Gravar-mail: Fluency and belief bias in deductive reasoning: new indices for old effects