iiinova — where your digital self keeps evolving

Reposted by Jitse S. Amelink

Nobel Prize @nobelprize.org · 05/10/2026

BREAKING NEWS The 2026 Nobel Prize in Physiology or Medicine has been awarded to Karl Deisseroth, Peter Hegemann and Georg Nagel “for their discoveries concerning light-gated ion channels and optogenetics.”

231059513

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

And we're off to the acknowledgements. #SNL2026 is a wrap! Great conference, learned a load and I'm going home with a packed brain!

020

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Putting a lollipop or lidocaine makes people read faster, but not faster in semantics - supercool finding! The connectivity shifts as well.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

A lot of data, but the nicest one was this slide:

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Last is Jacqueline Cummine from Alberta on the SMG. She says she has some provocative findings

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They show strong evidence for the role of the SMG for alexia and try to model that nicely integrating speech and reading models

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They used a classic reading artificial neural network architecture and lesioned it. It does a little bit worse than words. They lesioned the model and correlated it with imaging activation

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

He propses that the architecture for reading should be aligned with speech production, because our brains did not evolve for reading.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

He is followed by Peter Turkeltaub who discusses alexia in relation to primarily stroke.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

His data shows a strong role for the IPS and PCS, with overlap in most recorded neurons for reading and spelling

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Next up is Oscar Woolnough from UT Houston using intracranial recordings. He will focus on the IPS and PCS.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They followed this up with TMS (iTBS). General improvement in reading speed and connectivity was seen.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

You can dissociate phonological and semantic processing in dyslexia using TMS

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Higher activation in salience network in dyslexia and lower activation in the ventral visual network activation, as well as reduced between-network connectivity in dyslexia. Nice broader picture here!

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Next up is @sabrinaturker.bsky.social from Vienna. She focuses on reading & dyslexia and found hypoactivation in reading regions in dyslexia, with the difference getting bigger for complex words. Functional connectivity is also lower, functional connectivity is stronger on the dorsal route for words

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Potential implications include:

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

The third challenge is the lack of integration with models of speech

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

For reading, there's a dorsal (letter-by-letter) and ventral stream. Sachi Paul will focus on the supramarginal gyrus within the dorsal stream. She investigated reading recovery after stroke and found evidence for the SMG. The challenge is a plethora of tasks, and a lack of centrality in reading

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Last symposium of #SNL2026 on the role of the dorsal stream in reading. Let's go!

150

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

But still: very cool if this can be applied, and a potential shift towards network dynamics.

001

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

I missed a few critical slides. There are few methodological caveats: local values of the gradient shift if the embedding shifts. I am not sure how that fits with lesions, maybe the epilepsy field has dealt with this as well

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

The shift was stronger to the default mode network in post-stroke aphasia in the anterior LIFG. It related to better speech and worse writing. Really cool finding

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They did both lesion-based mapping and lesion-gradient mapping. The third gradient (control vs. DMN) shows the strongest association with post-stroke reorganisation.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

And last one Ramya Balakrishnan from York on functional connectivity gradients in post-stroke aphasia. Cool to see some gradient work here!

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Very cool data here, very nicely conducted study.

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They next looked into iconicity, so how symbolic the sign representation is (and naturally interpretable for non-signers). All verbs are encoded in the posterior MTC, with classifier verbs being the strongest

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Reading and sign are quite similar in brain activation. Verbs are more strongly encoded in the posterior MTC compared to nouns.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They embedded the semantics using word vectors as well as grammar and modelled features in the brain

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Third is Maria Zimmerman from Warsaw on Neural Encoding of Verbs in Natural Sign Language Narrative. They investigated brain activation in response to a 20-minute story from Edgar Allan Poe signed in Polish Sign Language

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They can somewhat significantly predict sentence structure and word decoding, with the signal happening around onset and the most relevant signal coming from all over the brain, but particularly the STG.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They tested priming of four different syntactic structures in experiment 1 and in experiment 2 they test sentence production in response to comics. They used classifiers with a held-out example.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

He uses a priming paradigm combined with stereo EEG (intracranial) in epilepsy patients.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Second is @adumbmoron.bsky.social from NYU with the title "Cross-modal decoding of syntactic frames from word comprehension to sentence production"

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They concludes an interactive and parallel model is a better fit with the findings

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

When words are presented together, they show a faster and more extensive activation and multiple processes seem to be going on

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They used MEG and feature-based meaning representation (from Binder et al., 2016) and applied time-resolved RSA

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Two competing models are the serial and interactive models:

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Experience of meaning starts from the temporal lobe and then gradually spreads across the brain.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Next up is the slide session, starting with Shaonan Wang from Hong Kong investigating word meaning using rapid parallel visual presentation, to remove the temporal dynamics.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Furthermore, Hamilton argues that TMS and tDCS should be applied at much higher doses (SAINT protocol in psychiatry) and that tDCS will be applied at home

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Future directions include: individualized stimulation sites, ML models to predict performance and use TMS to treat subacute aphasia and using tACS to target cortical oscillations

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

They then used network theory to understand TMS. Boundary Controllaby was associated with open-ended tasks and Modal Controllability was associated with closed ended tasks. Simulating a lesion in the LIFG linked back to the posterior inferior BA45.

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

There's some evidence for the role of plasticity in the context of TMS. Although I must say, that testing one SNP is not the gold standard for using genetic data. Pathway-based polygenic scores would be more reliable, and then it would require at least a few 100 participants.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

Neuromodulation works better for people who are worse off: which is the opposite of behavioural therapies.

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

TMS has a decent spatial resolution, tDCS doesn't, but sometimes messy can aid recovery. tDCS has been shown to have a significant effect (work from Fridrikson et al. and Tsapkini et al).

100

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

The possible explanation is the interhemispheric inhibition hypothesis. Possible caveats: anatomic specialization. Hamilton proposes instead it is a compensatory mechanisms.

110

Jitse S. Amelink @jsamelink.bsky.social · 02/10/2026

TMS stimulation of the posterior inferior BA45 also helped the most in improving performance in picture naming. And also helped improvement at 6 months follow-up.

100