The subject of intention and action has come up few times, so I thought I’d start a thread.
From my point of view as a cognitive neuroscientist, decision-making (which action to take) is best conceived of as a kind of winner-take-all arm-wrestling competition, in which competing programs (represented as networks of active neurons) of action exert a mutually inhibitory effect on on the other, while each receives excitatory input from various other other networks, each of which in turn are engaged in a kind of subsidiary arm-wrestling match with some networks and a mutually cheer-leading match with others.
The more activation in any one network, the greater the inhibitory effect it has on competing networks, and so the system is, in a technical sense, “chaotic” – two competing programs can be finely balanced at one moment, but once one gets ahead by more than a critical amount, its inhibitory effect on the other increase, reducing its activation and releasing its reciprocal inhibitory control. At this point, activation in the winner rises rapidly towards “execution threshold” – the point at which outflow to the muscles involved in the action are activated.
Of course this is a continuously looping process, and the actions can be as slight as an eye movement, which then brings new input to the decision-making process, or a gross-motor action, which also provide new input, so the decision-making process is constantly informed by new data. However, it is also informed by endogenous processes – processes that trigger activations in networks involved in goal-setting and reward prediction, and established through life-long learning, in which neural firing patterns that result in success become more probable and those that result in failure, or penalty, become less likely.
As the brain’s owner, of course, we call these processes “pondering”, “hesitating”, “deciding”, “exploring”, “testing”, “changing my mind”, “exercising will power”, “considering the long term effects of my actions”, “considering the effect of my actions on someone else”, etc.
Which is exactly what they are. But at a neural level they operate very like evolutionary processes, in which what replicates most successfully (neurally) is most likely to be repeated, and what replicates least successfully is least likely to be repeated. The interesting part is that this “neural Darwinism” takes place prior to actions actually being performed – and often the”winning” program does not actually reach execution threshold, but instead is fed back as input, so that we are able to imagine the results of our actions before we actually execute them, and use that information before actually allowing an action to take place.
That means that we, unlike evolutionary processes, are capable of intentional action. We can simulate the results of potential courses of action, and use those simulated results to inform the decision-making process. This allows us to take shortcuts, and pursue, in actuality, only those courses of action we deem likely to be successful. In contrast, evolution is stuck with trying anything that presents itself as an option, learning by actual, not simulated, errors. It cannot be said, therefore, to exhibit intentional behaviour, and is much slower and less efficient that we are. However, by the same token, it will often explore possibilities that a simulating – intentional – agent would reject, on the grounds that the simulations looked unpromising. As a result, some spectacular solutions are missed.
Which is why evolutionary algorithms are used by intentional designers – us – so that we can, intentionally, use the power of unintentional design to find solutions we ourselves would reject as not sufficiently promising to explore.
Llanitedave & pertrushka express very well what I mean by free will. Basically: I have free will whenever I am free to act upon my desires, in the widest sense.
And just in case you haven’t noticed it: this is not a semantic issue, as you seem to think, this is a fundamental question about the function of this libertarian free will that you think you have: you say that you need this libertarian free will to evaluate the perception, processing and responding to experiences produced by the *biological machine*. That means that the mechanism by which your libertarian free will comes by its perception, processing and responding would need to be independent from the *biological machine*. Otherwise it could not actually evaluate and arbit them outside of the reference frame of the machine itself. Or maybe there is a way in which it could. That would be very fascinating. Either way, in order to give any credence to the claim that this libertarian free will of yours does anything of consequence, you will have to explain how it accomplishes these feats of perception, processing and response (and a nice bonus would be to present some factual evidence that it actually can do the things you claim it does).
Ok, I can’t resist. This made my day.
Free will just as a general concept seems difficult enough to pin down. I don’t know that further proposals such as “libertarian” free will provide any added value. They certainly don’t add any rigor to the definition, if one exists.
Lizzie, this is an excellent OP. I have been thinking about what the word ‘want’ means from a cognition point of view. What do desires look like as a chain reaction or schematic view?
And lo, Moses came down from the mountain, Commandments in hand.
“I have here a set of guidelines … well, more like rules actually. Break these and God will be SERIOUSLY pissed. Of course, you have complete free will. You can break ’em or not; it is an entirely free choice, uninfluenced by any ‘programming’ you may consider yourself subject to in your present physical embodiment. So, how you gonna exercise your free will, punk? Do ya feel lucky? Well, do ya?”.
Quite so. We also have concepts like bi-polar disorder and obsessive-compulsive, which I think provide a more practical way of discussing the sliding scale of one’s ability to analyze and make choices.
