Dr. Grace Lindsay
Postdoctoral Research Fellow University College London
PhD in Neurobiology and Behavior Columbia University
Dr. Grace Lindsay was hooked on neuroscience before she even knew its name. In her high school psychology class, she was fascinated by topics of human behavior and the inner workings of the mind, but the class left her with more questions than answers. Grace realized that the focal point of her curiosity was not the thoughts or behavior per se, but rather the biological mechanisms underlying those thoughts and behavior. She googled “biopsychology”, wondering if that was even a field of study. Grace eventually came to realize that her piqued interest did indeed have its own name: neuroscience. This intellectual passion morphed into a serious career plan when she found out that most science doctoral programs pay their students. Thrilled at the idea of studying the brain as a job, Grace left high school hoping to eventually pursue a PhD in neuroscience. She went on to do exactly that.
As an undergraduate at University of Pittsburgh, Grace dove into clinically-relevant experimental neuroscience, working with a rat model of OCD and a mouse model of depression. Despite the university’s reputed strength in computational neuroscience, Grace originally had no interest. She thought computational neuroscience was based on the premise that the brain operated like a modern computer – quite a ridiculous assumption, in her opinion! Grace’s perception changed when she attended a talk by Dr. Brent Doiron, a University of Pittsburgh mathematics professor doing computational neuroscience research. He presented a model neuron described by a series of equations, then further showed how to combine the equations of multiple neurons and thus use math to understand brain circuit function. Grace was struck by the feeling that this was what she had been looking for since high school psychology – the bare-bone mechanisms underlying neural activity and behavior. She went on to do an undergraduate thesis co-advised by the professor who gave that lecture, dropping her double major in philosophy to focus on learning more math.
While attending a Society for Neuroscience conference towards the end of her undergraduate career, Grace became utterly overwhelmed by the size of the field. “It broke me!” she says. Although a neuroscience PhD was far from a new plan for her, she suddenly felt unprepared to write a personal statement for graduate school applications – how could she say with confidence what she wanted to study? Instead of applying to graduate school, Grace applied to research internships and fellowships to give herself time to discover her niche. She landed at the Berstein Center in Freiburg, Germany, where she focused on honing her coding and math skills and learning more about the different subfields within computational neuroscience.
With more of a focused plan for future research, Grace then joined the Neurobiology and Behavior PhD program at Columbia University, ultimately working with Dr. Ken Miller. Her thesis work centered on building computational models to understand how attention modulates perception. She configured a network of model neurons that, when given an input image, reported what objects that image contained. By mathematically modifying the amount of “attention” the network paid to particular objects in the image, Grace could improve the network’s performance in identifying those objects. Her results aligned with studies showing that attention affects animals’ neural activity and performance on visual tasks. Furthermore, Grace could more strongly enhance her model’s performance by modulating neurons deeper in the network. This result could explain why, in animals, attention has a larger effect on neural activity in higher-order brain regions compared to primary visual areas. Interestingly, Grace also found that if she manipulated the network to “look for” something that wasn’t there, the neural network would sometimes output a false positive. This phenomenon has been observed in humans as well – people told to look for particular objects in an image flashed on a computer screen sometimes think that they see the object when it’s not really there.
Grace is still using such deep neural networks to study the visual system, but now she does so as a postdoctoral Research Fellow at University College London (UCL). The fellowship position was created to bridge the Gatsby Computational Neuroscience Unit at UCL with the Sainsbury Wellcome Center (SWC) for Neural Circuits and Behavior. Grace is a perfect fit for this position, as she can combine her neurobiology background with her computational neuroscience interest and training. While her PhD work focused on top-down attention, she is now studying the feedback loops that modulate normal processing of incoming visual information. She is building computational models but is also working with a mouse vision lab at SWC so that the models can be directly tested with behavioral experiments, which can in turn inform the modeling process.
These days, when she isn’t building neural networks, Grace is writing or podcasting about science. She first became involved in science communication as a graduate student when she started blogging about new concepts or particular papers that she wanted to do a deep dive into. She found that blogging was a perfect way to help her digest the information while also packaging that knowledge into a shareable form. Later, she started a podcast, Unsupervised Thinking, that gives her the same dual satisfaction of learning while producing content. The podcast, however, is now on hiatus while Grace is writing a book! Thinking she might like to write a book someday, Grace had casually reached out to a publisher to start to understand the nuances of the author-publisher interaction and to get some feedback on her writing portfolio. Things quickly snowballed, and Grace now has a contract to write Models of the Mind, a book exploring the field of computational neuroscience for a general audience. It juxtaposes the future with the past: Grace explains how bits of math, physics, and computer science first started to intersect with neuroscience, and why computational neuroscience is considered by many to be the future of the neuroscience field. Perhaps in a couple years, a high schooler will read Models of the Mind and have an aha moment like Grace’s: they will see computational neuroscience as the perfect avenue for understanding the most fundamental mechanisms of the brain.
Listen to Nancy’s full interview with Grace on February 28th, 2020 below or wherever you get your podcasts!
