-be developed metric (Owen's formalism) or philosophic enquiry (Nick's & other's approach) - then what?

In fact, let's not limit ourselves to the present situation (because I suspect that the current answer is simply "Nothing. Identifying emergence is an end in it's own right"). What would you like to be able to do once you'd attached the "emergent" label to a phenomenon? What's your best case, your grand vision? Imagine the best of all possible worlds and tell me: what would you want to be able to do once that "emergent" label gets attached?

-- Robert

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Owen Densmore

Re: A question for the emergentists among you

Administrator

I find it odd that we're arguing about the value of creating a theory
for emergence. Follow me back just a few years.

<irony>
Lets see: why would we want a theory about Chaos. Its just when
things are messy, right? Poor Lorenz and his weather equations .. if
only he had be better with error calculations he would have certainly
gotten good results. Tisk tisk.

And the poor iterated logistics equation. Why on earth didn't they
try harder, clearly the series would eventually repeat. Sloppy,
sloppy methodology. And so what if it is weird, just forget it and
approximate.

I mean, what the heck does the sill word chaos mean anyway? .. its
just when things go wrong and the silly scientist or engineer is not
doing their calculations correctly, right?

No use in any of this. What good would it be for silly old Lyapunov
to succeed in defining chaos. His exponent certainly doesn't help my
work! And Feigenbaum! A brilliant guy, but moody over bifurcation
and whether or not it has hidden structure. What a waste!

You silly folks trying to formalize chaos and find good techniques for
non-linear dynamics are just wasting my time. It'll never work. Just
stick with the old proven ways and just perturb them a bit or be happy
with the first order approximation.
</irony>

I'm ashamed of you! Surely you are not against developing new
theories, right?

Maybe this makes it clearer:
Divergence is to Chaos as Emergence is to Complexity
.. thus studying emergence may do for Complexity what hard work done
by the Heros of Chaos did for their "silly science". What the heck is
Chaos anyway? Right.

Bad Friam! Bad Friam!

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Roger Critchlow-2

Re: A question for the emergentists among you

In reply to this post by Russ Abbott

On Sun, Oct 11, 2009 at 2:10 PM, Russ Abbott <[hidden email]> wrote:

Roger, Well said.

But there is a further question. Can anything be added to your (Mill's) statement that when you combine some things (e.g., combining a bunch of cows into a herd) the result has properties that the components lack. That is, what, if anything, can one say about those phenomena that exhibit this property? Do those phenomena have anything in common?

Wimsatt lists four heuristics for establishing "aggregativity" of properties: swap "identical" parts in the aggregate; increase or decrease the number of parts in the aggregate; take the aggregate apart and reassemble it; and freedom from non-linear interactions between parts. The heuristics aren't necessarily independent of each other, but neither are they necessarily dependent. So, there are four kinds of emergence which fail just one heuristic, six kinds which fail two different heuristics, four kinds which fail three different heuristic, and one kind which fails all four heuristics. So that's 15 different flavors of emergence, which is perhaps an overestimate, but Wimsatt is still soliciting for additional heuristics.

-- rec --

Russ Abbott

Re: A question for the emergentists among you

With aggregativity defined that way, Wimsatt notes that "Very few system properties are aggregative." Then what? Is the point that "emergence, defined as failure of aggregativity" has now been fully characterized? Problem solved? I wouldn't agree with that. I think there is more to say than just a negative definition.

An interesting example to which this approach might be applied is an ideal gas. Such a gas satisfies all the aggregativity conditions. Yet it has properties (the gas laws) that the individual components lack.

-- Russ A

On Sun, Oct 11, 2009 at 3:01 PM, Roger Critchlow <[hidden email]> wrote:

On Sun, Oct 11, 2009 at 2:10 PM, Russ Abbott <[hidden email]> wrote:
Roger, Well said.

But there is a further question. Can anything be added to your (Mill's) statement that when you combine some things (e.g., combining a bunch of cows into a herd) the result has properties that the components lack. That is, what, if anything, can one say about those phenomena that exhibit this property? Do those phenomena have anything in common?

Wimsatt lists four heuristics for establishing "aggregativity" of properties: swap "identical" parts in the aggregate; increase or decrease the number of parts in the aggregate; take the aggregate apart and reassemble it; and freedom from non-linear interactions between parts. The heuristics aren't necessarily independent of each other, but neither are they necessarily dependent. So, there are four kinds of emergence which fail just one heuristic, six kinds which fail two different heuristics, four kinds which fail three different heuristic, and one kind which fails all four heuristics. So that's 15 different flavors of emergence, which is perhaps an overestimate, but Wimsatt is still soliciting for additional heuristics.

