top of page

S3 Ep2 - Kate Lynch Transcript

Kate Lynch on Causal Explanation in Science

Samara Greenwood: Hi, and welcome back to the HPS podcast, where we discuss all things history, philosophy, and social studies of science for a broad audience.

I'm Samara Greenwood, your host and today on the podcast we have Dr Kate Lynch discussing the topic of causal explanation in science. Kate is a philosopher of biology, as well as a wonderful lecturer in HPS here at the University of Melbourne.

In today's episode, Kate introduces us to the difference between ‘causation’ and ‘causal explanation’, as well as difficulties involved in assessing what makes a good causal explanation.

Some of Kate's own research looks at causal explanations of death, including controversies and complicating factors that can be involved in determining what cause of death gets recorded.

Personally, I was fascinated to learn of the varied practical, political, and even social considerations that can shape what cause a physician ends up putting down on a death certificate. I also appreciated the way Kate uses concrete examples like this one to draw us into further discussion of the role of social values in science, right down to this core task of scientists producing causal explanations of natural phenomena.


Samara Greenwood: Hi Kate, and welcome to The HPS podcast.

Kate Lynch: Hi, thanks for having me.

Samara Greenwood: First, I wanted to know, how did you make your way into HPS?

Kate Lynch: Like so many other guests on this podcast, I came in a very roundabout way. I studied philosophy and science separately and then ended up fusing them together towards the end of my PhD.

In my undergraduate degree, I did psychology and philosophy as my majors. I then got very interested in wildlife and in zoology, so I thought I'd go back and study a biology degree. Then, I started a PhD in philosophy, so I ended up completing those two degrees at the same time.

My first job as a postdoc was in the biology department, then I moved back to a philosophy department, so I've just been switching back and forth. Now I'm in a history and philosophy of science department, which is perfect. I get to keep doing some research with biologists as well as doing more hardcore philosophy work.


Samara Greenwood: How was it doing two degrees, a PhD in philosophy of science and your biology degree, at the same time?

Kate Lynch: Yes, I wouldn't probably recommend it. I mean, it worked out fine, but it was pretty hectic at points.

It was helpful because it was a really good way to meet biologists. I was in there as a student, also volunteering in labs, and it really helped my philosophy PhD because I got an understanding of how the science actually works. I think if you don't have that experience, in a lab or talking to real scientists, there's this real distance as a philosopher of science.

You might think, why don't scientists just do it this way, this would give you such a cleaner result or make a better theoretical prediction? Then you go into a lab, and you realize there are all these pragmatic constraints that you have to do the experiment in a certain way. So, it grounded me on how the science actually works.


Samara Greenwood: Next, could you introduce us to the topic you are interested in discussing today?

Kate Lynch: So, I'm going to talk about causal explanation in science.

But before I do that, I need to jump in with an even bigger topic, which is causation. This is a huge topic in philosophy and in science. People are interested in causal relationships because you might want to know the causes of some disease or what caused the formation of Saturn's rings.

Philosophers, their main interest when it comes to thinking about causation is: What is it? What is this special relationship that holds between these two things, the cause and the effect?

What philosophers do is first distinguish what the cause and effect are. Are they objects? Is it billiard balls on the table? Or is it events? Is knocking over my cup the cause of the coffee spilling? Or are they variables? Smoking causes lung cancer, right? You specify what the relata are, and then you think about what the relationship between the relata is.

There are many different accounts in philosophy about what this special relation between the two things is, what the causal relationship is.

A very influential account is the counterfactual account. If the cause hadn't occurred, then the effect would not have occurred. You think about what would've happened if things had been different.

But under that accountant, you get this problem that there's just too many causes. Peter Menzies coined this term ‘the problem of profligate causation’ where for any given effect there's just too many causes.

Philosophers can agonize over what causation means and how to understand causal relata, but then we get to this secondary problem of causal explanation.

The reason that I'm interested in causal explanation, as well as causation itself, is because - you might be happy with an account of what causation is, but that doesn't seem to give us enough to talk about causation in the sciences. We still pick out some causes as more important than others. That's been my primary interest.


Samara Greenwood: What is causal explanation then?

Kate Lynch: It's a good question. I think people are still grappling with what makes a good causal explanation, right? Why you might pick out something as the cause. Why one thing counts as a causal explanation and something doesn't, or why maybe you say one thing counts as a better causal explanation than another.

