“Being part of this feedback learning process where I can help people grow and guide them is very rewarding to me, and a large part of my job.”
Adeen Flinker, Neuroscience PhD Program alum (entering class of 2005)
Adeen Flinker is an assistant professor of neurology and biomedical engineering at NYU Grossman School of Medicine, where he uses rare recordings of neural activity from patients undergoing brain surgery to understand how speech is perceived and produced. One of his major goals is to help patients — from improving pre-brain surgery language mapping, to developing therapies for patients with Parkinson’s disease who have speech deficits. His lab is even trying to develop neural speech prosthetics for people who are unable to speak.
Flinker grew up in Israel and was an undergraduate computer science major when he was introduced to neuroscience by his neighbor, former Helen Wills Neuroscience Institute (HWNI) faculty member Noam Sobel. After spending a summer doing research at UC Berkeley in Sobel’s lab, he says he “fell in love with neuroscience” and joined the Berkeley Neuroscience PhD Program.
Flinker did his PhD thesis in Robert Knight’s lab, where he studied speech and language using electrocorticography (ECoG). In this technique, which Flinker uses in his own research program today, electrodes are placed directly on the cerebral cortex of patients awaiting surgery for epilepsy. The primary purpose of the electrodes is to monitor and localize seizures to guide the surgery, but they can also be used by scientists like Flinker to obtain valuable data about how the human brain works.
After graduating, Flinker did a postdoc at NYU, where he uncovered differences in how the right and left hemispheres of the brain perceive different aspects of language. He then took a non-tenure track job at NYU’s School of Medicine running their ECoG program, which later developed into his current tenure track position. Flinker says that he wanted to be tenure track mainly so he could mentor PhD students.
Read our Q&A with Flinker to learn more about his research, career path, life outside of work, and why he says some of the best years of his life were at Berkeley. This Q&A has been edited for length and clarity.
Q: How did you first become interested in neuroscience?
A: My background is actually computer science (that was my undergrad), and I was interested in AI and natural language processing. I happened to have a neighbor, [former HWNI faculty member] Noam Sobel, who became a professor at UC Berkeley. I was an undergrad in Israel, and when he would visit, we would chat about a bunch of stuff. He kind of introduced me to neuroscience, and persuaded me to work with him over the summer as a research assistant. When I did, I really fell in love with neuroscience. Being at Berkeley, and interacting with the grad students, and doing some research really turned my passion towards neuroscience. That’s when I decided to apply to the [Berkeley Neuroscience PhD] program.
Q: How was your experience in the program?
A: I loved the program. Berkeley was really some of the best years of my life. I enjoyed the program thoroughly, had a great time personally, academically, and made a lot of friends. I met my wife at Berkeley. It was quite an influential part of my life.
I really cherished the scientific environment that I felt was top-notch in terms of the scientific level [and the] very supportive and positive interactions between PIs. Also, the mentorship I received felt more like a family than just a job. That’s something that’s stayed with me, and I try to create in my position and my lab — have a very familial feeling to the lab. Given that we all spend a long time of our day and our life in the lab, whether you’re doing your PhD for several years, or that’s your work and you’re going in every day, you want to enjoy it and have people that enjoy being there, and you enjoy their company.
Q: Tell me about the lab rotations you did before joining the Knight lab.
A: I was new to neuroscience, so I decided to learn as much as I can, and do kind of a breadth view. I rotated with Udi Isacoff working on light-induced ion channels; I rotated with Noam Sobel doing human olfaction psychophysics; and I rotated with Jose Carmena working on BCI [(brain-computer interfaces)]. In the end, I did a fourth rotation with Bob Knight, and I joined his lab.
Q: What research did you do in Bob Knight’s lab? Were there any interesting findings?
A: When I started out, I was thinking of perhaps doing some BCI with humans and ECoG. Moving towards my qualification exams and delving into the literature, I felt that’s not really where my passion was, and I wanted to move more towards language. Bob was very supportive. There were not many people in the lab studying language at the time, so I was spearheading some of that. I designed several tasks, [and] we had support from Nina Dronkers who was at [UC] Davis and the VA Martinez at that time.
We had a series of different studies, and I worked with ECoG patients at [UC] San Francisco, Stanford, and Johns Hopkins. So really all over, helping setting up different experiments including my own. And quite a few things came out. One of them, which I’m very excited about, [is that] we found that Broca’s area — which since the 19th century [was traditionally] thought to be a center for speech production — using the ECoG recordings, which have a great temporal and spatial resolution, we found that it’s actually not active when we when we’re speaking. It’s active right before, when we’re planning our speech. So it’s responsible for articulatory planning, sending information across cortex to motor cortices, and it’s motor cortex that does the actual articulation per se. That really was a head turner for me, and for many, I think, because it goes against the classical neuroscientific or neurolinguistic models of language production.
Q: What did you do after you graduated from Berkeley?
