Harnad, S. (2012) Alan Turing and the “hard” and “easy” problem of cognition: doing and feeling. [in special issue: Turing Year 2012] Turing100: Essays in Honour of Centenary Turing Year 2012, Summer Issue
The "easy" problem of cognitive science is explaining how and why we can do what we can do. The "hard" problem is explaining how and why we feel. Turing's methodology for cognitive science (the Turing Test) is based on doing: Design a model that can do anything a human can do, indistinguishably from a human, to a human, and you have explained cognition. Searle has shown that the successful model cannot be solely computational. Sensory-motor robotic capacities are necessary to ground some, at least, of the model's words, in what the robot can do with the things in the world that the words are about. But even grounding is not enough to guarantee that -- nor to explain how and why -- the model feels (if it does). That problem is much harder to solve (and perhaps insoluble).
“Many people have assumed that Turing had meant and expected the TT-passer to be a purely computational system”
ReplyDeleteI’m not sure that I understand why Turing didn’t expect the TT-passer to be a purely computational system and why that would even be a problem? A TT-passer (not including T3 and onwards) wouldn’t necessarily need to feel or even understand the arbitrary symbols it manipulates in order to communicate via email. As Searle has shown us, he could hypothetically communicate in Chinese through the use of rule-based manipulation of Chinese symbols that he does not understand; he points out, however, that that would not constitute cognition because he would not understand Chinese, it would just be computation. But, a TT-passer does not need to understand the symbols to fool a human right? I was under the impression that Turing believed that a purely computational system could pass the Turing test and can be considered to be “thinking” but in a different way than human’s think.
I think it's because Turing only spoke about T2 robots that are verbally indistinguishable and so does not entirely close the possibility of cognition being more than computation.
DeleteFirst, it is a speculation that a computer alone could pass T2. The symbol grounding problem suggests (and "Stevan Says") that only a T3 robot could pass T2.
DeleteAnd "Stevan (also) Says" that Turing was far too intelligent to not be aware of that.
It would still be true that a machine could pass T2; it just would not be a purely computational machine.
And Turing's core insight would still be that cognitive science should not try to affirm of one system what it denies of the other when they are Turing-indistinguishable in everything they can do (the "easy problem").
“So I do not believe that Turing was a computationalist: he did not think that thinking was just computation. “
ReplyDeleteAs Turing only spoke of the imitation as a T2 computer passing the Turing test, he left the possibility open to consciousness of being more than simply computation and therefore does not ascribe to computationalism. Furthermore, Turing only ever commented on the ability of robots to do what people do indistinguishably, not that they feel what people feel. The Turing machine, as brilliant as it is, only ever attempted to answer the easy problem by bringing about the notion of reverse-engineering. Although intuitively the hard problem would want to also be solved by reverse-engineering, as feeling is extremely subjective and because of the other minds problem, realistically, the hard problem cannot be answered by reverse engineering. Therefore, I agree with Harnad that Turing was not a computationalist as he allowed cognition to remain open to more than computation. But I also believe that Turing does not attempt and would not attempt to answer the Hard Problem with his Turing machines.
Turing explicitly said he would not try to address the problem of consciousness because it could not be addressed with his method.
Delete“Turing was perfectly aware that generating the capacity to do does not necessarily generate the capacity to feel. He merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wished to explain cognition. The successful TT-passing model may not turn out to be purely computational; it may be both computational and dynamic; but it is still only generating and explaining our doing capacity. It may or may not feel.”
ReplyDeleteThe hard problem of cognitive science is too difficult to solve, so we have accepted that the solving the easy problem must be the only viable method into understanding cognition. Harnad makes the case that Turing was not a computationalist, as he did not believe that all of cognition is computation. However, Turing did subscribe to the idea that insight into human cognition could only be obtained by reverse-engineering all that humans can *do*. Whether or not the Turing machine can feel is irrelevant, because knowing that can’t answer any questions that are capable of being answered.