So, in some cases, like lighting a match, it's fairly intuitive.

But there are other cases which are not as intuitive, and there are also cases where scientists disagree on what the causal explanation is. And so, philosophers of science are interested in, first, what are the features that tend to make a good causal explanation? Are those features justified in us picking out good causal explanations, or are they just some weird artifact of our psychology? And then, how do we account for disagreement? Is this because there's no universal standards for good causal explanations, or are some people mistaken about the features that they pick out.

There are a few contenders that philosophers of science, and also some social psychologists, have come up with for at least explaining why some causal relationships or some particular causal relata seem to be more important than others.

Samara Greenwood: So, what are they?

Kate Lynch: Right, so there's a few. Some of the ones that I think give us good explanatory power are things that James Woodward, a very famous philosopher of causation, has written quite a lot about. They tend to relate to ways that we can intervene on the world and have control over variables.

One of the reasons people think that we developed this complex causal cognition is because it's important for us to be able to understand which things we can intervene on in the world, which things we can control.

Woodward proposes specificity as a really important feature of causal relationships. So some causal relationships are going to be more specific.

For instance, if the effect of interest would be ‘turning on the radio so I can listen to The Hottest 100’. For non-Australian listeners, this is a music countdown that happens around this time of year every year.

So, I want to listen to this particular radio station. Well, what would cause that radio station to come on? We've got to turn the radio on. That's not a very specific cause, right? That switch you can turn on and you're going to be able to get lots of different radio stations. But then the dial, which adjusts…Gen Z is listening to this! You have to think about a digital dial here. I'm turning a knob in the air. It's just showing my age… You turn the dial to the different stations, and they're going to specifically map to the outcome that you want.

That's an instance of one type, at least, of specificity, right? It seems to be a better causal explanation that I set the dial to this station to get this output.

When you think about science, the kinds of causes that people pick out often have this feature.

One of the topics I'm interested in is causation in genetics. And one thing that people think is really special about genes is they're specific. You've got this very specific kind of genetic code that quite often very specifically maps to just one kind of protein or one kind of outcome.

That's not always the case, but this is a feature that's interesting. A kind of overall feature as to why people might pick out genes as more important. So that's one reason you might think some causes are better explanations than others, specificity.

Another one is stability. So how general a causal claim can be or how well that causal relationship would be maintained over changes in background condition, changes over time, changes to other variables.

For instance, if you have a virus that causes a set of symptoms, and everybody infected with that virus is going to get those symptoms, that's a really stable causal relationship. The virus is causing the effect (the symptoms), no matter someone's age, medical conditions, sex, any of those other kinds of variables. But if you have a less stable causal relationship…

So, a very contentious one is COVID 19, where the virus causes at least severe symptoms in a subset of individuals, right? So, if you think about COVID 19 causing death, for instance, there's a causal relationship there which is different depending on the background conditions - age, comorbidity, any other kind of background factors.

So what scientists are often interested in is finding these stable causal relationships that persist over changes in background conditions, changes over time, particularly in medical science.

If you want a really successful drug treatment - which again would be a causal relationship, if you're giving someone a drug, it's going to produce an effect - you want something that's really stable, that's going to work for people of all ages, sexes, different kinds of medical backgrounds, genetic backgrounds, diets, et cetera.

So stability, I think is another interesting and important way of pointing to a useful causal relationship.


Samara Greenwood: What about when you have an outcome that clearly has multiple causes? Like a knee injury, for example. How does that fit into the picture?

Kate Lynch: In human health, particularly in medicine, the more we're learning, the more complex we realize things are. This is where other kinds of factors might come into causal explanations.

You might think that some sort of pragmatic feature might be why you pick a cause for knee injury.

Say you're interested in causation because you want to prevent injuries. I don't actually know much about injuries, but maybe you tell people to run with soft shoes or to lift in a certain way, not play netball, whatever it is. You can point at those as kind of pragmatic causal influences that you can do something about.

I'm sure age is also a factor. The older you get, the more prone you are to these kinds of injuries. There's not much you can do about that. You can't intervene on someone's age. You can't tell someone to stop aging. So the focus of a useful causal explanation there would be the thing you can intervene on.


