Seems like a testable hypothesis: Prepare a "training" deck of cards with an A side and B side. Each side has a simple symbol on it. Create a second "test" deck

Seems like a testable hypothesis: Prepare a "training" deck of cards with an A side and B side. Each side has a simple symbol on it. Create a second "test" deck of cards which is identical, expect each A symbol is slightly shifted in meaning (e.g. horse -> donkey). The task is to predict side B from side A. If less dreaming leads to overfitting, we would expect REM sleep deprived probands to do better on the "training" set after learning with them and do worse on the "test" set, compared to probands without sleep deprivation.