Attribute amnesia (AA) describes a phenomenon whereby observers fail a surprise memory test which asks them to report an attribute they had just attended and used to fulfil a task goal. This finding has cast doubt on the prominent theory that attention results in encoding into working memory (WM), to which two competing explanations have been proposed: (1) task demands dictate whether attended information is encoded into WM, and (2) attended information is encoded in a weak state that does not survive the demands of the surprise memory test. To address this debate our study circumvented the limitations of a surprise memory test by embedding a second search task within a typical color-based AA search task. The search task was modified so that the attended attribute would reappear in the second search as either the target, a distractor, or not at all. Critically, our results support encoding of the attended attribute in WM though to a weaker extent than the attribute that is required for report. A second experiment confirmed that WM encoding only occurs for the attended attribute, though distractor attributes produce a bias consistent with negative priming. Our data provide novel support for a theory of memory consolidation that links the strength of a memory's representation with expectations for how it will be used in a task. Implications for the utility of this procedure in future investigations previously limited by single trial data (i.e., surprise question methodology) are discussed.

属性失忆症（AA）描述了一种现象，观察者无法通过突击记忆测试，要求他们报告他们刚刚参加并用于实现任务目标的属性。这一发现使人们对注意力集中到工作记忆（WM）中的编码这一著名理论产生了怀疑，对此提出了两个相互竞争的解释：（1）任务需求决定是否将出席信息编码为WM，（2）出席信息编码处于弱状态，无法满足意外内存测试的要求。为了解决这一争论，我们的研究通过将第二个搜索任务嵌入典型的基于颜色的AA搜索任务中来规避了突击记忆测试的局限性。修改了搜索任务，以便在第二次搜索中重新出现伴随属性，将其作为目标，干扰因素，还是根本没有。至关重要的是，我们的结果支持WM中人为属性的编码，尽管程度要比报表所需的属性弱。第二个实验证实了WM编码仅发生在有人陪伴的属性上，尽管干扰因素的属性会产生与负引物一致的偏差。我们的数据为内存整合理论提供了新颖的支持，该理论将内存表示的强度与对如何在任务中使用内存的期望联系在一起。讨论了此程序在以前受单一试验数据（即突击问题方法）限制的未来调查中的用途。第二个实验证实了WM编码仅发生在有人陪伴的属性上，尽管干扰因素的属性会产生与负引物一致的偏差。我们的数据为内存整合理论提供了新颖的支持，该理论将内存表示的强度与对如何在任务中使用内存的期望联系在一起。讨论了此程序在以前受单一试验数据（即突击问题方法）限制的未来调查中的用途。第二个实验证实了WM编码仅发生在有人陪伴的属性上，尽管干扰因素的属性会产生与负引物一致的偏差。我们的数据为内存整合理论提供了新颖的支持，该理论将内存表示的强度与对如何在任务中使用内存的期望联系在一起。讨论了此程序在以前受单一试验数据（即突击问题方法）限制的未来调查中的用途。