Ironically, I see William (and most creationists) being excessively limited in his ability to make choices because his landscape of options is so narrow and devoid.
Bingo. Well said.
Most certainly the mind uses an evolutionary algorithm. Using computers is merely and agumentation or extension of our thought processess.
To characterize the design of a GA as “unintentional” is probably a stretch. It searches for solutions based on the intelligent designer’s goals. The designer defines what constitutes a solution.
What is perceived as designed is deeply subjective and are so are design solutions. But despite the fact it is deeply subjective, in principle algorithms can be created (albeit clumsily) to create objects that we would recognize as designed. But the algorithm cannot construct these subjectively perceived artifacts unless it has specialized knowledge of what humans will affirm as designed.
To illustrate, say I wanted to build a Genetic Algorithm that could compose piano music. We can tie such a GA to either a mechanical or electric piano. The GA will have the freedom to explore melodies and harmonies. But unless we give it sufficient heuristic rules, it will not make anything resembling music. The more expertise we give the GA the more constructs it can make that might be perceived as designed music versus a baby pounding on the keys. This illustrates intentionality cannot be divorced from the GA for the very reason it is an extension of human conceptions of what constitutes design.
The problem of GA generating music is the same problem of GA generating structures that humans perceive as engineered artifacts. The GA (if that’s how life was created) would have to be loaded with the ability to construct objects that humans would recognized as designed. Suffice to say, the GA that made life would probably have to be much more capable than the GA that can compose symphonies.
Even though designs are subjectively perceived, the subjective perceptions still put constraints on how a GA can consturct such subjectively perceived designs. And those constraints express intentionality, not random accidents.
So how are you defining “intentional”, there, stcordova?
(And welcome to TSZ :))
stcordova:
Why? What do the perceptions of this particular twig on the Tree/Bush/Net of life have to do with the matter?
A Genetic Algorithm is a computer-based method of mimicking what nature does – produce random variation, apply some selection criterion, and replicate the surviviors. Nature does not dance to our tune; we ape it. The fact that certain individuals perceive life as designed does not make it necessary for the ‘algorithm’ that produced them to be loaded with the capacity to produce them (and the many and varied other forms that have an equal claim to being the current ‘end-result’ of the algorithm). Such ‘designoid’ points obviously exist within the ‘space of possible organisms’. Evolutionary theory would have it that the current points occupied by real organisms in that space are accessible, starting from the initial replicators, by serial amendment. You don’t have to front-load the algorithm, it just has to have a navigable search space. An algorithm front-loaded with the capacity to find every one of the 10-30 million current species would be equivalent to suggesting that every published novel was ‘front-loaded’ into the basic string ABCDEFGHIJKLMNOPQRSTUVWXYZ. (Yes, I’m aware that those novels are the product of Intelligent Authorship).
I think you confuse the roles of processing and raw data (input/output) in an algorithmic context. The ‘algorithm’ is loaded only with the ability to identify survivors from among the data points presented to it. The sole ‘goal’ of the Life Algorithm (as with any generational set in the GA) is to solve that immediate problem.
No, that isn’t correct. I have seen genetic algorithms that design antennas and that is something nature could not do.
But anyway if you had any evidence that nature could produce a genetic algorithm, you should publish it.
Oh, lordy! The mimicry is in using populations of digital ‘organisms’ and applying mutation and selection criteria, not in the particular application.
Hahahaha. You’re a card. No, I don’t have evidence that nature can produce “a computer-based method of mimicking what nature does”. Whatever makes you think I said it could?
Oh Lordy! That is question-begging.
Oh Lordy, Lordy! I didn't say that.
.
No it ain’t. That’s what GAs do – they use populations of digital ‘organisms’ and apply mutation and selection criteria. No questions are begged during their construction. Whether or not Nature works that way, computer-based GAs do.
Your own gift for mimicry aside, yes you did. “But anyway if you had any evidence that nature could produce a genetic algorithm, you should publish it.” As I had already stated a genetic algorithm to be a computer-based analogue of an (assumed) natural system (for which I have the support of no less an authority than Professor Wikipedia), then challenging me to produce evidence that nature can produce such is clearly nonsensical. You may not have meant that, but it is what you said.
Oh Lordy! It is question-begging because you have no idea, nor any evidence, that shows nature can produce a reproducing organism.
There ya go assuming that which needs to be demonstrated.
