Trademark Law's View of the Image: A Computational and Empirical Analysis
Trademark law maintains a remarkably primitive, incoherent, and unexamined approach to images, even as the number and importance of image marks has grown dramatically. The law has viewed image marks as unanalyzable. In practice, trademark law also privileges representational meaning over visual appearance when searching for similar image marks, but it reverses that priority when assessing image mark distinctiveness. It has additionally long assumed that businesses can draw from an effectively limitless supply of “image synonyms”-different images that convey the same meaning.
Drawing on advances in marketing theory and computational approaches to image analysis, we use machine learning models as a diagnostic tool to show that the PTO’s adjudication of image marks is systematically inconsistent and incomplete. Specifically, we show that the PTO routinely refuses registration of applied-for image marks on the basis that they are confusingly similar with already-registered marks even though the PTO has allowed the registration of numerous other marks more confusingly similar to those already-registered marks than the applied-for mark. We further show that registered image marks have become increasingly descriptive or generic of their associated goods and services but distinctiveness refusals remain exceptionally rare.
We propose three reforms. First, image similarity should be assessed under an explicit “sight and meaning” test that gives weight to both the representational meaning and visual appearance of image marks. Second, image mark inherent distinctiveness should be assessed under a hybrid of the two differing approaches that courts currently use. Third, trademark law must investigate, rather than assume, that the supply of image synonyms is inexhaustible. Throughout, we urge that trademark doctrine learn from computational methods of image analysis, seeking to demonstrate how these methods can serve trademark law’s overarching goal of facilitating fair and efficient competition.
