How Did I Choose My Major

I have finally confirmed my two majors in the first week of my sophomore Spring: Cognitive Neuroscience and Computer Science.

Cognitive neuroscience (CogNeuro) was the option I had long settled on in terms of academic goals. Before I turned seventeen, I already knew that, no matter what, I would want to take on memory as a domain within cognitive science. That once-vague intuition has now more concrete. I want to build a brain–computer interface system: one that can, by reading low-level basic life activities in real time (such as fMRI measurements of blood flow, EEG recordings of electrical signals, or probes detecting combinations of action potentials across multiple neurons), extract high-level human intentions, mental imagery, and concepts—especially at the level of memory. It should also be able to store these images and concepts in a reasonable format, and present them to others within a structured framework. The idea is based upon the concept of ROM (read-only memory) in William Gibson’s Neuromancer.

The justification for this academic goal can be divided into two domains. In a clinical context, for conditions such as Alzheimer’s disease, amnesia, and dementia—where memory loss is a defining symptom—this technology could help patients retrieve and preserve memory fragments that have not yet disappeared from their unconscious or subconscious. This would be beneficial both for patients’ rehabilitation and for communication between patients and their friends and families. In a personal and familial context, such a technology might preserve an individual’s memories even after death—and, through those memories, perhaps reconstruct parts of their consciousness—thus becoming one way for individuals and societies to resist death.

I still need to examine the specific technical pathways. For example: how should one choose between invasive and non-invasive brain–computer interfaces? Should the focus be solely on the medial temporal lobe, or on the whole brain? Which neural decoding algorithms are truly reliable? What counts as a “reasonable format” for memory storage? Should mental imagery or concepts be represented through language models or image models? These are questions I can only answer through sustained coursework, research, and inquiry—and CogNeuro is the first step in all of this.

The goal I have articulated belongs to the field of neuroengineering. It took me long to decide a second major that would both align with this field and serve as a practical tool for me.

I cannot claim that like computer science (CS) at all. The “coding” aspect of computing is, for me, merely a tool that conveniently solves many automation problems and assists future research in experimental design, implementation, and data analysis. I have realized that the only part of computer science I truly enjoy is algorithms, especially their mathematical derivations; I have little interest in their concrete implementation in code. In that sense, what I really like is theoretical mathematics.

I like many aspects of electrical engineering (EE). Its curriculum includes a substantial amount of physics, which conveniently helped me recover physics knowledge I had not revisited since high school, and even broaden it (indeed, just two courses in electricity, circuits, and signals in my sophomore Spring led me to concepts such as Maxwell’s equations and complex circuit components). I still like physics; had I not become a neuroscientist, I might have become a physicist. I also greatly enjoy the hands-on elements of EE courses: learning the principles behind signal amplifiers, building physical Arduino circuits, computing system transfer functions and Fourier series. I learn methods and apply them in practice.

Yet CS ultimately prevailed over EE. It prevailed because I do not yet have sufficient interest in designing interface hardware, but I am drawn to the software development, decoding algorithms, and other “soft” designs behind the interface. Moreover, CS is a universal major—one that maps onto work across almost all modern technological fields. If, due to limitations in my abilities or situational changes, I am unable to pursue research-oriented positions within academia, CS still leaves me a viable path forward.

Thus, the structure of CogNeuro as the primary major, CS as the secondary came into being.

So far, I have spent far more time in CogNeuro courses than in CS. I have completed 10 out of 17 CogNeuro courses, but only 6 out of 15 CS courses. This puts me in an awkward position: I do not want to finish my undergraduate degree by taking at least two coding-heavy courses every semester for the next four semesters, yet it seems to be an unavoidable trial. Every time I think about how the remaining twenty courses of my college career were all fixed on a single night in the spring of my sophomore year, with no room left for adjustment, I again feel an illusory suffocation, as though I were clamped down by a system. Unfortunately, I deserved this; a bitter smile is all I can do.

During my sophomore summer, I was fortunate enough to join Brown’s largest neuroengineering laboratory, where I met leading professors in the East Coast neuroengineering community and received a $2,500 stipend for a three-month internship. The lab’s core work focuses on restoring upper-limb mobility in patients with tetraplegia through soft robotic exoskeletons. My “mentor,” an Indian PhD student who grew up in the Davis, CA, was solid, hard-core, and dreamed of doing something big like I do. Through that summer, I adapted to lab routines and experimental workflows, and learned what researchers actually do, pay attention to, and strive for. For the first eight weeks, I worked on a peripheral task: reconstructing and developing a Unity game. The game was designed to train one-dimensional movements of various arm joints, allowing subjects to control a virtual human to follow another virtual avatar’s motions. When I took over the software, only a runnable shoulder-joint mode had been developed. I independently implemented an elbow-joint mode and a combined shoulder–elbow dual-joint mode. This became the largest codebase I had worked on in my two years of college—two C# files of roughly 800 and 400 lines respectively. During development, I also accompanied my mentor to different experimental sites to observe procedures, sometimes spending entire days there. It was, without doubt, a valuable experience.

Given the current situation, I cannot guarantee that my path toward a direct-entry PhD will be entirely unobstructed. I will need to apply to both industry and academia during the senior-year application season. If I receive offers from schools I like, that would be ideal; if not, working for a couple of years at a neuroengineering company before pursuing a PhD is also a viable strategy. After all, in fields like neuroscience, not having a PhD is effectively equivalent to having nothing. As for post-doctorate life, my sophomore-year self would hope to be at a fast-growing neuroengineering startup, working as an algorithm engineer in a technology-driven engineering lab, or leading a small team to produce exciting innovations in brain technology.

I may not have entirely escaped my former, purely “fear-of-death” motivation. I have merely rationalized it into what appears to be a concrete career plan. At times, I forget my original impulse amid all these academic efforts. But whenever I remember it, I rerun the entire chain of reasoning to convince myself that the choices I am making now are worth it. Or perhaps the fear of death is gradually fading as I begin to truly experience life, transforming into a curiosity about the unknown and a tangible sense of control over an intangible future. Death has suddenly moved much farther away from me. I am happy with what I am doing now—not because I have adequately prepared for events decades down the line, but because I am finally living in the present.