-- rec --

Roger Critchlow-2

Re: A question for the emergentists among you

On Sun, Oct 11, 2009 at 4:44 PM, Russ Abbott <[hidden email]> wrote:

With aggregativity defined that way, Wimsatt notes that "Very few system properties are aggregative." Then what? Is the point that "emergence, defined as failure of aggregativity" has now been fully characterized? Problem solved? I wouldn't agree with that. I think there is more to say than just a negative definition.

Very few system properties are aggregative, almost all system properties are emergent.

There are a lot of varieties of emergence to be dealt with. Maybe all the short, catchy slogans have been already been taken?

An interesting example to which this approach might be applied is an ideal gas. Such a gas satisfies all the aggregativity conditions. Yet it has properties (the gas laws) that the individual components lack.

Ah, let me count the ways: a simple hard sphere gas, as in Helium or Neon, which adds finite volume and van der Waals forces to the ideal gas; the diatomic hard sphere gas, as in Hydrogen or Oxygen, which adds rotational angular momemtum and vibrational energy; the asymmetric diatomic gas, such as Carbon Monoxide, where the center of gravity is off center; the polar asymmetric diatomic gas, such as Hydrogen Fluoride, which has a positive and negative ends; water vapor, which forms hydrogen bonded clusters; ionic gases; molecules with internal rotational degrees of freedom; oxygen and ozone in equilibrium with ultra-violet radiation; smog; vog; weather; solar wind; ....

These are all sorts of non-ideal gases; all, more or less, non-aggregate in their properties; all, more or less, introducing new ways to be emergent; all, more or less, reducible to aggregate properties if you add enough information about the particle structure and theory about inter-particle interactions.

This helps?

-- rec --

Roger Critchlow-2

Re: A question for the emergentists among you

On Sun, Oct 11, 2009 at 4:44 PM, Russ Abbott <[hidden email]> wrote:

An interesting example to which this approach might be applied is an ideal gas. Such a gas satisfies all the aggregativity conditions. Yet it has properties (the gas laws) that the individual components lack.

I read this better the second time through.

The gas laws are pretty well explained by the kinetic theory - that the gas is composed of atoms which have mass and velocity and the atom kinetic energies follow Boltzmann's distribution.

I suppose that one might call the Boltzmann distribution an emergent, but once one has any collection of individuals which have individual properties, one gets a distribution that describes the property in the collection, so it's a pretty low surprise emergent.

Now, there was an interesting paper in arxiv.org about systematic coarse graining of molecular dynamics simulations to compute non-equilibrium thermodynamic properties, http://arxiv.org/abs/0910.1467, which had some bearing on this,

-- rec --

Russ Abbott

Re: A question for the emergentists among you

Roger, I've lost track of what your point is.

I said that the attempt to find the appropriate abstractions to characterize emergence is valid science. Are you agreeing? Disagreeing? Neither? Both?

And what does Winsatt have to do with it? Are you saying that his aggregativity has captured the essence of emergence -- and that there is no more science left to do? That it hasn't captured the essence of emergence? (But then why did you mention it in the first place?)

So where are we with respect to whether or not it is worthwhile attempting to understand/chacterize emergence--your original question.

-- Russ A

On Sun, Oct 11, 2009 at 5:04 PM, Roger Critchlow <[hidden email]> wrote:

On Sun, Oct 11, 2009 at 4:44 PM, Russ Abbott <[hidden email]> wrote:

An interesting example to which this approach might be applied is an ideal gas. Such a gas satisfies all the aggregativity conditions. Yet it has properties (the gas laws) that the individual components lack.

I read this better the second time through.

The gas laws are pretty well explained by the kinetic theory - that the gas is composed of atoms which have mass and velocity and the atom kinetic energies follow Boltzmann's distribution.

I suppose that one might call the Boltzmann distribution an emergent, but once one has any collection of individuals which have individual properties, one gets a distribution that describes the property in the collection, so it's a pretty low surprise emergent.

