NCWIT 2012 Summit - Plenary 1, Nora Newcombe

May 22, 2012

NORA NEWCOMBE: Okay, well thank you very much. It's really a pleasure to be here. I remember well that panel where I met Lucy. And, actually, there's a picture of me circulating on the internet [laughs] from that panel, which I gave a talk in Niagara Falls a few months ago and the person who introduced me selected that picture because she said it was so much better than the usual headshot because I looked very animated. [laughs] I guess I was pretty animated at that panel. So, let's see if I, yeah. So, I called my talk Spatial Skills and Success in STEM, I guess I didn't know that I was suppose to somehow fit a C into STEM, although I'm not sure where you wanna put the C, [laughs] Thinking About Gender Differences. Basically, what I want to do is to give you, perhaps at greater length than I could on Capitol Hill and also updated by seven years, a version of the kinds of things I say when people ask me, "Are there in fact some "cognitive barriers to women succeeding in science, "and math, and engineering, and computer science?" I think you get further by really going into the science here than you do by just sort of indignantly saying, "Of course not." Because then they can say, "Well, but what about this and what about that?" So the truth is a little bit more nuanced than just, "Oh, of course not." But the bottom line is of course not. [laughs] So the Spatial Intelligence and Learning Center that I'm PI of is funded by the National Science Foundation. And I heard in the overview that Lucy gave that 50% of NCWIT's money comes from NSF. So I guess we owe a big debt to the National Science Foundation here. It's ending year six of a 10-year grant. And that will be it, it's not the kind of thing that they're planning to continue. They want to sort of seed the field and see where it goes from there. But our purpose is to develop the science of thinking about spatial thinking and to actually use that to change educational practice. So we do think that, as you do, that STEM skills are really important in the global economy. We do think that spatial skills are relevant to them, and I'll tell you a little bit about all that before I really dive into gender. Lucy found this difficulty, too. So I want to first say why spatial thinking is a sort of component of intelligence. When people say intelligence in the singular, some other people say multiple intelligences, and there can be a lot of academic debates about that that I don't wanna go into. But a lot of people, even people who want to use intelligence in the singular, really acknowledge that there is a kind of thinking that you would want to label spatial. And I actually think there are two rather different kinds of thinking that you would want to label spatial, and they both have evolutionary roots. So it's important to see them in this kind of biological context. One kind of spatial thinking is the kind that supports navigation. How do we get around in the world? How do we avoid predators? We don't think much in terms of predators these days, but in evolutionary adaptation we had to think about that. How do we find our food? How do we find our home? And this is a problem that we share with any organism that moves around in the world. So ants have this same problem, and they actually are quite good at spatial navigation. Birds, they do incredible jobs of migrating and so forth. So in that context, homo sapiens has involved certain particular kinds of adaptation that are different from the adaptations of other species. For instance, I've often wished that I had the magnetic compass that a bird has. But humans, sad to say, don't have a magnetic compass in their nervous system. But we have our own sort of set of principles for doing this. So Barron's stressing though the cross-species comparison, because lots of species have to do this. But there's a different kind of spatial thinking that is involved with tool use. And tool use is really one of the corners that our species has on the evolutionary market. So I have a picture there of a chimp using a stick to fish out a termite. And, yes, they do do that. And that's kind of cultural and it seems to be culturally transmitted. But it's really very humble compared to what we do. They don't, for instance, fashion the stick to be a better termite piercer than it could be naturally. So the kinds of things that we do when we chip away and make an arrowhead and attach it to a shaft and figure out how to launch it, this is really unique. And I use the term advisedly in its proper meaning. It involves things that we do with our hands to objects. And different parts of the brain, by and large, are involved in doing this kind of spatial thinking that are involved navigation. Actually, people sometimes talk about gender differences in both these kinds of areas, but most of the data that I'm gonna be talking about focus on the, focuses, focus, yes, because data is plural, [laughs] on the tool use kind of thing. This continues to be true in modern life. It's not just true for those long ago ancestors. We still need to figure out where to go. Now we use symbolic means. We try to interpret these very complicated signs. In terms of tools, we're not using arrows that much anymore, but we do use, for instance, power Tools to put together furniture. And we educate our children using spatial tools. If you're learning about the various layers in the canopy in the jungle, you make a diagram, and if you look at it upside down, you may or may not be able to read the label as that girl is doing. But these are the kinds of things the are supporting our learning about the world. And our use not only of tools for navigation, tools that are tools, but also tools to learn about other things such as layers in the jungle canopy. Not only is it important and sort of general everyday life, but spatial skills are really very relevant to the STEM disciplines. Here, what I have up is just a panoply of diagrams and figures from various of the sciences, whether from biology or from physics or from engineering and technology, or from geoscience, or from geography. We layout a lot of information in spatial ways. And we use this kind of thinking to learn from others. And also it's relevant in discovery. So when Watson and Crick came up with the structure of the DNA molecule, they were working with flat X-ray diffraction patterns that they got from a woman, that they got from Rosalind Franklin, but they were trying to imagine how those flat patterns gave them information about a three-dimensional structure. This is the kind of thing that people commonly cite when they say that spatial thinking is important. But psychologists always want data. We actually do have some really impressive data, I think, about how spatial skills predict success in the STEM disciplines. These data are taken from a longitudinal study that was launched in the 1950s when hundreds of thousands of American high school students took a wide array of cognitive tests. And they've been followed over time. I mean, most of them are now retired or retiring. We know quite a lot about what disciplines they went into and what their careers look like. Now, because there are lots of kinds of intelligence, and verbal and mathematical intelligence are distinct kinds of intelligence that obviously contribute to success, there was a statistical analysis done in which the team removed the contribution of verbal and mathematical intelligence in order to leave behind the extra contribution that spatial intelligence made in predicting whether or not these high school students would go into various disciplines. And what you can see, whoops, when I don't want that to happen, it does happen. What you can see in the bottom two bars is that both for engineering, so in terms of SWI, and for math and computer science which got lumped together, you were much more likely to go into those disciplines if you were high in spatial skill. That was also true for physical science. And then it really diminishes as you go out towards biological science and medicine. Distressingly enough, education goes to the left. So the people who are teaching K-12 or, a lot of those are probably frankly carried by K-8, are people who are not as confident in spatial skill. And that may be part of the puzzle of what kinds of support they're gonna need in introducing this kind of thinking to our children early. So the kinds of spatial tests that were given to all these teenagers in the 1950s were the following. There were four of them. The top one, which is kind of small, is a paper folding test where you have to imagine a piece of paper being folded along certain dotted lines, and that it can form one of those three-dimensional pictures that you see on the right. And you have to check the box that shows which one is the correct answer. There's mental rotation. The second row shows a flat, two-dimensional mental rotation task where you take the design on the left and you have to imagine which of the designs on the right could be made by a rotation in the plane. You're not allowed to flip it. There's also three-dimensional mental rotation tests, and I'll show you one of those in a minute. Then there's a mechanical reasoning test where you have to think about gears, and if one moves, how will the other one move? And then there's an analogies task where if one shape is to the other shape, is the other shape is to the other shape, then the third shape is to which of the bottom ones? So that's inference, but it's about spatial material. Now, interestingly, two of these tests show sex differences. Two of the four. So when anyone tells you, "Oh, there's spatial differences." One of the first things to say is, well, sometimes yes and sometimes no. So it depends on the test. And this is actually one of the scientific puzzles here. Why do some of the test show sex differences and others don't? There aren't any sex differences either on the bottom test, the inference task, or on the top task, the paper folding task. On the middle two, the mental rotation and the mechanical reasoning, there are. And they're quite large, too. Mechanical reasoning can be written off to, you know, there's a lot of content here, maybe boys have spent a lot more time with actual gears, you know, taking apart things and putting them back together. It's a little bit more mysterious when it comes to mental rotation. I wanted to lay that out. And then I wanted to tell you about some research that was actually done here at the University of Chicago that tells us that there are similar findings even back in early childhood. That is, that the spatial skill of young children predicts their mathematical skill, not decades later as was true in the Project Talent study which is the one I just told you about, but several years later. So the children took that test that you can see on the left where you imagine the two pieces that are at the top rotating and moving together. And then you have to select which of those four designs could be made by moving those four pieces together. And then, there. So the spatial skill measured at age five does predict the math task, it was something called approximate symbolic calculation. You