Because of the other minds problem, we will never be able to solve the hard problem (unless technology gets so advanced that there are mind reading machines in the future? But I highly doubt it…). The closest we will ever get to understanding cognition is if we reverse engineer a sensorimotor-computational system that can ground symbols to their real-world referents and can manipulate these symbols in a way that is behaviorally equivalent to a human.
You are mixing up the HP and the OMP...
DeleteI definitely confounded the other minds problem and the hard problem. The other minds problem refers to the fact that we will never know if other beings are capable of feeling, and the hard problem refers to problem of how and why we feel. They are related in that they both have to do with feeling, but they are not asking the same thing.
DeleteSomething I would like to add is that even if we were to solve the other minds problem (by some miracle, by God, by technology, whatever), this still would not allow us to solve the hard problem. The mere knowledge that others are capable of feeling does not tell us how and why we feel.
How many people in this field think the hard problem is solvable? The easy problem, though difficult, seems possible eventually. Right now, we're just doing a lot of correlation with neuroimaging but some of them do lead to causal relationships, which would explain the how and why we do what we do. So let's say one day, the easy problem is solved. As for the hard problem, I can't imagine the way we explain the easy problem would suffice for explaining the hard problem. I think this is what you meant in class by saying all the degrees of freedom are taken up explaining the easy problem. Thus, no space is left to explain the hard problem. There's nowhere to go since all of our ways to explain things have been used up with the easy problem. This is why I'm not sure the hard problem will ever be solved. That's not to be dismissive of the field, just realistic with the information at hand now. Maybe (hopefully) there will be a breakthrough down the road, allowing for more degrees of freedom to explain how and why we feel.
ReplyDeleteMore degrees of freedom would require more physical forces than the four we have (and will have used on the easy problem) already.
Delete"Turing was perfectly aware that generating the capacity to do does not necessarily generate the capacity to feel. He merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wished to explain cognition."
ReplyDeleteEssentially, as stated in the paper, the TT passing model can be a computational/dynamic hybrid and it may or may not feel. We whether it feels or not, we won’t know unless we are the T3 model. By solving the other minds problem and figuring out if what goes on in the T3 model’s head is what goes on in our own heads, we might be able to solve the hard problem of cognition. If we find out that what goes on in its head is similar to what goes on in our heads, we can assume (but probably still never know) that it feels, because we feel.
If we can't solve either the easy or hard problem by introspection (which allows us to feel [some of] what's going on in our own head), how would it help (for anything but the OMP) if we could "introspect" in someone else's head?
DeleteI don't know if I have much to add to this paper, but this is the summary of the course in a nutshell. It's a succinct and accessible means of getting a high level view of this course that maybe would have done me some good had it come sooner.
ReplyDeleteIn the context of the other readings this week I find this particularly impactful:
"Well Searle is not feeling the understanding of Chinese when he passes the Chinese TT. He can distinguish real understanding (as he understands English) from just going through the motions: just doing the doing."
Just doing the doing makes a lot of what we've read clear to me. While there isn't much if anything actually new in this, it does work to solidify a lot of what we've been talking about all semester.
"The contribution of Descartes' celebrated "Cogito" is that I can be absolutely certain that I am cognizing when I am cognizing. I can doubt anything else, including what my cognizing seems to be telling me about the world, but I can't doubt that I'm cognizing when I'm cognizing. That would be like doubting I'm feeling a toothache when I am feeling a toothache: I can doubt whether the pain is coming from my tooth -- it might be referred pain from my jaw -- I may not even have a tooth, or a mouth, or a body; there may be no outside world, nor any yesterday or tomorrow. But I cannot doubt that what it feels like right now is what it feels like right now."
ReplyDeleteThis brought up some confusion for me. We've discussed kicking Gabe, and if Gabe the T3 robot would feel pain. I can't remember some of the conclusions about that. I still don't think we should kick Gabe, but I can't remember if a T3 machine has pain receptors like humans do? If it does not have pain receptors, how can it really feel physical pain?
"Explaining how and why we can do what we can do has come to be called the "easy" problem of cognitive science (though it is hardly that easy, since we are nowhere near solving it). The "hard" problem is explaining how and why we feel -- the problem of consciousness -- and of course we are even further from solving that one."
