A new modelling framework reveals how neuromodulators reshape brain activity

Introduction and research question

Neuromodulators (chemicals in the brain such as dopamine, acetylcholine and serotonin) play a central role in altering how neurons behave, shifting brain states from attention to sleep. They are also key in disease processes - for example, dopamine loss is a hallmark of Parkinson’s disease.

Despite extensive research, neuroscience has lacked a unified model to explain how different neuromodulators reshape the dynamics of neurons across brain regions – that is, how they influence the range of activity states a neuron can express.

A unified modelling approach

A recent research study ‘’A unified model library maps how neuromodulation reshapes the excitability landscape of neurons across the brain’’, supported by the Virtual Brain Twin Project, addresses this challenge - aiming at establishing a framework to allow systematic comparison across neuron types, brain regions, and species.

The team analysed electrophysiological data from seven neuron types across five regions of the human and rodent central nervous system (CNS), under the effects of five neuromodulators - and built a reproducible open-source library of computational neuron models on GitHub.

In addition, they developed a modular Python workflow to describe the electrophysiological data (1) using the Adaptive Exponential Integrate-and-Fire (AdEx) models - a class of spiking neuron models, i.e. mathematical representations that capture how individual neurons fire in response to input signals - under both baseline conditions and during neuromodulation. A

The new framework for understanding neuromodulation can be a useful tool for experimenters and theoreticians, seeking to understand how brain chemistry shapes neural activity.

The researchers break it down for us

To explore the implications of this work, we spoke with Dr Ilaria Carannante, a postdoctoral researcher at NeuroPSI (Paris-Saclay Institute of Neuroscience) in Paris, working in Professor Alain Destexhe’s group. Her research bridges mathematics and computational neuroscience.

1. What problem in neuroscience were you trying to solve with this study?

Neuromodulators (such as dopamine, acetylcholine, serotonin, and many more) strongly influence how neurons behave. They change how easily neurons fire, how they respond to inputs, and how brain circuits operate during different brain states such as sleep, attention, or movement. Neuromodulators are also central to many brain diseases. For example, the degeneration of dopamine-producing neurons is a hallmark of Parkinson's disease.

Because of their importance, neuromodulation has been widely studied and several computational models have been proposed. However, we still lack a cohesive general picture of how different families of neuromodulators affect neuronal dynamics.
In this study, we asked whether the many reported effects of neuromodulators could be understood within a single, unified framework.

To do this, we collected electrophysiological data from published papers and modelled it using the Adaptive Exponential Integrate-and-Fire model (AdEx). This allowed us to compare neuromodulators' effects systematically and understand how they reshape the "excitability landscape" of neurons, in other words how neurons transition between different activity regimes.

2. What is the key insight that someone outside neuroscience should remember from this study?

A key insight is that many complex effects of neuromodulators can be described using only two basic mechanisms. Neuromodulators can either:

switch a neuron into a different mode of activity, or
scale an existing behaviour, without fundamentally changing the type of activity it produces.

In other words, neuromodulators move neurons around within a landscape of possible active states. Recognising these two simple modes helps simplify our understanding of how chemical signals regulate brain activity.

3. Is the novelty mainly methodological (the pipeline), conceptual (the framework), or biological (the findings)?

The work actually combine all three aspects.

We developed a modular, open-source pipeline that integrates experimental electrophysiological data analysis, neuron model fitting, dimensionality reduction, and dynamical system analysis (methodological).

Although AdEx models are widely used, there was previously no systematic pipeline to build them, and in many cases the parameters are manually adjusted. Here, instead, we employ systematic parameter exploration, and the framework is reproducible and reusable for other datasets. Thanks to the unified framework to interpret neuromodulators effect, we could show that neuromodulation operates mainly through two modalities: switching and scaling (conceptual and biological).

4. Why is reducing complex neuromodulatory effects to two dynamical modes (“switching” or “scaling”) an important advance?

Neuromodulation operates via extremely complex mechanisms involving many molecules, receptors and cellular mechanisms. Despite this complexity, we showed that the resulting dynamical effects can be grouped into two modes. This provides a simpler way to interpret how neuromodulators influence neuronal dynamics. This is especially useful when we build large-scale brain simulations.

5. How could this framework be integrated into EBRAINS services or modelling tools?

The framework is well suited for integration into EBRAINS because it provides a standardised way to represent neuromodulation in neuron models.

It provides a library of models ready to use.
It provides a pipeline to develop AdEx models (independently of neuromodulation). Users simply need to provide the electrophysiological traces (or corresponding features) and they are guided in the development of the models as well as their analysis.
It supports neuromodulation-aware large-scale simulations.

The code used to generate the results is available on Github.

Citation:

Guarino D, Carannante I, Destexhe A (2025) A unified model library maps how neuromodulation reshapes the excitability landscape of neurons across the brain. PLoS Comput Biol 21(12): e1013765. https://doi.org/10.1371/journal.pcbi.1013765

(1) Electrophysiological data: Experimental recordings of how real neurons behave electrically are then used to: