The ECSU will host a workshop at the ALIFE 2025 Conference in Kyoto, Japan. The 2025 Conference on Artificial Life, titled Ciphers of Life, will be held from 6-10 October 2025. The program can be found here. The ECSU's workshop TONAL 2025 can be found here.
Please direct any questions to the organizers at the following email address: ecsu.alife@oist.jp
Monday, 6 of October
First session (Hybrid)
1. “Opening Remarks“ by workshop organizers
2. "Playing Dice with the Reaper: How Viability Boundaries Become Fuzzy" by Connor McShaffrey
3. “Creativity in Learning-Driven Self-Organization: The Impact of Resets and Learning in Hopfield Networks“ by Natalya Weber
4. "Inspecting Neural Noise: Entropy in EEG as a Marker of Deliberate Decisions" by Milan Rybar [in-person]
5. "Noise in multi-scale alignment" by Mark James
Second session (all in-person)
1. "Indeterminism in Large Language Models: An Unintentional Step Toward Open-Ended Intelligence" by Georgii Karelin
2. “Stochastic Switching Promotes Open-Endedness in Sparse Random Boolean Networks” by Amahury Jafet Lopez Diaz
3. “Another Kind of Clay: Intelligence as the Shaping of Intrinsic Noise“ by Benjamin Gaskin
4. “Introduction to irruption theory and the role of noise in living systems“ by Tom Froese
5. Open discussion with speakers and invited guests
Indeterminism in Large Language Models: An Unintentional Step Toward Open-Ended Intelligence
What if noise is not merely tolerated by life, but embraced—perhaps even essential?
This workshop explores the generative and functional roles of randomness and noise in artificial life (ALife). Our central conjecture is that stochasticity is not just a background condition or a nuisance to overcome, but a vital and irreplaceable component of living systems—natural and artificial alike.
From random mutations driving evolution to chaotic dynamics in numerical simulations, unpredictability often sits at the heart of emergent complexity. In ALife, randomness appears in various forms: it may be exogenous (as procedural input), or endogenous (emerging from the system itself during its evolution). In either case, it poses a critical question: If artificial and natural life persistently co-exist with randomness, how might they benefit from it—or even thrive on it?
We invite discussion on whether some noise is actively generated or amplified by living systems and whether such noise plays a meaningful, causally efficacious role in the system’s operational closure. By better understanding these dynamics, we may be able to engage with artificial living systems in richer, more meaningful ways.
This workshop aims to build a community around exploring how ALife can better harness randomness and noise. We welcome contributions on topics including (but not limited to):
Random variation and novelty: Mechanisms by which randomness fuels creativity and open-ended evolution in natural and artificial systems.
Homeostasis through noise: Strategies for maintaining robustness and functional stability amid dynamic, chaotic environments.
Beneficial noise and flexibility: The ways in which variable perturbations enhance adaptive potential and responsiveness.
Intrinsic sources of randomness: Origins of noise within biological and artificial systems, and their operational significance.
“Edge of chaos” revisited: Historical and current perspectives on criticality, complexity, and the role of noise in system transitions.
Cross-scale integration: How noise contributes to alignment and integration across biological, cognitive, and social scales.
Engineering stochasticity: Techniques for designing or inducing beneficial randomness in artificial systems to support open-endedness and autonomy.
Dr. Tom Froese
Associate Professor at the Okinawa Institute of Science and Technology Graduate University (OIST), where he heads the Embodied Cognitive Science Unit (ECSU). His research spans theoretical, computational, and experimental approaches to life, mind, and sociality.
Dr. Mark James
Philosopher and cognitive scientist in ECSU, OIST. He specializes in how multi-scale dynamics shape health, identity, and behavior. As co-creator of the Wayshaping framework—a theoretically grounded approach to behavior change rooted in complex systems and embodied cognition—he explores change as a process of collaborative negotiation across biological, psychological, and social levels. He is especially interested in how noise enables alignment across these scales.
