I once gave a (perfectly awful) cognitive science lecture at a major centre for brain imaging research. The main project there, as best I could tell, was to provide subjects with some or other experimental tasks to do and take pictures of their brains while they did them. The lecture was followed by the usual mildly boozy dinner, over which professional inhibitions relaxed a bit. I kept asking, as politely as I could manage, how the neuroscientists decided which experimental tasks it would be interesting to make brain maps for. I kept getting the impression that they didn’t much care. Their idea was apparently that experimental data are, ipso facto, a good thing; and that experimental data about when and where the brain lights up are, ipso facto, a better thing than most. I guess I must have been unsubtle in pressing my question because, at a pause in the conversation, one of my hosts rounded on me. ‘You think we’re wasting our time, don’t you?’ he asked. I admit, I didn’t know quite what to say. I’ve been wondering about it ever since.
See also:
ABSTRACT: Visual categorization is thought to occur in the human ventral temporal cortex (VTC), but how this categorization is achieved is still largely unknown. In this Review, we consider the computations and representations that are necessary for categorization and examine how the microanatomical and macroanatomical layout of the VTC might optimize them to achieve rapid and flexible visual categorization. We propose that efficient categorization is achieved by organizing representations in a nested spatial hierarchy in the VTC. This spatial hierarchy serves as a neural infrastructure for the representational hierarchy of visual information in the VTC and thereby enables flexible access to category information at several levels of abstraction.
NOTE TO EVERYONE: Before posting, please read the other commentaries in the thread (and especially my replies) so you don't just repeat the same thing.
ReplyDeleteSorry to comment on this post but I really would like everyone to see the H01 dataset published on 2021,which is the most amazing and state-of-art 3D model, I think it might be useful to see some of current progress.
Deletehere is the link:https://h01-release.storage.googleapis.com/gallery.html
other avaliable brains include fruit flies can be found on the site https://github.com/google/neuroglancer
Delete**BLOGGER BUG**: ONCE THE NUMBER OF COMMENTS REACHES 200 OR MORE {see the count, at the beginning of the commentaries] YOU CAN STILL MAKE COMMENTS, BUT TO SEE YOUR COMMENT AFTER YOU HAVE PUBLISHED IT YOU NEED TO SCROLL DOWN TO ALMOST THE BOTTOM OF THE PAGE and click: “Load more…”
Delete________________
Load more…
________________
——
After 200 has been exceeded EVERYONE has to scroll down and click “Load more” each time they want to see all the posts (not just the first 200), and they also have to do that whenever they want to add another comment or reply after 200 has been exceeded.
If you post your comment really late, I won’t see it, and you have to email me the link so I can find it. Copy/Paste it from the top of your published comment, as it appears right after your name, just as you do when you email me your full set of copy-pasted commentaries before the mid-term and before the final.
——
WEEK 5: Week 5 is an important week and topic. There is only one topic thread, but please read at least two of the readings, and do at least two skies. I hope Week 5 will be the only week in which we have the 200+ overflow problem, because there are twice the usual number of commentaries: 88 skies + 88 skies + my 176 replies = 352!. In every other week it’s 2 separate topic threads, each with 88 skies plus my 88 replies (plus room for a few follow-ups when I ask questions.
I think there was a focus on functional specialization/localization, but that increasingly, scientists are looking for neural networks and connectivity rather than an allocation of a specific function in the brain, as outward behaviors are complex and usually need integration of multiple senses and processes to lead to that output. Especially in domains like cognitive science that focus on consciousness, attention, memory etc., compared to sensory integration that lead to a specific motor action which would require a functional location.
ReplyDeleteThe question Fodor raises is whether mapping WHEN and WHERE activity (whether local or distributed) occurs in the brain can help reveal HOW the brain produces our capacity to DO what thinkers can do. In other words, can it help reverse-engineer the brain's cognitive capacity?
DeleteMy inclination is to say that yes, by understanding the location and degree of distribution of the brain activity our underlying mental states CAN help reverse-engineer brain cognitive capacity. I agree with Fodor in the sense that no, reverse-engineering the brain itself is not - and does not relate to - the reverse-engineering of its cognitive capacity. However, a promising place to start for the reverse-engineering of anything is to localise its components. Assuming that brain activity can be thought of as a component underlying cognition, a natural place to start is to understand its characteristics. This is where knowing whether the brain “does different tasks at different places” or “questions about language in the brain” - which Fodor considers irrelevant to the question at hand- come into play. It is not to reverse-engineer cognitive capacity itself, but to better understand and perhaps eventually reverse-engineer the brain as the main component of this capacity. It seems logical that, as computation is implementation-independent, we can try to understand it by understanding one of the hardwares that it runs on which, in this case, is the human brain.
DeleteTo be the devil’s advocate, I'll try to defend the idea that we won’t be able to reverse-engineer the cognitive capacities of the brain based on the WHERE and WHEN of brain activation. To retake the analogy of Fodor concerning our understanding of the engine in our car, whether you know or not how the carburettor is linked to the air filter and the alternator won’t give you a further explanation of how each of those pieces activate the other ones. Sure, knowing what components activate simultaneously while doing a specific task could tell you what each part’s function is, as well as studying subjects with brain lesions, but as you said Ohrie, reverse-engineering the brain is not the same as reverse-engineering its capacities and we’ll still be unable to understand how it is that turning on for example the ventral pathway of vision, by leading to the temporal lobe, will make us recognize objects and why is it that activating the dorsal pathway leading to the parietal lobe will have different consequences.
DeleteI think mere correlations between brain activity and external behaviours do not provide the causal mechanisms necessary to solve the "easy problem". Neural imaging can reveal correlations, but Fodor emphasizes the need for causal mechanisms. Gathering data is a vital aspect of scientific research, but to fully understand and reverse-engineer cognitive abilities, it's essential to delve into the underlying mechanisms. I agree with Fodor who urges us to consider the limitations of brain localization data and to, instead, focus on better understanding the brain's causal mechanisms. Fodor also prompts us to consider that without a guiding hypothesis to formulate principles in cognitive science, the utility of neuroimaging in this context may be limited. Nonetheless, the brain remains a central focus in cognitive science, perhaps because there is currently no better alternative...
DeleteOhrie, kid-sib can't quite follow you: what is Fodor saying? What exactly do you agree with (and why)? and what do you disagree with, and why? I don't understand the difference between reverse-engineering the brain and reverse-engineering its capacities: What is reverse-engineering?
DeleteAdrien, same question. Does where and when parts are active help reverse-engineer the heart? If not, why not? If so, then why wouldn't it help in reverse-engineering a car? Is there a difference between reverse-engineering a heart or car, on the one hand, and a brain, on the other hand? If so, what is that difference? (There IS a difference.)
Miriam, kid-sib did understand your comment.
But isn't Turing Testing, even if only T2/T3, also reverse-engineering the (cognitive) capacities of the brain?
To answer the questions about reverse-engineering: where and when parts are active does help reverse engineer a heart. We can observe how valves open and close, and hence exactly how a heart completes its function of pumping blood through itself and the body. This wouldn’t help in the case of a car as a car has many systems working simultaneously. This would however work for one system of a car (such as air filtration). The difference between reverse-engineering these things as opposed to the brain is that in the heart and car, it is perfectly clear how each step leads to the outcome/function. In the brain, we are missing vital steps of how neural impulses cause behaviours. These steps are difficult to observe and difficult to reverse-engineer. Unlike Fodor, I do believe there is value to mapping different parts of the brain to functions (perhaps more general functions than just thinking about teapots) as understanding the parts is a step towards understanding the system.
DeleteGood points. To put it more compactly, (1) hearts pump blood and (2) cars propel passengers, but (3) brains pump everything that their passengers can do. So it's a lot easier with (1), and even (2), to match their function to their structure than it is with (3).
DeleteI love how Kid-Sibly Fodor’s text is, I’m sure it made him a good teacher in philosophy. If I understood correctly, he questions the usefulness of this type of scientific research by pointing out that what we know as the ‘hard problem’ (defined in class extensively) may never be solved, and furthermore questions the goal of trying to solve the easy problem. (The easy problem being the correlation between (1) the ‘when’ and ‘where’ of activity in the brain and (2) the ‘doing’ that thinkers can do, I think)
ReplyDeleteNo, the Easy Problem is not about correlation. What is the Easy Problem? And what is the Hard Problem?
DeleteThe "easy" problem of cognitive science is explaining how and why we can do what we can do, while the "hard" problem is explaining how and why we feel. I believe it is popular belief that mapping out the WHEN and WHERE of brain activity gives us insight into the easy problem. But as Fodor argues, there is no evidence in what neuroimaging is contributing to reverse-engineering the brain's cognitive capacity and solving the Easy Problem. Is this correct?
DeleteThe easy problem is how and why can we do what we do, whereas the easy problem is how and why can and do we feel. I think what Fodor is getting at is that while the specific data from function localisation experiments may not be something we exactly knew before (e.g., which brain region goes off when one thinks about teapots), its overall sentiment simply restates what we already knew, just scientifically. It does not, however, bring us closer to solving the easy problem, because it simply runs experiments of people doing arbitrary tasks that we know they can do. Cognitive science is not interested in the specifics, like how we can think about teapots (which one could argue the aforementioned experiment could provide insight on), but more so what gives us the capacity to do things we classify as abilities that humans have. The hard problem is not even brought up because function localisation does not even scratch the surface of the easy problem, let alone the hard one.
DeleteCsenge, "no evidence" is maybe putting it too strongly, but "still surprisingly little evidence" is perhaps closer to the truth.
DeleteJocelyn, similarly, "arbitrary tasks" may be a bit harder on experimental cogsci than deserved, but "not explaining the basic cognitive capacities but only tiny points of their fine-tuning" might be a bit fairer.
Using the example of reverse-engineering from class, if vacuum cleaners grew on trees and I wanted to build something that could do everything a vacuum cleaner does, I would start by taking it apart and seeing the parts that it has. If I knew nothing about how vacuums worked, just seeing that the motor and the fan are active when it is turned on might not be enough for me to then build my own working model. However, I think that most people would agree that this is a good starting point when looking for clues as to how and why a vacuum does what it can do. In this way, I do not agree with Fodor’s dismissal of neural imaging as a basis for reverse-engineering cognitive capacity.
ReplyDeleteHi Megan. I feel like what Fodor disagrees with the most, is that scientists are not trying to reverse-engineering cognition, but rather having fun with random mapping (in his opinion). I got the impression that he does not dismiss neural imaging completely is from his reply to Benjamin Martin Bly, where he says that brain imaging scientists are acting without trying to study cognition. I think he left an unsaid message, that he will agree with imaging method if it is for reasonable purposes.
DeleteMegan, good reply, and Fodor's example of a car is a very bad example, because you might eventually be able to reverse-engineer a car that way. But the brain is different: What does a car do? and what does a brain do? How and why might that difference matter?
DeleteTina, good reply. Fodor certainly doesn't say brain imaging is not important or informative in clinical work. But why does he think it isn't in the case of reverse-engineering cognitive capacity?
