Worried she won't ever be able to commit to one field of study, Dale Markowitz decides to go all in on a neuroscience project.
Dale Markowitz is an engineer and data scientist at OkCupid, where she spends endless hours contemplating the mechanics of romance and attraction. She graduated from Princeton University, where she bounced from physics to math to neuroscience before landing on a major in Computer Science. When she's not bugging people for stories about their online love lives, she can be found pondering math riddles or blogging at medium.com/@unquarked.
This story originally aired on April 7. 2017.
Story Transcript
It was the summer before my senior year at Princeton, and I was spending it in the computer systems research lab, which was on the third floor of the computer science building. It was this windowless, peopleless room. I was spending it studying something called HTTP network, caching proxies, which weren't approximately as fascinating as they sound.
I was so bored, I could barely get myself to go into the lab in the morning. I don't think I got anything done that summer, but the whole reason that I had committed myself to that was because in the fall, I wanted to apply to PHD programs in computer science, so I needed something to put on my application.
The fact that I was bored doing this really gave me pause about committing six years of my life to it, but at the same time, I feel like I had already flip-flopped what I wanted to do so many times.
I came into the university determined to be a theoretical physicist, then a year later, I decided, you know what, my true calling is math. Then a semester after that, I switched to computer science, and it wasn't just a matter of academic self-discovery, because I'd always had a sort of interest ADD.
Before, when I was younger, I'd done every extra-curricular activity you could think, from ice skating to ceramics, to acting and improv and guitar, piano, violin, bass guitar, saxophone. I don't actually retain any of these skills to this day so I'm not sure if the breath was worth it, and this was actually in pretty sharp contrast to my dad who in utero, knew that he wanted to program computers for a living; for a whole life.
He would sometimes say, "Dale, I think that you're sort of a mile wide and an inch deep." I was pretty insecure about this because I thought maybe I should just focus on something. I didn't want to give up research quite yet.
It was fall and I got on campus for my senior year, and it was time for me to pick the topic for my senior thesis. The senior thesis was a big deal because it was like the capstone of your time there at Princeton, you were supposed to have learned how to think critically, and for me, it was more than that because wanting to do this PHD, I wanted to actually prove to myself that I have the concentration, that I could focus on something for so long.
But I decided that since this was like the last hoorah, I did not want to work on HTTP network caching proxies anymore. I wanted to work on something that was easy to convince people was super cool. I thought about it for a little bit and I came up with the idea that I would do my research in neuroscience, because I figured that no matter who you are or what your interests are, brains make you think.
I did some research online and I found a professor in the neuroscience department who was doing something that I thought was really appropriate that I could do, that would still be computer science related, and it was a field called brain computer interfaces.
Brain computer interfaces usually begins with a device called EEG or an electroencephalogram, which is something that you put on your head and these electrodes on your scalp measure changes in electrical potentials there. From this, we can interpret the data and try to decide if you're concentrating or asleep or excited or bored, and the computer interface part comes from the fact that we can take this data and use it to control a computer.
I'd read about people looking at EEG signals to help people control bionic limbs or you concentrate really hard and a drone flies. I figured if I worked on that, I wouldn't have any problem justifying why this was worth my time.
I convinced this professor to let me work for him. He really needed a programmer to help him build this app, to make experiments easier to do. We sat down together and we decided on an experiment design which would work like this.
We would bring people into the lab and we would set them up with one of these EEG devices. Then I would build this really simple video game for them to play, where they would see a picture of a face, and all they had to do was click right if the face was a woman's face, or left if the face was a man's face.
They just had to do this for 20 minutes. While they were doing it, I would look at the electrical signals coming out of the EEG and I would try to interpret whether or not they were focusing on the task or not. Then I would use this data to give them feedback, to either encourage them to focus more or to punish them for not focusing enough. I felt like that was a really appropriate topic for me since I already had so many attention issues.
I spent a couple of months working on the original prototype and setting up the experiment, and then it was time for me to calibrate it, and so I was my own first test subject. I got this EEG helmet device and the way it worked was, there were these tiny sponges that you put on the electrodes, and you had to soak them in contact lens solution because that helped them make better contact.
I took this device and I put it on my head and I had to weave the electrodes underneath my hair, and the whole thing was pretty sticky and uncomfortable. Then I did my boring video game for 20 minutes and it was not a pleasant experience.
Worse, I realized very quickly as I started looking at the data, that the very subtle, quiet, neural signals of my attention, were being overwhelmed by all sorts of other electrical activity, like if I move my shoulders or if I blink too much or if I breathe too excitedly, that would just completely corrupt the signal.
I had to restructure the experiment so that I would be focusing, unblinking, unmoving for a minute, and then I would get a twenty second rest period where I could blink, and then I would do it again and this would go on twenty minutes.
