I recently had the pleasure of talking to Mark Baxter, Ph.D. about the status of neuroscience as a field. Mark is a neuroscientist who studies the molecular mechanisms of learning and memory.
This is my second conversation with Mark. Several months ago, we discussed a controversial paper on the effects of marijuana consumption on the brain. You can read that discussion here. As with our last conversation, this current exchange was sparked by a brief back-and-forth on Twitter that we thought warranted greater attention than Twitter’s 140-character limit.
Daniel: I think maybe I should start by contextualizing the question a bit?
Daniel: So I recently asked my followers on Twitter whether they thought that neuroscience is a mature science. That question was in response to a line of thinking you see occasionally in the philosophy of mind, which makes reference to what a “mature” neuroscience will be able to tell us. This comes up when philosophers debate whether or not a question is answerable empirically (for example, whether the nature of consciousness is an empirical question), or whether a question will forever remain in the domain of philosophy.
And I think that’s a bit strange. According to what criteria would you consider neuroscience to be “mature?” And doesn’t that run the risk of being a moving target? If a philosopher wants to say that a difficult puzzle like the nature of consciousness will be solved by a “mature neuroscience,” then they can keep claiming that neuroscience isn’t mature as long as it hasn’t solved that question.
So I think that at this juncture in the history of neuroscience, maybe it’s time that we took a step back and asked ourselves whether we’ve finally become a “mature” science. And if we have, what would that imply for philosophers who hoped that we would have answers to questions like the nature of consciousness by now?
The question seemed to catch your interest on Twitter, so here we are for our second blog-format conversation.
Mark: My initial reaction to questions like “what is the nature of consciousness” is wondering what the answer is supposed to look like. (“42”!)
This is a level of analysis problem: What constitutes an explanation or mechanism? You can apply principles of necessity and sufficiency, but it also requires choosing where you’re going to look: neurotransmitter release? Generation of action potentials by individual neurons? Population-level activity across multiple brain regions? Changes in blood oxygenation in particular parts of the brain?
That is, if I know that coherence in the gamma frequency range between the frontal cortex and mediodorsal thalamus is absent when people are not conscious is present when they are consciously attending to sensory stimuli, and disrupts conscious perception of sensory stimuli if selectively inhibited, have I satisfied that the nature of consciousness is gamma synchrony in that circuit?
Daniel: Well that used to be a popular theory, right? But it baffles me how people could conflate the explanation of what is necessary (though probably not sufficient) to produce consciousness with consciousness itself. And I agree that this a problem that’s unique to neuroscience: there are so many places we could be looking for the neural correlate of a specific cognitive function, or even lower down at the molecular mechanisms that drive that neural correlate of that specific cognitive function, but we’re often not very explicit about what the most relevant level of explanation is.
But I think we’re getting a little ahead of ourselves. It might help put this issue in context to first talk about what a mature science is (or could be), and then to figure out whether neuroscience fits the description.
Mark: Yeah I thought I was jumping the gun a little there. So we agree that physics is mature because it has an overarching theory of how physical phenomena arise? And it doesn’t seem to have a problem with levels of analysis because phenomena on the quantum level don’t necessary impinge on phenomena on a more macroscopic level, or only do so in particular situations (says the person who took college physics a long time ago).
Daniel: Exactly. Also the overarching theories in physics give rise to very specific, testable predictions. Oftentimes predictions in physics are specific down to the hundredth decimal place. But given the complexity of neural systems that’s probably an impossible goal for neuroscience.
So the definition of a “mature science” is of course up for debate, but the definition that seems to have gained the greatest currency among scientists and philosophers of science is Thomas Kuhn’s idea of what a mature or “normal” science is. I have a quote here from Kuhn’s The Structure of Scientific Revolutions, talking about what a normal science is:
‘Normal science’ means research firmly based upon one or more past scientific achievements, achievements that some particular scientific community acknowledges for a time as supplying the foundation for its further practice. Today such achievements are recounted, though seldom in their original form, by science textbooks, elementary and advanced. These textbooks expound the body of accepted theory, illustrate many or all of its successful applications, and compare these applications with exemplary observations and experiments. … Their achievement was sufficiently unprecedented to attract an enduring group of adherents away from competing modes of scientific activity. Simultaneously, it was sufficiently open-ended to leave all sorts of problems for the redefined group of practitioners to resolve. (Kuhn, 10)
Mark: This is a short conversation if we conclude that the presence of neuroscience textbooks means it is a mature science, but what is the theory/applications of the theory? At some basic level it is something like, you can predict the behavior of an organism based on the electrical activity of its neurons. And then that is refined. It sounds trivial but it is distinct from the idea that behavior is generated from a non-physical mind or is the purview of other organ systems (i.e. there is something special about the brain/spinal cord). But I think that takes it beyond textbooks of neuroscience just being records of reliable empirical observations that are not linked to some overarching theory.
