Simulating the Origins of Purposeful Behaviour in Evolving Protocell Ecologies

Origins of Life Posted 19 August 2025

Image: Flow reactor of the Araudia simulation platform which hosts cross-feeding protocell ecologies

Hydrothermal vent background artwork used with permission of Richard Bizley (bizleyart.com)

This computational modelling work explores a deep and fascinating question at the origins of life: how did the very primitive cell-like entities that preceded the first cells (protocells) develop the ability to survive in unpredictable and changing environments?

Instead of passively reacting to their surroundings, living systems, even simple cells, often act in ways that appear as purposeful or goal-directed. Often strikingly so!

This project investigates how minimal forms of agency (i.e. the capacity to act in a self-directed way) might have first emerged in protocells. Specifically, it looks at how protocells could have evolved internal regulatory systems that help them respond adaptively to environmental challenges, much like living organisms do today.

We developed (from scratch) a new computer model called Araudia which was broadly inspired by conditions thought to have existed at the origin of life, such as those in hydrothermal vent systems. In the model, protocells live and evolve in a simulated flow reactor, an artificial environment where nutrients are continually supplied and washed away. The protocells consume nutrients, grow, divide, and occasionally mutate. Importantly, they can live in a cross-feeding ecology, meaning that they can interact metabolically by exchanging chemical byproducts, which leads to complex interdependencies. The model spans three levels of analysis: metabolism (how cells process resources), ecology (how they interact with each other), and evolution (how populations change over longer timescales).

Image: Population and evolutionary dynamics in the Araudia platform

⬆️ Protocell population dynamics (top) and evolutionary lineage (bottom) in Araudia. The platform was developed in San Sebastián, Spain with "Araudia" being the Basque word for "set of rules/regulations". The platform investigates under what conditions protocells can evolve appropriate codes of conduct to dynamically deal with environmental changes over short time scales.

A key innovation in the model is the presence of a regulatory network in each protocell, implemented as a neural network dynamical system that can implement adaptive behaviour through evolutionary tinkering of the network weights. This network allows protocells not only to adjust their behaviour in response to immediate changes in nutrient availability but also to retain a kind of memory of past conditions, helping them anticipate or prepare for future shifts.

First results of our study (see our paper below) show that when two nutrients periodically fluctuate in the environment in anti-phase, "lac-operon"-like protocell regulatory systems that dynamically devote resources to metabolising the most abundant nutrient do indeed spontaneously evolve over time (for some ecologies).

Image: Emergence of basic protocell regulatory behaviour in an Araudia simulation

⬆️ In this Araudia simulation, after half time, a population of protocells learns to switch their internal enzyme levels (blue and red) up and down in synchrony with the external nutrient availability.

This research is a first step along the road to understanding how the first glimmers of purposeful biological life may have arisen from simpler chemical beginnings.

Araudia Model: How to Install and Use

https://araudia.readthedocs.io

Araudia Software Repository

https://bitbucket.org/ben_s_e/araudia (License: Open Source, GPL v3)

Presentations of This Work

Presentation slides (PDF) from the IAS seminar above are available too. ©2024 Ben Shirt-Ediss.

Journal Paper (in Special Issue)

Shirt-Ediss et al. (2025). Modelling the prebiotic origins of regulation & agency in evolving protocell ecologies. BioRxiv preprint. To appear in Transactions of Royal Society B Special Issue - Origins of life: the possible and the actual.

Mathematical Supplementary Material for the paper.

Techniques Used

Optimised numerical computing, optimised Gillespie stochastic simulation, ordinary differential equations (ODEs), neural ordinary differential equations (NODEs), dimensional analysis, linear programming, custom data analysis tooling, efficient data storage solutions, HPC computing, Python

This work was developed while I was a postdoctoral researcher at Donostia International Physics Centre, San Sebastián, Spain. Thanks to my co-authors for an intense and inspiring collaboration! I wrote the Araudia engine at Talent House, a great initiative.