don't really have to successfully do the calculation, you have to be able to estimate it three years later at age eight. Now, that relationship is, actually, mediated through knowledge of the number line. So lest you think that this is completely occult and how would this work, let me remind you that one of the most basic mathematical tools we have is the idea that numbers are arrayed on a line, that addition moves things rightwards, that subtraction moves things leftwards. And that's a spatial skill as well as a numerical skill. So being able to think that way is the mediating factor that allows you to predict the approximate symbolic calculation task. In fact, when you do the proper statistic, if I could ever get it up, there it is, the direct line from spatial skill to approximate symbolic calculation drops out. So it's all mediated through the number line. In terms of gender differences, on the same task, the one I just showed you, we once thought that they didn't begin until adolescence. And if they didn't begin until adolescence, some people thought, "Oh, that shows it's hormonal." Other people said, "Oh, no, that's because of intensifying "sexual pressures when you enter adolescence "and you start to date." But for better or for worse, I mean, those were two explanations, but neither of them can possibly be true because it turns out that gender differences, actually, are evident early. So I show here the full distribution, not just the average, because I do want to emphasize that any particular girl may well be better than any particular boy. So there's a great deal of overlap in these distributions. It's not the case that you can make predictions about any individual person. And that's a very important fact to keep in mind. Nevertheless, on the whole, it is true that the boys tend to score higher and the girls are overrepresented at some of the lower scores. And we also have found this in second grade, second and third garde on another mental rotation test. And this is, actually, pooled with a map reading task, but the results were identical for both of them. Now, one of the things that this graph shows you is that there are sex differences in the high SES group, socioeconomic status, and in the middle group, although they're a little smaller but they are still significant. There, actually, are no sex differences in the lower SES group. So it's interesting that that's true. We don't know quite why that is. But it's a fact that does lead one to think that some of this may be environmental. But I also want to emphasize that the sad fact here is that the low SES kids, those data are from the spring of third grade. Whereas, for the high and middle kids, those data are from the fall of second grade. So what this suggests is that it takes a year to a year and a half before lower status kids catch up with their higher status counterparts. So there are social status differences in this skill to worry about as well as gender differences in terms of social equity. Now, going along with the theme that, yes, there are some sex differences, but, no, there is not a sex difference in everything, another area of spatial thinking where people have found very robust substantial sex differences is horizontality and vertically. Now, you may realize that the middle photo is photoshopped. [laughs] You couldn't actually take that photograph. But not everyone really seems to realize that that's true, that the beer stein on the left, which actually has a weird sort of bottom is why it can be photographed stably without a hand holding it. But if you ask even adult men and women to draw a horizontal water line in a glass, the women, it's not that they totally don't know where it goes, but they draw it more than five or more than 10 degrees off. So they're unable to disregard the surrounding sort of influence of the glass to sort of orient themselves to the true sort of gravitationally defined, this is vertical, this is horizontal. And this is also true, my colleagues and I have found recently, for perception of slope under your feet. So when we put you in a room whose floor is tilted by five degrees, which is a fairly appreciable slope. It's about what a wheelchair ramp is sloped at. And there aren't any other cues for telling your way around the room. And we spin you around blindfolded and disorient you. Women are less able than men to use the feeling of the ground under their feet to reorient, to tell, basically, what's uphill and what's downhill tells me where something is hidden in the room because we had hidden something in the room that we had them look for. So it's a real sex difference. But there's another interesting absence of a sex difference to something that I think is very, very technologically relevant, namely cross-sectioning. This is very important in the geosciences which are very relevant to things like finding oil, that people in Washington, I think, care about. And both with a test that we devised to give to younger kids which you see on the left and a more complicated task which you see on the right, which can be given to undergraduates, people are not always good at this. In fact, one of the problems is that people can quite often be quite bad at them. But there are no sex differences. So men and women are the same. So there's a very interesting picture here. From a scientific point of view, it's really pretty intriguing. Don't ask me why we have this picture, it's really a work in progress. And I will say a little bit more about this. But I think people who either say there aren't any sex differences, or they're everywhere, they're both