The ability to experience intense emotional feelings when listening to music is a common experience among listeners (Goldstein, 1980; Panksepp, 1995; Sloboda 1991). These feelings can be experienced physically as chills throughout the body, especially in the dorsal neck-area and the upper spine (Goldstein, 1980). Not all music researchers refer to this phenomenon in the same way. Several commonly used terms include thrills, shivers and goose-bumps (Huron, 2006; Panksepp, 1995). In order to avoid confusion, I will adopt the French term frisson, as suggested by music researcher John Sloboda. Frisson, a moment of intense excitement, often manifests itself physiologically as the involuntary reaction of one's hairs standing on end.
 Relatively little is known about the phenomenon frisson. Several studies confirm that it is a common experience for those listening to music (Goldstein, 1980; Panksepp, 1995; Sloboda, 1991). Frisson is often attributed to a section of a musical work rather than an entire piece (Sloboda, 1991). More specifically, frisson is associated with particular musical events, such as unexpected harmonies and, to a lesser extent, sudden dynamic or textural changes (Sloboda, 1991). The entrance of a new part, especially if it is both in a high register and sustained, has also been associated with frisson (Grewe, Nagel, Kopiez, & Altenmüller, 2005; Panksepp, 2002). It has been hypothesized that such musical events are capable of evoking emotional responses, such as frisson, because they violate listeners' expectations (Meyer, 1956).
 Emotional reactions to music, e.g. frisson, are difficult to quantify (Sloboda, 1985). Electrodermal activity (EDA), measured as Galvanic Skin Response (GSR) or skin conductance (SC), has been shown to be one of the most reliable physiological measurements for intense emotional experiences (Rickard, 2004). Additionally, a correlation has been discovered between physiological and subjective frisson-related experiences; specifically, increases in EDA were positively correlated with perceived emotionality and tension (Steinbeis, Koelsch, & Sloboda, 2006). When a listener is actively attending to music, we can expect changes in EDA if they experience any intense emotional reaction. Studies have found, however, that frisson is not an automatic response, but rather the result of "attentive, experienced and conscious" listening (Grewe et al., 2005 p. 448).
 Some studies have shown that frisson is the result of unexpected musical events (Sloboda, 1991), but none have attempted to experimentally determine why and how these musical events elicit frisson-related responses. Further, there has been no research that has considered the effects of attention on frisson-related responses as the result of unexpected musical events. This study sought to experimentally test whether the listener's state of consciousness—being either relaxed or attentive—would have an effect on their ability to have frisson-related experiences, which were operationally defined as detectable changes in electrodermal activity (EDA). The hypothesis was that listeners in an attentive state of consciousness would be more likely to experience frisson-related responses than listeners in a relaxed state of consciousness.
 Fourteen undergraduate students (6 males, 8 females) were recruited from the student body at Elmhurst College in Elmhurst, IL. These students participated in the study to fulfill a course requirement. As noted on the sign-up sheet, participants were required to be at least 18 years of age and have self-assessed good hearing. Participants were randomly assigned to one of the experimental conditions.
 Music samples were presented automatically via an experimental software package (Neugraph) and heard through AKG-301XTRA headphones. EDA measurements were obtained using a Neurodyne Biofeedback System/3 multi-modality unit.
 The experiment utilized three musical samples. The first example was J.S. Bach's St. Matthew Passion (BWV 244) Part 2, No.54, "Barrabas!" This piece has been shown to create reliable frisson (Sloboda, 1991). The frisson-related musical event in this piece was an unexpected change in harmony. The duration of this sample was 02:09. The second example was Pat Metheny's Letter From Home. The frisson-related musical event in this piece was the entrance of a new part, a violin. The duration of this sample was 02:33. The third example was Pink Floyd's The Final Cut, which has also been shown to evoke reliable frisson (Huron, 2006; Panksepp, 1995). The frisson-related musical event in this piece is a sudden change in dynamics. The duration of this sample was 02:08.
 A 2 by 3 mixed design was used for this experiment. The between-subjects variable was one of two conditions: attentive or relaxed. The within-subjects variable was the musical event contained in each excerpt (unexpected harmony, unexpected dynamic and the entrance of a new voice). The dependent variable was operationally defined as a detectable change in EDA within a critical time period.
 Participants were seated in a comfortable chair in a psychology laboratory. The experimenter explained the procedure and asked participants to sign a consent form. The purpose of the study as explained to the participants was to measure the level of concentration through EDA activity; there was no mention of "frisson" or "chills" until after the experiment was completed. The EDA sensors were connected to the fingertips of the index and middle fingers on the participant's non-dominant hand. Once baseline physiological measurements were recorded, the experimenter started the automated experiment and left the room. Participants in both conditions were given instructions via audio prompts. The attentive group was asked to give their undivided attention to the music; they were also given basic information about each piece of music to help them with their focused listening task (i.e., title, composer and brief description). The relaxed group was instructed to take several deep breaths, close their eyes and relax deeply as music was played in the background. The relaxed condition was explicitly instructed to focus on the sound of their breathing and told that background music would be played for the sole purpose of blocking out external noise. Instructions were repeated before the beginning of each piece, which also allowed enough time for EDA measurements to return to baseline. The experiment was finished after each participant heard all three musical excerpts. The participants were then debriefed and thanked for their time. The total duration of each session was approximately 30 minutes.
