Abstract
Studies on the effect of music on task performance are contradictory about this relationship’s direction and valence. Task characteristics may be accounting for these inconclusive findings. Thus, this study employs effort to mediate music’s effect on task performance (objective and perceived) under a stressful decision-making process. This is a between-group experiment with three conditions: slow-tempo music, fast-tempo music, or no music. We designed a computer web interface, where participants did a stressful task. Results demonstrated that participants made a strong effort under the conditions with music. Hence, turning the music off under stressful activities is favorable in terms of performance. The article contributes to understanding the interaction between music and task performance, expanding the discussion within a stressful task.
Music exerts behavioral, cognitive, and emotional effects on humans (Kämpfe, Sedlmeier, & Renkewitz, 2011). However, the valence of the effect of music on task performance is inconclusive (Schellenberg & Weiss, 2013). Many studies illustrated that music improves performance (Beanland, Allen, & Pammer, 2011; Rauscher, Shaw, Levine, Ky, & Wright, 1994), whereas others found the contrary (Furnham & Bradley, 1997; North & Hargreaves, 1999). Others reported that music does not influence performance, or that both detrimental and improvement effects are present (Kämpfe et al., 2011; Schellenberg & Weiss, 2013).
These inconsistent results are tied to the task’s characteristics and required cognitive effort, as individuals have limited cognitive resources for performing a task and processing music (Kämpfe et al., 2011; Schellenberg & Weiss, 2013). Therefore, this study limits the task to investigate the effect of music on performance while acknowledging the role of effort. The study evaluated the influence of music during a stressful decision-making process and hypothesize that cognitive effort mediates the effect of music on performance.
The article provides theoretical and practical contributions to the literature. Theoretically, the discussion of the effects of music on task performance is expanded. The study consists of a highly stressful task conducted in a virtual environment under three conditions: slower tempo music, faster tempo music, and no music. Moreover, the study tested for the mediating effect of effort in the relationship between music and performance (using perceptual and objective measures of performance) to provide evidence of how music responses produce variations in listeners’ cognitive ability.
Practically, results have implications in designing spaces where music can or cannot help in decision-making processes. Understanding the interaction between music and task performance through participants’ effort facilitates better utilization of music.
For example, the COVID-19 pandemic has instigated virtually conducted activities, such as work, study, and shopping (Boumphrey, 2020); a trend that may remain. Hence, experiences should be adapted to such environments. Music may be present at home during demanding tasks. Against this background, investigating the effect of music on performance may be relevant to design virtual experiences.
Theoretical background
Ambiguous effect of music on task performance
Tempo is the speed at which the musical sequence progresses and the beat flows. It is measured in beats per minute (bpm), being the pulsations that occur in a minute: More (less) pulsations in a minute means a faster (slower) tempo, hence a beat that flows faster (slower). Tempo has captured the attention of researchers who evaluated the effect of music on task performance because it is a fundamental descriptor of music (Kämpfe et al., 2011; Schellenberg & Weiss, 2013). However, conclusions about how well tasks are performed when exposed to a given tempo remain unclear.
For example, cognitive abilities can be improved by listening to slow-tempo music because it reduces anxiety. In contrast, fast-tempo music can be distracting (Thompson, Schellenberg, & Husain, 2001). Besides, reading comprehension may decrease with fast-tempo music. However, slow-tempo music does not interfere (Thompson, Schellenberg, & Letnic, 2012).
Moreover, two parallel studies demonstrated conflicting results. First, Mayfield and Moss (1989) asked participants to record closing stock prices in a series of months and calculate percentual changes for each week in the given period and observed that music (i.e., fast tempo, slow tempo, and no music conditions) did not influence the quality of the performed task. However, when they replicated the experiment as a follow-up study, the participants displayed better performance under the fast-tempo condition despite being more distracted.
