Register | Sign In


Understanding through Discussion


EvC Forum active members: 65 (9162 total)
7 online now:
Newest Member: popoi
Post Volume: Total: 915,815 Year: 3,072/9,624 Month: 917/1,588 Week: 100/223 Day: 11/17 Hour: 0/0


Thread  Details

Email This Thread
Newer Topic | Older Topic
  
Author Topic:   Animal and Extraterrestrial Intelligent Design?
lpetrich
Inactive Member


Message 1 of 31 (38151)
04-27-2003 12:50 PM


I hope it's OK with the mods for me to discuss an article I'd written on that subject; I'll present an abridged version here.
It is Animal and Extraterrestrial Artifacts: Intelligently Designed? In it, I show that the scientific community has dealt with some important examples of nonhuman and nontheological designers.
I start off with animal behavior, noting George Romanes's "Animal Intelligence", which catalogued numerous purported examples. However, Romanes's work has been remembered mainly as an example of how not to do animal-behavior research.
Some species of spider build elaborate webs that seem as if the spiders had intelligently designed them. However, the work of Thiemo Krink shows that a simulated spider can build a web with the help of some rather simple algorithms. Furthermore, this simulated spider's performance can be improved with the help of some algorithms equivalent to evolution by natural selection.
Turning to other arthropod architects, like ants and bees, much of their behvior can likewise be explained by relatively simple algorithms. In fact, there is a whole discipline of "artifiicial life" dedicated to studying such algorithms.
But does such research apply only to very small-brained creatures? Beavers may seem to intelligently design their dams, as William Dembski himself seems to think, but beavers build dams by piling sticks and mud on wherever they hear rushing water.
But it's clear that even complicated instincts cannot explain everything, and learning can often supply valuable flexibility. But most forms of learning are difficult to call intelligent design. Baby birds can imprint on entities that do not resemble their mothers very much, and the famous horse "Clever Hans" was inadvertently trained to notice subtle cues from his master.
But there is a kind of learning that seems close to intelligent design, "insight learning". This term was coined by Wolfgang Kohler, who noticed that chimps pause for a while and then implement a solution to some puzzle. It is as if the chimps create a mental model of stacked crates or joined sticks, and then implement it. Which may be interpreted as performing intelligent design. But it's the behavior that provides the clue, not the "explanatory filter" of "what else can it be?".
Moving to extraterrestrials, the first speculation of design was Kepler's theory that the Moon's craters had been designed by its inhabitants. However, giant meteorite impacts have proved to be a much more reasonable hypothesis.
Mars's "canals" were a mistranslation of an Italian word for "channels", and some astronomers, notably Percival Lowell, worked out the features of their alleged designers in great detail. However, some astronomers, like Antoniadi, could never see them, and spacecraft pictures revealed no traces of them. They were perceptual artifacts.
However, one picture revealed a "Mars Face", which grew famous among a new generation of Martian-artifact-seekers, notably Richard Hoagland.
Beyond the solar system, in 1967, a certain Jocelyn Bell Burnell discovered some radio source that made pulses every 1.337 seconds. She and her colleagues soon discovered others, which they half-jokingly attributed to "Little Green Men". However, these sources showed no sign of being in orbit, and astrophysicists like Thomas Gold worked out a more plausible hypothesis: these "pulsars" are rotating neutron stars that are slowly spinning down.
Despite such failures, SETI efforts have continued, and search efforts have been directed toward some guesses as to what an ET signal would be like, notably that they are very narrowband signals near important radio spectral lines. These guesses are based on considering what would make a radio broadcast prominent to a possible listener, and not by invoking the Explanatory Filter.
These examples show that recognizing intelligent design is much more difficult than it might at first seem, and that Dembski's claim to have solved the recognition problem is very overoptimistic.
A parallel case is vitalism, the theory that living things have some special "vital force". It is nowadays thoroughly discredited, not from "mechanistic presuppositions", but from the massive success of mechanistic explanations.
So in conclusion, the mainstream of the scientific community has no presupposition against intelligent design. Instead, many seeming cases of intelligent design have turned out to be something other than that.

