B
- Status
B
Ben_Shipman
Guest
Dear Professor Feser,
A. Any artifact, including the human brain, could in principle be running any abstract program, at varying degrees of defectiveness. That is, Programmed Behavior - Actual Behavior = Defectiveness. Since only actual behavior is observable, we have a single equation in the two unknowns (Programmed Behavior and Defectiveness), so we cannot solve for either of them. Hence, we cannot know the program by observing the artifact.
B. We CAN know the program of an artifact by consulting its (human) programmer.
C. By knowing the program of the artifact, we know some of the intentions of the programmer.
D. If computationalism is true, then our consultation of the programmer constituted an observation of his brain.
E. If computationalism is true, then the intentions of the programmer constitute information about the program his brain is running.
F. If computationalism is true, then by B, C, D, and E, we have that we derived information about the programming of the programmer’s brain by observing that same brain. By A, this is impossible, so computationalism must be false.
In case KaLb, if it is granted that we “know” the relevant laws of physics, we can examine the structure of the artifact and derive rules that it is bound to follow*, hence its “program”.
Code:
I am sorry for the long delay in our correspondence. In returning to the problem, I'll state my purpose at the outset. I am trying to argue that the functional power of human intelligence might be artificially replicated - that this is at least a prospect worth taking seriously.
I had prepared a careful comparison of the capabilities of humans and computers to do addition, and an argument that the deficiencies of the latter relative to the former will disappear in the future. I will share it with you if you like.
But I fear I overlooked the essence of your contention in "Kripke contra computationalism". It seems to me now that you make the following argument:B. We CAN know the program of an artifact by consulting its (human) programmer.
C. By knowing the program of the artifact, we know some of the intentions of the programmer.
D. If computationalism is true, then our consultation of the programmer constituted an observation of his brain.
E. If computationalism is true, then the intentions of the programmer constitute information about the program his brain is running.
F. If computationalism is true, then by B, C, D, and E, we have that we derived information about the programming of the programmer’s brain by observing that same brain. By A, this is impossible, so computationalism must be false.
Code:
I'll grant for now that C, D, E, and F are valid. The fatal flaw in the argument is that A and B cannot both be true, when defining the words "know" and "program" consistently. My argument is as follows:
By the word "know", we might mean either "to have mathematical certainty" (we can label this definition Ka), or merely "to have high and justified confidence" (definition Kb). Likewise, by "program" we might mean either "the rules which the programmer of an artifact intended it to follow" (definition La), or "rules which explain the actual behavior of an artifact" (definition Lb).
Proposition A is true only if we use the former definitions for both terms (case KaLa). The argument is as follows:If we use definition Kb, then we can know the program (in senses of definitions La and Lb) by examining the structure of the artifact and using the reasoning below:
We want to find the simplest (most likely) explanation for the artifact. For instance, if the artifact resembles a computer, except that it trips the circuit-breaker when we plug it in, then we might decide between three explanations:
M. That the artifact, by some (necessarily highly complex) process, arose from materials more typically found in nature, without conscious agency.
N. That the artifact was constructed by humans in order to function as a computer, and by some (perhaps very simple) process developed an electrical fault.
O. That the artifact was constructed by humans in order to trip circuit breakers, yet structurally resemble a computer.
N. That the artifact was constructed by humans in order to function as a computer, and by some (perhaps very simple) process developed an electrical fault.
O. That the artifact was constructed by humans in order to trip circuit breakers, yet structurally resemble a computer.
Given our prior knowledge about humans, N qualifies as a simple explanation, since humans have simple reasons to construct computers. O is not a simple explanation, since circumstances motivating humans to behave in such a way would bear considerable explanation. For M to occur would require so many independent events that any explanation for those events would run to extreme length, indicating the correspondingly extreme unlikelihood of the scenario. The same reasoning applies when considering a computer running any specific code - the code may have bugs, but almost never so many as to make incorrect hypotheses about the programmer’s basic intent statistically competitive with correct ones.