wrong. One of the most important things that I wanna get across in this talk is that when Lucy used the word innate, innate often, to a lot of people, means fixed. Now, it doesn't actually technically mean fixed. You can have something present at birth and you can change it. But a lot of people think innate means fixed. And if you think that spatial skills are fixed, that is not true. Spatial skills can be improved. I did a meta-analysis which is a statistical literature review that aggregates findings a long time ago that showed this. Subsequent to that, I've done a number of training studies that show how much, in this case, mental rotation is what I'm graphing there, how much it can be improved. Now, the top line is in blue, and that's men. And they can improve a lot. I mean, they start out high, and that doesn't mean they're at ceiling performance. If you give them continued training, they still get better. So this is not a skill that we're maximizing in our society. The solid pink line is for women, who start high, we selected some women who start high. And then the dotted pink line is for women who start low. The women who start high also improve. And the women who start low also improve. Now there is a difference in the rate. So the people who start high, the men and women, improve more quickly at first and then a little less quickly later. Whereas, the women and men who start low... the women, we, actually, for some reasons that I won't go into couldn't get men who started low. Damn. They improved slowly at first and then they get better much more quickly. Oh, you know, the thing is that you have to be here, not where I thought, I think. Yeah, okay. So the big deal here is that if you were a woman who have low spatial skill, why did you come back and do this? You're bad at it, you are bad at it a week later, you're bad at it two weeks later. Well, the answer is we paid you to do this. [laughs] That's what happened in the study. But the message for, you know, situations where we can't necessarily pay people to try to get better is that the motivation is very important. And the faith that indeed they'll reach the sweet spot of their learning curve and that they will start to get better is very important. Now, we've recently done another meta-analysis, which is now in press, which continues to show that overall there are these big effects. And which shows that they are present for mental rotation, which is the second bar from the left, and also for the horizontality, verticality which is the bar on the right. In fact, the bar on the right is the highest bar there is. So even the two areas that I told you about that have sex differences could definitely be improved. Now, in terms of possible effects, I've illustrated this for engineering, but, as you saw, the effect size for spatial skill fostering going into computer science is just about as large as for engineering. So this is a hypothetical set of data that put together two data sets. First of all, we took from the longitudinal project talent study the thresholds for where we set, "You have enough spatial skill, "you could become an engineer." It doesn't mean you have to become an engineer, but were you to want to, you would probably be gifted enough at this kind of thinking that you would succeed. Then we right shifted that distribution by as much as we know from the meta-analysis. We can actually right shift it. So how much could be actually improved. Therefore, at the light area under the curve is the proportion of the population eligible to become engineers if we don't foster these skills. Whereas, the dark gray area is the proportion that we could add to the potential pool of people who could go into either engineering or computer science were we to foster these spatial skills K-12. And since we, actually, do need more engineers and we do need more computer scientists as Lucy mentioned, I think this is of national importance. And we also know this experimentally. Those of you who are fuzzy about experimental design are probably only 2% of the audience, but just for those of you who are, you may have realized that I've presented correlational data. And even though it's longitudinal data and it's controlled for a bunch of stuff, it's still not experimental data. We're beginning to accumulate findings which do the very, very, very toughest test, which is to take people, randomly assign them to being trained in spatial thinking or to be doing something else, and see if we can get them by virtue of having the spatial training to have hard endpoints like better grades in chemistry, or geoscience, or physics, or calculus. And although there aren't as many studies as one would like, it look like that is actually true. And there are also similar such experimental findings in children. But you'll note that both of these are under review. So this is a new enterprise and one we wanna press on just to make sure of our facts, but we think it's gonna turn out to be true. So now we want to get to the issue of whether sex differences are biological. And I have a photo of Larry Summers whose become very convenient as a whipping boy. [laughs] Still on the slides seven years later. Because he did suggest, inspired, I think, by Steve Pinker who was at Harvard and fed him a reading list, that spatial skills were biological. And for him that was correlated with immutability, which I just argued was false, and was supported, he thought, by hormonal and neurological evidence, and which is supposedly predicted by evolutionary theory. So I think I just disputed the first of those points. And I want to tell you a little