Replies to this message:
 Message 3 by Peter, posted 04-28-2003 7:42 AM lpetrich has not replied

  
NeilUnreal
Inactive Member


Message 2 of 31 (38153)
04-27-2003 1:55 PM


Good post. However, I would argue that in one sense, Dembsky is technically correct when he says beavers use intelligence to construct dams. Intelligence -- in the way most recent investigators conceive the term -- is made up of a wide spectrum of behaviors. It begins on a low end with things as simple as the construction of crystals and the folding of proteins, to simple sense-action behaviors and instincts, to learning, and finally to insight and self-awareness.
In this respect, "stack sticks where you hear running water," is part of this hierarchy of intelligence. Perhaps where Dembsky makes a mistake is in limiting intelligence to the more advanced categories of insight and self-awareness. ID, in this view, is more properly an acronym for "Insight-based Design."
When we talk about things like determining whether an exterrestrial radio signal indicates "intelligent life," the ID movement views it as a problem in placing something in a scholastic category (intelligent vs. non-intelligent); mainstream science, by contrast, is actually asking whether the signal is interesting in a way that other things we know about and consider common are not. For ID, the issue of intelligence is qualitative, for mainstream science it is quantitative.
This also explains the differing views towards genetic and evolutionary programming and modeling. ID views the results as invalid, since the involvement of a intelligence (us), has contaminated the process. Mainstream science doesn't see an issue, because it sees the barrier between intelligence and non-intelligence as artificial and susceptible to reductionistic investigation.
As a person of faith, I do not find the mainstream/reductionistic view threatening. However, I do understand the discomfort some religious people have with this view.
-Neil

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 3 of 31 (38194)
04-28-2003 7:42 AM
Reply to: Message 1 by lpetrich
04-27-2003 12:50 PM


I feel a need to point out that just because a simple
computer program can emulate something (anything) does
not mean that that is the way that the real thing
operates.
Spider webs are an interesting example.
What prompts a spider to form a web?
For me 'design' can only be 'intelligent' as it implies
some planning process in advance of implementation.
Even if the 'mental' process is simple, like:
Hunger triggers web-building.
(for example) it could be 'design'. It depends how/whether
spiders select a site for web-building and only build webs
when they require food etc.
A computer model is a model ... it's not the thing. At best
such models suggest that spiders have simple processing
capability (which is intuitive).

This message is a reply to:
 Message 1 by lpetrich, posted 04-27-2003 12:50 PM lpetrich has not replied

  
lpetrich
Inactive Member


Message 4 of 31 (38885)
05-03-2003 6:08 PM


Peter:
What prompts a spider to form a web?
For me 'design' can only be 'intelligent' as it implies
some planning process in advance of implementation.
Even if the 'mental' process is simple, like:
Hunger triggers web-building.
However, that is more likely some reflex action / instinct than serious planning. Like web building itself; spiders likely have some criteria for selecting good spots to build webs.
Exactly how spiders select web-building sites is uncertain, but they may do so in a fashion analogous to how honeybees select hive sites.
Honeybees choose enclosed spaces with certain dimensions, spaces that open to the outside world toward their bottom. They do that by looking for holes, flying inside of them for a while, and returning to the swarm. They may measure the spaces' dimensions by timing how long it takes to hit a wall, and repeating this measurement several times to obtain an average.
And when they return, they try to recruit other swarm members, with the more enthusiastic recruiters getting more fellow bees to follow. Whichever hive site gets the most of this kind of vote is the one selected. This suggests that honeybees have some sort of scoring system for would-be hive sites, with their scores calculated from their scouting efforts.

Replies to this message:
 Message 5 by Peter, posted 05-11-2003 5:53 AM lpetrich has not replied

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 5 of 31 (39676)
05-11-2003 5:53 AM
Reply to: Message 4 by lpetrich
05-03-2003 6:08 PM


The honey-bees behaviour doesn't sound like something that
would normally be put down to 'instinct' does it?
It involves measurment, lobbying, and a democratic process ...
Likewise, we do not know why a spider builds a web, or whether
they choose to do so or are 'programmed' to do so.
The main point I was making, however, was that a computer
model that produces similar results does not prove that that
is how it's done in the real-world system being modelled.
One of the major failings (for me) in most animal behaviour
studies is that humans seem to start with the assumption that
only 'we' have 'intelligence', so that option is ruled out
in other creatures rather than investigated properly.
'Instinct' covers a lot of bahvaiours in material I have read,
and yet some of the most complex and co-operative behaviours
seem to require communication and action based upon 'new
information.' Even something as seemingly simple as telling
your hive-mates where the good flowers are requires an encoding
of informtion on the one side, and an ability to decode it on
the other ... that sounds like languistic capability to me.