B
Ben_Shipman
Guest
That leaves, as promised, only case KaLa. But though A holds in this case, B cannot, unless we sacrifice A. The argument follows:
We could neither know the structure, nor discern the function, of an artifact in the absolute sense at the quantum level. But this is nothing special about brains - Heisenberg’s uncertainty principle applies to everything. At most, this allows free will to be defended against claims of physical determinism*.
**Unless, à la Roger Penrose, you claim that the brain performs quantum computation. In that case, thoughts might be physical but not fully observable. It should be noted that there is evidence against this theory - in biological systems, quantum coherence breaks down at a scale seemingly much smaller than the neural networks known to do information processing.
Sincerely,
Ben Shipman
P.S. Whether brain computation can explain, on one hand, the programmer’s mind - in the sense of an agent that observes events and produces actions, and on the other, his qualia, are notably distinct questions. My understanding of qualia is that - while they may be coupled very precisely to physical phenomena - they do not produce observable effects outside of the known laws of physics, hence intelligence as an observable phenomenon ought to be possible to analyze and replicate in the light of those laws. On the other hand, if the resulting entities lack qualia, then they arguably lack intrinsic worth. Hence the question of qualia may be at once morally crucial and practically irrelevant in regard to the issue of artificial intelligence.
There are at least two obstacles to knowing the programmer’s intentions for the artifact by questioning him. Firstly, he could be lying about those intentions. Secondly, and more importantly, he could be misrepresenting them unintentionally. A computational view of the programmer allows both of these possibilities to be expressed formally (though they exist in any case):
Let 1 be the programmer, and let 2 be the artifact he programs. Let PB[1] be the programmed behavior of the programmer, which we might idealize as his intentions. Let PB[1[2]] be the programmed behavior of the programmer for programming 2 (his ‘intentions’ for programming 2). Define PB[2] = C[PB[1[2]]], where PB[2] is the programmed behavior of 2, and C is a function that maps the ideal programming behavior of a programmer upon an artifact to the ideal behavior of the artifact once subjected to that programming behavior.
Code:
We want to find PB[2], and it will suffice for that purpose to find PB[1[2]]. Let Q[PB[1[2]]] be a query as to the value of PB[1[2]], and let PB[1[Q[PB[1[2]]]]] be the intended behavior of 1 in response to that query. Let F[1[Q[PB[1[2]]]]] be the value by which 1 intends his answer to differ from the sought value PB[1[2]]. Then PB[1[Q[PB[1[2]]]]] = S[PB[1[2]] + F[1[Q[PB[1[2]]]]]], where S is a function (assumed invertible) that maps information intended to be communicated to speech encoding that information.
But we are not done. Let AB[1] be the actual behavior of the programmer, and define D[1] = PB[1] - AB[1] as the defectiveness of the programmer. Then the programmer's actual response to your query will be AB[1[Q[PB[1[2]]]]] = S[PB[1[2]] + F[1[Q[PB[1[2]]]]]] - D[1[Q[PB[1[2]]]]], where D[1[Q[PB[1[2]]]]] is the value by which the programmer's intended response differs from his actual response.
If we know that the programmer is honest, forthcoming, and non-defective, then we have F[1[Q[PB[1[2]]]]] = 0 and D[1[Q[PB[1[2]]]]] = 0, so we can take S^-1[AB[1[Q[PB[1[2]]]]]] and get PB[1[2]]. But of course we cannot "know" such a thing in the absolute sense by talking to him.Now suppose that we could discern the programmer’s honesty by examining his brain. This already requires that the information content of the programmer’s thoughts exist in his brain. But that determines only the value of his falsehood, F[1[Q[PB[1[2]]]]]. To determine the programmer’s defectiveness (or likewise, his programmed behavior) we have to account not only for the internal inconsistencies of his neural structure, but for the inconsistency between that structure and the standard which we use to define his programmed behavior in the first place, which we call the programmer’s programmer. If that standard is something we can’t measure by examining the programmer, like his immaterial soul, or God, then we are stuck. If we instead grant that the programmer’s brain is its own programmer, we eliminate the problem of external inconsistency. But not only does this case (call it KaLa intersect KaLb) seem to identify the character of the programmer with the characteristics of his brain - thus granting the premises of computationalism - it denies A, since it defines the programmed behavior of his brain as something observable of his brain!
