Alternatively, a population of neurons together may represent a property, such as color, as a probability distribution (Pouget, Beck, Ma, & Latham,
2013). This could be accomplished if sensory input is encoded across a population of neurons with variable tuning (Finkelstein & Hood,
1984; Ma, Beck, Latham, & Poutget,
2006). Numerous authors have proposed that a representation of color space could be built in such a manner by cortical circuits that combine input from lower visual areas (Finkelstein & Hood,
1984; Brainard et al.,
2008; Kellner & Wachtler,
2013; Bohon, Hermann, Hansen, & Conway,
2016; Emery, Volbrecht, Peterzell, & Webster,
2017). For instance, midget ganglion cells multiplex chromatic and achromatic signals, an idea known as “double duty” (Shapley,
1990; Rodieck,
1991; Dacey,
2000). The relative strength of chromatic versus achromatic information varies between cells due to random wiring (Crook, Manookin, Packer, & Dacey,
2011; Wool et al.,
2018). Near the fovea, each midget cell would have an L or M cone in its center and variable cone weights in the surround, based on the cone types of its closest neighbors. Models explaining how randomly wired midget ganglion cells may subserve both chromatic and achromatic sensation have been described previously (Ingling & Martinez,
1983; Finkelstein & Hood,
1984; Finkelstein,
1988; De Valois & De Valois,
1993; Schmidt, Neitz, & Neitz,
2014; Sabesan et al.,
2016; Schmidt, Touch, Neitz, & Neitz,
2016; Wool et al.,
2018). Since each midget cell carries a unique chromatic/achromatic signature, the visual system could learn to associate different prior information with the output of each midget ganglion cell through experience (Benson et al.,
2014). The color reported when a single cone is targeted with light may be a reflection of that prior experience. Finally, this theory makes the testable prediction that color reports in the peripheral retina, where the centers of midget-cell receptive fields pool signals from multiple cones (Dacey,
2000), will be less tightly predicted by the cone type stimulated.