Intelligence in how priors shape learning

I am a postdoc with James Evans at the University of Chicago, studying how priors in humans, LLMs, and scientific communities shape the way each explores hypothesis space.

Humans observe sparse and ambiguous data, yet from it they develop powerful internal models to make inferences about the world. LLMs not only generalize to novel tasks from a few in-context examples; their still poorly understood priors also guide their reasoning into spaces humans have not explored. Both humans and LLMs have good priors that make their learning and search efficient (Of course, as the bitter lesson claims, search and learning can scale by using massive amounts of computation. But we might not want monkeys at typewriters to do it, or at least we should give them some tokens.)

Seen this way, science is also prior-guided learning and search. The knowledge it produces is jointly constrained by the structure of the world and the structure of the human mind. How, then, can we describe the space of all possible forms of knowledge (perhaps with a universal grammar of knowledge)?

I work on understanding intelligence through the structure of priors: what makes a good prior (inductive bias), how priors constrain and guide learning and search in individuals, models, and communities, and how to shape knowledge so that it fits the human mind.

Some of my work (some with 5-minute videos walking through the main results):

  1. How representational structure constrains online learning and search:
  2. Models of human behavior:

The pronunciation of my name is “HWAH-doang SHAWNG”.

I am on the job market

I am looking for full-time Research Scientist and faculty positions on using AI to model human behavior (including the scientific process) and AI alignment.

Want to collaborate or chat?

I’m always happy to collaborate. Whether you’re interested in my research, want to chat about science, or are figuring out how to do research in graduate school, shoot me an email. My own path has had a few detours, so I might have something to say. If you’re at UChicago, I’m usually at the Knowledge Lab (1155 E 60th St), Room 203. Feel free to just walk in.

My trajectory

In my PhD, I used neural networks as models and meta-learning as theory to understand human decision making, mentored by Robert Wilson at Georgia Tech, collaborating with Ji-An Li and Marcelo Mattar. Before that, I was mentored by Xue-Xin Wei at UT Austin on RNN models of working memory. Before that, by Da-Hui Wang at Beijing Normal University on spiking network models of working memory. Before that, I used EEG to study working memory and studied counseling psychology.

Some of my writing

I also write about cognition, science, and the concepts we use to understand the world. If something interests you, feel free to email me. I’d like to hear what you think.

I wish I had more time to write something meaningless.

Some interesting facts:

  • I enjoyed reading when I was young. My favorite writers are James Joyce, Milan Kundera, Jorge Borges, Franz Kafka, Dostoyevsky and Edgar Allan Poe.
  • I chose this stupid username (sakimarquis) as a teenager. It came from Saki and Márquez.
  • Yet I haven’t read much since high school. I suddenly lost my patience with long books.
  • I am addicted to computer games, but only when an exam is approaching. Since I have few exams to take, I seldom play them now.
  • I enjoy sick jokes and embarrassing short videos.
  • Using a second language is painful for me, mostly because I can’t help being sarcastic but I can’t do it well in English.
  • I am bad at calculating. I often mess up single-digit calculations, even with a pen and paper. This always makes me doubt myself as a researcher in computational neuroscience. (But LLMs also fail at simple calculations, so I am not alone.)