Robin:
LoL! You “choose” your position despite the lack of supporting evidence. IOW it is you who is being excessively limited in your ability to make choices.
Thank you for the welcome and the hospitality.
I really didn’t define it, like many things in the realm of inelligence, somethings are left as undefined terms. Even in math, the fundamental basis of many mathematical systems, there exist undefined terms. But not defining something doesn’t mean we can’t enumerate important characteristics.
Intent can be characterized by wants and desires or perhaps even needs. In real life we often have conflicting wants and desires. In the engineering world we call these “conflicting requirements”. For example we may want a battery to hold a lot of charge but we would like it to be light as a feather. The wants conflict. In the same way, intentions (based on wants) can conflict. In the case of conflicting intentions, it is hard to define what someone’s real intentions are since he may not know himself. I found that to be the case when my customers in the engineering world were providing me with conflicting (but well meaning) intentions. This problem of conflicting intents is most obvious to me in the investment world where risk often conflicts with reward.
Now with respect to GA or other examples of weak Artificial Intelligence, they are basically filters and in some cases conflict resolution mechanisms. The term that is sometimes used is “constraint propagators”. It is fair to say, they are basically glorified search algorithms exploring a space of possibilities.
Perhaps the most amusing example of how difficult it is to define intent was in the development of radom number generators in computer science. A certain element of capriciousness is desired in a random number generator. When these things were first designed, the generators gave some undesirable results. But then, I could poke fun a those pioneering engineers by saying, “isn’t capriciousness what you intended in the first place. It’s giving you what you intended because it gave you something you didn’t intend!” [the enigma was highlighted in Dembski’s essay “Randomness by Design”]
You’re such a nice person, I hate to offer a characterization that somewhat disagrees with this claim. It has some relation to the problem of designing random number generators that are capricious but not too capricious.
In searching for solutions we can, in principle, go through an exahustive approach by trying every possibility. But in practice this cannot be done for every problem because we have finite resources to solve the problem. If a problem is amenable to educated guesses without making an exhaustive search, then the genetic algorithm can be a cure. In formal language of computer science, these problems are often associated with NP-complete (non-polynomial time complete) problems. That is to say we can find adequate solutions to a problem without an exhaustive search.
But there is a subtlety. It has to do with whether there are right and wrong answers to a problem to begin with. For example, what defines music. It is hard give a right or wrong answer, but certainly we have intuitions on the matter!
In certain cases, there are right and wrong answers, or sufficiently good answers for an algorithm to solve. For example there maybe optimal solutions for the travelling salesman problem. A GA can arrive a the mathematical correct solution because we know in advance the structure of the problem and can thus front load the GA with this knowledge to help it make an educated guess.
But then there are open problems in Artificial Intellgence where it is hard to define “right” and “wrong”, and yet, it seems we have some notion of right and wrong. Examples are: algorithms that can make music, poetry, novels, architectures. We can eliminate what we define to be wrong (such as misspelling or violations of physics), but it is hard to define what is right, and then we have the notion that something are more “right” than others.
The space of designs allows for solutions where
1.”right” and “wrong” can be defined,
2. where the notions of “right” and “wrong” are very blurred.
Designed artifacts are found in both realms.
I’ll never forget that I just happened to have the TV on and Meditation from Thais by Massenet was playing. It seemed so different from many things that I heard until that point in my life. I thought to myself, “the composer must be a genius”. Time has borne out that many have viewed that piece as a classic, or genius if you will. But like many pieces of music, it is hard to define why it seemed “right”. And to answer your question about intent, in like manner, it is hard to define what “intent” really means, we can only grope for incomplete definitions.
GAs work. Do you believe that someone copied a mechanism that isn’t actually there / doesn’t actually work and stumbled across a method for finding (locally) optimal solutions? That’s up there with finding Titaalik by accident.
There is no need for an evolutionary algorithm to know what is right and what is wrong. There is only a need for variation and differential reproductive success. Evolution is not a search. When you characterize it as a search you are drawing bulls eyes after the arrow has landed.
As long as variation is possible (as long as small variations are not universally fatal) there will be change. For ID to be taken seriously it needs to demonstrate not just that gaps exist, but that there is no possible path linking cousins to a common ancestor. In other words, you need to prove that Thornton is wrong.
That’s really the heart of the argument. GAs are modeled after biological evolution. they don’t work on every problem,. They can’t be used to break ciphers. but they can explore connected gradients.
Now if biologically significant sequences are connectable, evolution will work. If they are cipher keys, the goddidit.