This message is a reply to:
 Message 4 by lpetrich, posted 05-03-2003 6:08 PM lpetrich has not replied

Replies to this message:
 Message 6 by NeilUnreal, posted 05-11-2003 1:35 PM Peter has replied
 Message 10 by John, posted 05-17-2003 11:27 AM Peter has replied

  
NeilUnreal
Inactive Member


Message 6 of 31 (39707)
05-11-2003 1:35 PM
Reply to: Message 5 by Peter
05-11-2003 5:53 AM


quote:
The main point I was making, however, was that a computer model that produces similar results does not prove that that is how it's done in the real-world system being modelled.
This is one of the purposes of building computer models. A researcher constructs an hypothesis about how a system performs some activity (e.g. how an animal behaves, how humans do mental arithmetic, etc.). A model, usually on computer nowadays, is built showing how the system might be constructed in order to perform the activity. Both the original subject and the model are then tested under novel conditions. If the behaviours are similar, the model of the system is considered robust. If the behaviours are dissimilar, the model of the system is discarded or modified.
For example, a model might be constructed of how an orb-weaving spider which favors parallel vertical attachments builds a web. The web-building behaviour of both the spider and model are then compared when forced to build webs without parallel vertical attachments.
quote:
One of the major failings (for me) in most animal behaviour studies is that humans seem to start with the assumption that only 'we' have 'intelligence', so that option is ruled out in other creatures rather than investigated properly... that sounds like languistic capability to me.
This was true for some ethologists of the extreme "Skinner" school of behaviourism, but is not generally the case in more recent times. Most researchers are doing exactly the opposite: investigating animal intelligence while recognizing that "intelligence" exists along a continuum from reflex through instinct to introspection. The modern view is that intelligence is neither an "elan vital" nor a behaviourist's black box, but a complex emergent behaviour.
In other words, many of your criticisms are valid and important, but they were recognized as such during the development of ethology and are now taken into account.
-Neil

This message is a reply to:
 Message 5 by Peter, posted 05-11-2003 5:53 AM Peter has replied

Replies to this message:
 Message 7 by Peter, posted 05-12-2003 6:39 AM NeilUnreal has not replied

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 7 of 31 (39793)
05-12-2003 6:39 AM
Reply to: Message 6 by NeilUnreal
05-11-2003 1:35 PM


Your opinion on computer modelling (and its underlying
assumptions) are exactly my problem with computer
modelling.
What the computer model does is match the input-output mapping
of the real-world activity with some or other degree of
precision. That does not mean the internal computations are
the same.
As to the other thing ... thanks for updating me, that's something
that has concerned me about behavioural research, but it's
been a while since I did any reading in that area.

This message is a reply to:
 Message 6 by NeilUnreal, posted 05-11-2003 1:35 PM NeilUnreal has not replied

  
NeilUnreal
Inactive Member


Message 8 of 31 (39809)
05-12-2003 1:57 PM


quote:
Peter: thanks for updating me...What the computer model does is match the input-output...does not mean the internal computations are the same
No problem, I just wanted to point out that the majority of researchers are aware of the (very valid) issues you raised and try to control for them. Unfortunately, the descriptions of these controls are rarely mentioned outside of the primary literature. Often, in abstracts and popular reports, only the results and conclusions are presented.
The technique I described for varying the input to both the model and the real-world subject are an attempt to make the internal operation of the model match the internal operation of the subject. This is what modelers mean when they use the term "robust" -- that the input/output of the model matches the input/output of the subject in a way that gives confidence that the internal operations of the model mirror the internal operations of the subject.
And sometimes, especially for very simple organisms like flatworms, behavior studies do attempt to model the actual neural behavior.
One can never be 100% sure of the results, because as you pointed out: "A computer model is a model ... it's not the thing." But that's science.
-Neil
[This message has been edited by NeilUnreal, 05-12-2003]