bit about the second and then more about what I think are some incredible contradictions in the third point. So just briefly in terms of the second point, I'm not even gonna talk about the neurological data because they're very weak. I really don't think there's a shred of clear evidence for neurological differences that would lead to these kinds of differences. There are some interesting hormonal findings, and I would not want to sweep them under the carpet. But they're very, very messy. You see them, then you done see them, it depends on the methodology. But more important, and this is actually in a way also anticipating what I'm gonna say about evolution, they're odd from the point of view of, if you're a woman, it seems that spatial skill is stronger at times, for instance in the menstrual cycle, when you have higher levels of testosterone or testoronizing type male hormones. I don't wanna get into technicalities of hormones. Basically, when you're menstruating, some of these skills tend to be a little higher than when you're ovulating. On the other side, among men, it's actually better to have less testosterone. I mean, I don't know if this is like a clue for people hiring male computer scientists that, you know, [audience laughs] heavy beards, forget it. [laughs] The point I'm trying to make is, it's very odd from the point of you of, if you're trying to use this to predict sex differences in an evolutionary framework, why would it go up with testosterone within the range of testosterone that you see in women and down with testosterone within the range that you see in men? You'd have to explain that in order to have a really consistent picture. And it isn't there. So the explanations really get to be very convoluted. So the remainder of the talk, I mainly wanna talk about an evolutionary approach and the idea that that somehow implies hardwired sex differences. And I am not a person who denies that there are some real sex differences. I mean, clearly human females lactate and human males don't lactate. And there can be indirect effects of that. So in some cases, anthropologically and cross-culturally, that has led to some sex type divisions of labor that are less in our society because we have fewer kids and we don't always nurse them, and so on and so forth. But in subsistence societies, it can be correlated with certain kinds of sex type divisions of labor. But what is true for spatial ability? There are, basically, two families of evolutionary theories that are pretty popular. And you probably already thought of them. One is that Man the Hunter. Well, aren't men higher at these skills because they have to hunt and they have to aim and they have to track, and they have to make tools in order to have something to aim with. So there's several things that are kind of correlated with hunting. And there's also a different story which is actually a little bit more consistent with sociobiology. Although Man the Hunter, well, they sort of interact. But sociobiologists mainly care about how many genes you leave to succeeding generations. So how many kids do you have. The idea is, spatial ability must somehow help men to leave more genes in the succeeding generations. You must somehow have more children if you have higher spatial ability. So this is a little weird-sounding. But let me tell you about the research that leads people to think that it's actually true. There are two kinds of a certain rodent, that you may or may not heard of called the vole. That's a picture of a vole. It looks a little like a rat, but it doesn't have a tail. So it's cuter because rat tail [audience laughs] is really obnoxious. And there's two kinds of voles, the prairie vole and the meadow vole. The prairie vole is pair bonded. They just find a mate and they live together and they make kids, and that's great. And they have equal spatial ability, which is tested, by the way, using a navigation test. Obviously, voles don't use tools. So you put them in a maze and see how they can get around, and they're equal. The meadow vole is polygynous, which means the female vole stakes out a territory during the mating season very separate from the territory stake out by some other female vole. And the male voles wander around trying to mate with as many of the female voles, within a limited period of time while they're ovulating, as they possibly can. So this is actually a navigational challenge because you have to basically calculate [laughs] the least distance to get you to as many female voles, within a limited time, as possible. And they do have higher spatial ability during the mating season only. And they actually even have a larger hippocampus, which is the part of the brain that supports navigation, at the mating season only. So it's actually a really interesting set of data if you care about about voles, but [audience laughs] the issue is, really, you know, to what degree this can be generalized. So let me first take Man the Hunter. And the idea of aiming, well, we saw that mental rotation is one of the biggest sex differences there is. Another one is this horizontality, verticality thing. Another one is slope perception. But aiming has nothing to do with that. In fact, people have actually tried to correlate aiming with mental rotation, and they aren't correlated. So you could just as easily say that the things where women are equal to men are related to aiming. But aiming, actually, it's basically a sensory motor kind of capacity, which has nothing much, as far as I can tell, to do with spatial skill. There's also the idea that it has to do with tracking where you go around