Social crowding is related to negative emotional experiences, concrete or low-level cognitive construal and even aggressive behaviors. Researchers have proposed that these consequences of crowding may reflect the defensive responses following violation of personal space. It has been demonstrated that personal proximity activates the amygdala which is sensitive to threatening information. Even a visually approached threatening face would increase people’s interpersonal distance preferences. On the other side, correctly recognizing others’ angry or fearful facial expressions plays a crucial role in social interaction. How does the feelings of crowded affect people’s reaction to angry or fearful expressions? To answer this question, the present study aimed to investigate the effects of social crowding priming on perception of threatening facial expressions within the framework of Signal Detection Theory (SDT). SDT allows for an inspection on both discriminability (d′: an index of perception sensitivity) and judgement criteria (β: an index of response bias) independently. Twenty-eight undergraduates participated in the experiment. A priming paradigm was applied to evoke the feelings of crowding with socially crowded pictures. And two additional types of priming pictures (visually clustered, and uncrowded) were used as control stimuli. In each trial, a face picture with either threatening or neutral expression was following a priming picture. Participants were required to judge whether the emotion on each face picture was threatening or not, for both the angry-neutral and fearful-neutral recognition tasks, respectively. These 3(crowding conditions) × 2(emotion recognition tasks) conditions were presented in separate blocks, with 10 threatening and 10 neutral face pictures in each block. After the recognition, participants also rated the emotional face for level of emotional strength with a 9-point scale. Automatic physiological responses, including skin conductance, finger pulse, finger temperature and respiration, were recorded continuously throughout the whole procedure by ProComp infinitiTM (Thought Technology Ltd, Canada). The physiological responses during a passive viewing phase preceding the emotion recognition task, while participants were viewing the three types of priming without any behavioral task, were analyzed and validated the manipulation of the visual priming of crowdedness. Results showed that social crowding affected the discriminability (d′) rather than the response bias (β) in the angry expressions discriminative task. Specifically, when being required to report whether a facial expression was angry or neutral, participants’ discriminability (d′) was reduced under the social crowding priming condition as compared to the uncrowded condition. However, in the fearful expression discrimination task, neither the discriminability (d′) nor the response bias (β) was affected by the social crowding priming. Furthermore, higher emotional strength rating was reported for angry expressions when primed with social crowding than uncrowded pictures. No significant difference between crowding conditions was found for the strength ratings of fearful or neutral expressions. In addition，compared to angry expression, fearful expression processing was related to higher discriminability, lower response bias, and shorter latency. These results indicated the selective effects of social crowding on the processing of angry facial expressions. When people feel crowded, it is possibly the early perceptual disruption rather than the response related bias that underlies the impaired discrimination of angry expression.