# Computational and neural underpinnings of decision-making in social contexts

> **NIH NIH K00** · HARVARD UNIVERSITY · 2020 · $82,432

## Abstract

Project Summary
Social learning—the ability to learn from others—is essential for adaptive behavior. Humans not only receive
social information, but also actively interpret it in light of the unobservable mental states of the person providing
it (e.g., their knowledge, goals, and preferences). This ability to reason about others’ mental states, known as
Theory of Mind (ToM), allows humans to capitalize on sparse, imperfect social information. The central
objective of this proposal is to characterize the neural computations that enable humans to use ToM to guide
decision-making. Specifically, this proposal integrates computational models of ToM with fMRI to ask how the
outputs of mental state inferences are represented in neural systems that support ToM and value-based
decision-making (Aims 1 & 2) and how these systems interact to support flexible behavior (Aim 3).
My dissertation work provides a computational account of how adults use mental state inferences to make
decisions that benefit the self (Aim 1.1) and others (Aim 1.2). In recently published work, Aim 1.1 found that
adults adjust their use of advice based on the advisor’s knowledge, intent, and strategy; participants’ behavior
was best described by a Bayesian ToM model that infers the value of a hidden option that is only known to the
advisor. Aim 1.2 found that adults can generalize others’ preferences to a set of novel options; these results
suggest that adults’ representations of others’ preferences are grounded in abstract, generalizable features.
This work has provided extensive training in computational models of cognition.
Work proposed during the F99 phase will use fMRI to test hypotheses about the neural implementation of
these computations. Building on Aim 1.1, Aim 2.1 tests whether neural value signals track the value of hidden
options that must be inferred from advice. Building on Aim 1.2, Aim 2.2 tests whether the representational
structure of patterns of activity shifts during choice, based on whether participants are considering their own or
their partner’s preferences and on which features the person for whom they are choosing values. This work will
provide extensive training in model-based fMRI and will capitalize on the tools and training available in my
graduate institution to bring this work to a high standard of transparency and computational reproducibility.
Work proposed in the F00 phase will examine how humans allocate cognitive resources to mental state
inference. This work will test the overarching hypothesis that humans choose between mental state inference
and other, simpler strategies by balancing mental effort with expected information gain. This work will build on
my expertise in model-based fMRI by providing additional training in connectivity-based fMRI methods,
information theory, and resource-rational cognitive models. In summary, the proposed work will yield new
insights into computational and neural underpinnings of decision-making in social contexts ...

## Key facts

- **NIH application ID:** 10189934
- **Project number:** 8K00MH125856-02
- **Recipient organization:** HARVARD UNIVERSITY
- **Principal Investigator:** Natalia Velez
- **Activity code:** K00 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2020
- **Award amount:** $82,432
- **Award type:** 8
- **Project period:** 2019-07-01 → 2024-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10189934

## Citation

> US National Institutes of Health, RePORTER application 10189934, Computational and neural underpinnings of decision-making in social contexts (8K00MH125856-02). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10189934. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*