Functional Anatomy of Face Processing in the Primate Brain

人类颞叶下皮层的病变会导致一种名为面容失认症的临床综合征，即无法识别熟悉的面孔。 来自猴子IT皮层的单细胞记录揭示了神经元的存在，这些神经元被面部的视觉图像选择性地激活。 最近，对人类和猴子的脑成像研究已经证明了面部选择性区域，在这些区域中，与其他物体类别（如非面部物体和地点）诱发的fMRI信号相比，面部诱发的fMRI信号大大增强。 然而，对这些fMRI定义区域内外的个体神经元反应知之甚少。 在一项研究中，我们研究了猴子IT皮层中fMRI识别的面部选择性区域之间的关系，相对于单个神经元和局部场电位（LFP）的选择性，位于这些区域内与外部。 我们证实，面部偏好神经元最集中在对应的fMRI识别的面部选择性区域。 然而，我们也发现了fMRI定义的区域之外的这种面部偏好神经元，浓度降低。 这两个群体的性质不同：位于fMRI定义的面部斑块内的神经元对面部表现出更大的选择性，与位于这些区域之外的神经元相比。 因此，fMRI识别的面部选择性区域对应于高比例的面部反应神经元，这些神经元对面部具有高度选择性。 这些发现有助于阐明fMRI定义的区域和其中的神经元过程之间的关系。 尽管在人类和猴子的颞叶和前额叶皮层中都发现了面孔选择性区域，但面孔选择性背后的神经回路仍不清楚。为了澄清这一点，我们研究了功能连接，这已被证明是一种有效的方法来揭示神经回路，在这些面孔选择性区域在休息，清醒状态的恒河猴。首先，我们通过对比猴子面部和非面部物体的fMRI激活来映射面部选择区域。两个面部选择性区域的下颞叶皮质通常被发现在每个半球：前面和后面的补丁。然后，这些动物接受了10分钟的静息状态扫描。 从每个半球的正面和背面补丁的静息状态的平均时间过程被用作种子的功能连接分析。 我们发现，一个种子放置在后面部补丁的一个半球与活动在后面部补丁的另一个半球和前补丁，前额叶面部选择区和杏仁核的两个半球。将种子置于前面部贴片显示出类似的功能连接，但连接强度不同：杏仁核较强，前额叶面部选择区域较弱。 这些结果表明，有一个功能网络之间的面孔选择性区域，这可以通过研究内在的，自发的fMRI信号波动检测。此外，功能连接似乎反映了潜在的神经解剖学。 我们之前已经证明，面部表情调节猴子杏仁核和视觉皮层中面部反应区域的fMRI活动。 具体来说，情绪的面部表情比中性的面部表情产生更大的激活，这种现象被称为效价效应。 我们接下来测试了杏仁核损伤会消除对视觉皮层的情绪调节反馈，从而破坏那里的效价效应的想法。 我们对猴子进行了选择性杏仁核损伤，然后使用功能磁共振成像比较这些动物视觉皮层内的效价效应与正常对照动物的效应。 测试了四种不同的面部表情：中性，攻击性（张嘴威胁），恐惧（恐惧鬼脸）和安抚（嘴唇）。 在对照组中，相对于中性面孔，具有情绪表达的面孔在面孔选择性区域产生了增强的反应，正如预期的那样。 恐惧的鬼脸表情总是引起最大的反应。 在杏仁核受损的猴子中，视觉皮层内的效价效应被大大破坏。 杏仁核损伤最严重的大脑半球最能证明这种破坏。 在这个半球，虽然面部选择性补丁被发现在IT皮层，他们的活动不调制面部表情。 相反，在三个半球与备用组织在杏仁核的前部（这是激活的中性面孔和调制的面部表情），在IT皮层的面孔选择补丁的效价效应。 和对照组一样，在前杏仁核不受影响的大脑半球中，恐惧表情引起的反应最大。 总的来说，我们的数据表明，杏仁核是在视觉皮层中看到的效价调节效应的来源。 这些结果对我们理解具有情绪内容的面部刺激的神经处理做出了重大贡献。

Lesions of the inferior temporal (IT) cortex in humans can result in the clinical syndrome termed prosopagnosia, an inability to recognize familiar faces. Single-cell recordings from the IT cortex of monkeys have revealed the existence of neurons that are selectively activated by visual images of faces. More recently, brain imaging studies in both humans and monkeys have demonstrated face-selective regions, in which the fMRI signal evoked by faces is greatly enhanced compared to that evoked by other object categories, such as non-face objects and places. However, relatively little is known about the individual neuronal responses within versus outside these fMRI-defined regions. In one study, we examined the relationship between fMRI-identified face-selective regions in the IT cortex of monkeys, relative to the selectivity of single neurons and local field potentials (LFPs), located within vs. outside these regions. We confirmed that face-preferring neurons were most concentrated in areas corresponding to the fMRI-identified face-selective regions. However, we also found such face-preferring neurons outside of the fMRI-defined regions, at decreased concentrations. These two populations differed qualitatively: neurons located within the fMRI-defined face patches showed greater selectivity to faces, compared to those located outside these regions. Thus, fMRI-identified face-selective regions correspond to a high proportion of face-responsive neurons that are highly selective for faces. These findings help to clarify the relationship between fMRI-defined regions and the neuronal processes within them. Although face-selective regions have been reported in temporal and prefrontal cortex of both human and monkey subjects, the neural circuitry underlying face selectivity remains unclear. To clarify this, we