Samuel A. Nastase

This study investigates how internal and external default attention modes interact by measuring brain activity during naturalistic video viewing and internal focus tasks, finding overlapping fronto-parietal networks involved in transitions between these states.

This study uses fMRI and HMM to show that narrative context modulates brain state dynamics and behavioral responses during story comprehension.

The paper investigates neural mechanisms underlying self and other mental state reasoning using fMRI and multi-voxel pattern analysis, identifying distinct brain regions involved in self-specific and general mental state processing.

This study uses fMRI hyperscanning to show overlapping neural networks for speech production and comprehension during real-time conversations, extending beyond the language network into social cognition areas.

The paper introduces a dataset of ECoG recordings from nine participants listening to a podcast, along with linguistic features and tutorials for analysis.

This study shows that cortical language areas are connected through a hierarchy of linguistic features, with higher-order regions using more abstract language features.

The paper evaluates how action, agent, and scene features predict cortical activity during natural vision and finds action features capture the most unique variance across the cortex.

The paper shows that behavioral models of action purpose and social content better predict neural representational geometry than other models in the action observation network.

This study shows that speakers' brains generate information-rich words through extra neural computation, evidenced by stronger neural activity before articulation, and that LLMs similarly require deeper computation for improbable words.

The paper shows that linguistic embedding spaces can capture neural alignment during conversations, revealing shared meaning between speakers and listeners.

This paper analyzes the functional specialization of transformer-based language models and finds similarities with the human brain's cortical language network.

The paper shows that larger language models better capture linguistic structure and predict neural activity, with performance peaking in earlier layers as model size increases.

The paper shows that aligning brains into a shared space improves LLM-based encoding model performance and denoising of individual brain responses.

This study examines how neural synchronization between speakers and listeners during storytelling reflects a 'herding effect' where listeners follow the speaker's brain activity patterns, enhancing narrative engagement.

The paper estimates individualized cortical functional topographies from naturalistic movies across different participants, scanners, and institutes with high fidelity.

The paper introduces the Individualized Neural Tuning (INT) model, which provides a fine-grained, reliable, and generalizable measure of brain functional organization across individuals.

This study compares representational geometries of deep convolutional neural networks, human cognition, and brain activity during naturalistic face processing and finds that late-intermediate layers better match cognitive and neural geometries than late layers.

The paper uses a deep speech-to-text model to predict neural responses during spontaneous conversations, revealing a distributed cortical hierarchy for speech and language processing.

The paper investigates how the brain updates memory representations of naturalistic events when new information changes understanding of the past, showing dynamic neural changes in the default mode network.

The paper uses fMRI to show a fixed sequence of response lags between brain networks during narrative comprehension, consistent across stories and supported by a computational model.

This study finds a correspondence between the layered structure of deep language models and the temporal structure of neural activity during language processing in the human brain.

This study uses real-time fMRI neurofeedback to influence interpretations of a naturalistic narrative, finding that higher decoding accuracy correlates with successful interpretation shifts.

The paper shows that brain embeddings from the IFG share geometric structure with DLM contextual embeddings, enabling zero-shot encoding and decoding of unseen words.

The paper uses fMRI and hidden Markov models to show that brain regions involved in narrative event segmentation also represent high-level event structure in music.

The paper introduces a new spatial covariance function for fMRI data called the brain kernel, which captures non-stationary spatial correlations and improves decoding and factor analysis.

This study shows that neural coupling between teachers and students during lectures correlates with learning outcomes, using fMRI and intersubject correlation.

This paper introduces hybrid hyperalignment, a method that aligns both neural response and functional connectivity data across brains using a single model, showing improved alignment for both types of data compared to previous methods.

Sharing happy stories increases interpersonal closeness by enhancing brain synchronization in specific brain regions, as shown through EEG and behavioral assessments.

The paper introduces a benchmark dataset of fMRI data from human subjects listening to naturalistic stories for evaluating language and narrative comprehension models.

In nine epilepsy patients, cortical activity carries next-word information up to 800 ms before onset, tracks GPT-2 surprise after onset, and fits GPT-2 contextual embeddings better than static ones.