About

Scott Linderman is an Assistant Professor of Statistics at Stanford University and a Faculty Scholar at the Wu Tsai Neurosciences Institute. His research focuses on machine learning, computational neuroscience, and the application of computational and statistical methods to decipher neural computation. His work aims at developing rich statistical models for analyzing neural data, addressing challenges related to extracting information from neural activity and understanding its relation to sensory inputs and behavioral outputs. Linderman's efforts have contributed to revealing latent structures underlying neural activity. He has been recognized for his contributions with the 2023 McKnight Scholar Award, a prestigious early-career honor for innovative investigators in neuroscience. Linderman is involved in the Stanford Center for Neural Data Science and has been interviewed about this initiative. His research combines developing new statistical methodologies with applying these methods to experimental data, advancing the understanding of neural computation through data-driven approaches.

Research topics

• Computer Science
• Neuroscience
• Psychology
• Physics

Selected publications

• Unifying Optimization and Dynamics to Parallelize Sequential Computation: A Guide to Parallel Newton Methods for Breaking Sequential Bottlenecks

  ArXiv.org · 2026-03-16

  dissertationOpen access

  Massively parallel hardware (GPUs) and long sequence data have made parallel algorithms essential for machine learning at scale. Yet dynamical systems, like recurrent neural networks and Markov chain Monte Carlo, were thought to suffer from sequential bottlenecks. Recent work showed that dynamical systems can in fact be parallelized across the sequence length by reframing their evaluation as a system of nonlinear equations, which can be solved with Newton's method using a parallel associative scan. However, these parallel Newton methods struggled with limitations, primarily inefficiency, instability, and lack of convergence guarantees. This thesis addresses these limitations with methodological and theoretical contributions, drawing particularly from optimization. Methodologically, we develop scalable and stable parallel Newton methods, based on quasi-Newton and trust-region approaches. The quasi-Newton methods are faster and more memory efficient, while the trust-region approaches are significantly more stable. Theoretically, we unify many fixed-point methods into our parallel Newton framework, including Picard and Jacobi iterations. We establish a linear convergence rate for these techniques that depends on the method's approximation accuracy and stability. Moreover, we give a precise condition, rooted in dynamical stability, that characterizes when parallelization provably accelerates a dynamical system and when it cannot. Specifically, the sign of the Largest Lyapunov Exponent of a dynamical system determines whether or not parallel Newton methods converge quickly. In sum, this thesis unlocks scalable and stable methods for parallelizing sequential computation, and provides a firm theoretical basis for when such techniques will and will not work. This thesis also serves as a guide to parallel Newton methods for researchers who want to write the next chapter in this ongoing story.

  PublisherOA PDFDOI

• Lifelong behavioral screen reveals an architecture of vertebrate aging

  Science · 2026-03-12 · 4 citations

  articleOpen access

  Mapping behavior of individual vertebrate animals across lifespan could provide an unprecedented view into the lifelong process of aging. We created a platform for high-resolution continuous behavioral tracking of the African killifish across natural lifespan from adolescence to death. We found that animals follow distinct individual aging trajectories. The behaviors of long-lived animals differed markedly from those of short-lived animals, even relatively early in life, and were linked to organ-specific transcriptomic shifts. Machine-learning models accurately inferred age and even forecasted an individual's future lifespan, given only behavior at a young age. Finally, we found that animals progressed through adulthood in a sequence of stable and stereotyped behavioral stages with abrupt transitions, revealing precise structure for an architecture of aging.

