Brain Inspired
Brain Inspired
BI 146 Lauren Ross: Causal and Non-Causal Explanation
Loading
/

Check out my free video series about what’s missing in AI and Neuroscience

Support the show to get full episodes, full archive, and join the Discord community.

Lauren Ross is an Associate Professor at the University of California, Irvine. She studies and writes about causal and non-causal explanations in philosophy of science, including distinctions among causal structures. Throughout her work, Lauren employs Jame’s Woodward’s interventionist approach to causation, which Jim and I discussed in episode 145. In this episode, we discuss Jim’s lasting impact on the philosophy of causation, the current dominance of mechanistic explanation and its relation to causation, and various causal structures of explanation, including pathways, cascades, topology, and constraints.

0:00 – Intro
2:46 – Lauren’s background
10:14 – Jim Woodward legacy
15:37 – Golden era of causality
18:56 – Mechanistic explanation
28:51 – Pathways
31:41 – Cascades
36:25 – Topology
41:17 – Constraint
50:44 – Hierarchy of explanations
53:18 – Structure and function
57:49 – Brain and mind
1:01:28 – Reductionism
1:07:58 – Constraint again
1:14:38 – Multiple realizability

Transcript

Lauren    00:00:03    Philosophers and scientists have been interested in causation for a long time. Mm-hmm <affirmative>, there’s, there’s at least two, two millennial. You know, there’s 2000 Aristotle is interested in causation and it’s a really important topic in, in philosophy. So it’s a lot of people have written on it. Part of what that means is that it’s very difficult to advance understanding. A lot of people have tried if mechanism is going to be a meaningful causal term, it has to be clear exactly what kind of causal system it refers to. If it can refer to any causal system imaginable, it loses its meaning. This gives us a taxonomy of distinctions among causation. And you know, the, the next step of work is to show how those distinctions matter, how are they gonna help us solve different problems and philosophy? And in science,  

Speaker 0    00:01:04    This is brain inspired.  

Paul    00:01:18    Hello everyone. I’m Paul. On the last episode of brain inspired, I had a discussion with James Woodward about his interventionist approach to causality and topics in his new book, causation with a human face. Uh, one of the things that we talked about was the wide variety of causal explanations. In this episode, I continue that discussion with a former student of gems, Lauren Ross, Lauren is a philosopher and an associate professor at the university of California Irvine, and she has taken Jim’s interventionist account of causality and used it as a tool to explore and dissect that wide variety of causal and non causal explanations. We discussed the most popular recent form of explanation in biology and other sciences, mechanistic explanations, uh, and some of the limits of mechanistic explanation. Um, then we go on to discuss other forms of explanation that Lauren has written about often in comparison to mechanistic explanations, like pathways, cascades, top, and constraints. I link to all the papers we discuss, uh, on the website at brand inspired.co/podcast/ 146, where you can also learn how to support the show on Patreon and or check out my online course, neuro AI, the quest to explain intelligence. Thanks for listening and thanks to you. Beautiful souls who support this podcast. Here’s Lauren,  

Paul    00:02:46    Lauren Ross, uh, philosopher extraordinaire, welcome to brand inspire. Thanks for being here.  

Lauren    00:02:53    My pleasure. Yeah. Thanks for the invitation.  

Paul    00:02:55    So this episode’s coming just off the heels of my conversation with Jim Woodward. And I know that you were, uh, a student of his, um, partially, I think you had a, a dual, uh, advisorship in, in your PhD, right?  

Lauren    00:03:08    That’s correct. Ken Schaffner and Jim Woodward. My two advisors.  

Paul    00:03:12    Okay. Uh, so Jim said some nice things about you at the very end of, uh, the last episode. And so we’ll see if that’s, uh, correct, but I wanna ask because there’s an MD behind your name also, so you have a medical degree and I just want to know what’s the deal with that? Why the medical degree and how did you come to philosophy?  

Lauren    00:03:33    Well, I have a medical degree because the original plan was to be a physician and that’s the standard kind of route. So for me in college, I was a pre-medical student and I’ve always been very interested in the sciences in particular, you know, biomedical, uh, neuroscience, medical sciences, and I was premed. So I went to medical school and essentially what happened, and this is a common trajectory for a lot of philosophers of science. I was immersed in the science and I started to have these extra questions about why the science was happening, the way that it was happening. I wanted to know more about the justification behind the kind of methods and the concepts. And so I started to ask these very kind of foundational questions. And, you know, I was doing this in a hospital in medical school, which is not the standard thing that a medical student at that point, and really a physician is doing.  

Lauren    00:04:42    So I could tell I had theoretical and philosophical interests, and I took those seriously. I finished the medical degree and I really didn’t know if residency in pursuing clinical practice was the best path versus, um, really taking my interest in theory of science and philosophy of science. Seriously. So I, I started a PhD program in philosophy of science. I was very fortunate to be at Pittsburgh, um, where Jim Woodward had just recently moved and, uh, Kenner was there. He’s actually another philosopher with a medical degree. And really, as I learned more about philosophy of science, it became very clear. It was a great fit for me. I, I really just fell in love with it. I learned a lot. And then the rest is history, went into academics and, and here I am.  

Paul    00:05:36    I mean, it’s so different though, than a medical practice. It seems like such an orthogonal path. I mean, did you find yourself like asking these kinds of questions and then no one would engage you in your pre-med, uh, classes or in the clinics or, you know, wherever you were that you couldn’t engage anyone with these kinds of questions or, or what?  

Lauren    00:05:55    Yeah. So the path of a clinician is interesting because before they ever get into the hospital and, you know, learn clinical practice, there’s a lot of, there’s a lot of learning that takes place. There’s a ton of classes, right? So there’s, you know, premed for me was five years, uh, medical schools, four, the first two years of medical school, you’re just taking tons of tests. It’s like, it’s like undergrad times a hundred. So you really don’t. So I completely agree. Clinical practice is almost the opposite of what I do now. And you kind of, it’s not a space where you, you always have the time to ask these, these questions they take. I mean, when I write a paper, it can take at least three months, you don’t have that kind of time in a clinical setting. Right. It’s more it’s, um, those are the front lines.  

Lauren    00:06:52    You’re, you’re really making, um, different types of decisions. So physicians and healthcare practitioners were very happy to entertain these questions when they had the time, but it’s not, that’s not their main job. And, you know, they don’t, they don’t have a lot of time, so completely agree. I mean, they’re almost just opposite ends of a, of a, uh, kind of spectrum of a kind of work one could do. So <laugh>, I mean, I, yeah, I guess I’m just lucky that I was able to explore. And then now I get to use my understanding of the reasoning from clinical practice, the reasoning from, you know, studying all of the science. And I can take that with me to philosophy of science to help my work there, but also to kind of cross the boundaries, to have discussions with scientists and clinicians. So, um, yeah, so very, very different fields and types of work.  

Paul    00:07:52    Okay. So, so you don’t, um, you don’t regret the, uh, the change, but, but it does have value in your, uh, philosophical practice.  

Lauren    00:08:01    It has a ton of value. I don’t regret it at all, but I also can’t say that I would recommend to someone interested in philosophy of medicine or philosophy of biology to go to medical school <laugh> oh God, for, for, uh, four years before this, it’s, um, not necessary. It’s a, it, you know, it’s a huge commitment, but I am, uh, it’s invaluable. It’s allowed me to, um, I mean, that was my background. I started, you know, I hadn’t taken that many philosophy classes when I started my PhD at Pittsburgh in philosophy of science. So, but I had a lot of kind of scientific background it’s, um, to say that it’s been, uh, instrumental in the work that I do is, is an understatement. I’m very, very lucky for that or grateful for it.  

