The effect of exposure to music stimuli on performance for tasks employing different systems of cognition

(1)

Msc Thesis Title: The effect of exposure to music stimuli on performance for tasks employing different systems of cognition

Name: Benjamin Rogerson

Student Number: 10392270

Study: Business Economics

Track: Managerial Economics & Strategy

ECTS: 60

Statement of Originality

This document is written by Student Benjamin Rogerson who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(2)

The effect of exposure to music stimuli on performance for

tasks employing different systems of cognition

Benjamin Rogerson University of Amsterdam

July 2017

Abstract: Agents are exposed to music in a variety of economic situations, including

situations which affect performance. Psychological research and neuroscience have shown that music stimuli can induce cognitive effects of various forms. To test whether exposure to music stimuli affect tasks employing different systems of cognition, an experiment is conducted in which participants perform a variety of cognitive task under economic incentives. The general findings of the paper were that exposure to music enhances performance on tasks that are relatively automatic, and worsens performance on those which employ more controlled cognitive processes. With the exception of a memory task, there was no evidence to suggest that performance under experimenter-selected music and participant-selected music differed. This is surprising given research that cognitive responses to self-selected music are more intense than exposure to control music.

(3)

1. Introduction

Much psychological research has demonstrated the range of ways in which music can affect various cognitive processes (Janata & Barucha, 2002, p. 121-135; Peretz & Zatorre, 2005, p. 89-114). Exposure to a music stimulus can have effects on both automatic processes and more controlled cognitive processes. This has important implications for economics as effects on cognition change the ability of agents to perform a range of cognitive functions that are essential in economic situations, like performing reasoning tasks or making decisions.

Studies in behavioural economics have employed various experimental methods to uncover the ways in which cognition is affected when performing multiple cognitive functions simultaneously (Pashler, 1998). To uncover the way in which cognitive processes are affected by each other, many of these experiments have been in the form of dual-task studies, or studies on exposure to stimuli (Shiv and Fedorikhin, 1999, p. 278-292). Studies of this nature enable a more comprehensive understanding of the nature of cognition in economic situations to be formed.

Economic experiments using music, however, have been limited, and have mainly focused on how exposure to music can affect consumption behaviour (Areni & Kim, 1993, p. 336-340; Miliman, 1986, p.286-289; Matilla & Wirtz, 2001, p. 273-289; Yalch & Spangenberg, 1990, p. 55-63). However, exposure to music stimuli in economic situations is not limited to situations of consumption. Economic agents are often exposed to music in situations where it may affect performance, such as in sports or the workplace (Atkinson, Wilson & Eubank, 2004; Karageorghis & Terry, 1997; Lesiuk, 2005, p. 173-191). Exposure to music in these performance related situations can have an impact on cognitive systems, which may affect performance. Consequently, it is useful to uncover the way in which performance under economic incentives is affected by exposure to music stimuli of different kinds.

The direction of research for this paper, is to expand dual-task behavioural economics studies of cognition to explore the interaction of music and performance. This is achieved using an experiment in which participants were exposed to different kinds of music stimuli whilst performing a range of tasks requiring different forms of cognition. The intention of this was to develop an understanding of how exposure to

(4)

a music stimulus can affect performance negatively or positively depending upon the type of cognition required for the task.

The main findings of the paper can be briefly summarised in the following way. There was evidence that exposure to a music stimulus increases performance for tasks that require lower levels of cognition. This supports research concerning how music induces emotion, and how enhanced emotion can be used to increase performance (Blood and Zatorre, 2001, p. 11818-11823). In addition to this, results supported the claim that for tasks requiring higher levels of cognitive processing, exposure to a music stimuli weakened agents ability to perform the task. This is, in accordance with theory, due to the decline in available higher order cognitive resources when listening to music (Janata & Barucha, 2002, p. 121-135; Peretz & Zatorre, 2005, p. 89-114).

When exploring the difference between experimenter-selected music and participant-selected music, there was little evidence to suggest that performance differs depending upon the kind of music stimulus. This, however, was with the exception of a memory task, in which performance under participant music was superior to that of experimenter-selected music. This is surprising given research that cognitive responses to self-selected music are more intense than exposure to control music (Blood and Zatorre, 2001, p. 11818-1182; North and Hargreaves, 1995, p. 77-93). The structure for the remainder of the paper is as follows. Firstly there is a literature review with three sections: theories of cognition, cognition and music, cognition and performance. Following this, the methodology is presented. This includes a description of the sample as well as a description of the experiment and reasons for how it was constructed and assessed. The hypotheses for the experiment are then formed based on theory from the relevant literature. This is followed by a results section with summary statistics, regression analyses and a discussion of the results. The paper ends with a summary of the conclusions and limitations of the experiment, as well as areas for future research.

(5)

2. Related Literature 2.1. Theories of cognition

In order to study the effects of music stimuli on tasks employing different kinds of cognition, an understanding of the theory behind cognition serves as a useful framework from which to build an experiment. Consequently, this section will explain the main trends and relevant papers on theories of cognition, including the frameworks that are used in the remainder of this paper.

A variety of theorists have developed dual-process theories of cognition in order to develop understanding of the structure of the cognitive process (Sloman, 1996; Evans, 1984; Evans & Over, 1996; Reber, 1993; Levinson, 1995; Epstein, 1994; Pollock, 1991; Hammond, 1996; Klein, 1998; Johnson-Laird, 1983; Shiffrin & Schneider, 1997; Posner & Schneider, 1975). Though the technical characteristics of these theories do not exactly match, these theories generally systematize into two types of cognitive system.

