Transcript of my Pi Day interview with Mathieu Blanchette

Dear readers,

Some of you have asked me to post transcripts of my interviews so that you could go through them by reading rather than listening to them. Today, I’m putting up my first transcript, of the interview I did with Mathieu Blanchette. I would like to thank my assistant for preparing the initial transcript. I will be posting the transcripts of the other interviews I did over the next few weeks, and I also got a couple of new exciting interviewees lined up, so stay tuned!

Transcript: Interview with Mathieu Blanchette

MP: Hi Mathieu! Thanks so much for finding the time to speak with me today.

MB: Hi Leonid! It’s a great pleasure to join you on this.

MP: Fantastic. I’m really glad we’re getting a chance to discuss things and I got this start question that I would want to ask you about your responsibilities as a professor of Computer Science at McGill University. Of course there are different aspects to what you do; there is teaching and there is research and there is administrative work so I’m wondering which of those you feel is the most important aspects of your work?

MB: Yeah, it’s a good question. It’s always a balance between the three and they are all interrelated as well. I would say that for me personally the aspect that I think has the most impact is the supervision of students, whether they are undergraduate students working on projects or graduate students, so that’s kind of at the intersection between teaching and research -because it’s the supervision of research projects. I feel that this is how in the long term I have the most impact because training a person to become an independent researcher means that once they are trained and then they get to do research that I’m doing, so it’s really kind of multiplicating the effect of our work. So the more qualified students I train the more research gets done, and so it’s not just the research that I’m doing with my students now but the research that they’ll be able to do in the future. That, I think, is the main impact and that’s what excites me most about the work. That is not to say that teaching like classroom teaching is not important but I think the most impact is through these closer relationships with students on specific research projects.

MP: Fantastic, so you mentioned you supervised undergraduate students, Masters students, I guess PhD students, probably post-docs as well. Out of these different groups, which would you say you enjoy working with the most?

MB: Well, it’s a good question. I must say that I think what I enjoy the most is to work with people who are very excited and dedicated to what they are doing, and often the group of students with whom I find the most of that is undergraduate students.  So I think at McGill and I think at many universities undergraduate students have the opportunity to get involved in research projects, and in particular during the summer. And I find that although these students might not have all the knowledge that more senior students might have, they have the excitement. The first time they do research projects they are really involved in it and they are really excited about it and they invest themselves completely into it and they’re bright students for the vast majority. I think in the area like where I’m working – bioinformatics – it is not the case that students need to have a very very deep knowledge of mathematics or computer science to be able to make an interesting contribution, so students can really become researchers and have an impact early on in their career, and this is where I find the most satisfaction. Often these students will go on to a Masters or a PhD and they’ll do great things and it will continue to be fun to supervise them, but the most kick I get is out of supervision of undergrads.

MP: Fantastic. Well I was just thinking back to my own experiences as an undergraduate researcher under your supervision –actually that was also my first time doing any kind of research.

MB: That’s right, I think you were among the first undergraduates or any students I supervised when I came to McGill if I remember correctly, and you really are among the people who got this trend started, but it has continued after you moved on and it continues to be the the newer, satisfying part of my work.

MP: Fantastic. Yeah and I guess this kind of brings up another question in my mind. So of course you know in the area of bioinformatics and computational biology it’s not so knowledge based and you don’t need to have necessarily a ton of background, I guess more important is the excitement and interest you have and the willingness to work hard to make a contribution. So thinking then of the more advanced students, let’s say PhDs and especially post-docs, do you feel that sometimes they get as, and I’m also thinking of myself, as a current post-doc, do you think that sometimes people at more advanced levels tend to get sort of set in their ways of doing things and tend to you know stick with things that have worked in the past rather than taking risks and really sort of trying out new things, new ideas, and so how do you feel about that?

MB: There is certainly a risk of that and it’s more comfortable to keep doing what you’re good at when you’ve become good at it whereas these younger undergraduate students are not formed yet. They don’t know what they’re good at and so they’re perhaps more open to trying very weird new ideas. On the other hand, I think that in a field like computational biology where first the technologies generating the data that are being analyzed, these technologies move very quickly, and so the problems change very quickly as well. What was an interesting problem five years ago may not be not so much of an interesting problem anymore, and so we really have to keep on our toes to be able to react, and that means using or taking new approaches to problems. And not just the technological advances, but the kind of questions that people are asking about biological systems are evolving very quickly as well, and so I can give you an example if you want.