Paul    00:08:52    I mean, this is kind of the era of, uh, paying for philosophers to pay attention to scientific practice and interact with that. Um, so that makes sense that, I mean, I don’t know what the normal background of a philosopher is. Um, do you know, in your experience, do people come from philosophy where, where do other people come from? Um, you said Ken was, had an MD.  

Lauren    00:09:15    Yeah. So for a philosophy of science department or program, the applicants come from two main areas. Typically either they come from a science background and then they have to learn philosophy when they get to the program or they come from a philosophy background and they have to learn the science, which  

Paul    00:09:33    Is better. So,  

Lauren    00:09:34    Um, neither they’re both. Okay. They’re both great. Yeah. It’s I mean, when you’re, when you’re doing it my way, you don’t know how to write until you get to <laugh>, you know, medical school doesn’t teach you how to write in a way that meets the high standards of, uh, philosophy and philosophy of science. So I had to learn that that’s one of the challenges of coming from the science part, you and, and you can’t just slip into describing science or being a science reporter. And that, that’s another challenge. If you know, a lot of science doing excellent philosophy of science is doing more than that. So, um, so yeah, both, both have their advantages and, uh, challenges, I’d say,  

Paul    00:10:14    All right, I was hoping to pin you down there. So, um, <laugh> so I, I, I, like I said, I just had Jim Woodward on the show and during, um, that episode, I referred to his book, making things happen as a classic. Um, and I, I don’t know if that’s true, but I just see it cited. Uh, and him, you know, just cited a lot in, um, various texts. But from my perspective, my guess is that he will be considered to be an enormous influence on the field of, of causality, at least. Do you share that view? And I, I know that there is a conflict of interest here since you were his underling there for a bit. So, uh, I don’t wanna get you in trouble of course, but, um, how, how’s your view on his influence?  

Lauren    00:10:58    Jim Woodward’s work is it has had a huge impact on the field of philosophy and philosophy of science. You can see this in the way that others describe his work in that, in that book, making things happen. This is referred to as the single most influential book on causation written by a contemporary philosopher, um, <laugh> and the, you know, the most important book on causation to appear decades. These are, um, you know, Jenna and Ishmail says this, Alison Gopnik says he revolutionized causation and philosophy. It is definitely a classic. Um, and it’s had a huge impact on the way that philosophers and scientists think of causation today. And it’s gonna be work that’s cited, in my opinion, in, uh, just the upcoming, uh, you know, we’re gonna be discussing his work for decades, if not, you know, centuries, I mean, one thing that can help put this in context is philosophers and scientists have been interested in causation for a long time.  

Lauren    00:12:12    Mm-hmm, <affirmative>, there’s, there’s at least two, two millennial. You know, there’s 2000 Aristotle is interested in causation and it’s a really important topic in, in philosophy. So it’s a lot of people have written on it. Part of what that means is that it’s very difficult to advance understanding. A lot of people have tried. So if you can, it’s a big deal. <laugh> and then the second thing to say is the, one of the biggest questions in this space is what is causation? How do we understand it? How do we define it? Mm-hmm <affirmative> Jim Woodward provides his own account of that. <laugh>, that’s a, that’s a huge undertaking and he’s done it very successfully by the standards of, um, philosophers in my field. So that’s just, that’s huge. And then you can, you can see this very easily in his work. His work is, um, I mean, he’s the kind of person that makes connections that no one saw before.  

Lauren    00:13:12    And once you see them, they make perfect sense. And it’s a kind of, his framework is a kind of tool that allows you to answer other types of questions. So there’s lots of philosophers who are not just interested in defining causation. They’re interested in all these other sorts of things, and they can use his framework to, to kind of see into other questions and to solve other problems. So it isn’t just that, oh, he did this one thing really great. We actually need and can use his account as a tool to do all sorts of other things. And this is partly what I do in my work. Um, and yes, uh, his work is brilliant. It’s had a profound impact on the current field. That’s going to continue for a very long time. And, um, it’s, it’s just amazing to see a scholar like him in the profession, cuz it’s, it’s just not common to see someone who can do that kind of work.  

Lauren    00:14:15    So there’s a lot more to say about, um, yeah, there’s a lot more to say about the particulars, but, um, but yes, he’s, he’s a giant in the field and his, I mean, we’ll talk about, <laugh>, we’ll talk about causal concepts, but there’s a sense in which once he, um, you know, once he introduced this framework, it’s just cascaded. Um, there’s been a lot of other, um, there’s a lot of philosophers that essentially can use it to solve other types of problems and to, um, to see kind of so much more about, um, how science gives us understanding of the world. So, so yes, and, and it, isn’t just that I think that way, because he’s my advisor. I, um, you know, I studied a lot of science before showing up to this, uh, you know, philosophy of science department and learning philosophy of science and his work just makes sense of how it makes sense of so much of how scientists actually reason causally. And so I use it because it’s just so compatible and so compelling with respect to how science actually works and how scientists actually think so. Um, yes. <laugh> um, yes, exclamation me.  

Paul    00:15:37    Okay. Well, another thing that he said is that right now is like the golden era of jumping into, um, the philosophy of causality, uh, and you know, the varieties of your work kind of illustrate that because you use that interventionist framework as the starting point for a lot of the, the, uh, philosophical writings that you have produced. And what I see in reading, you know, well, of course Jim’s work, but also, um, just the variety of your work is the, the range, the varieties of causality of explanation is, uh, is dizzying almost. And it’s hard to, it’s hard to keep afloat within all of the different, um, ways of going about explaining which what, you know, we’re gonna come to here in a minute, um, the varieties of explanation and, and causality and mechanisms and <laugh>, so I don’t know how you keep that all clear, but uh, nice job. I mean, do you, but do you agree that that is this, you know, the, the golden era here of causality?  

Lauren    00:16:44    I mean, I think that there are lots of different philosophical projects that are getting attention. Now, one thing that’s compatible with this comment is that there’s been a lot of current work on capturing the diversity of causal stuff in the world is one way to put it. So that the first question is what is causation? How do you define it? And Jim gives us a great way to understand that with making things happen, but that’s just the first question it’s really important, right? You can’t get to anything else until you answer that. The next question is, okay, great. That’s how you can distinguish true causal relationships for mere correlations, but we wanna know a lot more about the types of causal stuff in the world. The next question is what types of Causs and causal relationships and causal structures are out there, how are scientists identifying them?  

Lauren    00:17:40    How do they figure in their explanations? And that’s, that’s the complexity of the world that I feel like you’re referring to here. And it’s really important to be able to make these distinctions between just different types of causes and causal relationships and causal structures. And so, you know, Jim has done some of this work too. He refers to this as distinctions among causation, and this is just a huge space. Why, well, the world is really causally, diverse and capturing different types of systems is really important for understanding how science, how scientists understand the world and then also how they, um, refer to these structures and their explanations, and then just how, you know, how they explain different behaviors of different systems. So, um, that is just one big topic in the context of philosophy of science. There are many others, but, uh, but yes, there’s causation has, has received a lot of attention in philosophy for a long time. Right now it, it is probably at a peak in terms of, um, all of the focus and all of the attention it gets. So, yeah,  

Paul    00:18:56    I mean, as you know, as you’ve written, um, in multiple of your papers, the, I guess the modern era of explanation has been somewhat dominated by what’s I guess being called a new, um, mechanic, um, root of explanation, the MEChA, you know, call Craver and others, uh, the, um, mechanistic based explanation, where does causality sit in, uh, relation to mechanistic explanations?  

