Alakazam

An interactive MATLAB workbench for EEG and ERP analysis, made for learning the methods and for reusing the analysis.

Alakazam loads an EEG recording, applies processing steps to it one at a time, and keeps every result as a node in a tree: what was done, with which settings, and in what order. Nothing is a one-off. Every step can be looked at, changed, replayed on the next recording, saved as a template, or written out as a MATLAB script.

Read the manual (PDF) · Download a release · Example data

Made for education and for reuse

The focus of Alakazam is on education and on reusability in research.

For learning the methods

• See what every step does. Each step of an ERP analysis is a button with a dialog that names its settings, and each result is a plot one click away, so the effect of a filter, a reference or a rejection threshold is seen rather than taken on trust.
• Try the alternative beside the original. A second choice does not replace the first: it becomes a second branch of the tree, and the two results can be compared side by side.
• Follow the textbook in the application. The ERP CORE chapters of Steven Luck's open textbook, Applied Event-Related Potential Data Analysis, come as ready-made templates (library/templates/luck/), on the same data the book uses.
• Follow a published protocol. The workflow is the one Pütz, Span & Lorist (2025) set out step by step in STAR Protocols, from designing and recording an ERP task to preprocessing it and exporting the measures for statistical analysis (doi:10.1016/j.xpro.2025.103835).
• Understand why, not only how. The manual explains what each step computes, with which defaults and on what grounds, with the literature behind it. The same manual is behind the Help button.
• From clicking to programming. Export as Code turns a pipeline built in the interface into a readable MATLAB script, written as the EEGLAB calls it corresponds to wherever that is a faithful translation.

For reusable research

• A pipeline is recorded, not remembered. Drag a processed branch onto another recording, or use Apply to All Raw Files, and the same steps with the same settings run on every participant.
• Templates travel. A template (.alztemplate, plain JSON) carries a pipeline to another workspace, another study or another lab.
• Open data structures. Every dataset is an EEGLAB EEG structure, so it can be handed to EEGLAB, ERPLAB or FieldTrip unchanged. Raw files are never modified.
• Reports you can read and edit. The statistics are Quarto documents with R underneath, not a fixed printout, and the data-quality report states what each cleaning step did to each participant.
• Checked against published results. The ERP pipeline was compared step by step with ERPLAB on the textbook's data, and the frequency-tagging analysis with a published RIFT study; how close each comparison comes is written down in Docs/luck.md and Docs/dimigen.md.

What it does

• The ERP pipeline: filtering, re-referencing, channel repair, artefact rejection and correction (ICA with ICLabel, GEDAI), epoching, baseline correction, averaging, and amplitude, area and latency measures.

• Conditions in a few readable lines. Bins are described in a small language rather than a spreadsheet:

  % Targets count only when a correct response (201) follows in 200-1500 ms.
  epoch [-200,800] ms
  bin 1 "Target, related"   : 211|212 and next(201) within [200,1500] ms
  bin 2 "Target, unrelated" : 221|222 and next(201) within [200,1500] ms
  bin 3 "N400" = bin 2 - bin 1
  

• Statistics that fit the design: reports that choose the test the design calls for, from paired t-tests to linear mixed models, with Bayes factors and effect sizes, and cluster-based permutation tests over the scalp or the cortical surface.

• Data quality: trial loss per participant and condition, noise, the standardised measurement error and the dependability of each score.

• Frequency tagging: power, signal-to-noise ratio, phase-locking and coherence for steady-state and rapid invisible frequency tagging designs.

• Overlapping responses: regression-based deconvolution (Unfold) for free viewing, reading and fast presentation, and co-registration with an eye tracker (EYE-EEG).

• Source estimation on a template cortex (FieldTrip), with the limits of such an estimate stated beside it.

• Long recordings stay fast: a purpose-built viewer scrolls hours of multichannel data smoothly without hiding a single artefact.

Time-frequency power per condition.	Scalp distributions over time.	Overlap-corrected ERPs by deconvolution.

Getting started

You need MATLAB (a recent release, with uifigure support), the Signal Processing Toolbox and the Statistics and Machine Learning Toolbox. EEGLAB is set up by Alakazam on first run; other toolboxes (ICLabel, FieldTrip, Unfold, EYE-EEG, GEDAI) are installed the first time a step needs them, after asking. For the statistical reports you also need Quarto and R (Quarto comes with RStudio).

Clone the repository, or download a packaged release from the releases page, then in MATLAB, from the Alakazam folder:

startAlakazam

Point the workspace at a folder of recordings (EEGLAB .set, BrainVision .vhdr or ERPLAB .erp), or start with the openly available example data listed in DATA.md. The manual's Getting started chapter walks through a first analysis, from a raw recording to numbers you can analyse.

Documentation

Document	For
The manual (PDF, source)	Researchers and students: every step, report and setting, with its background. Also behind the Help button.
DATA.md	Where to get the example datasets the manual and templates use.
library/	Ready-made templates, bin scripts and measurement windows.
library/templates/luck/	The textbook chapters as templates.
dependencies.md	The toolkits Alakazam uses, their versions and licences.
DEVELOPER.md	Contributors: the architecture, adding a transformation, the tests.
CHANGELOG.md	What changed in each release.

Citing

When you publish an analysis made with Alakazam, cite the methods it used: EEGLAB for the data structure and the functions it wraps, and the specific methods named in the manual's section on each step (ICLabel, Unfold, FieldTrip and so on). Export as Code writes the complete pipeline with the settings actually used, which is the most precise methods description there is. The About box lists the papers and the version you ran.

Licence

Alakazam is free software under the GNU General Public License, version 3 (see LICENSE). The toolboxes it installs on first use keep their own licences, listed in dependencies.md; note that GEDAI's allows noncommercial research use only.

Source, issues and contributions: github.com/markspan/Alakazam.