Hello! I'm Carlos G. Correa.

I’m a postdoc at NYU, advised by Marcelo Mattar. My PhD was advised by Nathaniel Daw and Tom Griffiths at Princeton University, and used behavioral experiments and computational modeling to study how hierarchical representations and other kinds of structure can support efficient decision making.

My CV is here.

News

September 2024 - I started as a Postdoctoral Associate in the Mattar Lab at NYU!

May 2024 - I successfully defended my dissertation!

April 2024 - New conference paper! I’ll be presenting “Program-Based Strategy Induction for Reinforcement Learning” at CogSci this summer. Check the preprint out on arXiv.

November 2023 - New preprint! “Exploring the hierarchical structure of human plans via program generation” is out on arXiv. Read the twitter thread for an overview of the paper.

June 2023 - New publication! “Humans decompose tasks by trading off utility and computational cost” is out in PLOS Computational Biology. Here’s an image I made for the publication:

Caption: Dividing a task into simpler subtasks can make planning easier. At bottom, the task requires navigating a maze from the entry (at left) to the exit (at right; glowing arrow). At top, the maze is divided into subtasks, showing two key elements of hierarchical planning: an abstract plan between subtasks (boxes) and a concrete plan from the entry to the first subgoal (glowing circle). Dividing the task in this way makes it easier to plan. Our research article finds that people choose subgoals by balancing ease of planning with efficient behavior, consistent with our computational theory. Image Credit: Carlos G. Correa. License: CC-BY.

About Me

I live in Brooklyn, NY. I used to live in San Francisco, CA when I worked as a software engineer. I grew up in a small town in TX and went to college in Austin, TX. My parents are from a small town in Mexico. In my free time, I bake and garden and write programs.

Here are other places you can find me on the internet: [Google Scholar] [ORCiD] [GitHub] [Twitter]