Hadi Vafaii

Personal Website

Reveal Contact Info

Postdoc

Vision Science

Yates Lab

Current Research

How do brains infer the true state of the world from noisy, limited, and often unreliable sensory information? This challenge can be framed as an inverse problem: the world exists in some ground truth state that generates the sensory data collected by the brain, and the brain’s task is to infer this true state given only those observational projections.

In this sense, the brain can be thought of as an “”electric meatball inference machine.””

In my research, I build brain-like generative models that infer the hidden causes of their observations. Like the brain, these models use discrete and sparse latent representations, and reuse computational resources through recurrent connectivity.

I develop these models for two reasons: (1) as computational models of the brain, enabling me to formulate and test hypotheses about how neural systems infer their surroundings, and (2) to enhance machine learning by incorporating brain-like principles, hoping to make artificial systems more adaptive, and therefore, more intelligent.

At the Redwood Center, I collaborate with both the Yates lab and the Olshausen lab to advance this interdisciplinary research agenda.

Background

I am originally from Iran. I grew up in Tabriz, a mountainous city of about 2 million people with great food and snowy winters. After completing high school, I moved to Tehran and earned my BSc in physics from Sharif University of Technology in 2016. I then moved to the United States to pursue a PhD in physics at the University of Maryland, College Park.

At the beginning of my PhD, I was working in theoretical condensed matter physics, but switched to neuroscience halfway through. This came after I recognized that understanding the true nature of reality requires understanding the observer that makes inferences about it. Namely, the human brain.

After defending my PhD in 2023, I joined the Yates lab at the Redwood Center as a postdoctoral scholar in 2024 to pursue this inquiry.

My research has since focused on two central questions: (1) What are neural mechanisms and algorithms that brains employ to make sense of the world, given limited and noisy sensory information? And (2) can we draw inspiration from how brains solve this daunting inference problem to enhance AI algorithms?

Besides research, my hobbies include tennis and playing music. You can listen to some of my compositions here: https://mysterioustune.com/music/