The measurement and analysis of physiological data is part of the project of the Swiss National Science Foundation "Emotional processes in natural situations: a computer-based algorithm supporting the analysis of their physiological and expressive components" by M.L. Käsermann, A. Altorfer and K. Foppa. One of the project's main tools is the program VASC Analyzer which can automatically analyze large quantities of data for individual single-case studies or interpersonal comparisons. About 30 minutes of a conversation between two human subjects are examined. During the session, the spoken contents are recorded and the peripheral blood flow (PBF) is measured with a photoelectric sensor (plethysmograph) attached to a finger. The PBF data correlates highly to the heart beat and the constriction and dilatation of the small blood vessels that result from changes of arousal of the subject.

The usual procedure to analyze peripheral blood flow or heart beat is to measure the Inter-Beat-Interval (IBI) which means that a person with e.g. 60 heart beats per minute delivers 60 data points per minute. If the resolution of the data has to be higher, then the values between the peaks are extrapolated as a rough estimate. However, there lies much more information between the peaks. The program VASC Analyzer can extract the periodic information between the peaks of any wave of data. First the PBF data is transformed into the Vascular-Volume-Data (the quantity of blood which is circulating through the small finger blood vessels over time). This is done with a "Sequel Analysis" which operates with a combination of highpass and lowpass filters. These Vascular-Volume-Data display a greater density of measured, not estimated information like IBI algorithms.

An important principle of VASC Analyzer is to continuously build higher levels of abstraction without loosing the primary information. The lowest level is the PBF data and the highest level are classes of similar Vascular-Volume Motions which still show the time of occurrence, the initial values and the change of values of each member of the class. This is essential if these classes are to be linked to the original speech or video-tape. VASC Analyzer starts at pointing to relevant Vascular-Volume-Data Events with a motion-detection algorithm. These events are significant changes of the Vascular-Volume-Data flow. It is also capable of locating biologically unlikely events like artefacts or missing values. Local processes of arousal can consist of a group of several succeeding Vascular-Volume Events and are called Vascular-Volume Motions. The most important part is the classification of these Vascular-Volume Motions. Several algorithms of pattern and template matching are being used to build classes of similar Vascular-Volume Motions. Every motion within a class is kept together with its occurrence in time. In this way it is possible to focus on episodes in the recorded conversation with similar Vascular-Volume processes and episodes with extraordinary values. In this respect, communicative interaction and psychophysiological processes can be investigated by using methodologically linked systems of analysis which concern verbal utterances and emotional arousal [1].

References

1. Käsermann, M.L.; Altorfer, A.; Foppa, N.; Jossen, S.; Zimmermann, H. (1998). Measuring emotionalisation in everyday face-to-face communicative interaction. This volume.