iiinova — where your digital self keeps evolving

Benedikt Ehinger @benediktehinger.bsky.social · 24/07/2026

🧠 EEG / ERP researchers wanted! We are testing ways to visualize uncertainty in EEG topoplots ⏱ ~15min 💻 Best on Laptop, mobile possible 🎁 2 "Muse-2" headsets will be raffled among participants. Survey: www.soscisurvey.de/uncertain_to... #EEG #ERP #Neuroscience #DataVisualization

0117

Benedikt Ehinger @benediktehinger.bsky.social · 13/04/2026

Ha! That was fun. micropython is interesting I'll add the eye-movements later. It's going to be a demo for @unistuttgart.bsky.social Tag der Wissenschaft! #eyetracking #outreach

0152

Benedikt Ehinger @benediktehinger.bsky.social · 08/04/2026

LSLAutoBIDS: 1) integrates (LSL #EEG, behav, etc.) 2) manages (conversion to BIDS) 3) versions (datalad / git-annex) 4) publishes/archives (dataverse, private or public) No more dataloss & no stress in sharing data. Paper: doi.org/10.52294/001... Code: github.com/s-ccs/LSLAut...

160

Benedikt Ehinger @benediktehinger.bsky.social · 08/04/2026

When recording experimental data, have you ever: 1) overwritten data? 2) changed the experiment and forgot when? 3) had a student leave the lab leaving a mess of dataset? 4) noticed how everyone has their own way to transfer data? We published a solution: LSLAutoBIDS - open science by design. 🧵

12913

Benedikt Ehinger @benediktehinger.bsky.social · 09/12/2025

I'll probably need a week or so to fully digest this paper. Amazing work by Mohr, Geuzebroek and @spk3lly.bsky.social on disentangling contributions of V1,V2,V3 to ERPs+SSVEPs Beautiful illustrations throughout! doi.org/10.1162/imag...

372

Benedikt Ehinger @benediktehinger.bsky.social · 08/12/2025

Can we stop our #EEG experiments after 5 trials? Thought provoking paper: www.arxiv.org/abs/2511.23162 tldr: n=5 trials gave better cvR² than n=100% on average Kudos to the authors who went far beyond many other EEG papers in terms of additional tests & baselines (openreview.net/forum?id=c6L...)

1113

Benedikt Ehinger @benediktehinger.bsky.social · 23/09/2025

S-CCS Lab PhD Position 3+2 year 100% TVL-13 position in '26 - open topic on the intersection of combined EEG-EyeTracking, Statistical Methods, Cognitive Modelling, VR/Mobile EEG, Vision ... Apply via Max-Planck IMPRS-IS program until 2025-11-16 imprs.is.mpg.de Read: www.s-ccs.de/philosophy

EEG system photographed with backlight and glowing electrodes

01611

Benedikt Ehinger @benediktehinger.bsky.social · 22/08/2025

Actually, I do see some lateralisation in their publication

110

Benedikt Ehinger @benediktehinger.bsky.social · 29/06/2025

2/3 In the org-study we showed in several datasets, that deconv. removed EA signals - I'm happy about the comment showing that there seem to be at least some simulation parameter regimes were a naive deconvolution reduces EA, or potentially even blows up parameters due to collinearity

Top row, traditional ERP analysis in 4 datasets, bottom row after deconvolution, effects vanished after deconvolution.

100

Benedikt Ehinger @benediktehinger.bsky.social · 05/05/2025

🥳 New Unfold-Family documentation page! unfoldtoolbox.github.io/UnfoldDocs/m... A unified look, comparison to others, and some global perspective - check it out for #rERP & #EEG analyses! --- A good time to learn Unfold is this Friday: Free online Unfold workshop - www.s-ccs.de/workshop_unf... 🧠🧪

1115

Benedikt Ehinger @benediktehinger.bsky.social · 29/04/2025

Very impressive: Synchronous #eyetracking of dual-purkinje and AOSLO (for small saccades tested only) dual purkinje finds ~two times faster velocities - as far as I can see mainly because of lense wobble doi.org/10.1117/12.3... AOSLO has been around a bit, but this?! Wow! Ground truth data!

three graphs comparing velocities based on DPI and AOSLO tracking

020

Benedikt Ehinger @benediktehinger.bsky.social · 24/04/2025

trying my best... I do have internet, and I did try to restart My time/motivation to fix psychopybuilder is very limited as I mostly use opensesame