Besides, music may not influence performance. For instance, no significant effect was found in intelligence quotient (IQ) performance when participants listened to music while doing an IQ spatial test (Stough, Kerkin, Bates, & Mangan, 1994). Moreover, studies that investigated the effect of music on the performance of computing math problems show a lack of effect (Wolfe, 1983). Kämpfe et al. (2011), in a meta-analysis, suggested that both positive and negative effects of music on task performance may be present simultaneously. Thus, results on the effect of music on task performance are inconclusive (Schellenberg & Weiss, 2013).
Psychologists suggest that task characteristics (i.e., highly cognitively demanding or otherwise) should be explored to enhance the understanding of sensory stimuli’s influence on task performance (Chen, 2003). Hence, researchers endeavored to use task complexity to explain the contradictory results, arguing that a complex task requires high-cognitive effort (Cockerton, Moore, & Norman, 1997; Kahneman, 1973; Kämpfe et al., 2011; Lavie, 2005; North & Hargreaves, 1999).
Role of cognitive effort on task performance
Cognitive effort is the degree of limited cognitive capacity used when performing an information-processing task (Tyler, Hertel, McCallum, & Ellis, 1979). When a task requires a high level of cognitive effort (Lavie, 1995), task performance is expected to be reduced (Kahneman, 1973; Plebanek & Sloutsky, 2019). Along with the various findings on the relationship between music and task performance (Kämpfe et al., 2011), two perspectives explain the role of cognitive effort in this relationship, namely, the arousal hypothesis and cognitive load hypothesis (Thompson et al., 2012).
The arousal hypothesis states that listening to music while performing tasks increases performance by positively influencing mood (a calmed and relaxed state of mind; Schellenberg, 2005; Thompson et al., 2001). Cockerton et al. (1997) found increments in IQ test scores for participants that listened to relaxing music (vs. no music). Employees expressed feelings of happiness with their work and improved performance and mood while listening to their favorite music (Oldham, Cummings, Mischel, Schmidtke, & Zhou, 1995). Moreover, listening to music during tasks can reduce inattentional blindness as listeners are in a positive concentration mode (Beanland et al., 2011).
Research points to a psychological effect called the Mozart effect, in which tasks are performed better immediately after or while listening to Mozart’s compositions (Rauscher et al., 1994; Schellenberg & Weiss, 2013). Moreover, the effect of music on emotional states is observed for other stimuli, such as pop music (Schellenberg, Nakata, Hunter, & Tamoto, 2007; Schellenberg & Weiss, 2013).
The capacity model can explain the negative effects of music on task performance. In other words, individuals have a limited pool of cognitive resources for use during task performance (Baddeley, 2003; Kahneman, 1973). According to this perspective, cognitive and perceptual capacities are used to process stimuli and tasks (indifferently) until consuming the total capacity, such that cognitive capacity that is not used may process distractors that decrease task performance (Lavie, 1995). Thus, people should use full cognitive resources in demanding tasks to refrain from succumbing to distractors (Lavie, 2005), and effort should account for understanding how music influences task performance. In this manner, distractors can be controlled (Forster & Lavie, 2009).
However, scholars have suggested that auditory stimuli reduce task performance even when such a task requires high cognitive effort as they could be perceived separately from other tasks (Tellinghuisen & Nowak, 2003). Therefore, music competes for limited processing resources (Kahneman, 1973; Konecni & Sargent-Pollock, 1976), creating a cognitive interference known as “cognitive overload” (Norman & Bobrow, 1975). This overload leads to failure to process part of the information, which reduces task performance (Norman & Bobrow, 1975).
The cognitive demand of a task interacts with the arousal that music generates, which has a U-shaped effect on task performance (Konecni & Sargent-Pollock, 1976). Initially, the effect of music is consistent with the arousal hypothesis. However, as music produces high levels of arousal, it reduces the attention available for other tasks.
If a task requires additional cognitive effort, task performance is expected to be reduced (Kahneman, 1973). The effort is probably accompanied by feelings of frustration (insecurity, irritation), a sensation of an increase in mental demand (the amount of perceptual activity required to do some task), and temporal demand (the feeling of anxiety developed to conduct tasks in a determined time; Hart & Staveland, 1988).
