← home
simulate

path integration

Cataglyphis desert ant · dead reckoning · home vector

A Cataglyphis ant forages across open desert, tracking every step. It can travel 100 meters from the nest before finding food — then walks straight home, not by retracing its path, but by following a vector it has been continuously computing: every turn integrated, every step counted.

Press [displace] while the ant is returning to teleport it to a new location. The home vector doesn't change. Watch where the ant thinks the nest is.

speed
stateidle
steps0
home vector
displacedno
nest (actual)
food
ant
home vector
computed home

Press start. The ant will forage, then return home. Displace it mid-return to see what happens.

what's happening

The ant accumulates a displacement vector — the sum of all steps taken since leaving the nest, weighted by direction. This is path integration, or dead reckoning. The home vector is simply the negative of the displacement: if you've drifted 60m northeast, home is 60m southwest.

The direction comes from a sun compass (calibrated for the sun's arc across the sky). The distance comes from the legs — a step-counting odometer. When Wittlinger, Wehner & Wolf (2006) glued pig bristles to ants' legs to lengthen their stride, stilted ants overshot the nest by ~5m on the return trip. When they trimmed ants' legs shorter, stumped ants stopped ~4m short. Every group walked in the correct direction. Only the odometer was fooled.

Displacement experiments show something stranger. Take a loaded ant — home vector set, food found — and move it to a new location before it starts returning. It walks in the correct direction for the correct computed distance, arrives where home would be, and begins searching there. Even when the actual nest is nearby. The model wins over the senses. Within a single run, the internal computation is authoritative.

source

Wittlinger M, Wehner R, Wolf H (2006). The ant odometer: stepping on stilts and stumps. Science 312:1965–1967.

The Count — entry-338