Faculty Bibliography

Sentence production is the uniquely human ability to transform complex thoughts into strings of words. Despite the importance of this process, language production research has primarily focused on single words. It remains an untested assumption that insights from this literature generalize to more naturalistic utterances like sentences. Here, we investigate this using high-resolution neurosurgical recordings (ECoG) and an overt production experiment where patients produce six words in isolation (picture naming) and in sentences (scene description). We trained machine learning models to identify the unique brain activity pattern for each word during picture naming, and used these patterns to decode which words patients were processing while they produced sentences. Our findings reveal that words share cortical representations across tasks. In sensorimotor cortex, words were consistently activated in the order in which they were said in the sentence. However, in inferior and middle frontal gyri (IFG and MFG), the order in which words were processed depended on the syntactic structure of the sentence. This dynamic interplay between sentence structure and word processing reveals that sentence production is not simply a sequence of single word production tasks, and highlights a regional division of labor within the language network. Finally, we argue that the dynamics of word processing in prefrontal cortex may impose a subtle pressure on language evolution, explaining why nearly all the world's languages position subjects before objects.

OBJECTIVES/OBJECTIVE:Late-onset epilepsy has the highest incidence among all age groups affected by epilepsy and often occurs in the absence of known clinical risk factors such as stroke and dementia. There is increasing evidence that brain changes contributing to epileptogenesis likely start years before disease onset, and we aim to relate cognitive and imaging correlates of subclinical brain injury to incident late-onset epilepsy in a large, community-based cohort. METHODS:We studied Offspring Cohort of the Framingham Heart Study participants 45 years or older, who were free of prevalent stroke, dementia, or epilepsy, and had neuropsychological (NP) evaluation and brain magnetic resonance imaging (MRI). Cognitive measures included Visual Reproduction Delayed Recall, Logical Memory Delayed Recall, Similarities, Trail Making Test B minus A (TrTB-TrTA; attention and executive function), and a global measure of cognition derived from principal component analysis. MRI measures included total cerebral brain volume, cortical gray matter volume (CGMV), white matter hyperintensity volume (WMHV), and hippocampal volume. Incident epilepsy was identified through a review of administrative data and medical records. Cox proportional hazards regression models were used for the analyses. All analyses were adjusted for age, sex, and educational level (cognition only). RESULTS:Among participants who underwent NP testing (n = 2349, 45.81% male), 31 incident epilepsy cases were identified during follow-up. Better performance on the TrTB-TrTA was associated with a lower risk of developing epilepsy (hazard ratio [HR] .25, 95% confidence interval [CI] .08-.73; p = .011). In the subgroup of participants with MRI (n = 2056, 46.01% male), 27 developed epilepsy. Higher WMHV was associated with higher epilepsy risk (HR 1.5, 95%CI 1.01-2.20; p = .042), but higher CGMV (HR .73, 95% CI .57-.93; p = .001) was associated with lower incidence of epilepsy. SIGNIFICANCE/CONCLUSIONS:Better performance on the (TrTB-TrTA), a measure of executive function and attention, and higher cortical volumes are associated with lower risk of developing epilepsy. Conversely, higher WMHV, a measure of occult vascular injury, increases the risk. Our study shows that non-invasive tests performed in mid-life may help identify people at risk for developing epilepsy later in life.

OBJECTIVE/UNASSIGNED:This study details the design, efficacy, and usability of a novel wearable, wireless electroencephalography (EEG) sensor designed for extended-duration clinical monitoring in any environment. METHODS/UNASSIGNED:Simultaneous EEG recordings from REMI sensors and a conventional scalp-EEG recording system were conducted across two cohorts: 1) participants undergoing routine epilepsy seizure monitoring and 2) healthy volunteers performing tasks to induce common EEG artifacts. Comparative time and spectral-based analyses were conducted between the recording modalities. Sensor usability was also evaluated. RESULTS/UNASSIGNED:The temporal dynamics and signal morphology of artifacts and electrographic seizures were visually similar between the REMI sensor and conventional scalp-EEG. Additionally, spectral correlation between the two systems was high across all event types, ranging from 0.86 to 0.94. Patient-reported acceptance was also strong, with 69% of participants rating the sensors as comfortable to wear. CONCLUSIONS/UNASSIGNED:The REMI sensor showed strong agreement with conventional scalp-EEG in the signal characteristics of physiological artifacts and electrographic seizures. The positive comfort feedback further supports the REMI sensors' everyday utility. SIGNIFICANCE/UNASSIGNED:Although limited in electrode coverage compared to conventional scalp-EEG recording systems, the REMI sensor records comparable high-fidelity EEG data in both time and spectral domains. REMI sensor's recording quality and wearability facilitate extended-duration monitoring in everyday environments.

When we vocalize, our brain distinguishes self-generated sounds from external ones. A corollary discharge signal supports this function in animals; however, in humans, its exact origin and temporal dynamics remain unknown. We report electrocorticographic recordings in neurosurgical patients and a connectivity analysis framework based on Granger causality that reveals major neural communications. We find a reproducible source for corollary discharge across multiple speech production paradigms localized to the ventral speech motor cortex before speech articulation. The uncovered discharge predicts the degree of auditory cortex suppression during speech, its well-documented consequence. These results reveal the human corollary discharge source and timing with far-reaching implication for speech motor-control as well as auditory hallucinations in human psychosis.

