Mycelium Networks — The Forest's Underground Internet

A fungal network spanning a woodland hectare may contain 300 kilometers of hyphae in a single cubic meter of soil—yet, despite this density, carbon rarely flows directly between trees in the way popular science claims. What exists is not a “wood wide web” of tree talk, but something stranger: a distributed, slow-reacting substrate capable of electrical signaling, memory without a brain, and material self-assembly under Mars-grade radiation. The network is real. The story told about it is not.

The case: nutrient highways or accidental conduits?

Mycorrhizal fungi colonize ~90% of land plant roots, forming symbiotic interfaces where fungi trade soil nutrients (especially phosphorus) for plant-derived carbon. The hyphal network—sometimes kilometers per teaspoon of soil—creates potential continuity between roots. But continuity does not imply directed flow. Experiments introducing isotopic tracers show carbon can move between plants via shared networks, but median transfer is under 0.5% of a seedling's carbon budget. In a 2024 meta-analysis of 28 field studies, only 5 detected transfer—and none linked it to improved seedling survival.

Older "hub trees" are more connected in network models, but whether they actively subsidize younger trees remains unproven. The "mother tree" concept, popularized by Suzanne Simard’s 2011 TED Talk and Finding the Mother Tree (2021), relies on lab and simulation data. No field study has demonstrated preferential carbon allocation to kin via fungal networks. The narrative persists not because it’s falsified, but because it hasn’t been tested at ecosystem scale with controls.

Electrical signaling and proto-architecture

Where evidence strengthens is in electrophysiology. Mycelium conducts voltage spikes—0.5–2 mV, lasting up to 10 minutes—at speeds of 0.5–3 mm/s. These are not action potentials like in neurons, but share waveform properties with neural oscillations. In Omphalotus nidiformis, researchers observed spike trains clustering in patterns that, when analyzed via Zipf’s law, resemble syntactic structure in human language—50 distinct "words" by one metric (Adamatzky et al., FEMS Microbiology Reviews, 2025). This finding is contested: critics argue the analysis conflates randomness with syntax.

More robust is the 7-day oscillation discovered in Pholiota brunnescens (Scientific Reports, 2024). After 60 days of growth toward a nutrient source, a rhythmic electrical cycle emerged—coinciding with directional signal dominance across the network. No known biological clock explains this periodicity.

In spatial tests, Phanerochaete velutina mycelium grown over cross-shaped vs. circular wood block arrangements altered its decay strategy over 116 days—implying geometric recognition. When severed from a colonized food source, the mycelium later regrew toward the original location, suggesting spatial memory is encoded topologically, not electrically.

Fungal computing: memristors and biohybrids

In 2025, Ohio State researchers grew Lentinula edodes (shiitake) mycelium into functional memristors—devices that remember past voltage states. These fungal units switched resistance 5,850 times per second with 90% reliability, retained state after dehydration, and survived gamma radiation doses lethal to silicon chips (PLOS ONE, 2025). The melanin in fungal cell walls—not the hyphae themselves—is likely responsible for radiation resistance, a property first observed in Cladosporium sphaerospermum thriving inside the Chernobyl reactor.

At Cornell’s Organic Robotics Lab (2024), living mycelium was coupled to robotic actuators: electrical spikes triggered directional movement. This biohybrid system responded to light and moisture gradients in real time—proving fungal networks can command machines without digital intermediation.

These are not scalable computers. But they are proofs of concept for neuromorphic materials that self-repair, self-organize, and degrade cleanly—rare traits in conventional electronics.

What's contested

Is it communication or leakage? Most nutrient transfer via mycorrhizal networks may be passive diffusion, not active signaling. The burden of proof for intentional resource routing remains unmet.
Do spike patterns mean anything? While spike sequences are non-random, calling them “language” or “computation” assumes representational capacity. No study has shown mycelium forms internal models of its environment.
Is the network persistent? Field imaging shows mycelial connections are often transient. The continuous “internet” metaphor may be a lab artifact.
Mutualism or parasitism? Some fungi, like Armillaria, exploit the network to invade healthy trees. Others consume 30% of a plant’s carbon without reciprocation. The default assumption of mutual benefit lacks broad empirical support.

Why this has to do with other realms

Myco-architecture is a candidate for tech generation ship design not because it’s alive, but because it builds itself. NASA’s Mycotecture Off Planet project (Phase III NIAC, $2M) plans three-layer structures: ice for radiation, cyanobacteria for oxygen, mycelium for structure. In 2025, a fungal melanin-PLA composite was tested on the ISS (PNAS, 2025) and retained structural integrity under low-Earth orbit conditions. If scaled, astronauts could carry spores weighing under 1 kg to grow habitats on Mars—material, medicine, and life support in one biological package. This turns architecture into germination.

An open question

If a mycelium network can "remember" a food source after disconnection and redirect growth toward it, what is the physical substrate of that memory—synaptic weights, hyphal density gradients, or something else entirely?

Key sources

Karst et al. (2024) meta-analysis — to verify: exact citation pending peer-reviewed publication
Adamatzky, A. (UWE Bristol), fungal computing experiments (2020–2025) — foundational work on fungal spikes and logic gates
PLOS ONE (2025): “Shiitake Mycelium as Memristive Elements” — experimental demonstration of switching behavior
PNAS (2025): ISS mycelium composite test — first in-orbit validation of fungal material viability
Scientific Reports (2024): 7-day oscillation in Pholiota brunnescens — to verify: full dataset access

Abhishek