Replies to this message:
 Message 13 by Peter, posted 05-19-2003 6:22 AM NeilUnreal has not replied

  
lpetrich
Inactive Member


Message 9 of 31 (40501)
05-17-2003 11:06 AM


Peter:
The honey-bees behaviour doesn't sound like something that
would normally be put down to 'instinct' does it?
It involves measurment, lobbying, and a democratic process ...
Peter, do you mean to imply that honeybees have consciousness?
Likewise, we do not know why a spider builds a web, or whether
they choose to do so or are 'programmed' to do so.
And do you also mean to imply that spiders have consciousness?
The main point I was making, however, was that a computer
model that produces similar results does not prove that that
is how it's done in the real-world system being modelled.
True, but that's where principles like Occam's Razor and falsifiability come in. Let's consider two hypotheses for the formation of crystals:
Crystals form because their molecules seek the lowest-energy configuration, which is being part of a crystal lattice.
Crystals form because little elves carefully place the molecules into crystal lattices.
The first hypothesis is much simpler than the second in a certain way; it does not have the rabbits-out-of-hats hypothesis of those little elves. Furthermore there are some computer simulations that are consistent with it.
One of the major failings (for me) in most animal behaviour
studies is that humans seem to start with the assumption that
only 'we' have 'intelligence', so that option is ruled out
in other creatures rather than investigated properly.
Peter, I'm not sure what you want. The opposite? That every animal be presumed to have human-scale intelligence until shown otherwise?
George Romanes had become a subject of ridicule for very good reasons -- his approach was imprecise and anecdotal.
I'm not endorsing extreme Skinnerian behaviorism by any means; I think that internal mental states and processing are a legitimate subject for inquiry, even if they can only be inferred. But what internal states and processing can one reasonably infer the occurrence of?
'Instinct' covers a lot of bahvaiours in material I have read,
and yet some of the most complex and co-operative behaviours
seem to require communication and action based upon 'new
information.' Even something as seemingly simple as telling
your hive-mates where the good flowers are requires an encoding
of informtion on the one side, and an ability to decode it on
the other ... that sounds like languistic capability to me.
Except that I've had several years of programming experience under my belt, and I know that what may seem "linguistic" or whatever can often be produced in the total absence of conscious thought.
That bees have some sort of "memory" is very clear, but this "memory" is likely some sort of hardcoded slot, something like what goes on in a computer -- no consciousness needed.
Furthermore, I'd noted that if many animals are much smarter than they act, then they are causing themselves needless trouble by acting dumb.

Replies to this message:
 Message 12 by Peter, posted 05-19-2003 6:19 AM lpetrich has not replied

  
John
Inactive Member


Message 10 of 31 (40503)
05-17-2003 11:27 AM
Reply to: Message 5 by Peter
05-11-2003 5:53 AM


quote:
It involves measurment, lobbying, and a democratic process ...
Not really. The behavior can be mimicked with a fairly simple ruleset similar to that of DarwinBots, for example.
------------------
No webpage found at provided URL: www.hells-handmaiden.com

This message is a reply to:
 Message 5 by Peter, posted 05-11-2003 5:53 AM Peter has replied

Replies to this message:
 Message 11 by Peter, posted 05-19-2003 6:06 AM John has replied

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 11 of 31 (40624)
05-19-2003 6:06 AM
Reply to: Message 10 by John
05-17-2003 11:27 AM


That's my point ... mimiced ... doesn't mean that's
the way it's done in the real system.