  PublisherDOI

• Spontaneous behavior is a succession of self-directed tasks

  Zenodo (CERN European Organization for Nuclear Research) · 2026-02-07

  datasetOpen access

  Data and code from the paper "Spontaneous behavior is a succession of self-directed tasks." Behavioral data Files named *-behavior.parquet contain dataframes with processed behavioral data (MoSeq syllabes, shMoSeq states, pose information and other metadata). Supplemental Table 2 from the paper specifies which figure panels each dataset contributed to. Files can be loaded into python using the command pandas.read_parquet. Note that shMoSeq states were filtered and renumbered post-hoc to remove state with extremely low frequency; timepoints assigned to these low-frequency states are treated as missing data. All centroid locations, distances, heights, and 3D keypoint coordinates are in mm. 2D keypoint coordinates and object bounding boxes are in pixel units. "Egocentric angle to wall" is defined as the angle between the mouse's heading and a vector pointing toward the nearest wall location. "Wall direction" is the cosine of this angle. Neural data Files named *-neural.parquet contain neural activity for each recording. Each array has shape (n_timepoints, n_neurons). The data have been resampled so that the n'th row of the neural activity array corresponds to the n'th row of the behavior dataframe (after filtering the dataframe by session). For calcium recordings, activity levels correspond to Z-scored dF/F. For ephys data, activity levels correspond to spike counts within 1/120-second intervals (since behavior was recorded at 120 frames/second). In some datasets, neural data is missing from the first and last few seconds of each recording. Arrays have been padded with constant values to cover these missing timepoints. Model parameters File named *-model_params.h5 contain parameters for the final shMoSeq models derived from each dataset. The parameters can loaded to a dictionary using state_moseq.load_hdf5. In a subset of models, states were renumbered before downstream analysis. The file state_indexes.csv specifies the renumbering for each dataset. For example, the following row in state_indexes.csv would mean that state 3 in lesion-model_params.h5 corresponds to state 1 in lesion-behavior.parquet. dataset original index renumbered index lesion 3 1 Code top-bottom-moseq-shMoSeq-paper.zip contains code from the top-bottom-moseq github repo, which was used for preprocessing depth videos as part of the MoSeq pipeline. multicam-calibration-0.0.0.zip contains code from the multicam-calibration github repo, which was used for 3D keypoint tracking. state-moseq-0.0.0.zip contains code from the state-moseq github repo, which was used for estimating behavioral states.

  PublisherDOI

• Stiefel Manifold Dynamical Systems for Tracking Representational Drift

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-03-10

  articleOpen accessSenior authorCorresponding

  Understanding neural dynamics is crucial for uncovering how the brain processes information and controls behavior. Linear dynamical systems (LDS) are widely used for modeling neural data due to their simplicity and effectiveness in capturing latent dynamics. However, LDS assumes a stable mapping from the latent states to neural activity, limiting its ability to capture representational drift-gradual changes in the brain's representation of the external world. To address this, we introduce the Stiefel Manifold Dynamical System (SMDS), a new class of model designed to account for drift in neural representations across trials. In SMDS, emission matrices are constrained to be orthonormal and evolve smoothly over trials on the Stiefel manifold-the space of all orthonormal matrices-while the dynamics parameters are shared. This formulation allows SMDS to leverage data across trials while accounting for non-stationarity, thus capturing the underlying neural dynamics more accurately compared to an LDS. We apply SMDS to both simulated datasets and neural recordings across species. Our results consistently show that SMDS outperforms LDS in terms of log-likelihood and requires fewer latent dimensions to capture the same activity. Moreover, SMDS provides a powerful framework for quantifying and interpreting representational drift. It reveals a gradual drift over the course of minutes in the neural recordings and uncovers varying drift rates across dimensions, with slower drift in behaviorally and neurally significant dimensions.

  PublisherOA PDFDOI

• Cross-brain transfer of high-performance intracortical speech and handwriting BCIs

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-01-14

  articleOpen access

  Intracortical brain-computer interfaces (BCIs) that decode complex movements, such as handwriting and speech, can require substantial training data to achieve high performance. We investigated whether leveraging the neural activity recordings of previous users could reduce this initial data collection burden for new BCI users (an approach we call "cross-brain transfer"). Using intracortical recordings from five BrainGate2 clinical trial participants, we tested cross-brain transfer for both speech and handwriting neural decoders trained and evaluated on general, unconstrained corpora of spoken and written English. We found that cross-brain transfer improved decoding performance when training data from the target user was limited (< 200 sentences), and that dataset-specific input layers to the decoder were critical for combining data across users. Without trainable input layers, transfer failed and performed worse than training from scratch on target user data only. Finally, we measured the effectiveness of cross-brain transfer relative to training with (1) more data from the same user and (2) more electrode-permuted data from the same user, which simulates sampling from another brain with identical neural latent structure. In some cases (T16 speech, T12 handwriting), cross-brain transfer appeared as effective as additional permuted data from the same user, while in others (T12 speech, T15 speech) electrode-permuted data was more beneficial. Our results successfully demonstrate and characterize cross-brain transfer learning between multiple intracortical BCI users, for both speech and handwriting, using a general open-ended dataset not restricted to small sets of words or phrases. This work highlights a promising path towards addressing a key barrier to the clinical translation of BCIs, while clarifying when cross-brain transfer may be most beneficial and the decoder design choices needed to realize those gains.

