ruvector/examples/ecosystem-consciousness/RESEARCH.md

Ecosystem Consciousness: IIT Phi as a Food Web Integration Metric

Motivation

Integrated Information Theory (IIT) quantifies how much a system is "more than the sum of its parts" through the measure Phi. Food webs share a structural analogy: a resilient ecosystem cannot be decomposed into independent sub-networks without losing emergent function. This example explores whether IIT Phi correlates with ecological resilience.

Food Web Ecology Background

Trophic Structure

Ecosystems organize into trophic levels:

1. Producers -- autotrophs (plants, algae, coral) that fix energy
2. Primary consumers -- herbivores feeding on producers
3. Secondary consumers -- predators feeding on herbivores
4. Decomposers -- organisms recycling dead matter back to producers
5. Apex predators -- top-level predators with no natural enemies

Resilience and Redundancy

Ecological resilience depends on:

• Functional redundancy: multiple species filling similar roles
• Response diversity: different species respond differently to perturbation
• Connectivity: dense interaction networks buffer against single species loss
• Keystone species: removal causes disproportionate collapse

IIT as a Resilience Metric

TPM Construction

We model each species as a "state" and construct the transition probability matrix from energy flow weights:

TPM[i][j] = P(energy flows from species j to species i)

Row-normalization ensures each row sums to 1, giving a proper stochastic matrix.

Phi and Ecosystem Integration

• High Phi: the food web cannot be split into independent sub-networks -- every partition loses significant information about the whole
• Low Phi: the ecosystem decomposes into weakly connected modules -- removing one module barely affects the rest

Species Contribution

We define the "Phi contribution" of species k as:

C(k) = Phi(full) - Phi(without k)

Species with high C(k) are "consciousness keystones" -- their removal most reduces the integrated information of the web.

Expected Results

Tropical Rainforest (12 species)

• Dense cross-trophic connections and nutrient cycling
• Many redundant pathways between trophic levels
• Prediction: HIGH Phi, relatively uniform contributions

Agricultural Monoculture (8 species)

• Sparse, linear food chains
• Single crop dominates energy flow
• Prediction: LOW Phi, highly concentrated contributions

Coral Reef (10 species)

• Moderate connectivity centered on coral as structural keystone
• Removing coral should cause largest Phi drop
• Prediction: MODERATE Phi, coral has disproportionate contribution

Causal Emergence in Ecosystems

Beyond Phi, we compute causal emergence to ask: does the ecosystem have a "macro-level" description (e.g., trophic levels) that is more informative than the species-level description?

• High causal emergence suggests natural macro-level organization (trophic levels are real causal entities, not just labels)
• Low causal emergence suggests species-level dynamics dominate

Limitations

1. Synthetic data: real food webs have stochastic, seasonal dynamics
2. Static TPM: IIT assumes a fixed transition structure
3. Small system sizes: Phi is computationally expensive (exponential in system size), limiting analysis to ~15 species
4. Directionality: IIT Phi is defined for mechanisms, not flows -- the food web analogy is suggestive, not rigorous

References

• Tononi, G. (2008). Consciousness as Integrated Information: a Provisional Manifesto. Biological Bulletin, 215(3).
• May, R.M. (1973). Stability and Complexity in Model Ecosystems.
• Dunne, J.A. et al. (2002). Food-web structure and network theory.
• Hoel, E.P. et al. (2013). Quantifying causal emergence shows that macro can beat micro.