That is true, but it doesn’t solve the problem of the appearance of design in biology. Biological organisms are structured in ways that fit recognizable templates in the realm of engineering and computer science.
It matches templates that engineers use to satisfy the intentions of their customers or themselves.
You claim that life evolves, fine. Why does it have to exist at all starting from the origin of life? Why is there a genetic algorithm in nature in the first place?
One does not formally have to prove intentionality and purpose to recognize the fact that biological organisms are easy to characterize by templates that engineers use to satisfy customer intentions. We see the appearance of communication systems, feedback control, error correction, computation, information processing, compilers, artificial intelligence, real intelligence, sensory devices, storage devices, rube-goldberg machines, lock-key systems, login-password systems. We can’t project these templates onto random strings of numbers or piles of dirt.
The problem is why biological organisms evolved in ways that look like engineered artifacts. Random chance doesn’t produce these coincidences, and alogorithms will not produce these coincidences unless they are sufficiently front-loaded with intent, i.e. music and full-blown computers and living systems are not produced by algorithms unless the intentions are programmed into the algorithm in the first place.
For every GA that you can build that finds a designed solution, I can find substantially more that won’t! So which example of GA’s is more representative of real-life biology? GA’s that find solutions or GA’s that don’t? Answer: GA’s that don’t, i.e. Gammarus Minus and anti-biotic resistance.
Yes, I know GAs work. I also know that GAs do not have anything to do with nature.
As for Tiktaalik, I will go with what Shubin said and according to Shubin he did NOT find what he was looking for.
Except GAs are not modled after biological evolution. Ya see in order to model something you have to understand it.
LoL! In order for your position to be taken seriously it needs some positivce evidence to support it.
Good luck with that…
Oh. So where did the idea for genetic algorithms come from, then?
If you want to claim that insects are unnatural, you should support the claim.
It’s hardly necessary to publish the observation that genetic algorithms are part of the material world.
stcordova,
Biological organisms don’t look like “engineered artifacts”.
We humans use analogies to explain what we see and we’ve done that with biology.
Turning around and claiming our analogies now “equate” to those things we are describing is very misleading.
ID also claims “specification” as a characteristic of design that can be determined without referencing the intentions of the designer.
If I order two identical cars, one red and one blue, and I receive two blue ones, which one one was “specified” and which one wasn’t?
You can’t ask me, the “designer”, but you can look at the engineered artifacts, the cars themselves.
An interesting neuroscience and psychological experiment or observation worth considering. Among my colleagues, there seems to be a rather strong appreciation or ability regarding music. In my experience, the Darwinists least hostile to me have had keen musical abiliyt (like Allen MacNeill).
I mentioned Massenet’s music as an instance of intelligent design. I pointed out the substantial difficulty of developing an algorithm that could create such open ended designs as music. Below is a video of a performance of Massenet’s music. Life has intricacies which go far beyond this musical work (since life itself made the work possible). How can an evolutionary algorithm generate such things unless it is front loaded with sufficient information? At some point if an algorithm has sufficient front-loading, evolution is indistinguishable from creation!
If evolutionary algorithms are hard pressed to make works of art like this, how much more difficult will evolutionary algorithms have in making living systems capable of composing and performing music?
The idea for genetic algorithms came from humans- specifically humans who wanted to solve a problem.
drwho:
I claim that insects are natural in that they exist in nature. If YOU want to claim that nature produced insects, YOU have to support the claim. If someone wants to claim that insects evolved from some brine shrimp-like organisms via accumulations of random mutations, they have to support it.
Yes they are, just as the design is part of the material world. I knew you would understand.
stcordova,
The question you are asking applies to the designer as well.
To say that evolution can’t explain music is as valid as asking how the “designer” knew that Elvis should explode on the music scene in ’56 and Jimi Hendrix in ’66.
If evolution needs to explain “music”, you designer needs explain the timely arrival of Hendrix.
toronto:
Perhaps not to you. But regardless of that your position still can’t explain their existence.
Who cares what it looks like to you, produce a testable hypothesis and test it.
Hey ID explains the timely arrival of intelligent observers on this planet- just when the moon was in the right position to cause scientifically valuable total solar ecplipses:
Joe G,
“Specification”, a needed requirement of ID, requires intent.
Your position doesn’t explain the intent of your designer, leaving the ID “specification” argument useless.
Joe G,
The “The Privileged Planet” fine-tuning argument is one I could never understand.
If I pass a playground and see a small 8 year-old and a large 10 year-old on a teeter-totter I can claim that due to the finely balanced device, that the teeter-totter must have been fine-tuned for them.
