Robots Atlas>ROBOTS ATLAS
Reasoning

Decision model

1944ActivePublished: 17 November 2026Updated: 17 November 2026Published
Key innovation
Shifts the focus from prediction (what will happen) to action selection (what to do), tying the model's output to an objective/utility and to the consequences of the decision.
Category
Reasoning
Abstraction level
Pojęcie
Operation level
ModelInference
Use cases
A decision layer on top of a classifier (a cost-sensitive decision threshold)Control and agent action policies (MDP / reinforcement learning)Prescriptive analytics: business action recommendationsClinical and credit decision support with explicit risk

How it works

1) Define the decision space: the set of admissible actions/alternatives. 2) Represent state and uncertainty: input features and probabilities of states of nature (often from a predictive model). 3) Objective function: utility, cost, or risk assigned to (action, outcome) pairs. 4) Decision criterion: e.g. maximize expected utility, minimize expected cost, or min-max strategies when no distribution is available. 5) Rule/policy: a mapping from state to action, derived analytically (e.g. a Bayes threshold), by dynamic programming (MDP), or learned (an RL policy). 6) Execution and update: the decision is taken, the outcome observed, and the model may be refined (e.g. online learning).

Problem solved

It solves the problem of moving from knowledge about an uncertain future to a concrete action: a probability alone does not say what to choose when error costs are asymmetric or constraints exist. A decision model explicitly encodes the objective, costs, and rules, making decisions consistent, repeatable, and justifiable.

Components

Decision spaceDefines what can be chosen at all.

The set of admissible actions or alternatives from which the model chooses.

State and uncertainty representationProvides the information the choice is based on.

Input features and the probabilities of states of nature, usually supplied by a predictive model.

Utility / cost functionDefines what a good decision means.

A valuation of action-outcome pairs encoding preferences, risk, and the asymmetry of error costs.

Decision rule / policyThe actual action-selection mechanism.

A mapping from state to action, derived analytically, by dynamic programming, or learned.

Evolution

Original paper · 1944 · Princeton University Press · John von Neumann
Theory of Games and Economic Behavior (foundations of decision theory and utility)
John von Neumann, Oskar Morgenstern
1944
Expected utility theory
Inflection point

von Neumann and Morgenstern formalize rational choice via expected-utility maximization — the foundation of decision models.

1957
Markov decision processes (MDPs)
Inflection point

Bellman and dynamic programming provide a method for computing optimal policies in sequential decisions.

2015
Policies learned by deep RL

Deep reinforcement learning (e.g. DQN) makes it possible to learn decision models directly from raw observations.

Hyperparameters (configurable axes)

Decision criterionCritical

The optimization rule: expected utility, risk minimization, min-max, satisficing.

Maksymalizacja oczekiwanej użytecznościRequires a probability distribution over states.
Min-max (najgorszy przypadek)For decisions without a known distribution.
Action space typeHigh

Whether decisions are discrete (a finite set) or continuous — determines the optimization method.

DyskretnaE.g. approve/deny.
CiągłaE.g. dosage, control torque.

Hardware requirements

Primary

A decision model is a logical/mathematical layer; its cost depends on the underlying predictive model or optimization method, not on a specific accelerator.