studied the functional connectivity, which has been demonstrated to be an efficient method for revealing neural circuits, among these face-selective regions in rhesus monkeys in the resting, awake state. First, we mapped the face-selective regions by contrasting fMRI activation to images of monkey faces versus non-face objects. Two face-selective regions in the inferior temporal cortex were typically found in each hemisphere: the anterior and posterior face patches. Then, the animals underwent ten minutes of resting-state scans. Resting-state average time courses from the anterior and posterior face patches of each hemisphere were used as a seed for functional connectivity analyses. We found that a seed placed in the posterior face patch of one hemisphere correlated with activity in the posterior face patch of the other hemisphere and in the anterior patch, prefrontal face-selective areas and the amygdala of both hemispheres. A seed placed in the anterior face patch showed similar functional connectivity, but with different connectional strengths: stronger in the amygdala and weaker in the prefrontal face-selective areas. These results demonstrate that there is a functional network among the face-selective regions, which can be detected by studying the intrinsic, spontaneous fMRI signal fluctuations. Moreover, the functional connectivity appears to reflect the underlying neuroanatomy. We had previously shown that facial expressions modulate fMRI activity in face-responsive regions of the monkeys amygdala and visual cortex. Specifically, facial expressions of emotion yield greater activation than neutral faces, a phenomenon known as the valence effect. We next tested the idea that amygdala lesions would eliminate emotional modulatory feedback to the visual cortex, thus disrupting valence effects seen there. We performed selective amygdala lesions in monkeys and then used fMRI to compare the valence effects within the visual cortex in these animals to the effects in normal control animals. Four different facial expressions were tested: neutral, aggressive (open mouth threat), fearful (fear grimace) and appeasing (lipsmack). In controls, faces with emotional expressions relative to neutral faces produced enhanced responses in face-selective regions, as expected. Fear grimace expressions consistently elicited the greatest responses. In monkeys with amygdala lesions, the valence effects within the visual cortex were greatly disrupted. The most striking demonstration of this disruption was found in the hemisphere with the most complete amygdala lesion. In this hemisphere, although face-selective patches were found in IT cortex, their activity was not modulated by facial expressions. Conversely, in three hemispheres with spared tissue in the anterior part of the amygdala (which was activated by neutral faces and modulated by facial expressions), the valence effects in the face-selective patches of IT cortex were present. Just as in controls, in these hemispheres with anterior amygdala sparing, fear grimace expressions evoked the greatest response. Overall, our data demonstrate that the amygdala is the source of the valence modulatory effects seen in the visual cortex. These results make a significant contribution to our understanding of the neural processing of facial stimuli with emotional content.