Any female here?
-
@Heinz Doofenshmirtz said:
She's either a psychologist, a linguist, or in a very closely related field I'd guess...

That's a correct assumption. Her degree is in computer linguistics.
Prisoner's dilemma anyone? :D
As regarding to multi-tools or relationships? Sounds like a pretty kinky relationship... :D
Computer linguistics? Interesting! You're gonna have to clarify though... programming languages? Simulation and modeling of real languages? AI? Communications and networking protocols?
Prisoner's Dliemma; no kink... just decision making / game-theory.
-
Prisoner's Dilemma -- or Darwinian Altruism?
"Why can't we just be friends?"
-
When I studied a bit of computational linguistics it was related to parsing natural language, modeling semantics, grammar, etc, more than for example the compilers guys who were all about grammars of programming languages. It's a very interesting field. Although it seemed that the information retrieval people had made more progress than the people trying to parse language semantically. Really fascinating though.
-
Say what ?
-
@Heinz Doofenshmirtz said:
Computer linguistics? Interesting! You're gonna have to clarify though... programming languages? Simulation and modeling of real languages? AI? Communications and networking protocols?Prisoner's Dliemma; no kink... just decision making / game-theory.
I think the correct term in English is computational linguistics when I think about it an extra second. She did her master's level thesis on machine translation of compound words, using web search engines, actually. (Too bad the APIs now are so limited that it's not possible to do that kind of research from outside of the web search companies anymore.) So in other words, working on natural languages using machines. These days she doesn't get to use her education too much in her work, but linguistics still her passion and hobby. (And steel is neither my education nor my work, I just know a very friendly knife maker and metallurgist (PhD in tool steels) who has taught me enough to form prejudices. ;) )
I am of course familiar with Prisoner's Dilemma, just a bad attempt at humour. :) I like a way of summing up morality by way of that dilemma I once read: The goal of a system of morality is to make strangers choose the risky/trusting option in the Prisoner's Dilemma. In other words, make people assume a certain level of trustworthiness from strangers, thus creating a better society.
-
@user24 said:
When I studied a bit of computational linguistics it was related to parsing natural language, modeling semantics, grammar, etc, more than for example the compilers guys who were all about grammars of programming languages. It's a very interesting field. Although it seemed that the information retrieval people had made more progress than the people trying to parse language semantically. Really fascinating though.That observation is in my opinion one of the best examples of Dijkstra's truism about computers and thinking:
“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.”
We can do way more with computers by letting them do what they are good at, than trying to make them imitate us, humans, where we excel. Of course, this is slowly changing as transistor count skyrockets and we can use more and more algorithms reminiscent of how biological systems operate. Still, a good human chess player operates by pattern recognition, and is now routinely beaten by computers that map all possible outcomes sufficiently many steps ahead. The human uses its strengths, the computer is programmed to leverage what the computer is good at.
-
How did we get to here from asking whether ladies like knives?
How did we get to the moon from standing up on two feet?
Rock on, compadres/commadres ! -
When I studied a bit of computational linguistics it was related to parsing natural language, modeling semantics, grammar, etc, more than for example the compilers guys who were all about grammars of programming languages. It's a very interesting field. Although it seemed that the information retrieval people had made more progress than the people trying to parse language semantically. Really fascinating though.
That observation is in my opinion one of the best examples of Dijkstra's truism about computers and thinking:
“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.”
We can do way more with computers by letting them do what they are good at, than trying to make them imitate us, humans, where we excel. Of course, this is slowly changing as transistor count skyrockets and we can use more and more algorithms reminiscent of how biological systems operate. Still, a good human chess player operates by pattern recognition, and is now routinely beaten by computers that map all possible outcomes sufficiently many steps ahead. The human uses its strengths, the computer is programmed to leverage what the computer is good at.
I agree pretty much completely with your sentiment here.
However, I object to the ways that computer chess programs are pitted against humans, particularly notable ones, like Kasparov. I personally think the IBM Deep Blue team cheated, because they programmed all of his known recorded games into their system, and then refused to let him play a full tournament set of games against it. They also refused to let him see the system in action before the match.