in the world and you figure out, follow the animals, and you end up maybe quite a ways away from where you started. And then you also have to figure out how to get home. But the fundamental assumption here is that gathering, which is what women do in these kinds of hunter-gatherer societies, doesn't involve having to get around. So the sort of mental picture that people have of these groups, I think, is that women kind of roll out of the huts in the morning and the blackberries or whatever are right there and they just kinda gather them. [laughs] Well, the blackberries could be a long way off, too. In fact, studies of primeval hunter-gatherer societies today suggest you may have to wander quite a lot to be able to gather. Furthermore, we don't know that our ancestors hunted in the way that's envisioned this way. And there's actually quite a lot of evidence that they hunted instead by, you know, dig a hole, wait for an animal to fall into it. [laughs] Go to the stream, wait for them to drink then shoot them. These are smart things that homo sapiens is good at, but they don't involve spatial skill. And for toolmaking, all I have to say is that women in hunter-gatherer societies make tools, too. They weave, they make baskets, they produce pottery. So there is some sex typing in what you produce, but I can't see that making a pot is less spatial than making an arrowhead. So I kind of don't get what's going on here. In terms of the hypothesis, the man who gets around, the vole data are fine. But, as I said, there's the issue of generalization. So the big deal here, I think, is that humans live in social groups. So sidestepping the issue, up to what extent we are or are not pair bonded, to which the answer actually seems to be, you know, it's an enduring tragedy that we're kinda, sort of pair bonded but sort of imperfectly. So there's a lot of extra mate pairings. [audience laughs] 5% of children are estimated not to be fathered by their official father. But whatever the constraint is on fathering children out of wedlock, which is a sociobiologically good thing to do, it's not finding the woman. If you live in a village, finding her is not the problem that [mumbles]. Getting her to agree could be a problem, or evading the husband who's jealous could be a problem. But not finding her. So I don't think that that makes sense. In fact, as I said, we're probably only mildly pair bonded. So unlike the gorilla, we don't have a real polygynous mating system. And there's also some sociobiologists who talk about, well, there's female choice. So we might select someone to father our children if they're better hunters or if they can beat out other men, and, you know, bring home the mammoth or whatever the primeval equivalent of bringing home the bacon was. But that just gets back to me, back to my story about why Man the Hunter doesn't work. So, basically, there's no real reason to suppose that spatial ability is confined to men for evolutionary reasons. Most things that are sex differentiated in this way. So the lion grows a mane, the male lion grows a mane. The peacock grows a tail. Having a mane is a drag if you live in Africa. There's a cost. Having a tail is a real drag literally if you're fleeing from an enemy. So, typically, these kinds of displays that males have also have costs. It's an interesting fact. But since there's no obvious cost to spatial skill, this also doesn't fit the evolutionary picture. I thing I'm gonna skip this because it gets a little bit too much into the sociobiology stuff and just conclude that if sex differences in mental rotation have a biological cause, their evolutionary function is either not currently understood or maybe it's nonexistent. I'm not ruling it out, but I think the evolutionary people have a lot to do to shore up their story before it would really be believable. But what I wanna leave you with is the following. The idea that the causation of sex differences. Well, I agree that it's scientifically interesting. And, on occasion, I have done research relevant to it. It's not the socially important question. The socially important question is not whether this difference is in any way innate, or in any way biological, or in any way hormonal. The important fact is that it is not a fixed ability, that there is potential to benefit from experience, that there can be vast improvement for both the sexes. So the important fact is that we can bring both men and women more into that band that I showed you before where they are perfectly able to choose a technological discipline if they choose. Now, one of the barriers to that is anxiety. Now, I just wanted to mention this because I showed you that leftward bar for the elementary school teachers. Elementary school teachers have a lot of spatial anxiety, but one of the things we have done in my center is to show that it can be reduced. So this is the kind of area where cognitive psychology meets social psychology. And I know last year you had a plenary address from Josh Harrison. So these two fields are both part of the jigsaw puzzle. So overall, spatial skills are important to STEM learning, they can be improved, the causes of sex differences are not well understood, and more research is needed but it's not, I don't think relevant to the socially important fact that improving spatial skills could help to reduce gender and SES inequities in the STEM pipeline. And that's not to say that other factors aren't relevant as well, such as work-family balance. I think this is part of the picture. Thank you. [audience applauds] I left a little less time than I would have liked, but, yeah.

