Organismal Agency and Rapid Adaptation: The Phenopoiesis Algorithm for Phenotype-First Evolution

Evolutionary success depends on the capacity to adapt: organisms must respond to environmental challenges through both genetic innovation and lifetime learning. The gene-centric paradigm attributes evolutionary causality exclusively to genes, while Denis Noble’s phenotype-first framework argues that organisms are active agents capable of interpreting genetic resources, learning from experience, and shaping their own development. However, this framework has remained philosophically intuitive but algorithmically opaque. We show for the first time that organismal agency can be implemented as a concrete computational process through heritable phenotypic patterns. We introduce the Phenopoiesis Algorithm, where organisms inherit not just genes but also successful phenotypic patterns discovered during lifetime learning. Through experiments in changing environments, these pattern-inheriting organisms achieve 3.4 times faster adaptation compared to gene-centric models. Critically, these gains require cross-generational inheritance of learned patterns rather than within-lifetime learning alone. We conclude that organismal agency is not a philosophical abstraction but an algorithmic mechanism with measurable adaptive value. The mechanism works through compositional reuse: organisms discover how to compose primitive elements into solutions, encode those compositional recipes, and transmit them to offspring. Evolution operates across multiple timescales – fast, reversible phenotypic inheritance and slow, permanent genetic inheritance – providing adaptive flexibility that single-channel mechanisms cannot achieve.

💡 Research Summary

The paper introduces the Phenopoiesis Algorithm, a concrete computational implementation of Denis Noble’s phenotype‑first view of evolution. Traditional gene‑centric models treat organisms as passive carriers of genetic information; adaptation occurs only through mutation and selection, and any learning during an individual’s lifetime influences evolution indirectly via the Baldwin effect. Noble argues that organisms actively interpret their genomes, solve problems through development and learning, and can feed successful phenotypic solutions back into heritable material. Until now this idea has remained philosophical because no algorithmic framework existed to encode, store, and transmit learned solutions across generations.

The authors propose a dual‑inheritance system composed of (1) a genome – a conventional bit‑string encoding parametric information subject to mutation, and (2) an epigenome – a library of compositional recipes discovered during ontogeny. Each organism also maintains a current phenotype built from a fixed library of primitives (horizontal/vertical line segments, corners, etc.). During its lifetime the organism repeatedly (i) reads inherited recipes, (ii) decides whether to exploit them or explore new primitive combinations, (iii) constructs a candidate phenotype, (iv) evaluates fitness against the current environmental target using Intersection‑over‑Union, and (v) writes back any newly discovered high‑fitness recipe into its epigenome. This write‑back creates a bidirectional causal loop: phenotypic success directly shapes heritable material, contrary to the feed‑forward flow of classic evolutionary algorithms.

At the population level, selection operates on the best fitness achieved by each individual during its lifetime. Reproduction passes on a mutated genome together with the unaltered epigenome, allowing offspring to immediately reuse proven recipes. Thus two timescales are coupled: fast, reversible phenotypic inheritance (epigenome) and slow, permanent genetic change (genome).

Experiments were conducted on a 10 × 10 binary grid where the target shape alternated among L, T, plus, cross, and square. All shapes share the same primitive set, making compositional reuse possible. Three treatments were compared: (a) pure genetic evolution (no learning), (b) Baldwin‑effect style evolution (learning but no inheritance), and (c) the Phenopoiesis dual‑inheritance model. Results show that the Phenopoiesis agents adapt to each new shape roughly 3.4–3.8 times faster than the gene‑only agents, and significantly faster than Baldwin agents. Crucially, when epigenomic inheritance was disabled (learning only), the speed advantage vanished, confirming that cross‑generational transmission of learned patterns is the key driver.

The study demonstrates that organismal agency can be operationalized as an algorithmic mechanism with measurable adaptive value. By encoding learned solutions as reusable compositional recipes and allowing them to be inherited, evolution gains a multi‑layered inheritance architecture that combines rapid phenotypic plasticity with long‑term genetic stability. This work bridges the philosophical claim that “organisms write back to their genomes” with a concrete, experimentally validated computational model, opening new avenues for hybrid evolutionary algorithms, adaptive robotics, and the study of multi‑timescale biological evolution.