Of course we know that it was more or less a standoff (if you look at the actual record of win/loss/draw, and not the spin the CS/AI people put on it), but I also think the reason the IBM people wouldn't let him play a full 21 game tournament against it is they were afraid he'd figure out it's tendencies, adapt, and end up beating it. I also think the fault is partly Kasparov's because he was arrogant enough to assume their system couldn't beat him, and to agree to their one-sided terms.
This goes exactly with what you're saying about computers and people doing what they do best though. Computers by definition can only work on "well defined" problems, as Newell & Simon characterized them. People don't have those same limitations. We can make our own definitions and rules, unlike computers. The only thing that's even come close are the parallel distributed processing neural network systems that have been built, which themselves are far more capable than any traditional AI system ever has been, because they're modeled on how the nervous system works of course. But, even they have significant differences and limitations compared to real "wetware".
I was fortunate enough to be a philosophy minor as an undergrad at Cal Berkeley, as took Philosophy of Mind from John Searle. In my field though (cognitive and physiological psychology), the general outlook is decidedly materialist and reductionistic. I don't subscribe to the thinking = brain action theory most of my colleagues do. I subscribe to emergent property theory, that there's something greater that emerges from the complexity of brain function that goes beyond the simple sum of its actions as described by neurophysiology. I believe conscious is a real causal force in the brain and that it's not just a non-causal and epiphenomenal by product of brain function the way the vast majority of my colleagues do.
Ack! I've got lecture in 10 min. Gotta go. I'd love to ramble on about this more, but duty calls! Young minds to
indoctrinate! Er, I mean, teach! Yeah, that's it! :D -
@Steinar said:
We can do way more with computers by letting them do what they are good at, than trying to make them imitate us, humans, where we excel.Yes, I think that's very true. The bag of words / vector space approach is not particularly well grounded theoretically, and certainly isn't modeled on how humans do the job, but it plays to the strengths of the computer. It does have limits though. Even with SVMs and all the whizzy ML toys in the world, it's still very difficult to get a computer to understand the sense of a document. I still can't ask google very specific questions. But then it's optimised for keyword based document retrieval, not natural language querying. Maybe there are better systems out there.
@Heinz Doofenshmirtz said:
they're modeled on how the nervous system works of course. But, even they have significant differences and limitations compared to real "wetware".yeah, I mean ANNs are 'inspired' by real neurons, but like you say, the actual similarity is really only superficial (compare genetic algorithms, which seem to more closely model nature's way of doing things).
You studied under Searle? Impressive! (Also, another Philosopher-cum-computer-scientist? Cool!).
there's something greater that emerges from the complexity of brain function that goes beyond the simple sum of its actions as described by neurophysiology.
Speaking of Searle, this aligns closely with an argument I made against his Chinese Box thought experiment (in an undergrad philosophy essay), namely that while none of the individual components of the box 'understands', the system as a whole - the sum of the components - could be said to.
-
@Heinz Doofenshmirtz said:
I agree pretty much completely with your sentiment here.However, I object to the ways that computer chess programs are pitted against humans, particularly notable ones, like Kasparov. I personally think the IBM Deep Blue team cheated, because they programmed all of his known recorded games into their system, and then refused to let him play a full tournament set of games against it. They also refused to let him see the system in action before the match.
Of course we know that it was more or less a standoff (if you look at the actual record of win/loss/draw, and not the spin the CS/AI people put on it), but I also think the reason the IBM people wouldn't let him play a full 21 game tournament against it is they were afraid he'd figure out it's tendencies, adapt, and end up beating it. I also think the fault is partly Kasparov's because he was arrogant enough to assume their system couldn't beat him, and to agree to their one-sided terms.
Interesting, I didn't know these details of that match. I did know the terms were one-sided, but I didn't know how they were so. (Except part of the strategy was tiring out the human.)
I was fortunate enough to be a philosophy minor as an undergrad at Cal Berkeley, as took Philosophy of Mind from John Searle. In my field though (cognitive and physiological psychology), the general outlook is decidedly materialist and reductionistic. I don't subscribe to the thinking = brain action theory most of my colleagues do. I subscribe to emergent property theory, that there's something greater that emerges from the complexity of brain function that goes beyond the simple sum of its actions as described by neurophysiology. I believe conscious is a real causal force in the brain and that it's not just a non-causal and epiphenomenal by product of brain function the way the vast majority of my colleagues do.
