"So, the idea is that, if the brain's representation of the size of the object is what is being rapidly recovered with repetition, just changing the direction of the question from a 'bigger than' to a 'smaller than' question should not make a difference in performance," said Dobbins. "Since the subject can immediately recover the size of the object, he or she should be able to rapidly perform the comparison.
"However, if performance improves and neural activity declines because we rapidly recover our previous responses, then this sort of change should severely disrupt performance and preclude neural activity reductions because the subjects can't use their earlier responses when the question direction changes."
The researchers got the idea for the experiment from their own personal experiences with the tasks.
"We were going through these size comparisons, and we had this subjective feeling that after a few times, we weren't really thinking about the size of the object. We were just executing a response," he said. Such learning becomes automatic "such that at some point, you're just retrieving responses from memory; you're not actually performing the deliberations that were necessary on the first decision," Dobbins said. "It's like you know that two plus two equals four, not because you're working through the solution; you're recovering the answer from memory."
When the researchers performed their "cue-flipping" experiments on volunteers whose brains were being scanned by fMRI, "we saw a characteristic pattern of item-specific learning, in which performance would slow down when we flipped the cue around, but sped back up when we put it back the way it was," Dobbins said.