Alarm tones, music and their elements: A mixed methods analysis of reported waking sounds for the prevention of sleep inertia

Sleep inertia is a potentially dangerous reduction in human alertness and occurs 0 – 4 hours after waking. The type of sound people set as their alarm for waking has been shown to reduce the effects of sleep inertia, however, the elemental musical factors that underpin these waking sounds and their relationship remain unclear. The goal of this research is to understand how a particular sound or music chosen to assist waking may counteract sleep inertia, and more specifically, what elements of these sounds may contribute to its reduction using a mix methods analysis. Through an anonymous, self-report online questionnaire, fifty participants (N = 50) reported attributes of their preferred waking sound, their feeling towards the waking sound, and symptoms of sleep inertia after waking. This data enabled the analysis and comparison between these responses to define statistically significant interactions. Our results show that there is no significant relationship between sleep inertia and the reported waking sound, nor the subject’s feeling towards this sound. However, we found that the melodicity of a chosen waking sound does effect sleep inertia. A sound that is perceived as melodic, produces less sleep inertia in comparison to a sound considered to be neutral (neither unmelodic nor melodic). Furthermore, a secondary analysis reveals that this is an important factor for waking stimulus design as it suggests that the amount of perceived rhythm will affect the perception of melody, and in turn, may influence the severity of sleep inertia on a secondary level. Our results reveal that the inclusion of detailed descriptive terms (musical elements) in addition to macro classifications (e.g. “pop music”) for stimulus testing would benefit future research and our understanding of waking audio’s effects on sleep inertia.


Introduction
"The morning started disastrously. I slept through two alarms, one set for 0600 and another a half-hour later to remind me to take some CEO pictures. My body apparently went on strike for better working conditions." NASA astronaut journal report during orbit aboard the International Space Station. [1] Sleep inertia (SI) is a sleep-wake disorder characterized by low arousal and reduced cognition [2]. Initiated upon waking, SI's symptoms can last for seconds, minutes or hours, where extended SI may impact human performance in a variety of fields and occupations [3][4][5][6][7][8][9]. This has been highlighted as a likely factor in the 2010 Air India Express air crash disaster that resulted in 158 fatalities. It has been shown that the captain of the aircraft had recently woken from an in-flight nap just prior to the crash. The poor decisions made after napping were attributed to the disaster, and have been linked to the effects of SI [5].
There has been growing research interest into the mechanics and architecture of SI [10][11][12][13][14][15], however the modalities and means to activate waking remains in its early stages with respect to this issue. YouGov [16] report that out of 586 participants surveyed, 68.2% use a form of alarm for waking, of these, 23% use an alarm clock, 14.9% a clock radio, and 26.3% an alarm on a cell phone. These figures show alarms are still an important means to wake up, and given the 24-hour society in which we often live and work, the need for peak performance from our waking device, and the stimuli they produce, is advantageous to counteract the negative effects of SI.
Research in the field of human factors and psychology has provided initial insights into the application of countermeasures to reduce SI. Countermeasures are strategies to be implemented upon waking as opposed to methods that may consider circadian management, or pre-sleep hygiene techniques (routines to assist and promote sleep). Within the current literature investigating experimental SI countermeasures include light, temperature and sound [11,[17][18][19]. In the context of our research, two previous studies present findings on the effects of sound as a countermeasure for SI. Tassi et al [11] concluded that noise can reduce SI when deployed as an intense waking alarm [11], while Hayashi et al [19] discovered that 'high-preference popular music' as chosen by participants has the potential to reduce the intensity of SI after a short nap. Both studies support the use of sound and music as a countermeasure for SI, although many questions remain as to the auditory mechanics and aesthetics required for best practice design of such stimuli, hence, a consistent approach to minimizing SI through audio is yet to be determined.
Musical elements are components of music or sound that exist to assist in the description and production of music. These elements can be treated individually or combined to produce an infinite array of musical aesthetics, effects and compositions. Musical elements include, yet are not limited to, melody, rhythm, pitch, tempo and volume. These elements differ from sound types and genres as they afford the description of audio through musical characteristics and have the capacity to expose factors which may be neglected in sound type classification [20,21].
For one example, in the context of this research field, a subject may describe auditory stimuli as 'popular music', however, when reported as musical elements, the audio can be defined as 'very melodic', 'rhythmic', 'high pitch', 'fast tempo', and 'high volume'. From these descriptions, 'pop music' may now be analyzed and understood through musical terms to establish in greater detail what the respondent may actually be hearing, regardless of the genre or stimulus title. This mixed method research explores waking audio and its effects on SI through the deployment of a self-reporting online questionnaire designed to answer three primary research questions, (i) 'Do waking sound types counteract the effects of SI?', (ii) 'Do subjective feelings towards waking sound types counteract the effects of SI?', and (iii) 'Do the musical elements of waking sound counteract the effects of SI?'.
To achieve our objectives, several separate analyses where performed comparing the respondents' reported intensity of SI against their waking sound types, subjective feelings, and the musical elements of their waking audio. Additionally, further analysis investigates secondary level attributes of waking types and musical elements with respect to SI. Lastly, response time evaluation and alertness factors are evaluated, together with qualitative observations. This was stipulated to the subjects in the 'Invitation to participate' email distributed during the recruitment period. Potential participants were encouraged to undertake the study without bias towards music aptitude or waking method. All respondents (See Results) were 18 years of age and above consisting of males and females. Consistent with ethics, age and gender are for demographics only and are not subject to analysis in this study.

