GLM-HMM MATLAB Toolbox

A simple and modular MATLAB implementation of the Generalized Linear Model Hidden Markov Model (GLM-HMM), adapted from Ashwood et al. (2022). This toolbox allows fitting latent behavioral states to choice data using state-specific GLMs and EM inference.
see the following link for the authors notebooks (python implementation)
https://github.com/zashwood/ssm/blob/ster/noteboomaks/2b%20Input%20Driven%20Observations%20(GLM-HMM).ipynb

---

🧠 What It Does

• Infers discrete latent behavioral states (e.g., “stimulus-driven”, “biased left/right”)
• Supports 2AFC and Go/No-Go paradigms
• Uses EM algorithm with GLM-based emission models
• Works on both real and simulated behavioral data

---

🚀 Getting Started

Requirements

• MATLAB R2018 or newer
• No additional toolboxes required

---

🧪 Demo: Simulated Data

To test the toolbox on ground-truth simulated behavior:

run demo/Demo_simulated_data.m

---

---

🔁 Example: Fit to Your Own Data

Prepare two vectors:

• stimIntensity — stimulus strength per trial (e.g., signed amplitude)
• trialOutcome — binary trial outcome (0 = left/no-lick, 1 = right/lick)

addpath(genpath('path/to/GLM-HMM-MATLAB'))

params.GLMHMM.K = 5;              % number of latent states
params.GLMHMM.max_iter = 100;
params.GLMHMM.tol = 1e-3;
params.GLMHMM.pop_weigths = 1;
params.GLMHMM.display_output = 1;

Results = Fit_GLMHMM_GNGdata(stimIntensity, trialOutcome, params.GLMHMM);

---

📈 Outputs

Results contains:

• dom_state — most probable latent state per trial
• performance — GLM classification accuracy per state
• W — GLM weights (one vector per state)
• gamma — posterior over latent states (T x K)

---

📜 License

Released under the MIT License.
Free for academic and commercial use. No warranty provided.

---

🧩 Citation

Adapted from:

Ashwood, Z. C., Roy, N. A., Stone, I. R., et al. (2022).
Mice alternate between discrete strategies during perceptual decision-making.
Nature Neuroscience, 25(2), 201–212. [https://doi.org/10.1038/s41593-021-01007-z]