Samara Greenwood: Are there aspects of causal explanation that are hotly debated today?

Kate Lynch: Absolutely.

There's lots of examples in science about what the best causal explanation is. One very controversial one at the moment is how we identify causes of death.

Again, to go back to COVID 19, there's been this huge controversy about whether people die of COVID or die with COVID. That's because on every individual's death certificate you need to put down one underlying cause of death. So, you basically have to pick out the causal explanation for that individual's death.

If you think that the most important cause was COVID 19, then they died of COVID. If you think there was another more important cause, maybe the patient had cancer as well as presenting with COVID 19, then you would say they just died with COVID, it wasn't the causal explanation.

That's been very controversial for a whole variety of reasons, but it's also been super important because from a public health perspective, we want to be able to accurately track how many people are dying from such a serious disease, particularly when it's an infectious disease.

So there's a little bit of controversy about how the most important cause is determined in this case. Now it's very hard in an individual case to pick out causes like this because you can't do this kind of reasoning about ‘what would have happened otherwise’. You can't run one person's timeline again.

But there are other factors that come into play. So, the WHO, who write the guidelines for how to determine causes of death, say infectious disease gets priority on a death certificate. That’s because these are the kinds of things we really want to know and really want to be able to track to prevent future transmission and future deaths.

So, some infectious diseases like COVID, influenza is the same, would sometimes get causal priority on a death certificate. However, that is very controversial, of course, for people thinking about how to count COVID deaths.

Some people have suggested, why not put multiple causes of death, and give them weightings, say this one is more important, put a percentage value on them? But that kind of just bumps the problem. Where do you pull the percentage values from? How do you know for an individual that it was 80% cancer and 20% COVID? It's very hard to make sense of what those numbers would even mean. And doctors are already so overwhelmed with all their other duties.

There also seems to be space for values coming into play and this is something I'm getting more and more interested. For example, other reasons, beyond the pragmatic, that people might pick out particular kinds of causes to put on death certificates, and other diagnostic documents, is because it might help a patient in terms of getting the insurance benefit they need.

So, there's this kind of value-laden way of thinking about causation. There's some other evidence that doctors do think about the public health implications of what they're documenting. If they're tossing up between two causes, ‘which is the underlying?’, they might put down the one they think deserves greater attention, needs better funding. There's a study where that's come out by interviewing doctors.

Then there's also the role of things like stigma. So, you think about filling out a death certificate and putting an underlying cause of death that's going to impact the family. There are some causes of death, for example when they relate to self-harm or substance abuse, that doctors might shy away from having as the underlying cause of death.

There are also social conventions. There's been this big debate about whether obesity counts as a cause of death. It is something that you can record on a death certificate under the current guidelines, but very few death certificates have obesity as an underlying cause of death, even though there are some petitioning to say, ‘well, this is a feature that really is contributing to many, many deaths.’

There are all these social, moral, political reasons as well that feed into the way that we record causes in this context.


Samara Greenwood: Could you provide some examples of good versus flawed causal explanations?

Kate Lynch: I gave those examples of features which might lead to good causal explanations like stability and specificity because there's some rationale there. It's telling us something about the causal relationship relating to control or manipulations.

There's also lots of evidence that the way that we causally reason, the kinds of causal explanations we prefer, sometimes just boil down to things like being simple.

Tania Lombrozo at Princeton has done a lot of experimental research where she gives people different kinds of causal explanations and gets them to choose which they think is the best explanation.

Sometimes the simplest explanations are not the most likely ones to be correct, right?

So, she's devised these vignettes where there's these quite inaccurate but simple explanations and that's what people prefer.

This is where there's this bit of tension between the kinds of causal explanations we like. This is where philosophers mount their arguments for features about causal relationships. Do we like them for good reasons? Or do we just like them because they're simple, or perhaps because they reflect some other kinds of aspects of our causal cognition, like norm violation.

Joshua Knobe's done a lot of work on this, about the relationship between normative reasoning and causal reasoning. Quite often we might pick out something as causal just because it's abnormal or unusual.


Samara Greenwood: Is there an example of that? I'm interested in that. Maybe a little bit more explanation.

