Neuroreductionism: not dead yet (Reiner)

Daniel and I have been having a vigorous and healthy discussion behind the scenes about neuroreductionism, and his post on Pardo & Patterson's Philosophical Foundations of Law & Neuroscience has been an excellent opportunity for each of us to sharpen our thinking.  Because the discussion is important, I have chosen to write a new post on the same topic, lest it get buried in the comments section.   

The P&P paper is essentially an application of the idea that we neuroscientists regularly commit a mereological fallacy, this time applied to thinking about the law.  The notion was outlined in by Bennett & Hacker in their book Philosophical Foundations of Neuroscience, and is an important philosophical debate which is worthy of discussion.  As someone deeply steeped in the reductionist paradigm of neuroscience, I naturally find the mereological fallacy disturbing; if true, it would force me to rethink many of my most cherished ideas about the relationship between brain, mind, and behavior.

That being said, one of the benefits of being a seasoned scientist is that I am comfortable with the notion that my view of any particular issue might be wrong.  As with most scientists, I can be swayed in one of two ways: either by unimpeachable logic or by strong evidence.  For me, the P&P paper did neither.

In reprising Bennett & Hacker's arguments on the mereological fallacy, P&P provide a suitable outline but the logic is hardly unimpeachable.  To use but one example that stands out to a reductive neurobiologist such as myself, they state:

The upshot of this and countless other examples is that psychological attributes 
are essentially manifested on the behavior, reactions and responses of the living human 
being in the stream of life, not in whatever concomitant neural activity of his brain is to 
be found. This is the key to the mereological fallacy and the undoing of the reductive 
impulses of neurolegalists.  Behavior is something only a human being can engage in.  
Brain functions and activities are not behaviors (and persons are not their brains).  Yes, it 
is necessary that one have a brain in order to engage in behavior.71  But the reduction of a 
psychological attribute to a cortical attribute is a fallacious move from whole to part.  

I have outlined the problematic sentence in red: to my way of thinking, and, I suspect to philosophers as well, behavior is something that many species engage in (as I have stated to Daniel in our email correspondence, the behavior of a Paramecium is an interesting place to start this debate).  This might be construed as a minor quibble, and easily dispensed with, but if the point of the paper is to convince individuals such as myself that the mereological fallacy holds true, the logic needs to be unimpeachable.

Be that as it may, I yield to the notion that there may be more cogent presentations of the mereological fallacy that would be convincing to a reductionist such as myself, and look forward to philosophers wordsmithing their language in such a manner that an important concept such as this will ring true in my circles.  We are not there yet.

As for the examples cited by P&P, I am sorry to report that they represent an impoverished view of reductionist neurobiology.  In using lie detection, voluntary action and mental states, neuroeconomics, and moral decision making as areas of inquiry in which scientists step into the gooey mud of mereological fallacy, P&P choose poorly.  It is hard to imagine conclusions that are more contentious within the reductionist neuroscience community than these, and those critiques do not require one to have even heard of the mereological fallacy.  In short, these are areas of inquiry which are either very new and therefore not yet mature, or in which authors have made claims that far outstrip the data.  At times, both have occurred.  It is not that the science is weak (although sometimes it is), but rather that it is far too early to connect the dots in a manner that neuroscientists find more than provisional.  By basing their arguments on such observations, P&P fail on my requirement for strong evidence.

Let me be clear:  it is not the case that I do not anticipate that more mature explanations of these phenomena may come with time; Rather, it feels as if P&P have cherry-picked some of the weakest links in the field on which to build their arguments.  Such a crumbly foundation will not convince the skeptical rabble of reductionist neuroscientists.

Daniel asserts that P&P intended this paper "is intended as a radical challenge to some of the bedrock assumptions of contemporary neuroscience".  I laud their intentions, and remain open to further argument, but for the reasons outlined above, remain unconvinced.

Comments

[Scholar in Training]

I also found the P&P article not very convincing. They spend a lot of time suggesting that there is a mind-body disconnect, but never present any evidence to that effect. Beyond a conceptual move, where is the evidence that the mind is separate from the brain? We have lots of evidence to suggest the opposite (see Alvin Goldman on this).

With respect to law, I think they get the evidence/empiricial stuff completely wrong. I would have thought that the empirical claim that they talk about would be what most people would be interested in. I don't see why the fact that the evidence simply has 'inductive correlation with certain behavior' means that the value of the evidence diminishes. Are the authors suggesting that inductive evidence has no place in a court of law?