Lauren    00:19:26    Good. So the origins of this new mechanist framework are sometimes tied to a 2000 paper that’s written by mocker, Darden and Craver. And the idea in that paper in subsequent work is that the main way, if not the only way that causal explanations work in biology, neuroscience and medicine is basically to provide an explanation is to cite a causal mechanism in particular, the causal mechanism that produces the thing you wanna explain. So if you wanna explain a disease, find and cite the causal mechanism that produces that disease, this is, um, I mean, this is a body of literature that is just exploded. Yeah. And, um, this is the dominant way that causal explanation is, is currently understood. The, the way that it relates to causation is really interesting and complicated because a lot of these mechanists don’t actually talk about how to define causation. They’re interested in something else they’re interested in, um, how the features of a structure that you call a mechanism, like what is a mechanism? We know it has causes, we know that they causally interact. <laugh> we don’t need to talk about how to define that, but like, what are, what other features does the system have? How are those causes organized? Are they, are they hierarchical? Are the causes at a lower level with respect to the effective interest? Um, how much detail  

Paul    00:21:03    Do they kind of assume a single type of cause? Or is it just hands off that there’s no, no assumption necessary because it’s not in their purview to, um, describe  

Lauren    00:21:12    Pretty hands off. And some of them even agree with the same like account of mechanism, but they just don’t talk about causation. So in the MDC paper, they, gosh, I don’t even, so Craver’s work relies on an interventionist account of causation in some of Bechtel’s work on mechanism. It’s a powers based account of causation mm-hmm <affirmative> and just, you can find that and that’s, it comes up in Craver’s work. It just doesn’t even come up in other, in other types of work. So it really just, that’s not the, the main kind of question for them. The question has to do more with, um, what is a mechanism once you’ve accepted that it’s a bunch of causes mm-hmm <affirmative> what else, what else needs to be characteristic of a system to call it a mechanism aside from that we all kind of agree that mechanisms have more than one cause, and that they’re working together to produce an outcome.  

Lauren    00:22:19    The question is kind of like, okay, what else? Um, and what’s interesting, and what’s nice is that this work is motivated by scientific practice scientists use the term mechanism, just, um, all over the place in the life sciences. So it’s, it’s motivated by an attention to scientific practice. And the fact that scientists often use the term when they give an explanation, but it’s much more focused on what features are present in a causal structure. That’s a mechanism, um, as opposed to arguing or agreeing on what definition of causation kind of underlies that, that picture and that structure.  

Paul    00:22:58    Okay. Okay. Uh, again, I don’t wanna get you in trouble here, but my reading of the, um, mechanistic accounts of explanation, which is concerned with what counts as a good explanation fundamentally, um, is <laugh>. And I, my reading of maybe critics of the MEChA, uh, account is that they are in the business. They quote unquote, are in the business of moving the goal posts when it comes to what counts as a mechanistic explanation. And that it, that it over time, you know, other, um, counterexamples get subsumed by the mechanistic account. Is that a, is that because they’re moving the goalposts or is, are, is it too broadly defined or what am I getting there?  

Lauren    00:23:46    No, that is, that is something I would definitely agree with. I guess one helpful thing to say here is that the part of the background is that we think causes explain their effects. And so to provide an explanation that’s causal yeah. Involves citing the causes. And you could start with cases where you have a single main cause. So if you think of a disease that’s caused by like a single gene Huntington’s disease, for example, um, that’s a mono causal, well relationship or disease, right? What’s the, what’s the explanation for Huntington’s disease. It’s having this particular, um, genetic factor, the, I, the idea, the MEChA, the mechanism, I idea that is compelling. And it makes sense is, you know, that’s a pretty minimal explanation. We often want more, we want more of the details between that initial cause and the effective interest. Um, so, so I guess the answer is okay, that’s not a very complete explanation.  

Lauren    00:24:52    That’s not a very full explanation. Uh, an explanation that meets that kind of standard should cite a mechanism, which is a set of causes that are all working together to produce the outcome. And there’s a sense in which if someone asks, okay, you know, how does causal explanation work in neuroscience or in biology? How does causal explanation work? Um, if you say that causal explanation is mechanistic and you just mean by mechanism that there’s a lot of causes that all interact together, you’re not really saying all that much. Um, yeah, let’s see. This is a little, this is a little complicated. Okay. Here’s a, here’s another way that might make this more clear in the early accounts of mechanistic explanation mechanism was defined more narrowly. It had, it was a causal structure with very particular features. What were they typically lower level causes interacting to produce a higher level outcome, like parts of an engine or a machine.  

Lauren    00:26:04    Um, second lots of detail mechanisms. Can’t be sparse. Can’t be one, cause one effect can’t be a network needs to have good amount of detail. And then third, you need mechanical. You need mechanical interactions. So you can’t just say X causes, Y you need to say X bends, pushes poles compresses. So those are three pretty concrete features that make something, a mechanism that was partly how we started. And then the, the problem is, well, you can find other causal stuff that scientists cite that in their explanations that doesn’t have those three things. So that’s where you see this stepping back and back peddling where it’s like, okay, wait a minute. There’s other mechanisms. They, they can have lots of detail, but not always. They, they can have mechanical interactions, but not always <laugh>. And so that’s the backpedaling and that’s the more broad notion of mechanism.  

Lauren    00:27:05    The, the worry we have is that when you broaden mechanism so much, all that it means is causation and causal structure. It just means you’ve got a bunch of causes that work together, and that’s not going to give us a helpful understanding of a types of causal structures in the world or B how causal explanation works. Because, I mean, you’re just saying someone asks you, how does causal explanation work in neuroscience? You say it’s mechanistic. All you mean by that is that, um, there’s a bunch of causes you, you just mean many causes working together where the first, the way we started, you meant to causal structure, that’s hierarchical, fine grain detail, mechanical interactions, that’s clear. Um, so if, if mechanism is going to be a meaningful causal term, it has to be clear exactly what kind of causal system it refers to. If it can refer to any causal system imaginable, it loses its meaning.  

Lauren    00:28:08    And it’s, it’s, in my opinion, it’s essentially meaningless. And that’s really unfortunate because we, when we use mechanism, we often do think it means more than just any kind of causal system. It’s a particular kind, the rub and the work of a philosopher is to say, how should mechanism be understood? Um, what is it to say that a system is a mechanism? And, um, and yeah, so there’s been a, just a lot of debate about the right way to understand mechanism and really, um, the right way to understand causal explanations that involve many factors that are not just that kind of mono causal, um, type of framework.  

Paul    00:28:51    You’re not gonna get hate email for, for this exposition <laugh>  

Lauren    00:28:57    Um,  

Paul    00:28:58    Or do you already, is that the,  

Lauren    00:29:00    I mean, so I wrote a paper. I mean, I’ve, I’ve written a lot of papers that, that, uh, have revealed my views on this. And yeah, so, I mean, I wrote a paper on how another kind of causal concept that we find in science that’s different from mechanism is the notion of a pathway.  

Paul    00:29:18    Good. Okay.  

