Thank you to the authors @AIatMeta and @FondARothschild for their work, and we look forward to following its progress!.cc: @EvansonLinnea @JeanRemiKing.

❔3️⃣: For wav2vec, did you look at encoding and decoding at other regions of the architecture to see if encoding improves linearly with depth in the architecture?.

❔2️⃣: What about the architecture of the Llama model do you think allows it accurately encode word information in the first layer of the model?.

🤍3️⃣: Using weights from wav2vec and Llama to perform encoding/decoding was novel and paves the way for further research comparing LLMs to human cognition.

🤍3️⃣: Using weights from wav2vec and Llama to perform encoding/decoding was novel and paves the way for further research comparing LLMs to human cognition.

🤍2️⃣: Their demonstration of changes in language function (viz. being more distributed with age) was phenomenal.

🤍1️⃣: They collected one of the most fascinating datasets we have ever seen. The number of patients in the dataset across all ranges of neurodevelopment is truly a feat to behold.

Last week, @Daniel_Sexton16 (7th year Neurosurgery Resident at Duke University) presented @EvansonLinnea's new paper on comparing the development of language from childhood to adulthood and how it relates to LLMs. This 🧵 explores our thoughts (🤍 & ❔).

Thank you to the authors @NorthoffL, @uOttawa, and Heinrich Heine University Düsseldorf for your work, and we look forward to following its progress!. cc: Yasir Çatal, Kaan Keskin, @AngelikaWolman, Philipp Klar, David Smith, Georg Northoff.

❔3️⃣: Does the tradeoff between resting-state INT variability and rest-task modulation develop during early brain maturation? Understanding its developmental trajectory could reveal critical periods for interventions targeting temporal imbalances.

❔2️⃣: Could abnormal INT flexibility serve as a predictive biomarker for cognitive disorders marked by inhibited temporal processing and/or behavioral adaptability?.

❔1️⃣: To what extent do interpersonal differences in INT flexibility reflect individual cognitive characteristics? Do these patterns remain stable over time, or does INT flexibility change across the lifespan?.

🤍3️⃣: Computational modeling identified an optimal range of recurrent connection strength for INT flexibility. Insufficient connections prevent modulation, while excessive connectivity leads to rigidity, underscoring the need for balanced circuits in adaptive temporal processing.

🤍2️⃣: The inverse relationship between resting-state INT variability and rest-task modulation reveals a potential constraint mechanism. Regions with higher rest variability show reduced task-evoked changes, suggesting that intrinsic temporal dynamics may set response boundaries.

🤍1️⃣: This study demonstrates that intrinsic neural timescales accurately predict behavioral states with remarkable precision (86% accuracy in mice, 75% in humans), revealing that temporal integration properties contain robust state-specific signatures across species.

Last week, Nicole Liddle (@nicoleliddlee, CRS, Sr.) presented Çatal and colleagues’ 2024 paper on the flexibility of intrinsic neural timescales during distinct behavioral states. This 🧵 explores our thoughts (🤍 & ❔) .

RT @CoganLab: The Cogan Lab is looking for a new clinical research specialist/research assistant! If you are interested in speech, language….

The Cogan Lab is looking for a new clinical research specialist/research assistant! If you are interested in speech, language, and BCI and want to work with patients with implanted with intracranial electrodes, reach out!.

RT @Dukeneurosurg: #AANS2025 @AANSNeuro #DUKEatAANS

Thank you to the authors @CedarsSinai for your work, and we look forward to following its progress!. cc: @UeliRutishauser.@AMamelak.@kaminski_janek,.@singleunitgps.@MichaelKyzar00, Umais Khan, Chrystal Reed, Yousef Salimpour, Taufik Valiante, William Anderson.