Tracing the footsteps of Sherlock Holmes: Cognitive representations of hypothesis testing