Of course, discussing the hard problem of consciousness is a central part to cognitive science and related fields. We talk about it all the time. However, on a personal level, I feel like we can't solve this the problem empirically and I have doubts that there will ever be a solution to it. I don't see how it can be solved, and spending all this time discussing how hard it feels tautological. This is not to say it does not deserve some discussion, or in any way am I trying to dismiss the field. It just feels kind of futile to me.
Gabe, our T3 machine is most likely to have pain receptors, and thus there is a chance that Gabe feels pain in the same way we do. What we discussed in class, and what Turing argues, is that the Turing Test will not be able to tell you if Gabe is truly feeling. The Turing test focuses on the easy problem of cognitive science – how and why humans do what they do. The hard problem – how and why humans feel – cannot be answered by reverse engineering or creation of a Turing machine. As Professor Harnad states in his paper, “[Turing] merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wish to explain cognition.”
DeleteOn your second point, I completely agree that it seems like the hard problem is out of our reach. As we’ve discussed in class, after all our resources have been dedicated to explain how and why humans can do what we do, what is left to answer the question how and why we feel? If Turing’s hopes were that all that we could accomplish scientifically was answering the easy problem and not the hard problem, is it a waste of time to dwell on the hard problem? I don’t know the answer to that question.
Anna, if you did kick Gabe and then asked him if it did hurt, he would say yes (on the basis of the input-output equivalence necessary to pass T3). Regardless of whether he does or does not actually have pain receptors, by the other minds problem, there is no way for us to prove whether he does or does not feel pain. Related, if Gabe did have pain receptors and you were able to use some sort of recording / measuring device to show a physiological reaction to a kick, it would still would not provide answers to the hard problem - why and how it feels like anything at all. Pain may be related to and a response to doing but nonetheless at the bottom of it, it feels like something to be in pain. As Harnard, Turing (and Taylor has suggested above), reverse engineering cannot solve why the latter is so.
DeleteLike Taylor and Anna, I just admit it is a bit disheartening to accept that the hard problem is virtually unsolvable. However, I do think it is not a waste of time to dwell on the hard problem. Given that it is what makes us humans (and not zombies), I also do not think there will be a halt in efforts to solve it. Apart from remedying this existentialism, I wonder what else can we gain by knowing how and why we feel (particularly since we already know that at least one of us (oneself) is certainly sentient)?
“The physical CT does not imply, however, that everything in the physical world is just computation, because everyone knows that a computer simulation of (say) a plane, is not a plane, flying (even if it can simulate flying well enough to help test and design plane prototypes computationally, without having to build and test them physically, and even if the computation can generate a virtual reality simulation that the human senses cannot distinguish from the real thing -- till they take off their goggles and gloves).
ReplyDeleteSo Searle is simply pointing out that the same is true of computational simulations of verbal cognition: If they can be done purely computationally, that does not mean that the computations are cognizing.”
The way this was put was really helpful to me as to how we can dismiss the idea that cognition is just computation (re: Searle and the SGP), but still allow for a T2 or T3 simulation of cognition that can do everything we can do verbally (T2) or that can do everything we can do (T3). This allows for a computational model that approximates the output of cognition (ie a T3 that can do everything we can do, ignoring the hard problem), while still acknowledging that cognition is not just computation. Can we say that computation (barring the hard problem) might allow for an ‘underdetermined’ model for cognition?
As Stevan says, a T3-passing robot would also be able to pass T2. If we were able to build a machine that could pass T3 (capable of doing everything that we can do), I would agree that it would be an underdetermination of the easy problem of cognition. This is because it is weakly equivalent, just input-output equivalence. As you have already suggested via citing Searle and the above quote, computation is not all of cognition. A T3-passing robot would have to integrate both computation and sensorimotor capabilities. Due to this, I am unsure if it is right to suggest that just using computation to build a potential T3-passing robot could serve as a underdetermined model of cognition (given that I do not think it is possible to achieve input-output equivalence on the basis computation alone in the first place).
