Causal Architecture Dynamics Prior to Arrival of Self-replicators in a Model of Catalytic Networks Relevant to Origin-of-Life

Abstract: Agents that exert causal power in the world are thought to be the product of selection among diverse replicators; what is the causal structure of a medium before replicators appear, and evolution takes hold? We studied information-theoretic dynamics of the popular GARD model that captures several dynamics thought to be important at life's origin and found that causal emergence predicted the initial appearance of self-replication. Moreover, interventions that drove causal emergence up increased the longevity of self-replicators, while interventions that drove causal emergence down decreased the abundance of these self-replicators, suggesting causal emergence as a functional control knob. Thus, progressive increases in integrated causality are detectable in active media before evolutionary dynamics begin to operate, which may have implications for the origin of life across highly diverse scenarios and for understanding the forces driving the rise in causal power observed in the biosphere on Earth.

• The paper demonstrates that increased causal emergence predicts the onset of self-replicative dynamics in prebiotic catalytic networks.
• It applies the GARD model with time-lagged partial information decomposition, confirming notable correlation (73% runs) and significant p-values (<0.001).
• Intervention experiments establish causal emergence as a 'control knob' that can enhance the persistence and consistency of emerging self-replicators.

Causal Emergence and the Pre-replicator Phase in Catalytic Networks

Introduction

Understanding the transition from prebiotic chemistry to biological self-replication is central to origin-of-life (OoL) studies. Prevailing paradigms ascribe the emergence of agency to natural selection acting on replicators, yet the properties of the underlying pre-replicator medium—particularly its causal architecture—are poorly characterized. This paper analyzes the Graded Autocatalysis Replication Domain (GARD) model, applying information-theoretic measures of causal emergence to probe the dynamics preceding the advent of self-replicators. The findings suggest that increases in integrated causality within catalytic networks both anticipate and modulate the appearance and persistence of self-replicating molecular assemblies.

Methods: GARD Model and Information-Theoretic Analysis

The GARD model simulates the growth and fission cycles of molecular assemblies composed of multiple molecule types, capturing key features hypothesized to underlie prebiotic evolution, notably compositional (as opposed to sequence-based) heredity. Here, the causal emergence metric is computed via a time-lagged partial information decomposition (PID), detecting predictive information generated by the system as a whole but irreducible to its parts. Data preprocessing involves a centered log-ratio transform to address closure constraints in compositional data.

One hundred independent GARD simulations were performed, each tracking assembly composition and classifying system states as either self-replicator or drift, according to established similarity thresholds. The causal emergence measure ("$+$") was computed for each molecular step, and its temporal properties were compared to network-theoretic and dynamical systems metrics to establish its uniqueness.

Results: Correlation and Predictive Power of Causal Emergence

A key empirical finding is that, while aggregate causal emergence over all runs shows no significant trend, individual runs exhibit pronounced, punctuated spikes in $+$ just prior to the emergence of self-replicators. Across runs, 73% show positive correlation between $+$ and self-replication, with statistical significance in most cases.

Causal emergence is consistently higher during self-replicating states than in drift states (significant in 57/100 runs; combined $p < 0.001$ via Fisher’s method), and $+$ time-series reject white-noise nulls (Ljung-Box $p \approx 10^{-51}$), demonstrating persistent temporal structure. Notably, machine learning models using early $+$ trajectories as inputs significantly outperform baselines in predicting future self-replication, confirming that the information encoded in $+$ contains predictive signal regarding the system's transition to self-replicative dynamics.

Interventions and Causality

The pivotal claim of the work is that causal emergence is not only predictive but also mechanistically relevant. Intervention experiments demonstrate that systematically maximizing causal emergence after each assembly fission increases both the persistence (mean $874 \pm 233$ steps vs. control $716 \pm 198$) and consistency (Pearson $+$0 vs. control $+$1) of self-replicators, and leads to an accumulating increase in the probability of self-replication over time ($+$2). Conversely, minimization of $+$3 reduces both persistence and overall probability ($+$4 vs. control $+$5) and consistency.

Therefore, $+$6 acts as a functional "control knob": modulating the degree of causal emergence within a prebiotic system causally regulates the likelihood and robustness of emergent self-replicators.

Implications and Theoretical Significance

These findings challenge the canonical view that selection acts as the sole engine driving the rise of causal power in biological systems. The detection of increasing integrated causality preceding replication suggests that organizational and information-theoretic ordering principles, distinct from selection, underpin the earliest emergence of agency. This adds quantitative support to metabolism-first and information-based OoL scenarios and highlights the substrate-independence of causal emergence as a theoretically general principle potentially relevant to non-genetic, non-energetic, and even synthetic or artificial chemical systems.

Practically, these insights suggest routes for manipulating the emergence of replicators in engineered or natural systems. Applications may range from steering prebiotic chemistry in vitro, to the design of synthetic protocells, or even controlling dynamical transitions in non-biological complex systems. Further, the paper’s approach unifies dynamical, network, and information-theoretic methodologies, showing limitations of classical graph-based or entropy-based metrics in capturing causal structure relevant for the onset of biological functions.

Future Directions

The results open several lines for future research. First, examining whether analogous causal emergence signatures presage other major evolutionary transitions—e.g., genome compartmentalization or multicellularity—could establish whether this is a general prelude to increases in biological individuality. Second, leveraging $+$7 as an order parameter may improve the real-time prediction or even prevention of regime shifts (e.g., in ecosystem collapse or pathological states). Finally, the robust agnosticism of the measure to material implementation suggests utility in AI, nanotechnology, and other domains involving the transition from passive substrates to agentic systems.

Conclusion

This study demonstrates that quantifiable, punctuated increases in causal emergence are detectable in catalytic networks prior to the arrival of self-replicators. These increases both predict and modulate the onset and persistence of self-replicative dynamics. The findings support a view in which integrated causal structure serves as a precursor and enabler of biological agency, independently of genetic heredity or natural selection, and provide operational means to influence the emergence of replicators within complex dynamical systems.