[Daniel Goldberg]

Hey Peter,

These are fair points. However, as we've discussed, I'd like to see a more robust conceptual defense of behaviorism from those disinclined to accept P&P's arguments. The challenge is precisely to the tendency to think that the behavior of paramecia is in all ways comparable to the behavior of persons, the only salient differences being orders of magnitude greater complexity in the latter.

But we have minds, and, as near as we can tell, paramecia do not. Therefore, to someone inclined to agree with P&P, simply noting scientific criteria of behavior does reduce to material substrate is not going to be convincing, because the entire point of the challenge is to problematize the adequacy of those criteria as applied to humans.

You also fairly point out that P&P have challenged some of the most controversial and ill-accepted neuroscience as examples. Yet, I'd like to note that all of the examples they pick are well-represented among what they term the "neurolegalists." That is, it may be true that there are vastly more robust desiderata within contemporary neuroscience; but if so, that does raise the question of why it is that lie detection, voluntary action, neuroeconomics, and moral decisionmaking are four of the subjects that most have neurolegalists excited. I have commented on this very blog regarding what I see as the unfortunate tendency to reduce issues related to the neuroimaging of pain to detection of malingering. It is hard to imagine what 'social neuroscience' would consist of if we excluded these categories.

If indeed you are correct -- and I think you are -- perhaps the fault lies as much with those P&P are arguing against, who have seized on the exact examples P&P choose with relish, proclaiming all sorts of prophecies for how the (legal) world will be changed, as with P&P.

Still, in case it is not clear, I entirely respect your views, but would ask what it might take to dislodge you from your long, long held conviction regarding the criteria for (human animal) behavior?

Scholar,

Your criticism is confusing to me. P&P make quite clear from the outset they are presenting a conceptual critique. Theirs is not empirical in nature. They are challenging the foundational assumptions that underlie much contemporary neuroscientific practice. As such, I'm not sure that it is legitimate to question them for not citing evidence beyond a conceptual move; the conceptual move is, in fact, the point.

Moreover, P&P do not assert a mind-body disconnect. They note full well that body is a prerequisite for mind. They simply object to the notion that mind is nothing but body. Their theory, accordingly, assuredly relies on the existence of a mind-body connection.

But the evidence you are looking for is ample. Look at Howard Fields' work on pain. He has elegantly demonstrated, contrary to eliminative materialists like Woolf, that in fact our experiences of pain are modulated in crucial ways by our beliefs and by the meanings we attach to pain. This is what can help explain pain phenomena that materialists have long had great trouble accounting for (e.g., mystics who apparently exhibit no pain behavior and do not experience pain, phantom limb pain, all sorts of chronic pain problems, and many others). Fields's point is precisely that if we really want to understand the neuroscience of pain, we need to start by inquiring what pain means to us (which obviously implicates mind).

As to law, I agree that the empirical claims are what most people are interested. P&P think most people are wrong, because there are giant category errors in the concepts upon which the extent of the merit of that empirical evidence turns. I also do not read P&P as suggesting that inductive evidence has no place in law -- can you point specifically to a place in the text where you read them as saying as much?
_____________________________

In any way, I don't want to carry all the water for P&P. They are eminently capable of clarifying their own points and defending their claims. I am nevertheless most pleased to see interlocutors actually engaging these issues, because I tend to think that these and similar issues are some of the the most important ones to be discussing in neuroethics.

[Anibal]

I recomend this entry:

http://www.petemandik.com/blog/2007/12/12/your-brain-is-reading-this/

for a lively discussion on the Bennett & Hacker ´s mereological fallacy (that Pardo & Patterson use against strong neurolegalist) to see how philosophers criticise its underlying assumptions.

I´m with the idea that neuroreductionism is not dead yet as Peter says.

[Josh Weisberg]

Daniel--

You wrote:

>The challenge is precisely to the tendency to think that the behavior of paramecia is in all ways comparable to the behavior of persons, the only salient differences being orders of magnitude greater complexity in the latter.

It's not at all clear to me that any scientist (or philosopher, for that matter) thinks that the behavior of paramecia is in all ways comparable to the behavior of persons. Rather, they think that there is enough overlap of these phenomena to merit describing them in similar ways. Paramecia react differentially to the presence of certain stimuli in their environments. So do persons. Paramecia can alter future reactions based on previous simulations. So do persons. Paramecia can move towards nutrients and away from noxious stimuli. Etc. Do they overlap in ALL ways? Of course not. But is this close enough to merit use of the same sorts of terms?