MP: Sure, that would be great.

MB: During my career, I’ve been thinking of DNA sequences as a computer scientist would as a chain of characters, As, Cs, Gs and Ts that fit very nicely on any computer file and then can be analyzed in all kinds of ways.

MP: Sure.

MB: But more recently people, biologists have known all along that in fact a DNA sequence is not a chain of characters. It’s a molecule. And that molecule is basically a long string of smaller molecules that are the As, Cs, Gs and Ts and that molecule has a 3D shape and it’s folded inside the nucleus of cells, and that shape really has a major impact on how the information that’s inside that sequence, how the cell interprets that information. And so what we used to think of as a very linear set of As, Cs, Gs and Ts now becomes a very geometric object in 3D and the geometry matters a lot, and so we have to adapt to this. I was not used to thinking in terms of geometry but now I have to and so that’s one example where really the paradigm shift forces you to adapt your types of questions and the approaches you have taken.

MP: Absolutely. Yeah that’s a fantastic example. I think that as we know more of the technology progresses we are kind of forced to think in new ways about even such fundamental things as DNA molecules for instance.

MB: Right.

MP: Right. That’s a really good example. So I also have a question about sort of the way that you manage the projects and the collaborations that you’re involved with because you’re involved in a lot of projects and a lot of collaborations. So I guess specifically I am interested in, you know, is it ever the case that you feel there is too much going on, and what are your criteria for you know making decisions and making choices as far as the projects that you keep working on versus the projects that you leave behind perhaps or terminate.

MB: Well this is a very tricky question and I don’t think I have a very good approach to this. I am overwhelmed with all the things that are going on and I’m getting involved in, so I have the opportunity of getting involved in all kinds of projects that are all interesting but that all or many of them require very different approaches, require being familiar with areas of research that are completely different, and that’s really hard because each of these areas is moving very quickly. There are hundreds of papers published every month in each of these areas. And it’s impossible for one person to keep track with all these things. So on the one hand I rely very much on my students who are working in each of these areas to keep up with the literature and tell me about what they read, so that helps keeping me up to date. On the other hand I have to say no to some very interesting projects or people coming to me with ideas. At some point it would be doing them a disservice, to them or to my students, to commit to too many things because then I would not be able to well-attend any one of them. But it’s very hard, it’s so hard to say no to interesting projects and I don’t think I’ve become good enough at doing that yet. I am involved in more things than I can handle. There are some things that are, that get delayed a little bit, but with a good group of students and post docs and people like this we get through. And I think it’s important to push yourself in terms of getting involved in projects that might not be directly in line with your main line of research in your lab so that you get exposed to these ideas, the example I was giving you earlier about the 3D conformation of DNA itself was not something I would have thought about but when my collaborator Josée Dostie came to me with these questions I felt that this was something important to get involved in and that is taking time and it’s a lot of efforts but that’s how you really move things forward; otherwise you keep making small incremental steps to what you’ve already done.

MP: That’s right.

MB: I don’t have a very clear strategy here. I go by the guts. I guess it’s especially because it’s an area that is really growing quite quickly and there is a lot of new data, but also a lot of new subfields that are opening up on a regular basis as the technology progresses.

MP: That’s right.

MB: So I think you have to keep an eye open on your main line of research, but you have to be open to these kinds of things that are not exactly in line with this but that can inform or that can help towards that main direction that you have.

MP: And how would you describe the main direction that you’re pursuing and has that changed over the years?

MB: So the main direction that I am pursuing is to understand what is the function of different portions of the human genome. So the human genome it is a sequence of DNA of about three billion As, Cs, Gs and Ts. We kind of know the function of maybe one percent of this, which is the genes, but we don’t really know very well, we don’t have very much information about the pieces of sequences that are there to activate or repress genes when they need to be activated or repressed. Those are called regulatory regions. And much of the work that we’re doing is to develop computational approaches to better understand where these regions are located and how they work. And you might ask or people might ask how can computer scientists say anything about what’s going on inside a cell? Well, one approach that we’re taking is to study the evolution of DNA sequences, so we know today the DNA sequence of a human, but we also know the DNA sequence of several other species like a mouse or a dog or a cat. And by comparing these sequences we can learn pretty accurately what is the function, or we can predict what is the function of different portions of the genome and that requires the development of pretty sophisticated mathematical and computational approaches and that’s what we are after.

