For at least 40 years, there has been a recurring argument concerning the nature of experimental amnesia, with one side arguing that amnesic treatments interfere with the formation of enduring memories and the other side arguing that these treatments interfere with the expression of memories that were effectively encoded. The argument appears to stem from a combination of (1) unclear definitions and (2) real differences in the theoretical vantages that underlie the interpretation of relevant data. Here we speak to how the field might avoid arguments that are definitional in nature and how various hypotheses fare in light of published data. Existing but often overlooked data favor very rapid (milliseconds) synaptic consolidation, with experimental amnesia reflecting, at least in part, deficits in retrieval rather than in the initial storage of information.