Lauren    00:29:19    Yeah. So a pathway is a kind of causal structure. That’s different from you remember the three main features I just mentioned of mechanism pathways are different mechanisms need to have lots of fine grain detail. Pathways are very abstract. You just need the steps of the pathway and pathways are analogized to roadways and highways. A mechanism is often analogized to a machine. Mm. So a machine you drill down and you find the lower level parts, they all work together. Think of a watch mechanism, all the parts work together, they produce the outcome of interest, well pathways in biology, neuroscience. These are different types of systems. Um, you can think of it. They’re analogized to roadways where you have a kind of, um, you have a kind of linear that can get more complex, a kind of, um, linear chain of causal steps and something flows along that. So you can have neural pathways, you can have vascular pathways like blood vessels. You can have developmental pathways. And in each of these cases, there’s a kind of root that something can flow along and you can even have a map of the pathways, you know, metabolic pathways, ecological pathways, scientists will represent these connections with a map of connections between cause and effect and something flows along those steps, very different kind of causal structure than the more machine like watch mechanism. Um, and so  

Paul    00:30:45    There’s no hierarchy necessary in a pathway as well. Right?  

Lauren    00:30:48    So exactly. Yeah. There’s no hierarchy. The there’s no lower level parts. It’s the way that you study a pathway. So with a mechanism, the way you study the mechanism is you drill down to find the lower level parts. Once you fix the explanatory target with a pathway, one thing that scientists do is they can use tracers or tags and they can tag a cause and they can watch it as it moves through the system. Think about a system like a watch mechanism. We can’t study that system in that kind of way, because there’s no physical material that moves through the gears of the watch reliably. So part of what you see is that it’s important to keep track of these different types of causal systems. A because they have very different features, different behaviors, but B we study them in different ways. You, you, there’s some methods you can use for a pathway that you can’t use for a mechanism. Um, yeah.  

Paul    00:31:44    So how, how is, uh, a pathway different? So I, I, I guess what we’re gonna do is step through a few of the different, um, causal and non causal explanatory, uh, terms, concepts. Um, and so you’re talking about pathway. You’ve also written about cascades, uh, as an explanatory concept. So how is a pathway different from a, a cascade? And eventually I want to come to how, how to think about all these different varieties of explanations and causalities, um, with, within the brain, of course, uh, the easy topic we’ll save that for. Maybe we’ll go through a few examples beforehand.  

Lauren    00:32:23    Great. The notion of the cascade shows up in neuroscience biology. It, it actually shows up in many different sciences, physics, chemistry in biology. We have the blood coagulation cascade. We have ischemic, cascades and neuroscience. We have cell signaling cascades in our brain and in, um, I mean the transmission of COVID through the population is described in terms of a cascade. Um, we see different cascades and economics, physics chemistry with this causal structure. The important feature is amplification. So a cascade is a system where you have a small cause and you produce increasing amounts of the effect at every single step. So if you think of the snowball effect in the ripple effect, this is the analogy that scientists use. So when they talk about mechanisms, they often use the analogy of a machine. When they talk about pathways, they often use the analogy of a highway.  

Lauren    00:33:29    And when they talk about cascades, they will use the analogy of a snowball effect or the ripple effect, and even a cascade, like a waterfall where you have like a small stream of water at the very top. And it branches out at every single step. So cascades are kind of explosive is another way to think of it. Tiny cause at every step, the effect gets bigger and bigger and bigger. So COVID in the population. You have one patient, they can transmit this disease to two. Each of them can transmit it to two more and you have this fan out structure of amplification. Um,  

Paul    00:34:03    But can a cascade happen along a pathway or are they completely distinct?  

Lauren    00:34:09    It depends on what parts of the system you are interested in. If you think of a pathway in terms of a sequence of steps that are more linear, where you have something flowing along them, then that’s usually a kind of causal system that’s distinct from a cascade. So in a pathway, you’re not gonna emphasize amplification. Mm-hmm <affirmative> in a cascade. You are, and in a cascade, you might not emphasize the flow of something through the steps of the system. So with a, a pathway, you know, you have, you have blood moving along the blood vessel in the brain neural pathway, you have a signal, sometimes material moving along the neural tract, uh, nerve tract in ecology. We have pre predator relationships in an ecosystem and you have energy caloric energy moving through the steps. So with a pathway it’s like, it’s like cars on the roadway.  

Lauren    00:35:08    You wanna emphasize. There are, there is stuff moving along the sequence of causal steps with a cascade it’s less the, the focus isn’t on something is moving through these roots. It is amplification. Um, now the same system in the world could be described with either a cascade or a pathway or a mechanism, depending on what you’re interested in. I think that’s an important point. Um, so in blood coagulation, if you wanna know about one enzyme in that system and study it in detail, you can get a mechanistic understanding of it. But when you pan out and you wanna know how blood coagulation works as a process, and you wanna explain this tiny signal, the very beginning in the huge effect you have at the end, this huge clot, then you’re gonna see at kind of a larger scale. You’re gonna see something that’s a cascade type of system.  

Paul    00:36:01    OK.  

Lauren    00:36:02    So it’ll partly depend on, um, what kind of explanatory target one is one is interested in. And, um, and it, it could be the case that the same kind of physical system, the world can be represented with different types of causal structures or terminology, depending on what we’re interested in.  

Paul    00:36:25    How about topology? Let’s go through two more and then I wanna kind of bring it all into the brain and mind and behavior can top is topology causal. I mean, so obviously in neuroscience, um, mapping out the connections between neurons is a big deal. And you also, these days, a lot of people talk about dynamical structures and relate that to some cognitive function, dynamical structures of an interacting population of neurons. Um, and those are somewhat topological. So how, how do we think about topology with respect to causation  

Lauren    00:36:59    In many different ways? The standard way in the philosophical literature is actually to think of topology as providing non causal explanations, which is interesting. So if you think of topology as giving us information about mathematical features of a network, if the math is doing the explanatory work, actually those can look like non causal explanations. And if you, on the other hand, look at topology as a set of causal connections, then in my opinion, it can figure and provide causal explanations. There are also neuroscientists who discuss more complex explanations. So Danny Bassett is a great someone, um, who does work in this area and in their work. They often talk about having dynamics that they’re putting on top of a kind of network model. So there’s kind of combining many of these different, um, elements, maybe some kind of topological or network structure that has mathematical features, um, and then kind of combining it with a dynamical piece. So, um, the standard, so the standard view in Phil SI has been that topology is typically giving us a non causal explanation. I think we can actually see that sometimes it gives us a causal explanation, but there’s a debate about that. Um, that’s ongoing in the literature and the, the pathway concept is interesting, interestingly, related to some understandings of topology and, um, well, and constraints. So, sorry, I’m getting super off track here.  

Paul    00:38:54    <laugh>  

Lauren    00:38:54    That’s okay.  

Paul    00:38:56    Well, I wanted to jump in and just, you know, you were talking about non causal, topology and causal, the, the prospects of causal topology, and one may wonder why is it important to sort all of this out, what what’s at stake, um, and calling something causal or non causal. It almost, you know, you can interpret it almost as becoming a matter of semantics at some point at some level of granularity. So what is at stake? Why is it important?  

Lauren    00:39:26    That’s a great question. The reasons why we’re interested in causation often have to do with control. So if you identify a cause in the world, that should be something that gives you control over the effective interest and that’s useful for all sorts of reasons. Of course, we don’t just wanna provide explanations. We wanna actually do something about it, outcomes in the world when it comes to mathematical explanations. I don’t think they, um, well, I think they represent systems that we study differently. So cause identifying causal relationships in the world requires empirical work. I can’t just use math and look at a gene and figure out what the math, if it causes a, an outcome or not, I have to go in the world, maybe we’ll use an animal model and we’ll do the kind of intervention or manipulation or wiggling with math. You’re studying the world in a different way.  

