Despite continued acknowledgment of the importance of music in viewing films, empirical studies investigating the interaction of music with film are conspicuously absent. Evidence from a few isolated cognitive studies suggests that the relationship between music and film is additive. Little is known, however, about the physiology of emotional response to viewing film music stimuli. The present study utilized both self report as well as physiological indices to investigate the nature of the film and music relationship. Six-second films (having either negative or positive valence and low- or high-arousal) were paired with excerpts from instrumental classical music (pretested for valence and arousal). Results indicate a fairly straight-forward, additive relationship in terms of emotion self report. The modulating role of music on physiological reactions to film, however, was more complex. This study corroborates previous evidence regarding the subjective experience of viewing images with music. Physiological evidence, however, suggests that the interactions between music and film not always are predictable. The scientific examination of the physiological correlates of human emotional response dates back to the later nineteenth century, with the work of Darwin ( 1872/1969; Lewis & Haviiand-Jones, 2000). Since that time, many words have been written in an attempt to define what an emotion is-with varying degrees of success. Two general methods of classifying emotional response, however, have earned particular recognition: the categorical and dimensional approaches. The categorical or discrete emotions approach (e.g., Izard, 1977) posits the existence of several emotions (e.g., happiness, sadness, anger, fear, disgust) that can function either independently or as members of an interrelated group (Neumann & Waldstein, 2001 ). A major criticism of this approach, however, is the lack of an agreed-upon set number of these fundamental emotions (S loboda & J us Hn, 2001 ). In various iterations of the theory, this number ranges anywhere from seven through more than a dozen (Izard, 1993). Dimensional models of emotion date back over a century (Russell, 1997). Significant recent iterations include those by Russell ( 1980), Russell & Feldman-Barrett ( 1999), and Lang ( 1995). The dimensional model posits that every emolion can be expressed as a function of (usually) two core dimensions, frequently referred to as valence ana arousal. Valence is described in terms of a range of response from positive (happiness, pleasure) to negative (sadness, displeasure). Arousal ranges from low (indicating calmness, pcacefulncss) to high (excitement, energy). These two dimensions can be aligned orthogonally so as to produce four quadrants (Figure 1 ), each of which represents one of four general valence-arousal categories: ( 1 ) positive valence, low-arousal (PV/LA); (2) positive valence, high-arousal (PV/H A); (3) negative valence, high-arousal (NV/HA); and (4) negative valence, low-arousal (NV/LA). Specific emotions can then be plotted as a function of the two dimensions; anger is NV/HA, for example, while joy is PV/HA. Statistical evidence appears to favor such a two-dimensional, valence- arousal approach to emotion (Faith and Thayer, 2001). The dimensional approach is useful not only in classifying subjective (that is, verbal) labels of emotion, but also changes in physiology. Studies in which subjects view either still pictures (e.g., Lang, Greenwald, Bradley, & llamm, 1993; Lang, Oilman, & Vaitl, 1988) or moving images (Detcnbcr. 1995; Detenber, Simons, Detenbcr, Rocdcma, & Rciss. 1999; Simons, Roedema, & Rciss, 2000) suggest that the valence and arousal of the content of a visual image is consistently linked with changes in physiology-specifically, facial muscle activity, heart rate, and electrodermal activity. Facial muscle (electromyographic, or EMG) activity is measured from two muscles: zygomaticu. …