初级视皮质（V1）的神经元反应由简单的视觉刺激驱动，但是诸如注意、奖赏之类的认知功能和知觉学习训练可以调制V1的神经活动。在真实的生活场景中，中性的视觉刺激经常能够在经验的作用下带上情绪色彩，随之改变人对该视觉刺激的检测和辨别能力。fMRI和EEG的研究发现，视觉恐惧条件学习可以改变视觉皮质对刺激反应的活动水平，并且这种改变可以发生在V1。但目前的研究存在两个缺陷：第一，无法从单个神经元层面上描述V1反应特性的改变；第二，无法确定恐惧引起的V1反应改变是由高级脑区反馈所导致还是V1自己产生的。为了解决这两个问题，我们使用植入V1的电极阵列记录了2只猕猴在恐惧条件学习过程中神经元的活动。通过将特定朝向的光栅条纹作为条件刺激（conditioned stimulus, CS），并将其与恐惧/厌恶的非条件刺激（unconditioned stimulus, US）匹配呈现，我们发现在V1反应的极早期阶段（<40ms），具有CS特异性的恐惧相关信号几乎与V1的视觉反应同时出现，这提示我们恐惧相关信号源于V1本身，而非高级脑区向V1的反馈投射。我们通过实验验证了V1的恐惧相关信号为无意识的阈下信号，并且具有视野位置的特异性，从而进一步确认恐惧相关信号起源于自下而上的神经加工过程。此外，我们发现恐惧相关信号的产生与V1神经元的朝向偏好无关，即可以发生在任何V1神经元上，导致不同朝向偏好细胞的朝向调谐曲线发生不同的变化。我们还通过将CS光栅朝向与中性刺激（non-conditioned stimulus, NS）光栅朝向在实验过程中互换，考察了在反转学习过程中V1神经元反应的动态变化过程，发现只需要15-20次新的CS朝向与US匹配，V1恐惧相关信号就可以更新并迅速达到饱和。最后我们试图利用同步记录到的杏仁核与V1的神经活动进行相干性与格兰杰因果分析，以寻找V1可塑性变化的可能原因。虽然目前杏仁核与V1相互作用的结果仍有待进一步的验证和分析，但是基于V1的实验结果，我们提出了一个前馈调控的框架：恐惧学习可以通过改变前馈神经连接来影响视皮层神经元的编码特性，存储恐惧相关的内隐记忆，从而快速标记对即将来临的危险具有预示作用的视觉输入信号。

Neuronal responses in the primary visual cortex (V1) are driven by simple stimuli, but these stimuli evoked responses can be markedly modulated by non-sensory factors, such as attention and reward, and shaped by perceptual training. In real life situations, neutral visual stimuli can become emotionally tagged by experience, resulting in altered perceptual abilities to detect and discriminate these stimuli. Human imaging and EEG studies have shown that visual fear learning (the acquisition of aversive emotion associated with a visual stimulus) affects the activities in visual cortical areas as early as in V1. However, it remains elusive as to whether the fear-related activities seen in the early visual cortex have to do with feedback influences from other cortical areas; it is also unclear whether and how the response properties of V1 cells are modified during the fear learning. In the current study, we addressed these issues by recording from V1 of awake monkeys implanted with an array of microelectrodes. We found that the conditioned stimulus (CS) induced specific fear-related signals in V1 with a latency comparable to the visually-evoked responses (< 40ms). This suggests that the fear-related signals originate from V1 rather than feedback modulation from other brain regions. We also found that the fear-related signals were specific to the visual field location in a subliminal manner, which further confirmed a bottom-up mechanism. After grouping V1 neurons by their orientation preferences, we found that the fear-related signals were present in any group of V1 neurons, independent of their orientation preferences. By tracking the responses of V1 neurons during reverse learning, we found that the CS-specific signals emerged after 15-20 presentations of the new CS. We also attempted to examine the possible neural basis of V1 changes by conducting coherence and Granger causality analyses on simultaneously recorded activities from the amygdala and V1. We propose a feedforward framework whereby V1 responses are gated by subcortical structures and such gating mechanisms can be rapidly modified in the early phase of fear conditioning. Such a bottom-up mechanism could lead to formation of implicit fear memory in V1, allowing for proactive tagging of visual inputs that are predictive of imminent threat.