Evolution-Scaling Hypothesis

• The Evolution-Scaling Hypothesis is a framework formalizing how evolutionary processes repurpose local homeostatic agents to produce emergent global structures and power-law scaling.
• It explains pattern formation in systems from multicellular morphogenesis to urban infrastructures by evolving distributed error-minimization loops into collective, goal-directed behaviors.
• Simulations and biological experiments demonstrate that evolutionary tuning of networked systems leads to robust, adaptive scaling phenomena in metabolism, regeneration, and supply networks.

The Evolution-Scaling Hypothesis formalizes how fundamental scaling laws arise in evolving biological, artificial, and ecological systems through underlying evolutionary (or adaptive) processes. At its core, the hypothesis proposes that simple, local units or agents governed by homeostatic or adaptive dynamics are evolutionarily repurposed, combined, or tuned so as to produce emergent global structures, goal-directed behaviors, or regularities—often power-law scaling—that are not explicitly encoded at the microscopic level. The principle applies across diverse domains, from cell collectives and metabolic networks to supply infrastructures and information-processing systems.

1. Theoretical Foundations and General Principle

The Evolution-Scaling Hypothesis, as formalized in the context of the TAME (Technological Approach to Mind Everywhere) framework, asserts that evolution acts on distributed, cybernetic homeostatic controllers at the cellular (or agent) scale to construct higher-order collective behaviors and goal states. This process is not merely additive but involves the evolutionary "pivoting" of local error-minimization loops (e.g., homeostasis to setpoint) into coupled, multicellular agents able to navigate complex morphospaces and achieve tissue- or organism-level goals unavailable to their parts (Pio-Lopez et al., 2022).

More generally, the hypothesis posits that evolutionary dynamics can systematically drive the emergence of higher-level scaling laws—typically power laws of the form

$Y(x) = Y_0\,x^\beta$

where $Y$ is a macroscopic property (e.g., metabolic rate, network volume, collective computation), $x$ is a fundamental organizational parameter (e.g., mass, population size, cell number), and $\beta$ captures the constraint imposed by physiology, network structure, or evolutionary pressure (Hamilton, 26 Jun 2025, Shestopaloff, 2016, Shestopaloff, 2016, Meirelles et al., 2018).

2. Formalization: From Homeostatic Agents to Emergent Scaling

The prototypical example is found in multicellular morphogenesis:

• Cell-level homeostasis: Each cell $i$ maintains an internal setpoint $s_i$ and responds to deviations $e_i(t) = s_i - x_i(t)$ through corrective outputs $u_i(t) = K_i\,e_i(t)$ or, in complex implementations, via an artificial neural network (ANN) controller. Cells can communicate internal states, stress signals, and morphogens via gap junctions and signaling molecules.
• Collective evolution: Under evolutionary selection, these cells are evolved in silico to minimize a tissue-level error function, $E_{\text{tissue}}(t) = \sum_{i=1}^N \delta_i(t)$, where $\delta_i$ encodes whether a cell's state matches the collective morphological goal (e.g., the classic French Flag pattern).
• Emergence of higher-level goals: Through the aggregation of local error signals and the evolution of inter-cellular communication, the tissue acquires the ability to robustly reach, and recover, a target anatomical configuration. These emergent collective competencies are not simply the sum but a scaling-up of individual homeostatic drives (Pio-Lopez et al., 2022).

The generalization to metabolic, ecological, and infrastructure networks proceeds similarly: local agents adapt or evolve under resource constraints, and the resulting global distribution embodies power-law scaling relations whose exponents β reflect the ecological, physical, or developmental tradeoffs intrinsic to the system (Shestopaloff, 2016, Hamilton, 26 Jun 2025, Cheng et al., 2020).

3. Quantitative Examples and Simulations

Several model systems demonstrate the operationalization of the hypothesis:

(a) Morphogenetic Collective Agents

• Evolutionary simulations using ES-HyperNEAT/CPPN encoding evolve ANN controllers for all cells in a 2D tissue grid.
• Tissue-level fitness is defined geometrically (stripe accuracy in the French Flag task).
• Key quantitative results: evolved tissues achieve $>$94% accuracy in stripe patterning, robust recovery to perturbations, and long-term "allostatic" remodeling—emergent properties not directly selected for (Pio-Lopez et al., 2022).

