“…A biologically relevant learning mechanism needs to be able to function during both unsupervised and supervised learning conditions, including situations when unsupervised learning trials are mixed with supervised learning trials in unpredictable ways. We use an Adaptive Resonance Theory, or ART, category learning and recognition model to realize this property (Carpenter & Grossberg, 1987, Carpenter & Grossberg, 1993, Carpenter, Grossberg, Markuzon, Reynolds & Rosen, 1992, Grossberg, 1980a, Grossberg, 1999b. In order to learn and recall a category's name during supervised learning trials, the model incorporates teaching signals that can activate name categories, to which the object categories can be linked by means of associative learning (Figure 3g).…”