Leiter Reports: A Philosophy Blog

News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.

  1. VBBaltzly's avatar
  2. David Wallace's avatar
  3. Nat Hansen's avatar
  4. Steve Innsbruck's avatar
  5. Nat Hansen's avatar
  6. Paul Canis's avatar
  7. David Morrice's avatar

Rosenberg on the “threat” to our self-understanding posed by neuroscience

Brian Leiter

An accessible piece; an excerpt:

We’ve built our cultural, legal and political institutions on this theory that people’s actions are caused by choices made rational in the light of their beliefs and their desires….

But here’s the problem: the theory of mind we call carry around with us and use every day has no basis in what neuroscience—Nobel Prize winning neuroscience–tell us about how the brain works. Neuroscience has revealed that the theory is quite as much of a dead end as Ptolemaic astronomy. It’s been around for such a longtime only because it was the predictive device natural selection came up with, in spite of being fundamentally mistaken about how things were really arranged.

Eric Kandel, John O’Keefe, May-Britt and Edvard Moser, won Nobel prizes in 2000 and 2014 for experiments that showed exactly how the brain records information. Their work revealed it doesn’t do it in anything like the way the theory of mind says it does—in statements that represent the way the world is (beliefs) and ones that represent the way we want things to be (desires). This research program began with HM, the patient famous for being unable to acquire or store beliefs because of a lobotomy that went wrong and destroyed his hippocampus. The irony of this research is that it ended up showing that no ones’ brain acquires, stores, and uses information in the form of beliefs and desires….

But conscious experience is continually shouting out that belief/desire psychology is exactly how the mind does work. Introspection all by itself seems to refute the notion that we don’t have beliefs and desires with content that represent what we want and what’s available to attain our wants.

Alas, ever since Freud psychologists have diagnosed the illusions, delusions and confabulations in the mind’s eye and the mind’s ear, in the flow of experiences, feelings and sensations passing through consciousness. The theory of mind is just another one of these illusions, useful for survival and success in the Pleistocene, but a blunt instrument of limited predictive and explanatory power. It emerged out of the more fundamental mind-reading ability we share with other species and used to track predators and prey, threats and opportunities. That undoubtedly inborn ability combined with our unique gift, language, to generate the theory of mind. By colonizing consciousness spoken language turned it into a monologue of silent speech, tricking us that the meaning of spoken words is given by thoughts’ content when its just silent sounds passing through consciousness. Neuroscience shows that that in our brains the neural circuits neither have nor need content to do their jobs.

Thoughts from readers knowledgeable about this research?

Leave a Reply

Your email address will not be published. Required fields are marked *

28 responses to “Rosenberg on the “threat” to our self-understanding posed by neuroscience”

  1. Chandra Sripada

    First of all, folk psychology includes much more than beliefs and desires – it countenances categories such as goals, habits, urges, pleasures/pains, intentions, attention of various kinds, slips, boredom, memory, deliberation, know-how, mind wandering, emotions, moods, decisions, self-control, imagery, imagination, mental effort, and on and on… Stepping back and looking over the broad sweep of the last two decades of computational neuroscience, I would tend conclude the exact opposite of Rosenberg: the most striking thing about this huge body of work is the way it (for the most part) vindicates reams of folk psychological categories and explanatory patterns, including all of the preceding categories….

    Reply
  2. Mohan Matthen

    I can't see how these specific results threaten our self-understanding any more that the ancient realization that the mind is realized in the brain. John O'Keefe, Steve Nadel, and the Mosers showed how the hippocampus records memories spatially. So it's true that, as Alex Rosenberg says, they demonstrated that some of our thoughts are encoded in a way that is essentially different from the "language of thought." But showing that (some) beliefs are non-linguistically encoded is not the same as saying that we don't have beliefs. I have a belief about how my living room is arranged. This belief is embodied in an image, not in multiple sentences of the form "There is a couch between two armchairs." Even so, I still have a belief about how my living room is arranged. I could express this belief by drawing a labelled picture; I can't fully express it in sentences. (Liz Camp has some great work on this.) Alex makes a further claim: "Experimenters decode firing patterns. Rats don’t." This might be true, or it might not. But it has nothing to do specifically with the spatial coding of the hippocampal formation. Imagine that somebody looked into Broca's area and found firing patterns there that correspond somehow to sentential structure. It would be equally true that "Experimenters decode firing patterns. Rats (and humans) don’t." What I am saying in short is that there is no new threat posed by these discoveries. Or at least, I am not getting it from Alex.

