Expectations recorded in advance: do new connections appear when living neural tissue masters a new task?

AbstractThe Causa Sui theory distinguishes two ways in which a system changes when it learns: tuning, in which the strength of existing connections changes, and graph growth, in which connections appear that did not exist before. According to the theory, a machine is capable only of the first, living tissue of the second as well, and the genuinely new arises precisely through growth. In this record, before any experiment has been performed, we state what follows from this for cultured neural tissue — cultures on electrode arrays and organoids. We give a translation of the theory’s concepts into the language of tissue, a criterion for when a task counts as new, two expectations (lasting mastery of a new task is accompanied by the appearance of new stable connections; mastery proceeds in two stages, a fast one without new connections and a slow one with them), and the conditions under which we will consider the theory wrong. We do not propose an experimental method. The text of this record does not change; clarifications are published as new versions with a new date.