I find this very interesting. My background is from biophysics (though I have never worked professionally in the field), so I guess it is no surprise that I'm decidedly in the materialist camp. I liked this way PZ Myers put it: “I think consciousness is a product of self-referential modeling of how decisions are made in the brain in the absence of any specific information about the mechanisms of decision-making — it’s an illusion generated by a high-level ‘theory of mind’ module that generates highly simplified, highly derived models of how brains work that also happens to be applied to our own brain.”
Warning: From here on, just uninformed, personal opinion that will probably give a pro like yourself a headache. :)
Emergent properties is a fascinating thought, though. Personally, I think it's just an illusion from the way our minds work. We are so good at pattern recognition, that when a system reaches a level of complexity where we no longer can follow the workings of the system, we happily jump to the next, higher level of abstraction. Then, if we find order, systems, “beauty”, it's an emergent property, if we don't recognize anything, the system is chaotic.
I would love to hear more of your thoughts on consciousness as an emergent property, especially what feedback mechanisms you envision giving it causal properties. I have actually never heard about it the way you put it as more than loose ideas. (Biophysicists in general don't care much about consciousness, that's for the weird psychophysicists next door. A field where I always have found the name giving weird associations. (The psychophysicists at my university worked with colour and vision, btw.))
Ack! I've got lecture in 10 min. Gotta go. I'd love to ramble on about this more, but duty calls! Young minds to
indoctrinate! Er, I mean, teach! Yeah, that's it! :DAh, but the borders between those two are more fleeting than most of us like to admit, arent' they? :)
(Hm... I wonder if I should edit the subject... :) )
-
@user24 said:
The bag of words / vector space approach is not particularly well grounded theoretically, and certainly isn't modeled on how humans do the job, but it plays to the strengths of the computer. It does have limits though. Even with SVMs and all the whizzy ML toys in the world, it's still very difficult to get a computer to understand the sense of a document. I still can't ask google very specific questions. But then it's optimised for keyword based document retrieval, not natural language querying. Maybe there are better systems out there.For webseach, not really. You would have to analyse a large corpus, and design a recall and ranking algorithm, based on the first step of the analysis. The part where the commercial players (Google, Bing) really shine compared to what is known in academia isn't really in information retrieval, it is in scaling, efficient implementations. Also, a typical user query is just a couple of words, far too little for any really deep linguistics analysis, so the systems just start using heuristics like your search history, your geographical location, your presumed first language and so on. Which is the reason it is so hard to search for the unusual meaning of a word... There are few real business secrets in search, but the exact details of ranking algorithms are probably one of them.
(I work with search software, so this, as opposed to most of the stuff I post, is actually an informed opinion. :D )
-
Just so as folks know I am keeping an eye on this thread. Please, can we have no more posts that might make any of our valued members feel uncomfortable.
-
The Linguist/computer theory discussion, while excellent, might also be better spun off into another thread as well.
-
@Lynn LeFey said:
The Linguist/computer theory discussion, while excellent, might also be better spun off into another thread as well.Na, we don't mind a thread veering waaaay off topic if that's the direction the discussion takes. :D Do feel free to try and drag it back on topic if you like though.

-
This forum is known for thread derailing. :D I love it also!
Sent from Ash forum mobile
-
@Lynn LeFey said:
@Gareth said:
Na, we don't mind a thread veering waaaay off topic if that's the direction the discussion takes. :D Do feel free to try and drag it back on topic if you like though.
You ever hear the expression 'herding cats'? I think it might apply here. :D
I love that expression. :D Seems to apply to quite a few aspects of my life though. ::)
-
@user24 said:
You studied under Searle? Impressive! (Also, another Philosopher-cum-computer-scientist? Cool!).Speaking of Searle, this aligns closely with an argument I made against his Chinese Box thought experiment (in an undergrad philosophy essay), namely that while none of the individual components of the box 'understands', the system as a whole - the sum of the components - could be said to.
