Revised-TLDR:
Using the hidden multivariate pattern method we show how we capture 5 (!) recurrent EEG topographies in participants' decisions.

We show how these events parse the single-trial reaction time into:
- sensory processes
- attention orientation
- evidence accumulation
- motor execution

Next week on Thursday at 2pm CET I will present a workshop at #practicalMEEG2025 @cuttingeeg.bsky.social on our Hidden multivariate pattern method (hmp.readthedocs.io/en/latest/we...).

In this online workshop I'll show participants how they can estimate single-trial timing of cognitive events.

Preached for our #ERP lord and saviour HMP at #ICON2025
hmp.readthedocs.io

Wonderful time also at the decision in the EEG symposium with very interesting talks and panel discussion!

And if you want to hear about single-trial ERP modelling also come by my talk in the symposium "The Value of Methods of Analysis in Electrophysiology" (SY45, Tuesday 16:30) organized by Monica Wagner

One PhD and two post-doc positions are open in the group of Prof. Athina Tzovara in Bern to work on the neural mechanisms (#iEEG, #EEG) that support processing of auditory information.
Great scientific context, as well as landscape and city!

The time between these events align extremely well with the Piéron and Fechner laws. Even the electrode activity of the isolated topogaphies follows these laws. Even more crazy is that the buildup of the last decision event very clearly aligns with decision-making predictions.

In this opposing laws dataset, HMP decomposed the EEG signal into three clear topographies respectively associated with early visual, figure-ground separation and decision processes.

This accidentally generated two well documented but opposing effects: Pièron's and Fechner's laws. Increasing the overall contrast but keeping the difference constant makes you perceive the stimuli faster but also makes the judgment harder

The "clever" task design in the paper came through serendipity... To help a master student in need of a project, I quickly designed a contrast judgment experiment where we manipulated the overall stimulus contrast, but we kept the difference constant.

Started a new position (actually already 2 weeks ago 😁) at the Université de Lausanne with Audrey Bürki in the Statistics in Psychology group. Be ready for some #Stats, #chronometry and #electrophysiology

It was 3 wonderful years in NL at Groningen and Utrecht universities, thanks for everything!

Example screen for jupyter lab showing the interface with several windows opened, one text editor, one bash terminal, one notebook and a python console.

Personally it's everything that I need (not having copilot is a perk to me :))

Zotero now gives you the Figure as miniature if you hoover over a reference?? 🤯

Then allow me to raise, without contours and a better scale (-2 to 2 microV)

OK, both plots are left - right hemisphere EEG for a right index key-press

Left is when using single trial time of the key-press (RT)
Right same but using the single-trial "LRP" time

Outside of the difference in scale isn't the right one also closer to what you expect given motor homonculus?

In the paper we show three example applications on publicly available datasets:
- the P3
- the N2pc
- and a #decision-making dataset
We show how the exploratory decision-making dataset contains the same component as the N2pc and the P3 dataset.

This method basically turns any ERP into an estimated collection of small half-sines with different time and spatial representations.
Example where we only inserted half-sines in the time-series of each epoch (left plot), plot the average ERP (top) and the HMP-event centered ERP (bottom)

HMP assumes that single trials events in the E/MEG signal are 1) coherent transient topographies recurring across trials 2) in sequence after stimulus onset up to the response (e.g. visual before motor) and 3) that the time distances between events (loosely) follow a probability distribution.