DeleteIn this paper, Harnad gives a synopsis of the work done so far in solving the easy problem of cognitive science while also explaining our next steps (develop a robot with sensory-motor capacities that can pass the third tier of the TT). The easy problem can be solved, however, the hard problem, figuring out how and why we feel the things we feel is an impossible task. Because this paper is quite thorough, I thought it would be interesting to instead compare Harnad’s easy and hard problem of cognitive science to Chalmers’ easy and hard problems.
ReplyDeleteAfter briefly reading the second section of Chalmers’ Facing Up to the Problem of Consciousness which describes the easy problems and the hard problem, Chalmers’ easy problems, although more restrictive, seem to be similar to Harnad’s, as they try to explain various aspects of the things we do (Ex. Categorization, mental report, attention, etc.). Chalmers’ hard problem explores experience similarly to how Harnad uses feelings because a) they are synonyms and b) they both explain that all the things that we do are feelings as well. Where Chalmers differs from Harnad is the fact that Chalmers is concerned with consciousness rather than cognitive science overall. Because of this, it seems that both Chalmers’ easy and hard problem are one in the same because studying consciousness is simply studying feeling and is thus Harnad’s hard problem.
A question I have about Harnad’s easy problem, is besides clinical advantages, what would be the benefits of solving the easy problem? Although sci-fi makes it seem like T3 robots would make human life so much easier, I think it would cause a social crisis because people for the most part like being able to do the things that they do and in general need something to do in order to feel fulfilled.
I doubt whether T3 robots would want to be our servants (and whether we would want them to be!). I can’t imagine Gabe willingly driving me around in a rickshaw or baking me cookies whenever my heart desires. Like any other cognizing being, Gabe is the master of his own life, going about his own personal business. When asked in class whether we would kick Gabe, most of us said that we wouldn’t since we can’t be sure whether he feels or doesn’t feel, and would rather give him the benefit of the doubt. I don’t think many people would want to run the risk of enslaving a robot that could well be sentient.
DeleteOf course, once we’ve entirely reverse-engineered cognition, we may be able to subtract some functions and remove a T3 robot’s capacity to object and defend himself/herself, and that’s where some of the problems you mentioned might come in.
I think this paper is a very good outline and overview of all the things we've talked about in this course so far. It takes you through the sort of mental development that leads us to the hard problem - stemming from the Turing test. I don't have much to add because I think this is a very good summary of issues relating to the hard problem which has solidified many of my thoughts.
ReplyDelete"[Turing] merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wished to explain cognition." Following this quote I am left questioning if we ever can truly solve the hard problem. I think the general consensus is that the hard problem is unsolvable or at least seems that way right now since we are so far away from solving it in any way - in fact, we are even miles away from solving the easy problem. This leaves me wondering what the use for cognitive science is - are we just dancing around the real problem (the hard problem) which it seems we can never solve anyways? If the hard problem is unsolvable, where does that leave us? And what do we do from here?
Re: Flora Aldridge
DeleteHi Flora! I agree with you on many of the points that you bring up. I also really appreciated this article for its concise and kid-sibly explanation of this course's topics. As you and many other have stressed, the hard problem seems insoluble; however, I am sure that there is a reason why the "hard problem" is not called the "impossible problem". Though we are "just dancing around the real problem", there is still value in studying cognitive science because there are other applications of the (unexpected) answers we do derive. For example, many cognitive scientists conduct research in "consciousness" (which we know to be a weasel word for feeling) using neuroimaging and lesion studies that allow us to understand how awareness and attention function. I believe that studying the biological basis for consciousness does not necessarily solve the hard problem (nor the easy problem), but these (mis)steps still have important clinical applications.
Further, Professor Harnad uses symbol grounding and language to understand parts of the easy/hard problems. For instance, if we can determine the minimum set of grounded symbols needed to generate understanding, perhaps we can reveal more about how we categorize and do the things we do. I would not give up hope on cognitive science because evidently, researchers are finding ways to study "feeling" that do not diminish the complexity of the question.