Data collection
Data was collected via a self-report online questionnaire. The submitted data was captured digitally via the use of the online software system Qualtrics [22], where the questionnaire was contained and operated. Qualtrics is software specifically created for the undertaking of online questionnaires and surveys enabling researchers to design and implement their studies for ethically compliant distribution and data collection. The data obtained by Qualtrics is securely stored and only available for download and analysis by researchers with appropriate clearance.
Questionnaire -Development and Description Waking Sound (Section 3), and Sleep Inertia (Section 4), in total, requiring approximately ten minutes to complete. The ten-minute data collection was designed based on pilot results to minimize disruption to each subject during their natural 'day-to-day' waking routine, so as to maximise the ecological validity of the result.
Section 1 is an introduction to the study and clarifies the participants' obligations prior to completing the task. This includes a reminder that by completing and submitting the survey the participant has given their specific consent to partake in the study (S 1. Fig 1). Section 2 (Items 1-5, S 2. Fig 2) is comprised of Likert scale (unipolar Item 5, bipolar Item 4) and multiple-choice (Items 1-3) questions, which gather the demographic data of the respondents, their music appreciation, and musical aptitude. This data is paired with responses from Section 3 (Items 7-9, 12, Fig 1), and reported in the results under the following categories: Respondents (Table 7), Music Appreciation & Aptitude (Table 8), and Alarm Adoption & Application (Table 9).  No -Forward to Item 6.1.  Table 9) previously outlined in Section 1's description.
Item 10, 'Which most frequently used audio for waking up best represents yours?' requires the participants to nominate or specify the type of audio they use from six options provided.
This item is the essential component to determine each participants' waking sound type, and is elemental in formulating the analysis in response to primary research question (i) 'Do waking sound types counteract the effects of SI?'.
In designing the options for item 10, we first defined 'sound type' [23,24] as the most suitable overarching terminology in this context to define all audio that may be reported as waking sound stimuli (e.g. musical genre, auditory tone, white or pink noise, human speaking, the sound of the wind, or aircraft engine noise). Secondly, the first five options for selection (Alarm tone, Musical song, Instrumental music, Natural sounds) were elected as they are familiar descriptions of audio sound types which respondents may use for waking. To determine the categories, we surveyed available pre-set sound types and custom audio functionality provided by several device manufacturers [25][26][27][28][29][30][31]. The sixth option (Other) allows for the respondent to describe their specific sound type if desired. In sum, these options allow for the breadth of potential waking sound types respondents may report.
Operationally, if 'Musical song' or 'Instrumental music' are selected, the respondent is then forwarded to Item 10.1 and are requested to specify which genre represents their waking sound type. The categories of genre for selection have been adapted from the Short Test of Music Preference (STOMP) [32]. Similarly, when 'Radio' is specified the respondent is prompted for the specific station. If 'Other' is selected, the participant is requested for a description.
Item 11 is another key factor of the study and is employed to analyse primary research question (iii) 'Do the musical elements of waking audio counteract the effects of SI?'.
This item gathers each respondent's classification of their waking audio's musical elements which is used to establish a profile of the stimuli from a fundamental musical level. Each participant is required to rank the musical elements of their waking audio on a 5-point bipolar Likert scale that we developed. These specific ranks have been selected to afford descriptions of the participant's waking audio musical elements through subjective interpretations (e.g. negative, neutral, positive). Ranks 1, 3, and 5 are labelled (e.g. 1 = unmelodic, 3 = neither unmelodic nor melodic, 5 = very melodic), while rank 2 and 4 remain uncategorized to reduce respondent bias. When reporting and discussing the results, we included labels to ranks 2 and 4 for continuity which we have defined in the rank response codes shown below (Table 1).  [33], the SIQ has been researched and analysed to be a reliable measure of SI [33]. Our adapted SIQ begins with the question; 'After you wake up, to what extent do you…' [33] and is followed by each item.
All respondents are required to rate each item as either, Not at all = 1, A little = 2, Somewhat = 3, Often = 4, All the time = 5. In the context of this study, eight items were removed from the original SIQ questionnaire as they are already included in other items in the questionnaire, or are not a focus of this study. See Table 2.

