Advisor(s)

Kirstie Cummings

Committee Member(s)

Craig Powell
Jeremy Day
Lalita Shevde-Samant
Sofia Beas
William Marsiglia

Document Type

Dissertation

Date of Award

6-1-2026

Degree Name

Doctor of Philosophy (PhD)

School

Joint Health Sciences (Interdisciplinary)

Department

Neurobiology

Abstract

Interoceptive predictions are internal models through which the brain anticipates future bodily states based on past experiences and environmental cues. Under healthy conditions, these predictions allow organisms to anticipate physiological needs and proactively adjust behavior to maintain homeostasis. When these processes become dysregulated, however, environmental cues can trigger maladaptive physiological responses. Although interoceptive prediction is increasingly recognized as a key mechanism underlying adaptive behavior, the neuronal circuits responsible for generating these predictions remain poorly understood. The paraventricular nucleus of the thalamus (PVT) receives convergent inputs from brain regions involved in processing both interoceptive signals and exteroceptive cues, positioning it to integrate internal physiological information with environmental context to guide behavior. A subpopulation of PVT neurons expressing dopamine receptor 2 (PVTD2R+) is sensitive to fluctuations in physiological state, suggesting a potential role in encoding interoceptive predictions. Therefore, we sought to define the contribution of PVTD2R+ neurons to interoceptive predictions during goal-directed behavior. To address this question, we used fiber photometry to investigate the in vivo dynamics of PVTD2R+ neurons in mice performing a linear maze reward-seeking task. We found that PVTD2R+ neuronal activity increased during reward approach and decreased during task return, and that these responses were strongly modulated by physiological state, outcome relevance, and reward expectation. PVTD2R+ responses increased when animals approached physiologically relevant rewards, such as food during hunger or water during thirst, but were reduced when outcomes were physiologically irrelevant. In addition, PVTD2R+ responses were strongly modulated by reward expectation, decreasing during trials lacking reward-predictive cues and when the reward was omitted. Across training sessions, these approach and return responses emerged gradually, indicating that PVTD2R+ activity reflects learned cue-outcome associations. Additionally, PVTD2R+ responses scaled with reward magnitude, further supporting a role in predictive processing. Furthermore, chronic chemogenetic inhibition of PVTD2R+ neurons impaired performance in the linear maze task. Together, our findings demonstrate that PVTD2R+ neurons integrate physiological needs, environmental cues, and prior experiences to compute interoceptive predictions, highlighting these neurons as a critical neural mechanism for interoception that supports the perception of internal states and guides adaptive, motivated behavior.

Keywords

interoception;interoceptive predictions;paraventricular nucleus of the thalamus

Included in

Neurosciences Commons

Share

COinS