Predictive Agency

In plain language

A purely reactive system acts after a condition arrives. A predictive system uses an earlier cue—light before heat, a nutrient shift before toxicity, or a repeated temporal pattern—to prepare in advance. Predictive agency refers to the organism’s capacity to turn that relationship into action.

Why it matters

Preparation has costs. Producing protective machinery too early wastes resources; producing it too late may be fatal. Prediction can therefore be advantageous only when the world contains enough regularity to learn and the organism has a mechanism capable of retaining and using that regularity.

In my work

Experiments with Capsaspora manipulate whether cues carry reliable information about later stress. Comparing predictive and non-predictive cue histories makes it possible to separate the effect of a cue itself from the effect of its temporal meaning.

What to measure

1. The information a cue carries about a later condition.
2. The internal state change produced by that cue.
3. The cost of preparation when the predicted event does not occur.
4. The survival or reproductive benefit when the prediction is correct.

Further reading

• Knafo, Casacuberta & Ruiz-Trillo (2026), Beyond Diffusion: Bayesian Learning Strategies in Single-Cell Life.