Abstract
Recently we reported a counterintuitive phenomenon, termed attribute amnesia, in which participants failed to report a just-attended attribute (e.g., color or identity of a letter) in a surprise test (Chen & Wyble, 2015, 2016, Chen, Swan, & Wyble, 2016; Eitam, Yeshurun, & Hassan, 2013). Despite being unable to report the attended attribute on the surprise trial, participants could correctly report the same attribute on the first control trial (directly after the surprise trial) wherein their expectation had changed so that they now expected to have to report that attribute. These findings suggest that expectation plays a critical role in determining the reportability of attended information. Here we ask whether an expectation change would improve memory performance for only the attribute that had been unexpectedly probed by the surprise question (i.e. specific improvement hypothesis), or alternatively, would the change improve reportability of all attributes of that target object (i.e. general improvement hypothesis). To distinguish between these two possibilities we used a double-surprise test paradigm, wherein participants were probed on two different attributes of a target in two separate surprise questions. The results showed that once participants were unexpectedly probed about an attribute, their memory performance on the 2nd surprise question was improved on other attributes of the target. This was true even when 15 trials elapsed between the 1st surprise test and the 2nd. These results support the general improvement hypothesis and show that when expectations about information (ir)relevance are violated within a task, participants "broaden" their set to include multiple attributes of the object, rather than just the queried attribute. This change in strategy endured for at least several minutes.
Meeting abstract presented at VSS 2017