The power law in the amplitude spectra of natural scenes provides not just an efficient description of them but also a foundation for image processing. Psychophysical studies show that the the forms of the amplitude spectra are clearly related to the human visual performances. However, the underlying neuronal mechanism and computation that account for the perception of the natural image statistics is poorly known. We propose a theoretical framework for neuronal encoding and decoding of the natural image statistics, hypothesizing the elicited population activities of spatial-frequency selective neurons observed in early visual cortex. The predictions by the computational model are consistent to the experimental data reported in the previous study. Especially, the qualitative disparities between performances in fovea and parafovea can be explained based on the distributional difference over preferred frequencies of neurons. The model predicts that the frequency-tuned neurons have asymmetric tuning curves for the amplitude spectrum slopes.