P. B. Sederberg, M. W. Howard, and M. J. Kahana have proposed an updated version of the temporal-context model (TCM-A). In doing so, they accepted the challenge of developing a single-store model to account for the dissociations between short- and long-term recency effects that were reviewed by E. J. Davelaar, Y. Goshen-Gottstein, A. Ashkenazi, H. J. Haarmann, and M. Usher (2005). In this commentary, the authors argue that the success of TCM-A in addressing the dissociations is dependent not only on an episodic encoding matrix but--critically--also on its implicit use of a short-term memory store--albeit exponential rather than buffer-like. The authors also highlight some difficulties of TCM-A in accounting for these dissociations, and they argue that TCM-A fails to account for critical data--the presentation-rate effect--that dissociates exponential and buffer-like models.