Music may distract people when doing tasks as their brains find it difficult to conduct perceptual or attentional selection for musical stimuli; thus, their performance decreases (Oliver, Levy, & Baldwin, 2020). This effect is expected to increase when the task requires high-cognitive effort (Tellinghuisen & Nowak, 2003). Under a high-cognitive effort condition, even relaxing music may become stressful, annoying, and distracting, having a detrimental effect on cognitive performance (Iwanaga & Ito, 2002) or it can be as unpleasant and distracting as noise (Furnham & Strbac, 2002; Vasilev, Kirkby, & Angele, 2018). Hence, turning off music during a complex work is preferable to control the cognitive effort required to process stimuli and tasks (Groeneweg, Stan, Celser, Macbeth, & Vrbancic, 1988; Schwartz, Ayres, & Douglas, 2017).
To address the disparity of the effects of music on task performance, this study evaluates the effect of music on task performance under three conditions: faster tempo music, slower tempo music, and no music. In addition, this study suggests that effort mediates the relationship between music and task performance (Figure 1). Participants undertook a stressful task that frames the required cognitive effort. The task consisted of deciding a vacation plan in a web interface designed to be erratic, anticipating a higher effort to make decisions, and stress evidenced in the mouse behavior (Freihaut & Göritz, 2021; Grimes, Jenkins, & Valacich, 2013). When effort cannot be controlled in the presence of music, more errors should be expected, and performance gets reduced (Plebanek & Sloutsky, 2019; Tellinghuisen & Nowak, 2003). The anxiety produced by the lack of control of effort when doing tasks in online settings may be expressed in errors when doing the task, the number of scrolls, and the scrolls’ distance (Grimes et al., 2013). Hence, we expect that music increases the perceived effort when conducting the online task, which will decrease participants’ performance (number of errors, number of scrolls, and scrolled distance), such that the perceived effort mediates participants’ performance.

Mediating Effect of Effort.
Method
The experiment was conducted using a between-group design. Music (the independent variable) has three conditions: faster tempo, slower tempo, and no music. We tested the effects of music on the perceived effort, which mediates the effect on satisfaction with performance and task performance (Figure 1).
Participants
Students from a Colombian university (N = 105, female = 57%, mean age = 20.7 years, SD = 3.44) participated voluntarily in exchange for course credit. None of them were under the Psychology, Music, Computing, or Hospitality department. 64%, 30%, and 6% reported having no, between 1 and 5 years, and more than 5 years of musical training, respectively. Moreover, all of them reported being right-handed. They were randomly assigned to one condition of the independent variable (35 participants per condition).
Materials and stimuli
A virtual environment was created to design a vacation plan for a client with specific needs (Figure 2; for a sample of the interface, refer to the following link: https://youtu.be/X5wEk91u4Lc). The vacation plan was designed based on three client profiles: a 20-year-old party youngster, a family with an infant and a baby, and a shy student looking for a calm ambiance. Client profiles were randomly assigned to each participant. The task’s objective was to involve participants in a decision-making activity by selecting the destination, hotel, and transportation congruent with the client profile while managing the client’s budget.

Screenshot of the Application.
The task was stressful as the web page on which participants created a vacation plan was purposely designed to be erratic. Website elements were positioned so that a user was required to scroll down and up several times to find and select an option. After any decision (e.g., a hotel), the interface’s response was lagged and appeared as if the response was being processed. When the selected options were inappropriate for the client profile, the system flashed an error message, and the participant had to redo the vacation plan from the beginning. Hence, the activity was considered stressful as patience was required in selecting items due to the software’s slow response, repetition of the process, and bad design of the web page, such that attention was needed to achieve the task on time. Interaction with the interface was explicitly designed as stressful by emphasizing poor legibility and interacting with options in a challenging manner.
All participants used the same computer hardware (a 22-inch screen, a keyboard, a mouse, and headphones [Logitech H390] in a stereo system with a fixed volume).
