Abstract

Those familiar only with regular sampling theory must be somewhat surprised by recent research reports of the recognition of supra-Nyquist signals in near-peripheral vision and the detection of more peripheral sub-Nyquist signals by aliased energy.¹ Although in the first case the recognition could be based on the information in sub-Nyquist aliases and in the second case subsampling at later stages is possible, both effects are also consistent with irregular sampling and different detection mechanisms for the different regions. In particular, if we assume that detection is based on sensor units with Gabor-functionlike receptive fields, we can simply include units tuned to supra-Nyquist frequencies in the near periphery but omit high-frequency sub-Nyquist units in the far periphery. This argument also requires highly irregular sampling in the far periphery to scatter aliasing noise to lower frequency units. A quantitative assessment of these arguments is based on computer simulations of signal and noise output as a function of sensor unit parameters and sampling array irregularity.

© 1987 Optical Society of America