In this paper we summarize the proceedings of the Workshop on Humor and Cognition held at Indiana University's Center for Research on Concepts and Cognition on February 18 and 19, 1989. The principal type of humor considered, slippage humor, is first defined and contrasted with aggression-based humor. Next, a particularly clear variety of slippage humor, based on Douglas Hofstadter's notion of a frame blend, is presented. Given that a frame is a small coherent cluster of concepts pertaining to a single topic (similar to Victor Raskin's notion of a script), a frame blend is what results when elements are extracted from two distinct frames and spliced together to yield a new hybrid frame. Diverse ways of blending two given frames can produce varying amounts and types of humor, and some studies of this phenomenon are presented. A close connection between frame blends and analogies is pointed out. To make this connection more explicit, the Copycat domain -an idealized microworld in which-analogy-making can be studied and modeled on computer -is presented, and it is shown how jokes can be mapped into that domain, giving rise to a kind of abstract "microworld humor". The reduction of these phenomena to the Copycat domain helps to bring out the tight relationships among good jokes, defective analogies, and frame blends quite clearly. As a result, these relationships appear clearer in the real world as well. The notion that many jokes can share the same abstract structure is suggested, and the name ur-joke is suggested for the most abstract level of a joke. Several specific ur-jokes are presented, each one with a set of fully fleshed-out jokes based on it. We recount the group's collective efforts at translating two jokes from one subject matter to another, in an attempt to determine whether a joke's funniness is due more to its underlying-ur-joke or to its subject matter. This important question is, however, left open. There follows some discussion of Victor Raskin's overlapping-script theory of humor, which has many points of contact with Hofstadter's frame-blend theory, and then a summary of Salvatore Attardo's theory of a multiple-level analysis of jokes (closely related to Hofstadter "ur-joke hypothesis") is presented. Finally, a speculative theory by Gray Clossman about the adaptive value of humor is briefly addressed.