Suppose you presented the tone first, and then presented the letters. What would you expect letter report accuracy to look like as a function of SOA?