Georgii Karelin
PhD student in ECSU, OIST, with a background in theoretical astronomy. His interests include simple computational models (e.g., cellular automata, agent-based models) and the intersection between artificial life and astrobiology.
Cognitive Science, Indiana University
(coming soon)
Embodied Cognitive Science Unit, Okinawa Institute of Science and Technology
What role does noise play in creativity? We address this question by using the Self-Optimization (SO) model, which can be considered a third operational mode of the classical Hopfield Network. This model leverages the power of associative memory and noise to enhance problem-solving capabilities. Our findings indicate that this simple, learning-driven self-organized system can generate novel and appropriate solutions, thereby demonstrating creativity. We investigate how different hyperparameters influence the outcomes, allowing us to simulate and study the emergence of creative potential in artificial systems. We argue that the SO model shows how noise, rather than being a hindrance, can serve as a constructive force in learning-driven systems.
Embodied Cognitive Science Unit, Okinawa Institute of Science and Technology
Decision-making in humans is often studied through contrasts between arbitrary choices, which lack consequence, and deliberate choices, which are reasoned and meaningful. Electroencephalography (EEG) studies showed readiness potentials before arbitrary but not deliberate choices, suggesting distinct underlying mechanisms. We re-analyzed EEG data from a donation-choice task using multiple entropy and complexity measures.
Our results showed differences in post-stimulus entropy dynamics between decision types. Deliberate choices showed higher entropy and complexity in comparison with arbitrary decisions. These findings reframe neural “noise” as a functional component of cognition rather than a nuisance variable. Entropy does not merely quantify randomness in brain signals; it reveals how unpredictability is differentially recruited when decisions are meaningful versus arbitrary. We propose that entropy-based measures can serve as biomarkers of agential involvement, highlighting the role of noise in organizing adaptive behavior.
Embodied Cognitive Science Unit, Okinawa Institute of Science and Technology
Living systems change when patterns are perturbed just enough to loosen entrenched regimes without breaking coordination. This talk develops a theory of dosed noise as a mechanism for realignment across biological, psychological, and social levels. It frames noise, in this context, as a brief, well-timed fluctuation that opens a window for re-coordination of attention, affect, narrative, and action, Using familar examples, we show how endogenous and exogenous shocks and scaffolds can be tuned to canalize these transitions at different scales.
Embodied Cognitive Science Unit, Okinawa Institute of Science and Technology
Synergy between stochastic noise and deterministic chaos is a canonical route to unpredictable behavior in nonlinear systems. This letter analyzes the origins and consequences of indeterminism that has recently appeared in leading Large Language Models (LLMs), drawing connections to open-endedness, precariousness, artificial life, and the problem of meaning. Computational indeterminism arises in LLMs from a combination of the non-associative nature of floating-point arithmetic and the arbitrary order of execution in large-scale parallel software-hardware systems. This low-level numerical noise is then amplified by the chaotic dynamics of deep neural networks, producing unpredictable macroscopic behavior. We propose that irrepeatable dynamics in computational processes lend them a mortal nature.
Irrepeatability might be recognized as a potential basis for genuinely novel behavior and agentive artificial intelligence and could be explicitly incorporated into system designs.
The presence of beneficial intrinsic unpredictability can then be used to evaluate when artificial computational systems exhibit lifelike autonomy.
