7.1 Introduction

That men are machines (whatever else they may be) has long been suspected; but not till our generation have men fairly felt in concrete just what wonderful psycho-neuro-physical mechanisms they are.

William James (1893; 1842 − 1910)

7.1 Introduction

Despite the early work of William James and others before him, it took more than a century before emotions were widely acknowledged and embraced by science and engineering. However, currently it is generally accepted that emotions cannot be ignored; they influence us, be it consciously or unconsciously, in a wide variety of ways [521]. We are (indeed) psycho-neuro-physical mechanisms [312, 440], who both send and perceive biosignals that can be captured; for example, by electromyography (EMG), electrocardiography (ECG), and electrodermal activity (EDA). See Table 1.1 for an overview. These biosignals can reveal a plethora of characteristics of people; for example, workload, attention, and emotions.

Several studies have been conducted in the field of ASP, using a broad range of signals, features, and classifiers; see Table 2.4 for an overview. Nonetheless, both the recognition performance and the number of emotions that the classifiers were able to discriminate were disappointing. Moreover, comparing the different studies is problematic because of:

1.The different settings the studies were applied in, ranging from controlled lab studies to real-world testing;

2.The type of emotion triggers used;

3.The number of target states to be discriminated; and

4.The signals and features employed.

All in all, the conclusion has to be that there is a lack of general standards, which results in low prediction accuracy and inconsistent results. However, for ASP to come to fruition, it is important to deal with these issues. This illustrates the need for a well documented general framework. In this chapter, I set out to initiate its development, to explore various possibilities, and to apply it on a data set that will be introduced in the next section.

In the pursuit of emotion-aware technology, I will describe our work on the automatic classification of biosignals. Hereby, we will following the complete signal processing + pattern recognition pipeline, as was described in Chapter 1. For an introduction of the statistical techniques that will be employed throughout this chapter, I refer to Appendix A. The data on which the complete signal processing + pattern recognition pipeline will be executed is the data as has been discussed in Chapters 3 and 4. I will refrain from repeating the complete description of this data set here and will only summary it. For the complete description

117