From my perspective, the car is an artificial tool, and the brain's intelligence and capacities enable us to design and manufacture a car. For this reason, we could either engineer or reverse-engineer a car. In contrast, the brain is already inside our skull before we could notice it - it also enables us to create the car, an artifact - so reverse engineering, a top-down approach, is the only choice for us to know how the brain allows us to think or have other capacities.
DeleteAlso, Fodor thinks brain imaging isn't crucial in the case of reverse-engineering cognitive capacity, since our theorization about a specific region (or a group of neurons) may be from a single capacity but overwrite the other possible capacities of them. For instance, neuroplasticity could be the brain's capacity, bringing uncertainty to the process of theorizing through brain imaging. This is why sometimes the brain area would have slightly different mental functions that we think. In consequence, brain imaging could help us understand merely the single brain's current status; perhaps Fodor just want to remind the researchers to treat it as the crucial may easily lead the researchers to avoid overusing over-reliance on one technology in the aspect of reverse-engineering cognitive capacity.
"Top-down explanation" is not what is usually meant by "reverse-engineering." Forward-engineering is applying what is already known to building something useful that we want to build. Reverse-engineering is about trying to figure out how something, that is already built, works. And it's not because the same brain area might be involved in multiple functions, or might change or vary, that Fodor argues, correctly, that correlations between cognitive activities or capacities and where and when activity occurs in the brain do not explain how the brain produces the cognitive activity or capacity.
DeleteI think that Fodor raises good questions about the usefulness of mapping the brain, however I don’t believe in his apparent conclusion that these mappings are not necessary. The research being put into this area, to me, seems useful in understanding the basic networks that are going on in the brain during cognition. While I understand that this does not necessarily directly relate to being able to reverse engineer cognition to answer the ‘how’ question, I think it is important to know what we are looking for specifically when it comes to processes seen during cognition to properly answer this question. Also I think it is useful to note that these trials are not only non-invasive, making them a more accessible and ethical way to learn about the brain, but also can be used in many other aspects of understanding human behaviour that are less related to cognitive science that do still prove useful. I do however appreciate his willingness to question the common practices as he is correct in saying that there are limited resources that must be allocated in an effective manner.
ReplyDeleteYou're right that the fact that neuromaging is non-invasive is a great improvement, especially for nonhuman animals used in research and for clinical research and diagnosis. But the question of what neuroimaging is contributing to reverse-engineering the brain's cognitive capacity and solving the Easy Problem, there's no evidence yet. Even for apparent exceptions, like the discovery of "mirror neurons" (4a) -- which was done with invasive brain recording in macacques but turned out to be measurable non-invasively with neuroimagery -- it has identified where and when the mirror neuron areas are active, but not how they do what they do: enable us to imitate the movements of others (which we already knew the brain could do already, but not how).
DeleteWhile the non-invasive nature of neuroimaging marks a significant advancement, especially for animal welfare and clinical research, it indeed has its constraints. It excels in identifying the when and where of brain activity, as seen with the discovery of mirror neurons, but falls short in elucidating the underlying mechanisms, leaving the "how" unanswered. This gap underscores the need for a multifaceted approach in neuroscience research. Complementing neuroimaging with detailed cellular and molecular studies can offer a more comprehensive insight, helping to bridge the gap between brain activity localization and understanding the intricate neural processes involved in cognitive functions. Given these limitations, do you think the current emphasis on neuroimaging in neuroscience research is justified, or should there be a more balanced approach integrating various research methods?
DeleteI found it interesting to read on Fodor's concern regarding the incapacity to answer the "hard problem" by the use of neuroimaging. As if the sum of the parts of a brain could not answer the whole brain. And could even less answer how and why we think the way we think. But I am not sure to give him some reason on that point. And this is where I join you, Jenny. Because, for me, understanding a whole and its working comes by understanding all of its parts, to then be able to understand better the whole. And neuroimaging is extremely helpful for that: structural imaging helps understanding the parts, whereas functional imaging helps understanding the relation between these parts, in a bigger mechanism (the whole). And from my own perspective, this bigger mechanism could be associated to thinking. But I guess I'm wrong, since the "hard problem" is still a problem.... and I'm starting to look forward to the day when it will be solved...!!
DeleteHi Juliette, while I completely agree with you that understanding of a "whole" is possible through the understanding of its constituent parts - I do not think this is what Fodor was getting at. I do not think Fodor was asserting that one could not learn about the brain by knowing its parts, it seemed to me that Fodor's point was that by understanding the frontal lobe does X and the temporal does Y, we might come to understand that the brain does Z, but this tells us nothing about HOW the frontal and temporal lobes do X and Y, and HOW the brain does Z. Though, I do believe Fodor asserts his point too strongly. I believe it is useful to know what certain loci do, even if it is simply as a groundwork to test hypotheses.
DeleteI think this is an interesting topic to come back to after having done the week 11 skywritings around animal feeling and ethical research. It brings me back to the idea that neuroimaging should not be discounted as a useful way of performing such research. While I understand that there are no explicit links to its usefulness in reverse engineering the mind as we set out to do in cognition, I don’t think it means it is impossible for it to prove useful in the future. Especially if we hope to start decreasing the amount of harmful animal testing and experimentation that we are doing. I believe that this type of mapping could eventually produce results in a meaningful and ethical way.
DeleteI think Fodor is being too strict with neuroimaging. Because as we see in the previous reading of this week, that mirror neural system located close to the language system offers a unique insight into how language evolved. This is an example that brain imaging is being used to contribute to reverse engineering cognition. And chances are, many discovering are going to be made by serendipity. In which neural imaging can be a powerful tool if used well, as it has the capacity of giving us information to tackle how thinkers do things. Therefore, I think we should allow neural imaging some degrees of freedom to experiment with their hunch. However, I also do agree with Fodor that money in the scientific research field is a valuable resource, and that we should be more considerate when spending it.
ReplyDeleteDid we need to discover mirror-neurons to know we could (1) imitate movements and sounds, (2) understand what the goal of another's movement is, (3) understand the facial expressions and tone of voice of someone else, (4) understand what people are saying, and (5) understand when an other organism is frightened or in pain. These are all extremely important perception/production mirror-capacities that we already knew that we -- hence our brains -- have, but not how our brains do it. And we still don't know.
DeleteThis posting by Eugene G was inadvertently erased:
DeleteEugene G September 22, 2023 at 9:08 PM Based on what Fodor suggests, he appears to have a pessimistic attitude towards our current technology, specifically neuroimaging, deeming it a waste of time and money. However, I believe we should never halt our exploration, as discoveries and techniques are interdependent. Perhaps in the future, advancements in technology used in neuroscience will confirm the possibility of T4's existence. By then, we might either have already discerned our capacity for doing and feeling, or we'll have the opportunity to unravel it progressively, much like we are currently attempting to understand how ChatGPT functions. I'm concerned that this kind of discouragement might diminish the enthusiasm of the few candidates willing to dedicate their time to addressing both easy and hard problems. However, this skepticism can also serve as a reminder for researchers to stay true to their original objectives and not deviate from their primary goals.
Eugene, I understand the point about not discouraging technology, but none of the other points about T4, ChatGPT, and the Easy and Hard Problems. Can you say it kid-sibly, please?
DeleteI understand that the discovery of mirror neurons alone doesn't actually explain how neural activity allows us to imitate actions, empathize with others, etc., neither does the discovery tell us we have these capabilities, but isn't the discovery still a valuable step in reverse engineering these functions of the brain?
Deleteif it leads to a causal explanation...
DeleteThank you for saving my comment! Sorry and here is my revised version: I feel like by encouraging the development of technology, it will bring us the possibility to get closer to reverse-engineering. Taking ChatGPT as an example, even right now people have no idea how it works, but some have already argued that it has already passed T2 test.
DeleteEven though we have no idea how it functions, but this brings more chances to see closer, to find a way to do reverse engineering based on ChatGPT. The process of creation of T4 could be similar as the creation of ChatGPT, when there are breakthroughs on technology of neuroscience. And as previous article stated, to have the ability to understand others’ behaviors and emotions, in terms of feelings, is essential on human evolution.
At that time, under the achievement of T4, we either have already figured out human’s capacity of doing and feeling, or this provides us shortcut to ‘break it down’ and study how it works just as what we are doing to ChatGPT right now.
We do know how ChatGPT works: huge gulps of text, statistics, and next-word completion (though no one yet knows how that produces such good performance.) But GPT is a model and we're trying to reversenge how WE do it -- and that certainly can't be by swallowing the Big Gulp, because we can't!
DeletePerhaps neuroimaging plus 3-D printing could duplicate T4, but would that explain how T4 passes T3 any more than cloning the brain would? So kid-sib still can't quite follow your point.
I am wondering, as us humans grow up, we accumulate the information we are exposed to, can this process be the equivelance of the big gulp? But I guess it then leads to how we learn, which cannot be explained by how GPT processes it Big Gulp. Or can it be?
DeleteI agree with Fodor that the mapping of arbitrary cognitive functions to areas of the brain does nothing to explain those functions. But maybe the maps generated in this way have some use for cognitive science as data that we can test cognitive theories against. For example, if we built some model of cognition that processes the meaning of a sentence by invoking the sensory experiences related to each word’s referent, then we could affirm our faith that this model works the same way human cognition does by checking the neural imaging data to see if the brain processes sentences the same way (for example by sending signals to the areas of the cortex which process each kind of sensory input). Fodor argues against this point by saying that indiscriminate imaging is an inefficient use of limited scientific resources, but inefficiency is not the same as uselessness.
ReplyDeleteYou're right that neuroimagery could help test cognitive theories that make predictions about localization or sense modality. But then the cognitive theories themselves need to be contributing to reverse-engineering our cognitive capacity.
Delete“I take it to be currently the consensus that they have significant scientific import over and above their implications for medical practice.” I don’t think that’s very comforting to people still dying to the incurable diseases such as Fatal familial insomnia (FFI). Maybe brain imaging will be the key to producing synthetic sleep states. Basic research is and always has been beneficial for scientific development. Seems more useful to study the brain and how to save a person’s life than caring about what a black hole is doing 2000 light years away (no offence to astronomy but if we have to pick one).
ReplyDeleteMaybe reverse-engineering the functions of the brain won’t lead to a complete understanding of cognition, and other disciplines, such as philosophy, will help complete the explanation. Physical brain studies will likely be the how part of both the easy and hard problem in cognition, even though it may not help with the why part, we have to start somewhere. I like that Fodor wrote this thought provoking article and that he doesn’t give his final stance on whether we’re wasting our time and money and he admits that he’s still ”wondering about it”.
Fodor agrees neuroimagery is important and valuable for clinical purposes; he doubts that it will help reverse engineer cognition.
DeleteMoving on from the fact that the text was on purpose a bit aggressive to make the reader react, I liked the self-questioning that it brought up in me. Indeed, as a cognitive science student someone rarely says to you that what you are studying is not relevant. I agree with the author that the precise localization of functions might be less relevant than if they are, in fact, localized or not. However, both are very important to answer the question if we can reverse-engineer the brain’s cognitive capacity. Furthermore, how can it help differentiate with how machines think. The field of cognitive science as it evolves tends to side of non-localized cognitive functions. Therefore, can machines think – as they have clearly defined and localized functions ?