Data processing
The total number of initial respondents was 83 (N = 83) which was filtered omitting any inconsistent or incomplete responses, reducing N to (N = 73). Further, this study requires the analysis of waking sound stimuli, therefore the 'No waking sound' responses were disregarded, reducing N to (N = 50). The SI intensity is determined as the mode of each participant's response within the adapted SIQ questionnaire items in Section 4 (Fig 2). The raw data was initially filtered through Microsoft Excel [34], then imported to SPSS [35] for statistical analysis.

Statistical analysis Primary analysis
To  Table 3 contains the sequence of analyses, and each test we performed for the primary research questions.  Each test analysis was firstly trialled with a non-parametric contingency

Secondary analysis
For the significant (p < 0.05) results gathered from the primary analysis, we performed a secondary series of analyses. On this condition, we tested the appropriate items (10,11,13) against each significant result obtained. By conducting this sequence of analyses, we can respond to the main research questions by defining secondary level interactions between the primary results and each conditionally relevant item (I.e. sound type Item 10, subjective feeling Item 13, and music elements Item 11). These results provide data to be applied in the formulation and design of waking audio for SI in future studies. Each secondary analysis performed can be viewed in Table 4.  to determine significance. These tests analyse whether a report for how long it takes to completely wake up (Item 15), the period after waking to complete the questionnaire (Item 16), and time of day completing the test (Item 17) influence the intensity of SI. See Table 6.

