# The structure and dynamics of mental state representations

> **NIH NIH R01** · PRINCETON UNIVERSITY · 2021 · $364,500

## Abstract

PROJECT SUMMARY / ABSTRACT
Navigating the social world requires people to predict others' actions. This poses a significant challenge be-
cause people cannot directly observe some of the best predictors of action: others' internal mental states. In-
dividuals who can leverage information about these hidden causes of actions—by representing mental
states—can better predict those actions and more successfully navigate the social world. How do people (i)
represent the richness and complexity of others' invisible mental states, and (ii) use those representations to
make social predictions? We propose that people reduce the complexity of others' minds by attending to the
location of their mental states on a few key dimensions in a mental state “map.” We have previously used rep-
resentational similarity analyses (RSA) on functional neuroimaging (fMRI) data to show that people indeed
represent others' mental states using a simple, low-dimensional map. The structure of this map is defined by
three dimensions—rationality, social impact, and valence. Understanding how people employ this map will
provide key insights into how people predict others' actions. Aim 1: This proposal seeks to develop a compre-
hensive framework of mental state representations by first characterizing the structure of the map of mental
states.
We will
refine the structure by assessing how it adapts across new social contexts and modalities. We
will measure two structural features of the map—size and shape—using novel RSA methods on fMRI data. Di-
mensions that have universal social functions should hold a stable shape across all contexts and modalities;
dimensions that have specific, or contextualized functions should deform across context or modality. The size
of the space should expand to reflect the social relevance of the target. Understanding the structure of the men-
tal state map lays the foundation for understanding how people represent others' mental states. Aim 2: We
next explore how people leverage this map to make social predictions. We propose that the mental state map
encodes not only the location of others' current mental state, but also where in the map they will likely move to
Thus, people could make social predictions
We will use fMRI, large-scale experience sampling studies, and computational modeling
over behavioral data to establish that people indeed spontaneously model others' mental state dynamics, and
moreover, that these models make accurate social predictions. In both aims, we will test how the structure and
dynamic of mental state maps predict social functioning (or dysfunction). Using an individual differences ap-
next. by modeling the dynamics of others' mental states as paths
through this map.
proach, we will link our novel measures of structure and dynamics to participants' performance on a battery of
social cognition, social behavior, and social relationship measures. Taken together, this proposal
uses innova-
tive techniques (e.g., novel RSA metho...

## Key facts

- **NIH application ID:** 10176594
- **Project number:** 5R01MH114904-05
- **Recipient organization:** PRINCETON UNIVERSITY
- **Principal Investigator:** Diana Tamir
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2021
- **Award amount:** $364,500
- **Award type:** 5
- **Project period:** 2017-09-18 → 2023-05-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10176594

## Citation

> US National Institutes of Health, RePORTER application 10176594, The structure and dynamics of mental state representations (5R01MH114904-05). Retrieved via AI Analytics 2026-05-23 from https://api.ai-analytics.org/grant/nih/10176594. Licensed CC0.

---

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