At the most basic level, emotional responses to stimuli, including music, involve simple evaluations. Such evaluations are often measured with self-reports, typically by using rating scales with like a lot or extremely pleasant at one end, and dislike a lot or extremely unpleasant at the other end. If a perceiver likes one stimulus more than another, or considers one stimulus to be more pleasant, liking and pleasantness judgments extend readily to preferences.

These evaluative responses are more basic than so-called basic emotions (Ekman Reference Ekman1992b), such as happiness and sadness. This claim is supported by evidence of approach and avoidance behaviors in nonhuman species, and by the fact that brain imaging and lesion studies reveal differential activation solely on this basis. In the musical domain (for a review, see Peretz, in press), different brain areas are activated in response to consonant (pleasant sounding) and dissonant (unpleasant sounding) stimuli (Blood et al. Reference Blood, Zatorre, Bermudez and Evans1999; Gosselin et al. Reference Gosselin, Samson, Adolphs, Noulhiane, Roy, Hasboun, Baulac and Peretz2006; Koelsch et al. Reference Koelsch, Fritz, von Cramon, Müller and Friederici2006) and to music that sounds scary or threatening (Gosselin et al. Reference Gosselin, Peretz, Noulhiane, Hasboun, Beckett, Baulac and Samson2005; Reference Gosselin, Peretz, Johnsen and Adolphs2007). To date, however, there is no evidence of differential activation patterns in response to, say, happy and sad sounding music,Footnote ¹ probably because these responses do not map directly onto evaluations (i.e., valence; Russell Reference Russell1980). Although listeners typically exhibit a preference for happy-sounding over sad-sounding music (Gosselin et al. Reference Gosselin, Peretz, Noulhiane, Hasboun, Beckett, Baulac and Samson2005; Hunter et al. Reference Hunter, Schellenberg and Schimmack2008; Husain et al. Reference Husain, Thompson and Schellenberg2002; Schellenberg et al. Reference Schellenberg, Peretz and Vieillard2008; Thompson et al. Reference Thompson, Schellenberg and Husain2001), people often like and choose to listen to sad-sounding music. Indeed, listeners' typical preference for happy-sounding over sad-sounding music disappears as a consequence of manipulations that induce fatigue and frustration (Schellenberg et al. Reference Schellenberg, Peretz and Vieillard2008).

For many years, researchers have documented the role that exposure plays in stimulus evaluations (for a review, see Bornstein Reference Bornstein1989). For reasons that seem obvious from an evolutionary perspective, people and animals have an adaptive fear of the unknown (neophobia) that extends across modalities. After exposure to a particular stimulus reveals that it is relatively benign (i.e., with no adverse consequences), evaluative responses become more favorable (Zajonc Reference Zajonc2001). In line with this view, listeners respond more favorably to music and music-like stimuli they have heard previously compared to novel music (Peretz et al. Reference Peretz, Gaudreau and Bonnel1998; Schellenberg et al. Reference Schellenberg, Peretz and Vieillard2008; Szpunar et al. Reference Szpunar, Schellenberg and Pliner2004; Thompson et al. Reference Thompson, Balkwill and Vernescu2000), even if they do not recognize the stimuli (see Zajonc Reference Zajonc1980; Reference Zajonc2001). Because the to-be-exposed and novel stimuli are assigned randomly for each listener, favorable evaluations can be attributed to exposure rather than to stimulus differences. Such favorable responding is related to Juslin & Västfjäll's (J&V's) second mechanism (evaluative conditioning), but it differs substantially in that the association involves learned safety (Kalat & Rozin Reference Kalat and Rozin1973), which stems from the absence of negative consequences.

Liking often increases with additional exposure, a phenomenon that is typically attributed to perceptual fluency (Jacoby Reference Jacoby1983). On this view, a previously encountered stimulus is processed quickly and effortlessly, compared to a novel stimulus, because of the reactivation of an existing mental representation. When asked to make evaluations, people misinterpret this ease of processing as a favorable disposition toward the previously encountered stimulus. This perspective helps to explain increased liking as a function of exposure to stimuli that are aesthetically impoverished or highly controlled (e.g., line drawings: Kunst-Wilson & Zajonc Reference Kunst-Wilson and Zajonc1980; random tone sequences: Szpunar et al. Reference Szpunar, Schellenberg and Pliner2004). Nonetheless, positive misattributions should become less likely with further increases in exposure and explicit memory for the stimulus, such that processing fluency is an obvious consequence of exposure (Bornstein Reference Bornstein1989). For real music, however, there is evidence contrary to the hypothesis that liking and memory are associated negatively. Listeners often like pieces they remember (Schellenberg et al. Reference Schellenberg, Peretz and Vieillard2008; Szpunar et al. Reference Szpunar, Schellenberg and Pliner2004).

Berlyne's (Reference Berlyne1970; Reference Berlyne and Berlyne1974) two-factor model (which J&V mention in a different context) describes liking as a consequence of the arousal potential of a stimulus, which should be neither too great nor too small. Initial wariness towards a novel stimulus results from its arousal potential being too great. With additional exposure that has benign consequences, arousal potential is reduced to optimal levels. Finally, over-exposure leads to boredom as the arousal potential of the stimulus becomes less than optimal. In other words, Berlyne's first factor refers to increases in liking that accompany decreases in arousal potential, due to learned safety; the second factor refers to decreases in liking that accompany further decreases in arousal potential, due to satiety. Berlyne's model is under-specified in describing interactions between liking and memory, yet it explains increases in liking for music that accompany a moderate amount of exposure (e.g., recall when you heard The Macarena for, say, the third time), as well as decreases that occur as a consequence of over-exposure (when you heard The Macarena for the umpteenth time).

Indeed, there is abundant anecdotal evidence of increases followed by decreases in liking for music as a function of exposure. This inverted U-shaped function has also been documented systematically with real music (Schellenberg et al. Reference Schellenberg, Peretz and Vieillard2008; Szpunar et al. Reference Szpunar, Schellenberg and Pliner2004). Again, because the musical stimuli were assigned randomly to different exposure frequencies (i.e., 0, 2, 8, or 32) separately for each listener, the same stimulus was novel for some listeners, presented at moderate frequencies for other listeners, and over-exposed for still others. In short, the design ensured that liking ratings were independent of differences among stimuli and can be attributed solely to exposure frequency.

In sum, any consideration of mechanisms that underlie emotional responding to music must include exposure as a very basic mechanism, and learned safety, perceptual fluency, and satiety as sub-mechanisms that are related directly to exposure. As J&V acknowledge, failing to account for underlying mechanisms could lead to “inconsistent or non-interpretable” findings (target article, Abstract). This problem is likely to be particularly acute when a well-documented mechanism is ignored.