We explore how humans process written language and delve into potential similarities to large language models (LLM). Current LLMs use tokenization, a process that breaks text into discrete semantic units called tokens and processes them sequentially. We test whether human language processing operates through analogous discrete mechanisms, where fixation points during reading are semantic "tokens" that are processed in a neural hyperspace similar to LLMs, as we read from one to the next in discrete steps. If reading operates this way, we hypothesize that saccade (eye movement) duration should remain constant as the distance between words changes, and should not be affected by word spacing. This tests whether our brains read in "real space," where the time it takes for our eyes to move is directly proportional to the distance traveled, or in "hyperspace," where each saccade takes roughly a constant time, regardless of distance. We presented 150 sentences in 4 font sizes, manipulating the distances between “tokens,” or fixation points while recording eye movements. Results revealed consistent saccade durations across all font sizes, with only minimal increases observed at the largest saccade distances. This pattern supports a hyperspace processing model where linguistic decoding occurs through discrete temporal units rather than continuous spatial scanning. The findings align with recent eye-tracking research showing weak correlation between saccade size and fixation time.

Over the past few decades neuroimaging technologies have rapidly evolved to provide more accessible, non-invasive methods of neurological assessment, leading to earlier, more efficient medical intervention. The recent launching of Fast Neurite Tracer (FNT) software presents a simplified method of neuronal reconstruction that has hopeful implications for both research and diagnostic efforts. UCLA’s Brain Research and Artificial Intelligence Nexus (BRAIN) is a cross-disciplinary laboratory focused on developing the first cell type map of the mouse brain. This effort integrates 3D imagery produced by neural tracer injections with FNT connectivity reconstructions . Inferences and comparisons can thus be made across connectivity maps, which are uploaded into the Allen Reference Atlas for professional and public viewing. This presentation will dive into the process of hand tracing neurons, identifying components of basic morphology and regional circuitry. Creating a cell type map of the mouse brain is the first step in providing an interactive apparatus for health professionals to develop cell-specific neurological disease and assessment.

Respiratory depression and failure are life-threatening conditions and major contributors to mortality in opioid overdose, spinal cord injury, and other disorders. Emerging evidence shows that epidural electrical stimulation can restore autonomic and volitional sensorimotor functions, offering a promising therapeutic avenue. We demonstrated that cervical epidural electrical stimulation (CEES) can modulate respiratory rhythms and reverse opioid-induced respiratory depression (ORID) in both rodents and humans. Despite this potential, the molecular and circuit-level mechanisms behind CEES-induced respiratory modulation remain unclear. While brainstem respiratory central pattern generators (CPGs) are well defined, the spinal cord’s contribution to respiratory rhythm generation is poorly understood. Moreover, the transcriptomic basis of respiratory regulation has not been fully explored. Recent advances in multiomics and tracing techniques now enable deeper insight into these mechanisms. We used these approaches to characterize spinal neurons involved in respiratory rhythm by examining their intra- and intercellular properties. Through single-nucleus RNA sequencing (snRNA-seq), we identified transcriptomic alterations and cellular profiles following CEES-induced respiratory facilitation in the upper brainstem, lower brainstem, and cervical spinal cord. These findings advance our understanding of respiration’s molecular and physiological basis and support development of more targeted therapies for respiratory depression.

Alzheimer’s disease (AD) is a neurodegenerative disorder marked by cognitive decline, memory loss, and widespread impacts on patient independence and quality of life. Despite decades of research, the diagnosis of Alzheimer’s remains a complex and often delayed process, with no single gold-standard assessment tool. This project presents a meta-analysis of the most widely used diagnostic measures—including the Mini-Mental State Exam (MMSE), Montreal Cognitive Assessment (MoCA), and neuroimaging techniques—and evaluates their effectiveness in early detection and tracking disease progression. In addition, this work investigates how the heterogeneity of Alzheimer’s progression complicates therapeutic development and access. While emerging drugs and clinical trials show promise, our findings highlight significant gaps in population data, especially among BIPOC and low-income communities. These gaps affect both diagnostic equity and the generalizability of therapeutic recommendations. By exploring the current landscape of diagnostic tools and treatment research, this analysis emphasizes the urgent need for inclusive, longitudinal data and systemic changes to healthcare access. Ultimately, improving early identification and tailored intervention pathways for Alzheimer’s patients will require interdisciplinary collaboration, equitable clinical trial design, and community-centered policy shifts.