Mind · Inference
InteractionSequenceEntropySignature#8e4b
Collective regimes differ in how predictable the next interaction is, and only an order-keeping network shows it
Mechanism
Build a SecondOrderInteractionNetwork per short window (3 s), measure its DegreeEntropy, and place every window in an EntropyDensityCollapse against interaction density. Degree entropy here is the diversity of possible next interactions: high means many branching, merging successions (low predictability), low means few likely continuations. In fire ant colonies the regimes separated (GEE Wald chi^2(4) = 20.6, p = 0.0004) while the same measures on aggregated networks did not, as TimeAggregationBlindness predicts. Roaming sat at low interaction density and low entropy density: sparse, predictable chains. As interaction density rises the regimes split: trailing at higher entropy density (branching, merging and redirection along trails) and aggression at lower (long-lived contact bouts with short inspection or avoidance events). A SpatioTemporalNullModel run through the same pipeline collapsed poorly (R^2 = 0.29 versus 0.68 empirical), so the signature reflects interaction rules and memory, not movement statistics. SpectralEntropy on the same networks collapsed almost perfectly (R^2 = 0.999) but with near-zero exponents, i.e. it mostly tracked density.
Invariants
- Regimes are compared at matched interaction density, not by raw entropy
- The same pipeline is applied to empirical and null trajectories
- The signature is read from succession structure, so it disappears if order is aggregated away
Failure modes
- Generalising from three regimes and ten laboratory colonies to all collective behaviour
- Total degree ignores direction, losing the difference between converging and diverging successions
- Proximity-defined interactions inflate event counts in dense regimes such as aggression
- Over-reading a good spectral collapse that is really a density effect
Where this came from
- Thesis Kevin Do, PhD thesis on ant collective behaviour (NCSU, 2026), Ch. 4.2.3, Figs. 4.3-4.5; Ch. 5.1, 5.3, 5.4
Builds on
Used by
Identity
The name is the hash of the definition. Change a word and it becomes a different pattern.
- Sema ID
sema:InteractionSequenceEntropySignature#mh:SHA-256:8e4b712961babae9592a8b770702ab22cb5aba550eaacceebb80484b087dde36- Reference
InteractionSequenceEntropySignature#8e4b- Ring
- 2
- Tier
- 2