Within the stressful task, the participants were randomly assigned to one of the abovementioned experimental conditions. The songs used were instrumental as lyrics could influence listeners’ perceptions (Dahl, 2011). Two music experts performed a subjective study to estimate the classification of songs within each category, where the tempo (bpm) played the most important role; however, other features, such as mode, articulation of keys, and time signature, were also considered. After the experts’ classification, the songs’ tempo was confirmed using Metrónomo Beats, an application that measures tempo. The slower tempo condition mostly had songs in major keys, legato articulation, a mean tempo of 48 bpm, and a time signature of 4/4. Other songs under this condition had time signatures of 6/8 and 3/4. The faster tempo music condition was mostly in minor keys and staccato with a mean of 119 bpm; all songs had a time signature of 4/4. Conditions were named “faster-tempo” and “slower-tempo,” avoiding differences in perspectives among what a fast or slow tempo is but recognizing that songs in one condition are faster (slower) than songs in the other condition.
Songs within the slower tempo condition were Summy by Chamin Correa, The Long and Winding Road by Chris Cozensy, Once Upon a Time in The West by Francis Goya, You Needed Me by Gheorghe Zamfir, Open Arms by Javier Ricardo, Jeux Interdits by Nicolas de Angelis, Tears In Heaven by Sergi Vicente, Memory by The Bruno Bertone Orchestra, and Love Story by the Tokyo Kosei Wind Orchestra. For the faster tempo condition, the songs were Fuiste Mala by Adolfo and Gustavo Angel, Polka Caribeña by Rey Casas, Bamboleo by James Last, Andalucia by Paco Nula, Lluvia de Primavera by Raul Di Blasio, Soca Dance by Ray Hamilton and Orchestra, Buenos Recuerdos by Sergi Vicente, and Gypsy Earrings by Strunz and Farah.
We verified if the participants’ perception of the music tempo corresponded with the real tempo by the following manipulation-check: “You were listening to music while designing the vacation plan. Was the music slow or fast (Milliman, 1982)? Did you hear music playing?” Accordingly, the participants under the no music condition responded that they did not hear music playing, whereas 97% and 87% of those under the faster and slower tempo conditions, respectively, were consistent with corresponding treatment exposure.
The participants’ devices’ volume was adjusted to be identical in both tempo conditions for all participants. All songs from the slower tempo condition were matched to last 243 s, while the songs in the faster tempo condition lasted 216 s. The songs repeated on loop until the participant completed the task or after 15 min (900 s, the limit time for the task); hence, the number of repetitions of the song assigned to each participant was dependent on the length of time the participants took to complete the vacation plan.
Procedure
The participants stayed inside a laboratory in front of a computer screen using headphones, and no other distractor (i.e., cell phones) was allowed. At the computer lab, the participants were instructed to plan a vacation trip for a given client. The client profiles were randomly assigned an equal number of times. Participants began the task by reading the following instructions on the screen: The Best Vacation Plan: On the following link, you will plan a vacation trip for the following client [client profile]. You know how much money the person is willing to spend on the trip. Accordingly, you will be able to make decisions regarding the destination, transportation, room, and food budget. You may spend all available money or save the client a part of this money. What matters is that your decisions take your client to the Best Vacation Plan. You have 15 minutes!
Hence, participants had to design vacation plans congruent with the clients while managing the budget. The entire experiment was conducted in Spanish.
After reading the instructions, a random song from the playlist of a corresponding condition was played and repeated until the participants finished their task. Within this time, the participant finished by clicking on the “submit decision” link, or else the interface closed the application when the time was over. The application then appeared to be making calculations, and the following message showed up on the screen for 30 s: “Wait as we analyze all decisions.” The entire procedure took participants on average 492.1 s (SD = 227.4 s); 12.4% of the participants were unable to finish.
Measures
As the participants were designing the vacation plan, the software captured these three measures that accounted for task performance: number of errors, number of scrolls, and scrolls’ distance (dependent variables). The number of errors was set as the frequency with which the participants submitted an unsuitable vacation plan for the profile assigned. The number of scrolls over the interface was calculated as the number of times the mouse was scrolled to different parts of the interface. More up and down scrolls indicate uncertainty and having difficulties in decision making. Similarly, the distance covered when scrolling (measured in pixels) with the mouse on the interface implies that the task was unclear.
