Skip to main content

Ecology, Learning, and Evolution in PredPreyGrass

Morphological, Behavioral, and Life-History Perspectives (Clarified)​

This document is an extensive, self-contained synthesis of the discussion about what kind of evolution is (and is not) present in the current PredPreyGrass environment.

It explicitly clarifies the distinction between morphological (physiological) traits, behavioral traits, and life-history traits — including how energy-related parameters can belong to different categories depending on their role.

The goal is conceptual precision, not prescription.

1. Darwinian evolution: a minimal definition​

Darwinian evolution requires three ingredients:

  1. Variation
  2. Inheritance
  3. Differential survival and reproduction

Whenever these are present, selection operates.
The key question is which traits vary and are inherited.

2. Three axes of Darwinian evolution​

Darwinian evolution is most cleanly decomposed into three orthogonal axes:

  1. Morphological (physiological) evolution — what an organism is
  2. Behavioral evolution — what an organism does
  3. Life-history evolution — how resources are allocated over time

This triad is standard in evolutionary biology and maps cleanly onto artificial agent systems.

3. Morphological (physiological) evolution​

3.1 Definition​

Morphological evolution is selection on phenotype-defining physical and physiological traits.

These traits answer the question:

How costly is it for this body to exist and act?

They define the physics of the organism and shape the state-transition dynamics of the environment.

3.2 Morphological traits in PredPreyGrass​

In PredPreyGrass, morphological (physiological) traits include:

  • movement speed and movement cost
  • energy cost per action
  • basal metabolic cost per time step
  • efficiency of converting prey/food into energy
  • maximum energy storage capacity
  • sensing radius and observation geometry
  • action affordances
  • interaction mechanics (e.g. kill thresholds)

If two agents take the same actions in the same environment but lose energy at different rates, this difference is morphological / physiological.

4. Behavioral evolution​

4.1 Definition​

Behavioral evolution is selection on strategy, holding morphology constant.

These traits answer the question:

Given the same body, what actions are chosen?

Examples (biological):

  • foraging strategies
  • social and cooperative behavior
  • risk-taking vs avoidance

Examples (artificial agents):

  • policies
  • decision rules
  • action-selection strategies

4.2 Learning versus behavioral evolution​

A crucial distinction:

  • Learning modifies behavior within a lifetime
  • Behavioral evolution modifies which behaviors are inherited

Learning alone does not constitute behavioral evolution unless behavioral variation is heritable.

4.3 Behavioral evolution in PredPreyGrass (current state)​

In the current setup:

  • policies are shared per agent type
  • PPO updates policies continuously
  • offspring immediately use the shared policy

As a result:

  • behavior adapts via learning
  • behavior does not evolve via inheritance
  • no strategy can go extinct as a strategy

This is ecology with learning, not behavioral evolution.

5. Life-history evolution​

5.1 Definition​

Life-history evolution concerns traits that regulate timing and allocation across the lifespan, not immediate action choice.

These traits answer the question:

When should energy be spent on survival, growth, or reproduction?

They are often implemented as parameters, not decisions.

5.2 Life-history traits in PredPreyGrass​

In PredPreyGrass, life-history traits include:

  • reproduction energy threshold
  • reproduction energy cost
  • reproduction cooldown
  • aging rate or maximum age
  • starvation death rules

These traits:

  • strongly affect fitness
  • are heritable per agent type
  • are not under direct policy control

They therefore belong to life-history evolution, not behavior.

Energy-related variables can appear in both morphology and life-history, depending on their role.

6.1 Energy as physiology (morphology)​

Energy parameters belong to morphology when they describe costs of existence or action, such as:

  • basal energy decay per time step
  • energy cost per movement or attack
  • conversion efficiency from food to energy

These answer:

How expensive is this body to run?

6.2 Energy as life-history constraint​

Energy parameters belong to life-history when they regulate lifespan or reproduction timing, such as:

  • energy thresholds for reproduction
  • energy depletion leading to starvation death
  • age-related decline expressed via energy limits

These answer:

How long can this organism persist, and when can it reproduce?

6.3 Why the same variable can play different roles​

A single variable (e.g. energy_decay) may:

  • represent metabolic cost → morphology
  • represent aging or lifespan pressure → life-history

This is common in models but conceptually important to recognize.

7. What the current model implements evolutionarily​

Putting everything together, the current PredPreyGrass model implements:

  • ✔ ecological dynamics
  • ✔ death and reproduction
  • ✔ demographic selection
  • ✔ morphological (physiological) evolution between types
  • ✔ within-lifetime behavioral learning

It does not implement:

  • ✘ heritable behavioral variation
  • ✘ selection between strategies
  • ✘ policy extinction or fixation

The correct classification is:

Ecological population dynamics with learning and morphological selection

8. Why reproduction timing is not a strategy​

In the current environment, reproduction follows a rule of the form:

IF energy ≥ threshold AND cooldown passed:
    reproduce()

Key consequences:

  • reproduction is not an action
  • the policy cannot refuse reproduction
  • timing is determined by state, not choice

Therefore:

Reproduction is a mechanism, not a behavioral strategy

Policies can only influence reproduction indirectly via survival and energy acquisition.

9. Contrast with Genetic Algorithms and PBT​

9.1 GA / PBT​

In Genetic Algorithms and Population-Based Training:

  • the population consists of policies
  • policies reproduce, mutate, and are replaced
  • inheritance is explicit
  • poor strategies are removed

Selection acts directly on strategies.

9.2 PredPreyGrass​

In PredPreyGrass:

  • the population consists of agents (bodies)
  • policies are shared
  • reproduction does not transmit strategies
  • death removes individuals, not strategies

Thus:

Selection removes bodies, not behaviors

10. The minimal boundary to behavioral (policy) evolution​

The system crosses into behavioral evolution if and only if:

  1. Multiple policy variants exist within the same morphology
  2. Policy identity is heritable
  3. Differential reproduction changes policy frequencies

Mutation is optional; inheritance and selection are sufficient.

11. Diagnostic question​

A single question distinguishes ecology-with-learning from behavioral evolution:

Can two agents with the same body but different policies compete, reproduce, and change their relative frequencies over time?

  • No → ecology + learning
  • Yes → behavioral evolution

12. Final synthesis​

Darwinian evolution is not monolithic.

It acts on:

  • Morphology / physiology (what organisms are)
  • Behavior (what organisms do)
  • Life-history traits (how organisms allocate effort over time)

Your current PredPreyGrass model already captures:

  • morphology-based selection
  • ecological dynamics
  • learning-driven behavioral adaptation

Behavioral evolution begins only when strategies themselves become heritable and selectable.

Final takeaway​

Energy-related parameters are morphological when they describe physiological costs, and life-history traits when they regulate lifespan or reproduction.
PredPreyGrass currently implements ecological and morphological evolution with learning; behavioral evolution requires heritable strategy variation.