Research
My research aims are in two parts: How do people make sense of what they want? How do people assess the legitimacy of their value representations?
​​​
How do people make sense of what they want?
Humans do not have direct access to what we want. Instead, we are resigned to make inferences to estimate what we want. What are the relevant objects of our inferences? I am interested in how people learn from their internal affective signals experienced in certain environment, to form attitudinal representations over which possible worlds they would want to move towards and which possible worlds they would want to move away from.
​​
How do people assess the legitimacy of their value representations?
This latter question motivates my inquiry into value metacognition—our capacity to monitor and regulate our own value architecture. Through this lens, I study mechanisms such as regret, confidence, and introspective uncertainty as computational signals guiding how we monitor for the quality of our value representations. I hope to extend these ideas into value social metacognition - our capacity to form judgments about the quality of other people's desires - e.g., as shown in paternalism.
​​
Application
I am actively interested in applications of this work to strengthening the theory behind self-care practices in mental health and in AI Alignment (aligning AI with human values).
In mental health, many disorders—such as depression, anhedonia, and certain anxiety syndromes—may involve distorted or unstable representations of value. A better understanding of how people infer what they want and attempt the regulate their motivation could inform therapeutic strategies aimed at restoring coherence between values and goals.
In the context of human-AI value alignment, insights into the cognitive basis of value representation can contribute to AI alignment efforts: helping to design systems that are not only responsive to human preferences, but that respect the nuanced and evolving nature of human values and people's intuitive theories underlying value legitimacy.