I (long-term importance); and each Hebbian link is weighted with a probabilistic truth value. The equations of an ECAN explain how the STI, LTI and Hebbian link weights values get updated over time. As alluded to above, the metaphor underlying these equations is the interpretation of STI and LTI values

d in interpreting other link types im- plicitly as HebbianLinks: for instance, Similarity maps into SynunetricHebbian. and Inheritance A B maps into Hebbian A B. One may express these as inference rules, e.g. SimilarityLink A B <tv_1> I - SymmetricHebbianLink A B <tv_2> where tv2.8 = tvi.s. Clearly, tv

ds: e.g. if one has chosen a certain element A of the first blend input, then one may seek an element B of the second blend input that has a strong Hebbian link to A (if such a B exists). However, it may also be interesting to pursue different sorts of heuristics, using information- theoretic or other m