000

Benedikt Ehinger @benediktehinger.bsky.social · 15/04/2025

‼ Announcement: Online Unfold.jl workshop ‼ 📅 09.05.2025 💶 Free! 👉🏼 www.s-ccs.de/workshop_unf... ❓ rERPs, mass univariate models & deconvolution! #EEG #linearmodels #statistics @julialang.org #julia Organized with Romy Frömer @thechbh.bsky.social and the S-CCS lab @unistuttgart.bsky.social

0219

Benedikt Ehinger @benediktehinger.bsky.social · 10/04/2025

Animation of winsorized mean approaching the median CC-by me

020

Benedikt Ehinger @benediktehinger.bsky.social · 25/03/2025

🧵 4/5 A simulation needs only four basic ingredients: 1️⃣Experimental design 2️⃣Basis functions (components) 3️⃣Inter-onset distance distribution 4️⃣Noise Mix 🍸, add 🧂, and🪄 - enjoy!

already implemented noise types - add your own!

parameterization of Interonsetdistance distributions

plots of BOLD component parameterization

100

Benedikt Ehinger @benediktehinger.bsky.social · 25/03/2025

🧵 2/5 Some key features - Continuous & epoched EEG (via LMs/LMMs - also with overlap!) - Multi-subject & complex experimental designs - Multi-channel via forward models - Mix & match designs, components, noise... - Add your own (e.g., imbalanced designs, fMRI, pupil dilation...) - Extensive docs!

Simulated timeseries with two event types and three components

analysis results of ERP with continuous and categorical effects, and marginal effects plot of continuous effect

Multi-subject timeseries of 4 events with 20 subjects overlayed

multi-channel simulation with butterfly plot and two topoplots

100

Benedikt Ehinger @benediktehinger.bsky.social · 25/03/2025

🧵 1/5 Excited to share our latest @joss-openjournals.bsky.social 🧪 #paper: UnfoldSim.jl New @julialang.org package to simulate continuous event-based time series for #EEG & beyond! 📜 doi.org/10.21105/jos... 🛠 github.com/unfoldtoolbo... With @judithschepers.bsky.social, Luis Lips & Maanik Marathe

14215

Benedikt Ehinger @benediktehinger.bsky.social · 18/03/2025

Is this peak Burocracy? "Please provide a reason for using a bike" I'll admit, it was a - check notes - 7.82€ bike ride in the Netherlands. What should I answer: "because Netherlands", "Walking slow", "no elephant available"?

Formular mit Aufforderung: bitte eine Begründung für die Benutzung eines Fahrrades

5150

Benedikt Ehinger @benediktehinger.bsky.social · 20/01/2025

really like this interactive demo of short-lived #vision static.laszlokorte.de/persistence-... Looks like absolute noise until you move your cursor, then looks like noise again!

191

Benedikt Ehinger @benediktehinger.bsky.social · 25/11/2024

This is a _super_ cool preprint by Roman Kessler, Alexander Enge and Michael Skeide (none on bsky?) They show that preprocessing has little effect on whole-epoch decoding Very well done multiverse analysis! #EEG #decoding arxiv.org/abs/2410.14453

270

Benedikt Ehinger @benediktehinger.bsky.social · 20/11/2024

⚡Unfold.jl⚡ New optimizers & benchmarks. Up to 500x faster than the MatLab version. Fit a large overlap-corrected model with ~10k predictors and 128 channels in: ~10 seconds! A difference of 10s vs. 1.5h way-too-detailed-benchmarks: unfoldtoolbox.github.io/Unfold.jl/st... #EEG

084

Benedikt Ehinger @benediktehinger.bsky.social · 15/11/2024

"The modern regression toolkit" - I recently gave two talks (15' at #cuttingeegX and 50' in Gießen). You can find my slides on my blog, the video of cuttingeeg will hopefully land on youtube in the near future. #EEG #statistics #cognition #cogsci #psych 🧠🧪 benediktehinger.de/blog/science...

ERP with continuous effect, animated between adding temporal overlap and not

1308

Benedikt Ehinger @benediktehinger.bsky.social · 11/11/2024

I recently posted a demo how to combine plots in Julia with typst (latex is also possible). I now wrote a blogpost on it: benediktehinger.de/blog/science... #makie @makieplots.bsky.social #typst #visualization #plotting #latex #julialang #ERP #EEG

1134