MP: OK I guess then following up on that I know you mentioned that one percent of the genome is genes, how much would you estimate to comprise of the regulatory region? All the rest of it, or is it also a small fraction?

MB: Well, there’s still debate about that. There’s pretty clear evidence that there’s at least two or three times more DNA that is there for the regulation than there are genes. So if there’s one percent genes then there would be two or three percent of the DNA that might be regulatory. And then if that’s the case then the next question is well what about this ninety five/ninety six percent of the DNA that would not be genes and would not be regulatory regions, well why is that? And it might be there because it fulfills an important role that we don’t really know already and as research progresses we discover a role for more and more of these regions. And it’s also very likely that much of it is there not because it contributes anything to the function of the human cell but for other reasons. Because there’s all kinds of mechanisms that add basically random pieces of DNA to the genome; as long as they don’t hurt too much then they will just stay adhered to the genome and it looks like a large portion of your genome or anybody’s genome is made of this DNA that probably isn’t doing very much or anything at all to help. But it’s really, really hard to just look at the piece of DNA and say, oh yes that’s clearly the regulatory region whereas this one is just clearly not doing anything; to a human or to a computer it’s all just As, Cs, Gs and Ts, and so that’s the challenge – to recognize what is the function of each of these portions and which portions might not have any function.

MP: Sure. And if we were to, you know, go back to what you were saying earlier about the conformation and the importance of the three-dimensional structure, do you feel that that’s like understanding or if let’s say we do a thought experiment ok where we know exactly how you know the DNA sequence folds in a particular situation. Would that sort of give us, which of it are actually sort of doing stuff and which are not?

MB: Yes, it would, definitely. It would be extremely informative because right now there’s, if we just look at the DNA sequence as a chain of characters, there’s a lot of regions that look like they have all these signatures of a region that should be functional but when people test them in real settings they don’t do anything. And so what is one possible explanation for this is that in the three-dimensional confirmation of DNA inside the nucleus somehow these regions are prevented from doing the job that they could do and so they are kind of blocked inside or something like that.

MP: That’s right. That’s exactly right.

MB: And so knowing the three-dimensional structure it would be very important. That structure, by the way, is not fixed. It changes with time and with different, I mean your skin cells and your brain cells have essentially the same DNA but that DNA is not arranged in the same way in the nucleus and that’s in large part why they behave differently, and so this is really I think a very exciting area of biology that raises very challenging mathematical and computational questions. So I think it’s really a direction that has a lot of future, a bright future.

MP: Fantastic. Well I think that was a very interesting sort of thought experiment to do and also a very interesting set of question to explore. I definitely am starting to understand better now just from talking about this with you right now. But I was also wondering about this last thing that you mentioned when you talked about the mathematical models and the computational problems, so would you say, you know, how would you sort of describe the balance between those two, so is it, you know, more of the case that we need to build good mathematical models, or is it more important that we be able to sort of compute something even if it’s not quite necessarily the best description or the perfect description of the system that we work with, but we just need to get a computational answer, so how would you describe that tension between those two things?

MB: Well, there’s definitely a big tension between the two. I think one of the main challenges in my work is to translate biological questions into a more formal mathematical or computational question. And much of the success or the failure of a project lies in this translation of a biological question into a mathematical one. Now, there’s the tradeoff between the sophistication of the mathematical model versus the computational side; it happens every day. And it happens at several levels. One is that it’s typically easier. So first, I am dealing with large data sets of DNA sequences or three-dimensional conformations, and so we cannot think separately about the mathematical aspects and the computational aspects. If we want to go somewhere both of these things have to fit together. And that means; sometimes that means simplifying the mathematical aspects so that we can do some computation on it and get some answers on the large data sets that we’re talking about. And so typically the way I like to approach a problem is that we’ll try to translate the biological question into a mathematical question and come up with a mathematical model that we think is the most appropriate, irrespective of computational questions. Then try to develop the computational aspects that would allow us to study that model, and most of the time it’s not possible because there’s just, it’s too complicated and/or the data set is too large.

