Heart rate response to music: An artificial intelligence study on healthy and traumatic brain-injured subjects

Background and rationale: Investigation of the brain's emotional response to music is limited by methodological problems mainly related to the characterization of the emotions and concomitant brain conditions. In this study, artificial intelligence procedures were applied to identify significant music-induced changes in heart rate variability and to classify autonomic reactions to stimuli requiring complex brain operations. Both healthy subjects and traumatic brain-injury (TBI) patients were studied in order to test the method's validity. Methods: 16 TBI patents and 26 healthy subjects were requested to listen to selected music samples while the heart beat was continuously recorded. The parametric and nonparametric frequency spectra were computed on the heart rate and the spectra descriptors were entered into a 1-R rules (very simple classification rules) data-mining procedure. Data-mining procedures independently classified the heart-rate spectral patterns and the emotions reported by subjects as positive, indifferent, or negative. Results and conclusions: The data-mining procedures sorted the nu_LF descriptor as the spectral parameter that allowed clustering the emotions reported by the subjects as positive and negative. Classification by nu_LF was comparable to that by self-reported emotions, with an overall correct classification by author in 76.0% of controls and 70% of patients. The identification of negative and positive emotions was correct in 81.3% and 68.9% of controls and in 65% and 74% of TBI patients, without significant differences between healthy subjects and TBI patients. This observation suggests that autonomic concomitants of emotions are detectable in response to complex emotional stimuli. © 2008 Federation of European Psychophysiology Societies