“…These models offer little guidance, however, regarding how those representations actually enter into memory in the first place. Pre-existing semantic or similarity relations between items lead to stronger encoding of an association (Thomson & Tulving, 1970;Dosher, 1984;Dosher & Rosedale, 1991;Greene & Tussing, 2001), and interference only occurs between pairs that are made of the same types of items (e.g., word-word versus word-face pairs; Criss & Shiffrin, 2005), such that it is clear that associative information depends at least in part on the information contained in the associated items, especially relational features and features shared between items. While memory models of all stripes could easily implement this phenomenon, none of them predict it a priori.…”