Scalp to Source
Progress

Paths › path-oscillations

path-oscillations

Oscillations and spectra

For: Systems/cognitive neuroscience. 35 steps · ~30 h.

Why this order

The output here is a claim about a rhythm, and the hard part of that claim is not computing a spectrogram — it is showing that the rhythm is real, that the band it lives in was not chosen by the filter, and that the map is not one source seen from twenty electrodes.

Levels 0 and 1 in full, because everything downstream is spectral. L1.3 and L1.4 set the estimator and its trade-offs, L1.6 deals with the line that sits in the middle of the band people call gamma, and L1.7 introduces the separation of aperiodic background from genuine peaks — the lesson that decides whether “more beta” is a rhythm or a slope.

Level 2 in full, because the reference determines what a connectivity or topographic claim means (L2.3), the filter choices determine the band structure you will later interpret (L2.4), and ICA decisions can remove the very components an oscillation analysis measures (L2.6, and the over-cleaning pitfall).

Level 4 as the substance: what averaging hides (L4.1), the estimators and their resolution trade-off (L4.2, L4.3), baseline normalisation and ERD/ERS (L4.4), phase and coupling (L4.5), and then L4.6, which is where the level asks whether any of it was an oscillation. C4 is the worked example.

Then Levels 5.1–5.3 before the inference chapters, because volume conduction is the reason a sensor-space rhythm appears everywhere at once: L5.1 states the problem, L5.2 shows which connectivity measures survive it, and L5.3 gives the surface Laplacian, the cheapest defence against it. Deferring them past Level 4 keeps the estimator lessons together; putting them before Level 6 means the statistics are applied to quantities whose spatial meaning you already understand.

Level 6 selectively: L6.1 for the multiple-comparisons landscape that a time-frequency map makes urgent, L6.3 for effect sizes and power, and L6.6 for reporting. The rest of Level 6 belongs to the ERP and decoding paths.

Paths do not add content; they filter and order the ladder (§7). Lesson pages keep their own prerequisite links.

Lessons (35; about 30 h 25 min of lesson time, plus the capstone)

  1. L0.1 What EEG measures · 40 min
  2. L0.2 Electrodes, montages and the 10-20 system · 40 min
  3. L0.3 Amplifiers, sampling and recording · 45 min
  4. L0.4 Reading raw traces · 60 min
  5. L0.5 The artifact atlas · 75 min
  6. L0.6 Data hygiene and metadata · 45 min
  7. L1.1 Sampling, Nyquist and aliasing · 40 min
  8. L1.2 Time and frequency domains · 60 min
  9. L1.3 Power spectral density · 60 min
  10. L1.4 Windowing, leakage and zero-padding · 30 min
  11. L1.5 Filters: FIR, IIR, and what they do to your data · 75 min
  12. L1.6 Line noise · 40 min
  13. L1.7 Aperiodic and periodic components · 60 min
  14. L2.1 Loading data and BIDS · 45 min
  15. L2.2 Channel locations and bad channels · 60 min
  16. L2.3 Re-referencing · 50 min
  17. L2.4 Filtering in practice · 45 min
  18. L2.5 Artifact rejection strategies · 50 min
  19. L2.6 EOG regression and ICA · 90 min
  20. L2.7 ASR, SSP and alternatives · 45 min
  21. L2.8 Pipeline order and reproducibility · 60 min
  22. L4.1 Why time-frequency · 40 min
  23. L4.2 STFT and Morlet wavelets · 75 min
  24. L4.3 Multitaper and filter-Hilbert · 50 min
  25. L4.4 Baseline normalization and ERD/ERS · 60 min
  26. L4.5 Phase, ITC and cross-frequency coupling · 60 min
  27. L4.6 Is it really an oscillation? · 60 min
  28. L4.7 Statistics for time-frequency · 50 min
  29. C4 Capstone — Mu/beta ERD
  30. L5.1 Volume conduction: the central problem · 45 min
  31. L5.2 Sensor-space connectivity · 75 min
  32. L5.3 Surface Laplacian and CSD · 45 min
  33. L6.1 The multiple-comparisons landscape · 60 min
  34. L6.3 Effect sizes, power and precision · 50 min
  35. L6.6 Reporting standards · 40 min

Skipped prerequisites (optional reading)

Each of these is a prerequisite of a lesson this path includes. They are optional here, not removed — the lesson pages keep their prerequisite links — and the first three are the ones most likely to be worth the detour.

  • L3.2 The ERP and its components — a prerequisite of L4.1, which opens by saying what trial averaging hides. The averaging and √N argument, and the distinction between a latent component and an observed peak, are stated there.
  • L3.7 Statistics for ERPs — a prerequisite of L4.7 and L6.1. It introduces the cluster-based permutation test over channels × time; L4.7 adds a frequency axis to exactly that machinery, and reads much more easily once the two-dimensional case is familiar.
  • L3.5 SNR, trial counts and design — a prerequisite of L6.3. The standardized measurement error and the trial-count curve are what L6.3’s power argument generalises.
  • L6.2 Mixed models and trial-level data — a prerequisite of L6.6; read it if your design has crossed random effects or trial-level covariates.
  • L6.4 Circularity and analytic flexibility — a prerequisite of L6.6, and directly relevant whenever a frequency band or a time window is chosen after looking at the map.
  • L6.5 Decoding as inference — a prerequisite of L6.6; optional unless you report a classifier.

Lessons in order

  1. L0.1 · What EEG measures ~40 min
  2. L0.2 · Electrodes, montages and the 10-20 system ~40 min
  3. L0.3 · Amplifiers, sampling and recording ~45 min
  4. L0.4 · Reading raw traces ~60 min
  5. L0.5 · The artifact atlas ~75 min
  6. L0.6 · Data hygiene and metadata ~45 min
  7. L1.1 · Sampling, Nyquist and aliasing ~40 min
  8. L1.2 · Time and frequency domains ~60 min
  9. L1.3 · Power spectral density ~60 min
  10. L1.4 · Windowing, leakage and zero-padding ~30 min
  11. L1.5 · Filters: FIR, IIR, and what they do to your data ~75 min
  12. L1.6 · Line noise ~40 min
  13. L1.7 · Aperiodic and periodic components ~60 min
  14. L2.1 · Loading data and BIDS ~45 min
  15. L2.2 · Channel locations and bad channels ~60 min
  16. L2.3 · Re-referencing ~50 min
  17. L2.4 · Filtering in practice ~45 min
  18. L2.5 · Artifact rejection strategies ~50 min
  19. L2.6 · EOG regression and ICA ~90 min
  20. L2.7 · ASR, SSP and alternatives ~45 min
  21. L2.8 · Pipeline order and reproducibility ~60 min
  22. L4.1 · Why time-frequency ~40 min
  23. L4.2 · STFT and Morlet wavelets ~75 min
  24. L4.3 · Multitaper and filter-Hilbert ~50 min
  25. L4.4 · Baseline normalization and ERD/ERS ~60 min
  26. L4.5 · Phase, ITC and cross-frequency coupling ~60 min
  27. L4.6 · Is it really an oscillation? ~60 min
  28. L4.7 · Statistics for time-frequency ~50 min
  29. C4 · Capstone — Mu/beta ERD ~300 min
  30. L5.1 · Volume conduction: the central problem ~45 min
  31. L5.2 · Sensor-space connectivity ~75 min
  32. L5.3 · Surface Laplacian and CSD ~45 min
  33. L6.1 · The multiple-comparisons landscape ~60 min
  34. L6.3 · Effect sizes, power and precision ~50 min
  35. L6.6 · Reporting standards ~40 min