Using plants as biosensors to measure the emotions of jazz musicians

This chapter proposes a novel non-invasive, and privacy-preserving approach to quantify human emotions by tracking the action potential activations in plants positioned in the vicinity of humans. The production of human electrostatic discharge triggers changes in the electrical action potential values of a Basil plant (Ocimum basilicum) that acts as a highly sensitive movement and sound sensor. We construct a machine learning model to predict human emotions using features extracted from the plant’s electrical signals. The data is collected from a two-hour jazz rehearsal session with an orchestra of 19 musicians. Moreover, the emotions of musicians are also measured with two well-tested systems to collect ground truth. The first is facial emotion recognition using CNNs by obtaining camera-recorded videos of the musicians. The second is smartwatch sensor data of physiological signals like heart rate and body movement. On experimental evaluation, we obtain Pearson correlation values and regression coefficients for group facial emotions and plant Mel Frequency Cepstral Coefficients (MFCCs). Our machine learning approach achieves a training and test accuracy of 69 percent and 64 percent, respectively.