On the left is a photograph of a real snowflake. Most people would agree that it was not created intentionally, except possibly in the rather esoteric sense of being the foreseen result of the properties of water atoms in an intentionally designed universe in which water atoms were designed to have those properties. But I think most people here, ID proponents and ID critics alike, would consider that the “design” (in the sense of “pattern”) of this snowflake is neither random nor teleological. Nor, however, is it predictable in detail. Famously “no two snowflakes are alike”, yet all snowflakes have six-fold rotational symmetry. They are, to put it another way, the products of both “law” (the natural law that governs the crystalisation of water molecules) and “chance” (stochastic variation in humidity and temperature that affect the rate of growth of each arm of the crystal as it grows). We need not, to continue in Dembski’s “Explanatory Filter” framework, infer “Design”.
The patterns below, also have six-fold rotational symmetry, and the process that created them is also one in which no two are alike. However, despite this, they were, in fact, designed. By me. I wrote the program that generated them, and I can generate as many as I like. The chances that two will be identical is pretty low (though possibly not as low as that of two snowflakes). I did this by first of all “designing” a law (one that ensures six-fold rotational symmetry), and then by “designing” a stochastic algorithm that randomly generates “ice” by drawing from a built-in probability distribution.
Clearly, applying the Explanatory Filter does not easily allow us to infer design in the second case, but that is not a problem – Dembski does not claim that the ID detecting methods he proposes will not produce false negatives, he only claims a good record for true positives. And in any case, that isn’t what I want to discuss in this post. I’m not asking people to infer which were designed and which were not. I know that the second set were designed and the first was not.
What I’d like to discuss is how the processes differ. Both involve a law (natural in the first, designed by me in the second), and both involve stochastic processes (natural in the first, designed by me in the second). But we would probably agree that the first was the result of a non-teleological process, regardless of the fact that when the conditions are right for snow, snowflakes of a reliable general pattern form, while the second are the result of a teleological process, namely my intention to make snowflake-like patterns for Christmas cards (yes, I know I’m late) and for this post.
(Have a merry Christmas all, by the way!)
At the moment, I’m reading Dembski’s book, Being As Communion. I was interested to see that he uses “teleology” more or less interchangeably with “intelligence”, which is a change from the definition he used to use (“by intelligence I mean the power and facility to choose between options”), and which unambiguously entails the concept of “intention”, something he back then explicitly claimed was outwith the domain of science (I profoundly disagree), only coming “back on the table” after “intelligence” (old definition) has been established. He also, in Being As Communion, uses “design” in the sense of “pattern” rather than as in “by accident or design”. So under his current usage, “Intelligent Design” means “Patterns produced by teleological processes”, which I think is actually clearer.
So I am curious now about his view of the difference between what he characterises as “materialism” and his own view (and interestingly, he places Nagel on the same side of this perceived divide).
I think that Dembski would say that, as a materialist, I could avoid the conclusion that my ersatz snowflakes are the result of a teleological process by claiming that they are, nonetheless, the outcome of interactions between matter in my body and brain, and that thus they are not essentially different from the non-teleological snowflake because I am not really an intentional being – my sense of intention is illusory.
Whereas a non-materialist, or at least someone not a priori committed to materialism would say (as I understand Dembski’s thinking here), would regard the second as a special case of a process (teleology) operating within the world in a way that may also be apparent in such phenomena as the Origin of Life, possibly the evolution of the bacterial flagellum, and possibly in the “fine-tuning” of our universe to be life-friendly.
There are a number of things that could be said about this, but the point I want to make in this post, is that I do NOT think that the intentional processes by which I generated the second lot of snowflakes are illusory. I think there is a real and major distinction between the processes that created the real snowflake and the processes that created the artificial ones (although I will note in passing that often the way we infer artifice, i.e. intelligent design, is that the results are not as complex as the real thing!)
So what is that distinction? What is the property of teleological processes that makes them different from non-teleological ones?
I suggest that the answer is fairly straightforward: a teleological process, whether stochastic or under strict control), entails some kind of prior representation (of something to something, possibly itself) of what the end result of the process will be like. The only advance “representation” of a snowflake is that inherent in the laws and probability distributions that govern its emergence from non-snowflakeness. Whereas had you stopped me before I’d finished writing my MatLab code and asked me what I was doing (in the middle of the night, when I couldn’t sleep) I’d have told you quite clearly: “I’m trying to write some code that will generate snowflake patterns, and I’m fiddling about with possible distributions my code will randomly draw from trying to find one that tends to the most snowflake like patterns”.
