Mistaking the instance for the rule: A critical analysis of the truth-table evaluation paradigm