  PublisherOA PDFDOI

• Spontaneous behavior is a succession of self-directed tasks

  Neuron · 2026-01-28 · 1 citations

  articleOpen access
  PublisherOA PDFDOI

• Minions: Cost-efficient Collaboration Between On-device and Cloud Language Models

  ArXiv.org · 2025-02-21

  preprintOpen access

  We investigate an emerging setup in which a small, on-device language model (LM) with access to local data communicates with a frontier, cloud-hosted LM to solve real-world tasks involving financial, medical, and scientific reasoning over long documents. Can a local-remote collaboration reduce cloud inference costs while preserving quality? First, we consider a naive collaboration protocol where the local and remote models simply chat back and forth. Because only the local model reads the full context, this protocol achieves a 30.4x reduction in remote costs, but recovers only 87% of the performance of the frontier model. We identify two key limitations of this protocol: the local model struggles to (1) follow the remote model's multi-step instructions and (2) reason over long contexts. Motivated by these observations, we study an extension of this protocol, coined MinionS, in which the remote model decomposes the task into easier subtasks over shorter chunks of the document, that are executed locally in parallel. MinionS reduces costs by 5.7x on average while recovering 97.9% of the performance of the remote model alone. Our analysis reveals several key design choices that influence the trade-off between cost and performance in local-remote systems.

  PublisherOA PDFDOI

• Extracting task-relevant preserved dynamics from contrastive aligned neural recordings

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-11-13

  preprintOpen access

  Abstract Recent work indicates that low-dimensional dynamics of neural and behavioral data are often preserved across days and subjects. However, extracting these preserved dynamics remains challenging: high-dimensional neural population activity and the recorded neuron populations vary across recording sessions. While existing modeling tools can improve alignment between neural and behavioral data, they often operate on a per-subject basis or discretize behavior into categories, disrupting its natural continuity and failing to capture the underlying dynamics. We introduce C ontrastive A ligned N eural DY namics (CANDY), an end-to-end framework that aligns neural and behavioral data using rank-based contrastive learning, adapted for continuous behavioral variables, to project neural activity from different sessions onto a shared low-dimensional embedding space. CANDY fits a shared linear dynamical system to the aligned embeddings, enabling an interpretable model of the conserved temporal structure in the latent space. We validate CANDY on synthetic and real-world datasets spanning multiple species, behaviors, and recording modalities. Our results show that CANDY is able to learn aligned latent embeddings and preserved dynamics across neural recording sessions and subjects, and it achieves improved cross-session behavior decoding performance. We further show that the latent linear dynamical system generalizes to new sessions and subjects, achieving comparable or even superior behavior decoding performance to models trained from scratch. These advances enable robust cross-session behavioral decoding and offer a path towards identifying shared neural dynamics that underlie behavior across individuals and recording conditions. The code and two-photon imaging data of striatal neural activity that we acquired here are available at https://github.com/schnitzer-lab/CANDY-public.git .

  PublisherDOI

• Identifying multi-compartment Hodgkin-Huxley models with high-density extracellular voltage recordings.

  PubMed · 2025-06-25

  preprintOpen accessSenior author

  Multi-compartment Hodgkin-Huxley models are biophysical models of how electrical signals propagate throughout a neuron, and they form the basis of our knowledge of neural computation at the cellular level. However, these models have many free parameters that must be estimated for each cell, and existing fitting methods rely on intracellular voltage measurements that are highly challenging to obtain in vivo. Recent advances in neural recording technology with high-density probes and arrays enable dense sampling of extracellular voltage from many sites surrounding a neuron, allowing indirect measurement of many compartments of a cell simultaneously. Here, we propose a method for inferring the underlying membrane voltage, biophysical parameters, and the neuron's position relative to the probe, using extracellular measurements alone. We use an Extended Kalman Filter to infer membrane voltage and channel states using efficient, differentiable simulators. Then, we learn the model parameters by maximizing the marginal likelihood using gradient-based methods. We demonstrate the performance of this approach using simulated data and real neuron morphologies.

  PublisherOA PDF

• Dynamax: A Python package for probabilistic state space modeling with JAX

  The Journal of Open Source Software · 2025-04-03 · 6 citations

  articleOpen access1st authorCorresponding
  PublisherOA PDFDOI

Recent grants

• Towards a Unified Framework for Dopamine Signaling in the Striatum

  NIH · $32.4M · 2019–2025

• Towards a Unified Framework for Dopamine Signaling in the Striatum

  NIH · $7.2M · 2019–2024

Frequent coauthors

• Alex H. Williams

  New York University

  37 shared

• Bernardo L. Sabatini

  Howard Hughes Medical Institute

  22 shared

• Aditya Nair

  California Institute of Technology

  20 shared

• Sandeep Robert Datta

  Loughborough University

  19 shared

• Thomas Panier

  Sorbonne Université

  18 shared

• David J. Anderson

  Howard Hughes Medical Institute

  18 shared

• Jeffrey E. Markowitz

  The Wallace H. Coulter Department of Biomedical Engineering

  17 shared

• David M. Blei

  16 shared

Education

• PhD, Computer Science

  Harvard University

  2016

• BS, Electrical and Computer Engineering

  Cornell University

  2008

Awards & honors

• 2023 McKnight Scholar