Epilepsy affects approximately 1% of the world population. Patients have recurrent seizures, increased physical and psychiatric comorbidities, and higher mortality rate than the general population. Over the last 40 years, research has resulted in 20 new antiseizure medications (ASMs) approved between 1990 and 2018. In spite of this, up to one-third of patients (~ 1 million patients in the USA) have drug-resistant epilepsy (DRE), with little change between 1982 and 2018, a period of intense new ASM development. A minority of patients with DRE may benefit from surgical treatment, but this specialized care remains challenging to scale. Therefore, the greatest hope for breakthroughs for patients with DRE is in pharmacologic therapies. Recently, several advances promise to change the outcomes for patients with DRE. Cenobamate, a drug with dual mechanisms of modulating sodium channel currents and GABA-A receptors, achieves 90-100% seizure reduction in 25-33% of patients with focal DRE, a response not observed with other ASMs. Fenfluramine, a serotonin-acting drug, dramatically reduces the frequency of convulsive seizures in Dravet syndrome, a devastating developmental epileptic encephalopathy with severe DRE. Both drugs reduce mortality. In addition, the possibility of DRE prevention was recently raised in patients with tuberous sclerosis complex, a relatively common genetic form of epilepsy. A paradigm shift is emerging in the treatment of epilepsy. Seizure freedom has become attainable in a significant proportion of patients with focal DRE, and dramatic seizure reduction has been achieved in a developmental encephalopathy. Coupled with a rich pipeline of new compounds under clinical development, the long sought-after breakthrough in the treatment of epilepsy may finally be in sight.

Recent research has used large language models (LLMs) to study the neural basis of naturalistic language processing in the human brain. LLMs have rapidly grown in complexity, leading to improved language processing capabilities. However, neuroscience researchers haven't kept up with the quick progress in LLM development. Here, we utilized several families of transformer-based LLMs to investigate the relationship between model size and their ability to capture linguistic information in the human brain. Crucially, a subset of LLMs were trained on a fixed training set, enabling us to dissociate model size from architecture and training set size. We used electrocorticography (ECoG) to measure neural activity in epilepsy patients while they listened to a 30-minute naturalistic audio story. We fit electrode-wise encoding models using contextual embeddings extracted from each hidden layer of the LLMs to predict word-level neural signals. In line with prior work, we found that larger LLMs better capture the structure of natural language and better predict neural activity. We also found a log-linear relationship where the encoding performance peaks in relatively earlier layers as model size increases. We also observed variations in the best-performing layer across different brain regions, corresponding to an organized language processing hierarchy.

OBJECTIVE/UNASSIGNED:This study investigates speech decoding from neural signals captured by intracranial electrodes. Most prior works can only work with electrodes on a 2D grid (i.e., Electrocorticographic or ECoG array) and data from a single patient. We aim to design a deep-learning model architecture that can accommodate both surface (ECoG) and depth (stereotactic EEG or sEEG) electrodes. The architecture should allow training on data from multiple participants with large variability in electrode placements and the trained model should perform well on participants unseen during training. APPROACH/UNASSIGNED:We propose a novel transformer-based model architecture named SwinTW that can work with arbitrarily positioned electrodes, by leveraging their 3D locations on the cortex rather than their positions on a 2D grid. We train both subject-specific models using data from a single participant as well as multi-patient models exploiting data from multiple participants. MAIN RESULTS/UNASSIGNED:The subject-specific models using only low-density 8x8 ECoG data achieved high decoding Pearson Correlation Coefficient with ground truth spectrogram (PCC=0.817), over N=43 participants, outperforming our prior convolutional ResNet model and the 3D Swin transformer model. Incorporating additional strip, depth, and grid electrodes available in each participant (N=39) led to further improvement (PCC=0.838). For participants with only sEEG electrodes (N=9), subject-specific models still enjoy comparable performance with an average PCC=0.798. The multi-subject models achieved high performance on unseen participants, with an average PCC=0.765 in leave-one-out cross-validation. SIGNIFICANCE/UNASSIGNED:The proposed SwinTW decoder enables future speech neuroprostheses to utilize any electrode placement that is clinically optimal or feasible for a particular participant, including using only depth electrodes, which are more routinely implanted in chronic neurosurgical procedures. Importantly, the generalizability of the multi-patient models suggests the exciting possibility of developing speech neuroprostheses for people with speech disability without relying on their own neural data for training, which is not always feasible.

Effective communication hinges on a mutual understanding of word meaning in different contexts. We recorded brain activity using electrocorticography during spontaneous, face-to-face conversations in five pairs of epilepsy patients. We developed a model-based coupling framework that aligns brain activity in both speaker and listener to a shared embedding space from a large language model (LLM). The context-sensitive LLM embeddings allow us to track the exchange of linguistic information, word by word, from one brain to another in natural conversations. Linguistic content emerges in the speaker's brain before word articulation and rapidly re-emerges in the listener's brain after word articulation. The contextual embeddings better capture word-by-word neural alignment between speaker and listener than syntactic and articulatory models. Our findings indicate that the contextual embeddings learned by LLMs can serve as an explicit numerical model of the shared, context-rich meaning space humans use to communicate their thoughts to one another.