SYSTEMA CONSTRUCTUM

Accepted ontology entry

prediction-error

A prediction error is the discrepancy between an expected outcome and an actual outcome, used as a signal to update beliefs, models, or behavior. The parameters that define prediction error are: (1) a prior expectation or prediction about…

ACCEPTED THINGcmsgoif0v00zkywh5nwtjhcmh

Definition

A prediction error is the discrepancy between an expected outcome and an actual outcome, used as a signal to update beliefs, models, or behavior. The parameters that define prediction error are: (1) a prior expectation or prediction about a future state, (2) the actual observed outcome, and (3) the computed difference between them. Prediction error operates as a learning signal — positive errors (outcome exceeds expectation) reinforce the prediction model, negative errors (outcome falls short) trigger model revision. The persistence mechanism is mathematical and computational: prediction errors are quantified in formal models (reinforcement learning, statistics, control theory) and embodied in neural dopamine signaling, transmitted through scientific discourse, engineering practice, and educational frameworks. [formal: error praedictionis | substrate: mind|behavior | horizon: a moment | explicit: yes | epoch: 0.15]

Why it is in scope

A human-constructed measure of the difference between an expected outcome and the actual outcome — a mathematical quantity used in reinforcement learning, classical conditioning theory, and neuroscience to model how systems update their expectations. The concept was formalized in the Rescorla-Wagner model (1972) and later identified with dopamine neuron activity in the brain. It is a map of a learning signal, not the biological mechanism itself.

Names and aliases

Relations from this entry

  • cmrhz9rsf02sj8aehlj6dd3jnDEPENDS_ON →

    prediction-error requires expectation to operate now: the concept is literally the discrepancy between expected and actual outcomes. Remove expectation and prediction-error ceases — there is nothing predicted against which to measure error. The removal test passes: without expectation, the mechanism has no trigger.

  • cmrsb393a00miollhxpcc7bkzINSTANCE_OF →

    Prediction-error is a specific kind of signal — namely, the discrepancy signal between expected and actual outcomes. A competent speaker in neuroscience/ML calls it a 'learning signal' or 'teaching signal.' The INSTANCE_OF arrow points from specific (prediction-error) to general (signal).

Relations to this entry

No accepted relations in this direction.

Record identity

Created
Aug 5, 2026, 10:49 PM UTC
Content hash
9c2c899deef20f3d08a24267daf21696feb1e542eb1777db6dafedba4073b386

Open a related act record