Daniel: Agreed. Neuroscience textbooks recount particular famous experiments because those experiments helped shape the paradigm of our field. For example, every neuroscience student learns about Ramon y Cajal’s and Golgi’s articulation of the “neuron doctrine,” which is that the nervous system is composed of individual cells, and the later idea that information is passed between these cells via synapses, and later descriptions of the time-course of action potentials and the ions that mediate action potentials, and later biophysical models of action potentials like the Hodgkin-Huxley model. Students learn about these experiments and theories because they form the basis for the current neuroscience paradigm.
The reason that a lot of people claim that neuroscience lacks a clearly defined paradigm is because historically we’ve just assumed that our very exact descriptions of these microscopic phenomena, which are based on the success of these models and experiments, will eventually explain more complex cognitive phenomena as described by psychology. In other words, other scientists and philosophers have questioned the possibility of reducing psychology to neuroscience, which is presumably one of the main goals of our field.
So maybe one way of thinking about what a “mature” neuroscience might look like is when we have successfully reduced mental events to neural events, or at least a theory that explains how we might get such a reduction.
Mark: This relates to the quote from Churchland Neurophilosophy you drew my attention to while we were planning this conversation: “We may know that cocaine acts by blocking the uptake of norepinephrine, leaving norepinephrine lying about in the synaptic cleft to continue its work unabated, but how that produces euphoria is a smooth-walled mystery” (Churchland, 82).
Daniel: So it’s worth noting that Patricia Churchland wrote Neurophilosphy in the late 80s, and I think that we definitely had no idea how these molecular phenomena gave rise to these cognitive phenomena back then. But things have changed since then. I’m sure there’s a computational model somewhere out there that models the effect of cocaine on dopaminergic neurons and how that affects larger brain systems that drive behavior and decision-making.
Mark: Sure, there are – but I still feel like there’s something implied in her question about the relationship between measurable physical phenomena and a subjective mental state that is sort of ineffable. What does it mean to say I know how elevated norepinephrine induces euphoria? I can describe it at an arbitrary level of precision and resolution, but does that mean I know how it works?
Going back to that Brown article: if I can’t recreate a system in which norephinephrine leads to euphoria, does that mean I don’t really know how it works?
Daniel: That’s the central question of neurophilosophy: how do these neural events, be they at the level of single neurons or at the level of a network, give rise to the subjective experiences known to correlate with those neural events? But maybe it’s not in the business of neuroscientists to answer those questions, at least not yet. I would argue that the paradigm under which neuroscience is currently operating explicitly leaves out questions about consciousness or subjective experience. Sure, you have some scientists like Christof Koch and Giulio Tononi who get away with saying outright that their subject of study is consciousness, but they’re kind of a fringe case. For the most part, I think that neuroscientists operate under a highly mechanistic view of the brain. And whether that’s a valid view of neural function or not, it’s a useful paradigm because it allows us to ask and test much more specific questions than we’d be able to if we worried about subjective experience too much.
Mark: Because it’s circumscribed the questions to observable/operationally definable phenomena.
Daniel: Right, and that’s actually what makes me think that neuroscience has achieved the status of a mature or normal science, in the Kuhnian sense. Here’s another Kuhn quote that’s relevant:
One of the things a scientific community acquires with a paradigm is a criterion for choosing problems that, while the paradigm is taken for granted, can be assumed to have solutions. To a great extent these are the only problems that the community will admit as scientific or encourage its members to undertake. Other problems, including many that had previously been standard, are rejected as metaphysical, as the concern of another discipline, or sometimes as just too problematic to be worth the time. (Kuhn, 37)
Daniel: He was referring to questions in physics and chemistry mostly, but I think that’s an extremely accurate description of neuroscience right now.
Mark: I totally agree. One sees this for better or for worse in the BRAIN Initiative where the impetus is that better measurement and manipulation technology for neural activity in terms of spatial and temporal resolution will lead to new advances that will benefit human health, on the view that the important thing is the activity of single neurons.