MP: Sure.

MB: But then, you can make principled choices about what aspects of your sophisticated mathematical model do you want to give up on or what kind of approximations you want to do so that you know what you’re giving up on.

MP: And I think there’s a lot of pressure in our field to get results out quickly.

MB: That’s true, absolutely true. When there’s a big study on autism that identifies certain genes that might be involved in something, I’m just taking this as a random example.

MP: Sure.

MB: There’s lots of people involved, there’s millions of dollars that have been invested, and there might be competing groups who might be getting their results out before ours, so that there’s a pressure to do relatively quick analysis so we can get a paper out quickly, so the leaders of these projects might not want a mathematician or a computational biologist to take two years to come up with a solution which is what it might take if we wanted a really satisfying, mathematically solid solution, so there’s this pressure that is happening all the time.

MP: Sure, yeah.

MB: And I think it’s important to resist that pressure to some extent and to say: well, I need to be able to come up with a reasonably good mathematical model and the computational aspects that go with it, maybe not perfect but that’s a fight that’s going on every day in my work, or not a fight, but a tension that’s quite difficult to resolve.

MP: Absolutely, and do you feel that the increase in computing power tends to alleviate that tension, or is it actually the case that, you know, with the increase of computing power we also get an increase in the amount of data that’s coming in and so the problems also become harder, perhaps they become harder faster than our resources actually increase?

MB: Right. I think having a lot of computing power is good, and sometimes it’s good enough, meaning you don’t have to be too clever about how you solve a particular problem because you can just throw a lot of computers at it and you’ll get your results. And that’s fine, and that’s useful to be able to move on to more interesting questions, so if you don’t have to spend months optimizing a particular program so that it runs close enough, that’s a good thing.

MP: Right.

MB: I think though that what you were saying towards the end of your question, it reflects the reality that if you think in terms of DNA sequencing power, like a machine that I could have on my desk can generate a billion pieces of DNA, and if you look ten years back the cost of having done that would have been probably ten million dollars, whereas now it’s a thousand dollars.

MP: Right, yeah.

MB: And so this aspect of computational biology has changed very quickly. The amount of data that can be generated very quickly now is probably a million fold more than it was ten years or ten thousand or a hundred thousand fold more than it was ten years ago, and the computing power has not scaled to that extent. So data generation increases a lot faster than computing power and that’s one big concern, and the other one is the sophistication of the questions that we want to ask, which require more and more advanced algorithms and mathematical approaches, which means more need for computation. And so we’re not about, so the computing power is not about to get rid of the need for sophisticated math and computer sciences.

MP: Right, OK. Fantastic. I think that really describes the tension quite well, and it’s definitely something that I’ve also experienced a few times when, you know, things really do become a little bit stressful because there’s a pressure to, you know, analyze things quickly, but at the same time you know analyzing things well sometimes requires a lot more time than we actually have, and so definitely, you know, some of that tension is present. So I wanted to ask you about another aspect of your work, which is the teaching. So you got the very prestigious Leo Yaffe teaching award in 2008, and you were awarded that by the Faculty of Science at McGill. And so I guess my question there is this: what are some of your secrets in the way you approach teaching, and, you know, how did you do that and how did you get that award?

MB: Right, well there’s no secret, I think. Teaching is something I love doing, and I think that helps doing a good job at it for sure. To me, the most important part about teaching is not really conveying advanced concepts in computer science or mathematics, but it’s conveying the excitement for, or the interest for why these concepts are useful. So once somebody understands, somebody gets excited about a particular question then they’ll want to know how to solve that question and they’ll be willing to listen to you explaining some more advanced computational or mathematical concepts, and so on, whereas if I just go in front of a class and I say, well this is what a binary search tree is and here are the properties, it’s not so exciting. Because why do I care? And so I try, and it’s not always easy, but I try to spend as much time explaining or motivating why a concept is needed as explaining itself. I think, especially now, with all the information being available on the internet about all these things, what students need the most is motivation and more than somebody who will just read the textbook in front of them, basically. And so this is what I’m trying to do and it’s not always easy. There are some basic concepts in computer science and math that are hard to motivate, especially in the context of the knowledge that students have at that point, but this is what I’m trying to do.