Participants completed a survey after the decision-making task. The questionnaire was presented in Spanish (participants’ native language). The experienced level of effort was evaluated using the NASA-TLX, a subjective perceptual measure of the workload that a person experiences during tasks that require certain levels of effort (Hart & Staveland, 1988).
The scale rated how demanding the task was from the participants’ perspective using the following items: mental demand, temporal demand, effort, and frustration. The item for physical demand from the original scale was excluded because it did not apply to the task. A 10-point Likert-type scale ranging from low (1) to high (10) was employed. Cronbach’s α for the Task Load Index (TLX) scale based on our data were .77.
The participants’ satisfaction with their task performance was evaluated using two questions that appeared at the beginning and end of the questionnaire: “How well did you perform during the task?” and “How will you rate your overall satisfaction with your performance on the task?” A 7-point scale ranging from bad (1) to excellent (7) was employed. From our data, Cronbach’s α for the satisfaction scale was .76. Finally, the study controlled for music likability in the two music conditions (slower and faster tempo) using three items: likable, pleasant, and good (Macinnis & Park, 1991). The likability construct yields an average response of 3.9, and no significant differences were noted between the two tempo conditions for these variables, F(2, 68) = 0.89, p > .1.
Participants were not asked how stressed they were. It was expected that the stress state was evidenced in the behavior with the mouse as significant research shows that stressed participants tend to reflect their state in the use of the mouse in a computer, such that they do more scrolls and increase their scrolled distance (Freihaut & Göritz, 2021; Grimes et al., 2013).
Results
Preliminary analysis
One-way ANOVAs checked if there were differences between the three client profiles, and confirmed that the client profile did not influence participants’ experience of effort toward the task (TLX), F(2, 102) = .83, p = .44,
Besides, a one-way ANOVA showed that the manipulation (i.e., no music, slower, and faster tempo) affected the participants’ experience of effort toward the task (TLX), F(2, 102) = 6.14, p = .003,
Table 1 provides the descriptive statistics for each condition across dependent variables (variables 1–5) and correlations between dependent variables. TLX and satisfaction’s negative relationship indicated that the participants decreased their performance as the perceived effort increases.
Descriptive statistics and correlations.
SD: standard deviation; TLX: Task Load Index.
p < .001.
Mediating role of effort
The second step evaluated the mediating effect of TLX of the participants (i.e., perceived effort) using PROCESS for SPSS 26 (Hayes, 2013); this is a two-step process described in Table 2. For these models, the experimental condition with no music was at the intercept as a comparison point. The first step in the mediation procedure shows that both music conditions (slower and faster tempo) influenced the mediator, perceived effort (B to TLX).
Path coefficients of the mediation model.
B: Unstandardized beta coefficient (standard errors are enclosed in parentheses); S: satisfaction construct; TLX: Task Load Index.
p < .05; ***p < .001.
In the second step, the complete models include the effect of music and effort on the dependent variables (B to Errors, Scrolls, Distance, and Satisfaction) to show the mediation effect (Table 2). After having music (slower and faster tempo) and including TLX in the model, the music conditions lose their effect for the four mediated models. All effects of music on the dependent variables are explained through TLX, which becomes the only significant variable in the model, thus supporting the complete mediation of music through perceived effort on task performance or satisfaction with the task.
Moreover, the Sobel test statistics were significant (Table 3). In the relation between the music conditions and the dependent variables (errors, scrolls, and distance of scrolls), TLX functioned as a third variable that explained the relationship between the independent and dependent variables. When introducing TLX in the models, there was a significant reduction in music’s effect, either with slower or faster tempo (Table 3, Sobel’s Z). The beta coefficients in the indirect effect confirmed that the majority of the effect of the music conditions on the dependent variables is explained through the effect of perceived effort (TLX). Figure 3 presents the direct and indirect effects.
Indirect effects.