6.
7.  Table 10. There was a significant interaction between the reported melody of the waking sound, and the reported waking sound rhythm X 2 (16, N = 50) = 50.32, p < 0.001. Subsequent analysis (Fisher's Exact Test) produced X 2 (N = 50) = 37.38, p < 0.001. The Post-hoc, Adjusted Residual testing revealed several significant interactions between the reported melody of the waking sound and rhythm of the waking sound. These results are shown in Table 11. No significant statistical interaction between these three cases and the intensity of SI were reported. See Table 12.  Existing research supports audio's potential to counteract the intensity of SI [11,19]. In this research context, 'excitative popular music' has been shown to negate SI [19] which provides the first insight into a sound types effect on SI, though many questions concerning the specific To rationalize this phenomenon is challenging considering both the lack of specific research in this context, and the absence of descriptive detail (including music elements) concerning auditory stimuli for waking. Research shows that sound can increase and maintain arousal, and attract human attention [5][6][7][8][9][10][11], however, music elements (specifically melody) in the context of counteracting SI is unknown. We hypothesize that stimuli perceived as melodically neutral may be interpreted as an auditory ambient variation of their counterpart (melodic).
When compared to strong melodic material this neutral classification is less likely to gain the human center of attention, may induce less arousal, and lead to reduced cognition, all of which, are symptoms of SI [2]. For situations where a stimuli's melodic content is increased, its auditory ambience transitions to salience, increasing arousal, cognition and attention, subsequently reducing the effects of SI. For this analysis and building on available research, we propose that the melodic content of waking audio may be an essential musical factor in counteracting SI, supplementing holistic waking sound types as previously understood.
Further, this result supports the requirement for detailed descriptions and inquiry of auditory test stimuli, to inform analysis and discussion in this research field.
Given the finding that melodic content of waking audio can influence SI, we now discuss this musical element and its statistical significance to waking sound types (e.g. alarm tone, natural sounds). Granting that SI is not directly affected in these conditions, we define which categories of waking sound types may have the ability to influence SI on a secondary level.
Our secondary analysis results show that an 'unmelodic' waking sound has a significant interaction to the category 'alarm tone', a 'very melodic' report is significant to the category 'musical song', and, 'melodic' is significant as a musical element ranking with respect to 'natural sounds' (See results, Test analysis No 8. Music element -Melody (Item 11) vs Sound type (Item 10)). From these results, we hypothesize that the musical mechanics and aesthetics of each sound type attributes to its perceived melodicity ranking. For example, as a function of the traditional sounding alarm (to wake sleeping humans through auditory intervention), by design this category of stimuli typically consists of a static rhythm, an insistent tonal center, and a salient aesthetic (e.g. a relentless beeping sound) [7,8]. 'Musical song' incorporates melody as a dominant feature which is evident in popular western music, examples include, yet are not limited to, The Beach Boys 'Good Vibrations' [36], and The Cures 'Close to me' [37]. The relationship between 'melodic' and 'natural sounds' is more difficult to define. One hypothesis we present is bird song at dawn (a ubiquitous ecological waking sound common in many cultures [7]) for their melodic qualities associated to a variety of species [38], and by virtue that birds typically exist in natural environments. Similarly to the investigation of music elements and sound types in this study, bird song is categorized through song 'type' and the 'elements' which articulate the song type. [39].
The resulting interactions between melodicity and waking sound types may prove beneficial for the initial formulation of affective waking stimuli. We have shown that the melodicity of waking stimuli can counteract SI, and as melody is a musical element of significance with respect to particular waking sound types, we suggest that these factors must be carefully considered when designing optimal sounds to counteract SI. For instance, and referring to the results we have obtained, future investigation may study the 'melodic' element of 'natural sound' types for counteracting SI.
The quantity of rhythmic content in waking sound stimulus has a consistent and significant interaction to the perception of the reported melodicity (positive or negative) of waking audio. Our results show that when the rhythmic content of the waking stimuli is decreased, the reported melodicity is reduced (Table 11). In contrast, as the rhythmic content of the waking sound is increased, the melodicity of the stimuli is perceived in this manner also (Table   11). Additionally, the data reports an anomaly between the interpretation of 'very rhythmic' and melodicity. Specifically, if the rhythmic content of the waking sound is 'very rhythmic', then the perceived melodicity of the stimuli is reported as 'unmelodic' and 'very melodic' with equal probability (See Table 11). We hypothesize that this result is a by-product of the composition's increased rhythmic elements, whereby a perceptual threshold is introduced, rendering the melodic and rhythmic interactions of the composition musically too complex for humans to clearly define as 'very melodic' or 'unmelodic'. These results align with Boltz's [40] findings suggesting an intrinsic relationship between melodic and rhythmic perception whereby temporal microstructures of compositions hold a primary role in the cognitive processing of melodic relations.
With respect to time duration evaluation of alertness, time period after waking to respond, and time of day with which the questionnaire was completed, there is no statistical interaction between these classes and SI (See Results, Test analysis No. 13,14,15). Our results indicate that SI intensity is independent of these factors highlighting the potential for waking stimuli to be as affective regardless of SI duration and time of day. Positive implications in this context include waking stimulus for shift workers, emergency response personal, and drowsy driving. However, further research is required in context to reinforce these understandings.
For example, a Psychomotor Vigilance Test [41] could be utilized in conjunction with the SIQ in response to test stimuli to substantiate the results obtained and allow comparisons.
Primarily a technical limitation in the context of online testing, this obstacle is being minimized with the increasing availability of accessible software to accomplish this task.
Gorilla [23] is one online software system enabling the design and production of online questionnaires with embedded interactive capabilities and data logging.
The demographic data collected from the respondents in this study produced two key findings for referencing in waking sound development. Firstly, a mobile phone is the most frequently reported device for communicating the participant's waking sound (84%, Table 9). Coupling this with estimates reporting that the audible frequency range of these devices ranges between 900 Hz -10,000 Hz [42], this is central when developing auditory stimuli, as the human auditory hearing range spans 20Hz -20,000 Hz [18]. Secondly, when considering the design of waking stimuli for SI and the aesthetic treatment of volume, sixty-six percent (66%) of subjects' report that they employ a constant volume for waking. Specific research in this domain would clarify the benefits of auditory design targeting the most appropriate aesthetic volume treatment for waking sounds in the context of SI.
The primary analysis showed a significant statistical interaction between melodicity and SI. By dissecting the musical elements of the waking sounds melodic content in the secondary analysis, we conclude that the rhythmic attributes of the audio have a significant interaction to the perceived melodicity of the stimulus, which may impact SI through elemental musical interactions. We define these as key components to examine in the understanding of music and how its elements in combination can be assembled to produce the most effective waking sound stimuli to counteract SI. Future research based on these results may investigate rhythmic interactions with the perception of 'melodic' musical elements to inform the design and development of 'natural sound' types for SI.