Small minds.
Real dinner.

Follow the trail. Then move the feast.

Start with enough food. Compare local memory with fading traces. Now choose “The feast moves” and watch what happens when a useful memory outlives the thing it remembers. A dense cluster can be wasted motion.

When does local signaling earn its energy cost?

What the little world actually does.

Every body can sense nearby food, collide with the same walls, and spend a finite energy supply. Carrying slows movement. Patches deplete when visited; food counts only when it reaches the nest. Traces persist in the environment and decay. The pooled controller can coordinate only observations its own bodies have collected.

This closes the gap between a search toy and a foraging mission. Information has a location, a lifetime and a price. Increasing population also increases the budget; the population slider is a scaling experiment, not proof that many small brains beat one large brain.

每个个体都要找到食物、搬回巢穴并支付能量。踪迹会衰减，食物会耗尽，拥挤也会拖慢队伍。请比较同一种子，尤其留意过期信息何时开始帮倒忙。

Keep the inconvenient outcomes.

The simplest trace rule failed this test. Across 24 abundant worlds, local memory returned 57.54 food units on average; fading traces returned 39.79. Pooling observations returned more food, but in the hazard regime it also lost 33 of 40 bodies on average. More coordination can mean more exposure.

Traces may improve discovery reuse, but stale signals, crowding, and scarcity can erase that gain.

24 paired seeds in each predeclared regime. Intervals are 2,000 resamples of paired primary-outcome differences. Null values are excluded from numeric aggregates and retain explicit valid counts. No biological or physical validation is implied.

One chosen seed is an illustration. The downloadable batch retains every world and treatment; null means unavailable. A confidence interval describes the sampled synthetic worlds, not the real world.