INTERACTIVE MODELFIELD NOTEANIMAL NAVIGATION
The Star Compass Bench
- Study
- Field note
- Review
- Peer-edited by the club review board
- Article
- 13 sectionsabout 20 min
- Files
- LaTeX source26 kB
Three small models of the mechanism, not the animal. Rotate a projected sky and watch a compass heading rotate with it, exactly as the moths did under the real rig. Then compare the naive strategy of following a single star against how little the real moths actually drifted. Then take the Earth's own magnetic field down toward the level the experiment needed before it could trust the stars as the only cue left.
Model 1. Rotate the sky, watch the heading follow
The dial on the left is a stand-in for the inside of the 50 cm arena: a scattering of stars plus one elongated band standing for the Milky Way. Drag the sky rotation slider and the whole pattern turns, the way the projector could rotate the real stimulus in the lab. As long as the pattern stays coherent, a fixed offset from it (learned once for each season) gives a stable compass heading, drawn as the arrow on the right-hand readout dial. Scramble the stars and that offset stops meaning anything.
This is a mechanism sketch, not a replay of real trial-by-trial data: it assumes a fixed offset between star pattern and target bearing, which is the simplest hypothesis the six-trial result (Figure 1 of the field note) is consistent with. The paper does not claim the mechanism is this simple; it only rules out simpler ones.
Model 2. The naive strategy against the real one
Suppose a moth simply fixed its heading relative to a single star and never corrected for the sky's own rotation. Earth turns roughly 15° per hour, so that strategy should drift steadily. The paper's outdoor trial measured the same 95 moths at two real times, 20:49 and 23:42, and found a heading shift of only 10° where the naive strategy predicts about 43°. Drag the clock forward and watch the two predictions pull apart.
The real line is a straight interpolation between the paper's two measured points (5° at 20:49, 355° at 23:42) and is drawn only to compare against the naive prediction; the paper reports the two endpoints, not a continuous path between them. The naive line is entirely the club's own construction, not a quantity anyone measured.
Model 3. How far the field had to be pushed down
Earth's surface field runs roughly 25,000 to 65,000 nanotesla depending on latitude. The experiment's three-axis Helmholtz coils had to push that down far enough that no plausible magnetic sense could still read a direction from it. The paper's own magnetometer checks confirmed the coils held the field below 100 nanotesla around the moth. Slide the coil current up and watch the residual field, and the reduction factor, fall.
- Paper's own value
- < 100 nT, magnetometer-checked, within 4 cm of the moth
- Club's arithmetic
- 25,000–65,000 nT ÷ 100 nT ≈ 250–650× reduction
What to take away from pushing the sliders
Two things land harder here than in prose. The first is how binary the rotation test is: there is no in-between state where a coherent sky produces a half-confident heading, because the model (and the moths) either lock on or do not. The second is how small the gap really is between 10° and 43° once you watch both lines drawn at the same scale — a factor of more than four, accumulating for less than three hours, that either a rigid pattern or a genuine clock has to be cancelling out.
Neither model reproduces the paper. Dreyer and colleagues recorded actual flight headings from tethered animals and actual spikes from actual neurons; what is here is the arithmetic and the geometry underneath their result, left where you can drag it yourself.