Foraging through multiple target categories reveals the flexibility of visual working memory
Foraging through multiple target categories reveals the flexibility of visual working memory
复制标题
DOI：
10.1016/j.actpsy.2017.12.005
复制
发表时间：
2018-02-01
影响因子：
1.8
通讯作者：
Kristjansson, Arni
中科院分区：
文献类型：
--
作者：
Kristjansson, Tomas;Kristjansson, Arni
视觉注意文献中的一个关键假设是，模板，积极维护在视觉工作记忆（VWM），引导视觉注意。因此，一个重要的问题涉及VWM的性质和能力。根据负载理论，可以同时激活多个搜索模板，并且容量由总负载而不是模板的精确数量确定。通过另一种解释，在任何给定的时间，只有一个搜索模板可以在视觉工作记忆中处于活动状态，而其他模板处于辅助状态，但不影响视觉选择。我们通过以两种方式改变视觉觅食任务中40个目标和干扰物的数量来解决这个问题：1）固定干扰物数量，涉及两种干扰物类型，而目标类别从一到四个不等。2)固定颜色数（7），因此，如果目标类型是两个，干扰物类型是五个，而如果目标数量增加到三个，干扰物类型是四个（等等）。这两个账户做出了不同的预测。在单一模板帐户下，我们应该预期目标类型增加到两个时会产生较大的转换成本，但目标类型增加到两个以上时转换成本不会增加太多。负荷帐户预测开关成本随着目标类型数量的增加而近似线性增加。结果是，开关成本大致线性增加，在这两种情况下，与负载帐户。结果进行了讨论，在最近的建议，工作记忆反映挥之不去的神经活动在不同的网站上操作的刺激，在每种情况下，研究结果显示神经沉默的工作记忆表示。
A key assumption in the literature on visual attention is that templates, actively maintained in visual working memory (VWM), guide visual attention. An important question therefore involves the nature and capacity of VWM. According to load theories, more than one search template can be active at the same time and capacity is determined by the total load rather than a precise number of templates. By an alternative account only one search template can be active within visual working memory at any given time, while other templates are in an accessory state but do not affect visual selection. We addressed this question by varying the number of targets and distractors in a visual foraging task for 40 targets among 40 distractors in two ways: 1) Fixed-distractor-number, involving two distractor types while target categories varied from one to four. 2) Fixed-color-number (7), so that if the target types were two, distractors types were five, while if target number increased to three, distractor types were four (etc.). The two accounts make differing predictions. Under the single-template account, we should expect large switch costs as target types increase to two, but switch-costs should not increase much as target types increase beyond two. Load accounts predict an approximately linear increase in switch costs with increased target type number. The results were that switch costs increased roughly linearly in both conditions, in line with load accounts. The results are discussed in light of recent proposals that working memory reflects lingering neural activity at various sites that operate on the stimuli in each case and findings showing neurally silent working memory representations.