I am, in other words, selecting my actions so as to execute those with the greatest probability of producing an outcome that matches some prior template.
In yet other words, I start by imagining a snowflake, then I set about experimenting, trying things out, rejecting those that don’t work very well, dreaming up different way sof doing it, until I end up with a reliable series of snowflakes.
And even as a so-called “materialist”, that process is very different from the one that produces a real snowflake. The “intentional” component is not illusory – it can be objectively detected as being present in my actions, and not present in the processes that I hope will give us a White Christmas this year.
Intention, in other word, is perfectly real, and “real teleology”, as Dembski calls it, is perfectly compatible with the view he likes to call “materialism”.
[edited to fix grammar!]
Well, I would agree. My point was a bit more trivial really – that if the word “teleology” is to have any utility, it’s definition should include some things and not others.
Otherwise what does it denote?
One other point I should make – this is primarily addressed to Mung and William:
I was not suggesting that Dembski ever thought that Darwinian evolutionary processes were intelligent. My point was that given his definition of intelligence, which has some merit, in my view, they are included, and do in fact explain what Dembski wants intelligence to explain – the fact that biological organisms are exquisitely optimised for survival and reproduction.
Even if he DID intend (heh) “intention” to be implicit in his definition (“the power and capacity to choose between options in order to achieve some intended goal”), then my point stands. A system with the capacity to choose between options with NO intended goal can still result in “information” as Dembski defines it – a specificiable pattern that is unlikely under the null of no optimising process.
I don’t see any reason for adopting Mungian teleology myself. Either it’s a simple definitional quirk, which seems to reduce to “If there is a current state A that transitions to a state B at a later time, teleology is involved”, or it is a statement about the nature of reality (“as we have goals, so does something else”) which appears to require empirical support to me.
See also ‘intent is a force’.
With no goal how does the system determine which option to choose? Is the choice merely random? Can a random act even be considered a choice?
peace
Funny, I just had the following thought this morning rising from bed, before I’d seen this OP:
Sure, no two snowflakes are exactly alike. What’s so amazing about that?–No two ANYTHINGS are exactly alike!
fifthmonarchyman,
Sure, why not? If the system has no ‘option’ other than to fall into one pathway or another. A razor will not indefinitely remain balanced on edge. It has no choice as to whether there is a choice to be made.
eta: I am obviously not saying that the razor has intent. But this nonintentional use of the word ‘choice’ is common and legitimate – for instance in QM.
According to Aristotle, it is the nature of causation that all things (substances, beings) have a final cause. For each particular thing, it has a specific final cause by virtue of the kind of thing that it is.
Or so the Aristotelians would have us believe.
Personally, I think that one can learn a great deal from studying Aristotle’s metaphysics but that it is ultimately unacceptable in light of 21st century sciences.
Well, in the case of natural selection, what is chosen – selected – are those variants in the population that are best able to exploit the environment’s resources and avoid its hazards.
Well, it depends what you mean by “random”. It’s stochastic, but biased in favour of one outcome over another.
Depends on what you mean by random. You can certainly talk about a “random selection” – we try to do that intentionally in science!
And to you too 🙂
So how would you define “teleology”, KN? What do you think the word mostly means?
I have a question about the real snowflake verses the fake copies that EL created.
Could we model snow flakes so well that it would be impossible to distinguish real snowflake images from the models?
peace
So we do have a goal then. Correct?
A goal to better conform to the environment.
Sounds like striving toward a goal to me. What am I missing?
peace
Does a pair of die choose to land on snake eyes 2.77…. percent of the time?
peace
I think it mostly means the purposive striving of organisms that must maintain some structural invariance over time relative to their thermodynamic and material openness to their environments.
On my view, the origin of life is the origin of teleology. Contra Aristotle, there is no abiotic teleology.
So does that mean you think intentionality, in the sense of aboutness for mental representations or directed actions, is inherited from the overall intentionality of the organism and so there is no need for a separate explication to naturalize the intentionality of these two?
I saw your reference to Millikan’s pushmi-pullyu concept earlier. It has occurred to me that this bears a superficial resemblance to the tight coupling of the representations involved in PP for action/perception (corresponding to imperative/descriptive), but I have not tried to think that through in detail.