ReplyDeleteOrganisms can think, and organisms (including humans) are machines (cause/effect mechanisms). So machines can think. The question is HOW? Can neuroimagely help?
DeleteI think Fodor had a pessimistic view on neuroimaging, and looking into when and where activity occurs in the brain. I think it might not immediately give light to how we could reverse-engineer the brain, but to reverse-engineer the brain, wouldn’t functional specialization/localization be a really important component since the brain is not just one thing, or equipotential? When he said “It occurs to me that maybe we’re heavily invested in finding answers to which we don’t know the corresponding questions.”, I took it as he found it pointless to researching neural localisation. But I think that finding any sort of answer, even the ones to which we don’t know the corresponding questions, will give rise to more questions which could help us understand more about the brain, or at least maybe guide us to the right questions that would help reverse-engineering the brain’s cognitive capacity.
ReplyDeleteWell, yes, let's keep imaging (not just for the clinical but also for the cognitive), but maybe not while we hold our breaths; maybe a bit of computer-modelling (Strong C/T Thesis) might help speed things along too...
DeleteAlthough Fodor makes an extreme point on neuroimaging, I understand and somewhat agree with where he is coming from. Neuroimaging is important, especially for understanding neurodevelopment and what makes a healthy brain. However, the point Fodor makes is that the where in the brain thinking occur doesn’t seem to matter. I do agree with him that in reverse engineering, where things occur seem like a strange question (what does it give us as tools)? Beyond that, I believe when we focus to much on specialized locations of the brain, we often miss out on the emergent properties of the individual parts as a system. In doing so I think we tunnel vision and may miss out on what is important. On the other hand, knowing that specific functions occur in specific areas is useful as it may allow us to study the cytoarchitecture of said brain area (potentially giving us components to reverse engineer some parts of cognition). Overall, I believe there needs to be a balance between studying the smaller fine grain details of the brain and where they are located and the emergent properties of the system as a whole.
ReplyDeleteSee reply to Selin, above.
DeleteFodor brings up some excellent points regarding neuroimaging and why it may not give us all the answers we need regarding one of the main questions we need to answer as cognitive scientists. Namely, it cannot give us a causal explanation for how and why humans are able to do all the things we are able to do (the Easy Problem). I believe he grants that neuroimaging has important clinical applications (allowing surgeons to preserve cognitive capability in patients for example) but he does not believe it will help us solve the Easy Problem. In this respect, I agree with Fodor. For example, consider the mirror neurons (MNs) from the last reading. We located in the brain groups of neurons we believe are responsible for specific human behaviours (language comprehension, empathy, motor imitation, etc.) but we are still left with the question: how do these neurons give rise to those abilities? We have no causal explanation, we’ve only correlated their firing to the display of the behaviours in question.
ReplyDeleteImagine we reverse engineered a car in the same way neuroimaging aims to reverse engineer human cognition: by identifying the pieces of the car that are responsible for the various aspects of capacity to move people around in space. If we were then asked “how do those parts, combined thusly, CAUSE the car to move people around?”. We couldn’t answer that without discussing the laws of physics, that the car is turning chemical energy into kinetic energy and exerting a force on the ground through its tires and by Newton’s Third Law, the ground exerts a force back on the car and it moves forward. By only examining the components of the car, we could not tell the story above. Equally, by only examining which components of the brain are tied to specific behaviours, we cannot say causally how those behaviours arise from those components. A question I’m left with that I’d love to hear people’s thoughts on: what other methods for investigating human capacities should be involved in telling the causal story and what pieces of information would they bring that neuroimaging misses?
Hi Stephen,
DeleteI agree with most of what you're saying; I think Fodor's argument has sometimes been misunderstood as an argument against neural imaging as a whole, when in reality it is only an argument that it will not help us solve the easy problem. He still acknowledges its potential for clinical application. Fodor mainly had issue with the scientists that were finding correlations between brain functions and brain regions and then going on to say that they were contributing to the easy problem, when in fact there is nothing being shown about any of the brain's causal mechanisms. Finding correlations doesn't help us identify HOW the brain activity gives rise to function.
As for what other methods we have available to us in our attempt to find these causal mechanisms, it may be the case that computing - combined with information that we're gaining from cognitive modeling - could be a better direction to explore than cognitive modeling on its own.
Stephen, good points, but I think that Fodor is wrong about the car: "Auto-imaging" could help us (a little) with reverse-engineering cars (if cars grew on trees rather than being forward-engineered by us). Not the whole causal story, but some pieces of it. See my replies about hearts, cars, and brains, and how function resembles structure in two of them but not the third. (And I don't think real auto inventors had to design cars from basic physics, bottom-up.)
DeleteAdam, good reply. (But in general, although correlations are not causal explanations, they might (sometimes) suggest causal hypotheses,)
The quote from this article that sums it up for me is: "we’re heavily invested in finding answers to which we don’t know the corresponding questions." I find in psychology we are taught that activity level of neurons in one area indicates that is where is the cognition is happening, but this article made me think differently, that this doesn't explain how neurons lighting up in one area creates cognition. In terms of reverse-engineering, it also doesn't get us any closer. I don't think it's totally irrelevant to understanding cognition, but as Fodor points out, allocating a huge amount of funding to trying to reverse-engineer cognition by looking exclusively at localization of activity does then seem like a waste. It might be more useful to integrate some of these ideas with what computational models tell us about cognition.
ReplyDeleteGood points. But Fodor over-reaches with his critique.
DeleteThe question raised by Fodor is whether functionally mapping the brain (temporally and spatially) is relevant for reverse-engineering the brain's cognitive capacities. He agrees that it is relevant for clinical purposes and that it matters in determining whether mental functions are neurally localized or not. His concern is that neuroimaging is monopolizing funding, researchers, and public attention without contributing significantly to cognitive science. I found his questioning interesting and well-founded.
ReplyDeleteIn my opinion, the first part of the answer is that neuroimaging techniques are the only and most effective non-invasive techniques currently available to investigate the brain. They do indeed have many limitations and may lead us to believe that functions are more localized than they really are. However, I still think that neuroimaging is useful in reverse engineering the brain's cognitive capacities. By localizing functions, it becomes possible to associate them with specific neurological structures (types of neurons, specific organization in layers, connectivity with other regions, etc.). Additionally, activity correlation can help identify networks, and repeated neuroimaging allows for the study of the effects of development and learning on the brain. Hence, I believe that we can't directly infer the HOW of our cognitive capacities with neuroimaging, but it can either strengthen or weaken specific theories by testing their predictions.
And neuroimaging is used to test some toy cogsci models that have implications for location and for timing (as in laterality and reaction-time experiments). So Fodor is a bit over dismissive. (Although Fodor is a philosopher, he used to collaborate with cognitive psychologists in psycholinguistic experiments, years earlier. But his worst day was when he tried to criticize Darwinian evolution...) (Week 7)
DeleteI would hope that Jerry Fodor is not right in saying that neuroscientists ‘are wasting their time’ as someone who wants to pursue a career in neuroscience. I think localization of senses, tasks, and functions are extremely important to out understanding of how the brain works, which will continue to help us treat brain traumas, disease, as well as will allow us to reverse engineer brains in the future. Though Fodor has some interesting takes, I think he’s missing something on this topic.
ReplyDeleteFodor does not doubt neuroimaging's clinical importance, just it's contribution to reverse-engineering cognition (so far).
DeleteIn this reading, as others have said in their skywritings, Fodor believes that neuroimaging is less valuable for reverse engineering cognition than it is for clinical purposes, and I would like to use a analogy to explain the insurmountable barriers that neuroimaging has in reaching out reverse engineering cognition. The purpose of neuroimaging is based on the neuroscientists’ interest in mapping the brain to localize mental functions in the brain. In other words, the work done by neuroimaging is more like exploring the new land of the brain, scrutinizing the causal relation between the brain and the mind, and creating an interpretive guide map for this new land. As such, it has nothing to do with helping to reconstruct this continent; a map of the continent can only guide us to know how to walk on this land without getting lost; it does not help to know how this continent was formed and built.
ReplyDeleteAnd correlations don't explain causation (though they can suggest hunches.).
DeleteI think many of us are in disagreement with Fodor because it seems as if he is saying that utilizing brain imaging to map brain activity isn’t relevant to understanding and explaining cognition. Perhaps he is indeed saying this, but I think a more important takeaway from his article is not that brain imaging isn’t useful, but rather, we’re carrying out all these brain imaging studies without a defined question, or strong purpose (as he says, “it occurs to me that maybe we’re heavily invested in finding answers to which we don’t know the corresponding questions”). I agree with his point that we shouldn’t be expending time, money, and energy on mapping activity and correlating it with various functions/behaviours, if we’re not invested in also coming up with hypotheses and explanations for why these correlations exist. As people have said previously in comments above, brain imaging is likely useful if we use it to test our hypotheses for causal mechanisms behind these brain functions, but perhaps Fodor is correct in saying it’s worthless to simply just observe where activity occurs.
ReplyDeleteAnd imaging is infinitely preferable to invading...
DeleteI found the following quote to be interesting: "But the issue is academic in the invidious sense since in fact there’s no good reason to think that similarity of psychological functions generally predicts similarity of brain locations or vice versa." I find this to be a bold claim, but overall his argument concerning the inability to reverse engineer cognition by looking at correlates in the brain alone makes sense. I'm not sure if I believe this fully though. For example, if we find certain brain areas to be responsible for the physiological state of hunger, and we find that a person who has damage to those areas would not feel hunger, do we then conclude that those brain areas predict the conscious feeling of hunger?
ReplyDelete"Responsible for" (and even "causes") does not explain HOW...
DeleteBut, apart from the fact that it feels-like something to be hungry (the HP), does the fact that removing or damaging a part of the brain eliminates or damages the appetite explain how the brain causes hunger? And that's not even the EP, if appetite regulation is vegetative rather than cognitive.
We obviously want to solve the easy problem, that is how and why humans are able to do all the things we are able to do (the Easy Problem) and Fodor believes that brain imagining is not useful to solve it. But isn’t it still important to know regardless of if it will solve the easy problem? I agree with him to a certain extent that the resources could be used for something more important but I disagree with his car example. I think it is still a valuable step that we need to help us reverse engineer and we do need to start somewhere.
ReplyDeleteI agree. While brain imaging may not provide all the answers to the "easy problem" of consciousness itself, it serves as an essential foundation. Understanding the neural correlates of various mental processes and functions is a critical step towards unraveling the mystery of consciousness. It provides the building blocks and empirical data that research can build upon. However, I share Fodor's perspective on the absence of a well-defined research strategy in such studies. The fact that there are brain imaging researchers who appear to prioritize collecting experimental data without a clear direction/specific research questions prompted me to reevaluate my own stance on this issue.
