A highly replicable model of achromatic contrast sensitivity based on individual differences in optics and spatial channels: robust consistency in factor structure across >6 very different datasets

For 40 years, individual differences and factor analysis have been used to investigate the underlying variable structure of achromatic contrast sensitivity in hopes of understanding underlying spatial mechanisms. Some perceive these factors to be volatile and difficult to replicate. However, the robust stability and consistency of underlying factors are shown here. Six+ studies were considered and reviewed, each using some methodologies that differed from the other studies (e.g,, adult vs. infant participants, psychophysical vs EEG/sweep-VEP thresholds; photopic vs. scotopic illumination; data collected by author vs archival data of others). The datasets provide a consistent and congruent set of results: 1) Below 2.25 c/deg, two factors correspond to Wilson’s channels A and B. (Wilson’s A and B channels were estimated using modeling masking data decades ago, with A and B of the 6-channel model peaking in at approximately .7 and 1.3 c/deg in adults, and at predictably lower spatial frequencies in infants). 2) Above 2.25 c/deg, a third factor corresponds to optical variation’s influence on contrast sensitivity. That is, individual differences in contrast sensitivity are primarily determined by optics. This third factor was recently validated by an analysis of CSFs collected using adaptive optics. 3). For scotopic vision, there is an additional fourth factor tuned to very low spatial frequencies. This third factor is consistent with findings from adaptation and masking studies. The existence of a highly replicated factor structure confirms that contrast sensitivity variability below 2.25 c/deg is determined by two "Wilson" channels (adults with photopic viewing) or more channels (infants, adults with photopic viewing). Above 2.25 c/deg, optics determine variability in CSFS. Unless adaptive optics methods are used, CSF variability should not be used to estimate channel properties above 2.25 c/deg.