Kate Lynch: In one of the experiments that he did, Knobe constructed these vignettes where there are professors and administrators. There's a whole heap of pens left on a desk, say in a university faculty office. The pens are meant for the administrators and the administrators are allowed to take the pens. The professors are not supposed to take the pens. But they do anyway. The pens are slowly dwindling, and then there are two left. One professor and one administrator take the last pens. Then an administrator needs to use a pen. There's no pen.

You ask the question, well, what caused the problem of there being no pen?

Now, in some sort of strict sense, both the professor and the administrator caused the problem. But because it's the professors that violated some kind of norm, in this case a social norm, people tend to pick out the professors as being the cause of the problem [not just morally responsible, but causally responsible as well] with there being no pens.

Then some of the research I've been involved in with social psychologists is about particular kinds of causal relata that we tend to privilege, maybe in a biased way. Genetics is this good example.

A lot of social psychologists think that we have implicit essentialist biases. Before we knew anything about genetics, this was a feature of human reasoning. We think of things as having essences. We think humans have essences, trees have essences, right? There's something internal, within an object or an organism, that makes it what it is. It’s kind of deterministic and invariant.

People tend to think about genes in this way, so genes fulfil this ‘essence’ placeholder role.

Because we tend to think about genes in this special way, that's another reason we might pick out genes as being the most important cause.

If you think about the causes of sexuality, people think that your sexuality is really a part of who you are, it's part of your essence. It's kind of intuitive and it makes sense to think about that as genetically caused. There's something that feels uncomfortable about thinking about social origins or environmental origins in that example. But that doesn't necessarily mean that's the way the evidence stacks up. In fact, in that example, it's very hard to pull apart what the causes are.

What we've found in some of these experiments that I've done with social psychologists is that not only the general public are prone to these, what are called genetic essentialist biases, where they pick out genes and think about really strong causal relationships with genes in an unjustified way, but even professionals tend to do this.

We've done some experiments with medical students who have these biases towards picking out genes and that influences the way that they would diagnose a patient and the kind of treatment recommendations that they would make.


Samara Greenwood: That is really interesting. Is there potentially some link there to a human need for certainty and foundations for things? As in, that feels more solid than a more complex, conditional, contextual explanation. I'm just finding that in other areas.

Kate Lynch: I think it's a very human thing, we're going to try to find a heuristic if we can.

This gets back to Lombrozo's work on simple explanations being preferable. When people think about causation, they might fall into thinking about determinism because it's a lot easier to think about determinism than to think about things being interactive and complex and multi-causal.

In science we hope we can break away from that a little, but of course we're all human, we all have these biases, we all use heuristics. So even with scientists who are very careful, often in their experimental designs and the conclusions they draw, even in those contexts, you sometimes see people slipping into these errors when they make causal reasoning errors or even just having really overblown causal claims about things.


Samara Greenwood: You've talked a little bit about how this topic relates to some of the research areas you're working in. Were there any others that you'd like to tell us about?

Kate Lynch: There’s kind of very two very separate ways of thinking about causal selection or causal explanation.

There are the philosophers who are trying to find ways to justify picking out causes. What are some good grounds for a causal explanation? Is it stability or specificity or some feature of the relationship that tells us something about the world?

And then there's the psychologists and the experimental philosophers who are interested in what's influencing [our selection of causes]. Is it because they violate perceived norms? Is it because they're simple? Is it because there's something pleasing about them? Are there some biases influencing what we pick out here?

It would be nice to think of a way of unifying those two very distinct research areas. One surprising way of doing that, that I'm working on at the moment, is to think about the way that values influence all sorts of things in science.

There's this huge literature, and I know people have talked about this at length on the podcast, on how values are embedded in science. Across all aspects of science, at every step of the process, there are these implicit values that can make a difference. I think what has been overlooked is the way we think about causal explanations is also influenced by values.

Philosophical things - like stability and specificity - you might think about them as epistemic values. They fit into the epistemic or cognitive values framework, where it's giving you some indication of the knowledge that you might gain by looking at causal relationships. Then you have these non-epistemic values like moral, ethical, social values - like norm violation - or aesthetic values like simplicity. They also seem to make a difference. So, I actually think the ‘values in science’ literature might be a really nice way of unifying these two very separate areas of research.


Samara Greenwood: That seems like a great area for future research. Okay, so now jumping to a broader level, how do you think research on causal explanation - like you've been describing to us - might be interesting to a range of scientists? How might it inform their work?