 A physiological measurement increase within a critical time period was used as the operational definition for a frisson-related response. For the EDA data, the peak (baseline corrected) was defined as the highest value within a five second window from stimulus onset (critical musical event), as suggested by Dawson, Schell and Filion (2000). The baseline was defined as the average measurement of the two seconds before stimulus onset. The EDA data were carefully screened for artifacts, but none were found within the critical time periods. Because the EDA scores differed greatly from a normal distribution, a log transformation was applied to the data (see Tabachnick & Fiddell, 2007). See Figure 1.
Figure 1. EDA response as a function of Condition (Attentive/Relaxed) and Musical Event.
 Descriptive statistics for the data-set (Table 1) showed that unexpected harmonies and unexpected dynamics caused a greater EDA response for the attentive condition and that the entrance of a new voice caused a greater EDA response for the relaxed condition. The data were analyzed using a 2 (condition) by 3 (musical event) mixed analysis of variance, with the alpha level set at .05. The test revealed no significant results for condition, F(2,24) = .543. or musical event, F(2,24) = 1.735. Therefore, the null hypothesis failed to be rejected and suggested that frisson-related events can occur regardless of one's state of consciousness.
Table 1. Mean EDA response as a function of Condition and Musical Event.
|Unexpected Harmony||Unexpected Dynamic||Unexpected Entrance|
|Attentive||4.327 (SD 6.88)||.146 (SD 1.17)||1.987 (SD 4.23)|
|Relaxed||1.679 (SD 2.93)||.098 (SD 1.39)||2.446 (SD 6.73)|
 The purpose of this study was to determine if listeners in an attentive state of consciousness would be more likely to experience frisson-related responses than listeners in a relaxed state of consciousness. The results showed that the two conditions varied for the various musical events, but that these differences were insignificant.
 Descriptive statistics showed that unexpected harmonies elicited the biggest change in EDA and caused the biggest difference between conditions; unexpected dynamic changes elicited a smaller EDA change than unexpected harmonies; and the EDA data for the unexpected entrance of a new voice were inconclusive. This information is consistent with Sloboda's (1991) finding that unexpected harmonies, and to a lesser extent unexpected dynamics, are capable of eliciting frisson-related responses. The results do not support Grewe et al (2005) finding that the entrance of a new voice can also elicit a frisson-related response.
 Consistent with previous research, frisson-related responses are highly variable and are difficult to recreate in a laboratory (Grewe et al., 2005; Panksepp, 1995; Sloboda, 1991). EDA responses ranged from subtle to extreme and it is likely that the results would be more consistent if the participants were familiar with the musical examples (Rickard, 2004). Because of this high variability, frisson could be considered a subjective post facto contextual phenomenon, i.e. the variability is likely due to both subjective and contextual factors. Subjective factors include one's ability to experience frisson, musical background, personality and musical preferences whereas contextual factors include things such as listening environment, mood and time of day.
 This experiment did not find evidence to support the claim of Grewe et al. (2005) that frisson is the result of "attentive, experienced and conscious" listening. Attentive listeners, who were encouraged to focus on the examples, were not more likely to react to frisson-related musical events than relaxed listeners. Further, being in a relaxed state of consciousness actually caused a greater EDA response to one type of musical event, the entrance of a new voice, although this difference was not significant. Therefore, it seems unlikely that attentive and conscious listening are requirements for frisson.
 If both attentive and relaxed listeners can experience frisson-related responses to musical events, perhaps the musical event itself is capable of reorienting the listener's attention. This would mean that frisson-related musical events might function as "attention-grabbers." One possible explanation for this response is that our brains are highly tuned to detect musical structures with a low probability of occurrence (Temperly, 2007). The idea that frisson-related musical structures function as mental-orienting responses seems to be one of the most economical theories in explaining the findings of previous frisson research.
 A mental-orienting response is similar to a startle response, but is less obvious and abrasive. Research has shown that startle responses can be caused by unexpected musical events—such as dynamic changes—and that the perception of these responses ranges from arousing and beneficial to adverse and negative (Ravaka & Kallinen, 2004). Such research provides evidence for the subjective nature of frisson.
 Limitations of this study included the difficulty in determining one's state of conscious without directly asking participants. It is therefore possible that those in the attentive or relaxed conditions were actually in another state of consciousness. Further, the complex nature of frisson cannot be solely monitored and detected by measuring EDA. Future research needs to determine a way to record physiological and subjective data in a manner that is congruent with one's typical listening environment. Panksepp & Bernatzky (2002) have suggested the use of thermal-imaging—a non-invasive technique which holds great potential for future frisson research.
 A greater understanding of frisson could provide composers with the ability to elicit strong emotional reactions to their music. Additionally, frisson research could advance theories about music and emotion which debate music's ability to merely resemble or actually elicit different types of emotion (Krumhansl, 1997). Finally, an understanding of frisson could contribute to various theories of human emotion, which would generate numerous commercial and non-commercial applications.