Circular inference in neuroscience: The dangers of double dipping