Code:
That concludes the argument. But a general feature should be noted. Premise A does not hold for cases where the intentions underlying an artifact's programming are invested in the artifact itself. Since computationalism holds this scenario to be realized in regard to the brain, one cannot begin an argument against computationalism by presuming its impossibility.**Unless, à la Roger Penrose, you claim that the brain performs quantum computation. In that case, thoughts might be physical but not fully observable. It should be noted that there is evidence against this theory - in biological systems, quantum coherence breaks down at a scale seemingly much smaller than the neural networks known to do information processing.
Sincerely,
Ben Shipman
P.S. Whether brain computation can explain, on one hand, the programmer’s mind - in the sense of an agent that observes events and produces actions, and on the other, his qualia, are notably distinct questions. My understanding of qualia is that - while they may be coupled very precisely to physical phenomena - they do not produce observable effects outside of the known laws of physics, hence intelligence as an observable phenomenon ought to be possible to analyze and replicate in the light of those laws. On the other hand, if the resulting entities lack qualia, then they arguably lack intrinsic worth. Hence the question of qualia may be at once morally crucial and practically irrelevant in regard to the issue of artificial intelligence.
M
Michael19682
Guest
Let me state from the outset that I do not fully comprehend the context of the remarks in bold italics underlined. However, to me there is a third obstacle that is left out of your conclusions and that would confound even the neural test if the world were paranoid enough.
The programmer might be telling the full truth without deception and totally straightforward. The subject [perceiving] however might not believe the truth or as you specify, intentions. The Apostles had all fit this bill to a Tee. In your example, without a 100 percent truth to compare the brain image results to, even that neural test would fail to persuade on the basis of no reliable correlation. and so goes the case against even clear examples of faith and miracle…, both of which are among the hallmarks of [an] immaterial intellect.
what do you say – overall intent of your remarks aside?
M
Michael19682
Guest
@Ben,
Also, a kind of piece wise remark here with regard to addition. The operation of addition is likewise a psycho-perceptual phenomenon. With regard to the principles in the letter you wrote, take any whole number: say, a group of 7 apples. One can create such a group by various combinations taken 2 at a time. 7+0, 6+1, 5+2, 4+3. Depending on who is doing the questioning and what his mental conditioning is, “How do you create that group of 7 apples?”
Honest answer: “I simply took 7 apples from the bushel and put them down on the table.”
Yet the interrogator remains convinced that this is false. He is sure it was by taking 3 apples from the bushel, then returning to the bushel a few hours later to take the other 4 and put them on the table next to the 3. In fact he draws a jury to trial and convict his mark for lying. The man responds in his self defense by noting various perceptual illusions, and also something akin to the Stroop Effect. He claims his interrogator is so in the custom of seeing the numbers 4 and 3 together – because he lives on 43 Raisin Boulevard, that he automatically assumed that was the way (4+3) the group was formed.
So really, 7 is an arbitrary grouping in this case. While we may not be so deranged as to say that 4+3 != 7, there is always the theory to contend with that this is moot for a grouping that started out at face value with no need for additive appreciation. Perhaps the truth of 7 indeed holds different secrets depending on WHO sees it and HOW it is seen. In such a case, at the level of human experience, 7 does not always equal 7. If two numbers that are equal are unequal, what can be said about the possibilities for machine addition?
Also, a kind of piece wise remark here with regard to addition. The operation of addition is likewise a psycho-perceptual phenomenon. With regard to the principles in the letter you wrote, take any whole number: say, a group of 7 apples. One can create such a group by various combinations taken 2 at a time. 7+0, 6+1, 5+2, 4+3. Depending on who is doing the questioning and what his mental conditioning is, “How do you create that group of 7 apples?”
Honest answer: “I simply took 7 apples from the bushel and put them down on the table.”