MP: Yeah, you know, I can definitely tell from the fact that, you know, the first computer science class that I took that you taught, you definitely succeeded in making it really interesting and exciting, and I was actually not completely sold at the time on computer science as an idea because I was actually still playing with the idea of doing math and physics, and then shortly after that I of course decided to switch to mathematics and computer science. Not sure if I mentioned that to you before, but that was, sort of, definitely one of the motivating factors, because I realized how interesting that could be.

MB: I am glad; I hope you don’t regret it.

MP: No, it was definitely the right decision for me, although I still think that physics is really fascinating.

MB: Oh yeah, the most fascinating thing is the intersection of all these areas.

MP: Perhaps, yeah. So I guess another question I wanted to ask you is about language, and this is perhaps somewhat controversial, although perhaps not really so. It’s always been an interest of mine to sort of explore this idea of, you know, how can we have a community in the sciences and the mathematical sciences, especially in the life sciences that is inclusive of native speakers of different languages, and in particular, for yourself as a native speaker of French, I was wondering how easy is it for you to, you know, have to basically do most of your research, you know, writing and publishing in English, and what are your thoughts on, you know, well, first of all, is there a need to make things more inclusive, and if so, how would you start bringing that about, or how could we start to think about making that happen.

MB: Yeah, that’s a hard question. It’s not a question I would ask myself very often because most of my career, although I am a native French speaker as you’re saying, all my work has been done in English and then most of my studies have been done, well most of my graduate studies have been done in English, and at this point, actually, it’s easier for me, actually, to communicate science in English than in French but, that being said, I think it would be a very useful to be as inclusive as possible in terms of languages, and of course the tension is between accessibility of information, say if I write something in French, only people who can read French can understand what I’m writing, and that might just be 10% of the scientific community. Nonetheless, the ability to write in English is probably a challenge for many people in countries outside of Canada and the US and that probably is slowing down the advancement of research. So I don’t know if there is a way around this. I think, and my hope is, that automated translation tools are getting better and better. They are far from perfect, but they are getting better, and you can hope that they’ll get pretty good, pretty soon, and at that point it might become possible to publish something in French and have it translated automatically to something that will be actually readable and understandable by somebody in another language.

MP: I guess another sort of relevant fact to the discussion that I just wanted to bring up, and I am not exactly sure on the statistics for this anymore, but it’s actually the case that in the last few decades, the relative proportion of all research in science that has been communicated in English versus other languages has increased quite dramatically, because even as late as let’s say the middle of the 20th century, there was a wide variety of French and Russian of course, Spanish, and German and perhaps even Chinese and Japanese research journals that were mostly published in those languages. And all that has very quickly sort of become absorbed into English language journals. I guess one of the things I was thinking about was that’s sort of a necessary consequence of everything becoming more global.

MB: Yes, I think, well, I don’t really see a way around it. As unpleasant as it is, the accessibility of information now has become so easy and so important. If it took me two months to order a journal written in Japanese and have it translated so that I can understand what it is talking about, things would go a lot slower. And honestly, I would probably not do it. I would be too lazy or in too much in a hurry, and so, that paper would not be read by me.

MP: Sure, that’s fair. Yeah, well there is definitely hope and I know there is a lot of work being done on automated translation tools as you mentioned, so perhaps one day we will get to a point where we can actually translate things back in a high quality in all these different languages.

MB: And eventually that might even be better than somebody who is not a great English writer try to write their paper in English. It might be better to write it in their native language and then have a really good translation, so that might get there.

MP: And then I have another couple of questions that are related to human genetics and human genomics, which you’re an expert on. So, the first question that I wanted to ask around that is, what do you feel is the payoff like with work in regard to human genetics and human genomics, and, you know, how sort of realistic are the expectations that a lot of people are currently placing on this line of work?

