Cole Hurwitz

This study examines age-related changes in neural and behavioural variability during a decision-making task in mice, finding increased neural variability and reduced variability quenching in specific brain regions.

This paper improves neural decoding by exploiting correlations across trials and sessions, showing better accuracy for four behaviors in mice using interpretable models.

This paper introduces PhysMAP, a framework that uses multimodal electrophysiological data to identify and study cell types in vivo, showing improved alignment with transcriptomic cell types and circuit dynamics.

This study evaluates the reproducibility of electrophysiological measurements across ten labs using standardized protocols and finds that variability in electrode targeting and statistical methods affects reproducibility.

This paper introduces the BAND model to extract neural dynamics for feedback-driven motor corrections during reaching, finding that corrections are encoded in small neural variability in M1 cortex and are transient and sensory-driven.

This paper introduces NEMO, a multimodal contrastive learning method for classifying cell types and brain regions from electrophysiological data, achieving state-of-the-art results on two opto-tagged datasets and the International Brain Laboratory Brain-wide Map dataset.

This work introduces a contrastive learning framework for extracellular data, showing improved performance over specialized methods in spike sorting and cell-type classification.

This paper introduces a spike sorting-free decoding method using density-based modeling to improve neural decoding performance without explicitly clustering spikes.

This paper introduces NP Ultra, a high-density electrode that improves neuronal recording yield, detection, and cell type classification in mice.

DARTsort improves spike sorting by incorporating drift tracking, outperforming Kilosort 2.5 on simulated data with drift and noise.

The paper describes a modular data architecture for a neuroscience collaboration, enabling data contribution, access, and analysis across multiple labs.

SpikeInterface provides a unified framework to run, compare, and benchmark spike sorting algorithms, reducing manual curation and enabling comprehensive benchmarking of automated methods.

SpikeForest benchmarks spike sorting algorithms using a curated database of electrophysiological recordings with ground truth, providing interactive results to guide optimal sorter selection.

This paper introduces a Bayesian model using variational autoencoder for localizing spike sources in extracellular recordings, showing improved accuracy over heuristic methods.