Seamless Cognition: Visual Processing at the Intersection of Different Cognitive Faculties

不同认知能力(即知觉、注意力、视觉短期记忆-VSTM，长期记忆-LTM)的无缝整合是人类思维成熟的标志之一。在孤立地理解这些过程方面已经取得了很大进展，但关于它们如何相交的知识有限。 例如，视觉LTM和VSTM对表征精度具有相同的限制，但尚不清楚是否相同的因素对跨LTM和VSTM的表征精度有影响。我的实验室通过研究记忆干扰的过程，测试了有关记忆精确度和可及性的问题：竞争信息扰乱了与任务相关的处理。在LTM研究中，相似性假设认为，一个人的记忆损伤程度随着干扰信息和记忆内容之间的相似性而增加。记忆损伤有不同的形式；在受到干扰后，学习的信息可能会被保留下来，但精确度会降低，或者完全被删除。 之前的研究缺乏将这两种选择分开的决心，因为它们使用的是离散的、强迫选择的反应。使用混合建模的连续自由回忆任务估计擦除和退化的不同贡献，我们发现高度相似的项目的干扰导致记忆擦除，而中等相似的项目的干扰降低了记忆的精度。参与者通常会抵制高度相似的物品的干扰。因此，我们证明了相似性假设并不普遍适用：视觉相似性的性质可能会使天平倾向于牺牲记忆精度而有利于记忆可获得性，反之亦然。 我现在正在提出一个新的研究计划，它将揭示认知的不同方面是如何相互影响的。我的建议是，关注颜色空间中的某个特定值可以增强对该值的处理，并减弱对高度相似但不相同的颜色值的处理，类似于环绕抑制。在相关的工作中，我们将展示对任务无关新颖性的检测将调节VSTM的精度。最后，我们将使用集合统计方法来表明早期视觉皮质可以代表看不见的对象。我和我的学员将从行为和不同的神经成像方法(例如，EEG和fMRI)来测试我们的假设。 这项研究计划的首要目标是确定能够实现无缝认知的基础认知和神经过程。我们的主观体验很少会强调从一种认知能力到另一种认知能力的转换；相反，我们会毫不费力地在不同的时间尺度上进行各种认知操作。这些不同的认知能力可能具有相反或互补的加工偏向。我们将展示这些过程是如何相互加强或减弱的。

The seamless integration of different cognitive faculties (i.e., perception, attention, visual short-term memory - VSTM, long-term memory - LTM), is one of the hallmarks of the human mind's sophistication. Much progress has been made in understanding these processes in isolation but knowledge about how they intersect is limited. For example, visual LTM and VSTM have the same limit on representational precision but it is unclear if the same factors contribute to the precision of representations across LTM and VSTM. My lab has tested issues around memory precision and accessibility by studying the process of memory interference: the disruption of task-relevant processing by competing information. In LTM research, the similarity assumption holds that one's degree of memory impairment increases with the similarity between interfering information and memory contents. Memory impairments come in different flavors; following interference, studied information may be retained with degraded precision or erased completely. Prior studies lacked the resolution to disentangle these two alternatives, as they used discrete, forced-choice responses. Using a continuous free-recall task with mixture modeling to estimate the separate contributions of erasing and degrading, we found that interference from highly dissimilar items resulted in memory erasure and that interference from items of intermediate similarity degraded memory's precision. Participants typically resisted interference from highly similar items. Thus we demonstrated that the similarity assumption does not hold universally: the nature of visual similarity may tip the balance towards sacrificing memory precision in favor of memory accessibility, or vice versa. I am now proposing a novel program of study that will uncover how different aspects of cognition inform each other. I am proposing that attending to a specific value in colour space enhances the processing of that value and attenuates the processing of highly-similar but nonidentical colour values, akin to surround suppression. In a related line of work, we will show that the detection of task-irrelevant novelty will modulate VSTM precision. Finally, we will use an ensemble statistics approach to show that early visual cortex can represent unseen objects. My trainees and I will test our hypotheses behaviourally and with different neuroimaging approaches (e.g., EEG and fMRI). The overarching goal of this program of research is to identify the foundational cognitive and neural processes that enable seamless cognition. Rarely does our subjective experience highlight a switch from one cognitive faculty to another; instead, we effortlessly access various cognitive operations across different timescales. These different cognitive faculties may have opposing or complementary processing biases. We will show how these processes mutually accentuate or attenuate each other.