DeleteNot invasive, no harm done. The issue of research funding priorities is beyond the scope of this course. (I'd vote for research on what to do about Climate Change -- and how to get people to do it!)
DeleteCan mapping out the brain help reverse-engineer the brain’s cognitive capacity? This is a question that professor Harnad mentioned in an earlier reply. At this moment it seems promising, since to my understanding the general belief among neuroscientists is materialism. I do however like Fodor’s question; ‘do we need to know everything scientifically?’. Is making the map and completely reverse-engineering the brain really necessary? It’s important for us to keep trying to understand the brain for clinical reasons, but don’t we want to maybe leave some mental functions up to mystery?
ReplyDeleteHi Fiona! Personally, I don’t think we need to completely map out the brain to reverse engineer cognition. The major point I took away from Fodor’s critique of neuroimaging was that simply observing where something occurs in the brain does not necessarily tell us how these areas do what they do (and what this means for our ability to cognitively do what we do).
DeleteI found this discussion very interesting so I read Professor Harnad’s paper that is linked to this page, where he formulates this question in another way. Professor Harnad explains that “we can observe what the computer is doing on the outside. . .but that does not tell us what program (algorithm) it is executing.” This helped me understand why we may not need to be so completely focused on neuroimaging to reverse-engineer cognition—maybe there is a more creative alternate approach we can use to determine what ‘program’ the brain is executing, not just what brain area is active at a given time.
But if computationalism is wrong then what's going on inside is not a program (or not all just a program)...
DeleteI thought that Jerry Fodor’s questioning about why researchers are spending so much time and resources trying to identify what specific areas of the brain are associated with specific mental states that a person might be having was a valid form of questioning. In fact, the point of brain research is to identify what area of the brain is processing a specific mental function, but what is the purpose of that? What information are we looking to extract, or what questions are we trying to answer with that research? Can this research explain the hard problem of cognition which is how and why we feel? Even if we manage to identify the brain areas that are associated with feeling hunger, or with feeling sad or happy, this isn’t really telling us something about how these sensations are occurring. I guess we could say that they are the result of neurons firing and sending signals, but that does not offer a clear explanation to the hard problem. However, I believe that this research does offer an answer to the easy problem of cognition because, through reverse engineering, it allows us to know how and why we can do what we do.
ReplyDeleteLocalization is very important for clinical neuroscience and neurology. It certainly can't solve the hard problem of feeling, but can it even help solve the "easy" problem of reverse-engineering cognitive capacity? Fodor thinks not. Why do you think it can? Now?
Delete
DeleteWhile I don’t think that it offers a complete answer to the “easy” problem of reverse engineering cognitive capacity, I think that it can help answer it. First, knowing where something is can give us the building blocks of knowing what portion of the brain to do more research on how to reverse engineer that process. He also touches on, and quickly dismisses, the idea of it being utilized in a clinical sense, but I think it is more important than he gave it credit. In a clinical sense, locating the area of the brain causing the person to feel and experience what they are is the first step to knowing where to do more research. If every time someone had an issue, we had to look at the whole brain rather than being able to pinpoint exactly where it is coming from. It would be a waste of time and resources (as he claims he is not wasting resources). I believe localization is a step in the right direction for finding an answer to the “easy problem,” and he dismisses it too harshly; however it is not the only factor, even not the most important factor, that needs to be looked into.
While Fodor emphasized the significance of neuroimaging in clinical applications, he expressed a strong opinion about its limited utility in comprehending cognition. Neuroimaging, they argued, falls short as a means to unravel the "why" of cognitive processes because it primarily examines the tangible components of the human brain. This made me think about the limited yet valuable role played by neuroimaging in exploring the unconscious mind which poses greater challenges for comprehension compared to the conscious mind. Similarly, while neuroimaging may not be the key to understanding the complexities of the "hard question," it can potentially contribute by providing valuable data on brain activity.
ReplyDeleteBoth "mind" and "conscious" are weasel-words: "Felt brain state" and "unfelt brain state" are better approximations. What is the conscious mind and what is the unconscious mind?
DeleteI thought the "conscious mind" would be what we're aware of —our thoughts and experiences we can think about.
DeleteThe "unconscious mind" would refer to mental processes, memories, and emotions that occur below the level of conscious awareness —like automatic reactions or hidden feelings.
I once asked a grad student why we are focused on such a narrow aspect of cognitive science and whether there are people out there doing more 'meta' studies. He told me that researchers are currently learning as much as they can about the different aspects of cognition (attention, decision-making, etc.) and that the goal is to tie them all together to explain cognition as a whole. It didn't make sense to be then and it doesn't make sense to me now and the ending of this reading made me remember this interaction. I agree with the author that trying to understand a particular thing extensively can sometimes be pushed to ridiculous levels when one loses touch with the goal. I think it would be useful for CogSci researchers (but also researchers in general) to take a step back and think more about the intention of their research projects and think about how it would help us answer the 'why' and 'how' questions that we are so eager to answer.
ReplyDeleteI'm not sure I understand the author's point in the following passage on page 4 and would appreciate it if someone would confirm if I have it right:
''But why [...] does it matter where in the engine the carburetor is? What part of how your engine works have you failed to understand if you don't know that?''
From my understanding, he is saying that knowing the theoretical/abstract idea of the 'how' suffices and that we don't need to know the concrete underlying mechanisms of it. But isn't the whole point of CogSci to understand the ''how'' concretely so that we could understand it and potentially replicate it?
Although I agree with him that brain imaging studies are disproportionately endorsed, I disagree with him on some of his points, because, unlike him, I quite like knowing where exactly the carburetor is :)
I think I just understood what Fodor meant; I think he was talking about implementation-independence (i.e., that we don't need to know about the hardware, we just care about the software).
DeleteFodor might or might not be a computationalist; hard to tell (he did co-publish with Pylyshyn, who definitely is.) Either way, it's not clear that correlating DOINGs with when and where brain activity occurs do not explain how the brain is doing the doing.
DeleteGood points, but the main reason the brain is much harder to reverse-engineer than a heart or a car is not because of the brain's neuroplasticity but because the heart's and the car's structure resembles its function, whereas the brain's does not (or hardly). The brain's function is to do everything the body can do.
ReplyDeleteHow can one simplify thinking, carried out by such a complex and phenomenal organ as the brain, into rudimentary regions that are specific to thinking about teapots or taking naps? Its insulting to the capacities of Fodor’s own brain! Cognitive science hasn’t yet answered the Hard Problem – this we can all agree on. But Fodor’s stance is inflammatory and unrealistic if he believes that basic research in cognitive science isn’t meant to be translational! I am entirely in agreement that research questions need to be grounded in strong hypotheses. But there is absolutely a value to brain maps (he so quickly grazes over its immense importance for clinical assessment). In fact, Fodor’s argument supports research investigating anatomo-functional relationships in the brain, which have the potential to reveal which structural areas are involved in specific aspects of cognitive function. Now, the brain is not modular, but it’s a step towards reverse-engineering and trying to understand the ‘how.
ReplyDeleteCan you give kid-sib a sample of how localization could explain how that location produces the cognitive capacity it is correlated with? (e.g., mirror-neurons)
DeleteI can give a kid-sib and non-vegan friendly example: Lesion studies in non-human primates. When some researchers have worked with patients that have brain tumours, they’ve noticed consistent patterns of difficulties across subjects depending on where the tumour/lesion was. For example, frontal lesions lead more to executive difficulties (kid sib: the way you can focus, decide, organize your movements to get dressed in the morning), whereas medial temporal lesions lead to memory difficulties (kid sib: remembering what the doctor’s name is). This motivated some researchers to make precise lesions in non-human primates and they were able to localize certain specific functions to specific areas of the cortex; meaning, if a specific area was lesioned in the monkey, the monkey would no longer be able to do a certain task, even if they were able to do it before the surgery and on control tasks.
Delete“I wonder why anybody cares,” writes Fodor, commentating about the apparent uselessness of expending our scientific resources, funding, and years upon years of research on studying the brain, more specifically the neural localization of mental functions. To me, although some of Fodor’s rationale is valid, his complete dismissal seems ignorant of how significant progress in this realm has been in not only enriching our understanding of global brain function, but also in finding the causes and potential cures for countless neurological disorders, which serves the greater good of society and allows for novel interventions. It is a common trend to see neuroscientists disregarding philosophers and philosophers disregarding neuroscientists, which is frustrating to see as a cognitive science student. I am not saying I entirely disagree with all of Fodor’s points, and I certainly see his concerns about the possible futility of neuroimagery in the realm of reverse-engineering cognition. Rather, it is his complete dismissal and the lack of leniency in his writing that threw me off slightly, as it frankly sounded like just another philosopher with a personal vendetta against neuroscience.
ReplyDeleteYes, Fodor is over the top. But he did not question neuroimagery's contribution to clinical neuroscience. Can you give kid-sib an example of how it could help reverse-engineer cognitive capacity?
DeleteIn this reading, Fodor was getting at whether the abundance of neuroscience research coming out, specifically on functional specialization, is actually important or beneficial. This is a question I have considered before for a couple of reasons. Firstly, much of our research on brains using animals involves rather cruel and inhumane methods. Two of my friends in labs at McGill have been directly involved in the killing of mice. To me, this means that there is a real responsibility for the research that is being churned out from the use of animals to lead to actual human progress - development of new treatments and medications - otherwise these animals are being hurt and killed completely in vain. Secondly, a fair amount of the science that is constantly being generated goes completely unseen by the public. Many scientific papers are hidden behind paywalls and are difficult for the average person to understand due to the number of acronyms, long sentences, and academic writing. This means that science is mostly being read by other scientists, in which case all of the possible benefits of this knowledge being passed down to the general public are not reaped.
ReplyDeleteI certainly agree with you that invasive animal research needs a lot more justification than what people put into their grant applications, and welfare committees approve.
DeleteBut Fodor is questioning whether NONinvasive neuroimagery research provides information (what does that mean?) about how the brain produces cognitive capacity.
Yes, research in most fields of science is written primarily for other researchers to use and build upon. I'm a fervent Open Access advocate, but I don't think that the general public is interested in reading basic research. That's what popular review articles are meant to provide, isn't it? Raw research is peer-to-peer, not to-public.
On the other hand, everyone would definitely be better off if researchers wrote in a more kid-sibly way.
The public benefits of scientific research are reaped in its applications (and, to a certain extent, its educational use). But the target readership is the usership: fellow scientists. Research is collaborative and cumulative.
Information is the reduction of uncertainty, and Fodor is questioning whether research using noninvasive neuroimagery actually reduces the abundance of uncertainty we have about how the brain does what it does (spoiler alert: he doesn’t think so).
DeleteI agree that research is intended to be peer-to-peer, I suppose I more meant that because of the un-kid-siblyness of how research is written, combined with lack of access, the scientific world is very closed off from the public. While in some ways this is necessary and important, I also think that misinformation is rampant on the internet, and making online journals and peer-edited publications more accessible may aid with the cultivation of a better educated, informed society.