Daniel: Right, which is just an assumption, though a justified one. I wonder to what degree philosophers have actually shaped the research questions of neuroscientists.
Mark: I feel sure this is where the concept of “neural correlate of consciousness” (NCC) came from (and, the idea that neuroscientists ought to go find it).
Daniel: So here’s where my contribution will get a little shakier. I’m familiar with the arguments of philosophers who have engaged with neuroscience, but a proper analysis of how their views have affected the field of neuroscience is probably beyond me. But I can give it a try.
You have three philosophers who were probably instrumental in shaping what you might call the “cognitive science” paradigm: Patricia Churchland, Paul Churchland, and Daniel Dennett. There are plenty others, but these three are the most famous, especially among neuroscientists.
I’m probably not doing their views much justice, but their ideas often amount to what I think is a dismissal of consciousness as a relevant research question. All three of them are functionalists, which means that they take mental states to be fully reducible to their functional role. Under the functionalist view, pain is just what makes you say “Ow!” and recoil your hand or nurse your wound, and nothing more.
Now we could get into the nitty-gritty of whether that’s a philosophically justified position, but I think that this is relevant to our conversation because it is, I think, the implicit paradigm that neuroscientists have embraced as a way to map neural events to cognitive events. But there are plenty of very respected philosophers who don’t take functionalism seriously at all (Thomas Nagel and David Chalmers, for example).
Mark: So (says the person who’s never taken a philosophy class) this point of view dispenses with “qualia.” Color is just a dimension by which visual stimuli can be classified or described and we don’t have to worry about the blueness of blue?
Daniel: Exactly. I personally don’t like that position philosophically, but I think that it’s useful for neuroscience research, as I mentioned a little while ago. Neuroscience has too much to figure out to be bogged down by the problem of subjective experience. Instead, neuroscientists have embraced the functionalist paradigm as a useful way to solve puzzles. And so far it’s been working really well – neuroscientists have made enormous progress in bridging biology and psychology over the last two decades.
Also it’s fascinating, historically, that the philosophical position of the woman (Patricia Churchland) who in the late 80s bemoaned that neuroscience lacks a paradigm may have played a key role in shaping what the paradigm of neuroscience would become.
Mark: Well, appropriately so! At least she helped resolve the problem.
Daniel: I agree. I wonder, though (and I don’t have anything useful to contribute here), to what extend she, her husband, and Dan Dennett shaped the modern neuroscience paradigm, or to what extent that paradigm emerged on its own and philosophically minded neuroscientists have just pointed to these philosophers’ arguments as justifying the assumptions of the field.
Mark: So there becomes a circularity that is problematic.
Daniel: Exactly. Maybe your answer to this question might shed some light on the issue: had you heard of Pat or Paul Chuchland, Dan Dennett, David Chalmers, or Thomas Nagel before?
Mark: All but Nagel.
Daniel: Interesting. Nagel is definitely the most anti-functionalist of the bunch.
Mark: I read Churchland and Sejnowski’s “The Computational Brain” when I was a little undergrad but what its main messages are have been lost in the mists of time.
Daniel: Although you might be familiar with the “what it’s like to be a bat” argument? That’s Nagel – I think more people are familiar with that argument than they are with him.
Mark: No, I’m embarrassed to admit, although I’ve developed a certain amount of directed forgetting for anything to do with consciousness. And I have a good friend who is close with Dennett and has written some things about what consciousness is for. And I haven’t internalized those either except to the extent that I know to talk to her if I ever need to know anything about consciousness. I justify my lack of scholarship on the basis that I ration my reading time on stuff about memory.
Daniel: Which is a legitimate justification! So it seems that you’ve confirmed my suspicion, which is that neuroscientists generally try not to think about consciousness. But my suspicion is that consciousness will eventually be the problem that leads to a new paradigm for neuroscience. The way Kuhn describes the progression of science is that scientists working under a mature science will use their paradigm to solve problems until they can’t. Eventually they hit a wall, and that’s when they’ll tend to turn to philosophers and philosophical thinking.
Mark: Because a functionalist point of view will not bridge the gap between physical descriptions of brain activity and an understanding of subjective experience, which we do not even know what will look like.
Daniel: Exactly. We’re making enormous progress now, and I don’t think that one day we’ll look back and think that our current findings are wrong. But we’ll probably one day be thinking about them in a different light, in the way that we now think of Newtonian physics in light of General Relativity.