AUDIENCE MEMBER: I just got the microphone, hi.

NORA NEWCOMBE: Oh, there you are, hi.

AUDIENCE MEMBER: Right here. Thank you for a really fascinating sharing of your research. So it's about this same time, 1989, when one of your first analysis, I think, were on the slides about gender differences and spatial cognition. One of my colleagues at MIT Media Lab ran a study on gender differences and spatial cognition and correlated it to elementary school boys and girls, low income, African Americans and Hispanic kids in their ability to do 3D programming. And I'm bringing this in because I think it really relates to what we are about here. And so what she showed is that indeed where some things that had no differences between boys and girls, but on some test did show a difference. But she also demonstrated, and I know, it was 30 years ago. I don't really remember the exact details, but she did demonstrate that learning how to program, how to code, it was Logo programming language but also 3D space that she created to program in 3D, actually was a form of training to eliminate the gender differences. And so while we see in your slides this pretty fascinating correlation between people who have strong spatial cognition with their attraction to engineering and computer science, I think that research from back then also demonstrated, her name is Judy Saf-ter. Her research at MIT, her PhD thesis showed that if we will get kids very early on to program, especially 3D programming, we can actually increase not just their engagement in computing and programming but also their spatial cognition that will eventually make them better programmers, which is kind of like what we're saying. Do you know anything about that research? Did anybody carry on that?

NORA NEWCOMBE: I don't know about that particular study, and if you could remember the name or send me the reference, I'd love to get it. Part of the work of our center is in fact to come up with a variety of means to improve spatial skills both in little kids and in undergraduates and everything pretty much in between. And we have a lot of good tricks, and, you know, if I had yet another talk to give, I could talk about, you know, just some homely things like play with blocks and do jigsaw puzzles and so on and so forth. But the idea that you get kids to do 3D programming, that's fascinating, that's great.

AUDIENCE MEMBER: This is the area that I'm interested in also is the, you're defining spatial skill as relating to the actual envisioning of objects or molecules or mechanical drawings that represent reality. Is that correct?

NORA NEWCOMBE: Mmm hmm.

AUDIENCE MEMBER: So aside from the use of the number line, you're not using visual aids to try to model a mental process like programming. So 3D programming aside, I'm interested in the correlation between spatial skills as you define them and the ability to produce programs, computer programs, knowing that frequently the representation of an algorithm is a visual representation. And in my experience of teaching computer science, generally I find I'm a tremendous visual learner. Many of my best students are visual learners. And whether it's a URL or a flowchart or a data transfer diagram or whatever, a visual representation of an algorithm generally shows the mastery of that algorithm, so the correlation between spatial skills and that ability.

NORA NEWCOMBE: Right. Now that I understand what you're saying, I think I totally agree. One of the slides that I sometimes show and didn't show in this presentation actually makes the point that the spatialization of non-spatial information is a very important learning tool. So every time we produce any kind of graph or any kind of flowchart, we can be putting down inherently non-spatial information such as, you know, economic information, how much people earn over time, or this kind of thing, and spatializing it. And it's often much easier to grasp what's happening from those kinds of representations than it is from words. I'm actually a great believer in those kinds of data displays. I don't know much about programming or teaching programming, but I do think that the fact that, in the longitudinal study there was a prediction to computer science, tells us there must be some kind of link of that kind because why else are you getting it? So I would like to know more about the linkage you see in teaching computer science, but yes.

AUDIENCE MEMBER: I enjoyed your talk very much. I actually have a quick comment and a question to follow up. I'm Catherine Di-de-an, and I was actually at the talk that Larry Summers gave. And he had actually three points. One was this is your mathematical ability, and that the fact that young men tested at much higher level, at the higher levels than young women did. But his other two arguments were that women did not want high power jobs because they did not want to work more than 40 hours a week and that there were social and cultural differences that excluded them. But what's interesting is actually if you look at the data that was presented at that meeting in January 14th, that there, actually, even the women who test high in those levels of mathematical or spatial, roughly less than a third of them go on to computer science and engineering. So actually having that spatial ability and testing well doesn't necessarily predict you going into the field. And there actually are ethnic and racial differences among certain groups. Like Asians, you won't see the differences within gender than you do in Caucasians. And sometimes you don't see them in other ethnic and racial group. So I think this is a very interesting topic, but I think we need to acknowledge that there may be some cultural and social expectations that can influence people even who have that interest that go beyond that. And I just wanted your response to that.