On a more personal note, I also believe that thinking about the hard problem allows us to critically analyze our conceptions of empathy and our relations with other species. In Week 11, we had the opportunity to discuss topics like animal sentience and the other minds problem through the easy/hard problems. Can other species feel the way we feel? Why and how are humans regarded as the only feeling entities? These questions ultimately shape how perceptions of and actions towards other individuals/animals.
Hi Vivian,
DeleteI appreciate your responses to my questions and I do agree with you that there's value and importance to examining questions relating to the hard problem. I agree that certain sectors of cognitive science are doing beneficial work for clinical research/medical fields which is surely important. I'm just not sure if we will ever be able to utilize these methods (or similar methods to ones being employed in cognitive science research) to solve the hard problem. On a different note - I think your last point is really interesting and I agree with you that there is definitely value in studying 'feeling' as it allows us to empathize with other animals and/or people. I think studying the hard problem, or at least thinking about it, allows us to perhaps be more analytical and aware of our feelings. Additionally, I think it makes it clear to us that in fact the only thing that matters is feeling!
Hi Flora! I agree with you (and William) that we do not know how to attempt to solve the hard problem. Another way to frame the same question (as you did too) would be can humans ever solve the problem of why and how we feel, given that it is an uncomplemented category, or we do not know how it is to not feel? Is there an experimental setup we can imagine in which both a group of T3 biorobots and human infants are learning about the world in a similar manner? The biorobots and humans are similar in every aspect, other than the fact that the human children can feel. Could we then compare their rate of progress across time and see if the capacity of feeling provides any advantages to learning, survival or any specific cognitive function? Would that help us understand why we feel/ or what is the advantage of feeling? Or would we be not left with any differences between the two groups? The course seems to indicate that we would not, because explaining doing uses all the degrees of freedom. But do we have evidence for that?
DeleteThis paper clearly summarizes consciousness and computation (most of PSYC 538): the distinction between the easy problem and the hard problem, OTP, Turing, Searle, Symbol Grounding Problem, and Church-Turing Thesis. Harnad leaves us with a definition of both the easy and hard problem. Where are we to go from here?
ReplyDeleteThroughout all of the sky readings, there have been several propositions for methods on how we are to answer the easy and hard problem, such as heterophenomenology, fMRI, etc. Nevertheless, we come to the same conclusion: the approaches are not suitable or sufficient for solving the how and why. The paper leaves me asking an enduring question… what should we do now?
Really enjoyed reading Stevan’s article explaining the hard and easy problems of doing and feeling, and the very simplified kid sib narrative. It made it easier to comprehend some of the main takeaways from the course, other than the important topics of OMP and sentience.
ReplyDeleteOverall, I still find the distinction between grounding and feeling fascinating. Sensorimotor grounding is what we feel (or an important component that leads to feeling), but not feeling itself. We seem to require an additional capacity or force to feel. Is feeling hardware dependent? Grounding words requires a minimal set of grounded words, and hence memory. And memories seem to vary among individuals, leading to different brain (or hardware connections). As such grounding, and hence feeling seems to be hardware-dependent. Is it accurate to say that? That we all probably have different feelings about the same object/stimuli?
Sensorimotor grounding is connecting symbols to what they refer to in the physical world. Feeling is a part of this - as it feels like something to connect a word to something we see, touch, taste, hear or smell, but you're right that it is distinct.
DeleteI don't know if we can say that we can say that sensorimotor grounding leads to feeling though. Harnad states that sensorimotor grounding is necessary to pass the Turing Test, but that only covers the "easy problem". We still don't know what leads to feeling.
Hey Anna! Sensorimotor grounding is a part of our doing capacity but doesn't tell us anything about why and how we feel, and I don't think it's right to say that it is what we feel. It allows us to infer meaning for example and it feels like something to understand but it is not feeling itself. I think you mix up feeling with all of our doing capacities when you say that we all probably different feelings about the same object /stimuli. We have different emotions, reactions, thoughts about objects and stimuli but all of these are part of our "feeling" capacity. It feels like something to react, think etc. Like Emily says, these doings are part of the "easy problem" which are functions of feeling, which we have to be careful to separate from the "hard problem" of how and why it is that we feel.