    Reply
  3. Herodotus

    I just finished the main argument of his book. I simply don't see how the neuroscience he talks about refutes the theory of mind. He makes a lot of weather about how the neural structures in rat brains (and, by extension, human brains) don't themselves "represent" the contents of beliefs like that Paris is the capital of France. But functionalist philosophy of mind (which, I take it, is the best articulation of folk psychology we've got) doesn't require that at all. Functionalism says that the belief that p is to be identified with a particular causal role (vis-a-vis other beliefs, desires, etc.), and then there will be something physical in the brain (viz., a neural structure) that realizes that causal role; it *doesn't* say that there will be something physical in the brain *that represents the capital of France*. This is just a misunderstanding.

    Functionalism provides a programme for reducing the belief that p; this is a programme for reducing *all* of it: the token attitude (the belief), the content (that Paris is the capital of France, say), *and* that attitude's *having* that content. None of it should be left over in the physical realizer. And so neuroscience, the study of the physical realizers of our mental states, should be finding any leftovers.

    I'm happy to be shown that I don't understand something, because frankly I think it more likely that I misunderstood Rosenberg's argument than that he made such a basic mistake about the philosophy of mind.

    Reply
  4. Herodotus

    Sorry, I meant to say "neuroscience, the study of the physical realizers of our mental states, should **not** be finding any leftovers".

    Reply
  5. John Schwenkler

    I agree with Chandra and Mohan. This is an inaccurate description of both the theoretical posits of commonsense psychology and the state of neuroscience with respect to human rationality. There is indeed a scientifically informed threat to the idea that humans act for reasons that we ourselves endorse and understand — but it comes from social psychology (and, yes, from Freud), not from anything we've learned about the flow of information in the human brain.

    Reply
  6. John Turri

    I agree with several of the posts already. Commonsense psychology is richer than the linked article gives it credit for. To focus on one example, mentioned in Chandra Sripada's thoughtful list, a central category in commonsense psychology is memory. Neuroscience fully vindicates that category while, of course, revealing complexities that escape commonsense.

    Reply
  7. alex rosenberg

    Yup…that Rosenberg, he's probably never even heard of teleosemantics…

    Reply
  8. alex rosenberg

    And how do rats remember? There are three chapters on that topic in Rosenberg's book.

    Reply
  9. John Turri

    I haven’t read the book but don’t need to in order to know that memory is not an illusion and has a strong basis in neuroscience. Commonsense cognition is wrong in many ways but not in positing memory.

    Reply
  10. Kelby

    A propos Herodotus — on the one hand, the points made so far are so basic that interpretive charity would imply we've misunderstood Rosenberg's argument. On the other hand, a solid n% of philosophy of mind consists of basic mistakes, so…(naturally people disagree about whose work falls in that n%).

    I'm wondering — sincerely — about the counterfactual case; if these results refute folk psychology, what kind of alternative results would have counted as a vindication?

    Reply
  11. Herodotus

    I allowed that I might have simply misunderstood the argument 😉

    And I wasn't describing teleosemantics, which from what I know is a programme for providing a conceptual analysis of what it is for a mental state to have the content it has. Functionalism per se doesn't say what that is and can remain agnostic on the question, as long as the true analysis doesn't foreclose on giving a complete (at least in principle) causal specification of the mental state in question (and I can't imagine what sort of analysis would foreclose on that anyway).

    I'm beginning to wonder whether we're at all on the same page here.

    Reply
  12. Rollo Burgess

    I certainly wouldn't count as knowledgeable about this research; it sounds very interesting to me, although I don't quite see why it's so devastating to our traditional conception of mind.

    What jumps out at me is that this type of finding does seem to contradict the lazy analogy between brains & minds on the one hand, and computers on the other. Storing memories in the form of electrical/physical structures retrievable as propositions is certainly what computers do. This seems to me to go towards Hubert Dreyfus's Heideggerian point about minds being skillful coping mechanisms thoroughly disanalogous to computers.

    Reply
  13. Jeremy Pober

    My sense is that the issue depends, at least in part, on what is meant by belief and desire. I'll discuss the point in terms of belief, but I take what I say to aplpy to desire as well.