Lauren    00:40:27    And at least you’re bringing in this, this other kind of way of inquiring into the world. Um, you know, I can, you can look at a network model. In some cases you can look at the math and with just that alone, you can identify what’s explanatorily relevant for the outcome. So at least one main reason that’s practical is the way that we study and learn about the world. The, the other reasons are just that philosophers of science, of course wanna figure out how explanations work. Yeah. And if they’re not all causal, they care about that. But for me, methodology is gonna be a key, um, piece. What are the kinds of methods we use to get the information, we need to explain things in the world and they seem to be quite different from the causal to the non causal cases.  

Paul    00:41:18    Okay. So, um, let’s talk about constraint. So actually this was originally my, uh, interest in having you on the podcast is your work on constraint as an explanatory factor, um, causal and non causal. I, I got interested in constraints from reading works like Terrence deacons, um, incomplete nature. And, uh, it works about biological autonomy, this concept of closure of constraints. And I started wondering like what in neuroscience could I, you know, apply this to how to think about constraints, um, in terms of brain activity, uh, and how that brain activity relates to mind and behavior. So maybe you can just describe first what, um, your ideas about constraints and how they can be causal and, and non causal versions of constraints. This is all, it’s all, uh, dizzying, I think is the, the right word for me anyway.  

Lauren    00:42:16    Good. I mean, it, it, it is for us too. And I think part of what the job of a philosopher of science is, is to clarify in this case, right, what is a constraint? What does it mean to call a factor, a constraint? Um, and what’s interesting is, so Mark Lang is a philosopher who has done great work on constraints and how they figure in explanation. And Sally has Langer is, um, a philosopher who has done work on social structural explanations and their structure is a constraint. And she will talk about the constraining influence of social structure on individuals. So part of what we see is that in some explanations in science constraint talk is coming up and we, and we wanna know, um, what’s going on here. And one thing that’s interesting is that it looks like they provide a very different kind of explanation than the standard type we’re used to.  

Lauren    00:43:09    So the standard type of explanation we’re used to is a causal explanation. We’re at least a standard causal explanation. How does this work and what’s going on? Well, the cause is explaining different possible outcomes. It could be two main outcomes or a whole set what’s interesting is with constraints. You’re often explaining why something is impossible and you’re explaining this border of possible and not possible. So that is a very different kind of explanatory target Lang will refer to these sometimes as impossibility explanations. And in the context of biology constraints will show up in explaining limits on the size of organisms on the planet. So we can’t have animals that are, you know, the size of sky of skyscrapers. We can’t have a, we can’t have insects that are the size of elephants and that because there are constraints on how, um, large they can be.  

Lauren    00:44:10    There’s also constraints at the lower level. You can’t have, um, mammals that are too small for various reasons. And so what’s interesting is a constraint. There is playing a very different kind of role in the sense that the explanatory target is different. It’s telling you what can’t show up. And this is interesting, of course, because we look out into the world, we see this diversity of, um, of structures, um, and organisms, right, or looking at organisms on the planet, but then there’s also some limitations we see. And so constraints are figuring in these cases and explaining those limitations. And then with social structural explanations, social structure, we think is limiting individuals in society and playing a kind of constraining role on what they can and can’t do. And this is where it comes up in has Langer’s work. So constraint, I mean, this is another example of almost like causal diversity or causal complexity.  

Lauren    00:45:08    There’s lots of different types of explanatory factors in types of causal factors. Some constraints are causal. And part of what’s happening in this work is capturing how they’re different from other types of causal and explanatory factors and how the explanations are different. And in this case, one of the main differences is this impossibility feature. You’re not explaining why these different outcomes that can all occur. Well, you’re not explaining why one of them shows up versus another. You’re explaining why a whole bunch of stuff can’t show up effort. Mm-hmm <affirmative>. So the, the impossibility feature is a unique feature of a lot of these constraint based explanations.  

Paul    00:45:52    So, so what about the sort of on the opposite end of the constraint <laugh> spectrum, I suppose, is the idea of enabling constraint. Um, so like to get to do any work, right? You can’t have an open system, you have some sort of constraint, um, to do the work, right. Um, you know, like a blood vessel constraints, the flow of blood, that sort of thing. Uh, and so I want to ascribe this sort of like positive, um, role to constraint, um, as opposed to just thinking it of it as a limitation, right. Limiting what, what can happen because is, is the term enabling constraint? Is that a real term? Am I making that up?  

Lauren    00:46:32    No, that is a real term. And I think this, this hopefully captures the guiding role of constraints too. Okay. So, I mean, if you think of riverbank, right. And the sense in which they’re guiding the flow of the water, same with blood vessels, same with, uh, neural, uh, tracts and neural pathways. There’s a sense in which this is very much guiding some object to a final destination. So yeah, that is the, that is the kind of flip side of this when you limit, you’re also, <laugh>, you’re also guiding, right, right. You’re saying you’re not gonna go over there, but you are gonna go over here. And the, the interesting thing is when a constraint is guiding so much that it dictates which outcome will show up, then it’s just playing a bigger role in the explanation. So, you know, if you think of, um, a set of blood vessels, they’re actually really interesting explanations in, um, biology.  

Lauren    00:47:35    So if you think of a, um, the vasculature in our body, if there’s a clot in the leg of a human, that Clott ends up causing damage in certain parts of the body and not others, it’s often going to cause damage in the lungs, it’s going to basically block the vessels in the lungs. It’s not gonna block the vessels in the hands, not in the brain, not in, you know, the gallbladder in the lungs, not anything else. And the reason for that is the, the vessels are guiding it to the lungs first. That’s the way that our vasculars vasculature is set up. So the, you very much agree that the yeah, part of the limiting factor of constraints is also the sense in which they have this like guiding or structuring role in terms of, um, guiding kind of what the final outcome will be, or in many cases kind of where, um, an entity will flow or the kind of decisions that an individual can make.  

Paul    00:48:39    I mean, is it wrong? I, I really wanna say that our brain architecture or the connectome, is it wrong to think of that as a constraint? What, what other ways what’s the right way to think about the structure of let’s, you know, neural networks wet or, uh, machine learning or, or whatever.  

Lauren    00:48:59    I think in many cases, it, it does make sense, and it’s very useful to think of them as constraints and the way that they show up in the explanations we provide will just differ. Um, but that’s partly why we’re so interested in them. If two RAs of the brain are connected to each other, that matters for all sorts of reasons. And it mean maybe it’s a little easier to see this in thinking of not just the brain, but the nervous system altogether, right? If you have a lesion in a certain part of the brain and that part of the brain is connected up to your leg muscle, right. Um, the fact that that connection exists is going to kind of limit what that lesion will produce. Obviously, things get a lot more complicated in the brain, but you know, many experiments in animal models, involved inducing lesions to kind of see what happens and part of how we understand those situations is that the circuitry and the connections and the wiring of the system are constraints that are figuring in, um, how a lesion produces a particular kind of outcome. So it’s, yeah. I mean, obviously the analogy of computers get brought gets brought up a lot in circuitry, but there, there are similar, there are similarities. I think the question will be the brain is so complicated. We have different explanatory targets. We have to be very clear about what exactly we want to explain. And then that will help us to see how constraints matter for the explanation if they do it all. Um, yeah, they’re there, but do they matter for the explanation? Depends on what you wanna explain.  

