Importantly, we must balance such massive scale in data, and the benefits of recording large neural yields with strong hypotheses, models, and rigorous statistical approaches to ensure we aren’t just "data-rich" but gaining knowledge.

What’s next? We outline a future defined by "big science" collaborations inspired by fields like Astronomy. We’re approaching terabyte-scale data, and will see more multimodal datasets. We envision a "team science" approach as the path forward.

Today, we’ve moved from single neurons to the ensemble era. Researchers need to bridge brain-activity and behavior by interpreting activity from thousands of neurons simultaneously by leveraging spike sorting, dimensionality reduction, machine learning, and neural decoding.

Early work was often bottlenecked by analysis: E.g., in the 1930s, Fourier transforms were painstakingly calculated by hand until mechanical "automated" devices were invented. This history shows that analytical tools often lag behind recording tech, a gap we’re still facing

A massive thank you and congratulations to you, Keiland, for leading this humongous project. We hope this review paper is a fun and useful read to those interested in data analysis of neural activity data.

Keiland Cooper@kwcooper.bsky.social · 3mo ago

From mechanical gears in the 1930s to deep learning in the 2020s, the way we study the brain has been a race between technology and analysis. How do we close the gap? Our new paper, doi.org/10.1080/2694... traces this journey, offers resources, and notes a plan for the future.