    Some philosophers have what we might call a 'thin' view of belief, such that any state with informational/doxastic content counts as a belief. Eric Mandelbaum, who claims that any representational state which plays a belief-like role in inference and action, is an exemplar of this sort of view. Other philosophers have what we might call a 'thick' view of belief, such that an informational/doxastic must meet various rational/normative criteria (e.g. accessible to consciousness, endorsed on reflection) in order to count as a belief–John McDowell would be an exemplar of this sort of view.

    I take neuroscience to raise problems for the thick, but not the thin, view of belief. I've never seen a theory of motivated action in the neuroscientific literature that doesn't posit something like information states and goal states, which, on a 'thin' view, would count as beliefs and desires (since they are amenable to belief-desire explanation). But I've also never seen a theory of motivated action in neuroscience that posits states with the properties beliefs and desires are claimed to have by 'thick' views.

    Reply
  14. Skef

    Just to offer one simple observation:

    Discussion of belief in philosophy generally centers on the role it plays in reasoning. Any *explicit* reasoning presumably makes use of what is often called "working memory". There are probably various forms of implicit, subconscious reasoning, but what kinds of memory are involved in that can be left open.

    That understanding of belief therefore places fewer restrictions on the nature of long-term memory than one might think. What is necessary is some kind of exchange or translation of whatever is stored into a form that can serve as a belief in the process of reasoning. And of course, some kind of "indexing" to allow the relevant beliefs tend to come to mind when needed.

    Since this question isn't very central to what is at stake with belief and reasoning, I'm sure plenty of articles talk about beliefs as if they are "stored" in memory as propositional units. But that can either be treated as shorthand or as an aspect of those articles that is wrong but easily corrected.

    Reply
  15. John Schwenkler

    Of course neuroscience doesn't posit neural states that have a propositionally structured content and stand in the space of reasons. That's because neuroscience simply isn't in the business of making such posits. (Notice however that we are *constantly* learning new and better ways to "decode" how information is stored in patterns of neural activity — e.g. the technique of multi-voxel pattern analysis gives a way to look at how patterns of activation are distributed across multiple regions of the brain, rather than identifying only local differences, as on the traditional univariate approach.) More importantly, on McDowell's own account the place to look for beliefs (and decisions, perceptual experiences, etc.) is *not* at the level of neural activity or anything like it — so it's hardly fair to say that problems are raised for *his* view by findings (or, rather, a lack of findings: as I indicated above, the difference matters greatly) that are *exactly what his account predicts*. Maybe this aspect of "thick" views of belief shows that these views are somehow spooky or non-physicalistic, but in that case any reasons for finding them objectionable won't have anything to do with neuroscientific discoveries — the reasons would rather be philosophical, and they need to be defended as such rather than by appeal to the authority of science.

    Reply
  16. Smith&Jones

    The threat might be a bit overstated if put in terms or what neuroscience has already shown about beliefs etc. But it might still be that content does not or will play an explanatory role in neuroscience and other cognitive sciences. We should consider that possibility and think about what that might mean for folk psychology. Frances Egan has some really interesting recent work on this.

    For what it’s worth, I personally love Rosenberg’s book and popular pieces. I always thought an alternate title for the book could have been “Materialism On The Rocks”. He pulls no punches, drops the academic hedging, and just simply states what he thinks and why. My own views often lie pretty far from his, but this holiday season I guess I’m thankful that some people are still willing to write like that.

    Reply
  17. Robert Lee

    From the neuroscience point of view…

    Looking past the usual overreach, the ideas put forth are fashonable if not mainstream. That doesn’t make them right of course, but, there are some fairly solid elements to it. Moment to moment, there simply isn’t time for us to reason. There is no way around that. Instead we have a trainable intuition. Belief (again, from a neuroscience point of view) likely is part of that intuition and probably conforms to what one of the commenters called the “thin” version. Reason is probably one of, or one set of, the training strategies. I am forced to say “probably” and “likely” repeated because there this no way to directly get at (i.e. measure) these ideas in current neuroscience.

    However, we are not simply moment to moment creatures! We have years to ponder, cogitate, reflect, muse, and consider. Thus, while reason is a slow, post hoc training mechanism, it’s got time on its side.

    So we have an issue of time scale, which, to my extremely limited training in philosophy, has not been addressed. Can I be held accountable for my past training or am I blameless because I didn’t have time to think? The law, again to my extremely limited knowledge, does seem to address this to some degree with for example manslaughter, murder and premeditated murder.

    Reply
  18. John Schwenkler

    "Moment to moment, there simply isn’t time for us to reason."

