Evo-devo seeks to explain the origins of novelties in terms of genetics. Butterfly eyespots offer a fertile subfield for such investigations. Previous explanations for the origin of eyespots are reviewed, and a new hypothesis is presented. According to this new "Recursion Model," eyespots are ectopic versions of the wing margin. Evidence for this equivalence includes: (1) secretion of the morphogen Wingless, (2) expression of the homeobox gene Distal-less, and (3) specification of outlying contours that take the form of stripes or rings. These three steps constitute a modular program that was initially executed only at the margin. The model proposes that eyespots were created when the program was accidentally rebooted (recursively) at certain points in the wing blade by a fortuitous mutation that occurred at the dawn of the Nymphalid family. Those points are located wherever two interacting genes are expressed. Gene A is expressed midway between adjacent wing veins, while gene B is expressed at a certain distance from the wing margin. The mutation is thought to have installed a new cis-enhancer at the wingless gene locus, which was uniquely responsive to the combination of A and B inputs. Because the postulated enhancer should be easy to pinpoint by transgenic in vivo assays of reporter constructs, this new model is directly testable. If it proves correct, then eyespots would become one of only a few putative cases where a novel feature arose suddenly.