Discriminating incipient Alzheimer's disease (AD) from major depression (MD) and age related memory decline is a challenge in clinical practice. Since AD is characterized by an early loss of neuronal and functional plasticity, a dynamic test strategy, such as the testing-the-limits (TtL) approach, that measures learning capacity can be a helpful diagnostic tool. To evaluate this, a short recognition paradigm consisting of a pre-test (baseline) and two post-test conditions with an interposed encoding instruction was developed and administered to individuals with incipient AD (n = 19; Mini Mental Status Examination (MMSE) 26.5), patients with depressive disorders (n = 11; MMSE 30), and healthy controls (n = 11; MMSE 30). In addition, participants completed a set of traditional neuropsychological tests that focused on the subjects' cognitive baseline performance. Intergroup comparisons (Kruskal-Wallis, U test) revealed significantly higher post-test failure rates in AD patients. Pre-test performance of MD and AD patients did not differ. Intra-group comparisons (Friedman, Wilcoxon test) showed that all three subject samples benefit from intervention in post-test 1. In contrast to MD and healthy individuals, who revealed significantly lower failure rates in post-test 2 compared to the pre-test, AD patients did not improve. Out of the 15 traditional test scores obtained, only two discriminated simultaneously between AD and each of the other groups. Our data confirm the finding of an impaired cognitive plasticity already present in very early stages of AD and illustrate the efficiency of a dynamic test approach in identifying incipient dementia.