Now, there was an interesting paper in arxiv.org about systematic coarse graining of molecular dynamics simulations to compute non-equilibrium thermodynamic properties, http://arxiv.org/abs/0910.1467, which had some bearing on this,

-- rec --

Eric Charles

Re: A question for the emergentists among you

In reply to this post by Robert Holmes

Robert,
(Building a bit off of Roger and Owen...) Not to be trite, but the answer is obviously that different people have different reasons for wanting to discuss "emergence". Some of the reasons would match your criterion for usefulness, others wouldn't. One reason for doing this, that receives right criticism on this list, is a sort of pure nominalism - "we just want to name things so we can pretend we understand them". Only a slight step away from this is a desire to define and name the thing and then stop (with no pretense of understanding). I don't think anyone on this list is doing either of those things, but there seems to be a lot of suspicion that some (or all) might be. Other goals may be deemed more laudatory depending on your disposition.

That said, I suspect few have a goal as concrete in its usefulness as what you are looking for. I suspect that most people's goals can be divided into two kinds:

1) Those who wish to define "emergence" because they suspect we will be able to determine which of those things we care about are emergent and which are not. These people presume that a good definition will allow us to continue as usual with most things we care about, while identifying better ways to analyze and discuss those few things that are emergent. That is, we will know from the start that certain ways of treating those things (mathematically, scientifically, metaphorically, etc.) will be insufficient, and we might even be able to identify ways of treating those things that are acceptable for all emergent phenomenon. Think of this maybe like the legal distinction between a tort and or a crime... it's nice to know which you are accusing something of, because the best way to proceed differs by type. This is very useful to know, even if you don't have in mind yet any particular things you are accusing someone of.

2) Those who wish to define "emergence" because they suspect we will find that everything we care about is emergent. That is, there are people who suspect that our ways of analyzing most everything is too simplistic and based on false assumptions, and that we globally need more sophisticated ways to discuss and analyze topics of interest. The situations I am most familiar with that are in dire need of revision are simplistic notions about perception, cognition, and development. The ideas that you can reasonably talk about "genes for" single phenotypes, that perception can be reduced to sensation, or that cognition can be separated from physiological processes on the one hand and social processes on the other, are all surely silly. I can't think of a good metaphor for this one, sorry.

Certainly there are people on the list with other goals (at the least, more specific goals), but hopefully will be a satisfactory answer to your question nonetheless.

Eric

On Sat, Oct 10, 2009 09:58 AM, Robert Holmes <[hidden email]> wrote:

What's the point of determining whether a phenomenon is emergent or not? What useful stuff can I actually do with that knowledge?

In other areas of my life, classification can have actionable consequences. For example, I can use the sophisticated pattern-matching algorithms and heuristics embedded in my brain to work out that the three animals wandering through my house can be categorized as "cats" and not "dogs". And that is useful, because it tells me that I should buy cat food and not dog food when I go to PetCo.

So what is an equivalent example with emergence? Once I've attached the "emergent" label to a phenomenon, then what?

-- Robert
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Nick Thompson

Re: A question for the emergentists among you

In reply to this post by Robert Holmes

Dear all,

I am clearly being shunned. I keep trying to answer robert and nobody pays the slight attention to my attempts at answering. Next I will find my porch light shot out. After that my barn will be burned.

Answer #1

let me take a quick whack at this. Before the recent "epigenic" revolution we focussed only on which genes we had, not on the arrangement of timing of events during development. It's example, I think, of the heurism of the emergentist viewpoint.

Answer #2

Following wimsatt, the puffiness of pancakes is emergent because it depends on the order of mixing the ingredients. You mix the dry ingredients together, you mix the wet incredients together and THEN you mix the wet with the dry. Similarly, with a bread maker you dont want to mix the yeast with the salt, or with the water, in the first instance.

If you are making pancakes from a recipe, because text is linear, the steps always appear in an order. For instance, most start with the flour and then add the baking powder and the salt, then the sugar, etc.. For pancakes, the order of these steps does NOT make a difference. Similarly, there are spacial instructions that dont make a difference: "In a separate bowl...". it instructs you to mix the wet ingredients, but you can put the sugar and the salt in witht the eggs and the milk and the oil, if you like. Try to add the baking powder to the wet ingredients or to add it AFTER you have mixed the wet and the dry, and you have trouble.

It is these sorts of facts that make the puffiness of pancakes an emergent property, and knowing on which sorts of temporal and spatial arrangements the emergent properties of a meal depends is what makes a flexible and skillful cook.

