Born out of a recurring need to perform similar analyses across multiple projects and datasets, this toolkit provides a small library of Matlab functions and scripts for neural and behavioral data analysis. The repository also includes functions for calculating various neuronal selectivity indices that I used throughout my career. "NeuroIndexes" is an ongoing side project aiming to evolve into a library of clear and reusable code useful for teaching and research purposes.

This repository offers my Matlab implementation of the powerful "Bootstrap-t" procedure for estimating confidence intervals via resampling. Designed for versatility, the code allows you to apply Bootstrap-t-based confidence interval estimation to the output of any user-defined Matlab function, given its inputs. It's especially useful in situations where traditional methods for estimating confidence intervals fall short. The repository includes both serial and fast parallelized versions of the Bootstrap-t algorithm to optimize computation time.

This Zenodo repository contains the source code and data for our automated method, designed for laminar identification and depth estimation of cortical recording sites during in vivo extracellular recordings with multichannel silicon probes. Unlike traditional histological procedures, which are time-consuming and subject to human judgment, our template-matching algorithm offers an automated and reliable alternative. The repository includes the core Matlab code for implementing this method on new datasets, as well as the original data from our published article in J. Neurophysiol. 

In this review, we provide a comprehensive overview of the state of research (up to 2023) in both experimental and theoretical domains, exploring how unsupervised learning enables the visual cortex to develop efficient and effective sensory representations, primarily leveraging the statistics of its inputs. We aim to contextualize the recent efforts in bridging neuroscience and artificial intelligence on this fascinating topic, highlighting the interplay between enlightening experimental observations and insightful theoretical ideas. This review could be valuable for both neuroscientists and machine learning practitioners interested in the unsupervised learning algorithms shaping the neuronal networks of the sensory cortex.

This repository contains code from my side project, which focuses on exploring the behavior of large language models (such as Llama-3b-Instruct) to better understand their decision-making and information processing through the lens of 'textual psychophysics.' I am passionate about applying neuroscience-inspired mechanistic interpretability to these systems. I believe that subjecting these models to behavioral experiments, analogous to those traditionally conducted on human or animal subjects, and probing their internal representations as an electrophysiologist would do, can help foster a much-needed connection between the study of biological and artificial brains.