Imaging black boxes with multiple stochastic output

A black box is modeled by decomposition into an unknown but finite number of independent noise sources and of independent signal sources. Signal sources carry signals in strict relation to an input signal vector. Output signal vectors are transformable into a system-state vector by means of stochastic properties of an assembly of independent noise sources. By proper scaling such system-state vectors are measurable. The internal organization structure can be imaged by the correlation of components of the state vector. This correlation reflects overlap of so-called activity profiles which are defined by the fraction of signal sources and of noise sources used in the formation of a response. In accordance with the two kinds of sources, there are two complementary images, for overlap of noise sources and for overlap of signal sources, respectively.