Fodor argues that much funding is allocated to brain-imaging technologies used to map out where and when neural responses happen in the brain. However, these technologies are more expensive than most other tools used in psychology research, both in time and money. I feel like it distributes resources to the “wrong” experts. For example, when studying memory-encoding processes, funding the development of new brain-imaging technologies would mean paying teams of engineers, computer scientists, neuroscientists... when they are not the most qualified experts to study the topic. Also, these hundreds of thousands of dollars allocated to ONE project could instead be distributed amongst multiple teams with different approaches.
ReplyDeleteGood points.
DeleteDrawing from Fodor's skepticism of the brain-centric perspective in understanding cognitive capacities, I think that it is important to reflect on the increasing reliance on neural networks modeled after the human brain. The mapping of "when" and "where" of neural activity doesn't necessarily illuminate the question of "how" and "why" of cognition, and that question is one of the core points of the reverse engineer problem. All in all, reading from the previous discussion, I keep in mind that research in all areas is still important--interdependent--and may inadvertently answer the problem.
ReplyDeleteNeural networks are used in modelling cognitive capacity, especially learning, but they are not very brainlike.
DeleteIf we want to investigate the mind-body connection, I think there are valuable insights to be gleaned from the study of functional localization, particularly with respect to language. If we assume that language has some relation to higher-order cognition and even consciousness, I think understanding its anatomical foundations in the brain provides valuable insights, that stretch beyond being merely scientific confirmations of things we already roughly knew as the author objects. Specifically, the strong connectivity of semantic regions of the brain (the anterior temporal lobe) to Broca’s area and the broader premotor cortex can, in light of reading 4a, be used as potential support for a mirror-neuron model of symbol grounding. Simply put, I do think functional localization can provide significant information on the nature of the mind-brain connection that might not always be immediately obvious - and I think that analyzing the symbol grounding problem with reference to the proximity of language centres to the premotor areas is a good example of this.
ReplyDeleteYes, but proximity (i.e., nearness) does not explain how.
DeleteI find Fodor's dismissal of the importance of scientific validation a little concerning. He humorously illustrates his point with Pavlov's response of "Now we know it -scientifically-," suggesting that seeking scientific validation for things we already understand is somewhat trivial. However, I think scientific validation is crucial in many cases. Our intuitions and subjective feelings about certain phenomena may not accurately represent their true nature. For example, Playing music does not equate to understanding music theory like a musicologist does. And driving a car does not equate to understanding car mechanics like an automotive engineer does. Our world contains numerous phenomena that defy our natural intuitions regarding how they function. For instance, the discovery that there is not a specific "teapot area" in the brain does not lead scientists to doubt the existence of mental states related to teapots. Instead, it suggests that their conceptualization of mental states may differ from how they genuinely operate. In my opinion, pursuing scientific validation for concepts we merely "intuit" is not entirely devoid of merit. Scientific validation can help us to better understand the world around us and to correct our misconceptions.
ReplyDeleteI agree with you that scientific validation can help us better understand the world around us, but it certainly doesn't always. I think Fodor's main point is that brain localization research is taking away a large chunk of funding from other avenues of studying the brain that could be more effective or useful (beyond helping surgeons not cut the wrong bits). Likewise, when he brings up the Pavlov example, the cost of doing such research involves "rather horribly, drilling holes into the mouths of dogs" to figure out ~scientifically~ something we already know. I could be wrong, but I don't think he has a problem with scientific research like brain imaging in general, or that science is the wrong epistemology here, he just thinks that research involving localization is not useful to solving the problems of cognitive science. I think he's got a point. That being said, fMRI is non-invasive and doesn't involve mistreating animals or slicing up any critter's brain, so it might be difficult to find a better option. There have been some suggestions above about computational modeling as an alternative, but I'm not sure whether or not that limits us to a computationalist view.
DeleteA little bit extra on my first point (specific to fMRI) about how scientific validation can, but doesn't always, help us understand the world around us: http://prefrontal.org/files/posters/Bennett-Salmon-2009.pdf
DeleteA researcher put a salmon in an fMRI to make a point that our standard statistical thresholds and methods can sometimes give wacky results. It certainly doesn't negate fMRI as a means of study, but it makes you think about taking scientific validation at face value, and it's a bit funny.
Computational modelling does not imply computationalism: what is the difference?
DeleteA computational model could be modeling the physical parts of the brain, too, not just the computational states. So as I understand it, computational modeling could be useful as a tool to study the brain and as a stepping stone to reverse-engineering, but it doesn't count as reverse engineering itself because even though it considers the physical properties of the brain, it's all still simulated. ... right?
DeleteFrom my understanding, the difference is that computational modeling would fall under the umbrella of "Weak AI" (or Strong Church-Turing thesis) in which computers are a tool to study cognition, and allow for simulation or modeling, but don't represent the full picture. By contrast, computationalism holds that consciousness can be fully explained as computations.
DeleteHaving computational models programmed to complete a task doesn't imply that all consciousness is computation.
I think the utility of computational modelling is a little bit more subtle than demonstrating whether the cognition is computation or contributing to reverse-engineering the brain. It seems to me that computational modelling is an incredibly effective way to test specifically how certain cognitive tasks are completed by the brain, rather than how they are instantiated in the brain itself. Cognitive scientists are very good at posing theories as to how a cognitive process is algorithmically completed in the brain, and computational modelling is equally as good at testing whether that theory has merit. If we can implement a computer program that completes tasks according to a candidate algorithm, and it completes these tasks to a degree representable of human performance, then perhaps the brain completes the task through a similar algorithm/process. Cognitive modelling side-steps the problem of neural localization and physical implementation altogether, and instead focuses on what Fodor would argue is a more worth research endeavor: how the brain does what it does, not where it does what it does.
DeleteI have a lot of sympathy for Fodor's annoyance over the over-promising of neuroimaging. Neuroimaging will not tell us what cognition is for one simple reason: knowing where, when and what does not tell us why. We can understand that a certain brain region has more electrical impulses when we think about a teapot but this will never tell us why those electrical impulses create the effect of thinking of a teapot or even if they do create it or are simply related. Fodor does go a bit too far in that he thinks there is no use in studying the brain in any way we can for cognition. We have good reason to believe that the brain and cognition are intimately connected and it seems eminently possible that studying the brain may lead to insights that could help us understand the nature of cognition.
ReplyDeleteIf where and when won't tell us HOW, how is it "eminently possible that they will, after all? Do you have a hypothetical example for kid-sib?
DeleteA good point I should have been more clear. I don't think that purely studying the brain can ever tell us what the nature of cognition is but I think that knowledge about the brain might help generate hypotheses or test them. Let's say we generate a hypothesis that cognition happens because of some physical dynamical process 'X' and we know that 'X' would require the system 'Y' studying the brain for the existence of 'Y' would help evaluate that hypothesis. On the other hand, lets say we find a system 'Z' in the brain which we know to be causally connected to process 'W'. We could then hypothesize that 'W' is responsible for cognition in some way and then evaluate that hypothesis.
DeleteFodor's position arises from his perceived irrelevance of "what" and "where" in our task of understanding cognition. The neuroscientific exploration of the brain's functional localization essentially only operates as a categorization method. Obviously there is value in understanding these processes in a clinical setting, but as far as we're concerned, there don't bring us any closer to reverse-engineering. Even as far as the easy problem is concerned, this neuroscientific inquiry does not progress that exploration. It certainly begs the question of what other neuroscientific methods exist that are more effective ways of not just eliminating false hypotheses but actually solving the relevant problems.
ReplyDeleteI don't understand your last sentence.
DeleteOne sentence that particularly struck me was: "But why (unless you’re thinking of having it taken out) does it matter where in the engine the carburettor is?"
ReplyDeleteI was reminder of Marr's 3 level analysis. If I'm not mistaken, he says that in order to understand the brain and for that matter, any machine, one needs to do three levels of analysis: computation, algorithm and implementation.
I think what Fodor's paper highlights is that neuroscience, by itself,
has very significant limits as to what it can explain.
Adding Marr into the mix, it seems neuroscientists "skip" a few steps, going straight to the hardware or the implementation! Back to the first quote, I think this highlights neuroscience's intense focus on the hardware...
Hello Emma, this paper also reminds me of Marr’s three-level analysis. Similar to Fodor’s advocacy, Marr’s theory points out that any level alone is inadequate for understanding the whole system, and we should pay equal effort to all three levels. Obviously, people are paying way more attention to the implementation level.
DeleteIn my opinion, studying on different levels should complement each other. Taking mirror neurons as an example, by simply neuron-imaging studies, we can only know its function of movement imitation. While we try to discover the mechanism behind, i.e. the algorithm, we learn that it helps with intention understanding, and involves more functions.
AI systems have made significant advancements in translating brain activity into continuous text streams, using fMRI scan data while individuals listen to or even silently imagine stories. These systems employ decoders trained to associate brain activity with meaning, utilizing large language models like GPT-1, a precursor to ChatGPT. Additionally, other models can reconstruct specific images from fMRI data corresponding to the part of a movie the participant was watching. This progress might lead us to a form of "mind-reading," an advancement made feasible by the utilization of MRI to map cognitive functions to specific brain regions. While this approach doesn't reverse-engineer the brain, it has evolved from mere functional mapping to the decoding of brain activity, warranting its consideration rather than dismissal.
ReplyDeleteWhat I find particularly fascinating is how AI models like ChatGPT, whose inner workings remain partially elusive to us, are capable of decoding and assigning meaning to neural activity – a process that mirrors our limited understanding of how meaning is grounded in the brain. It's perplexing because we're creating systems that can arguably pass T2, yet the mechanisms behind their success remain enigmatic (mystery of the big gulp and whether it allows them to “cheat” through recognizing syntax patterns). These systems can decode our neural signals and assign them meaning, and can seem to assign a form of meaning to the text they generate (not just statistical parrots), but we cannot grasp the missing piece of the puzzle; where does this meaning assignment come from!
Fodor believes that studying the localization of brain functions is a waste of resources and not actually effective in understanding how cognition works. He thinks that knowing "where" in the brain certain functions occur doesn't necessarily answer the deeper questions about how the mind works. However, I still think that doing research in all areas, and just doing what we can in general is still important to improve our understanding about the human body. There might be limitations to it, but it might lead us to a discovery that helps us get a step closer to being able to reverse-engineer our brain.
ReplyDeleteHi Andrae! Although I do agree with you on the premise that studying the localization of brain functions is not a complete waste of time and potential, I must say I do agree with Fodor in a sense as well. When it comes to funding and pure man power (regarding the availability of grad students and researchers), it definitely does not seem like the best use of time or resources to continue studying brain localization using brain scanning techniques. It almost seems counterintuitive to continue this type of research that has yet to yield any significance when it comes down to understanding human understanding and consciousness. There are definite limitations to continuing down this narrow minded road of localization and function. With this we have not been able to determine the why or the how, just that it is. It makes me wonder whether these localization studies are favored for the hope that one day it will lead to a revolutionary finding, or whether they are just favored because they yield success for the intended findings.