This message is a reply to:
 Message 10 by John, posted 05-17-2003 11:27 AM John has replied

Replies to this message:
 Message 21 by John, posted 06-01-2003 7:54 PM Peter has replied

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 12 of 31 (40625)
05-19-2003 6:19 AM
Reply to: Message 9 by lpetrich
05-17-2003 11:06 AM


I'm not implying that they do, I'm saying that it has
been an untested assumption that they don't.
Not sure how you would test for consiouness anyhow.
In most other studies such bias would tend to invalidate
the results. i.e. basing conclusions on unsupported assumptions.
Back to computer models ... suppose you have two possible
mechanisms for something, neither of which require the supernatural
but one is more complex than the other.
Both a complex computer model and a simple computer model can
achieve consistent results.
Or suppose that one theoretical mechanism cannot even be modelled
fully with current software technology.
Is it justified to say that because the simple answer can be
modelled, that that's what must happen?
In the case of physical processes we may get close ... but even
the most complex mathemetical models of physical phenomena are
not 100% accurate ...
In the 'not sure what I want..' well basicalyy it is studies that
are untainted by a 'human superiority' assumption, and unbiased
in approach to what is going on to provoke behaviours.
If we start from a mechanistic model for ALL animals then studies
are inapproriately biased (the same way as they would be if
we assumed that ALL animals had high-orders of intelligence).
Phrases like 'human-scale intelligence' are founded in that
same self-superior bias.
If someone designed a program that passed the Turing test would that
mean that human thought is algorthmic?
Or would it mean that the program was cleverly designed to
exhibit the same features?
The model is not the thing ... it can aid understanding but should
not be consfused with a description of the actual mechanism.
Which animals (apart from some humans ) act dumb, and what are
your criteria for this dumnbness?

This message is a reply to:
 Message 9 by lpetrich, posted 05-17-2003 11:06 AM lpetrich has not replied

  
Peter
Member (Idle past 1479 days)
Posts: 2161
From: Cambridgeshire, UK.
Joined: 02-05-2002


Message 13 of 31 (40626)
05-19-2003 6:22 AM
Reply to: Message 8 by NeilUnreal
05-12-2003 1:57 PM


If I build a soft-nueral network with sufficient
internal connections, and supply it with sufficient
training data I can match pretty much any input-output
relationship.
Does that mean that everything operates via a nueral network?

This message is a reply to:
 Message 8 by NeilUnreal, posted 05-12-2003 1:57 PM NeilUnreal has not replied

  
NeilUnreal
Inactive Member


Message 14 of 31 (40659)
05-19-2003 1:11 PM


Peter-
Good points and questions. I want to take some time to formulate a good reply.
-Neil

  
lpetrich
Inactive Member


Message 15 of 31 (40695)
05-19-2003 6:39 PM


Burden of Proof?
It seems to me that Peter is trying to shift the burden of proof onto whomever claims that most animals do not have human-scale intelligence or something similar. It is as if Peter is following in the footsteps of George Romanes and his book Animal Intelligence.
Which has been remembered as how not to do animal-behavior research.
I think that the burden of proof should be otherwise. One ought to prefer hypotheses of relatively simple mechanisms without conscious thought unless there is some good reason to believe otherwise. And I reach that conclusion because of the great success of hypotheses of consciousness-free mechanisms, and also out of concern that conscious thought is being used as a sort of deus ex machina, a rabbit out of a hat that can explain essentially anything.
As an example, let us consider chimpanzee "insight learning", first observed by Wolfang Koehler some decades back. Chimps would pause and then quickly attempt to implement some solution to the inaccessible-banana problem, like stacking crates. The pause-and-implementation sequence suggests that the implementation was being planned during the pause, which in turn suggests that chimps have some mental-modeling ability. Much like our species, even if not as well-developed.
However, most other species do not have that sort of problem-solving ability, at least as far as can be determined. Thus, it is unlikely that spiders, ants, bees, and beavers build their well-known structures with the help of mental models, and the burden of proof ought to be on those who claim otherwise.
The closest human analogy to their behavior is certain jokes about following a few simple rules in order to survive in unfamiliar social environments. Such as surviving in the Navy:
If it moves, salute it.
If it doesn't, paint it.

Replies to this message:
 Message 16 by Peter, posted 05-21-2003 9:49 AM lpetrich has not replied

  
Newer Topic | Older Topic
Jump to:


Copyright 2001-2023 by EvC Forum, All Rights Reserved

™ Version 4.2
Innovative software from Qwixotic © 2024