How does the information-geometric measure depend on underlying neural mechanisms?

Abstract To analyze how information is represented among neuron groups, a recently presented information-geometric method (IGM) has attracted growing attention. However the detailed properties underlying the information-geometric measure have not yet been elucidated because of the ill-posed nature of the problem. Here the underlying neural mechanism of the information-geometric measure is investigated with an isolated pair of model neurons. For the symmetric network, the information-geometric measure is solely dependent on the underlying anatomical connections between the recorded neurons. For the asymmetric network, however, the information-geometric measure is dependent both on the intrinsic connections and on the external inputs to it. In other words, there are multiple neural mechanisms corresponding to the same information-geometric measure. In addition, the relation between IGM and conventional cross-correlation is also investigated.