I hold a doctorate in cognitive and physiological psychology. I received my BA in psychology at Berkeley with a minor in philosophy. I currently teach things like cognitive psych, physiological psych, sensory perception, etc. For my dissertation, I did a comparative test of two major theories of visual attention in terms of their ability to predict human performance in a novel visual search task. I also used that task to provide evidence for the idea that brain mechanisms of ocular and saccadic control are directly related to covert shifts of visual attention.
I'm familiar with the "systems" objection to the Chinese Room. Searle's reply is quite simple and straightforward. If the system "understands" Chinese, then anything can be a system and can be considered to "understand" whatever command we give it, or action we perform on it.
So, my pen is also a "system"; I put it on the table with the intent that it stay there and not do anything until I pick it up to write with it, and it does so. Therefore, according to the systems explanation, my pen understand the command "stay there until I pick you up".
When I do pick up my pen and write with it, my pen dispenses ink where I want it to and so it can be said that my pen "understands" the command "dispense ink where I put you in contact with a surface". Both of these examples are clearly hog-wash; in neither case is it valid to say that my pen "understands" either command, at least not in the way a person would understand them. According to Searle, that is no different than what the Chinese Room is doing "as a system". As a result, the "system" doesn't do any understanding at all.
-
@Steinar said:
I find this very interesting. My background is from biophysics (though I have never worked professionally in the field), so I guess it is no surprise that I'm decidedly in the materialist camp. I liked this way PZ Myers put it: “I think consciousness is a product of self-referential modeling of how decisions are made in the brain in the absence of any specific information about the mechanisms of decision-making — it’s an illusion generated by a high-level ‘theory of mind’ module that generates highly simplified, highly derived models of how brains work that also happens to be applied to our own brain.”Warning: From here on, just uninformed, personal opinion that will probably give a pro like yourself a headache. :)
Emergent properties is a fascinating thought, though. Personally, I think it's just an illusion from the way our minds work. We are so good at pattern recognition, that when a system reaches a level of complexity where we no longer can follow the workings of the system, we happily jump to the next, higher level of abstraction. Then, if we find order, systems, “beauty”, it's an emergent property, if we don't recognize anything, the system is chaotic.
I would love to hear more of your thoughts on consciousness as an emergent property, especially what feedback mechanisms you envision giving it causal properties. I have actually never heard about it the way you put it as more than loose ideas. (Biophysicists in general don't care much about consciousness, that's for the weird psychophysicists next door. A field where I always have found the name giving weird associations. (The psychophysicists at my university worked with colour and vision, btw.))
Ack! I've got lecture in 10 min. Gotta go. I'd love to ramble on about this more, but duty calls! Young minds to
indoctrinate! Er, I mean, teach! Yeah, that's it! :DAh, but the borders between those two are more fleeting than most of us like to admit, arent' they? :)
(Hm... I wonder if I should edit the subject... :) )
An emergent property is by definition where a property is created or arises from a system because of some form of synergism or interaction that takes place as a product (unintentional or otherwise) of the workings of that system.
I agree that we are basically pattern recognizers, and really good ones at that. It also makes sense to me that once we get to a certain level of pattern recognition we extrapolate that to levels that are unjustified by our direct sensory data and experience. But, I don't see that as a problem. Rather, I see that as evidence of the emergence of the property.
In the case of consciousness, the system is basically feeding back on itself in ways that are a by-product of evolutionary adaptations to specific adaptive problems. Various modular subsystems in the brain increased in complexity to deal with a wider variety of inputs, or equivalently an increasing amount of uncertainty of input, and in the process, those modules started communicating with one another in various ways as an indirect result of their separately increasing complexity. This manifests as recursive feedback or processing if you will. Because of its highly recurrent nature, the brain is able to process information about its own processing. Or to put in in terms of pattern recognition, to recognize pattern recognition.
The twist to this, which is related to what you mention about not being able to follow the workings of the system, is that our cognition is transparent to us for the most part. In cognitive psych we refer to this as "cognitive penetrability", or how far into our own cognitive processing can we extend our awareness and control of our cognitive processing. The answer is - not very much at all. Much of what the brain does it does so automatically and autonomously, and for good reason; survival value. We wouldn't be very successful if we had to turn on a whole bunch of stuff every morning like color perception, visual motion capture, digestive processing, balance and hand-eye coordination, and so on. Also, our cognitive mechanisms wouldn't be much of an advantage if we had to distinguish between the functioning of those systems and the information they're processing. So our cognition, and especially our perception, is largely transparent to us. Even though what we experience is actually the product of the functioning of the brain and nervous system, we "just see", or "just hear", or "just feel", and so on. Evolution has selected cognitive systems whose workings themselves don't get in the way of being able to process and respond to the information they take in.