As I understand it, this is not true of quantum entities like photons or electrons. (even properties like position don’t work due to uncertainty and the probabilistic nature of the quantum state)
You would not ask that if you lived in Canada and had to shovel the real stuff.
I think it would be possible to model the snowflakes so well that they would be indistinguishable from a photograph of the real thing.
I didn’t try that hard – I didn’t attempt to simulate anything much like the actual process, although I could have done. But mine differ from real snowflakes in that they just have hexagonal symmatry. The 60 degree angle isn’t coded into the patterns along the arms, only into the fact that there are six arms 60 degrees apart.
I also picked out the most snowflake ones! And did a bit of post processing as well, to smooth them a bit.
Here’s the code (it’s in MatLab, but probably readable to non-MatLab coders as MatLab is pretty transparent):
Size=100;SnowFlakeNumber=1
for iFlakes=1:20
close all
X=[randn(Size,1);randn(Size,1)/100]
Y=randn(Size*2,1);
Starter=[X,Y;-X,Y;X,-Y;-X,-Y];
Arm1=complex(Starter(:,1),Starter(:,2));
Angles=angle(Arm1);
Rads=abs(Arm1);
Angles2=Angles+pi/3;
Angles3=Angles2+pi/3;
Arm2=Rads.*exp(i*Angles2);
Arm3=Rads.*exp(i*Angles3);
SnowflakeX=[real(Arm1);real(Arm2);real(Arm3)]
SnowflakeY=[imag(Arm1);imag(Arm2);imag(Arm3)]
figure('Position',[100,300,500,500],'Color',[1 1 1]);
scatter(SnowflakeX,SnowflakeY,'*','w')
set(gca,'Color',[0 0 .3]);
axis([-4,4,-4,4])
drawnow
Choice = questdlg('Do you want to save this snowflake?', 'Keep?','Yes','No','Quit','No')
switch Choice
case 'Yes'
saveas(gcf, ['Snowflake_' num2str(SnowFlakeNumber) '.png'])
SnowFlakeNumber=SnowFlakeNumber+1
case 'Quit'
break
end
end
OK, in that case Darwinian evolution has a goal! Welcome to the Dark Side 🙂
fifthmonarchyman,
No. But if you throw a pair of dice (in an environnment that has gravity and a surface) you will get a result. Given a set of alternative possible results, that can be viewed as a system ‘choice’, whatever number actually comes up on that throw. Aggregated throws will converge on a distribution, dependent on the fairness of the dice.
WJM said:
EL said:
Really? Haven’t you agreed with me in the past that all science does is create models? Isn’t what science refers to as a ‘force” (like gravity) simply a designation for a model – a description – of behavior? If not, then what does “force” mean, scientifically speaking?
I fixed what EL said:
This is one of the problems with EL trying to “interpret” and “characterize” what Dembski is saying. I run into the same problem with EL, keiths and others here all the time. I’m right here on hand to explain what I mean, and even then EL rarely shows any comprehension of what I mean when I say certain things and then answer questions.
I understand what Dembski is saying and what he is doing when he uses different terms to describe the same essential thing – I’ll do the same thing to just avoid repeating the same term over and over and over. I know that intelligent design, intention, real teleology – these are all different terms that refer to the same essential thing.
When ones main purpose, however, is to find fault with someone’s argument or point of view or to use what they say to advance their own point of view, then one focuses on specific terms and how they could be interpreted in a way favorable to their own thesis – consciously or not.
From what I gather, Dembski is making an argument that corresponds to a description of my views that I gave here some time ago using different terminology. I used the term “psychoplasm” to describe what the essential substance of the universe was; Dembski apparently uses the term “information”. In both our views, I would think, intention/intelligence manipulates and directs information/psychoplasm. Therefore, what information/psychoplasm appears as in the universe is directed by some form of intention/intelligence and cannot be the cause of intention/intelligence.
Both information (the substance) and intelligence (the architect) would be essential aspects of existence.
How about if we used your program in drive a (hypothetical) 3D snow printer to create the output snowlflakes made from real water and any required impurities?
Would that be indistinguishable from naturally-created snowflakes?
I would say the output would differ, because you are not attempting to model the scientific theory of how snow forms, and so it is likely there there will be situations which produce snowflakes which your model cannot produce, or conversely.
An interesting implication for this general type of question is whether models of neural processing could reproduce a mind. But here we don’t need a 3D printer, I would say, if we agree that the neural system is doing information processing. Then the question is whether a simulation of information processing using an equivalent model is indistinguishable from the real thing.
