Interested in joining the lab? Below are some opportunities to get involved :)
All interest inquiries about joining the lab are accepted through this google form. Unfortunately, we have been receiving a large volume of AI-generated email inquiries and are unable to respond to individual emails at this time (see this post explaining why). If you are genuinely interested in joining the lab, please fill out the form instead, and we'll reach back to you if there seems to be a match with one of the available research opportunities!
Research Assistants
The RA opportunities in the lab vary each semester and are only available to current UC Berkeley undergraduate and masters students.
There are two exciting RA opportunities in the lab:
Learnability. How do we decide when there is something to be learned, and persist on trying to learn even when the immediate success is unclear (i.e., the problem is quite hard)? The goal of this project is to model how we do (and should) judge learnability, and the common failure modes of this kind of judgment -- 1) illusion of control (when we think we've learned something when we actually haven't) and 2) learned helplessness (when we think there is nothing to be learned yet there is). The student will lead this project and gain experience in computational cognitive modeling, mentored directly by Prof. Dubova. A qualified student has to have technical and/or mathematical background (ideally, be a computer science or cognitive science major at UCB) and interest in cognitive psychology or philosophy of science. Please fill out this form to indicate your interest and mention you are interested in the "learnability" project.
Recognition ability of overparameterized (excess capacity) models (joint with Aly lab). The goal of the project is to assess different types of recognition abilities in models of different complexity. This project sits at the intersection of cognitive psychology, computer science, and statistical learning. Background reading: this, this and this. The student will lead this project and gain experience in computational cognitive modeling, mentored directly by Prof. Aly and Prof. Dubova. A qualified student has to have technical and/or mathematical background (ideally, be a computer science major at UCB) and interest in cognitive science. Please fill out this form to indicate your interest and mention you are interested in the "recognition ability" project.
PhD students
If you are a current PhD student at UC Berkeley and would like to get involved in the lab, please email Marina or fill out this form. We welcome UC Berkeley PhD students at our lab meetings, if you want to be added to our lab meeting mailing list, email Joshua.
The lab is looking to admit one or two graduate students to start in Fall 2027. Students from a range of backgrounds (including psychology, computer science, and philosophy) are welcome to apply. Prospective students should apply to the Department of Psychology at UC Berkeley (see here for more info, application deadline is November 16th). You don't have to email me before applying; if you'd like to indicate your interest before applying, you can do it by filling out this form (this is totally optional).
The lab works on many different topics, and a big part of your decision to apply to work with us and our decision to admit you is whether we are driven by the same scientific questions and research methods. Below are things that can help you decide if the lab is a good fit for you:
There are many questions that you can apply to work on in the lab. These questions include: scientific concepts, conceptual influences on scientific inquiry, conceptual revolutions, process vs. entity ontologies, causal learning, exploration vs. exploitation in science, illusions of understanding, levels of explanation, limits of learning, complexity vs. parsimony in theory building, actor-critic models of scientific reasoning, measurement, analogical reasoning, epistemic diversity, performativity/looping effects, and many others.
Even though lab works on some seemingly 'abstract' or 'fuzzy' questions, we believe they can be rigorously approached with cognitive methods such as computational models, and experiments with humans and, in particular, scientists. We welcome students with different backgrounds (cognitive science, psychology, computer science, philosophy, and others), as long as they are interested in employing some of these cognitive methods and approaching their questions through a cognitive lens.
We will not admit a student who is primarily interested in LLM vs. human comparisons, mechanistic interpretability of LLMs, etc. We are broadly interested in (biological and artificial) minds, and would like to focus on improving our understanding of cognitive mechanisms and principles that go beyond a specific type of model or architecture. If you are interested in studying LLMs, please check out our friends at the colala lab, computational cognitive science lab, and berkeley nlp group.
We are not developing new AI for science assistants. We worry that we may not yet understand the discovery processes well enough to thoughtfully scale them up via automation (in other words, what are we automating?). This doesn't mean that we think human minds are superior, but rather we believe that someone has to carefully study the processes of discovery before deploying our intuitions about these processes at scale. Therefore, the lab focuses on studying the cognitive processes of discovery, and we hope that this knowledge could ultimately inform the development of new tools for science. If you are interested in automating science, please check out our friends at the automated scientific discovery of mind and brain lab.
Postdocs
We are looking to potentially recruit a postdoc with a flexible start date. If you're interested, please fill out this form. These topics of interest include (but are not limited to) scientific concepts, conceptual influences on scientific inquiry, conceptual revolutions, process vs. entity ontologies, causal learning, exploration vs. exploitation in science, illusions of understanding, levels of explanation, limits of learning, complexity vs. parsimony in theory building, actor-critic models of scientific reasoning, measurement, analogical reasoning, epistemic diversity, performativity/looping effects, and many others.
Other (e.g., visiting researcher)
We are open to hosting research visitors who'd like to spend some time interacting with the lab and working with Marina. Please email Marina or fill out this form to indicate your interest.