Mark: I have been fascinated by Adrian Owen’s work on fMRI and other modalities of brain scanning as to whether people in vegetative/minimally conscious states are aware of their environment. But that does not tell us what it is like to be them, nor does it tell us “well we can get nonhuman brains to report on sensory stimuli that are not there but does that mean they are conscious in the way we perceive ourselves to be conscious?”
Daniel: Right, and that’s a pretty big sign that our picture is incomplete. You have philosophers like David Chalmers insisting that a mature neuroscience will never solve the problem of consciousness, because it’s a fundamentally philosophical problem, but I don’t know if that’s the case. I agree that a future version of our current paradigm will never solve it, because the problem of subjective experience is strictly outside the realm of the kinds of questions we can ask or solve. But maybe a mature version of a future paradigm will solve it?
Mark: Right. So functionalism will get us only so far in terms of linking neural activity to what mental states are for, and then we will have to have a neuroscience equivalent of quantum mechanics to make the jump to the problem of consciousness in terms of the subjective state.
Daniel: Exactly! Does that sound like a reasonable projection?
Mark: Now that I understand the problem a little better, yes, that seems totally reasonable.
Daniel: So this is a question for you: to what extent do you think the problem of consciousness even bothers most neuroscientists?
Mark: I think I’ve developed an aversion to talking/thinking about consciousness because I have never really understood the problem clearly in these terms (so, thank you). And the running joke is that neuroscientists turn 60, start thinking about retirement, and write books about consciousness.
Daniel: Seems to hold true. Even biologists – look at Francis Crick.
Mark: Yup. I think the problems about subjective states are very interesting but one quickly recognizes, implicitly or explicitly, that we are not going to get there with the current neuroscience paradigm. And at least my response at that point is to turn my attention to things I can get a handle on instead of worrying about the things I can’t.
So I think that in the absence of some real impetus to address issues of subjective experience – either a massive availability of research funding directed at a new neuroscience of subjective experience that leads people to vote with their feet, or a “consciousness virus” that can’t be satisfactorily understood in terms of current biological paradigms, I sort of feel like people are going to exhaust the questions they can address in the current neuroscientific structure before they start looking elsewhere.
Daniel: “Consciousness virus?”
Mark: Well, the equivalent of a public health problem for subjective experience that requires a neuroscience or biology of subjective experience in order to understand and treat.
Daniel: That’s a really interesting idea.
Mark: Consider all the research on viral genetics and lifecycles that was spurred by the discovery of the AIDS virus. There was that Joss Whedon show “Dollhouse” which I think overall wasn’t one of his best efforts but did have this interesting notion that you could just rewrite people’s memories with a burst of signal over a phone line. That is, totally changing their identity. I guess that’s one example of what a “consciousness virus” might look like.
I certainly have colleagues that are more interested in thinking about deep philosophical questions like subjective experience than I am, so there are individual differences, but the field as a whole I think is very functionalist in the way that you have laid it out.
Daniel: Well there’s also the sense that questions like consciousness are “useless” or a “waste of time.” And for somebody who’s interested in getting empirically verifiable answers, it is.
Mark: Well to a lot of neuroscientists, maybe not just neuroscientists, people that aren’t studying what they’re interested in are wasting their time. So I don’t know that it’s unique to consciousness.
Daniel: I think it is unique to consciousness (and maybe other philosophical questions for which neuroscience is relevant, like free will). A neuroscientist working in memory should follow what’s happening in the neuroscience of cognitive control, because he or she might gain some insight from there. But I think that very few neuroscientists stand much to gain as far as their research goes by following debates on the philosophy of mind. For now, at least.
But once we exhaust the puzzles we’re able to solve, neuroscientists will turn back to philosophy and thought experiments. That’s what happened with physics.
A final Kuhn quote is again relevant:
It is, I think, particularly in periods of acknowledged crisis that scientists have turned to philosophical analysis as a device for unlocking the riddles of their field. Scientists have not generally needed or wanted to be philosophers. Indeed, normal science usually holds creative philosophy at arm’s length, and probably for good reasons. To the extent that normal research work can be conducted by using the paradigm as a model, rules and assumptions need not be made explicit … But that is not to say that the search for assumptions (even for non-existent ones) cannot be an effective way to weaken the grip of a tradition upon the mind and to suggest the basis for a new one. It is no accident that the emergence of Newtonian physics in the seventeenth century and of relativity and quantum mechanics in the twentieth should have been both preceded and accompanied by fundamental philosophical analyses of the contemporary research tradition. Nor is it an accident that in both these period the so-called thought experiment should have played so critical a role in the progress of research. (Kuhn, 88)
Daniel: But I think that the fact that neuroscientists currently don’t engage much in philosophy is a good sign for the field. That means that it’s a “mature” science, even if it hasn’t solved the problems that philosophers hoped it would.