DeleteFodor's critique of brain imaging research as data collection without a clear conceptual framework may overlook the incremental and exploratory nature of scientific inquiry - looking into how one does what he can. As in the subject of Cognitive Science, I've learned that studying the brain is undoubtedly essential for gaining insights into how it functions to support various mental processes. While identifying specific brain regions for specific tasks is valuable, it doesn't provide a complete understanding of the brain's inner workings. To truly comprehend how the brain operates, we must delve deeper into the mechanisms that underlie cognitive processes, examining neural networks, synaptic connections, and information processing. One example I could think of is putting together a recipe, first the ingredients, then the steps in cooking. In essence, brain studies offer valuable pieces of the puzzle, but a comprehensive understanding of how the brain achieves its cognitive feats requires a holistic approach that combines neuroscience with cognitive science, psychology, and other disciplines.
ReplyDeleteOn that topic, how do we best explain to others, like Fodor, why and how we study what we study?
DeleteI think this view is one-sided. First, a PhD friend of mine once said pessimistically that most of the current clinical neuroscience research is, pessimistically speaking, “raising experimental mice.” The one-year cycle is almost the entire academic career of a master/phd school. Neuroimaging is a good practice to simulate the brain, speed up research, be more humane and achieve T3 level in biology. As Dr.Harnad mention in previous post, I hope to explain such problem in detail.
ReplyDeleteThe current method of combining anatomy and neuron imaging can better model the brain and realize 3D printing. Further progress in materials science may gradually realize T3 level reverse engineering in biology. It is now possible to thermally print mouse brains, and a popular field of deep learning is to generate 3D structures through photo reverse engineering, called NERF (Neural Radiance Fields). Therefore, neuroimaging is the most likely combination to help realize reverse engineering with the help of state-of-art AI. This is more humane because we can do it with neural network technology without having to do a lot of invasive dissection. Similarly, this article was published in 1999, so it is not technically advanced enough. I sincerely invite everyone to visit H01 dataset, https://h01-release.storage.googleapis.com/gallery.html to see such accuracy. And amazing work. Is neuroimaging really of no value in reverse engineering? I don't think so, it's worth every penny.
Please visit H01 dataset!
I think Fodor raises some important points particularly concerning resource(time, money and expertise) restraints and waste. While I agree that reverse engineering from an fMRI doesn't seem plausible, I personally don't find that this should make neuroimaging any less interesting. To call it boring and frequently a waste of time because it doesn't bring us closer to understanding consciousness seems like a silly point to me. Yes, it is no way of reverse engineering as it stands right now, but many other techniques in the field haven't yet had much success yet either. Does this make them boring? Useless?
DeleteTo me, the knowledge we are acquiring from neuroimaging scans is still of incredible interest, particularly in clinical populations in highlighting differences in brains of those with disorders(be it mental or neurological). Often it is these differences that can tell us the most about what certain areas of the brain are responsible for.
The comment above suggests that the more understanding about brain activity, the more likely we are to creating T3(though I think they may have meant T4 as it revolved around biological tissue?). They mentioned it is already possible to thermally print mouse brains, and that photo reverse engineering is becoming popular. To me, it seems ages away that we would manage to achieve this with a human brain(much more complex, and less certainty on how it works), but it highlights the importance of understanding functional brain activity and brain structure in reverse engineering a T4. This begs the question of course that we discussed in class of "Is a T4 even necessary", but either way, I thought this was an interesting point.
*Please correct me if I am wrong and help me understand why this would be T3 and not T4.
Thanks for replying me, and reverse engineering from FMRI is actually helpful. Researchers can reconstruct images, music and texts. Here are some interesting papers to read, several groups include one I participated paid attention to it. This might help understand the computation role and connect modern neural network to determin how computation works in our brain.
DeleteMusic:https://arxiv.org/pdf/2307.11078.pdf#:~:text=We%20reconstruct%20music%20from%20fMRI,resembles%20the%20original%20music%20stimulus.
images:https://www.biorxiv.org/content/10.1101/2022.11.18.517004v3#:~:text=Reconstructing%20visual%20experiences%20from%20human,models%20and%20our%20visual%20system.
DeleteNLP: https://www.nih.gov/news-events/nih-research-matters/brain-decoder-turns-person-s-brain-activity-into-words
DeleteI think that Fodor's critique, while too strict at times, is necessary in the field of neuroscience. These are questions that I think a neuroscientist must hold in mind, especially when considering conducting studies on animal models. Is this question essential? Is the experiment I'm about to run capable of providing an answer to this question, or restricting the parameters in which this questions applies? It still seems pretty ludicrous to reject neuroimaging altogether, especially considering the fact that neuroimaging is not the naive "thinking about teapots has a place in the brain, is it close to the place that's activated when I'm taking a nap?" that Fodor makes it out to be. The study of the brain is more and more dependent on programs mapping out widespread patterns of activation in human models, and recording neural patterns of activation, or the genetic underpinnings of neurons in animal models, and importantly, all of this (or good neuroscience that is worth doing) is based on behavioral models. So, yes, the question "does brain function have a localisation" is essential, and it's a question that got the field where it is today, but the field is in many ways past that now.
ReplyDeleteEven though Fodor had extreme and not so relevant arguments against localization research, sometimes I feel the same when reading neuroscience papers. Lots of neural areas are related to specific functions, as well as neurotransmitters (e.g., for a specific psychiatric condition almost all neurotransmitters are involved, it doesn't seem that they're "localized" at all.) I don't get how it can be linked with reverse-engineering cognitive capacity, since we don't know the causal relationship between 1- a brain area lighting up in an fMRI and 2- its reverse-engineerable capacity. It seems that localization of brain functions still has great utility for clinical/research areas, yet I don't quite understand how it can serve a role in trying to solve the easy problem, if we don't figure out why is one function is related to a particular area. (I personally struggle to see how the purpose of a function/the answer to “why” can just be neurological.) Neuroimaging studies are interesting and intriguing for many reasons, but I feel like it drifts away from having solid theoretical questions based on causality, since most research seem replications of correlational relationships between functions and specific brain regions.
ReplyDeleteWhile Fodor made some solid points in his article, mainly about allocation of research funding, I disagree with him that there’s no value in studying functional specialization. I understand that from his perspective of studying cognition, it may not matter where a function is performed, just that we can do it, however for those outside of his specific field of study these are important topics to research. He mentioned the clinical relevance to neurosurgeons which I feel cannot be overstated when discussing the importance of this research, however I believe that understanding where functions are localized would ultimately help in the mission of “reverse-engineering” the brain. Even if we don’t want to make a 1:1 copy of the brain, understanding the various pathways that are activated during cognition would surely be useful, right?
ReplyDeleteWhile Fodor does make a compelling points, it seems like the philosopher has a quite superficial understanding of the study of neuroscience. I agree with him on many levels, but as someone that studied some neuroscience, I think he is omitting, either for argumentation purpose, or ignorance, some subtilities of this science. One of those might be, to connect back to the last skywriting, the role of mirror neurons, thus at a specific locus in the brain, which, by connecting them to the process of ‘the gestural theory of speech origin’ gave us some crucial understanding as to how language evolved. Which, as the author is a philosopher, might have some repercussions on how we think of language in a philosophical way like : what is the nature of language?.
ReplyDeleteThis sentence in the article impresses me:“Science is expensive, and it’s largely publicly funded, and there’s never enough money to do all the research that might be worth doing. In particular, brain imaging is expensive compared to other ways of trying to find out about the mind. If you put your money (which is to say: our money) into the elaborate technology required to establish neural localisations of mental functions by imaging techniques, you almost certainly take it out of other kinds of psychological research. Likewise in respect of the time and money that is required to train people to do the science; graduate students, too, are a limited resource.”(Fodor 2)
ReplyDeleteThis reminds me of a social phenomenon. It is obvious that science is making human progress. Science is the basis for creating value and creating a better future. Scientists have contributed a lot to the world, but why do they get so little in return? Compared with Internet celebrities and celebrities, although they have made achievements within their scope of work, sometimes they only need to attend a 2-hour event to earn tens of millions of dollars, which is dozens more than what scientists pay for a project.I assume this question as the relationship between value and income.
I think that this unequal income, as well as the social and government's low investment and funding for science, has also led to fewer and fewer talented people who are willing to contribute to scientific research. Although you can explore something if you love it, the bottom line is, after all, we are human beings and we need to live.
A better word would be to make the results of your projects provide more and better value to the world we live in, which requires more exposure, but the reality is not like this.
The above are my personal thoughts and feelings, as well as the question I have always wanted to explore, how to make scientists more and more rewarding.
There’s something really interesting about the following claim: “If the brain does different tasks at different places, that rather suggests that it may do them in different ways.” While describing the empiricist perspective that aligns with the idea that the brain is viewed as “equipotential”, Fodor made me think about our discussion last class on neuroplasticity (the apparent capacity of any intact part of a functional brain to carry out the functions which are lost by the destruction of other parts) and weak equivalence. From a weak equivalence perspective, even though cognitive processes might appear different or be organized differently within the brain, they could still lead to similar behavioral outcomes or task performance. He then tried to redirect the debate pointing out that understanding where cognitive processes are located in the brain doesn't matter significantly for the study of cognition. I feel like he has a point because even though neural networks and deep learning algorithms draw inspiration from how the brain processes information, they are not very brain-like and the money could be indeed invested for different approaches to understand “how” and “why” we think the way we do.
ReplyDeleteThis article brought a different perspective to the work that is often seen as the prerequisite of treating a disease: mapping the disease and isolating it to specific targetable regions. For instance, when trying to model the neuroanatomical correlates, the approach is to assemble a model of the disease within the brain based on a large amount of data of brain regions that are correlated to affected behaviors. However, as mentioned, such studies only locate “where” the disease starts as well as in response to what. The problem remains that this is still just a correlation. This information is useful to an extent for clinical purposes, as mentioned by others, however it may take many trial and error repeats to find the correct treatment: we don't know exactly at when something went wrong in the brain, so we need to test many different targets and dosages hoping we’ll find the right one. I thus agree with Fodor that localization on its own does not help reverse engineer a cognitive process, however perhaps with the help of computation we can go through a significantly larger amount of trial and errors (simultaneously analyzing the large amounts of data to search for a pattern) until we find an effect. Then, despite not knowing exactly why its working, we do know how the medication works on its own, and we can use that to take a step towards deciphering how the cognitive process on which it had an effect may work.
ReplyDeleteA rough analogy to the car example - if you didn't know anything about how a car worked, but you needed to find the right gasoline to make the car drive. You could potentially just test every different type of gasoline (assuming the car doesn’t break down in the process) until you were lucky and found the right one. Then, knowing a lot about that type of gasoline, you could infer something about how the car works. Of course this is a very inefficient process, but perhaps with computational modeling it can be made faster and less painful.
