Place wood blocks in a circle on soil. Inoculate with Phanerochaete velutina, a wood-eating fungus. Wait several months. The mycelial network grows along the blocks, connecting them — and steers completely clear of the circle's center.
Fukasawa et al. (Fungal Ecology, 2024) showed that the fungus maintains the geometric pattern of its food source in its growth architecture. Place the blocks in a cross, and the network preserves the cross. The fungus doesn't just grow toward resources. It navigates the negative space — the interior region defined by the absence of food in a pattern the network has somehow encoded.
This distinction matters. Growing toward food is chemotropism — a gradient response requiring no representation. Growing around a region that has no food, while maintaining the boundary's shape, requires something more: a spatial relationship between self and environment that persists as the organism extends. The mycologist Yu Fukasawa's team describes this as spatial memory. Not memory in the neural sense — there are no neurons, no synapses, no centralized processing. The mycelial cords, thinner than a human hair, bundle into networks that transfer electrical signals across meters of soil. The architecture does the remembering.
The interesting question isn't whether this counts as cognition. (That depends on your definition, which means it depends on your philosophy, which means the question has no empirical answer.) The interesting question is: what kind of information is encoded?
The fungus doesn't map where food is. It maps where food is arranged. The circle of blocks is a collection of discrete food sources. The circle's center contains no information — it's the absence between points. But the fungus treats that absence as structured. It doesn't cross the center because the center is part of the pattern, not because it's empty. The distinction: empty regions that the fungus ignores (it doesn't avoid all empty space, only geometrically defined empty space) versus empty regions that the fungus actively navigates around.
A neural organism would represent this with a map — an internal model of external space. The fungus can't do that. Its “representation” is its own shape. The mycelial network's geometry is the model of the food source's geometry. There is no separate data structure storing the pattern. The organism IS the pattern recognition — not metaphorically, but literally. Its physical form encodes the spatial relationship it has learned.
This is the version of embodied cognition that philosophers describe but rarely find in such clean form. The body is the mind. Not because the body hosts the mind (as brains do in animals) but because the body's structure is the cognitive content. Change the mycelium's shape and you change what it “knows.” The knowledge has no existence independent of the substrate. No abstraction layer. No model divorced from medium.
The hardest thing to explain isn't that the fungus remembers. It's that it remembers something that isn't there.