So we have a built-in blindness of our own pattern recognition ability, as you mention above. Some evolutionary biologists, such as Dawkins, have suggested this is where things like the belief in God comes from. It's a product of our highly evolved ability for pattern recognition, combined with recursive processing in the brain, that allows us to perform causal reasoning, and to learn those associations of events and their effects. We then naturally apply this process to ourselves, and because we aren't privy to the workings of our own cognitive and perceptual systems, we end up making a "leap of faith" so to speak, and we propose the existence of supernatural and/or spiritual entities such as God, spirits, souls, etc. as a way of explaining who we are, where we came from, as well as how things work in the world.
Getting back to consciousness, I believe it's that recursive aspect of pattern recognition that produces consciousness in the brain. It's an irrefutable fact that's the case. Extensive studies of patients with various forms of selective brain trauma have shown that very specific cognitive and perceptual deficits occur as a result, and that manifest as gaps in consciousness, particularly for perceptual consequences of traumatic brain injuries. Since the brain can process information about its own processing, basically, that processing can then feedback and affect processing in other ways, in other areas, and so on. So that is where the causal efficacy of consciousness comes from. Searle says we'll eventually develop scientific methods and theories that will let us find just how this works in the brain, as do philosophers like Patricia and Paul Churchland. (They just differ in terms of how they explain consciousness in the workings of the brain from Searle. I'm very much in the Searle camp, which also is very sympathetic to the ideas of cognitive neuroscientists like Donald Hebb.)
I myself am rather skeptical of this idea. I think that the phenomenon of consciousness isn't that highly localizable to one part of the brain, but that it's a distributed function that depends on the total overall patterns of activity produced in the brain. I also think there is a powerful interactive component to it as well, based on the social nature of our species, and the effect that's had on our evolutionary history. (Social forces can place selective evolutionary pressures on us just as well as purely mechanical and consequential environmental factors.) I think an essential component of consciousness is that it takes place between individuals in a social context. Of course, this is essentially an unanswerable question, because we have no way of knowing if someone who would never have any social experiences, any language learning and so on, would possess consciousness.
But, in sum, that's my take on it. Consciousness is a product of both the complexity of the brain causing the development of highly recursive feedback processing mechanisms (70% of all neurons in the brain talk to other neurons in the brain, and only about 30% actually send signals to or receive them from the body), and that this process has also been shaped by our social nature and the effect that social nature has had on our evolutionary history. In short, we've selected ourselves to have consciousness to facilitate social cooperation and productivity, as it also increases our survival as a species.
One last thing. The closest thing we know of in the brain as a "center of consciousness" is the prefrontal cortex. Neurophysiological studies and imaging studies have shown this is where we seem to do our "thinking" and decision making. Back in the 40's and 50's, when lobotomies were performed on schizophrenics as a way to try and cure them, it also turned people into zombies for all intents and purposes. They wouldn't eat, sleep, etc. or do anything else on their own. They had to be prompted and guided to even the most basic things like go to the toilet. Based on this, it seems to indicate pretty strongly that something important to consciousness is going on there in the prefrontal cortex, but again, we have no way of being certain about this, because consciousness is a subjective and private experience. We have no way of knowing for sure if lobotomy patients no longer have conscious experiences, of if it's just that their ability to motivate their own behavior has been lost.
Because of this, I don't think we'll ever be able to scientifically quantify consciousness or its contents and qualitative experiences. The best we'll be able to do is produce a very high degree of correlation between brain function and what looks like conscious and intentional behavior, which we've already done a pretty good job with. Consciousness, because of its very nature, being subjective, transitory, and ephemeral, means we'll never be able to have a truly quantitative and scientific theory of consciousness.
-
Whoa! Was that from your Master's thesis in abstract biophysical philosophy?
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login
