<i>The generative model</i>.

This schematic illustrates how hidden states cause each other and sensory outcomes in the interoceptive and exteroceptive domain. The upper row describes the probability transitions among hidden states, while the lower row specifies the outcomes that would be generated by combinations of hidden states that are inferred on the basis of outcomes. The green panel specifies the model’s prior preferences; namely, the sorts of outcomes it expects to encounter. Please see main text for a full explanation. Although this figure portrays interoceptive and exteroceptive outcomes as separate modalities, they were in fact modelled as combinations–so that the prior preferences could be evaluated (this is necessary because the preferred physiological outcome depends upon the visual cue). In this model, the precisions are denoted by Greek letters and control the fidelity of various probabilistic mapping is (i.e., the likelihood or A matrices and the transition or B matrices).