    What's the evidence for this? Cognitive psychology has a lot to say about "slow" processes of reasoning wherein, as Skef says above, representations are manipulated in working memory. And cognitive neuroscience has a lot to say about the neural underpinnings of these processes, and also about the differences between how information is "stored" in e.g. the hippocampus and how it is made available for explicit use.

    Reply
  19. Robert Lee

    First let me say that moment=150 milliseconds which is pretty short. 150 ms appears to be a rough boundary between "reaction" and more complex responses. Consciousness sits all the way back at 500 ms. I've got to get on a plane and a ref isn't leaping to mind so I have to get back to you.

    Reply
  20. David Mathers

    'But conscious experience is continually shouting out that belief/desire psychology is exactly how the mind does work.'

    One obvious question is how it does this without having exactly the kind of propositional content that Rosenberg says doesn't exist. (Or if they conscious experience doesn't have it, but merely makes plausible the relevant conclusion about how the mind works, how that conclusion can be getting drawn in the mind without this sort of content.)

    It's not that I think this question is necessarily going to prove impossible for Rosenberg to answer, so much as that when it becomes clear that the radical, eliminativist view can still make sense of this kind of talk (i.e. talk that for example, correctly attributes to human being's an inclination to endorse a particular theoretical view about how things work), the view will look far less scary and radical than the rhetoric with which it's presented makes it seem. (This is compatible with it still being inconsistent with many cherished beliefs, just less of them.) Or to put it another way, I expect any account of why this kind of talk (the talk in the quoted sentence I mean) is in some sense ok, will inevitably prompt the question 'why isn't this just a perhaps somewhat revisionary reductive account of belief, on which believing that P is just being in a state that gets you appropriately characterised as having whatever attitude towards P Rosenberg thinks we have towards the claims of folk psychology, that physicists have towards the claims of physics, and so forth?'

    Reply
  21. John Schwenkler

    That we sometimes act without deliberation is not something we needed neuroscience to teach us. Aristotle understood it pretty well, and it's hardly surprising from the perspective of commonsense psychology. It's also compatible with acting in a way that is made rational by one's beliefs and desires.

    Reply
  22. Jeremy Pober

    You are quite right that McDowell, as a non-physicalist, is claiming that his view of the mental is that it is distinct from the object of study of neuroscience, and that, consequently, the way to attack his view is to provide philosophical arguments for physicalism, rather than empirically grounded arguments from physicalism. It would have been better for me to say that the challenge is to 'thick' views that are also physicalist, which I take to be a non-empty category (Fodor, arguably, belongs in this camp).

    But I have to disagree that neuroscience (at least broadly construed to include cognitive/systems neuroscience) is not in the business of making posits about e.g. propositionally structured content. For instance, a main claim in Larry Barsalou's work on situated conceptualization is that concepts are represented modally, rather than symbolically, and are not structured in the way that propositions are. He sees himself as attacking the exact sort of propositional theory of thought you seem to think isn't in the purview of psychology.

    More generally, I take it that philosophy and psychology both are in the business of making theoretical posits to explain the nature of thought and behavior. Part of the reason that theses like the existence of a sui generis space of reasons are interesting is because they have testable consequences: they predict that we think and act in certain ways, and these claims are either confirmable or falsifiable. Thus, I take it that philosophers of mind/psychology who endorse some sort of physicalism are all making claims which are answerable to the findings of experimental psychology.

    Reply
  23. John Schwenkler

    I didn't mean to suggest that McDowell's position wasn't open to empirical challenge, but only that the relevant challenge couldn't come from findings about the way that that information is represented in the human brain. THAT is a matter that his position is entirely silent on.

    Reply
  24. alex rosenberg

    Now this is the beginning,and in fact even the middle, of a great and illuminating discussion! Thanks.

    Reply
  25. Robert Lee

    answering both John Schwenkler and David Mathers…

    A common if not prevailing view in neuroscience is that there should be neural correlates to, well, just about everything that goes on in the mind. Conversely, if there are no neural correlates then said behavior/perception is illusory. Said another way, emergent properties/behavior aren’t “real.” This seems to be where Rosenberg is coming from. This is view also underlies most of the neuroscience attacks on reason, consciousness, free will, spirituality and so on.

    I’ll let you all argue over the degree to which this view is ridiculus. Do note that emergent properties are problematic across all of biological research as they tend to defy probing by the scientific method. As such, arguing that emergent properties are “real” is to fly in the face of a central premise of science.