And here we reach the core of Bennett and Hacker's Wittgensteinian thesis. They claim to be able to tell when a certain conceptual line has been crossed--that is, when some evidence is criterial rather than merely inductive. And thus they can pronounce to the scientists that they've all illicitly crossed this line and thus engaged in various "fallacies." But unless you've already bought into this (to my mind) deeply discredited view of conceptual analysis--how, exactly, does one establish the criteria? Has anyone ever convincingly done so for ANY term?--there is no reason to accept their view. Scientists (and everyone else) constantly modify and extend their use of terms. Further, the very meanings of terms can shift as new scientific results come to light. The idea that Bennett and Hacker (and to my mind, it's really Hacker, but whatever) can pronounce on these things from their armchairs and thereby cast doubt on wide swaths of science strikes me as the worst kind of philosophical over-reach. (Wow. I don't know what it is about this stuff that gets me so rhetorically fired up! Something about B&H's presentation, maybe. This is coming off too snarky...)

This is certainly not to say that there aren't crucial conceptual issues in cognitive neuroscience, neuroethics, etc. As Peter rightly points out, there are, and they're being addressed in the relevant literature, on case-by-case bases. The problem with B&H is the global sweep of the thing, and its foundation in bad-old-fashioned ordinary language conceptual analysis.

[Daniel Goldberg]

Hey Josh,

I'd like to note that while I tend to share P&P's views in many particulars, defending Bennett & Hacker to the core is not my project. Rather, I prefer Searle's account of the role of subjectivity in consciousness, which gets to similar places in different ways, I think.

That said, on the merits:

Admittedly, my rhetoric in the sentence you quote is imprecise. But the clash is, I think evident in your response: of course paramecia and humans can be compared in some respects, but the analogy breaks down when we come to issues of consciousness and mind and their relationship to human behavior. It is not close enough, in P&P's mind, whereas it is in Peter's, and, I am presuming, in yours.

Respectfully, I am not sure B&H need to be able to precisely demarcate where to draw the conceptual line. It is enough to establish that there is in fact a line between specifying criteria for what counts as consciousness and compiling inductive evidence of human behavior. If such a line does exist, then the unhesitating leap from the latter to the former is problematic.

I'm also not sure I understand why such a view is deeply discredited; I understand perfectly well that this view is widely rejected by a vast majority of philosophers of mind and neuroscientists alike. B&H, P&P, Searle, and myself, if I may join such company, are all perfectly aware that arguments against the reduction of mind to brain are in the serious minority. On my reading of these authors, we just happen to think the majority has got this one wrong, and I admit to being unconvinced by arguments to the contrary thus far (largely but not exclusively for reasons I've documented here on on my own blog).

[Josh Weisberg]

Daniel--

I haven't had a chance to read Searle's criticisms of B&H, but I recall that he was not at all into their approach, so much so that he didn't have time to attack Dennett. Which says a fair bit about how little he thought of B&H!

Agreed that the issue turns on the question of being close enough. My claim is that this should be decided in individual cases, and so long as everyone is clear about the meanings of terms (which, of course, is not always the case!), scientists can use terms in whatever way is most useful. But, look, what's in a name? Would the problem go away if scientists said "paramecia shepresent" instead of "represent," where 'shepresent' is like representation in various specified ways but not others? I'm not sure what good this language policing really does.

I don't think that merely "drawing a conceptual line" is good enough for B&H. They need something strong enough to show that we've slid into nonsense--that our extended usage literally does not say anything. I completely agree that an "unhesitating leap" can be problematic, but not all leaps are like that. For example, saying that a paramecium represents some aspect of its environment seems ok to me because of similarities to what we do when we represent our environment--we have states that play similar functional roles. Yes, of course, there are differences, but unless I go on to try to tell you about the rich inner mental life of paramecia, I don't see the harm. And the benefits of such an explanation are great. We can understand the paramecia, we can see a possible line from them to us, we can see what sorts of mechanisms allow for primitive representations and consider if they can scale up, etc.

As for what is deeply discredited, I didn't mean anti-reductionism (though, needless to say, I don't agree with it). Rather, it's a priori armchair conceptual analysis, of the kind practiced by Hacker. The idea that we can read "criterial" conceptual connections off of ordinary language usage is out of fashion, to say the least. How do we justify this or that criteria? Maybe a criteria is just a well-entrenched empirical claim, but one that can (to our surprise) be overturned. And how can we, from the armchair, tell the difference between these possibilities? And, even if we can overcome that problem, what are some specific criteria for mental terms? And how do we settle disagreements over what those specific criteria are? This approach to language and meaning had thankfully collapsed by the late 60's, though Hacker is trying to revive it. There is nothing new in Hacker--just warmed-over Wittgenstein applied to neuroscience.

More interesting, I think, is Searle. He rejects B&H, but denies reductionism. I think he's on difficult and unstable middle ground, but it's an interesting question. I always though he was more down on machines than brains, but for sure he's not part of the reductionist orthodoxy.