MB: Right. So, I think the impact is several fold. There’s a lot of work that’s being done on identifying genes or mutations that would be associated to particular diseases, so I could have my genome sequenced now for a few thousand dollars, and I could know potentially that I have certain mutations in certain genes that maybe at this point would not really cause me any trouble, but might mean that I would be more subjected to heart disease later on, which might mean that, but if I exercise well or I eat well, then I can reduce my risk also. So this aspect of personal genomics where we can measure the real risks of having certain diseases, I think we’re not quite there yet, we’re getting there. And I think that’s going to be an important aspect. The choice of treatment for certain diseases should also depend on the genetic makeup of the person who is being treated. People realize more and more that certain people will have very severe side effects with a particular drug, whereas the drug works perfectly well for others, and if we could tell ahead of time and test the person who would be a candidate for that drug, and tell whether that person will have the bad side effects or not, that would be a great way to allow better treatments. So those are two of the things that I think in the long run my work, and the work of many researchers in the field could contribute to it, and it’s just two examples, but they are interesting.

MP: What would you say the timelines are for this, because certainly, when the human genome project was first completed, there was very widespread optimism about our ability to solve all diseases within the next decade or two, and now it’s been over a decade and we still haven’t solved any of them, well maybe that is not completely fair, but we have only solved a handful of fairly straightforward diseases so far.

MB: That’s very true. I think that the real need to get there is on the mathematical and computational scientists. So, one of the big challenges toward taking the understanding of genomics and bringing it to have real biomedical implications was the ability to read people’s DNA, just to tell what mutation they might have. Now that is done. You can sequence your DNA very easily and cheaply, but the challenge that we’re facing is interpreting this set of mutations that everybody bears, to tell which ones might be associated with what kind of consequences, and that’s a statistical/mathematical question.  And this is where things have moved very quickly in the past few years, and I don’t think it’s going to be solved in the next three or four years, but I think that in the next ten or fifteen years you will be able to get a pretty decent picture of the risks that you might be exposed to in terms of your genetic makeup.

MP: OK well that sounds very promising so I guess we’ll hopefully re-visit this conversation in ten years or so.

MB: Right; I may be in trouble!

MP: But I would certainly not hold you to this. Well another sort of speculative question, even more speculative than the previous one, that I wanted to ask you about, is this idea of being able to recreate ancestral species, like relatives of humans, or perhaps ancestors to humans, from their DNA sequence. In particular, as I am sure you know, there was a proposal from George Church who has basically managed to extract DNA sequence from Neanderthal fossil and is looking for a person who is willing to give birth to a Neanderthal baby, and so what are your thoughts about that, given that genomic reconstruction, ancestral genomic reconstruction, is a particular expertise of yours.

MB: Right, so I think having the ability to look at DNA sequences from Neanderthals, or from mammoths, or the ability like we are working on to infer what the ancestral DNA sequences were, I think this is very interesting and very informative about questions like, what is the function of the different regions of our own genome. And so, the question of using this information to learn more about how our genome works, I think that is extremely valuable. Now I don’t really see the point of trying to grow a living Neanderthal human or a living mammoth. I would not support this idea, mostly first for ethical reasons, but also for more very practical questions like why, what good could that do?

MP: I was going to say that the argument could be that potentially, we might be able to, not just understand their biology but also something about the way that they actually lived and the way that they actually did things, and we could perhaps conduct some experimental studies that would help us learn even more about, not just the biological functions, but also, you know, more physiological things and so on.

MB: Yeah I understand that. I certainly think the same ethical standards that are being applied to humans today should at least be applied to these new revived species, that I don’t know the gains are worth the risk or the obstacle.

MP: OK, so I think I definitely agree with that, in the sense that the ethics are slightly murky there as soon as you start to get into these ancient species, but I think we do need to hold ourselves to at least the same level of ethical standards as we do when we do research on our own species.  So then, with that being said, I guess I would just like to conclude with the following question, which is: suppose that you are talking to somebody, a young person who is interested in a career in the mathematical sciences, and potentially having some interest in biological applications, what kind of advice would you give to that person, and in particular, would they do best to focus on learning mathematics, or would they do best to focus on learning biology, or a bit of both. What advice would you give to somebody like that?

MB: Right, well, I think that although the area of computational biology or mathematical biology has both terms in it, biology and math, it’s important that people are really good at one of the two aspects and good enough in the other.

MP: I see.

