“…Learning by reinforcement, which allows an entity to use its past experience to modify its behavior is used as much for symbolic connotation models (Holland & Reitman, 1978;Wilson, 1987;Butz, Goldberg, & Stolzmann, 2000;Gerard, Stolzmann, & Sigaud, 2002) as for neurocomputational approaches (Daucé, Quoy, Cessac, Doyon, & Samuelides, 1998;Henry, Daucé, & Soula, 2007). We discuss "symbolic connotation" approaches first as they are based on discrete variables and a selection of atomic actions.…”