    For the specific case of neural correlates of belief, I am quite skeptical that they would ever be found. You can’t measure the belief directly because you can only measure activity and the activity is the decision itself not the basis for that decision. Theoretically, you might be able to measure a change in belief as a memory/learning mechanism. Practically, it occurs over some much time that it is probably impossible to separate it out of all the other changes occurring in the same time frame.

    To argue against my own opinion, I believe that there have been FMRI measureable changes in “thinking” following psychedelic exposure. If you regard those types of changes as changes in belief then the FMRI data might be the neural correlate or at least a downstream effect implying that the is a correlate.

    Reply
  26. David Mathers

    Robert Lee:

    It's hard for me to know what to make of that, without knowing how they are using 'emergent property'. Is being a battle an emergent property? It's not a property of any of the things that take part in the battle, after all. What more is required for emergence than that? Or was Gettysburg not real? Even if there's a sensible meaning of 'real' on which it wasn't, if beliefs are as real as battles, I don't think the sense in which they're not real is likely to constitute a revolution in how humans conceive of themselves.

    I also don't see why we should expect neural correlates for beliefs in any straightforward sense. Obviously, what someone believes depends importantly (maybe even entirely, though I doubt that; I think the environment is likely involved too) on what's going on in their brain. But that doesn't *necessarily* mean we're going to be able to identify some particular brain region associated with each possible human belief, or even necessarily one associated with 'belief' in general. (I'm not denying their could be, I'm just saying it doesn't follow from the fact that the overall state of the brain fixes, among other things, what I believe.) I know this is a hoary example, and I'm sure neuroscientists are aware of it (or I certainly hope they are), but, right now, there are a huge number of things that my laptop is processing information about. But it's not the case for all of them that there's a location in the laptop associated with processing information about exactly that. Nor need there be any equivalent for each of them to something like the famous Crick and Koch proposal about consciousness involving a particular rate of firing. There's not a distinctive rate of changes of internal state associated with opening a webpage, relative to booting up, or changing the desktop background photo. Or at least, there needn't be. The point here isn't that the brain works much like my laptop; that would be highly surprising. The point is just that we can't know, just because X depends on what's going on inside Y, that we ought to be able to associate particular regions, or particular patterns of activity within Y with X. (There might be specific reason to expect there will be in the case of the brain, but that requires some brain-specific argument.)

    Reply
  27. Rollo Burgess

    Re letting everyone argue over the extent to which this view is ridiculous; it won’t be much of an argument. The idea that emergent properties are not real, and are instead illusory, is just flatly wrong.

    This would presumably mean that everything is illusory, based on our best of our current knowledge, unless it corresponds directly to a term in the Standard Model, or is gravity. I don’t know if anyone actually defends this view, but a lot of contemporary philosophy of physics makes heavy use of emergence and absolutely does not try to the argue that the emergent phenomena are unreal. There are in fact many sophisticated and important ideas which are naturally expressed at the level of the emergent phenomena – including the whole of neuroscience itself.

    Arguing that these phenomena are real and not illusory is not to ‘fly in the face in a central premise of science’, it is to adhere to the definitions of the words ‘real’ and ‘illusory’.

    That these phenomena ‘defy probing by the scientific method’ would not in any way license an inference that they are not real, it is more likely to be due to any number of other factors such as – we haven’t yet worked out how to study them scientifically, or that we have restricted the term ‘scientific’ in such a way as to exclude methodologies that are in fact appropriate to studying them. To argue from this methodological point to an ontological point about their non-existence is a very extreme and crude form of logical positivism. Neuroscience may follow this approach as a methodological principle, but to the extent that it is metaphysically committed to it, so much the worse for neuroscience. For the record, I don’t think it is so committed.

    Reply
  28. Howard

    Two comments:

    1) To illustrate in a simple way a point already made by several commentators, the fact that when we look inside a computer, all we see is various kinds of circuits and no software doesn't entail that the computer isn't running software. The analogy with Rosenberg's argument should be clear.
    2) G.E. Moore once maintained that a philosophical argument couldn't convince him that he didn't have two hands, as any component of such an argument would be more doubtful than the proposition that he had two hands. Replace 'philosophical argument' with 'argument from neuroscience' and the proposition that he doesn't have two hands with the proposition that he doesn't have beliefs and desires and you get a good rebuttal to Rosenberg's argument.

    —–
    KEYWORDS:
    Primary Blog

    Reply

Designed with WordPress