Detection of anger with and without control for affective computing systems

Feedback from affective computing interfaces should improve awareness and self-regulation of negative emotional states, such as anger. The aim of this study was to identify cardiovascular and electroencephalography-based indicators of anger, in combination with level of control. Forty participants were split into four experimental groups: anger/no control, anger/control, neutral/no control and neutral/control. Anger (anger state vs. neutral state) was manipulated via verbal mood induction. In addition, participants were exposed to a computer-based problem-solving task where the keyboard either worked correctly (control) or malfunctioned (no control). Various psychophysiological variables (including blood pressure, cardiovascular impedance, electroencephalography and facial electromyography), in addition to self-report variables, were obtained. Statistical analyses of self-report variables indicated that manipulation of anger and control was successful. Blood pressure and electroencephalography were found to be sensitive to the anger with no control state. Implications of the study are discussed in relation to the development of biocybernetic systems to monitor different categories of anger.