Memory monitoring and memory performance: linked closely or loosely?

An experimental test of the metamemory hypothesis claiming a substantive positive relationship between memory monitoring and memory performance is provided. Forty-eight university students learned a set of abstract nouns. They were either instructed to make use of monitoring activities (experimental group), simply to learn the words (control-1) or to use non-metacognitive learning strategies (control-2). Prior to recall, the subjects had to predict the number of items they expected to remember. The experimental predictions were derived on two levels. First, prediction accuracy should be best for the experimental group. This is a necessary prerequisite for a valid test of the second prediction, namely that memory performance, too, should be best for the experimental group. Whereas the first prediction was confirmed, the second, vital for the metamemory hypothesis, was not. Thus, this hypothesis did not stand up to an experimental test, but would have been confirmed if tested by the correlation between prediction accuracy and memory performance.