An agent-based framework is proposed to describe the emergence of complex intelligent systems, starting from active matter and progressing towards increasingly cognitive/intelligent systems. The distributed, concurrent information processing by different types of agents—from physical, chemical, and biological entities to ecosystems, and social systems—in this approach bridges multiple levels of organisation. It provides an interdisciplinary synthesis that explains the role of agents in shaping emergent behaviours as foundations of cognition and intelligence, through developmental and evolutionary processes. The framework offers new insights into the organisation of natural agents and the evolution of natural and artificial intelligent systems.

The capacities we associate with agents originate in active matter, which manifests at various scales, from particles and molecules to entire ecosystems. Phenomena like self-assembly, self-organization, and autopoiesis represent systems’ innate ability to self-maintain and drive increasing structural and functional sophistication [1].

At the most fundamental level, Hewitt’s Actor Model [2] allows us to think of particles and molecules as computational agents involved in a continuous message exchange. Such frameworks show how richly patterned behaviors can emerge from relatively straightforward interactions at different scales. For example, in the molecular domain, as described by Mathews et al. [3], networks of molecules and cellular signaling pathways exhibit forms of memory, problem-solving, and adaptive reprogramming,which are rudimentary cognitive features.

Recognizing cognition as a defining characteristic of living agents connects it directly to the chemical and biological foundations of life [4]. Even relatively simple organisms, such as bacteria, process information from their surroundings, adapt their behavior accordingly, and thus engage in a basic form of cognition [5]. These fundamental computations underpin the emergence of more advanced cognitive processes in complex organisms.

In synthesising of this broad spectrum of agent-based approaches, we arrive at a coherent conceptual framework to investigate fundamental questions: How does intelligence arise naturally? What roles do agents and cognition play in ecosystems and the evolutionary process? How are material agents organised into hierarchies of complexity, and can artificial intelligence extend the cognitive reach of humanity?

Agent-based thinking offers a powerful, integrative tool for understanding understand both natural and engineered systems, illuminating the conditions under which intelligence and complex behaviour emerge. These perspectives set the stage for future research directions, increasing our grasp of the evolving interplay between life, information, and cognitive technologies.

References

1. Dodig-Crnkovic, G. (2017) Cognition as Embodied Morphological Computation. In Vincent C. Müller (ed.), Philosophy and theory of artificial intelligence 2017. Berlin: Springer. pp. 19-23.
2. Hewitt, C. (2010) Actor Model of Computation: Scalable Robust Information Systems. https://arxiv.org/abs/1008.1459
3. Mathews, J., Chang, A. J., Devlin, L., & Levin, M. (2023). Cellular signaling pathways as plastic, proto-cognitive systems: Implications for biomedicine. Patterns (New York, N.Y.), 4(5), 100737. https://doi.org/10.1016/j.patter.2023.100737
4. Ford, B.J. (2023) The cell as secret agent—autonomy and intelligence of the living cell: driving force of Yao. Academia Biology;1. https://doi.org/10.20935/AcadBiol6132
5. Miller, W. B. (2023) Cognition-Based Evolution. Natural Cellular Engineering and the Intelligent Cell. Routledge. Taylor & Francis. https://doi.org/10.1201/9781003286769