Fodor’s sentiments conveyed in his enjoyably conversational and snide writing style I think can seem on the surface like big claims with big calls to action to change the way we do neuroscience but are actually quite simple and—though interesting—not revolutionary. What he means to say about the recent (and still ongoing) trend of functional brain mapping is that it’s not going to help us find the ‘how’ of consciousness, though of course, it will help the medical fields. I think he expects more pushback about that claim but personally I think it’s evident enough through simple reasoning (i.e. his slightly flawed car example) but that does not at all mean we should redirect efforts in neuroimaging away from functional brain-area studies because they serve the field of medicine. So to me it just seems a quaint and fun read that serves to change people’s perspective about what our reasoning for doing this kind of brain-imaging science should be.
ReplyDeletesuch a great read! main thoughts on this: New York Times probably likes publishing neuroimaging studies because the results are more accessible to the average New York Times reader. it is much more interesting and satisfying to look at a rendering of the brain where certain areas are "coloured in" than to look at indecipherable graphs. this isn't the main point of this article of course, but I think that it is interesting to look at how what is accessible to the reader (or the tax payer) guides which research is done. The one thing this diary entry is missing is the phrase "stamp collecting"
ReplyDelete
ReplyDeleteI found this text to be quite intriguing. Until now, I had always perceived technological advancements in Brain Imaging and Functional Localization through Neural Imaging in a positive light. The author presents compelling arguments, emphasizing the cost and resource-intensity of such research, the potential influence of technology availability on motivations, and the significance of pinpointing exact brain locations from a materialist perspective.
However, the reference to Bernard Shaw's fable, "The Little Black Girl in Search of God," left me somewhat skeptical. While his experiments with classical conditioning did prove something “we already knew, they also laid the foundation for behaviorism, which is still relevant in psychotherapy to this day! Moreover, his discovery not only aids our understanding of learning and memory processes, but they also allowed practical applications, notably in the understanding of phobias, as well as their treatment.
Furthermore, those technologies enable us to explore the repercussions of neural damage or dysfunction, opening doors to the treatment and management of neurological disorders. Thus, while I concur that theoretical understanding is integral to scientific inquiry and merits increased focus, I contend that data obtained from neural imaging remains crucial for theory development, testing, and treating or preventing neurological disorders. These technologies serve as empirical instruments that bridge the gap between the abstract realm of philosophy and the tangible world of physical brain structures and activities. Empirical and theoretical understanding should ideally inform and enrich each other in a balanced research approach.
I sympathize deeply with many of Fodor's ideas, particularly suspicion of neuroimaging. He brings up how it doesn't explain the why, although to me it's equally important that it doesn't show "how". Knowing where the carburetor is in an engine gives one very little to understand how it actually works.
ReplyDeleteAlthough what most stuck with me was his empiricist/rationalist issue.
Plato, in many ways THE rationalist, says Timaeus 47a:
" Vision, in my view, is the cause of the greatest benefit to us, [...] and has given us not only the notion of Time but also means of research into the nature of the Universe. From these we have procured Philosophy in all its range ."
The quote above fits in a larger argument about how vision is the first gift which can allow humankind to reason. Noticing regularities, position, distances of the stars was a necessity for astronomy, which he saw amongst the highest of reasoning.
Neuroimaging, being nothing more than extended vision, could then in this rationalist frame, be seen (ha) as the first step to allow for further inquiry onto the workings of the mind. After all, observing the retrogradations of planets allowed us to infer that the universe doesn't center on earth.
I'm not sure how useful this skywriting is, if at all... probably not. But going back to the origins of rationalism is where my mind went in all of this, which endears me to the Neuroimaging, which I am still at an impasse over
To my understanding, in “Why the brain?” Fodor is highly critical of neuroimaging and correlational research because it lacks the ability to address “how” exactly certain mental processes arise and work. In connection with my previous skywriting on mirror neurons, I had similar thoughts regarding correlational and brain mapping studies— I had mentioned how although associating parts of the brain to functions like the ability to understand was an interesting first step in research, the results still seemed to leave more questions than answers. For example, observing certain actions leads to high activity in mirror neurons; therefore we know there must be some sort of relationship, but what exactly is this “activity” and how does it work?
ReplyDeleteI personally would still say that neuroimaging has some important applications worth investing in, especially in applications like medicine. Similarly, Fodor doesn’t seem to completely dismiss the importance of neuroimaging and the localization of brain functions— he briefly mentions its importance in medical fields like neurosurgery. However, whether it is relevant to understanding the ‘how’ questions that Cognitive Science involves, is the bigger debate he introduces.
In this reading, Jerry Fodor questions why functional localisation of the brain is so important for discovering how our minds work. In the passages, Fodor states, “Likewise for anyone who cares about how much of the mind’s structure is innate (whatever, exactly, that means). If you think a lot of it is, you presumably expect a lot of localisation of function, not just in the adult’s brain but also in the infant’s. Whereas, if you think a lot of mental structure comes from experience (whatever, exactly, that means), you probably expect the infant’s brain to be mostly equipotential even if the adult’s brain turns out not to be”. This quote shows how two differing perspectives on the mind’s structure (innate vs. experience) bring out two differing conclusions about our mind. As I was reading this and thinking about the answer to the question, it reminded me one of Massimi’s argument in her book ‘Perspectival Realism’. Massimi argues that perspectival representations open a window on reality, which leaves room to make inferences about the space and what’s inside it. In a similar manner, by figuring out the localization of functions in our brain, it may not directly give us an answer to how our minds work, but with more accumulated information about brain’s functional localisation, it can open up different perspectives / points of view that can lead to small discoveries of how our minds work.
ReplyDeleteFodor allocates that a more diverse exploration of mental functions is needed and I would agree. I found his points on empiricism and rationalism to be quite interesting. Empiricism emphasizes the equipotentiality of the brain and that several parts of the brain could be responsible for certain mental functions. It favors 'nurture'/ experience and learning to shape cognition. On the other hand, rationalism emphasizes functional localization (which Is what modern neurological research is focusing on). The greater philosophical conversation about our mental processes is of significance in ongoing explorations of how the cognitive processes work-- just as studies about the brain's physiological processes do (this is important as we do not know of the homogenous/heterogenous nature of mental functions currently).
ReplyDeleteI'm not sure why Fodor thinks that it should be "good enough" for neuroscientists to know whether mental functions are neurally localized, nevermind where they are neurally localized. The quite graphic analogy of "drilling holes in the mouths of dogs to show that expecting food makes them salivate" despite "already knowing that" just to "know it scientifically" cannot be applied to the majority of the goals of neuroimaging. Indeed, I believe that most neuroscientists pursue a study in functional localization only if this study could teach us something we DON'T know... Sure, they will go into it with a hypothesis, but almost never something as obvious as "dogs salivate because they're expecting food". Usually, there is good reason to believe that the study might prove the hypothesis completely wrong. Thinking back to reading 4a, for example, we have seen the same areas of the brain light up for producing speech, hearing speech, writing and reading, and thus have good reason to believe that these areas must be part of the system responsible for language. We also know that writing and reading engage other areas of the brain that are not engaged in speech production or comprehension. And then we run a neuroimaging study on whether sign language engages those latter areas used in written language but not speech, perhaps hypothesizing that written language does in fact cause these areas to light up, which at least in the past was reasonable to believe. After seeing, however, that these areas do not light up, this study would suggest that speech and sign language are each more fundamental than the visual word form processing required for reading and writing, and that sign language is more similar to speech than it was previously thought to be.
ReplyDeleteHi Jordan, I just wanted to chime in on your first point as I was going to mention something similar. I completely agree; Fodor's view that we may be discovering what we already know, just to "know it scientifically," cannot accurately apply to many of the areas that neuroimaging is trying to elucidate. I think there is a marked difference that sets the study of functional mapping (and possibly consciousness) through various scientific methodologies apart; we effectively do not know what we are tangibly looking for and lack the clearer frameworks based on existing knowledge that many other disciplines possess. With this said, I think Fodor is missing this factor in his oversimplified view.
DeleteI think Fodor is definitely overstating in his claim that neuroimaging is basically not very useful in reverse-engineering the brain. In order to understand the brain to the level that it can be reversed-engineered, I don't think researching when and where activity in the brain occurs is completely useless because it could give us a foundation of how thinking then occurs. It is hard to get to the how without knowing the when and where. I think that especially since neuroimaging is also non-invasive, it can be particularly useful. Since reverse-engineering the brain has not yet been done, who is to say what exactly will be helpful in leading to that?
ReplyDeleteFodor examines some valuable points in this essay. Brain imaging and associated training, analysis, etc. is all incredibly costly, and it is important to evaluate exactly what we are getting out of it.
ReplyDeleteBy localizing the mechanisms and pathways, and what is interacting during specific tasks, we can gain more information about the “how” of things. It is also important on the non-philosophical end of things where neuroanatomy research is essential to finding certain neurological disorders and their treatment, it is not all done in the name of ~finding the cells where consciousness lies~. But Fodor brings up a good point of - we don’t even really know that we are going to be able to find what we are looking for, and if consciousness and other mental activities are even possibly localized in the brain. The research looking into this relies on the baseline assumption that we can. And if we can, Fodor claims it does not really matter. I do not completely agree with this, but I do think some functions should be prioritized over others. However, I think all of these studies should be taken with the grain of salt that they may not find what they are looking for, or that it may be different than what current findings present, as to not get caught up in what we think we “know”.
I'd argue that regardless of the fact that the research may never localize consciousness and other mental activities in the brain, the benefits of mapping brain functions are overwhelmingly positive. You touched on the value of mapping brain function in order to treat disorders, which I believe itself already warrants funding for such research, however, it is immensely valuable when treating brain injury as well. The ability to predict what type of functions may be impaired by different injury/trauma to the brain is incredibly valuable information for a whole host of different fields and industries. Ultimately, regardless of localizing consciousness, brain mapping research has the ability to benefit all of society in many different ways.
DeleteThe fundamental question Fodor asks, as many have said, is "whether mapping WHEN and WHERE activity (whether local or distributed) occurs in the brain can help reveal HOW the brain produces our capacity to DO what thinkers can do". His hypothesis, it seems, is that this desire to localize all brain abilities is not as effective for scientific understanding of cognition as one might hope, and its only apparent use is in knowing whether different brain abilities are localized to specific regions at all. To some extent, Fodor is right, that knowing whether localization exists is the first, and fundamental, insight we must research.
ReplyDeleteHowever, one point I have not seen mentioned, and is quite fundamental to me, is the cost of energy transfer in the brain. If two regions of the brain are very close, it indicates that they are probably relevant to one another, since information transfer from one to the other (in the form of neurochemical signals) is less energetically costly, than if these regions of the brain were farther apart. Therefore, if we find out that regions that seem quite different are quite close to one another, or if regions we think similarly about are quite distant, this should certainly give us insight into the actual functioning of different brain regions. This, of course, assumes that there is such a thing as functionally different brain regions; but Fodor himself agrees that the empiricists, who defended this idea, were proven right by neuroimaging research.
Therefore, although Fodor's concern is very well justified, and much of neuroimaging, today, seems useless at best and actively harmful at worse, it is false that some mapping of specific brain structures, and the implied distances between different brain structures, cannot give us insight into the actual functioning of different brain regions, and the brain as a whole.
Although I think Fodor raises a good point, and does so clearly and accessibly, I think that his ultimate argument is misguided. He struggles with the state of neuroscience being one preoccupied with correlation, which does nothing to directly get at either the easy or the hard problem. I think this point stands. But, I think that there is value in having correlational data as a supplement to larger questions, which are what I believe a significant portion of neuroscientists actually work on. I am suspect of the argument he makes at the end of the paper, where he suggests that neuroscientists have no real direction or clue of what they’re asking. Certainly, the questions can get bogged down, but I think that the fact that he believes this is a symptom of problems with scientific communication, rather than with scientific inquiry.
ReplyDeleteJerry Fodor asks why it matters to know where exactly things happen in the brain. I believe one reason is that, in order to determine who's right between empiricists and rationalists, investigating the location of different concepts in the brain is a useful tool. It's possible, for example, that concepts like 'dogs' and 'German Shepherds' are both obtained through the same processes but are acquired differently than the concept of 'elephant.' Examining which part of the brain is active when thinking about one concept versus another can be useful in this regard. Furthermore, while it's possible to study these phenomena without directly examining the brain, if we agree that scientists are materialists and that each mental state has a neural equivalent, then discovering the locations of these mental states in the brain provides stronger evidence for their existence.
ReplyDelete**BLOGGER BUG**: ONCE THE NUMBER OF COMMENTS REACHES 200 OR MORE {see the count, at the beginning of the commentaries] YOU CAN STILL MAKE COMMENTS, BUT TO SEE YOUR COMMENT AFTER YOU HAVE PUBLISHED IT YOU NEED TO SCROLL DOWN TO ALMOST THE BOTTOM OF THE PAGE and click: “Load more…”
ReplyDelete________________
Load more…
________________
——
After 200 has been exceeded EVERYONE has to scroll down and click “Load more” each time they want to see all the posts (not just the first 200), and they also have to do that whenever they want to add another comment or reply after 200 has been exceeded.
If you post your comment really late, I won’t see it, and you have to email me the link so I can find it. Copy/Paste it from the top of your published comment, as it appears right after your name, just as you do when you email me your full set of copy-pasted commentaries before the mid-term and before the final.
——
WEEK 5: Week 5 is an important week and topic. There is only one topic thread, but please read at least two of the readings, and do at least two skies. I hope Week 5 will be the only week in which we have the 200+ overflow problem, because there are twice the usual number of commentaries: 88 skies + 88 skies + my 176 replies = 352!. In every other week it’s 2 separate topic threads, each with 88 skies plus my 88 replies (plus room for a few follow-ups when I ask questions.
I think this reading was a little confusing to me. I understand that Fodor is poking and prodding at brain research, and I guess he was highlighting that our increased interest with employing brain imaging is based on an aim to mapping brain functions to brain structures, or the localization of behaviors and functions to brain structures. But I did not understand the empiricism vs functionalism debate to an extent. I understand that functionalists obviously tend to describe functions of the brain as corresponding, or localized to certain structures, but what do empiricists believe then?
ReplyDeleteThis comment has been removed by the author.
DeleteHi Mal, I will attempt to answer your question. From my understanding, correct me if I am wrong, functionalism focuses on how mental processes contribute to adaptive behavior, rather than its adherence to brain structure. Fodor is critical of the reductionist approach that seeks to map mental functions one-to-one onto specific regions of the brain. He argues that the brain is not a ‘neat and tidy’ modular system where each function corresponds to a specific area. Empiricists may be more open to the idea of localization, as long as it can be empirically supported through observations and data analysis. They are associated with the idea that knowledge comes from felt sensations (‘sensory experience) and observable behaviors.
DeleteFodor’s writing on the current trends of brain localization and specialization is an interesting one. He makes the claim that determining functional localizations in the brain is not really relevant in understanding the processes of the brain. In addition, I think that if we were to consider neuroplasticity, the ability of the brain to fill the role of areas that have been damaged/destroyed with other areas, Fodor’s point is well justified. Although neural imaging can be useful and will probably play a role in reverse-engineering cognition (how and why we can do what we do and feel), it is certainly not the entire answer. My question though is, if we cannot really find much answers from studying brain localization, what can we study about the brain that will guide us closer to the reverse-engineering of cognition?
ReplyDeleteFodor’s paper is largely about opportunity cost. Fodor explicitly states his disapproval of the trend of brain imaging, looking at the phenomenon through a more philosophically tuned lens. It seems he’s less concerned with WHERE phenomena are taking place—in fact he thinks that too many resources are being allocated to this mission—and more concerned with HOW and WHY they are taking place. He ties a ribbon around this thought with a comment from his colleague: “You think we’re wasting our time, don’t you?” After reading this paper, it certainly seems he thinks so.
ReplyDeleteReading Fodor’s article, I kept thinking about Oliver Sacks’s book “The Man Who Mistook His Wife For A Hat”, in which Sacks discusses his patients’ (mostly neuro-psychological) symptoms and how treating these patients led to new discoveries about the brain and human behavior. I find that this is similar to what Fodor was saying in terms of not having a hypothesis. Sacks didn’t have a hypothesis, he had a patient, symptoms, and access to the brain of the patient. Sacks’ efforts were medicinal in practice but it did add a lot to the cognitive science field through his curiosity as well.
ReplyDeleteI understand how neural imaging might be seen as a scientist with a camera but no hypothesis. However, I do think, at the end of the day, it is a useless debate. There are findings that scientists have reached through neural imaging that are significant and that we can see in the Times. Obviously, neural imaging isn’t a waste of time, to answer Fodor’s questioning in the last paragraph.
We have understood almost everything about our bodies: how our heart is able to pump blood to our entire body, how our glands react to different physical stimuli, how we digest our food, etc. Years of scientific research and we are still not even a meter deep in understanding the mind. I think it is very important that all and every research method dedicated to understanding the mind should be encouraged. Especially if there is proof that it has already helped answer questions. There might be no hypothesis in neural imaging, like Fodor said, but there is a question to be answered.
Hi Nazlı, I think you have a great point on how Sacks' research gives us more understanding of the brain. But I think his research is mainly out of clinical purpose (i.e. to treat his patients), rather than investigating human behaviour. So Sacks did not aim to study the brain to bring more insight to Cognitive Science, his discoveries of human behaviours are just a by-product. And this is what Fodor is criticizing as he says "serendipity is a frail reed". You can't always rely on serendipity in scientific research. If the same discovery about human behaviour can be made using a different strategy than brain imaging, then why should we hold on to this specific method?
DeleteI think Fodor is oversimplifying brain imaging and localization of functions to phrenology. His description of brain localization gives me the feeling that he thinks localization is finding a brain region that is responsible for a specific mundane function (such as thinking about a teapot). This is more like the out-dated phrenology, not modern localization. Nowadays, brain imaging and localization show that the relationship between a function and the brain is more complex. It is not always one-on-one mapping, but involves complex neural pathways and associations between different cortices.
ReplyDeleteI understand that Fodor’s criticism of brain imaging is that it doesn’t really provide any new insights in Cognitive Science, but merely provides an additional prove to what we already knew. Fodor points out that scientists using brain imaging for functional localization are finding answers without a question. But I think this is because scientists don’t have enough information about the brain to formulate a question, and functional localization is gradually finding its question through its research. I do agree with Fodor that if brain imaging research continues to drift aimlessly, then it would be a waste of money. The current problem is that we do not know when we can find a hypothesis.
I agree with Fodor’s criticism of neuron localization. For example, the technique of Optogenetics allows us to establish correlations between specific groups of neurons, or even individual cells, and certain functions. However, it's too costly and doesn’t serve any clinical purpose. It only establishes the correlation, but does not help to explain how the function is achieved.
ReplyDeleteOn the other hand, I learned that geneticists have developed the entire genome map, which is really helpful for understanding genetic expression. I’m wondering if one day we can map out every brain cell, how it can help us to answer the question of cognition. I think it would be helpful to understand the brain of an organ, but not , because it’s not reverse engineering.
After reading Fodor’s paper, I felt like his main point was that the purpose of neural imaging and where structures were located was kind of useless. Fodor essentially asks what part of understanding your mind would you have failed to understand if you didn’t know the location of where the processes of your mind happened. When I was reading this, I kind of agreed with his point, until I thought about the problem of reverse-engineering. I think that it is important to understand the location of specific anatomy in the context of trying to reverse-engineer human cognition. In general, I think that Fodor’s main frustrations were not that neuroscience is useless, but rather rooted in the fact that people get obsessed with miniscule details about structure and location when that is not even confirmed to be the cause of cognition.
ReplyDeleteThe study 'The functional architecture of the ventral temporal cortex and its role in categorization' reveals three key insights about its role in visual categorization. First, the VTC's functional representations are orderly and aligned with brain gyri and sulci, indicating a consistent organizational structure. Second, a spatial hierarchy within the VTC corresponds to an information hierarchy, suggesting that different spatial scales in the VTC enhance category processing efficiency and flexibility. Third, the VTC exhibits superimposition of representations, facilitating both integration and segregation of information for efficient processing. These findings highlight the importance of VTC's structural organization in supporting its computational and representational functions in visual categorization.
ReplyDeleteIn the article, “Can neuroimaging reveal how the brain thinks?”, Harnad used an analogy between the brain and a computer used to show how neuroimaging (observing the hardware) can’t tell us how cognition is done (the software) requires assuming that cognition is computation. But it seems that cognition is in fact the result of a dynamical process (electrical, chemical). So, why can’t observe the brain help us investigate cognition just as it is the case with other vegetative functions? It is said that “the causal mechanism [underlying cognition] can’t be just vegetative either. Cognizing is not just moving.” because the difference is that cognition is “not only what we actually do, but also what we are able to do”. But I still don’t understand why if cognition is dynamic (just as vegetative functions), we wouldn’t be able to find causal mechanisms between the brain and cognition.
ReplyDeleteI propose we consider neuroimaging not as an end but as a means to a more broader exploration. Although it is true that locating a function in the brain doesn’t necessarily unravel the mystery of cognition, this knowledge acts as a critical piece in a larger puzzle. It helps guide us through the mapping of the brain as a whole. This however leads to a critical point of contemplation: should we pivot our focus from the where and instead to the why and how of the brain functions. Instead of just pinpointing areas of activity, can we leverage these findings to delve a little deeper into the mechanisms of all of these processes? How could we use the insights gained from neuroimaging to create a more integrative approach by combining it with other scientific disciplines to help better understand the brain as a whole?
ReplyDeleteFodor's candid reflection on the chase for brain images compels me to question the essence of what we're actually capturing. I can't help but wonder if we are indeed collecting data that propels our understanding, or merely basking in the glow of colorful scans. Fodor's unease resonates with me, underscoring a critical examination of our scientific pursuits. Am I, along with my peers, becoming too captivated by the quantity of data rather than its qualitative leap towards unraveling cognition's complex tapestry? His anecdote serves as a sobering reminder to ensure that each neural snapshot taken is a step toward clarity, not just another picture in an ever-growing album.
ReplyDelete