A: I moved to New York City and I did a postdoc with David Poeppel. During my PhD I had an F31 NIH grant, and then I used my time to write an F32 to work with David. That really funded my postdoc and my transition to New York City. The project that I wrote the grant for and we completed has to do, again, with speech and language, but tackling the question of hemispheric asymmetry. So when I’m talking to you right now and you’re hearing speech, what side of the brain is doing the heavy lifting? And what kind of lifting is it doing? Is it the left side? Is it the right side?
It turns out that speech perception is bilateral in nature. But there’s a very long literature showing that there’s asymmetries. There are different cues when I talk to you that the left is picking up more than the right, and the right more than the left. So I came up with a computational technique in space in order to quantify what these differences are.
I was very influenced by work coming out of UC Berkeley — work from [HWNI member] Frédéric Theunissen’s lab. Taffeta Elliott and Frédéric Theunissen came up with this paper about the modulation domain of speech. It was a behavioral study, but the results really struck me as something that could be the foundation of perhaps how the same types of features that were shown [in the paper could be what] the left and right process differently. I used that type of approach in my own computational approach in order to ask questions about how the left hemisphere may be integrating more temporal information or temporal modulations, while the right is integrating more spectral or frequency modulations.
There were a series of studies that came out in Nature Human Behavior [that] included psychophysics as well as MEG [(magnetoencephalography)], which is a non-invasive technique that we did at NYU with David Poeppel, as well as ECoG recordings that I did at Long Island Jewish [Medical Center]. So starting from computational technique and going to psychophysics and to non-invasive and invasive measures, we were able to show that although sometimes it’s very small changes, it’s the changes in temporal aspects of speech versus changes in frequency aspects or spectral modulations of speech that are driving asymmetries in the left and right side of the brain.
Then I was offered a position at the School of Medicine here [at NYU].
Q: Is that where you were doing your postdoc?
A: No, I was doing a postdoc at NYU downtown. They’re actually very separate business units and schools. I had a lot of background in ECoG, and they were looking for somebody to take over the position here. They interviewed me and offered the position, which at the time was a non-tenure track position, so it was quite a risky move on my part. But I wasn’t really ready to go on the market, and I wasn’t really ready to give up academia. And it’s a pretty rare opportunity to run an entire ECoG operation if you’re not an MD, and especially an operation that’s as high throughput as NYU, which is one of the largest in the world.
Q: Did your position turn into a tenure track position?
A: It did. It took a lot of work; it basically took a lot of grants. But after I was able to provide sufficient grants, they moved me to a tenure track position. And the main reason I wanted tenure track was [that] you can’t really be a primary mentor of PhD students at the school of medicine if you’re not tenure track, and that was something I really wanted to do — to mentor PhD students.
Q: Tell me about the research you do in your lab now.
A: The lab is predominantly focused on ECoG work in speech production. I do have a line of work where we’re also working with Parkinson’s disease patients, both behaviorally as well as intraoperative recording in single units. In general, the technique is rare neurosurgical recordings in humans relating to speech and language production.
Right now, there are several postdocs and several PhD students in the lab. We’re studying quite a few questions [including]: What happens when you’re monitoring your own speech during production? And what regions are helping you attune to different cues and correct your speech if you hear something that’s weird or incorrect?
For example, right now we’re talking, and if I had headphones on and there was a delay in Zoom, I could hear an echo of myself. If it’s a small echo or a short delay, you’re fine. But if it’s a long delay, that really interferes with fluent speech production and I’ll stop speaking, or maybe I’ll slow down or slur or even stutter. And interestingly, [for] people who stutter, the opposite effect happens. If you introduce a long delay in their feedback, they actually start speaking fluently for a while. What networks are involved is really non-delineated, and I would say unknown.
So one project that we did is we used this online delayed auditory feedback, where we introduced these online delays randomly with patients, and were able to show what regions of the brain are processing this error. And surprisingly, we found an area of motor cortex — the dorsal part, so the superior part of motor cortex — that is not actually recruited for articulation. Like auditory cortex, it’s processing auditory errors, and allows you to compensate for this weird delay when you’re speaking. So it’s processing auditory cues like auditory cortex, in order to facilitate fluent speech production. And that actually just came out today [Ed. note: this interview was conducted on February 3, 2022] in PLOS Biology.
Q: For people who are not familiar with ECoG, can you talk a little bit about the technique?
A: Most of the research done with humans is ethically constrained to non-invasive techniques, and invasive techniques are usually only done with animals. However, there’s a golden opportunity to work with neurosurgical populations who have to have implanted electrodes for their clinical treatment. We’re not deciding where electrodes are placed, but we piggyback on what the clinicians are doing.
