Note that passed the optical system, images are transformed to a form of natural log-polar mapping* such that scaling in the Cartesian domain becomes analogous to a translation.
This mitigates greatly the effect of scaling in human perception. The fact that the radial dimension is logarithmically mapped means a 10-fold scaling corresponds to a single-unit translation on the logarithmic scale. My understanding is that this allows human vision to be very good at scale-invariant pattern recognition.
*See:
G. Sandini, and V. Tagliasco, ”A anthropomorphic retina-like structure for scene analysis”, Computer, Graphics and Image Process- ing, Vol. 14, 1980, pp. 365-372. or
E. Schwartz, ”Anatomical and Physi- ological Correlates of Visual Computation from Striate to Infero-Temporal Cortex” , IEEE Trans. on Systems, Man, and Cybernetics, Vol. SMC-14, No. 2, March/April 1984, pp. 257-271.