My view is that the two are not distinguishable.
By information processing, I mean something quite general, eg as Piccinini defines it (pdf), and not Fodor-style computation using abstract symbols.
I am also assuming that embodiment is accounted for somehow in the simulation, eg it drives a synthetic human and reacts to the equivalents of hormonal and elector-chemical environment. I’m also not limiting the simulation to the brain only.
fifthmonarchyman,
Regarding all biased processes as striving towards a goal would be somewhat … eccentric. If you drop a magnet in a roulette wheel, you produce a biased distribution on subsequent spins.
All that happens in NS is that one variant consistently produces more surviving offspring than the other in the given environment. Both variants have the same ‘goal’ – to survive and reproduce.
That’s interesting–thanks. I guess to be accurate, I’d have to have said something like ‘no two macroscopic anythings are exactly alike.’
What do you mean its biased? What is the bias? How do you determine the bias?
phoodoo,
I presume you are talking to me. You mean roulette wheels, NS, or both?
Bias and its absence will both produce evolutionary results.
If you have two independent variants, and assume (for simplicity) a consistent distribution of offspring numbers when tested in isolation, 3 things can happen if you put them together in one combined population at a 50/50 ratio:
1) They produce the same offspring numbers. One will be eliminated by drift.
2) A exceeds B. B will be expected to eliminate A, with a certainty dependent on the extent of the differential.
3) B exceeds A. A will be expected to eliminate B, with a certainty dependent on the extent of the differential.
2 and 3 are actually the same. There is basically either bias or there isn’t, but in both cases evolution will occur. The no-bias case is actually akin to balancing a razor on edge. It cannot stay poised at 50/50 indefinitely, as one can demonstrate mathematically, computationally and experimentally. It’s basically the principle behind the chemostat, which is a lab workhorse for purifying a cell line. There’s no reason to suppose this does not happen in nature too.
Yes, that’s correct. I think that the intentionality of mental representations needs to be explained in terms of the intentionality of the organism. Here I’ve been deeply influenced by Okrent’s argument, contra Millikan, that organismal goals are explanatorily prior to biological functions; we can’t explain what the proper function of a heart is without a prior understanding of the goals of the organism that a heart’s proper functioning enables. Likewise with brains.
However, I do allow that the intentionality of language and lingual thought is importantly different from that of non-linguistic animals. Getting really clear on that is what I’m trying to do now.
That’s a really interesting connection! But in Clark, there’s merely a very coupling between directives (predictions) and descriptives (prediction errors), and not the very same representation carrying out both functions — right?
What I do like about Clark’s work is that, except for not taking evolution seriously enough, he offers a detailed model for the proper function of brains: the proper function of brains is to implement computations that minimize the discrepancy (subject to other constraints) between the relevant-for-goal-satisfaction information that can be actively sampled from the energetic flux and the behaviors that are likely to be conducive to satisfying those goals.
One aspect of Clark’s book I found deeply dissatisfying is that he builds a model of what brains generally do, but all his data are based on studies of human brains. But therefore he cannot explain what makes human brains different from other kinds of brains.
Of course the real differences between human beings and other animals are not going to be found at the neurological level, but those differences do have neurological correlates and we want to understand those as well.
Allan Miller,
I think my question was simpler than this. What is the bias, and how do you know what the bias is?
To me, that seems pretty close to connecting teleology with homeostasis — except that homeostasis is not limited to organisms.
BruceS,
The harder question would be whether a fully functional synthetic mind could be built in less than eighteen years, which is how long it takes bio-culture.
Right — whereas what I want to say here is that teleology is just the kind of homeostasis that is distinctive of organisms, and that makes living things different from dead ones, or life from non-life.
Allan Miller,
Seems like what you have said is that either A exceeds B or B exceeds A. What was the bias?
OK, if that’s what you meant, sure. All scientific concepts like “force” are models. I thought you meant that the referent for the word “force” *was* “a description of behaviour”, which didn’t make much sense. But, sure, that’s exactly what I think – that we don’t have direct access to reality, what we have are models. I think Dembski is saying something quite similar, except he’s going one further, and saying (I think) that it’s the “model” (in my terminology) or the pattern by which we know a thing is there i.e. the “information” (in his) that is the sense in which a thing can be said to be “real”. That the model IS the reality. Which is sort of neat, if somewhat solipsistic.
OK. Have you read this book, by the way?