Answer #3

Like any definition, a definition of emergence will serve us if it calls attention to crucial aspects of the systems we are curious about. So, for instance, let us say that we are curious about the whole properties of a system and we adopt Wimsatt's definition that a syetem is emergent if its properties are dependent upon the arrangement of its parts, either in time or in space. We are led to consider the robustness of the system properties against variations in the arrangements of its parts. Some system properties are senstiive to the arrangements of some part entities, not sensitive to others. Against some small changes in part arrangements, the system is strongly buffered; against others, it is highly senstiive.

Further, we are led to inquire how the parts come to be arranged in a manner that facilitates emergence. Notice that arrangements of anythng are already at least nominally emergent properties. Organismic development is just a cascade of emergents, each emergent becoming part of an arrangement for the purposes of facilitating the next emergent in the cascade.

I dont think you get to any of these questions by thinking reductionistically, and unless the questions are asked, you don't get the answers.

Nick

Nicholas S. Thompson

Emeritus Professor of Psychology and Ethology,

Clark University ([hidden email])

http://home.earthlink.net/~nickthompson/naturaldesigns/

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: 10/10/2009 12:28:36 PM

Subject: Re: [FRIAM] A question for the emergentists among you

All,

Following wimsatt, the puffiness of pancakes is emergent because it depends on the order of mixing the ingredients. You mix the dry ingredients together, you mix the set incredients together and THEN you mix the wet with the dry. Similarly, with a bread maker you dont want to mix the yeast with the salt, or with the water, in the first instance.

If you are making pancakes from a recipe, because text is linear, the steps always appear in an order. For instance, most start with the flour and then add the baking powder and the salt, then the sugar, etc.. For pancakes, the order of these steps does NOT make a difference. Similarly, there are spacial instructions that dont make a difference: "In a separate bowl...". it instructs you to mix the wet ingredients, but you can put the sugar and the salt in witht the eggs and the milk and the oil, if you like. Try to add the baking powder to the wet ingredients or to add it AFTER you have mixed the wet and the dry, and you have trouble.

It is these sorts of facts that make the puffiness of pancakes an emergent property, and knowing on which sorts of temporal and spatial arrangements the emergent properties of a meal depends is what makes a flexible and skillful cook.

Nick

Nicholas S. Thompson

Emeritus Professor of Psychology and Ethology,

Clark University ([hidden email])

http://home.earthlink.net/~nickthompson/naturaldesigns/

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: 10/10/2009 12:06:14 PM

Subject: Re: [FRIAM] A question for the emergentists among you

Robert's original question was "What's the point of determining whether a phenomenon is emergent or not?" I don't think there is a point. That's not the issue. The point of the discussion is that some properties seem to exist at a macro-level (every time I use that word now, I worry that Glen will attack me for it) but not at a micro level. If that is a frequently occuring phenomenon, it makes sense to ask whether there is something common to all instances of such phenomena. I think that's the point of the discussion. It's really a matter of scinece: here are a number of somewhat diverse phenomena that seem to have something in common. Can we come up with a characterization of what it is -- and if so does that offer any insight into how the world works?

-- Russ A

On Sat, Oct 10, 2009 at 10:26 AM, Owen Densmore <[hidden email]> wrote:

On Oct 10, 2009, at 7:58 AM, Robert Holmes wrote:

What's the point of determining whether a phenomenon is emergent or not? What useful stuff can I actually do with that knowledge?

In other areas of my life, classification can have actionable consequences. For example, I can use the sophisticated pattern-matching algorithms and heuristics embedded in my brain to work out that the three animals wandering through my house can be categorized as "cats" and not "dogs". And that is useful, because it tells me that I should buy cat food and not dog food when I go to PetCo.

So what is an equivalent example with emergence? Once I've attached the "emergent" label to a phenomenon, then what?

-- Robert

My interest is pretty theoretical. I'd like to reduce it to some sort of formal setting, like computer science does with its three classes of computing devices (FSA, Pushdown Automata, TM), then see if I could discover simple properties of "complex" systems, emergence among them.

As an example: Emergence could be a computational complexity class .. one that has has no "short cut" towards "solving" it. Game of Life is often used as such an environment. It has several trivial initial conditions that are pre-computable .. i.e. you can analyze the system and predict the result before running it. But this is not true in general. Finding the conditions separating the two would be useful.