That made the data a lot cleaner, but as I started to build my algorithmic model, I realized there still wasn't enough signal there and I couldn't tell whether or not I was concentrating. This made me quite concerned because I'd already spent so much time, and I didn't have much time left for the thesis.
I went to my advisor, Ken, and I said, "Ken, I don't know if this is working."
He said, "If you have a very noisy signal, something that helps is that you just collect more data, so instead of doing the experiment for 20 minutes, do it for 30 minutes."
I thought, "That sounds very unfun, but I'll just recruit somebody else to do it." I got my boyfriend, Joe, and he volunteered his scalp, and I hooked him up to the device and he did it for thirty minutes. He didn't blink, but when I looked at all of that data, even thirty minutes worth, I couldn't make any sense of it and at this point I had only two months left, and I was starting to really flip out.
I went to Ken and I said, "Ken, we have only two months left. We have to do something else. We have to cut our losses. This isn't going to work. Our very premise about why this would work are flawed,"
And Ken said, "Slow down. You just need to collect more data. Do it for an hour."
I thought, ‘Ken, you haven't done this experiment before."
I said, "I feel like there are ethical problems with that. It's so boring and it's so uncomfortable and physically strenuous."
Then he said, "But we're paying them, right? We're paying them $40 an hour, right?"
I said, "Yeah."
He said, "And you want to graduate, right?"
I said, "Yeah, I want to graduate."
There wasn't much that I could do to fight that, but I had to find a new participant because I wasn't going to do it and my boyfriend wasn't going to do it. I sent out this email to the university list serve, with the subject line, 'Get paid $40 an hour to participate in a brain computer interface experiment'.
My experiment slots filled up instantly because everybody thought they were going to be a part of this sci-fi minority report-like thing.
My participants came into the lab and I started putting the contact lens solution on the device and wiring it up into their hair, and then I started to explain to them what they were going to be doing for the next hour.
Frankly, I felt pretty judged because I knew that they knew, that I knew that I had been fairly misleading in that email, but I had to graduate. Not all of my participants even made it through the whole thing.
One girl gave up after 20 minutes because she had this throbbing headache and another boy, he sat through the whole thing, except he said, "I had to throw away the garbage” because he was sure he fell asleep at least twice."
I looked through their data and some of them had definitely blinked when they were not supposed to have blinked. But even the hours’ worth of data that was reasonable-- It was kind of funny. I built this model to predict whether they were concentrating and the model was 52% accurate. What that means is that if you came into my lab and I wired you up with this device and I watched your brain for an hour, and then you asked me whether you had been concentrating or asleep, I was 2% more likely to be right than if I flipped a coin and said, "Heads, you're asleep."
But at that point, there was no point in going back to Ken because I only had two weeks left, and there was no way I could change the experiment. I had agreed to run so many participants in the study so I need to recruit people faster and faster.
It was too hard to go through the list now because of scheduling, so I had to dip into my friends and I had them come into my dorm room at 2:00 a.m. I had to lie to them a little bit about what they would be doing there and how cool it would be, but nobody said that there wouldn't be a human cost to science.
I collected all my data, I finished and it was the weekend before the thesis was due. I lock myself in the basement of the neuroscience building and I had to write up 50 pages of a report, and it's funny because I know that science is not about results, that no results are just as useful as positive ones, but it's hard to say that you did nothing when you spent an entire year and hundreds of dollars on an experiment.
I spent a lot of time in that thesis, talking about the previous work and the future directions and how that two percent accuracy was really promising.
Then I printed and then I had to have it bound in leather because at Princeton, all thesis are kept in the archives forever, so that you and your children and your children's children can always look back on your contribution.
I got it printed and bound and I got it sealed with the Princeton seal and I submitted it. I graduated, which was great. Then something else interesting had happened during that whole time, which was that I had been applying to my PHD programs and I got accepted at the University of Washington.
During that last week of school, when I submitted my thesis, I got this email from my would-be PHD advisor and I think that he really wanted me to come because I'd gotten two scholarships to go there, and he was emailing me asking what projects I was excited to work on.
I thought back on my thesis, and I figured, you know what? Even though I didn't get many results, I probably got through the toughest past of science, which is when you're working on something that you suspect is garbage, but you just have to keep on pushing through to confirm that it really is garbage.
I figured that if I had done that, then I maybe could survive six years of a PHD program. But part of me wondered, maybe there are types of people in this world that just really love finishing things. They like putting that last period at the end of their conclusion section, and submitting their thoughts to the archives of their lives forever, and maybe I'm more of a beginnings person and a middle person.
Maybe that was my calling, and perhaps PHD programs already had enough of these finishers and I would be better spending my life elsewhere. So I wrote back to my professor and I told him, "I don't know how to say this, but I have to turn down the offer. I can't join you next year in Seattle."
I moved back to my childhood house in New Jersey and I had absolutely no idea what I was going to do the next year, but part of me thought that maybe that was all right, because if there was one thing that I was good at, it was beginnings. Thanks.