MB: I would say becoming excellent in math and then learning the biology is certainly a good way to go, or becoming excellent in biology and learning enough math is alright too. Being too thin on both sides I think is a problem. I think, particularly if you want to become a cutting-edge researcher, I think you need to be excellent at one of the two.

MP: I see, so basically maybe just focus on the one that they are most interested in as their sort of main area of expertise, and then just develop enough of an understanding of the other one, would that be a fair assessment?

MB: That would be my advice. It’s tempting to want to learn about everything and physics could be on your list too, right, and statistics and chemistry, because all these things are important, and it’s important to have at least a minimal understanding of these things, but I think if you want to really push science, you have to be aware of these things and be excellent at one of them.

MP: OK well on that note I would like to thank you again for taking the time to speak with me and I think this has been a fantastic conversation. I have certainly learned a lot from it and gotten a lot of interesting ideas from discussing these with you. I am looking forward to checking back hopefully, not in ten years but definitely will be checking back to see where things are.

MB: Well it was a great pleasure to talk to you and it will be a great pleasure to talk again in ten years or tomorrow if you want.

MP: Absolutely, well thank you so much.

MB: It’s a great pleasure.

Mathematics: attitudes matter!

Dear readers,

It’s been a long time since my last post, over 3 months in fact! I’ve been dealing with a lot of personal challenges during that time, but I’m happy to report that they are now behind me and I’m getting back on my weekly posting schedule again!

Today, I’d like to share with you part of an excellent piece I found in a newsletter from Annie Murphy Paul, who has a blog about education that I recently subscribed to (and highly recommend), It reviews the latest research revealing links between parents’ and teachers’ attitudes towards mathematics and the students’ success in learning them. I especially liked this piece because it sheds some new light on possible reasons for the underrepresentation of women in mathematical professions.

Elizabeth Gunderson, a researcher at the University of Chicago, and her colleagues recently published an article in the journal Sex Roles that examined the “adult-to-child transmission” of attitudes about learning—in particular, how mothers’ unease with mathematics may be passed down to their daughters. Parents’ “own personal feelings about math are likely to influence the messages they convey about math to their children,” Gunderson notes—and kids will readily recognize if these feelings are negative. Becoming aware of our anxiety is the first step toward stopping such transmission in its tracks.

Previous studies have looked at how parents’ stereotypes (“boys are better at math, and girls are better at reading”) and expectations (for example, holding sons’ academic performance to a higher standard than daughters’) affect their children’s orientation toward learning. Gunderson takes a different tack, suggesting that parents may influence their offspring’s attitudes in two more subtle ways: through their own anxiety, and through their own belief that abilities are fixed and can’t be improved (expressed in commonly-heard comments like “I’ve never been good at science,” and “I can’t do math to save my life”).

Research shows that school-aged children are especially apt to emulate the attitudes and behaviors of the same-sex parent—a source of concern if we want to improve girls’ still-lagging performance in traditionally male-dominated fields like science and mathematics. If mom hates math, a young girl may reason, it’s O.K. for me to dislike it too.

Teachers aren’t immune to negative feelings about learning, either. In fact, studies show that undergraduates who study elementary education have the highest math anxiety of any college major. Instructors who are uncomfortable with mathematics feel less capable teaching the subject, research indicates, and are less motivated to try new and innovative teaching strategies. A study by cognitive psychologist Sian Beilock, published in the Proceedings of the National Academies of Science in 2010, demonstrates how teachers’ unease with math can influence the students in their classrooms.

Beilock and her coauthors evaluated 52 boys and 65 girls enrolled in first and second grade and taught by 17 different teachers. At the beginning of the school year, there was no connection between the students’ math ability and their teachers’ math anxiety. By the end of the year, however, a dismaying relationship had emerged: the more anxious teachers were about math, the more likely the girls in their classes were to endorse negative stereotypes about females’ math ability, and the more poorly these girls did on a test of math achievement.

Adults who want to avoid passing on pessimistic attitudes about learning can do more than simply watch their language (no more “I’m hopeless at math” when the dinner check arrives at the table). They can jump into the subject they once feared with both feet, using their children’s education as an opportunity to brush up on their own basic skills. Learn along with your kids, and you may find that math and science, or writing and spelling, are not so scary. And let kids know that it’s always possible to change and improve our abilities—you being a prime example.