DeleteIn this paper, also as we previously suggested in the class, Harnad says that cognitive science is a subject that investigates “the reverse-engineering of the capacity of humans (and other animals) to think”. Here, Harnad also suggested that “Turing's methodology for cognitive science (the Turing Test) is based on doing.” I have no problem with these two claims independently, but when considering together it raises my question of what could be counted as the phenomenon to reverse engineer with in cognitive science. Since when we conduct reverse-engineering, it has to start with an observable phenomenon, does this definition itself has restricted the domain of investigation that Harnad himself objects to? As Harnad says that "What is thinking? It is not something we can observe. It goes on in our heads. We do it, but we don't know how we do it. We are waiting for cognitive science to explain to us how we -- or rather our brains -- do it." If thinking is something un-observable, then how could be reverse engineer it? What else except the explicit behaviour of doing, just as Dennett and Turing suggested, could be counted as a phenomenon to reverse-engineer with? What would be the validity of such phenomenon across individuals, as one concern may be that this phenomenon cannot be intersubjectively confirmed between individuals as Husserl would demand (If there is confirmation, there is still a behaviour of confirmation, i.e. doing).
ReplyDeleteI feel like the distinction we draw about doing and thinking seems sometimes tenuous to me. Is thinking also a kind of doing, a kind that TT cannot explain. I feel like by completely separating the two, we are maybe still vulnerable to the problems of dualism. If doing is just doing and thinking is just thinking and the does not intercept, then from where we acted out our doing? Why does the doing match out what we intended in thinking?
ReplyDeleteAlso, this discussion of doing and thinking reminds me of how we talked about the uncomplemented category in class. I understand that feelings are subjected to this problem, and I wonder if we can have a state of “not-doing”, i.e. the total negation of doing. If we are actively aware of the category we are in is “not-doing”, aren’t we still doing something, i.e. reflection? Therefore, I wonder when reading the analysis of this paper, how should I understand the relationship and the demarcation of being and doing.
Hi Minjuan!
DeleteI might be incorrect, but from what I gather from your reply, I feel as though the distinction you refer to between thinking and doing is not as strong as you make it out to be. I feel as though we can argue that thinking is in fact a case of 'doing' something. The problem we run in to is the paradox of 'not doing' or 'not feeling' -- frankly I don't think we know what either feels like, and we cannot really speak to them as such. Doing is a ubiquitous and continuous process because cognition (thinking) falls under its umbrella (and so does feeling, by extension).
I wrote this out so sure of my answer, and now I believe I've only just managed to confuse myself!
“Explaining how and why we can do what we can do has come to be called the "easy" problem of cognitive science (though it is hardly that easy, since we are nowhere near solving it). The "hard" problem is explaining how and why we feel -- the problem of consciousness -- and of course we are even further from solving that one.”
ReplyDeleteThis whole course seems (at least under my impression) to be build around the idea of “reverse engineering” thinking (broadly understood). Computation was a prime candidate to try to explain all the capacities and all that organisms can do. Even if was strongly suggested (if not demonstrated) by Searle argument that computationalism (the idea that cognition **is** just computation cannot be true, I feel that the need to explain humans and animal as machines is still very present.
Maybe, from a “scientific” point of view, there is no other mean to “explain” what we can do. But science has the unfortunate tendency to treat everything as “object”. Humans, personality, animals, everything becomes something that you can and should be able to calculate, measure, etc. From this point of view, even “feelings” are treated as objects. But since you cannot measure them (except if you believe in a fifth force), it is not a big surprise why the hard problem seems insoluble (from a scientific point of view), you are not even sure what you are measuring. Is “feeling” only a big continuous object that eludes physical measurement? Are there multiple feelings? Is it non of that? Who knows? Of course this is speculation, but the point being that cognitive science may not be the best way to provide an organized discourse about feeling that give humans the impression that they “understand” it (given the underdetermination, we know that no explanation are the “right” one, so we agree that scientific theories are more “tools” than “truths” per se).
