# Toward a Computationally-Informed, Personalized Treatment for Hallucinations

> **NIH NIH R21** · YALE UNIVERSITY · 2021 · $183,750

## Abstract

PROJECT SUMMARY / ABSTRACT
Despite advances in the computational study of perception, there has been remarkably little progress in
understanding perceptual abnormalities like hallucinations. In recent work, we identified abnormalities in
information processing that underlie hallucinations using cutting-edge computational modeling of perception.
We adopted a hierarchical Bayesian framework, which views perception as a constructive process wherein
prior knowledge of the environment is combined with incoming sensory information to build an internal
representation of one’s surroundings. The weight each of these sources exerts during perception is dependent
upon its precision (or reliability). Thus, within this framework, hallucinations—percepts in the absence of
corresponding incoming sensory evidence—may arise from an over-weighting of prior knowledge in
comparison to incoming sensory evidence. To demonstrate this, we used classical conditioning to safely and
reversibly induce hallucinations of simple tones in those both with and without auditory verbal hallucinations
(AVH). Those with AVH were roughly five times more susceptible to this Conditioned Hallucinations effect
because of a tendency to weigh prior knowledge more than incoming sensory information during perception.
This relative weighting of priors versus sensory evidence during perception depends critically on cholinergic
signaling: acetylcholine biases perceptual inference toward sensory data and away from priors. Thus, in voice-
hearers with abnormally high prior weighting, enhancing cholinergic signaling could result in fewer
hallucinations. We propose to characterize the effects of cholinergic signaling on the perceptual,
computational, physiological, and clinical signatures of hallucinations. Principally, we hypothesize that: 1)
Decreasing cholinergic tone with scopolamine (an M1 cholinergic receptor antagonist) in healthy participants
will result in exhibit higher prior weighting, more conditioned hallucinations, and more prior-related
brain activity compared to placebo; 2) Increasing cholinergic tone with IV physostigmine (a reversible,
centrally-acting cholinesterase inhibitor) in patients with daily hallucinations will result in decreases in prior
weighting, conditioned hallucinations, and clinical hallucinations compared to placebo; and 3) These
effects will depend on the existence of high prior weighting at baseline assessment. Our goal is use the
knowledge generated to take the first steps toward a computationally-informed, personalized treatment
approach to hallucinations.

## Key facts

- **NIH application ID:** 10159329
- **Project number:** 5R21MH122940-02
- **Recipient organization:** YALE UNIVERSITY
- **Principal Investigator:** Albert R Powers
- **Activity code:** R21 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2021
- **Award amount:** $183,750
- **Award type:** 5
- **Project period:** 2020-05-05 → 2024-04-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10159329

## Citation

> US National Institutes of Health, RePORTER application 10159329, Toward a Computationally-Informed, Personalized Treatment for Hallucinations (5R21MH122940-02). Retrieved via AI Analytics 2026-05-24 from https://api.ai-analytics.org/grant/nih/10159329. Licensed CC0.

---

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