What Makes A Movie Scary? Do Sound or Video Contribute More To Fear?

Bekah Bowie

Wofford College, Department of Psychology


Macintosh HD:Users:rebekahcbowie:Desktop:a59375599f9ea2d185ca68440bc91fd4.jpg


            Loud, unpredictable dissonant and minor chords. A person slowly walking to look around a corner in the dark where something ominous awaits.  Someone running and screaming.  All are characteristics of horror movies.  But what exactly makes a movie scary?  Blumstein, Davitian, and Kaye (2010) investigated the characteristics of music in different genres of movies in an attempt to answer this question.  They found that when compared with films of another genre, horror films had fewer frequency transitions and contours, but had more non-harmonic sounds and loud screams, particularly those of a female.  From this, they concluded that filmmakers specifically use nonlinear analogues, such as large amplitudes and subharmonics, in their scary movies to elicit fearful emotions. 

A follow up study by Blumstein, Bryant, and Kaye (2012) analyzed the effects of neutral and distorted nonlinear musical compositions on emotions and their impact when paired with a neutral video.  Each participant in the study was instructed to rate the stimulus on a scale of -5 to 5 on its arousal (ability to evoke emotional stimulation) and valence (whether the stimulus was perceived as positive or negative).  The results showed that music containing distorted nonlinearities alone could elicit a fear response that was both highly arousing and negative.  However, when the neutral video was paired with the “scary” music, the resulting affective response was suppressed.  The video led to a decrease in the initial arousal, although the perceived valence of the stimulus was less affected.  One explanation for this increase in arousal while listening to nonlinear sounds is that the musical progression is unpredictable, and thus difficult to habituate to.  Increased emotional responses to nonlinearities also appear to have a biological basis, supported by a study conducted by Mende, Herzel, and Wermke (1990), which demonstrated that humans respond more to infants if their cries possess nonlinear traits than if their cries were lacking these characteristics.

When studying the effects of video games on emotions, Geslin, Jégou, and Beaudoin (2016) found that images that are less saturated, less bright, and contain fewer colors induce a greater sense of fear than their converse.  Additionally, Nanda, Zhu, and Jansen (2012) found that videos that were novel, contained objects with pointed features or sharp-angled contour, or showed images of people expressing emotions of fear induced higher levels of arousal and fear than those that were familiar, contained curved objects, and showed people expressing neutral emotions.  The amygdala was shown as the primary brain structure involved in processing fear, anxiety, and emotion with other contributors that include the insula, thalamus, and hippocampus (see Figure 1).

It is apparent that both music and imagery contribute to emotion and arousal, however, which has the greatest effect on eliciting a fear-related response?  The current study seeks to investigate whether music or video clips of scary movies induces the greatest levels of fear indicated by changes in arousal.  It is hypothesized that music, rather than video, contributes the most to the fear response.  As suggested by Blumstein et al. (2012), musical nonlinearities characteristic of horror films are associated with sounds produced by humans and non-humans in times of distress and danger, leading to a biological fear response.  Additionally, music alludes to a lack of predictability that is difficult to perceive with a video alone.

The current study consisted of sixteen undergraduate females students.  In order to detect changes in arousal as an indicator of increased fear, galvanic skin response (GSR) and heart rate were measured.  GSR can be calculated by gathering information about the electrical resistance and electrical voltage between electrodes placed on the left pointer and middle fingers (see Figure 2).  Heart rate was measured using electrodes placed above both inner ankles and on the right inner wrist below the thumb.  Before each stimulus presentation, a 45 second baseline was recorded as a means of comparison.  The participants were instructed to sit still as they watched or listened through headphones to a 45 second clip of a scary movie they had never seen before.  Three movies (Burnt Offerings, The Amityville

Horror, and The Descent) were included in the study to create eight total stimuli (four video and four audio clips) that were shown in alternating order.  Four repeated-measures ANOVAs were performed to analyze the data.  A p-value of .05 was used to determine significance.