So in severe refractory epilepsy, untreatable by medication, the last resort is to map out where the seizures are coming from via two surgeries. In the first surgery, a grid of electrodes or multiple electrodes are implanted inside and on top of cortex in order to monitor seizures. After implantation, the patient just sits in the hospital room for a week watching TV, talking to family — just waiting for seizures. When the seizures hit, the clinicians can map out the center or the foci of these seizures, and map out what should be removed in the subsequent surgery — what should be resected to treat the epilepsy.
During this time, they have all their electrical activity from these electrodes monitored by the clinical hospital systems. The patients, the majority of the time, are just sitting around on their phone, watching TV, talking to family. That’s where my team comes in; we consent them for research. If they’re up to it, we ask them, ‘Instead of watching TV, do you mind working with us for an hour here and there?’ Most of them are very happy [to do it] because they’re bored out of their mind and they can stop at any time. When they’re up to it, we run a battery of different tasks. Most of the tasks we run in the lab are related to speech and language production and perception.
Q: What are the big picture goals of your research?
A: There are multiple goals. One is just to answer basic questions about how speech and language information is transferred from different parts of cortex, from temporal cortex to frontal cortex, and how that subserves fluent speech production.
The other is to try to feed back to patients. A lot of the work we do actually has some clinical avenue to it. We’re trying to correlate neurophysiological measures with clinical electrical stimulation mapping. We’re trying to improve clinical mapping of language pre-surgically. We’re also trying to create neural speech prosthetics — so using the ECoG measurements to create a neural network that can create synthetic speech for patients who cannot speak, as well as working with Parkinson’s disease patients where delayed feedback, for example, has a therapeutic aspect and can help them with speech deficits that are either caused by Parkinson’s disease or from a deep brain stimulation surgery itself.
One clinical aspect, for example, is [that] to map out language surgically, surgeons stimulate cortex. They can’t just measure activity. They have to see what areas are critical for speech output or language comprehension by ‘zapping’ or running an electrical current through different regions while [the patient is] doing a task. This takes a long time, it’s painful, and it can evoke seizures, even in non-epileptic tumor patients. But it’s necessary for the surgeons to map out the regions that when you stimulate [them], the patient is unable to produce speech or comprehend language.
We’re trying to predict these areas that are critical [for speech and language] from our recordings that are much shorter and don’t involve running current through different regions of the brain. If we’re able to do that, we could record a short session of a task (10, 15 minutes), run different computational models, and provide a map of critical regions for the surgeon or the clinician that can either replace their stimulation mapping, or at least guide their stimulation mapping if they’re constrained on time. We want to show them the hotspots — they should try this pair or that pair [of electrodes] — because they don’t have time to map out all the pairs of electrodes.
Q: What do you enjoy about your job?
A: There are several things I enjoy. It’s important — you want to be in a job that you enjoy. I really enjoy mentoring. I really enjoy providing feedback and watching mentees grow, whether postdocs or graduate students. Being part of this feedback learning process where I can help people grow and guide them is very rewarding to me, and a large part of my job.
I also really like the hands-on interaction with patients at the hospital, and with clinicians. And working with very rare data that kind of gives you a ground truth sense to it. These are direct recordings and I feel very close to the scientific process, getting these very rare signals.
Lastly, I have tried to move some of the research to a more translational aspect, because I really enjoy having something that’s not just purely academic, but could maybe in a shorter timeframe feed back to help patients. Which is why we’re driving these projects on predicting electrical stimulation mapping or doing these delayed auditory feedback tasks, learning about them, so we can perhaps help different populations where this could be a therapeutic approach.
Q: Do you have time to do anything fun outside of work, any hobbies?
A: I have a wife and two kids, so that’s a lot of time for me. But yeah, I do a bunch of martial arts, that’s always been part of my life. Before COVID I used to love traveling a lot. We just traveled recently to Mexico for the first time since the pandemic.
A big part of my life is my family and my children. That’s one thing I do like about academia where, depending on what stage [you are], you can sometimes manage and juggle that more freely than industry. But that’s not always the case; I guess there are pros and cons to both.
I also bike to work a lot. Anything I can do for exercise and to survive type 1 Diabetes. I’m a coffee connoisseur — I really like making good coffee. We have nice coffee machines in the lab. I like fixing the coffee machines; I like fixing things in general. Tinkering and fixing are some of my hobbies.
Q: Do you have any advice for students who want to go into tenure track positions?
A: First, I will say follow your passion and your heart, whether it leads you to an academic position or a non-academic position. I don’t think one is better than the other. Go towards a position that is good for you, and is fulfilling.
Then if you are looking for a tenure track position … I mean, I honestly don’t know if I’m the best person to give advice, because I did not go on the market. I didn’t take a tenure track position; I took a non-tenure track position. So I guess my advice would be perhaps not to do that [laughs] because it’s very risky. It worked out for me, but you never know. On the flip side, I would just say, follow your passion, follow life where it takes you and things will work out, whether it’s a tenure position or a non-tenure position or industry. As long as you’re fulfilled and doing something you enjoy, things will work out.