[Peter B. Reiner]

Daniel:

I am sorry to point this out, but there appears to be a bit of a 'bait and switch' going on here. The B&H language that I objected to relates to behavior. You say, "Respectfully, I am not sure B&H need to be able to precisely demarcate where to draw the conceptual line. It is enough to establish that there is in fact a line between specifying criteria for what counts as consciousness and compiling inductive evidence of human behavior." Consciousness? When did that "concept" creep into the discussion? [Frankly, after more than 30 years of studying the neurobiology of behavioral state, I am not sure that I know what we mean by that term, and am certain that it is bandied about much, much too loosely by philosophers and neurobiologists alike.] But B&H specifically attack the use of the term behavior, not consciousness. I would never suggest that the paramecium has consciousness, and would be willing to admit that it is unclear whether species other than humans have it (whatever it is). But behavior? With your indulgence, I would like to introduce to the discussion a set of arguments from our email correspondence of 11 Feb 2009:

Daniel: Thus, while it may or may not be fair to label what the paramecium does 'behavior,' that notion of behavior, which reduces action to rote, automatic processes, fails in their view to account for what humans actually do in context of behaving.

Peter: On this point, I must object strenuously!! I will paraphrase (liberally) from a discourse I had with my colleague Grant Gillett (who would be in a better position than I to discuss mereological fallacies) during a lunch-time discussion of free will. Imagine the life of the paramecium. It is swimming in the primordial goo, taking in chemical information about its environment. It detects a gradient of food towards its left, and this sets off a series of chemical reactions that cause its cilia to move such that the entire organism moves to the left. Has it made an informed decision? I answer YES!! Has it acted upon that decision? I answer YES!! Has it exhibited a primitive form of organized behavior? I answer YES again!!

Having had four additional days to consider the matter, I stand by my comments on paramecium behavior. At least until convinced otherwise.

[Apologies to Grant Gillett if I have misquoted him. The conversation was over a year ago and I am working with fallible synapses whose ability to reconstruct precisely what was said are subject to routine failure.]

[Daniel Goldberg]

Josh & Peter,

I tend to think that the blog format rapidly outlives its usefulness in discussing these kinds of matters. So, I'll issue one last set of replies on this, though I do want to reiterate that while I tend to agree with B&H and P&P, defending all of their claims to the nth degree is not really my project.

(If you want to discuss Searle, that's a different matter).

So, my last set of comments on these matters:

Josh writes:

"I don't think that merely "drawing a conceptual line" is good enough for B&H."

I'm not sure if it is good enough for B&H or not, but drawing that line is sufficient to do the kind of conceptual work I am interested in. If indeed there is a conceptual line that exists between mind and brain, it follows that there are significant problems with simply crossing that line without argument or evidence. And I am very comfortable asserting that this is exactly what is done in much contemporary neuroscience, and even more so in social neuroscience.

"I completely agree that an "unhesitating leap" can be problematic, but not all leaps are like that."

True, but I think that this particular 'unhesitating leap' is. And if there is a conceptual line between mind and brain -- whether or not that is "enough" for B&H -- then I think that social neuroscience has some extremely serious conceptual problems that need to be dealt with.

I don't have much to say about what has been discredited about B&H's method; but, being a devotee of later W., I think we could to a lot worse than to apply that analysis (warmed over or not!) to a great many topics, including the conceptual foundations of neuroscience . . .

Peter,

Yes, I was using the terms 'behavior' and 'consciousness' loosely, and it is fair to point that out. But respectfully, I don't think anything of consequence turns on that looseness. There are two options. Either P&P or B&H would reject the entire notion of behavior as applied to the paramecia, or they would reject the analogy in the salient respects to human animal behavior. In either case the reason for the argument rests on the importance of mind and consciousness to human behavior.

The argument challenges the very criteria for what counts as (human animal) behavior to begin with, and that argument, on my reading, absolutely turns on the significance of mind or consciousness. Either behavior is simply not possible without mind (in which case scientists are wrong to term the activity of the paramecium 'behavior') or some kinds of animal behavior is possible withoud mind (in which case it is correct to term the activity of the paramecium 'behavior') but human animal behavior is not (because humans have minds).

But I do think it is interesting to see you note that you are not sure that we know what is meant by consciousness, and that it is bandied about too loosely. I actually tend to agree with you on both counts, (as does Searle -- he says as much), though I certainly don't think we can simply excise it from any adequate account of human behavior. That much contemporary neuroscience begins by doing so is exactly the framework that is being challenged.

In any case, this has been a gratifying discussion, and I do hope to continue it via other media.