The lab has been renewed by HHMI until 2032! Yippeee! On for more science!
SEEKING POSTDOCS: The lab has recently embarked on very new unexplored directions. We are seeking postdocs interested in joining us in the effort to lead the field into what we (boldly? foolishly?) think will be the future of cognitive systems neuroscience.
For a brief (6 min) video summarizing the lab’s latest work and future directions, click here.
—
The brain is a very densely interconnected network. What signals are brain regions sharing? What do those signals mean? How do brain regions coordinate with each other to produce cognition and behavior?
To address these questions, we are conducting very large-scale recordings across the brain, using many simultaneous Neuropixels probes, while rats perform well-controlled cognitive behaviors (Bondy*, Charlton*, Luo* 2024). These large-scale simultaneous recordings let us assay internal signals — signals that are not necessarily time-locked to external events, but that are internally generated, timed, and are defined by coordination across neurons. Such signals comprise what we sometimes like to call “the internal conversation of the mind.”
Using single-probe recordings, we recently discovered one such signal, a biomarker for the moment when a subject makes up their mind about a decision. We call this signal “nTc”, for “Neurally-inferred Time of Commitment.” It is detectable in populations of simultaneously recorded neurons even if the subject made up their mind covertly, meaning they might not overtly indicate their commitment to a decision until many hundreds of milliseconds later (Luo*, Kim*, 2023). Our first major finding with the multiprobe Neuropixel recordings is that nTc, computed from the activity of neurons in a frontal region, marks a sweeping state change across the brain, observable in all the regions we recorded (Bondy*, Charlton*, Luo* 2024).
These results indicate that internal signals play a major role in neural activity, and suggest that they will be critical towards understanding the brain. Indeed, we have known for decades that most brain activity looks like noise, yet is structured across neurons (e.g., Arieli 1996; Fiser 2004)– which is exactly what nTc looked like before we understood what it is. Perhaps much neural activity consists of undiscovered internal signals. Can we discover more of these internal signals? Doing so requires characterizing and elucidating structure in simultaneously recorded neural activity.
This is the right time for such an endeavor: First, advances in recording methods make it possible to record from ever more neurons simultaneously, giving us better and better statistical power to detect structure in single-trial neural activity. In a collaboration with Tim Harris, we are planning to record using his next-generation “NXT” Neuropixels probes as soon as they are first produced in Spring 2025; with eight simultaneously inserted probes, we expect to record, electrophysiologically, from 6,000 to 12,000 neurons simultaneously, from tens of brain regions across the brain. Second, and critically, the AI revolution gives us unprecedented tools to discern structure in such large-scale activity. A major part of our efforts is to develop new AI-based neuroscience analysis tools that are commensurate with the new large-scale datasets being generated (Koukuntla 2024, Kim 2023). Third, doing this in the context of well-controlled cognitive behaviors, which my lab specializes in (e.g., Pagan 2024), gives us many quantitative features with which to compare those internal signals, which will help us understand their meaning. Indeed, these three elements together are what made it possible to discover nTc. We believe that the intertwined combination of all three is the future of cognitive systems neuroscience.
Come join us as we work to define what that future looks like. We are seeking postdocs with an appetite for risky, technically complex, and fundamental new directions. (Strong programming and quantitative skills required.)