As subjective artistic creations, artistic paintings carry emotion of their creators. Emotions expressed in paintings and emotion aroused in spectators by paintings are two kinds of emotions that scholars have paid attention to. Traditional studies on emotions expressed by paintings are mainly conducted from qualitative perspectives, with neither quantitative output on the emotional values of a painting, nor exploration of trends in the expression of emotion in art history. In this research we threat facial expressions in paintings as an artistic characteristics of art history and employ cognitive computation technology to identify the facial emotions in paintings and to investigate the quantitative measures of paintings from three emotion-related aspects: the spatial and temporal patterns of painting emotions in art history, the gender difference on the emotion of paintings and the color preference associated with emotions. We discovered that the emotion of happiness has a growing trend from ancient to modern times in paintings history, and men and women have different facial expressions patterns along time. As for color preference, artists with different culture backgrounds had similar association preferences between colors and emotions.