Yet the interrogator remains convinced that this is false. He is sure it was by taking 3 apples from the bushel, then returning to the bushel a few hours later to take the other 4 and put them on the table next to the 3. In fact he draws a jury to trial and convict his mark for lying. The man responds in his self defense by noting various perceptual illusions, and also something akin to the Stroop Effect. He claims his interrogator is so in the custom of seeing the numbers 4 and 3 together – because he lives on 43 Raisin Boulevard, that he automatically assumed that was the way (4+3) the group was formed.
So really, 7 is an arbitrary grouping in this case. While we may not be so deranged as to say that 4+3 != 7, there is always the theory to contend with that this is moot for a grouping that started out at face value with no need for additive appreciation. Perhaps the truth of 7 indeed holds different secrets depending on WHO sees it and HOW it is seen. In such a case, at the level of human experience, 7 does not always equal 7. If two numbers that are equal are unequal, what can be said about the possibilities for machine addition?
I
inocente
Guest
Whoa
. Cartesian mind/body dualism isn’t biblical, it has nothing to do with revelation, it was invented by a philosopher. A number of Catholic posters do seem to confuse Descartes’ substance dualism with Christian tradition, but the CCC says the soul is the form of the body, not some separate, independent paranormal ghost in a machine.My bible keyword look-up gives 563 occurrences of “spirit”, but even if you quote all of them it still won’t make Descartes a latter-day Prophet.
Y
You
Guest
I can honestly say I don’t know what you are talking about. I had in mind that the creators spirit, as an immaterial thing, speaks to the mind so the mind must be accessible to immaterial things.Whoa
My bible keyword look-up gives 563 occurrences of “spirit”, but even if you quote all of them it still won’t make Descartes a latter-day Prophet.
I
inocente
Guest
I’ve debated this subject a number of times here and perhaps assumed you knew about things such as the homunculus fallacy, it comes up a lot in philosophy of mind.
Folklore varies, and isn’t a good guide to systematically thought-through philosophical positions.
In post #85 you set your belief in God in opposition to science, and belief based on gaps in knowledge is called god-of-the-gaps by theologians. That’s a fact whether you like it or not, I can’t change the history of theology.
There’s a world of difference between the on-off pulses emitted by neurons and a modulated analog signal.
As for “You still need a mind to interpret what you are seeing on the screen”, that’s kind of classic homunculus fallacy, for instance see:
en.wikipedia.org/wiki/Homunculus_argument
en.wikibooks.org/wiki/Consciousness_Studies/The_Conflict#The_homunculus_fallacy_in_philosophy_of_mind
logicallyfallacious.com/index.php/logical-fallacies/105-homunculus-fallacy
Now, please. The stickies say it’s never acceptable to question the sincerity of another’s beliefs, presumably for the obvious reason that it’s an easy way to duck and dive, and can just lead to trading insults. But I answer to Christ, not to every wannabe pope on the internet, so calm down, I get that you have a temper. Arguments are against what is said, not the person saying them, sorry if that was unclear. I look forward to a nice response, if you’d still like to talk. Thanks
.
I
inocente
Guest
:ehh:
You appear to be equating humans with Almighty God. Our minds are not His Mind.
(On the other hand, if you’re saying you can’t see any difference between Descartes’ substance dualism and Thomas Aquinas’ doctrines on the soul (the Church’s dogma), perhaps a generous Catholic would post some links for you - I wouldn’t want to get caught up in trying to teach Catholics on Church dogma, and couldn’t do a good job anyway.)
Y
You
Guest
inocente, look.
John 16:13
But when he, the Spirit of truth, is come, he will teach you all truth. For he shall not speak of himself; but what things soever he shall hear, he shall speak; and the things that are to come, he shall shew you.
He has, you would agree an immaterial mind.
His mind, he said, speaks to us and can even show us things that have not happened yet.
So you have to tell me how an immaterial mind, of his, speaks to your mind.
How does the immaterial interact with the material?
Is it possible?
I
inocente
Guest
Didn’t know you’re a fan of the KJV. More NIV myself.
Your question implies you think there is an immaterial substance of mind, along with a material substance of matter, and the two cannot interact. But there are lots of problems with that.
