Abstract: To recognize emotional attitudes of body language accurately in human-computer interaction,a surface electromyography based emotion recognition model of head movements is proposed.Aiming at analyzing attitudes of agreement and disagreement expressed spontaneously by nodding and shaking head,we recorded the surface electromyographic signals from splenius capitis,sternocleidomastoids and trapeziums of 8 participants.Using one-way ANOVA,10 features of electromyographic time domain parameters,with significant differences between two attitudes,were extracted as input parameters.The emotion recognition model of head nodding and shaking was constructed using Elman neural network.Finally,the performance of the model was compared with other two emotion recognition models using BP neural network and support vector machine.Experimental results show that correct recognition rates of our model on the test set with agreement and disagreement emotional attitudes are more than 96%,which demonstrates the reliability of the presented model and method in this paper.