Retina-Centric Model of Visual Processing: Cascading Photon Transduction Without Cortical Image Reconstruction

Current visual neuroscience proposes that the cortex reconstructs visual images from a compressed retinal input of 1.2 million nerve fibers transmitting ~10 Megabits/second. This framework creates a thermodynamic puzzle: cortical visual areas consume 4 Watts while retina consumes only 0.5 Watts, yet cortex is theorized to perform the computationally intensive task of image reconstruction. We propose the Retina-Centric Model (RCM): the retina is not a passive photoreceptor array but an active neural preprocessor that performs sophisticated 130-million-to-1.2-million compression via variable-speed photochemical cascades. The cortex does not reconstruct images; it integrates pre-processed information for binocular fusion, temporal coherence, and conscious access. RCM is grounded in:- Documented metabolic constants: ATP retinal = constant (~10⁸ ATP/s, Okawa et al. 2008); ATP cortical = constant (4W, Wong-Riley 2012)- Quantified temporal dynamics: Photochemical cascades accelerate 4-10× in milliseconds with light (Okawa et al. 2008)- Verified damage mechanisms: Retinal phototoxicity is photochemical (ROS-mediated, not thermal; literature consensus)- Measurable adaptation timescales: Gradual light changes (atardecer/amanecer) imperceptible across 20-60 minutes; abrupt changes (cinema/cave) perceptible in seconds- Falsifiable predictions: ATP constancy, ROS correlation with damage, cascade acceleration, cortical ATP independence from light The model explains clinical observations (rapid post-LASIK recovery, optogenetic acuity ceiling, myelination-constrained plasticity, amblyopia irreversibility) without invoking cortical compensation or image reconstruction. RCM suggests that visual system organization follows the principle of metabolic parsimony: delegate processing to the stage with lowest energetic cost per unit computation.