Further study

Goal of this page. A short map of where to go beyond nSTAT — the topics this toolbox does not implement, with primary references.

See also: the glossary defines every term used across the concepts track, with HTML anchors for direct linking.

What nSTAT covers

The implemented topics are organized in the concepts index:

  • spikes and the LFP from a microelectrode;

  • spike trains as point processes and the encoding GLM;

  • the LFP and spectral analysis;

  • goodness-of-fit (time-rescaling) and decoding (PPAF / PPHF / clusterless);

  • the state-space GLM and EM-trained latent state-space models;

  • functional connectivity (ensemble GLM, Granger);

  • uncertainty and confidence intervals;

  • rhythmic firing and the clinical microelectrode.

Each page pairs intuition with the matching nSTAT objects, runnable snippets, and primary references.

What nSTAT does not — and where to learn it

  • Population geometry and dimensionality reduction. nSTAT ships only a PCA sketch (population geometry). For the standard tooling see Gaussian-Process Factor Analysis (Yu et al. 2009) and the dimensionality-reduction guide (Cunningham & Yu 2014); for the dynamical-systems view see Vyas et al. 2020.

  • Deep-learning encoders and decoders. The bridge page From filters to deep learning draws the line from the PPAF to modern sequence decoders — what carries over (encoding underlies decoding; goodness-of-fit and uncertainty still matter) and what changes, with pointers into the literature.

  • Spike sorting. nSTAT consumes already-sorted spike trains. For raw acquisition to sorted units, use a dedicated tool such as SpikeInterface, then bring the results in via the interop bridges.

  • Vendor-format I/O. The interop bridges (nstat.extras.interop.{neo,nwb,pynapple}) read Spike2 / Blackrock / Plexon / NEX / TDT and NWB into nspikeTrain / SignalObj.

See also