Figure 2.

Schematic detailing the quantities that define the free-energy. These quantities refer to the internal configuration of the brain and quantities that determine how a system is influenced by the environment. This influence is encoded by the variables $\tilde{y}$ that could correspond to sensory input or any other changes in the system state due to external environmental forces or fields. The parameters α correspond to physical states of the system that change the way the external forces act upon it or, more simply, change the way the environment is sampled. A simple example of these would be the state of ocular motor systems controlling the direction of eye gaze. $p(\tilde{y}\mid \vartheta ,\alpha )$ is the conditional probability of sensory input given its causes, $\vartheta \phantom{7}$, and the state of effectors (i.e., action). $q(\vartheta ;\lambda )$ is called an ensemble density and is encoded by the system’s parameters, λ. These parameters (e.g., mean or expectation) change to minimise free-energy, F and, in so doing, make the ensemble density an approximate conditional density on the causes of sensory input.