It would mean that your mind couldn’t control your body. It would mean your immaterial mind wouldn’t be able to see through your material eyes or hear through your ears. I guess it would also put a stop to miracles if an immaterial God couldn’t interact with his material Creation.
But a bigger problem is that if God is made of this immaterial substance, He must be formed out of that substance, and therefore the substance must preexist God, and another god must have created God.
Besides which, the scientific hypothesis that we think with our brain is morally good, since it has led to medications to help those suffering from clinical depression, schizophrenia and other illnesses, helped understand how brain injury often changes personality, how to mitigate the effects, and so on. It has brought benefits, and most Christians I know (who are mainly Catholic, as that’s where I live) would rather go to a scientifically trained doctor than a faith healer.
If some other folk have to change their notion of how we relate to God, that will be a small price to pay imho, as they must have been wrong anyway.
Y
You
Guest
Does that mean God cannot speak to your mind? You cannot hear him, in your mind?
B
Ben_Shipman
Guest
In my argument, I handle things case by case. One of the cases is that it is known by the examiner that the programmer is honest and non-defective. Another case is that it is not known. I cover both.
I also handle the case where the programmer’s brain needs to be compared to an external reference, and the case where it does not.
The issue isn’t whether the (name removed by moderator)ut of a function is determined from the output. It is trivial in the case of addition that it isn’t. The dispute over addition vs. quaddition is whether the output of a function can be truly determined from the (name removed by moderator)ut, and I address that point in my argument.
B
Ben_Shipman
Guest
I don’t see what the dispute over God’s communication with man is about.
The Divine Nature is immaterial; It is not made of matter.
Jesus’ human body is material, but it is not essential to His existence. Without it, He would still be God the Son.
Neither are our bodies, strictly speaking, essential to our existence, since without them, our souls still live.
And of course God can communicate with us, and of course he can do so by altering the material structure of our bodies. He is God, after all.
There is no problem with the immaterial communicating with the material.
The Divine Nature is immaterial; It is not made of matter.
Jesus’ human body is material, but it is not essential to His existence. Without it, He would still be God the Son.
Neither are our bodies, strictly speaking, essential to our existence, since without them, our souls still live.
And of course God can communicate with us, and of course he can do so by altering the material structure of our bodies. He is God, after all.
There is no problem with the immaterial communicating with the material.
G
GEddie
Guest
Apart from what Descartes held, the immaterial or spiritual human mind was not invented by him. It was taught by Aquinas, nay, the Church Fathers, nay even, by our LORD Himself.
Fear not those who kill the body but not the soul" (includes mind).
“This day shalt thou be with me in Paradise.”
Etc.
It is in fact, the teaching of soul-sleep, which Inocente seems to hold to, that is an innovation. It was not taught prior to the 1800s.
GOD bless and ICXC NIKA
Fear not those who kill the body but not the soul" (includes mind).
“This day shalt thou be with me in Paradise.”
Etc.
It is in fact, the teaching of soul-sleep, which Inocente seems to hold to, that is an innovation. It was not taught prior to the 1800s.
GOD bless and ICXC NIKA
B
Ben_Shipman
Guest
@GEddie
I just want to make a distinction. I believe that the human capacity for intelligent behavior is materially based, and thus replicable in an artificial context. At the same time, I believe that the human capacity to inwardly experience things is due to the immaterial soul.
Hence I think it might be possible to build a machine that has the power of human or superhuman intelligence, but which lacks a soul, and so lacks any inward experiences, though it might very much appear to have them. It would be a so-called philosophical zombie. Though I don’t know whether God would bestow a soul on an AI or not.
For clarity on this thread, I suggest maintaining a distinction between the exterior intellect, which is judged solely from how an agent interacts with its environment, and the interior intellect, which consists only of the inner experiences of the agent.
I just want to make a distinction. I believe that the human capacity for intelligent behavior is materially based, and thus replicable in an artificial context. At the same time, I believe that the human capacity to inwardly experience things is due to the immaterial soul.
Hence I think it might be possible to build a machine that has the power of human or superhuman intelligence, but which lacks a soul, and so lacks any inward experiences, though it might very much appear to have them. It would be a so-called philosophical zombie. Though I don’t know whether God would bestow a soul on an AI or not.