A similar thing happened to me at Sun: we were trying to build an event distribution scheme for an early window system that would work well in a multi-tasking environment (unix). It was really slow. One of our team spent time resolved that its computational class was non-polynomial. We started over.

I hate to say it but as much as I despise the flower child philosophic, I've gotten some interesting ideas out of the book. The difficulty is the signal to noise ratio is pretty poor.

-- Owen

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Roger Critchlow-2

Re: A question for the emergentists among you

In reply to this post by Russ Abbott

On Sun, Oct 11, 2009 at 7:11 PM, Russ Abbott <[hidden email]> wrote:

Roger, I've lost track of what your point is.

My point was that Mill spent a few pages defining what became know as emergence, and that everyone since has known exactly what he was talking about.

Your question was: what can you say beyond what Mill said?

My answer is that you can say what Wimsatt says: here are four heuristics that should allow you to identify emergents by their failure to be simple aggregates.

I said that the attempt to find the appropriate abstractions to characterize emergence is valid science. Are you agreeing? Disagreeing? Neither? Both?

Sure, characterization of emergence is science, whether emergence is epistemological, ontological, or some unholy amalgam of the two.

And what does Winsatt have to do with it? Are you saying that his aggregativity has captured the essence of emergence -- and that there is no more science left to do? That it hasn't captured the essence of emergence? (But then why did you mention it in the first place?)

I'm saying that I accept his distinction between aggregate and emergent properties and the heuristics he proposes to implement the distinction. The essence of emergence is that the whole can be more than the sum of the parts.

Beyond that I believe that science is a brute force enumeration of the ways that the whole can be the sum of the parts, and of the ways that the whole can be more than the sum of the parts. The second part of the enumeration is going to be much longer than the first.

So where are we with respect to whether or not it is worthwhile attempting to understand/chacterize emergence--your original question.

My original question? Where was that?

-- rec --

Douglas Roberts-2

Re: A question for the emergentists among you

In reply to this post by Nick Thompson

That was a "quick whack"?

We operate on different plateaus. In different dimensions, more likely. On different planets, certainly.

I was hoping for something more along the lines of

"Once I've attached the 'emergent' label to a phenomenon, I now know that I can apply the following scientific methodologies to solve my problem:"

a) ...
b) ...
c) ...

--Doug

On Sun, Oct 11, 2009 at 8:25 PM, Nicholas Thompson <[hidden email]> wrote:

let me take a quick whack at this.

[...]

Nick Thompson

Re: A question for the emergentists among you

In reply to this post by Robert Holmes

roger will give a more complete answer, but let me just say that I think wimsatt would say that in point of fact, the idealness of ideal gasses exists only in the models. Aggregativity is for him a useful fiction. How a fiction can be a fiction and still useful, is the kind of issue dennett struggles with in his chapter.

By setting the 4 criteria for aggregativity, wimsatt directs our attention to each of the ways in which it can fail.

Nicholas S. Thompson

Emeritus Professor of Psychology and Ethology,

Clark University ([hidden email])

http://home.earthlink.net/~nickthompson/naturaldesigns/

----- Original Message -----

From: [hidden email]

To: [hidden email]
Cc: [hidden email]

Sent: 10/11/2009 4:45:59 PM

Subject: Re: [FRIAM] A question for the emergentists among you

With aggregativity defined that way, Wimsatt notes that "Very few system properties are aggregative." Then what? Is the point that "emergence, defined as failure of aggregativity" has now been fully characterized? Problem solved? I wouldn't agree with that. I think there is more to say than just a negative definition.

An interesting example to which this approach might be applied is an ideal gas. Such a gas satisfies all the aggregativity conditions. Yet it has properties (the gas laws) that the individual components lack.

-- Russ A

On Sun, Oct 11, 2009 at 3:01 PM, Roger Critchlow <[hidden email]> wrote:

On Sun, Oct 11, 2009 at 2:10 PM, Russ Abbott <[hidden email]> wrote:

Roger, Well said.

But there is a further question. Can anything be added to your (Mill's) statement that when you combine some things (e.g., combining a bunch of cows into a herd) the result has properties that the components lack. That is, what, if anything, can one say about those phenomena that exhibit this property? Do those phenomena have anything in common?

