Large language models (LLMs) capture long-range contextual structure in natural language and have recently been shown to align with the human brain's contextualized linguistic encoding. This makes them a promising computational probe for studying how context-dependent linguistic information is represented during natural speech perception. Speech perception often occurs in multi-talker environments, where attention must dynamically select among competing streams, yet how contextual information from attended and unattended speech is neurally encoded remains underexplored. Here, we investigate how auditory attention modulates neural tracking of context-dependent linguistic representations using electrocorticography (ECoG) and stereoelectroencephalography (sEEG) recordings from three epilepsy patients engaged in a two-conversation "cocktail party" paradigm. To model neural responses to attended and unattended speech streams, we used contextual word embeddings generated by large language models. We find that LLM-derived features reliably predict brain activity for the attended stream and that contextual information from the unattended stream also contributes to neural prediction. Importantly, these contributions extend beyond low-level acoustic features and shallow syntactic information, and depend on the surrounding linguistic context. Moreover, neural tracking of the unattended stream reflects shorter-range contextual integration than that of the attended stream. Together, these findings indicate that neural responses to speech reflect context-dependent linguistic representations from multiple concurrent speech streams, with attention modulating the depth and timescale of contextual integration. Our results highlight the utility of LLMs for probing higher-level linguistic representations in complex, naturalistic listening environments.

Effects of CHD2 depletion on the growth of H3K27M DMG xenografts.

CHD2 regulates neuron-induced proliferation of H3.1K27M DMG cells.

FOSL1 colocalizes with CHD2 to regulate cell viability and gene expression.

Validation of hits from CRISPR/Cas9 screens.

CHD2 and FOSL1 regulate H3K27me3 and H3K27ac in H3.1K27M DMG cells.

Analysis for DEGs of CHD2-depleted H3.1K27M DMG cells.

CHD2 binds to the gene regulatory elements of axon-guidance and synaptic genes.

Calcium imaging in brain slices with SU-DIPG36 xenografts in hippocampal CA1.

Analysis for shared DEGs between CHD2 and FOSL1 depletion in H3.1K27M DMG cells.

OBJECTIVE:There is a need to develop a comprehensive, formalized, and globally applicable epilepsy surgery curriculum in order to help standardize the quality of epilepsy surgery practice in the world. METHODS:The Epilepsy Surgery Educational Task Force of the ILAE developed a competency-based epilepsy surgery educational curriculum comprised of four domains: Diagnosis, Counseling, Pre-surgical Work-up, and Surgical Techniques. To evaluate this educational curriculum, a survey questionnaire consisting of 11 items was sent out for feedback to ILAE and IESS members. The survey questionnaire asked respondents to rate the degree of importance of each competency on a five-point scale that ranged from "extremely important" to "not at all important" and provided space for free text comments. All competencies that were rated as "slightly important" or "somewhat important" by more than 10% of responders, as well as all free text feedback comments, were fully reviewed, and the final curriculum was adjusted accordingly. RESULTS:One hundred and twenty-two responses (n = 122) were submitted by the ILAE and IESS communities. Seventy-one percent (71.9%) of the responses came from neurosurgeons of varied experience and training levels. Ninety-four percent (94.4%) of responders stated that they "support a curriculum to determine the competency of neurosurgeons participating in the care of people with epilepsy in my country," and 94.3% confirmed that they "would be interested in taking or would recommend an ILAE course based on these competencies, that would provide an ILAE certificate of completion." Relevant changes within the curriculum were made based on feedback. SIGNIFICANCE:The newly developed ILAE comprehensive epilepsy surgery educational curriculum has a high level of support from neurosurgeons and neurologists around the world and shows potential for broad acceptance and implementation.

Supplementary Figures S1-S6 and figure legends.

We have generated a single-cell RNA sequencing atlas of peripheral blood and ventricular CSF in idiopathic normal pressure hydrocephalus (iNPH) patients totaling 140,207 single-cell transcriptomes. We found proinflammatory alterations in peripheral blood and CSF monocytes in iNPH patients with lower baseline cognitive function. We also identified CSF cell populations likely representing periventricular sloughing of degenerating neuroglial cells. Our findings suggest possible immune dysregulation in the blood and CSF of iNPH patients.

