Through vibe coding, ChatGPT has been making stunning publication-quality analysis and figures for me, w/ detailed README-notes, logs. All scripts/functions are well organized, named, and annotated. It often does things that would take a good programmer a few days. It is truly impressive and scary.

Recently showed K. Harris neural data recorded during a task and asked whether dimensionality reduction was the right approach. His view: first try hard to understand what drives the neurons, using linear and nonlinear regression models that account for events, kinematics, time, and other variables.

The Transmitter @thetransmitter.bsky.social · 3w ago

I fear that placing too much emphasis on a specific interpretation of dimensionality, or treating dimensionality as an end-all quantification of some aspect of neural computation, may lead us down the wrong path, writes @mattperich.bsky.social. #neuroskyence www.thetransmitter.org/neural-dynam...

Writing a paper; here's my workflow: 1. Statistics in R in a single Results.qmd file: methods and interpretation sit alongside each code block, results printed in html. 2. MATLAB scripts for figures; Adb. Illustrator for cosmetic cleanup. 3. Writing in TeXstudio, pulling statistics from Results.qmd.

There should be more discussion of how behavioral trials are initiated in experimental design. Many experiments impose trial onset at random, but recent work by @davidrobbe.bsky.social suggests that self-initiated trial times in rats follow a long-tailed Wald distribution: 10.1101/2024.05.31.596850

We developed DANT to match units from the same neurons from chronic Neuropixels recordings. The key idea: cluster units with similar waveforms and firing statistics, then iteratively refine matching and drift correction. We tested it in rat cortex and striatum during reaction-time behavior.

Patterns@cp-patterns.bsky.social · 3mo ago

Online Now: Density-based longitudinal neuron tracking in high-density electrophysiological recordings #datascience

Online Now: Density-based longitudinal neuron tracking in high-density electrophysiological recordings #datascience

Density-based longitudinal neuron tracking in high-density electrophysiological recordings

Tracking the same neurons across days or weeks is essential for understanding learning, stability, and plasticity in the brain but remains technically challenging in chronic electrophysiology due to probe drift and unit turnover. Huang et al. present DANT, a framework using density-based clustering for longitudinal neuron tracking that improves matching yield while maintaining accuracy across Neuropixels recordings in cortex and striatum during trained motor behaviors in freely moving rats.

dlvr.it

2025 was the year we ramped up chronic Neuropixels experiments (even wrote a paper on unit tracking), and it gave me a lot of hope for the future. But I just learned that Neuropixels won’t be sold to China anymore. If you have easy access to this technology, please don’t take it for granted.

Wait, some of you don’t like that neuroscience is such an opinionated field, unlike machine learning, which can be judged by numbers?

I almost write everything in R Studio now (using R Markdown or Quarto): daily log, manuscript, notes, paper reviews, talks, data presentations, etc.