A 2 x 4 (Stimulus [Audio, Video] x Trials [1-4]) repeated-measures ANOVA was performed to analyze the GSR percent change from the baseline for the peak-to-peak value, or the difference between the maximum and minimum values of arousal.  There was a significant main effect of stimulus, F(1, 15) = 8.488, p = .011 (see Figure 3).  This indicates that the audio stimulus (M = 2.648, SE = 0.321) had a greater peak-to-peak value, and therefore a greater change in arousal, than the video stimulus (M = 1.688, SE = 0.247).  There was also a significant interaction between stimulus and trial, F(3, 45) = 4.457, p = .008.  There was no significant main effect of trial F(3, 45) = 2.335, p = .087.

A 2 x 4 (Stimulus [Audio, Video] x Trials [1-4]) repeated-measures ANOVA was performed to analyze the GSR percent change from the baseline for the peak area.  There were no significant main effects or significant interactions, all Fs < 3.866, all ps > .062.  A 2 x 4 (Stimulus [Audio, Video] x Trials [1-4]) repeated-measures ANOVA was performed to analyze the percent change from the baseline for average heart rate.  There were no significant main effects or significant interactions, all Fs < 2.446, all ps > .076.  A 2 x 4 (Stimulus [Audio, Video] x Trials [1-4]) repeated-measures ANOVA was performed to analyze the percent change from the baseline for the maximum heart rate.  There were no significant main effects or significant interactions, all Fs < 2.211, all ps > .100.


The results of the study did not suggest any differences in arousal between the auditory and video stimuli.  Although there was a significant difference for the peak-to-peak value, such that the auditory stimulus evoked greater arousal than the video stimulus, there was also a significant interaction between stimulus and trial.  Further evaluation of the results indicated that there was a large change from baseline for the first trial for the auditory stimulus with little differences between the two stimuli for subsequent trials.  One explanation for this could be that counterbalancing was not implemented in the experiment, so the participants were all exposed to the same auditory clip at the start of the trial.  Therefore, it is no surprise that the first stimulus encountered after 45 seconds of silence would elicit an increase in arousal, especially because the stimulus was relatively loud in the participants’ ears.  It was also noted that participants returned to baseline more quickly after each stimulus presentation with decreased arousal for each trial as the experiment progressed.  This trend suggests that the participants habituated to the stimuli with each presentation. 

Future experiments should focus on increasing the arousing nature of the video and audio clips used as well as incorporating counterbalancing to ensure that there is no order effect.  Another improvement would to have participants rate their level of fear after each stimulus presentation.  This was attempted in the current experiment, but many participants forgot to complete their ratings midway through the experiment.  This methodology could be refined with a cue to complete the fear evaluation after each stimulus.  Information on the participants’ perceptions of the stimuli would yield more conclusive results as to the nature of the physical arousal indicated by changes in GSR and heart rate.  Additional experiments could also analyze the changes in heart rate and GSR for paired audio and video stimuli.  This could give insight as to whether there is a synergistic effect in regard to arousal and fear when scary videos and music are paired together.


Blumstein, D. T., Davitian, R., & Kaye, P. D.  (2010).  Do film soundtracks contain nonlinear analogues to influence emotion?  Biology Letters, 6(6), 751-754.

Blumstein, D. T., Bryant, G. A., & Kaye, P.  (2012).  The sound of arousal in music is context-dependent.  Biology Letters, 8, 744-747.

Geslin, E., Jégou, L, & Beaudoin, D.  (2016).  How color properties can be used to elicit emotions in video games.  The International Journal of Computer Games Technology, 2016, 1-9.

Mende, W., Herzel, H., & Wermke, K. (1990).  Bifurcations and chaos in newborn infant cries.  Physics Letters A, 145, 418-424.

Nanda, U. Zhu, X., & Jansen, B. H. (2012).  Image and emotion: From outcomes to brain behavior.  Herd, 5(4), 40-59.