School of Systems Science and Industrial Engineering, Binghamton University
Open-ended evolution (OEE) refers to the capability of a system to continually generate novel and increasingly complex structures or behaviors without settling into fixed or cyclic states. Biological systems exemplify OEE through their potential to evolve emergent properties such as new cell types, metabolic pathways, or novel phenotypic traits. Despite widespread interest in theoretical biology, there is still no widely accepted conceptual or formal definition and measurement framework for open-endedness [1]. Using Random Boolean Networks (RBNs)—a generalized model of discrete dynamical systems—we introduce a novel metric for quantifying open-endedness that extends previous approaches [2]. This metric distinguishes genuine novelty from mere noise by capturing the persistence and emergence of new attractors across various RBN logics. Recognizing that open-endedness is inherently a product of evolutionary processes, we further explore the underlying mechanisms driving this phenomenon [3]. Our preliminary analysis evaluates five candidate mechanisms contributing to open-ended evolution, highlighting stochastic switching as particularly effective in promoting increased open-endedness within RBNs (see Fig. 1). Inspired by biological evidence that stochastic switching serves as an adaptive survival strategy in fluctuating environments [4], we critique prevailing approaches to open-endedness focused solely on developing adaptive syntactic structures. These approaches typically emphasize the evolution of formal rule sets, pattern generation, and complexity growth within predefined symbolic frameworks. Instead, we argue that achieving genuine OEE requires the incorporation of adaptive semantic processes [5]—that is, mechanisms by which systems construct and evolve their own relationships to the external world through the development of new sensors, effectors, and interpretive mappings.
[1] Pattee, H. H., & Sayama, H. (2019). Evolved open-endedness, not open-ended evolution. Artificial Life, 25(1), 4-8. [2] Gershenson, C., & Fern´andez, N. (2012). Complexity and information: Measuring emergence, self-organization, and homeostasis at multiple scales. Complexity, 18(2), 29-44. [3] Stepney, S., & Hickinbotham, S. (2024). On the open-endedness of detecting open-endedness. Artificial Life, 30(3), 390-416. [4] Acar, M., Mettetal, J. T., & Van Oudenaarden, A. (2008). Stochastic switching as a survival strategy in fluctuating environments. Nature genetics, 40(4), 471-475. [5] Cariani, P. A. (1989). On the design of devices with emergent semantic functions (Doctoral dissertation, State University of New York).
History and Philosophy of Science University of Sydney, Australia
In 1952, John von Neumann presented a series of lectures in which he contended that error, thus far viewed as “an extraneous and misdirected or misdirecting accident” to the role of logics in automata, should instead be understood as an “essential part of the process.” These lectures, on “the synthesis of reliable organisms from unreliable components,” were concerned with the differences between biological computing and the machinic. Here we will take this point of departure and extend the analysis by proposing a theoretical synthesis concerning the nature of noise in biological systems across scales: molecular, genetic, cellular, individual, and even social. We will present a coherent view of life, agency, and intelligence across these scales as being essentially concerned not with the elimination of error and noise but rather with the harnessing and constraint of this fundamental activity. To this end, we will describe an account of intelligence in particular with reference to Howard Pattee’s biosemiotics: whereby meaning in biological systems comprises the selected function of physical symbols as constraining the dynamics of a system. This involves our extending Pattee’s work from genetic and linguistic systems to sensory processes, thus taking the signals produced by transduction events—whether chemical or neural—as symbolic constraints in this sense. Finally, taking up this framework, we will aim to show its fruits by examining a few case studies; with a particular focus on the nature and role of intrinsic activity in neural organoids; we will see how what some early studies suspected to be pathological activity may, when seen through this lens, have implications for the design of artificial neural architectures. Our aim, in sum, is to derive an alternative paradigm to the passive input–output view which was inherited from the mechanical and computational schools and sees intrinsic activity as noise; instead proposing an account in which intrinsic activity is the clay from which intelligence is made.
Embodied Cognitive Science Unit, Okinawa Institute of Science and Technology
I will present the basic axioms and concepts of irruption theory, which is a novel nonreductive framework for capturing agency in living systems. In a nutshell, what motivates this theory is following train of thought: if agency is taken to be efficacious, and if it is not reducible to underlying non-agential factors, then the way that this irreducible efficacy shows up at that underlying level is in terms of unpredictable deviations from physiological tendencies that would otherwise take place. The principled derivation of these deviations - irruptions - provides a fresh perspective on the source of the noisiness of living systems, and highlights the essential role noise plays in the self-organization of adaptive behavior.
