Synaptic histamine shapes the neurocomputational dynamics of human learning

Michael Colwell, Fin van Uum, Philip J. Cowen, Marieke Martens, Michael Browning, Helen C. Barron, Catherine J. Harmer, Susannah E. MurphybioRxiv 2025 · DOI

Abstract

Histamine was the first canonical monoamine identified in the mammalian brain, yet arguably remains the least understood in its mechanistic contributions to human behaviour. Using a first-in-class causal probe (H3R inverse agonist pitolisant), we show how increased synaptic histamine shapes offline and online temporal-hippocampal dynamics, sustaining learning-related activity and polarising retrieval computations. At a broader level, elevated histamine adaptively shifts neurocomputational strategy under high cognitive load, while stabilising value updates during aversive reinforcement learning. These findings uncover a mechanistically grounded influence of this underexplored system on human neurocomputation, supporting its therapeutic potential in psychiatry.

Introduction

While neurotransmitters like dopamine and serotonin are well-studied for their roles in reward and mood, histamine has often been relegated to the periphery of neuroscience, primarily associated with peripheral allergic responses. Originating in the tuberomammillary nucleus (TMN) and projecting widely across the cortex, the histaminergic system's causal role in human learning has remained elusive due to a "translational gap"—a lack of safe, selective pharmacological tools for human research. This study bridges that gap, positioning histamine as a critical modulator of the brain's neurocomputational architecture.

Methods: Probing the Histaminergic System

The researchers employed a randomized, double-blind, placebo-controlled design with 60 healthy participants. To selectively increase synaptic histamine, they utilized pitolisant, a first-in-class $H_3R$ inverse agonist that blocks the autoreceptor "brake" on histamine release.

Experimental task phases and the histaminergic circuit connecting the tuberomammillary nucleus (TMN), mammillary zone, and hippocampus.

The study integrated several advanced techniques:

  • Pharmaco-fMRI: Tracking BOLD signals during memory and learning tasks.
  • Computational Modeling: Utilizing Drift Diffusion Models (DDM) to analyze decision-making parameters (e.g., drift rate and decision policy) and Q-learning to model reinforcement learning.
  • Receptor-Weighted Connectivity: Connectivity data was weighted by regional $H_3R$ density maps derived from PET scans to ensure specificity to the histaminergic system.

Key Results: Connectivity and Computational Shifts

Strengthening the Mammillary-Hippocampal Bridge

A primary finding was the significant strengthening of functional coupling between the hippocampus and the mammillary zone under increased histamine. This circuit is a canonical pathway for memory processing. A machine learning classifier trained on these $H_3R$-weighted connectivity patterns could distinguish between the drug and placebo groups with 88.5% accuracy.

Functional connectivity between the mammillary zone and hippocampus is significantly increased in the active group compared to placebo.

Enhanced Memory and Neural Persistence

Boosting histamine led to improved recognition accuracy ($p = 0.0048$) and more efficient evidence accumulation (increased drift rate). Crucially, histamine sustained activity in the medial entorhinal cortex (MEC) during "offline" rest periods following learning, suggesting a role in stabilizing memory traces before consolidation.

Adaptive Computational Strategies

Under high cognitive load (n-back tasks), histamine increased "non-decision time," indicating an adaptive strategic slowing that allows for better resource reallocation. In reinforcement learning, it reduced the learning rate for losses, effectively stabilizing behavior in aversive environments by making value updates less reactive to negative outcomes.

Discussion and Implications

These findings reveal histamine as a sophisticated "tuning knob" for human cognition. By stabilizing memory and allowing for flexible strategic shifts under pressure, the histaminergic system presents a compelling target for treating cognitive deficits.

Limitations and Future Directions

The study focused on a healthy young cohort; effects may differ in clinical populations with pre-existing histaminergic dysfunction. Additionally, as a single-dose study, the long-term effects of chronic $H_3R$ modulation remain to be explored. Future research must also disentangle the roles of $H_3$ autoreceptors versus heteroreceptors on other neurotransmitter systems. Despite these questions, the research provides a robust causal framework for understanding histamine's role in human neurocomputation, highlighting its significant therapeutic potential for psychiatric and neurological disorders.

Summary updated

Abstract

Histamine was the first canonical monoamine identified in the mammalian brain 1-6, yet arguably remains the least understood in its mechanistic contributions to human behaviour. Using a first-in-class causal probe (H3R inverse agonist pitolisant), we show how increased synaptic histamine shapes offline and online temporal-hippocampal dynamics, sustaining learning-related activity and polarising retrieval computations. At a broader level, elevated histamine adaptively shifts neurocomputational strategy under high cognitive load, while stabilising value updates during aversive reinforcement learning. These findings uncover a mechanistically grounded influence of this underexplored system on human neurocomputation, supporting its therapeutic potential in psychiatry.
BibTeX
@misc{colwell2025synaptic,
  author = {Michael Colwell and Fin van Uum and Philip J. Cowen and Marieke Martens and Michael Browning and Helen C. Barron and Catherine J. Harmer and Susannah E. Murphy},
  title = {{Synaptic histamine shapes the neurocomputational dynamics of human learning}},
  year = {2025},
  doi = {10.1101/2025.10.05.680501},
  howpublished = {bioRxiv},
}
Figure