MIT OpenCourseWare

13: Color Perception (Cont'd), Motion Perception

In this lecture, Josh McDermott wraps up the series on color perception before moving on to motion. He revisits the foundational idea that objects appear colored because their surfaces reflect specific wavelengths, but emphasizes that the reflected light is a product of both surface reflectance and the illumination spectrum, creating a complex visual problem. The talk delves into the physiology of the three cone types—short, medium, and long—explaining how they compress a high‑dimensional spectral signal into three neural responses. This dimensionality reduction gives rise to metamers, physically distinct spectra that look identical to the eye, and underpins modern color‑reproduction technologies. McDermott then bridges trichromatic theory with Hering’s opponent‑process model, showing how cone outputs are recombined into red‑green, blue‑yellow, and achromatic channels at the retinal ganglion‑cell level. Illustrative examples include spectra from daylight, tungsten bulbs, and fluorescent lights, and how the same objects (leaf, orange, tomato) maintain consistent perceived colors across these illuminants. He cites Hering’s 1872 observation on color constancy and demonstrates classic visual‑illusion experiments where identical gray patches are perceived as blue or yellow depending on surrounding context, highlighting the brain’s inferential solution to the illumination‑reflectance ambiguity. The implications are far‑reaching: understanding metamers informs display calibration and printing, while insights into opponent processing and color constancy guide computer‑vision algorithms and augmented‑reality rendering. Recognizing the limits of cone‑based information also underscores why certain visual disorders arise and how future neuro‑tech might augment human color perception.