On the course of forgetting in very long-term memory.

For practical reasons, little information is available about memory across very long learning-retention intervals. To determine the time course of forgetting in very long-term memory, a memory test was constructed that sampled events (former one-season television programs) that had occurred during a single year from 1 to 15 years ago. Updated versions of the test were administered each year for 9 consecutive years, and a forgetting curve was then calculated by superimposing the results from the nine tests. The best fitting function to the curve was monotonic and nonlinear. The findings show that forgetting in very long-term memory can be gradual and continuous for many years after learning.