Paul    00:50:44    I think I want them to matter. So I was gonna ask you, like, what if, if, if there’s sort of a hierarchy hierarchy hierarchy of, um, you know, explanations, right. And if there is, it seems like constraint has long been at the bottom of that hierarchy. If it belongs there at all, is it right to think of these different modes of causal explanations and non causal explanations as a hierarchical structure who like you, where there’s a winner, there’s a best kind and a worst kind. And  

Lauren    00:51:16    For me, I don’t think of them in a hierarchy in that sense, I’m pretty open to their being different types of explanations that are distinct and that are full and provide important understanding and where they can’t be subsumed by one single picture of explanation. So I’m pretty, I’m pretty open to their being many different, legitimate types of explanations. And the, some of the hard questions are, well, which kind do you need in this context or with this explanatory target, and what level of detail do you need? Often? The, one of the questions that comes up is how low do you need to go? You know, are all behaviors of the brain best understood with lower level physics or chemistry? Are we, do we need to go down that low or do we need to go down to the cellular level? Can we just stick with a, a network model that’s more abstract.  

Lauren    00:52:20    So for me, there isn’t a hierarchy of better or worse explanations. It’s a, it’s a set of species <laugh> that are all different and you do need to justify why this one is the best way to provide an explanation in this context and the rationale behind it needs to be very clear, but I, um, and they can be complicated, right? The constraint can configure. I mean, constraints can configure in non causal explanations, but causal ones two, and we can grab information from many different levels to explain a particular behavior of interest. So I think they can just get very complicated. Um, but it isn’t a one size fits all approach for me. And part of the challenge, which you mentioned earlier is capturing different types and clarifying the rationale that, that underlies them.  

Paul    00:53:19    Here’s what I want you to solve for me. Um, it’s just a, a minor little, uh, so there’s this classic, um, problem, I suppose, in neuroscience relating structure to function and like Eve martyr has said that structure is necessary, but not sufficient to understand function, um, because of things like multiple realizability, which maybe we’ll talk about here in a moment as well. Um, and so what I really want is like the pluralistic fine, but the sort of structure of the various types of causal and non causal explanations for how to think about how function, uh, derives from structure, how to connect function and structure, because there’s this divide. And I don’t know that we really know how to think about it. Um, yet, does that make sense? I, I feel like there’s room for, um, these different types of explanations, causal, non causal, et cetera, to fit together, um, to somehow weave together and, uh, enable a satisfying account of the relation between structure and function.  

Lauren    00:54:31    Yeah. This is really interesting because in my view, I don’t think function comes up or is important in every single context of biology and neuroscience and medicine. I mean,  

Paul    00:54:49    What, what do you mean? What, what,  

Lauren    00:54:51    So it’s one thing to know that something causes something else mm-hmm <affirmative> and why, what more dysfunction add to that, I guess. Right. Okay. Um, and you know, so, I mean, my background is in medicine, we were more studying, um, you know, which relationships were causal, what caused a disease. Um, it didn’t matter so much what the, you know, what the proper function of the system was. We just wanted to know, given this outcome, what’s going on here, what’s causing what’s causing this. And once you get the causal answer, I don’t know why you need to know so much. So, you know, suppose our bodies developed to serve a particular function. Um, uh, you know, that could be the case, but in the context of medicine and many areas in biology and neuroscience, I mean, there’s this behavior, this patient isn’t doing well. We wanna know why do we care?  

Lauren    00:56:03    Whether that’s the way we evolved. I mean, not, you know, not really, we just wanna know, like this person is suffering or they have this outcome what’s going on, what’s explaining it. How can we fix it? Um, or, you know, you can think of it in a more generic biological context. Here’s a fruit fly, it’s got black eyes, red eyes, white eyes, why what’s causing this, what’s explaining it. The function is a kind of another domain. Did it, you know, maybe more in like evolutionary context, which I think I’d focus on less. Um, so for me, many of the explanations and many of the appeals to structure are simply focused on a particular explanatory target. And you can think of that as a why question? Why does the fruit fly have red eyes as opposed to white eyes? Um, why does this patient have this behavioral outcome as opposed to another, why is there, you know, blood pressure at this level versus another, it’s not always clear that functional information is going to help you answer those questions. And that’s just a set of questions that this literature is typically more focused on. And I fully agree that you can’t use structure to get information about function. Um,  

Paul    00:57:29    You can in limiting limiting cases, but when you’re talking about such a large population of neurons, for example, uh, it’s, it just becomes unwieldy.  

Lauren    00:57:39    Is there a way that function figures in explanatory questions and science that I didn’t kind of capture, and that is like present in these cases that you’re interested in?  

Paul    00:57:50    Yeah. I, I mean another way. So, you know, this is, it’s an impossible question. Another way to ask it is, uh, it’s a different question, but what I really wanna know and what a lot of people really wanna know is I suppose, how brain and subjective experience are connected. And I’m wondering if there, so I don’t, you know, that might not have, have to do with function, so to speak, but is there <laugh> optimism that, uh, these kinds of causality approaches have any promise to connect those two things that have been forever distinct and such a conundrum?  

Lauren    00:58:32    Yes. So at least the way that I think about it, um, yes. So one of the interesting puzzles we get into with causation is downward causation and causation between levels that shows up in many sciences is really interesting. You may think subjective experience is something at a higher level and maybe stuff in the brain is at a lower level. And the reasons for why this is puzzling is sometimes we have an account of causation in mind where it it’s kind of like billiard ball causation, the cause needs to be connected to, or it needs to like hit the effect. And that needs and subjective experience is pretty difficult to think of as a billiard ball, right? What’s hitting subjective experience. Another place that shows up very different context is social structure. If we think that social structure causes individual behaviors, it’s super puzzling in many cases because sci scientists will say, how does social structure get under the skin to cause the disease or the outcome, because we’re kind of expecting a mechanism in the sense of physical connections.  

Lauren    00:59:39    One of the advantages of the interventionist framework is it shows you, that’s not always, maybe ever the right way to think about causation. It’s more in terms of causes, giving control over their effects, don’t need to know about, or have physical interaction between them in that kind of way. So if you’re subjective experience, if that’s the explanatory target and you can search for and identify candidate causes such that when you manipulate them, it changes subjective experience. That’s gonna start looking a lot like a causal relationship on an interventionist framework. And that’s a kind of first step into getting a sense of the causal relationships. We often want more information about the causal picture, but same with social structure, right? The idea is that you don’t need to have the billiard ball causation understanding of social structure to individual outcomes. It’s not the right way to think about causation.  

Lauren    01:00:35    The way to think about it is when you change social structures in a society, in the right kind of way, what you see is changes in the outcomes of individuals. And there’s a strong, stable connection between those and that is giving you that’s evidence for causation. Um, the, so the advantages of the intervention to count is you can capture downward causation. All of a sudden it doesn’t look puzzling anymore. And, and we talk about it all the time in a way where we think it makes sense, but we don’t, we haven’t had the account of causation that captures that and the interventionist account does. And then it allows us with levels to get over some of these worries we’ve had about causation spanning levels. So I think it can play the same role with thinking of what causes and explains subjective experience.  

