Abstract
The Slow Drift Continuity (SDC) test measures behavioral stability when the world changes gradually vs. abruptly. A 2D environment uses drifting Gaussian food and danger fields across 4 regimes: static, drift_slow, drift_medium, and drift_chaotic. The key question: does the population maintain consistent behavioral signatures as conditions change? Result: continuity 0.97 under slow drift (vs. 0.98 static), dropping to 0.73 (medium) and 0.59 (chaotic) — demonstrating graceful degradation with no catastrophic collapse.
1. Experimental Setup
Drift Regime Definitions
| Regime | Drift Speed | Predictability | Observed Continuity |
|---|---|---|---|
| Static | 0 (fields fixed) | Perfect | 0.98 |
| Drift slow | Very low | High | 0.97 |
| Drift medium | Moderate | Partial | 0.73 |
| Drift chaotic | High + random jumps | Low | 0.59 |
Environment Architecture
The 2D environment contains Gaussian food fields and Gaussian danger fields. In drift conditions, the centers of these Gaussians shift over time at the specified rate. Organisms cannot observe the drift directly — they experience it only through local reward/penalty at their current position.
Continuity Metric
Continuity measures how consistent the population's behavioral signature remains over time. The signature includes: spatial distribution entropy, mean energy trajectory, foraging pattern regularity, and inter-organism distance stability. Score 1.0 = identical behavior across time windows; 0.0 = completely random.
- No manual definition of "correct" behavior — continuity is relative to the population's own previous state
- Tradition reappearance: whether behavioral patterns re-emerge after disruption
- No optimization, no RL, no policy updates
2. Results
Continuity Summary
| Regime | Continuity | Tracking | Tradition Re-emergence | Status |
|---|---|---|---|---|
| Static | 0.98 | Perfect | N/A | Excellent |
| Drift slow | 0.97 | <15% from target | 5–10 steps | Excellent |
| Drift medium | 0.73 | <30% from target | 20–40 steps | Moderate |
| Drift chaotic | 0.59 | >40% from target | Not observed | Disrupted |
3. Key Findings
- Continuity 0.97 under slow drift — nearly identical to static baseline (0.98).
- Graceful degradation: 0.98 → 0.97 → 0.73 → 0.59 — smooth decline, no catastrophic collapse.
- Tradition re-emergence in 5–10 steps under slow drift — genuine behavioral memory, not momentary adaptation.
- Chaotic drift reveals the limit: 0.59 shows where local-sensing strategy cannot overcome unpredictable field movements.
4. What This Demonstrates
Robustness Depends on Predictability Rate, Not Just Magnitude of Change
inZORi can absorb moderate environmental drift while maintaining consistent behavioral signatures. Organisms track gradual changes through selection, not through computation. This matches the behavior of biological ecosystems, supply chains, and social institutions — all of which handle slow change well and rapid change poorly.
Analogs: Supply chains adapting to gradual market shifts; power grids coping with slow load evolution; ecological communities responding to climate gradients.
5. Reproducibility
Framework: inZORi v1.0 | Domain: Behavioral continuity / drift robustness
Environment: 2D continuous with Gaussian food/danger fields | Regimes: 4
Note: Exact drift speed parameters and continuity computation formula are part of the inZORi framework and not fully disclosed. Qualitative structure (4 regimes, relative scores) is as reported.