But it is not my “main purpose” to “find fault” with Dembski’s argument. My purpose in reading his book is to try to understand what he is saying and decide whether or not it makes sense to me. That might, inter alia, involve finding a fault if a fault is to be find. It might also involve discovering errors of my own. Thus far I like his book, or would if it weren’t his straw-man characterisation of “materialism” that he takes constant side-swipes at.
Yes, they are quite similar. I infinitely prefer “information” to “psychoplasm” though, especially as Dembski defines “information” quite carefully.
The bias is towards organisms that have genotypes that give them an advantage in the current environment.
Lizzie,
Can you explain what is this bias you describe?
Yes, of course.
A population of organisms reproduces with some variance. In each generation, some die before reaching maturity, others die before producing many offspring. Others live long enough and successfully enough that they produce lots of offsspring and those offspring thrive.
A lot of this is due to luck – “random chance”; some organisms just happen to miss disasters and chance upon good food supplies. However, it isn’t entirely “random” – there is a “bias” inbuilt, in that some of the chances of an organism’s survival is due to the versions of the genes it inherits – maybe some that help it run faster, or digest a widely available food more efficiently.
So the system as a whole “selects” the next generation largely “randomly”, but but is slightly “biased” in favour of genotypes that result in organisms who happen to have genes that help them exploit the current resources and avoid its hazards.
And it is this “bias” that Darwin called “natural selection”.
phoodoo,
No, I said that either there is a differential (A and B are essentially interchangeable labels), or there isn’t (A=B). Either way you get evolution, but with bias you get evolution by Natural Selection.
If there is a differential, that is a bias in production of offspring numbers. If there is no differential, there is no bias. Are you saying that there is never a bias (ie all types produce the same mean offspring numbers in all circumstances)? Or you want a simple answer to ‘what was the bias’ that is applicable to every situation everywhere? I think you have that. The bias in the general case is the differential in offspring numbers between competing types. In each specific case, the causal agency is different – cold, predation, and so on.
Elizabeth,
Lizzie, you just restated that there is a bias, which favors some organisms. I already got that you are stating there is a bias-that is not what I am asking you. I am asking you, in living organisms what is the bias, how do you determine this presumed bias?
I am not asking you to restate your theory.
Allan Miller,
Allan,
What you are saying then is that bias means differential. The words are interchangeable? Differential means the higher differential has a bias? That theory is thus un-falsifiable.
So you don’t determine the bias until after the results, and the results can’t be falsified, because the results are the proof.
I am not expecting a fully synthetic mind to ever be built.
Well, it’s not always easy, but one way you can do it is to figure out what the “expected value” is for the prevalence of an allele, and if the observed prevalence is a lot higher than that, you can infer, with a certain p value, that it was the result of selection, i.e. that it conferred some benefit in some past or present environment.
If you can also find out what it does, that can help – maybe it makes beaks larger, for instance, and you note that larger beaked birds can access the most common kind of seeds more easily.
William J. Murray,
I don’t think this is particularly fair to Elizabeth but it does highlight the importance of rigorous definitions. As Elizabeth points out, evolutionary processes meet the criteria set by Dembski’s own definition of “intelligence”. Since Dembski doesn’t often engage with his critics directly, we can only go by what he has written.
Can you point to any specific errors in Elizabeth’s understanding of Dembski’s definition?
So likewise you cannot determine who is taller until you get the result of a measurement, that result can’t be falsified( unless you measure again), because the result is the proof who is taller? Is that your objection?
I think the only way of doing it would be to set it up to evolve.
OK, but I’m still not seeing definition of teleology there.
To fully describe the force acting upon an object, you must describe both the magnitude (size or numerical value) and the direction.
The Meaning of Force
phoodoo,
It’s not a theory, it’s a definition. If a die is weighted, it is biased. If not, it isn’t. If one type produces more offspring than another, there is similarly a bias. Hard to deny, really, though I can see you’re having a good go (again).
You keep tootling this trumpet. If you have a weighted die, it is biased before you make a single throw. If there is a reproductive differential, then there is a bias before anything has reproduced. Regardless whether anyone looks or not. Are you saying that there has never been a differential in offspring numbers, in any environment, ever? Or that, if we can’t go back and measure it, it might as well not have happened? Strange how bacteria with different growth rates compete, innit? Even if you don’t know beforehand which is faster.
There are only two possible results for competing types: a differential or no differential. There is no other option. In both cases, evolution occurs.
I was asking for a definition of teleology.