Unannounced memory tests are not necessarily unexpected by participants: test expectation and its consequences in the repeated test paradigm.

In memory research, many paradigms take advantage of repeated testing. One phenomenon that is revealed through this procedure is hypermnesia, a net increase in memory performance over repeated tests. While this effect is robustly found, a consensus about the underlying mechanism is still pending. This paper investigates whether test expectancy may have contributed to this circumstance. The present research demonstrates that it may not be assumed that unannounced memory tests come as a surprise to participants. Based on the violation of fundamental conversational norms as well as the informative function of experimental procedures, systematic discrepancies between participants' expectations and experimenters' announcements may occur. Following identical instructions, test expectancy was shown to be a function of the experimental procedure (Exp. 1). Anticipation of an additional memory test did not affect hypermnesia; however, it did affect item fluctuation: Those participants who expected (vs. did not expect) another test showed reduced forgetting and--at the same time--reduced reminiscence (Exp. 2a and b). Consequently, our results show that test expectation does affect memory performance. It remains open, however, why this effect occurs and whether this generalizes to other research paradigms that apply "surprise" recall tests, which may not truly be surprises.