Within the two state theory (TST) for stochastic resonance (SR) we analize
two different aspects: (a) the extension of the TST in order to include pot
ential asymmetry (i.e.: the states show different stabilities): (b) the eva
luation of transition rates for systems whose stationary distribution is no
n Gaussian. We apply the results of (a) to study the role of the potential
symmetry for SR in bistable systems, observing that the signal-to-noise rat
io increases with the symmetry of the potential of the system indicating th
at it is this feature that, governs the optimization of the response, We ap
ply the results of (b) to discuss SR in situations where we call assume tha
t the noise is non Gaussian, and discuss its relation with experimental res
ults in sensory systems.