Innovations in electrophysiological recordings and computational analytic techniques enable high-resolution analysis of neural traveling waves. Here, we present a protocol for the detection and analysis of traveling waves from multi-day microelectrode array human electrophysiological recordings through a multi-linear regression statistical approach using point estimator data. We describe steps for traveling wave detection, feature characterization, and propagation pattern analysis. This protocol may improve our understanding of the coordination of neurons during non-oscillatory neural dynamics.For complete details on the use and execution of this protocol, please refer to Smith et al.1

Auditory foundation models, including auditory large language models (LLMs), process all sound inputs equally, independent of listener perception. However, human auditory perception is inherently selective: listeners focus on specific speakers while ignoring others in complex auditory scenes. Existing models do not incorporate this selectivity, limiting their ability to generate perception-aligned responses. To address this, we introduce intention-informed auditory scene understanding (II-ASU) and present Auditory Attention-Driven LLM (AAD-LLM), a prototype system that integrates brain signals to infer listener attention. AAD-LLM extends an auditory LLM by incorporating intracranial electroencephalography (iEEG) recordings to decode which speaker a listener is attending to and refine responses accordingly. The model first predicts the attended speaker from neural activity, then conditions response generation on this inferred attentional state. We evaluate AAD-LLM on speaker description, speech transcription and extraction, and question answering in multitalker scenarios, with both objective and subjective ratings showing improved alignment with listener intention. By taking a first step toward intention-aware auditory AI, this work explores a new paradigm where listener perception informs machine listening, paving the way for future listener-centered auditory systems. Demo available.

Human brain tissue studies have historically used a range of metrics to assess RNA quality. However, few large-scale cross-comparisons of pre-sequencing quality metrics with RNA-seq quality have been published. Here, we analyze how well metrics gathered before RNA sequencing (post-mortem interval (PMI) and RNA integrity number RIN) relate to analyses of RNA quality after sequencing (Percent of counts in Top Ten genes (PTT), 5' bias, and 3' bias) as well as with individual gene counts across the transcriptome. We conduct this analysis across four different human cortical brain tissue collections sequenced with varying library preparation protocols. PMI and RIN have a low inverse correlation, and both PMI and RIN show consistent and opposing correlations with PTT. Unlike PMI, RIN shows strong consistent correlations with measurements of 3' and 5' bias, and RIN also correlates with 3,933 genes across datasets, in comparison to 138 genes for PMI. Neuronal and immune response genes correlate positively and negatively with RIN respectively, suggesting that different gene sets have divergent relationships with RIN in brain tissue. In summary, these analyses suggest that conventional metrics of RNA quality have varying degrees of value, and that PMI has an overall minimal but reproducible effect on RNA quality.

Approximately 60-80% of all glioma patients experience seizures, and a significant number of these patients are unable to achieve seizure freedom with conventional anti-epileptic treatment alone. One suggested mechanism for these tumor-associated seizures (TAS) is through disrupted presynaptic and postsynaptic GABAergic signaling. Postsynaptic GABA response is determined by neuronal intracellular chloride state, largely regulated by the KCC2 chloride transporter, which is significantly downregulated in the peritumoral region. Recent studies in our group using our diffusely infiltrating glioma mouse model have implicated glioma-induced mTOR activation as a mechanism underlying local neuronal hyperexcitability and GABAergic dysfunction. Using this model, we hypothesize that neuronal chloride regulation in the tumor region is mTOR-dependent. We show via gene set enrichment that KCC2 is downregulated in excitatory neurons in the peritumoral region, and is recovered by mTOR inhibition, supporting this mechanistic link. We subsequently performed whole-cell patch-clamp electrophysiology on peritumoral excitatory neurons, demonstrating these neurons exhibit a depolarized resting membrane potential (RMP) and GABA reversal potential. Importantly, inhibiting mTOR signaling restores a hyperpolarized RMP and GABA reversal potential, normalizing these glioma-induced effects. We are translating these findings to surgically excised human peritumoral cortical tissue from patients undergoing resection for high grade glioma. Acute slices are treated ex vivo, and electrophysiologic recordings are performed via microelectrode array. Immunohistochemical (IHC) analysis after 6 hours of treatment with an mTOR inhibitor shows a significant reduction in neuronal mTOR activation and an increase in KCC2 expression. We then correlate these IHC findings to local field potential, multi-unit activity, and single unit neuronal activity isolated from electrophysiologic multi-electrode array recordings of tissue under different treatment conditions. Our findings reveal a mechanistic link between glioma-induced alterations in neuronal mTOR signaling and neuronal inhibitory dysfunction in both mouse and human glioma and point the way toward therapeutic strategies for treating pharmacoresistant glioma-associated epilepsy.

Sound structures such as phonemes and words have highly variable durations. Therefore, there is a fundamental difference between integrating across absolute time (for example, 100 ms) versus sound structure (for example, phonemes). Auditory and cognitive models have traditionally cast neural integration in terms of time and structure, respectively, but the extent to which cortical computations reflect time or structure remains unknown. Here, to answer this question, we rescaled the duration of all speech structures using time stretching and compression and measured integration windows in the human auditory cortex using a new experimental and computational method applied to spatiotemporally precise intracranial recordings. We observed slightly longer integration windows for stretched speech, but this lengthening was very small (~5%) relative to the change in structure durations, even in non-primary regions strongly implicated in speech-specific processing. These findings demonstrate that time-yoked computations dominate throughout the human auditory cortex, placing important constraints on neurocomputational models of structure processing.