We reveal patterns of circadian variations in our collective emotions from the analysis of 800M anonymous microposts collected on Twitter across the United Kingdom. We use methods of Fourier analysis and statistical analysis to investigate periodic events in the use of words denoting a positive emotion, a negative emotion, or fatigue. Our research reveals two strong patterns: while the positive emotions show a bi-phasic behavior, peaking once in the morning and once in the night, the negative emotions reach their minima in the morning period, they gradually increase during the day to reach a peak late at night. Our research confirms previous results reported on diurnal mood variations and shows finer aspects of mood variations: sadness and the positive emotions are found to vary in response to the change of activity between the weekend and the weekdays, these two emotions are also found to vary between the seasons, showing interaction with the onset of sunlight exposure. The variations of anger and fatigue remain more stable in response to these changes in the external environment.
In the context of hormonal changes that occur during the day, anger shows a pattern that inversely mirrors the known circadian variation of plasma cortisol concentrations. While our study design prevents us from establishing a causal link between hormones and emotions, our research makes apparent interesting hormonal correlates such as cortisol. New research will be needed to investigate further, and separate the effect of other potential clues such as sleep pressure building up during the day, or diurnal changes in our social environment.
To the best of our knowledge this study is the largest of its kind, the patterns of circadian variations were sampled across the 54 largest cities in the United Kingdom and averaged over the course of four years. We reveal accurate patterns of variation for meal times, as well as other natural zeitgebers related to seasonal changes such as light or temperature. These methods of sampling and statistical analysis that we demonstrate on the social media provide valuable tools for the study of our emotions and for the understanding of their interaction within the circadian rhythm. We hope this study will encourage other researchers to use the social media to help in our understanding of the brain and mental health disorders.