D: Scrolled distance; E: errors; FT: faster tempo; S: Satisfaction with results; Sc.: number of scrolls; ST: slower tempo; TLX: Task Load Index.
p < .05.

Mediating Role of TLX. No Music Condition is in the Intercept. Final Standardized Regression Betas and Adjusted R2 are Reported, Whereas p-values are Placed After Semicolons. Initial R2 Values are Enclosed in Parentheses for the Model Without Mediation. Brackets Indicate 95% Confidence Intervals.
Discussion
Results show that music (faster tempo or slower tempo) did not influence task performance directly compared to a condition with no music. However, when perceived effort mediated the relationship, the study found that music (regardless of tempo) decreased task performance. Low levels of performance were observed in terms of physical measures (i.e., errors of the participants when selecting profiles, number of scrolls, scrolled distance using the mouse) and perceptions (i.e., satisfaction with results). Under the stressful task, music does not worsen the results; its negative influence on performance is explained by perceived effort (Figure 1). This study supported the notion that as participants perform stressful tasks that require high cognitive effort, the presence of music increases their cognitive effort, decreasing task performance (Tellinghuisen & Nowak, 2003).
The main contribution of the study is elucidating the inconclusive findings on the effect of music on task performance by illustrating that the participants’ perceived effort may mediate the effect of music on performance. Thus, the study supports the notion that music may cause additional cognitive overload, simultaneous with other cognitively demanding tasks (Kahneman, 1973; North & Hargreaves, 1999).
Groeneweg et al. (1988) suggested that the effects of music on task performance should account for the task characteristics. In this experiment, the task was designed to be stressful. Hence, music increased the cognitive effort required for decision making, as it may become a distractor (Mayfield & Moss, 1989). Likewise, Iwanaga and Ito (2002) and Schwartz et al. (2017) stated that although music may be relaxing in the performance of repetitive tasks under stressful situations, music could be perceived as noise. Thus, turning off music is preferable because it increments perceived effort.
From a pragmatic perspective, the study infers that results are relevant for designing environments where different tasks are performed. Designers of environments where stressful tasks occur should aim to reduce the effort produced when doing tasks, but if the effort cannot be controlled, they may aim to reduce background noise and prevent using music that increases participants’ cognitive effort, thus decreasing their performance (i.e., clients, employees, or students). Designers should measure potential participants’ perceived effort on a given task and create environments that help address the task’s complexity. Music may enhance some emotions when used to accompany light and fluid (i.e., not overloaded) decision-making processes. Within this context, different levels of cognitive effort may occur, for example, shopping. Groceries may be picked according to a list (requires no effort). However, certain products pose as those that consumers find difficulty deciding but want to purchase (requires considerable effort). After recognizing such products, the shopping experience can be segmented, and localized music can accompany moments that require less effort but additional emotional arousal.
A second practical implication is the effect of music on virtual environments, which could be related to social networks, work-from-home, virtual education, and shopping experience. Given our findings, music has a detrimental effect on online performance or decision making if such a task requires a significant cognitive effort. This is particularly important because after lockdown due to the COVID-19 pandemic, virtual activities may be performed more frequently (Boumphrey, 2020).
The study presents certain limitations that serve as a direction for future research. Our experiment was conducted in a controlled laboratory environment, thus minimizing external issues that would naturally affect the participants’ task performance (i.e., background noise). However, laboratory experiments’ results may be taken with caution as more external validity is needed to conclude about them (Borgatta & Bohrnstedt, 1974). The effects of music and mediation of perceived effort on task performance can be explored further under stressful situations in natural environments. In a natural context, music can help as a protection mechanism that reduces the effort placed on a task and enhances emotional well-being (according to the arousal hypothesis). Thus, under a real stressful situation, the effect of music and effort may differ from our study.
A limitation is that we designed a stressful and highly cognitively demanding task, but participants were not asked about their stress level. It was expected to observe stress in using the mouse, but future research should ask participants about their stress level. Moreover, future research could control how distracting the music is to get a deeper understanding of how music interacts with effort and performance. In addition, future studies can contrast our results by developing neutral or enjoyable tasks.