Paul    01:01:29    Why does, um, I was gonna ask you this with regard to multiple realizability, but, and so maybe we can lead into that. Why is reductionism sort of the default? I think if I wasn’t scientifically taught, I, I don’t know if I would be a quote unquote reductionist, but the, the new mechanic, um, explanatory, uh, approach is reductive, right? It’s, it’s solely reductive, right? But there’s this huge anti reductionism movement right now in neuroscience. Um, and I was trying to picture a world where reductionism was not the default where the default, uh, was, you know, downward causation or, you know, what have you it’s like, which seems very unnatural. And reductionism seems way more comfortable is that you, you think we, we are just naturally more comfortable with reductive explanations.  

Lauren    01:02:23    I think there’s a few things going on here. I do think that we’re more comfortable with them for reasons that might have to do with culture and the types of stories that we tell about the world. If you tell a story that’s machine, like you’re encouraged to think of lower level parts, and we’ve been comparing the natural world to machines for a very long time. I think another, I mean, it’s also the case that science is a, a community. So if you’re raised in a community yeah. In society is too. If, if the way that we’re raised is people are saying while the lower level stuff explains the higher level stuff, it can be hard to shake that.  

Paul    01:03:05    I think you think you, sorry, you don’t, you don’t think that the, uh, success of physics, uh, has anything to do with it had just popped into my head that, you know, physics was so successfully reductive for so long that maybe, you know, that’s sort of the gold standard. Right.  

Lauren    01:03:21    I think what would explain it is the fact that we had serious insights in physics early on in ma in humankind study of the world. So, you know, psychology, social science, biology, that’s showing up a lot later, right? We don’t learn about genes until much later. Yeah. But there’s a study of physics much earlier. I think that has to do with it. But if you look at the success of explanation, I mean, we’ve got a whole heck of a lot of them in medicine, you know what explains COVID where there’s a lot of diseases that we can explain. There’s a lot of, you know, economic outcomes, social outcomes. We can explain in those levels that physics doesn’t give us any, any handle on. So there’s a lot of success in explanations that are outside of physics. I don’t think it’s the success, but I do think that there is this picture, that physics is fundamental.  

Lauren    01:04:16    And sometimes that means physics involves tiny stuff. And sometimes it means physics gives us a complete, full understanding of the world. And one, one thing that I sometimes think is the reason for the reductive view is that there’s a confusion between two things sometimes and someone, sometimes it’s common for people to think that when you’re not a reductionist. Um, so let me try to phrase this a little better. <laugh> so there’s two things that you could have, right? One of them is to deny that any system in the world like a, like a, um, disease state in an individual, or suppose you’re interested in a kind of phenotype in a fruit fly or a Guinea pig, suppose you’re interested in phenomena at those scales. It’s one thing to say that those things are made up of physics at a lower level, that there’s some kind of lower level micro, structural detail.  

Lauren    01:05:27    If you go into those systems and you look down, there’s physical, there’s physics stuff there, right? You can go all the way down. That’s just, physicalism no one should deny that. And sometimes people think that that’s what’s at stake. It’s not mm-hmm <affirmative>, it’s not, we’re not denying that there’s lower level physics stuff in these systems. The question is what’s providing the explanation is that lower level detail, what causes this behavior of interest and what explains it. So the question is what information is causally relevant and explanatorily relevant to the system. And that’s, what’s at stake in these questions, it’s explanatory reduction. So the, the suggestion is that for some things we’re interested in, in biology, medicine, neuroscience, social structure, this the real explanatory power it’s coming from things at a higher level. That’s where multiple realizability comes in here too, because you can have a cause that’s multiply realized by different lower level physical details.  

Lauren    01:06:30    Um, and it, in a way that makes it difficult, if not impossible to C the lower level physics, but basically the idea is, and one of the ways I describe this, or the way I think about it is it’s not how low can you go? It’s what gives you control. So it’s what gives you control over that explanatory target. Sometimes it’s stuff at a higher level and it’s not physics. So I sometimes think that it’s common confuse that kind of suggestion, that explanations and relevant causes are, are at a higher level with that kind of physicalism claim, the physicalism claim. Isn’t, what’s being debated. It’s what information gives you explanatory power and it’s causally relevant, but there’s lots, there’s lots of, uh, there’s lots of reasons. That’ll be really interesting to think more about, about why, um, it’s so easy to be reductive as scientists philosophers in an everyday life.  

Paul    01:07:35    Yeah. I mean, it’s, in some sense, there, there seems to be an easing of people’s, uh, stance toward reductionism because it, it, it does seem like something that’s kind of hard to let go of until you come to a comfortable level with thinking, um, anti reductionist. So, yeah. Anyway, um, let me backtrack and ask you, uh, a bit more about constraint <laugh> mm-hmm <affirmative>. So one of the properties of constraint is that they are relatively fixed, right? Constraints are relatively fixed relative to the, to the process that they’re constraining, but does, does that need to be the case? Can’t you have a, a constraint that is, um, transient and or itself a process. And can you have two mutually constraining processes? Is it, is it right to think about it like that as a possibility? Or is this, am I asking two abstract a question here also?  

Lauren    01:08:33    Um, you’re not at all <laugh> if you, no, uh, this, I mean, I feel like this discussion is very abstract in a comfortable way for okay.  

Paul    01:08:41    For fosterers  

Lauren    01:08:43    <laugh> yes. And, uh, well, no, I think that’s a great question. Yes. I do think that in many cases, when we think of something as a constraint, we also think of it as relatively fixed. And the way that I think about this is it isn’t that it’s entirely fixed in a strict sense where it can’t be changed, but that it is more fixed than other types of factors that might figure in the explanation. And that’s why it matters to suggest that it’s fixed. So if we think of a fixed constraint, like, um, like a maze and a rat or a mouse, that’s like in a maze moving through it, or, you know, a human walking through a building like the walls, um, and the hallways in the rooms of a house, we think of the walls of the maze as constraining the movement of the mouse.  

Lauren    01:09:38    We think of the walls of the house as constraining our movement as we move through the house. And it’s very easy for us to kind of think of those walls as fixed. And when you compare them to our choice to move through the space, we think of them as just more fixed than that choice. So, um, we can change the walls though, when we can change different types of factors that we think of as constraints. It’s just, we kind of forget that we can sometimes, and we can ignore constraints in our explanations because we think they’re fixed in ways that’s problematic. So with social structure, this a similar thing is happening. Um, where if you think that public policies are kind of fixed in place, then you’re gonna sometimes background them in your explanation, as opposed to considering what would things be like if they were to change or if they were different.  

Lauren    01:10:34    So I think the notion of fixed is capturing how the constraint differs from factors that are not constraints. We think of them as more fixed, but, um, and they are, they are harder to change and there’s different reasons for why they’re harder to change. Often they change over longer time scales, right? So the walls of the house, those aren’t gonna change as quickly as my decision of where to go. Same with public policies. If an individual has like three options in the current public policy, they can make that decision on a smaller time scale than changing the policy. So there’s, um, there are, I think good reasons for why we think of constraints as more fixed than the other factors, but that isn’t to say that they can’t be changed. And I think appreciating the fact that they can be changed is important for getting a kind of fuller picture of how these explanations work,  

Paul    01:11:32    What, uh, how many more different varieties of causation are there to, um, tease apart? Like what, what are the next five years of your academic life gonna be like, are you, it seems like you’re just, you can just keep picking out different varieties and situating them within the philosophy of science, explanation, explanatory, uh, approaches. Are, are you gonna run out?  