For clarity on this thread, I suggest maintaining a distinction between the exterior intellect, which is judged solely from how an agent interacts with its environment, and the interior intellect, which consists only of the inner experiences of the agent.
M
Michael19682
Guest
You missed my point about changing subjectivity. If the (name removed by moderator)ut itself is unavailable for determination, prior to (name removed by moderator)ut, the output is equally indeterminate.
It is akin to an endless recursion. The base case is indeterminate, thus when unwound the function has no “external” reference to anything; and thus the function always calls itself again. E.g.,
Like in a maze traversal where you reach the final impasse [of a particular route] and start popping the stack [unwinding] back to the entry point only to find that an altogether different maze has “grown” around the original entry point. The stack is now empty, but you are in the middle of maze with an uncharted entry you can’t reach. This entry point is like a randomized pointer. IT could be anywhere because it has shifted while you were traversing out into the maze.
Biological processes that incorporate growth exhibit this dilemma. Why can the effects of medicines change over time? or side effects develop that didn’t appear at the first or successive doses? It’s because the body subject changes in the interim between the first and the sampled dose in question.
B
Ben_Shipman
Guest
Let me make sure that I understand you. You are saying that the ‘program’ of a human mind cannot be determined because it has no sharply defined (name removed by moderator)ut, no external specification, and it isn’t possible to learn the person’s “program” by asking them?
M
Michael19682
Guest
No (name removed by moderator)ut that can be predicted or determined exactly, or as you say, no external specification [that can be agreed upon universally]. Asking the program what it is doesn’t work because the question itself is (name removed by moderator)ut that the existing system will “interpret” according to an unknowable internal environment in a variable manner. A mind is not a static database that is queried and produces an unbiased report, or information that is a copy of contents. People react to questions in ways that cannot be predicted or determined at all, especially over time and after the “superficial” responses have been elicited.
B
Ben_Shipman
Guest
This is a different argument against seeing the brain as hardware and the mind as software than the one Feser used, and it is a better one. I myself do not identify the interior experience of being as any abstraction of the brain, the way that software is an abstraction of the configuration of the hardware running it.
Nor do I think that in the brain the line between hardware and software is well-defined; even the number of neurons in the brain is changeable.
Nevertheless, I do suggest that the function of the brain might be stochastically predicted if the internal environment of the brain were known. You would need a computer much more complex than the brain, but in principle a computer could simulate the brain, given its atomic configuration and highly realistic boundary conditions (a virtual body and environment).
Now, I’ll grant that even a typical person might be the beneficiary of enough supernatural grace that the computer simulation of the person would fail to match their actions closely (maybe the simulation would ask why it suddenly felt empty inside). I conjecture, however, that the simulation would still display intelligence, because I believe intelligence to be a function of neural structure.
Further, an artificial brain (not a simulation of a natural brain) might learn by adapting the strength of its synaptic connections but not by growing or eliminating neurons. In the case of an artificial brain, then, one might be able to take the number and topology of its electronic components as fixed hardware, and the strength of its synaptic connections as software.
Nor do I think that in the brain the line between hardware and software is well-defined; even the number of neurons in the brain is changeable.
Nevertheless, I do suggest that the function of the brain might be stochastically predicted if the internal environment of the brain were known. You would need a computer much more complex than the brain, but in principle a computer could simulate the brain, given its atomic configuration and highly realistic boundary conditions (a virtual body and environment).
Now, I’ll grant that even a typical person might be the beneficiary of enough supernatural grace that the computer simulation of the person would fail to match their actions closely (maybe the simulation would ask why it suddenly felt empty inside). I conjecture, however, that the simulation would still display intelligence, because I believe intelligence to be a function of neural structure.
Further, an artificial brain (not a simulation of a natural brain) might learn by adapting the strength of its synaptic connections but not by growing or eliminating neurons. In the case of an artificial brain, then, one might be able to take the number and topology of its electronic components as fixed hardware, and the strength of its synaptic connections as software.
- Status