Figure 1.

Identifying effective connectivity from restricted functional connectivity. (A) The G matrix relating unknown noise to the unobserved “neural” signals. Using the connectivity model of 1.6, the matrix G determines the covariance of the signal space via GG^T. Connectivity in G is directional in the sense that Gⁱ influences G^k via β but G^k does not influence Gⁱ. (B) Graphical depiction of the relationship between the common factors ε and the signals x implied by the pattern of values in (A). The variable xⁱ is only influenced by εⁱ while x^j is influenced by εⁱ and ε^j, and so on. (C) Effective connectivity between the signals. The effective connectivity depicted in C is a direct consequence of the matrix G in (A). See the text for the solution.