I then watch as they switch positions, and amazingly, they are just as well balanced.
Did someone fine-tune the teeter-totter again while they were changing sides?
This is why the fine-tuning argument is so bad.
Systems stabilize due to the static and dynamic characteristics of the components whether they are children playing or planets in orbit.
Geez we don’t know the intent of the designer’s of Stonehenge. However we can tell it was intentionally designed and constructed.
Yes Toronto, I am sure that you cannot understand “The Privileged Planet”. But then again you think this all “just happened” so what does it matter?
That’s easy. Nature is the only thing known to have produced anything.
Of course. Intelligent design is an observable biological phenomenon.
For all of you so engrossed with evolution-
Engineers have used trial and error for quite some time. Thomas Edison was proud of his “99% perspiration 1% inspiration”- what GAs allow us to do is get rid of the perspiration and have a computer run all the T&E until a suitable solution is found.
The evo-tunnel vision is showing….
Really? Science by bald declaration, as usual. Unfortunately nature has not been known to produce a living organism.
Yes, it is.
Joe G,
What matters is that “you” cannot understand the fine-tuning argument either.
I’d expect someone who claims to understand the ID position to have the answer to the fine-tuning argument ready, since a dynamic “stabilized system” type of question should have been expected.
No, what matters is that your position doesn’t have anything so you are forced to attack ID.
I find it entertaining.
I also find it entertaining that you change the subject from eclipses to fine-tuning
Yeah, by design.
In what way is it finely balanced? If I see a large kid on one end and a small kid on the other, if the fulcrum is centered, I would expect the large kid to be seated on the ground and the small kid to be propped up in the air. There wouldn’t be any balancing.
What, other than nature, is known to exist? There’s no bald declaration. Whenever we’ve found causes and explanations for things, they’ve always been natural.
You’re not one of these people who seriously blames the gremlins when your car breaks down, are you?
So nature produced nature? Unfortunately science says nature had a beginning.
All you have are bald declarations.
Or artificial.
No, I am more than capable of keeping my car running.
[audio src="http://ersimages.com/ecards/beet2.mp3" /]
[audio src="http://ersimages.com/ecards/invention.mp3" /]
We can assert that a certain mechanims (like GA’s or evolution) cannot reasonably arrive at certain designs without sufficient front loading. Whether that means there is an intelligent designer or not is a separate question.
An intelligent designer would be a sufficient explanation (assuming he exists in the first place), whether he is a necessary explanation is the point of contention. If it is a necessary explanation, then biological systems need an intelligent designer, and the intelligent designer exists or existed once upon a time.
You can detect design without knowing why it exists at the time it exists. There is no need for such explanations to infer design — numerous archaelogical exist for which we have no access to the INTENT of the design. Knowing intent is a sufficient but not necessary condition to recognize design. However “unknown” designs (like stonehenge), mimic patterns of artifacts that are produced with known intention, and that is how we recognize design.
I recognized great works of music sometimes on the first hearing without even realizing it was written by a great composer. I did not have to know the designer to know these works were designed by a great mind.
When engineers build communication systems, there are senders and receivers. Most system require substantial engineering of both sender and receiver. The ability to compose (send) and recognize (receive) great music is very much analogous to the ability to engineer senders and receivers. Evolution must be sufficiently front loaded to make such a communication system possible. Natural selection ought to select against such metabolically expensive rube-goldberg machines of life, and empirically speaking, it appears it will since we are expected to go extinct quite soon relative to bacteria or the lowly cockroach.
And that’s how GA’s work in the biological world versus in the imaginations of evolutionists. Natural selection is notorious for destroying designs, not creating them. Given that’s the behavior of GA’s in biology implies a mechanism other than GA’s created the salient features of life.
Joe G,
Science students will find ID entertaining too.
Well, yeah- science students find science very entertaining
stcordova,
If you accept as a hallmark of design is that it “appears” engineered by human standards, then you cannot discount evidence against it that also applies to human standards.
One of the hallmarks of human design is timing, which appears as complete designs and also its sub-components.
If the designer does not know when to introduce a biological artifact, the future he is planning will not “come to pass”.
IF…. you have a designer who “cannot” see the future, front-loading is impossible.
In order to show the designer is in “control”, you have to show evidence of the ability to forecast future events, before the fact.
If the designer does nothing for future biological planning, that is evidence that he was not responsible for the biology we see.
I hope you understand my argument now.
No foresight means no ability to design “specific” biological information.