The point that I try to communicate is that everything related to feeling is maybe best taken by philosophy than by cognitive science per se. In my view, science and philosophy both try to “give meaning” (in the sense of “it matters”, I hope this is not too “weasel-like” since I cannot give a better account of what I mean) to existence, either by giving humans a sentiment of understanding of their world, or of themselves.
I agree here with you on the fact that the answer to the hard problem may be found in philosophy but not in cognitive science.
Delete"In my view, science and philosophy both try to “give meaning” (in the sense of “it matters”, I hope this is not too “weasel-like” since I cannot give a better account of what I mean) to existence, either by giving humans a sentiment of understanding of their world, or of themselves."
Indeed, this is something that humans like to do, trying to give meaning to all of our experiences and understanding how everything in the universe works. And this course showed us that this understanding may not be accessible for human beings . So why don't we just focus on "being" instead of trying to understand and attach meaning to everything like on how and why it is that we feel. I think that these questions should be left aside. Even if they were answered, then what?Will it change anything to the way we live and the way we are?
Hello! Here is my reply:
Delete“So why don't we just focus on "being" instead of trying to understand and attach meaning to everything like on how and why it is that we feel.”
Humans are curious by nature, we want to explore the unknown and try to learn more about ourselves. Science is about causal explanations. Through cognitive science, we are hoping to find the causal mechanism that explains why and how we feel.
I understand both of your points. I too believe that the hard problem is unsolvable by cognitive science, but I do not think we will get any close to solving the hard problem by philosophy. Philosophy will try to entertain the questions of how and why we feel things, but I don’t think it will help us learn anything about feeling.
"He (Turing) was perfectly aware of the possibility that in order to be able
ReplyDeleteto pass the verbal TT (only symbols in and symbols out) the candidate system would
have to be a sensorimotor robot, capable of doing a lot more than the verbal TT tests
directly, and drawing on those dynamic capacities in order to successful pass the
verbal TT."
I think this point shows what individuals like Dennett are missing in their thinking and approach to cognitive science. Turing, who both "teams" credit as being a major player in cognitive science, says that it goes beyond verbal accounts to be able to successfully pass the TT. Feeling and experience play a pivotal role in thinking about the hard problem and easy problem because as Searle pointed out, there's something more than just simple computation going on, thus alluding to the symbol grounding problem. Thinking back to Dennett's paper, I'm realising more and more all the ideas that this team's approach is negating, and why heterophenomenology is not a liable approach to properly answering any question we're concerned with.
Very good observation! I was thinking the same exact thing, by proposing a technique like heterophenomenology, Dennett completely misses the point and closes his eyes on the symbol grounding problem. While reading his paper I felt like I was reading Turing when we was describing a T2 machine that had verbal capacities and that, if it could successfully pass the Turing test would be a good enough candidate for cognition. And we saw during this course, what are the problems with computationalism. Behaviorists like Dennett and computationalists like Turing face the same problems of failing to account for our grounding capacities and on the how and why it is that we feel.
Delete“Turing was perfectly aware that generating the capacity to do does not necessarily generate the capacity to feel. He merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wished to explain cognition. The successful TT-passing model may not turn out to be purely computational; it may be both computational and dynamic; but it is still only generating and explaining our doing capacity. It may or may not feel.”
ReplyDeleteThis part of the text really got me thinking. On the first day of class, Professor Harnad pointed out that there is an “easy” problem and a “hard” problem of cognitive science. He explained that the easy problem was all about how and why we do the things that we do. He then explained that the hard problem was all about how and why we feel what we feel. From the beginning of this class, I was looking forward to discussing the hard problem of cognitive science. Talking about the easy problem was extremely interesting, but I always looked forward to the part of the course where we would discuss feeling and get insight on how we may solve the hard problem. Once we reached this part of the course, I was quite surprised at how little we know about how and why we feel. After thinking about everything that I have learned in this class, I do not believe that the hard problem can be solved. I believe that if we do ever successfully manage to reverse engineer a mechanism that is able to do everything that we are able to do, we will never be able to get it to feel everything that we feel as well (let alone know if the mechanism is actually feeling at all!) because we will have exhausted all the causal degrees of freedom. Reading that Turing himself doesn’t even think that we will be able to come close to solving the hard problem is quite disappointing for curious minds. If anything, he is trying to motivate people to focus on solving the doing and completely abandoning solving the feeling aspect of cognition.