Wimsatt lists four heuristics for establishing "aggregativity" of properties: swap "identical" parts in the aggregate; increase or decrease the number of parts in the aggregate; take the aggregate apart and reassemble it; and freedom from non-linear interactions between parts. The heuristics aren't necessarily independent of each other, but neither are they necessarily dependent. So, there are four kinds of emergence which fail just one heuristic, six kinds which fail two different heuristics, four kinds which fail three different heuristic, and one kind which fails all four heuristics. So that's 15 different flavors of emergence, which is perhaps an overestimate, but Wimsatt is still soliciting for additional heuristics.

-- rec --

Eric Charles

Re: A question for the emergentists among you

In reply to this post by Nick Thompson

Ah.... can I change the requested line a small amount?

"Once I've attached the 'emergent' label to a phenomenon, I now know that I CANNOT apply scientific methodologies to the problem that treat the phenomenon as if:

A) it is a simple aggregate of the ingredients
B) its final state was determined by any individual ingredient
C) its final state was determined by any number of 'purely internal' or 'purely external' factors
D) its final state can be adequately described purely by reference to a lower level of analysis "

I wish I could give a positive answer rather than a negative one, but of course we won't have that until the task is done (under some notions of what the task is).

Better that time?

Eric

On Sun, Oct 11, 2009 10:42 PM, Douglas Roberts <[hidden email]> wrote:

That was a "quick whack"?

We operate on different plateaus. In different dimensions, more likely. On different planets, certainly.

I was hoping for something more along the lines of

"Once I've attached the 'emergent' label to a phenomenon, I now know that I can apply the following scientific methodologies to solve my problem:"

a) ...
b) ...
c) ...

--Doug

On Sun, Oct 11, 2009 at 8:25 PM, Nicholas Thompson <nickthompson@...> wrote:

let me take a quick whack at this.

[...]
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Eric Charles

Professional Student and
Assistant Professor of Psychology
Penn State University
Altoona, PA 16601

Nick Thompson

Re: A question for the emergentists among you

In reply to this post by Robert Holmes

Doug,

you wrote ====>

"Once I've attached the 'emergent' label to a phenomenon, I now know that I can apply the following scientific methodologies to solve my problem:"

Well, the experimental method or the comparative method, depending upon the domain we are dealing with. One has to tease apart the effect of the configuration of elements from the effects of the simple presense of the elements.

Here's an example: Once upon a time, many people assumed that a set of properties possessed by groups of monkeys occured because of the manner in which groups were organized. Forty years ago, hoping to demonstrate this, I did a sseries of experiments in which the members of an artificial social group were convened dyad by dyad ... in other words the group had never met as a group but all the potential dyads of the group had met and had an opportunity to behave. Then I summed the dyadic behavior accross all the itneractions and wrote it up as if I was describing a group in the field. The variables of interest were indistinguishable. Therefore, I supposed, triadic, tetradic, n-adic etc., intereactions were not essential to the traditionally observed patterns in the variables of interest. So far as these animals were concerned and these variables being in a social group was just like meeting all the other members of the group one by one.

Now, I dont really believe this to be true, but that was the answer I got, and had I not tired of running a monkey concentration camp, I would have continued research on this subject, and it would have been a study of the emergence of social order in monkeys.

Nick

Nicholas S. Thompson

Emeritus Professor of Psychology and Ethology,

Clark University ([hidden email])

http://home.earthlink.net/~nickthompson/naturaldesigns/

----- Original Message -----

From: [hidden email]

To: [hidden email]

Sent: 10/11/2009 8:43:13 PM

Subject: Re: [FRIAM] A question for the emergentists among you

"Once I've attached the 'emergent' label to a phenomenon, I now know that I can apply the following scientific methodologies to solve my problem:"

a) ...
b) ...
c) ...

--Doug

On Sun, Oct 11, 2009 at 8:25 PM, Nicholas Thompson <[hidden email]> wrote:

let me take a quick whack at this.

[...]

David Eric Smith

Re: A question for the emergentists among you

Nick, hi,

I can't really summon the energy to be part of the emergence thread,
but for this particular post, you may wish to keep an eye on
publications coming out from Flack, deWaal, Krakauer, and
collaborators including Ay and deDeo, on primate interactions. They
have some very strong analysis showing that a very large component of
group power structure and the functions associated with it, such as
policing, is mediated by the response of individuals to dyadic
interactions between others, and very explicitly _not_ to merely the
members who participate in the dyads. They have tested a variety of
p-to-q responses, and find a very strongly significant signal in the
1-to-2 response (i.e. individual responds to dyad), with higher-order
interactions apparently well explained by the composition of 1-to-2,
and an equally strong absence of signal for any of the other
elementary levels, or for any single strong explanatory excess of any
higher-order p-to-q above its dependence on the 1-to-2.

