How are spatial filters used in fovea and parafovea?

The information used in the output of the visual filters was examined for a blur discrimination task for which it had previously been claimed that the best human performance was with pedestal blurs of intermediate magnitude. This conclusion was verified for parameters leading to asymptotic performance for foveal and parafoveal locations. The nature of the computations underlying these visual discriminations was then assessed experimentally and by simulations. The results suggest that a frequency-based model in its simplest form cannot explain these findings. A model based on the spatial properties of filter outputs is discussed.