Kate Lynch: I think most scientists are engaging in some form of causal reasoning without really thinking about how they're coming to those conclusions.

One thing that philosophers of science like myself often do is, we look at the science, we look at the causal claims scientists are making, and then go and look at their methods and see whether the kinds of claims they're making are justified.

One example of this is - I've done some work with Maureen O'Malley on microbiome research, looking at the methods that have been used and the causal claims that people are making and showing that the claims are overblown, or they don't quite match up.

If you take some of these Woodward dimensions, they're not very stable causal relationships. We don't have specific causal relationships in a lot of that field yet. So, you might want to hold back on making strong causal claims.

Also, thinking about there being different kinds of causal relationships and different ways of representing causal relationships that could be useful for scientists, rather than just going from the standard research methods that they're used to doing, whether it's a case control study or an RCT [Random Controlled Trial] or whatever it is, and then saying, ‘we've got a significant result here that means it's causation’.

Now, I know scientists are often very careful about not reporting causation. They talk about associations and correlations instead. And it's often that translation from scientific article to popular media article where things get overblown. What I've shown with Maureen is that even at the scientific level, if you go and read these scientific papers, people are not really being careful enough in the way that they describe their research results.

[00:20:33] Samara Greenwood: Even having some reflexiveness about what you're emphasizing as a causal relationship when there's a range of potential ones to focus on, having some sense of what you emphasize personally.

Kate Lynch: I think a contribution from philosophy of science to scientists is just to be explicit about what's often going on implicitly.

There's nothing wrong with scientists having a context of inquiry - having this or that reason for picking out a particular cause. But disagreements occur in science when two groups of scientists have these different implicit contexts of inquiry.

For example, if one is interested in control and manipulation and another's interested in prediction, that might lead to them both picking out different kinds of causes as the most important cause. Then they might fight over it. But, if you think about why they disagree, that might help further some of this scientific research.

It's the same when it comes to the public. Quite often the public can be sceptical of scientific research. Like the cause of ADHD. Some people are staunchly, ‘it's genetic’. Or there's a very controversial causal explanation of ADHD - is it just complex trauma from early childhood? Giving a developmental account. The scepticism that comes from the public to scientists picking out causes might be better understood if we recognize that there are these reasons that you might pick out some causes over others.


Samara Greenwood: Yes, it is that sense of context about it. Why are we interested in this cause or relationship? That's directing us to a primary cause for this purpose, which is different from what you're talking about and that's why we're talking past each other. Those kinds of conversations are useful.

Kate Lynch: That's right. And that is something that I think a lot of scientists are quite hesitant about. They think about things being context dependent - that this is going to just devolve into everything being all relative, there's no scientific objectivity. That's something that scientists are quite uncomfortable with. But you don't have to go all the way down that track just to think that context of inquiry might influence what causal explanations we prefer or the kinds of things we study.

Again, it's this ‘values in science’ literature. There are some things that are going to guide what we think are important in science, and that's inevitable, and it's not necessarily a bad thing either.


Samara Greenwood: No, and what's emphasized versus put to the background in a particular case. It can be more subtle than we often think about.

In what ways might this research be of interest to the general public?

Kate Lynch: The cause of death stuff in particular, I often find people who have had their own personal experience, “When my family member died, this was on the death certificate, but really this was the cause.” They have this personal connection with the importance of recording one cause over another, and perhaps if there was a greater understanding of why doctors might put one cause over another - they might be tracking something important from a public health perspective, there are constraints over what doctors can put on these death certificates, because of the way results are tabulated and feed into public health data. Again, there's this context of inquiry that informs the kinds of causes that are documented.

Perhaps if there was a greater understanding there, there'd be less animosity, less disappointment - better communication between scientists and medical professionals and the public.

Samara Greenwood: Thank you so much, Kate, for being on the podcast. It has been an absolute delight.

Kate Lynch: Thanks for having me.


Samara Greenwood: Thank you for listening to season three of The HPS podcast. If you're interested in the detail of today's conversation, you can access the transcript on our website at Stay connected with us on social media, including Bluesky for updates, extras, and further discussion. We would like to thank the School of Historical and Philosophical Studies at the University of Melbourne for their ongoing support. And thank you for joining us in the wonderful world of HPS.

We look forward to having you back again next time.


bottom of page