What I have said here is an oversimplification of a longer and more
complicated story involving several forms of interactions (fights,
subordination signals, etc.) with inter-related but distinct dynamics
and timescales, so I haven't done most of it justice. I don't know
how much of the new 1-to-2 work is currently published or on the SFI
working paper list. Some of the earlier papers explaining what
quantitative definitions they attach to the notion of power, and its
relation to policing and other group-coherence attributes, is out in
Nature and several behavior journals, and probably mostly available
from the authors' webpages. All of this work is in various stages of
development, write-up, or submission, and some of it may be presented
in talks as the year wears out. So one way or another it should be
available either now or soon.

Just a topic of interest as a bit of science.

All best, and I do find much of the larger argument interesting and
thoughtful,

Eric

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

glen e. p. ropella-2

Re: A question for the emergentists among you

In reply to this post by Eric Charles

Thus spake ERIC P. CHARLES circa 10/11/2009 09:13 PM:
> "Once I've
> attached the 'emergent' label to a phenomenon, I now know that I CANNOT apply
> scientific methodologies to the problem that treat the phenomenon as
> if:

Excellent modification. I do have a (speculative) positive answer,
though. I've just been waiting to see if anyone else put it forward.

My answer to Robert's question is: Once I trust that a phenomenon is
emergent, I can be more confident in the assumption that the phenomenon
can be used as a mechanism in a layer of abstraction that generates
coarser phenomena.

If a phenomenon is NOT emergent, then, in order to build an adequate
description of the whole system, I must include the details of the
mechanism that generated the phenomenon. I.e. any abstraction of those
details will be inadequate or impoverished... the abstraction will be
too easily punctured. If, however, a phenomenon is emergent, then I'm
under less pressure to delineate each detail of its mechanism and can
get away with encapsulating the phenomenon in a coarser abstraction.

The _use_ to which such a categorization would be put is the method of
replacement in, for example, modeling and simulation. If we need a more
"sciency" method, then we can talk about compressibility. I might be
able to claim that systems exhibiting emergent phenomena are _more_
compressible than those without them.

Note that the above is about emergent phenomena, not emergent
properties. I still think the concept of an emergent property is either
useless, self-contradictory, or just confused.

--
glen e. p. ropella, 971-222-9095, http://agent-based-modeling.com

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Roger Critchlow-2

Re: A question for the emergentists among you

Nice. That sort of turns Bedau on his head without rearranging his features much. Where he is saying that an emergent process cannot be compressed into a smaller computation than a full simulation, you're saying for given computational resource the full simulation of an emergent process gives you the most "complexity" for your buck.

-- rec --

On Mon, Oct 12, 2009 at 9:57 AM, glen e. p. ropella <[hidden email]> wrote:

Thus spake ERIC P. CHARLES circa 10/11/2009 09:13 PM:

> "Once I've
> attached the 'emergent' label to a phenomenon, I now know that I CANNOT apply
> scientific methodologies to the problem that treat the phenomenon as
> if:

Excellent modification. I do have a (speculative) positive answer,
though. I've just been waiting to see if anyone else put it forward.

My answer to Robert's question is: Once I trust that a phenomenon is
emergent, I can be more confident in the assumption that the phenomenon
can be used as a mechanism in a layer of abstraction that generates
coarser phenomena.

If a phenomenon is NOT emergent, then, in order to build an adequate
description of the whole system, I must include the details of the
mechanism that generated the phenomenon. I.e. any abstraction of those
details will be inadequate or impoverished... the abstraction will be
too easily punctured. If, however, a phenomenon is emergent, then I'm
under less pressure to delineate each detail of its mechanism and can
get away with encapsulating the phenomenon in a coarser abstraction.

The _use_ to which such a categorization would be put is the method of
replacement in, for example, modeling and simulation. If we need a more
"sciency" method, then we can talk about compressibility. I might be
able to claim that systems exhibiting emergent phenomena are _more_
compressible than those without them.

Note that the above is about emergent phenomena, not emergent
properties. I still think the concept of an emergent property is either
useless, self-contradictory, or just confused.

--
glen e. p. ropella, 971-222-9095, http://agent-based-modeling.com

============================================================

FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

123