Datasets & Tools
The most direct path into BCI research is getting a classic decoding pipeline running on a public dataset. This chapter collects the major neural datasets, decoding benchmarks, open-source toolchains, and learning resources as the practical backbone for the whole book.
How to use this chapter. This chapter is a "hands-on checklist" — it does not assume Chapters 1–13 have been finished, and you can dip into it at any time. Pairing it with the earlier chapters is more efficient, though: pick up a decoder in Chapter 04 or 05 → reproduce it on the matching task in Neural Latents Benchmark: NLB / FALCON; learn a signal modality in Chapter 03 → grab the ready-made preprocessing pipeline from Open-source Tools: MNE / EEGLAB / CEBRA; read about a real system in Chapters 06–07 → find the authors' public code and courses in Leading Labs & Learning Resources. Every actionable piece of the BCI chapter lives here.
Recommended reading order. Total newcomers should read this chapter in reverse: first use Leading Labs & Learning Resources to pick a track (Stanford NPL for motor cortex and intent decoding, Caltech for posterior parietal and high-dimensional decoding, Pitt for ICMS and embodiment, BrainGate for clinical, Shanechi for DPAD and affective BCI), then assemble the toolchain via Open-source Tools: MNE / EEGLAB / CEBRA, and finally pick a specific task from Neural Latents Benchmark: NLB / FALCON as your first hands-on project. Researchers with a direction already chosen can jump straight to the benchmark.
Chapter contents:
- Neural Latents Benchmark: NLB / FALCON — Neural Latents Benchmark and FALCON NeurIPS 2024
- Open-source Tools: MNE / EEGLAB / CEBRA — preprocessing and decoding code bases
- Leading Labs & Learning Resources — Stanford NPL, Caltech, Pitt, BrainGate, Shanechi, and more