It’s a staggering amount of data to work through, and requires sophisticated computational knowledge, Blair explains. “We code up software programs to pull out the things we think are important, and then we analyze that stuff. We go through several stages. We get the replay files (which are compressed) into a form that we can deal with. And then we put all that data in big databases and run other software programs to cull the databases for the information we’re interested in.” Through IRMACS, the lab is able to handle this kind of complex computational work using WestGrid (computing clusters that use thousands of computers hooked together to run parallel operations).
Cognitive Science, Psychology
Faculty Profile: Dr. Mark Blair, Cognitive Science
When cognitive scientist, Mark Blair, was on the job market back in 2005, he explains that among a few different job offers, SFU stood out as an institution that balanced support for research and teaching. “In large part, students are why we are all here. I chose SFU partly because it was a school that was supportive of research and research opportunities but it also emphasized teaching and I’m interested in how we open up opportunities for undergraduates.”
Blair runs SFU’s Cognitive Science Lab, and a visit to the Lab webpage shows how he facilitates collaboration and academic community amidst SFU students in Cognitive Science, Psychology, Biology and Biochemistry and Health Science. Alongside detailed descriptions of Skill Craft (the lab’s project that uses the videogame StarCraft2 to analyze cognitive ability and skill acquisition) are detailed bios of both graduate and undergraduate students in “The Team” section. The numerous co-authored publications listed also demonstrate a strong focus on collaboration and the lab’s findings regarding age and cognitive ability, led by Blair’s graduate student, Joe Thompson, hit mainstream media in Spring 2014.
Blair completed his MA and PhD in Cognitive Psychology at Arizona State University. Of his own undergraduate academic career, he explains that he cobbled together a Cognitive Science education at University of Maryland where he received both a Bachelor of Arts in Philosophy and a Bachelor of Science in Psychology. He said he’s always had a “philosophical bend” in approaching questions and research with that careful contemplation philosophers are known for, but that he equally values the way science tends to approach questions with an empirical intention of answering them. “There’s a satisfying solving of the mystery when you’re taking a scientific perspective that doesn’t necessarily happen when you’re taking a philosophical perspective.”
The best thing about Cognitive Science, according to Blair, is that the discipline tends to combine these approaches and that’s incredibly valuable for a student. “Philosophers are really good thinkers, they’re careful to define their terms,” he says, “I think that’s what we really need in science. It’s training that helps you think more clearly and express complicated ideas in written language—a skill that’s beneficial for those of us doing scientific research. Blair says Cognitive Science undergraduates are exposed to how “philosophers think about a problem, writing papers and thinking about things people wrote a hundred years ago” while simultaneously, when taking psychology courses, they are asked to “think hard about recent data.” Cognitive Science students are required to take courses in Philosophy, Psychology, Linguistics, and Computing Science and this structural interdisciplinarity, Blair says, gives students “great breadth” and helps them to “see the big picture.”
The Cognitive Science Lab’s research is shifting the “big picture” for researchers interested in human skill acquisition and skill development. Blair says SkillCraft’s use of StarCraft2 as a case study is providing datasets that haven’t ever been available to researchers. The lab’s interest in StarCraft2, he says, is similar to the biologist’s interest in studying the fruit fly: “It is representative of other kinds of things. And it’s really straightforward. There are many methodological advantages.” Previous studies of expertise, Blair explains, have only really gone in two directions: comparing non-experts with experts or bringing in participants and teaching them a simple skill. There are problems with both approaches because, as Blair says, in the first you are “seeing the beginning point and endpoint and making a gigantic assumption about everything that happens in the middle” and the second is “studying a really detailed record of a fairly simple skill.” This is not to mention the pragmatic challenge of getting participants into a lab for a big enough sample.
As a solution, studying StarCraft2 is ideal. Because it is an online game, players do not have to be physically present to participate, which works logistically. It also involves fairly complex strategizing and the computer mediated environment provides, as Blair explains “detailed records of literally every action that a player takes when they play. And we also have whole replay sets from various people, and all the replays they’ve ever had. We have every step they’ve taken along the way.” Access to this kind of data is unprecedented. “In terms of skill learning we’ve never had a complex skill with the ability to track every single step along the path to acquiring it,” Blair says. Access to computer-recorded, digitized archives of players’ skill development means the project provides the opportunity to study cognitive processes in intricate detail. With thousands of people across varying levels of expertise, Blair says they have information for approximately 240,000 individual games.
Blair says this type of research is the future of cognitive science and the applicability of such kinds of studies are immense. In modern medical surgery, for example, some surgery is done with the aid of a computer robotic arm and doctors are sitting at something akin to a video game console during the procedure. As Blair notes, there is evidence to support that surgeons who have played video games are better than those who haven’t: “they make fewer mistakes; perform surgical techniques faster than surgeons with no videogame experience.” The impact of Blair’s work in the Cognitive Science Lab takes it one step further: “Imagine that you create a detailed record of every surgery that you do. Now you can spot potential problems, you can compare across surgeons and look at more global problems in the training. You can begin to say, ‘you know what, the variability of this surgeon’s measures is scary—maybe it’s time for them to stop.’”
Betty Leung, a member of the lab team who has since moved onto graduate studies in Linguistics at McGill, took the gaming data in yet another direction, analyzing the linguistics of the game (StarCraft2 records when people say things to each other in the game). Leung was interested in the linguistic question of how you can understand whether the chat is positive or negative and whether people are being nice to each other. Blair explains that, logically, within 240,000 games, one cannot go through the chat records by hand; instead, the lab used a linguistics model that had been built for analysing movie reviews, and applied it to video game chat. As Blair notes, this approach could have a significant impact on online communities struggling with how to deal with bullying and negative behavior of users. The lab is building models of “chat parsing” that could be useful in monitoring these communities: “Imagine having an automated way of saying, ok, this person is being a jerk, a lot. They can then be banned or reprimanded. […] Or if someone is really nice all the time we can give them bonus points in the context of the game.”
With so much data to process in the SkillCraft project, Blair says it’s difficult to predict what his next research project will be. As someone who likes sketching and drawing, though, he says he would like to develop a project on the cognitive science of drawing, using eye-tracking to study the processes involved with learning to draw. “People who can draw, it’s almost like magic. People who don’t draw, their drawings might look like they’re the drawings of a ten-year old. But when you watch someone experienced draw, they make what looks like ten marks and it looks like a total person.” Tablet and graphic arts technology is rapidly developing and—as with StarCraft2—the tools we have are able to track and record detailed movements, empowering scholars like Blair and the team he is leading at SFU’s Cognitive Science Lab to study big data in new ways.