Lauren    01:12:00    <laugh> <laugh> I mean, <laugh>, I don’t, well, there are many different, there are many other different types of questions that a philosopher of causation can ask that don’t always have to do with different types. One way. I think about the, so I’m not, I’m not worried about running out <laugh> um, <laugh> I mean, look, in some ways it’s really important to just capture this diversity because the standard view has been yeah. Mechanism. It’s a one size fits all. This is the way it works. This is, and I mean, with salmon, the causal structure of the world is mechanistic. That is, that is causality in the world. So, you know, in my work and in other work, this is just being more appreciated that that’s not the best way to understand how scientists think of causality in the world, that there are these many different types of systems.  

Lauren    01:13:02    Of course, we have to say why they matter the way that I think about three types of these distinctions is first we have a bigger causal system, like a mechanism pathway cascade. That’s a big causal structure. There’s a lot going on there. A second type is just a simple pattern, like a causal motif or a causal pattern they’re sometimes called. And this is just the organization of how causes are organized together. So you can have a linear chain, you can have a positive feedback loop. You can have branching structures, you can have a bow tie structure. You can have a final comment pathway. That’s another set of distinctions and they’re more simple, just causal patterns. And then if you go one step down, you can talk about different types of single causes. So with Woodward, you talked about causes that are inva or stable mm-hmm <affirmative>, and you can think of, and talk about causes that are fast, or that are slow. Some causes produce their effects, irreversibly and others produce their effects. Reversibly. You can undo it. That’s a set of distinctions at the level of a single cause. So that is one way to think of these differences. You can have it at the level of a pretty complex causal structure or just causal organization, or just single causes. And this gives us a taxonomy of distinctions among causation. And, you know, the, the next step of work is to show how those distinctions matter, how are they gonna help us solve different problems in philosophy and in science?  

Paul    01:14:38    Um, here’s what I want to ask you with respect to multiple realizability is what is, uh, and I think that this follows from what we’ve just been discussing, what is, uh, causal heterogene and how does it, um, relate to multiple realizability? Maybe that’s one way to, we can discuss your work on that.  

Lauren    01:14:56    Oh, good. So the way that I think of causal heterogeneity is that if you pick an explanatory target like a particular disease, I’ll give the example Parkinson’s disease. This is a helpful case. I think there was, um, there were a lot of studies on Parkinson’s disease that at one point suggested that different patients with the same disease got it as a result of completely different causes. So suppose you have three patients with the same Parkinson’s disease phenotype in one of them, it was caused by a single gene in the second, it was caused by an environmental factor could be caused by pesticides. And in a third, it was caused by a combination of stuff, genes and environment. That’s a situation of causal heterogeneity, the way that I define it, because the same outcome, different instances of it can be produced by completely different causes. So, um, and what’s interesting is in medicine, we actually don’t like it when diseases have that kind of feature, right?  

Lauren    01:16:04    <laugh> we want the, we want there to be a kind of unified causal process that produces the disease. So that’s causal heterogeneity, multiple realizability is related to this, but it’s a bit different in this case. I mean, multiple realizability, isn’t a causal notion it has to do Withers. So what this means is that if you have a phenomenon at a higher level that that phenomenon or entity of interest, if you drill down that there are going to be different, lower level micro, structural details that could be present in different examples of that phenomenon. So one case that comes up, I mean, you can, it’s almost easy to do with anything. Um, one example that comes up is, um, cigarettes. So think of, so cigarettes in lung cancer is a causal relationship smoking in, in lung cancer. But if you think of just the cigarette itself, this is an example that Elliot sober philosopher biology brings up that cigarette can be multiply, realized by different types of carcinogens, different combinations of them, different types at a lower level, and the realization relationship.  

Lauren    01:17:25    And I mean, with a neuron, some of the best examples of this come from neuroscience, right? You can have different neurons that produce the same firing behavior, but they have very different types of ion channels. They have very different concentrations of them along the neuron, and those are micro structural differences. So multiple realizability is a situation where the same higher level entity can be realized by different things at a lower level. Another classic example is money. So money is something that can be realized by different things, right? You have the dollar, you have the Euro, you have you other communities, um, right. Other types of, you know, gold or precious metals. So that realization relation matters for explanation, because if the higher level entity gives you a better explanation for the outcome than the revisers, then that can be one reason to be non reductive essentially, or to give an explanation that appeals to higher level stuff, as opposed to lower level stuff.  

Paul    01:18:32    Why is it a bad move though, to let’s say, in the, um, the cigarettes case, right? Mm-hmm <affirmative> not appeal to just say, okay, there is a, a family, a set of lower level explanations that we can just, you know, consider like this is a reductive viewpoint, right? Consider the 30, let’s say, you know, it could be 30,000, but why not, you know, consider them as sets of, uh, reductive explanations rather than appealing to the higher level.  

Lauren    01:19:01    Good. One reason is if you appeal to all of the sets, sometimes there are so many that you would be citing 300 different things. Yeah. So you’re actually take you a really long time. So you go to the doctor and you’re like, Hey, what caused my disease? And the doctor’s like, take a seat. I gotta tell you <laugh> so take a really long time. It also, isn’t clear how to understand a disjunctive set of causes with an interventionist framework, cuz it’s not clear how to intervene on a disjunction. Mm  

Paul    01:19:33    Mm-hmm  

Lauren    01:19:34    <affirmative> um, and then another, um, another reason is it doesn’t tell you what they all have in common and that’s kind of, you know, that’s kind of what we wanna, we wanna know what unifies them. So, and also if you wanna make changes, right? Suppose you wanna prevent lung cancer in the population. Are you gonna tell people either a stop smoking or B stop inhaling carcinogen, you know, a one and a two and a three or this other combination of a no, you’re just gonna say smoking is the better causal variable. It gives you control over all of the cases of the disease. If you just focus on and then it tells you what, what kind of unifies them together. So, um, there will be different reasons in different, in different cases. But if you think of intervening on stuff in the world, and if you could just intervene on, on smoking at the higher level, you get control over all the disease, all the disease caused by cigarettes.  

Lauren    01:20:35    If you start to try to intervene on carcinogens, um, you, you can’t get that same kind of control. So for me, I’m an interventionist, causation and control are very much related. If a factor gives you that kind of broad level control over all cases of the outcome of interest, it’s an indication that it’s actually a better, more important cause for the outcome. It has more explanatory power than the kind of splintered factors at the lower level that are, uh, numerous. And that it would be hard to kind of use as targets to, to make that kind of change or treatment or, or prevention.  

Paul    01:21:16    All right. That’s I think that’s a good place to end Lauren. Thanks so much for talking with me. And I don’t know if we, um, clarified all the issues for everybody, but of course people can learn more because you, you write really well. And so it’s, it’s clear in the papers as well. So thank you for your work on causality and explanation, uh, and, and all that you do and continued success.  

Lauren    01:21:38    Thank you so much. Thanks for the invitation and discussion. Paul,  

Paul    01:21:57    I alone produce brain inspired. If you value this podcast, consider supporting it through Patreon, to access full versions of all the episodes and to join our discord community. Or if you wanna learn more about the intersection of neuroscience and AI consider signing up for my online course, neuro AI, the quest to explain intelligence, go to brain inspired.co to learn more, to get in touch with me, email Paul brain inspired.co you’re, hearing music by the new year. Find them@thenewyear.net. Thank you. Thank you for your support. See you next time.