NORA NEWCOMBE: Well, two points. One, just about the Project Talent data. It actually does work in Project Talent when you separate the men from the women. So you find the same kind of threshold effects and the same kind of prediction that you find for women as for men. Now, the more important point that you make, I think, that actually Larry Summers was also making in a somewhat ham-handed fashion, is something that I eluded to at the end when I mentioned work-family conflict. I do think that the threshold that's posed by spatial skills is really just a kind of threshold or gatekeeper kind of effect. There are other gatekeeper effects that I did not mention, one of which I really believe in although I haven't personally examined the data, but I think there are data there that women tend to see these disciplines as not helpful and not social. And it's very important, I think, and you guys probably know more about this than I do, to contextual, you know, if your bridge falls down, people die. This is like socially bad for people or, I'm not quite sure how you do it, but. So there are other gatekeeping factors. In other words, you don't wanna say that any one factor is the only factor. And, recently, Steve Ceci and Wendy Williams at Cornell have been making the argument that the work-family conflicts are really the key. And I don't think that's correct. I think they're very important, I just don't think they're the only thing that's going on.

AUDIENCE MEMBER: My first impression is, first of all, why is mental rotation important? And when I think about it, I wonder if there's data and whether or not people who do well on mental rotation, or don't do well on mental rotation, do well when they're dealing with the kinesthetic problem. So when they have a 3D thing in their hand, is there data that shows that they do much better with spatial knowledge when they're working with the model that's in 3D rather than doing something mentally? And it even made me think of even auditory spatial awareness, thinking of the distance between musical notes or something like that and how those types of things might be different if they were looked at across genders.

NORA NEWCOMBE: Right, right. I don't know about the auditory part, but people in general do much better at these tests if they do have kinesthetic information. Interestingly, that's not just because when you turn something, oh, you know, the answer pops out at you. You can give something to somebody to hold in their hand and it's under the table. So you can't actually see it. But the fact that you are turning it sort of drives part of the motor cortex of the brain to help you to mentally turn it and to get the correct answer. In fact, when people do these mental rotation problems, in echo MRI setup so we know which bits of their brain are lighting up. You often, although not always, get motor cortex involvement. In fact, the kinesthetic component of doing mental rotation or any other of these mental transformations is actually very important. But the first part of your question, which is why is it important at all? I didn't go through the many, many correlational studies, I just looked at the Project Talent one because it's the most impressive because it's really longterm longitudinal. But there's a lot of correlational studies that do suggest that it's correlated with success not only in the STEM disciplines but also, you know, things we might care about in, you know, a different kind of way like being a good dentist or being a good surgeon, or these kinds of things. So yeah, it is correlated.

AUDIENCE MEMBER: Fantastic talk, thank you for it. You didn't spend much time talking about transfer. As I understand spatial abilities, for a long time we thought you could practice, let's say Tetris, it would improve your score on Tetris, but it wouldn't transfer to any other spatial. What was the breakthrough on that and what have we learned about creating transfer from one domain to another from your work?

NORA NEWCOMBE: Well, the breakthrough was... I'm not sure the conclusion that you couldn't get transfer was ever correct. [laughs] In other words, I think you always could get transfer, there were just some studies that were widely cited that failed to find transfer. In those studies, there are actually reasons for that. In some cases, they didn't train very much, so it just wasn't enough training. An interesting reason, which you'll probably appreciate more than most other people in the audience, has to do with the nature of your control group. So it has to do with what your quote, unquote non-spatial activity is. And it also has to do with how many spatial tests you give as pretest and your post test. Because taking the test itself is a form of training. So it's hard to get that kind of transfer if what you have to beat is a control group that is also getting training through taking the test. And we actually rigorously show this in our meta-analysis that's in press. So I think that's an important methodological point. Retesting effects in this area are actually enormous. A psychologist called Tim Salthouse has compared, psychologists are always worried about retesting effects, and they're bigger in the spatial domain than in any other domain. So you have to worry about it more there. Okay, well, thank you very much. [audience applauds]