Computational Neuroscience

How does a brain learn, and then remember?

I build mathematical models of learning and memory and test them against large-scale neural recordings — where theory, data, and machine learning meet.

Assistant Professor · Department of Neuroscience · Brandt Institute of Technology

About

I'm a computational neuroscientist. My lab asks how the brain turns experience into lasting change — how a spiking network reshapes itself to hold a memory, and why some memories stabilize while others fade.

We work in both directions: building theoretical models of synaptic plasticity, then testing them against large-scale recordings from collaborators' labs. Increasingly that means borrowing tools from machine learning — and asking, in return, what the brain can teach us about learning algorithms. The goal is models that don't just fit the data, but explain it.

plasticitymemoryspiking networksneural dataML × neuroscience

Selected Publications

Recent work

A representative selection. Full list and preprints on the CV and lab site.

Nature Neuroscience · 2026

A plasticity rule that predicts which memories survive sleep

A single local learning rule reproduces overnight consolidation across three published datasets.

Raman, P., Liu, S., & Cortez, M.

Neuron · 2025

Drift without forgetting: stable readout from unstable representations

How a population can keep a memory legible even as individual tuning slowly drifts.

Raman, P. & Adeyemi, K.

NeurIPS · 2024

Biologically constrained recurrent networks learn faster with fewer samples

Architectural priors from cortex improve sample efficiency on sequence tasks.

Cortez, M., Raman, P., et al.

eLife · 2023

A shared geometry of learning across motor and spatial tasks

Low-dimensional structure in neural activity generalizes across two behaviors.

Raman, P., Okonjo, T., & Liu, S.

The Raman Lab

Open models, open code, open data

Everything the lab builds ships as documented, reusable code. Our plasticity-rule library has been downloaded by groups on four continents, and we maintain a public benchmark for comparing biological learning rules against machine-learning baselines. We train students to write code other people can actually run.

6Lab members
14kLibrary installs
100%Open-source

Teaching

Courses

  • Computational Neurosciencegraduate · spring
  • Models of Learning & Memoryadvanced seminar
  • Python for Neural Datamethods · open to all majors
  • PhD & postdoc mentoringrecruiting · see contact

Talks & Service

Beyond the lab

  • Invited talk, Computational Neuroscience Summit2025 · main meeting
  • Tutorial, Open Neuroscience Summer Schoolplasticity models · 2024
  • Associate editorTheoretical Neuroscience Reports
  • "What AI gets wrong about memory"public lecture · 2024

Get in touch

Students, collaborators, and press.

Prospective PhD students: please include your research interests and CV. Journalists: note your deadline in the subject line.

replies within ~2 business days