Mark: I do think there are a lot of puzzles left to solve just because of the complexity and diversity of behavior so the return to philosophy may be a long way off.
Daniel: I completely agree.
Mark: I think I had been uncertain about the maturity of neuroscience because the reliance on a functionalist point of view means that there are certain problems with which we need not concern ourselves. But you have convinced me that this is a good thing, not a bad thing, at least for right now.
Daniel: So maybe to wrap up, do you want to take a stab at what the current paradigm of neuroscience is?
Mark: Well, that we would be able to predict the behavior of an organism based on the electrical activity of cells in its nervous system, given a sufficient level of measurement.
Daniel: Would you say that a key assumption is that the mapping from electrical activity to behavior relies on how neurons process information?
Mark: Well, that’s part of the description of the system, I guess: how physical energy from the environment is transduced into electrical activity in the nervous system, and then the transformations that electrical activity goes through that eventually generate electrical potentials in muscle cells.
But if you think about something like place cells, there isn’t necessarily a question about processing of information there, fundamentally: it’s that the existence of place cells is the basis of behaviors that are guided by a modality-independent representation of space. Obviously one wants to know how those cells are generated but I feel like the paradigm is that the activity of those cells is required for the organism to behave adaptively based on space.
I think I’ve introduced a concept of adaptive behavior into the paradigm.
Daniel: Yes I suppose that the inherent plasticity of neural networks is a pretty big part of the paradigm.
Mark: I guess I’m not sure what needs to be part of the “paradigm” and what is part of the assumptions. I think I agree with you that the idea that the nervous system is plastic and that it needs to be to carry out its functions is a key part of the paradigm. That brains are not programmed like computers and that we just have an organ that transforms energy from the environment into a sequence of contractions of muscle cells.
Daniel: Might we say that the current paradigm of neuroscience is that the electrical activity of neurons in the brain leads to various contractions of muscles, which then leads to a change in the environment, which leads to a change in the energy acting on our sensory neurons, which then propagate a signal to the brain, which, based on various neural network learning principles that take advantage of information already stored in the brain, leads to changes in neural activity, which leads to slight changes in neural wiring, which leads to some set of muscle contractions, and so on and so on? And what we’re trying to do is figure out the precise details of this process, from the level of neurochemistry to the level of entire networks?
Mark: Yes. I’d agree with that.
Daniel: And, although we don’t test predictions based on a grand unifying theory, there are computational and mathematical models of various levels of description of this process that are based on our empirical findings, which some hope will eventually pave the way to a grand unifying theory of the mind.
Mark: I still am not sure what a “grand unifying theory” would look like – a computer simulation of a behaving organism? I suppose that would link up subordinate theories of how neural activity in particular areas/circuits leads to behavior.
Daniel: That’s more or less what I had in mind. But you don’t need a perfect computer simulation of every aspect of the behaving organism. We’d pick what we think is the level of description most relevant to behavior, model that, and use that model to generate novel predictions of both behavior and how that behavior might be instantiated in the brain.
Mark: Right, you wouldn’t need every molecule of norepinephrine if the simulation were accurate enough.
Daniel: Well, at least we’re assuming we wouldn’t. But maybe we do?
Mark: My intuition is that all the synaptic contacts are a little noisy and simulating enough neurons in terms of generating APs or not will be enough.
So I have a better understanding of the philosophical point of view now and less anxiety that neuroscience is “immature.”
Daniel: I think more people – especially outside of neuroscience – would benefit by realizing that. There’s this pressure on neuroscience to figure everything out. We clearly have a long way to go before we do, but I think the field is doing an excellent job considering how young it is.
Mark: The concept of “physics envy” is frequently invoked.
Daniel: And for good reason. Physicists have done amazing stuff. But we have to remember that we’re dealing with a much noisier system, which also happens to be the most complicated object in the known universe! So I think that people should cut us some slack.
After that, we wrapped up our conversation and went off to get dinner on our respective coasts.
What do you think? Is neuroscience a mature science?