I also don’t think that there is a way around the other mind’s problem. I don’t think there will ever be a way to confirm what and if other living beings are feeling anything at all. The fact that we cannot even confirm that other species are able to feel makes it all the more difficult to believe that we will be able to explain why and how we ourselves feel.
All of this to say, I believe that we will not be able to solve the hard problem of cognition… and if we ever are able to do so, it will be in an EXTREMELY long time. (I would like to remain at least a little hopeful because I am extremely curious).
In Harnad’s 2012 adage to the work of Turing, he has essentially gone over the earlier portion of the class, outlining our road; from the Turing Test, to computationalism, to the Chinese Room Argument and finally to the solving of the “easy” problem and the presentation of the “hard problem”. Rather than just summarize the paper, I’d just like to say that the reverse engineering paradigm of cognitive science has been quite a whirlwind for my humanities-conditioned mind. Sometimes I even ask why? Not “why do we feel?” or “why do we do what we do?”, but simply, why do we want to recreate ourselves? I have the upmost respect for cognitive scientists who theorize the T3 and T4 robots that might one day enter our worlds – and I think with the advent of advanced AI and big data, this path is more relevant then ever. However, I can’t help but having some cautionary reservations about the attempt to recreate(I assume that is the purpose of reverse-engineering) something in the image of human selves(an ironically biblical phrase, but i think it might speak to some purpose after all, although perhaps my earlier assumption is incorrect?). (please understand that I do not mean this in a deprecatory or negative way, I just sometimes feel kind of lost with it all, especially when it comes to the grim atmosphere of the hard problem..)
ReplyDeleteHi Matthew! I agree with you that this enterprise of understanding the mind, answering questions such as how and why we do what we do, and how and why we feel is quite overwhelming at times. I think your question about the will to create ourselves outside of ourselves is valid, but I don't agree with you that the goal of cognitive science is to re-create ourselves per se. I think the basis of cognitive science is related to questions of identity and it is particularly complex to answer those questions. Hence, to know whether we are successful at answering the questions, we can only rely on the observation of what we are able to build via reverse-engineering. In other words, the robots that we are trying to build, for example to replace some workers on dangerous jobs and the like, are very different from the robots that we are trying to build in cognitive science (to do perfectly what humans do, as well as humans do), specifically because the former are ends in and of themselves, whereas the latter are closer to a proxy that will serve as proof in our grander attempt at understanding the mind.
Delete“Turing was perfectly aware that generating the capacity to do does not necessarily generate the capacity to feel. He merely pointed out that explaining doing power was the best we could ever expect to do, scientifically, if we wished to explain cognition.” I feel like for us when we read about all of these past discoveries it seems so obvious and we are quick to judge, but Turing was really innovative and of course he did not try to sold the capacity to feel problem, he probably already knew that we don’t have what it takes to solve it.
ReplyDeleteRe: Veronica Rusu
DeleteI think most of us in this class would agree with you and the fact that Turing was a genius. My only issue here is that I disagree with the idea that cognition can be reduced to what we do. How we feel is also, in my opinion, a critical aspect in understanding what we do. Why do people react differently to the same situation, emotionally? How does that impact their behaviour? Why does the same emotion elicit different behaviours in different people? Why do some people cry easily while others are incapable of doing so, regardless of the emotional turmoil they may be experiencing? There are so many questions related to the human experience that are tied to emotions, I don’t really see how or why we can settle at “doing” as a satisfactory explanation for cognition.
I think emotions are inherently hard to study scientifically because they cannot really be seen or measured, only some of their subsequent physiological impact can. I think we need a new Turing, one that would be capable of coming up with the computer equivalent of feelings, only then could we claim to be on the right tracks to explain cognition as a whole.