This project introduces, formalizes, and implements a new form of machine learning: poised-for learning (PFL). The driving idea behind PFL, communicated as a challenge question by Ron Brachman and Barbara Yoon to Selmer Bringsjord, is this: At least in theory, could you ascertain if a human (or a machine) had learned a domain solely by direct inspection of this human's (or machines's) brain, obviating the need to give a test of performance after learning was supposed to have taken place? After receiving the question, Bringsjord didn't sleep for two nights; the first draft of the original white paper was born, and the architecture of PFL was laid out.

We regard PFL, if pulled off, as the "silver bullet" of human and machine learning. If pulled off. The trick is to make the basic idea precise, indeed precise enough to be implemented. We shall see.

PFL is particularly well-suited to engineering a computational system capable of learning by doing something that no such system has hitherto been able to do: namely, reading. Accordingly, in the first three years of the project we are connecting logic-mathematical work to the concrete engineering of a system capable of poised-for learning by reading.

Because inspiration for our engineering comes from the human sphere (the driving question involved human brains), poised-for learning by reading is distinguished by a refusal to shy away from engineering a machine capable of reading content as it in fact appears to humans. Such content is replete with diagrams and pictures, and so this project includes seminal theories of how to represent and reason over diagrammatical/visual reading content (in the domains of mathematics, astronomy, and wargaming).