(b) Bioenergetic Scaling in Food Chains

• Metabolic allometric scaling ($B \propto M^\beta$) is shown to result from the optimal evolutionary partitioning of resources within a food chain.
• β depends on geometric, biomechanical, and ecological parameters, integrating limb length, skeleton mass, and maximal speed. Empirically derived exponents for mammals, reptiles, fish, and birds all align closely with model predictions, confirming that interspecific scaling exponents emerge from evolutionary optimization across the network (Shestopaloff, 2016).

(c) Urban and Supply Network Scaling

• Water supply networks, cities, and other complex infrastructures display scaling laws (e.g., network volume vs. pipe length $V \propto L^{5/4}$) at both the individual (historical/temporal) and population (cross-sectional) levels.
• Individual adaptation trajectories parallel the population scaling law, demonstrating that global scaling at the macrosystem level is the aggregate result of sequential, path-dependent local adaptations ("projectiles" along an evolutionary track) (Cheng et al., 2020, Meirelles et al., 2018).

System Type	Evolved/Observed Scaling	Mechanism
Morphogenetic tissue	Stripe accuracy $\sim$ 95%, emergent long-term repair	Cell-level homeostasis scaled by evolutionary selection and communication
Animal metabolism	$B \propto M^\beta$; $\beta = 0.87$ (mammals)	Evolutionary optimization of energy partition in food chain
Urban networks	$V \propto L^{5/4}$, $d̄ \propto L^{1/8}$	Individual dynamic evolutions cumulate into population-level scaling

4. Biological and Experimental Evidence

The hypothesis is supported by both simulation and biological experiment:

• Planarian regeneration: After induced headlessness, a fraction of planaria spontaneously repattern head structures long after stabilization—mirroring the simulated tissues' ability to undergo unprompted morphological correction driven by intrinsic stress dynamics. Neural reformation proceeds via distributed signaling akin to evolved networked agents (Pio-Lopez et al., 2022).
• Allometric scaling in unicellulars and multicellulars: Direct measurements of metabolic exponents across unicellular taxa and comprehensive model reconstructions for mammals show close correspondence to evolutionary-scaling predictions, in both maximal and basal metabolic states (Shestopaloff, 2016).

5. Implications and Predictions

The Evolution-Scaling Hypothesis explains why a multitude of biological, ecological, and engineered systems exhibit robust power-law scaling laws or universal scaling exponents across orders of magnitude:

• Goal scaling: Simple homeostatic loops are sufficient for the evolutionary emergence of novel, higher-order goal-directed behavior at collective scales, without the need for explicit coding of global objectives.
• Robustness and allostasis: Evolved networked systems display robustness to acute perturbation and can maintain long-term stability or execute delayed, spontaneous remodeling (allostasis), similar to the biological phenomena of regeneration and repair.
• Generality: The same evolutionary rules underlie scaling in pattern formation, metabolic networks, social infrastructure, and collective computation; predictions extend to biofilm patterning and somitogenesis (Pio-Lopez et al., 2022).
• Design principles: Synthetic systems incorporating minimal homeostatic control and shared "stress" signals are predicted to recapitulate the evolvability, robustness, and self-repairing attributes of biological morphogenetic agents.

6. Broader Context and Limitations

The unification of control theory, active inference, and evolutionary developmental dynamics under this hypothesis highlights that emergent large-scale regularities—such as error minimization, scaling exponents, and robustness—arise as "free lunch" outcomes of multi-scale homeostatic architectures shaped by selection, not as direct targets of selection themselves. The framework thus provides powerful explanatory and predictive tools for both natural and engineered complex adaptive systems.

Limitations include reliance on sufficient inter-agent communication, the need for continued selection or environmental pressure to maintain stability, and the dependence of specific scaling exponents on detailed system constraints and boundary conditions. Extensions are expected to yield further insight into the scaling and emergence of cognition, behavior, and structural complexity across the phylogenetic and organizational hierarchy (Pio-Lopez et al., 2022).