Integration of perception and conception in the human brain

在任何一天，我们都与数百个物体互动。这种能力对我们的生存至关重要。当我们看到一个物体——比如吹风机——我们能够毫不费力地识别它，并以符合其功能的方式使用它。例如，我们不会把它误认为是手持式电钻，尽管它们有相似的视觉外观和相似的手部动作。相反，尽管胶带和胶水看起来不一样，手的动作也不相似，但我们毫不费力地意识到，它们都是将两件东西粘在一起的合适对象。因此，在任何给定的一组物体中，许多视觉特征将是高度相似的，并不总是与物体的识别有直接的对应关系。那么，我们如何能够理解我们所看到的呢？在一系列相互关联的项目中，我们将解决这个问题。我们将专注于一个核心假设：理解和与物体互动的能力取决于一种特定的神经编码。具体来说，我们将论证，因为无论是感知特征还是概念特征都不能完美地诊断一个物体的身份，当它们单独考虑时，大脑必须整合感知和概念特征来做出高度具体的物体表征。我们将展示，当面对许多分散手头任务的特征时，这些集成的表示对于以灵活的方式行事是必不可少的。项目一将使用尖端的神经成像技术来绘制大脑中的信息模式。我们预计将观察到一个综合信息代码，这样当参与者思考物体的视觉属性时，大脑活动也会反映这些物体的概念属性（反之亦然）。我们预计会在周围皮层观察到这一点，这是大脑的一部分，它既连接着感知的重要区域，也连接着概念处理的重要区域。因此，它处于整合这些信息流的有利位置。项目2将调查当这个集成信息代码被破坏时，行为是如何受到影响的。具体来说，我们将研究由于急性损伤（例如感染或肿瘤）而导致的鼻周皮质损伤的患者，以及那些由于阿尔茨海默病的早期阶段而导致轻微的鼻周损伤的患者。项目3将研究不同的大脑区域如何相互沟通，以了解神经编码如何根据任务需求动态适应。项目4将首先研究对象知识是如何形成的——重点关注信息是如何从多种形式集成到一个对象上，从而创建一个连贯的表示。对于每个项目，我们将创建大型刺激集，以精确地表征对象的感知和概念特征。这些刺激集可能对其他研究人员有帮助，所以我们将免费分享它们。

In any given day, we interact with hundreds of objects. The ability to do so is central to our survival. When we look at an object - a hairdryer, for example - we are able to effortlessly recognize it and use it in a way that matches its function. We don't, for example, mistake it for the handheld power drill, despite the fact that they have a similar visual appearance and engage similar hand movements. And conversely, despite the fact that tape and glue neither look alike nor engage similar hand movements, we have no trouble recognizing that they would be both be suitable objects for the task of sticking two things together. Thus, in any given set of objects, many visual features will be highly similar and will not always have a direct correspondence with an object's identify. How, then, are we able to understand what we see? In a series of inter-related projects, we will address this question. We will focus on a core hypothesis: that the ability to understand and interact with objects depends on a particular kind of neural code. Specifically, we will argue that because neither perceptual nor conceptual features are perfectly diagnostic for an object's identity when considered on their own, the brain must integrate perceptual and conceptual features to make a highly specific object representation. We will show that these integrated representations are essential to behaving in a flexible manner when faced with many features that are distracting to the task at hand. Project 1 will use cutting-edge neuroimaging techniques to map information patterns in the brain. We predict to observe an integrated information code, such that when participants think about the visual attributes of objects, brain activity will also reflect the conceptual properties of those objects (and vice versa). We predict to observe this in the perirhinal cortex, a part of the brain that is connected both to regions important for perception, as well as to regions important for conceptual processing. Thus, it is well-placed to integrate these streams of information. Project 2 will investigate how behaviour is affected when this integrated information code is compromised. Specifically, we will work with patients who have perirhinal cortex damage due to an acute injury (e.g., an infection or tumor), and also those who have subtle perirhinal damage because they are in the early stages of Alzheimer's disease. Project 3 will investigate how different brain regions communicate with each other to understand how neural coding dynamically adapts based on task demands. Project 4 will investigate how object knowledge is formed in the first place - focusing on how information becomes integrated from multiple modalities to create a coherent representation if an object. For each project we will create large stimulus sets that precisely characterize the objects' perceptual and conceptual features. These stimulus sets may be helpful to other researchers, and so we will freely share them.