Evolving Functional Symmetry in a Three Dimensional Model of an Elongated Organism

Ben Jones, Yaochu Jin, Bernhard Sendhoff, Xin Yao

Introduction

Bilateral symmetry—the mirror-image organization of the left and right sides—is a defining characteristic of the vast majority of animal life. In evolutionary biology, the transition from radial symmetry (exhibited by organisms like jellyfish) to bilateral symmetry (seen in fish, insects, and mammals) represents a pivotal "symmetry-breaking" event. While the fossil record documents the occurrence of this transition, the functional pressures that drove it remain a subject of intense debate. This research investigates whether bilateral symmetry is an inherent requirement for efficient directional movement in a three-dimensional environment by evolving a simulated organism from a "blank slate."

Model Architecture and Methodology

To explore the emergence of symmetry, the researchers developed a 3D physical simulation of an elongated, eel-like "animat." This model serves as a platform to study the coupling between an organism's physical body plan and its nervous system.

Physical Model

The animat is composed of eight segments, each structured as a block of mass points connected by springs. These springs function as muscles, governed by Hooke’s Law with added damping to simulate the realistic physics of movement in a viscous medium like water. The evolutionary algorithm was tasked with determining the "active motor configuration"—essentially deciding which specific springs in the 3D lattice would be controlled by the neural network to produce movement.

Physical model of the 3D animat showing mass-spring segments (a), the full elongated body (b), and the mechanism of muscle activation (c).

Neural Controller

The animat's behavior is governed by a Continuous Time Recurrent Neural Network (CTRNN). These networks are particularly well-suited for creating Central Pattern Generators (CPGs)—neural circuits capable of producing rhythmic motor signals (such as those needed for swimming or walking) without requiring external sensory input. A key feature of this model is spatial encoding: the synaptic weights between neurons are determined by their physical Euclidean distance, mimicking the biological constraints found in real nervous systems.

Evolutionary Results: The Emergence of Symmetry

The researchers utilized an Evolutionary Algorithm (EA) to optimize both the neural architecture and the muscle configuration. The fitness function was straightforward: maximize the distance traveled by the animat within 200 time-steps.

Functional Symmetry Convergence

The most significant finding of the study was the spontaneous emergence of functional symmetry. Although the evolutionary process was free to select any combination of active muscles, every successful run resulted in a bilaterally symmetric configuration. In all six independent trials, the system evolved muscle pairs situated on opposite sides of the body (e.g., top/bottom or left/right).

Examples of evolved bilaterally symmetric motor configurations, showing pairs of active (black) and inactive (white) muscles across different planes.

Movement Strategies and Fitness

The evolved agents did not exhibit the smooth "S-curve" undulation typical of real eels. Instead, they developed a "crumple and extend" motion. By contracting segments and then pushing against the simulated water, they achieved highly effective forward propulsion. The evolutionary trajectory followed a "staircase" pattern: long periods of incremental improvement punctuated by sudden, massive leaps in efficiency whenever a new symmetric innovation was "discovered" by the algorithm.

Discussion and Insights

The study highlights the inextricable link between the nervous system and the body plan. The neural architecture did not evolve in a vacuum; it reflected the physical constraints of the mass-spring body and the environmental demands of swimming.

Furthermore, the results suggest that evolution favors configurations that exploit "passive dynamics." The physical properties of the springs allow the body to return to a neutral state naturally, reducing the neural energy required to maintain posture or reset for the next stroke. While all segments showed rhythmic CPG activity, the front and middle segments acted as the primary drivers, while posterior segments functioned as interneurons to propagate the signal.

Conclusion and Future Directions

This research demonstrates that bilateral symmetry is a robust and perhaps inevitable solution for efficient directional movement in 3D space. Even without pre-defined templates, evolution consistently converges on symmetric motor plans to maximize swimming performance.

Future work will aim to refine the simulation by introducing energy costs for muscle contraction, which may lead to more biologically recognizable undulatory patterns. Additionally, the researchers plan to investigate the actual transition from radial to bilateral symmetry by starting with a "starfish-like" radial model and evolving it into a directional swimmer, providing a real-time look at the symmetry-breaking process.

Summary updated

BibTeX
@inproceedings{jones2008evolving,
  author = {Ben Jones and Yaochu Jin and Bernhard Sendhoff and Xin Yao},
  title = {{Evolving Functional Symmetry in a Three Dimensional Model of an Elongated Organism}},
  booktitle = {ALIFE},
  year = {2008},
}
Figure