Stanovich and West (2000, p. 658) used the characteristics from research on these two channels to group them into the neutrally phrased system 1 and system 2.

The key properties of the two systems are as follows. System 1 is characterized as an automatic process, and is relatively undemanding of cognitive capacity. System 2 represents the various characteristics of controlled cognitive processes; these processes are relatively more demanding of cognitive capacity.

In light of dual-system theory, and more recent psychological research on intuition and reasoning, Kahneman (2003, p. 1450-1452) extended the model for use in behavioural economics.

His model made two key developments for understanding cognition. Systems 1 and 2 became respectively referred to as intuition and reasoning. In addition to this, he included an extension of the model to include a third cognitive channel: perception. Kahneman argued that all of the process characteristics of system 1 are attributable to this system of perception. It is thus possible to use information about the working processes of intuition to develop understanding of the working processes of perception, and vice versa. Whereas the content of systems 1 and 2 are

(6)

conceptually based, the content of perception is stimulus bound. Kahneman argues that the processes of perception and system 1 are effortless, whereas the processes of system 2 are effortful.

In order to develop understanding of the interaction and demands of these different cognitive systems, psychologists have performed a range of dual-task experiments (Pashler, 1998). Dual task studies enable researchers to observe the way in which cognitive processes disrupt each other (Kahneman, 2003, p. 1451). For example, when agents are required to perform a demanding cognitive task, this weakens their ability to perform other demanding cognitive tasks simultaneously. Kahneman uses the example of a driver who converses whilst driving. The degree to which he can converse effectively depends upon the attention to which is being paid to the driving task.

Dual-task studies have been used to uncover how the effectiveness of cognitive systems can be limited when agents perform multiple activities (Pashler, 1989). As pointed out by Kahneman (2003, p. 1451), a study by Gilbert (1989, p. 189-210) showed that when agents were performing a demanding mental activity, such as remembering several digits, this weakened their ability to perform other reasoning tasks.

Behavioural economists have built upon these dual-task approaches from psychology in order to develop an understanding of cognitive processes in economic situations. For example, dual-task studies have been used to make inferences about consumer choice (Gabarino and Edell, 1997; Hoch & Lowenstein, 1996; Loewenstein, 1996; Luce, 1998; Luce, Bettman & Payne, 1997; Shiv & Fedorikhin, 1999).

A relevant example of a dual-task study in behavioural economics on consumption behavior is a paper by Shiv and Fedorikhin (1999, p. 278). They ask participants to memorise a seven-digit number, and then observe the differences in choice between a healthy and unhealthy snack. They find that when undertaking the demanding cognitive process of remembering several digits, consumers are less able to make choices that are in their long-term benefit, and instead give in to more impulsive desires (like choosing a cake instead of fruit). This paper is a good example of how a study can be used to develop understanding of the interaction between perception,

(7)

and systems 1 and 2. Whereby system 2’s monitoring of system 1 (the degree to which the agent able to overrule the desire to eat the cake) is limited by the demands of the memory task.

This paper also demonstrates how dual-task approaches are not limited to variations in the task, but can also be studied using variation in the stimulus. Shiv and Fedorikhin (1999, p. 279) refer to this as stimulus-induced affect. Researchers who study the interaction of cognition and stimuli propose that reaction to stimuli induce various kinds of cognitive events (Berkowitz, 1993; Ledoux, 1987). These can be generalized to mainly automatic, rapid processes (that would belong to system 1) and slower higher order cognitive processes (system 2).

2.2. Research on the cognitive effects of music

Despite the various psychological and behavioural economics studies on how cognitive processes are affected by stimuli, music is one stimuli which has been used sparingly in economic experiments. This may in part be due to the lack of consensus among psychology researchers in how music information is processed by an agent. This makes it difficult to construct experiments from which concrete conclusions about cognition can be drawn.

Though results of research are broad, much psychology research suggests that exposure to music creates cognitive effects of many forms, across many levels. (Janata & Barucha, 2002, p. 121-135; Peretz & Zatorre, 2005, p. 89-114). Using the framework provided by Kahneman’s (2003, p. 1451) three channels of cognition, there is research to suggest that exposure to a music stimulus in the perception channel can have effects on system 1, through emotion, and system 2, via more controlled processes. This research will be elaborated below.

Focusing first on system 1 effects, a large amount of psychological research has focused on whether music can induce emotion, and how this can be brought about. A study by Konecni (2008, p. 120) showed that music can induce emotional reactions such as joy, sadness, anger and fear, but these reactions can only take place when mediated by associations to the music. Thus emotional responses to music are limited to exposure to music that the participant has already been in contact with. This idea supports research by North and Hargreaves (1995, p. 77-93),

(8)

in which they showed subjects enjoy music more when it is more familiar to them. They show that repeated exposure to more familiar music corresponds to a greater effect on brain activity than that of control selected music. This idea is further endorsed by Blood and Zatorre (2001, p. 11818-11823), who find using a neuroscientific approach of positron imaging tomography, that self-selected music is more successful in inducing emotion than music selected by another.

With the aid of further developments in neuroscience, research has also shown that cognitive responses to music are not limited to emotional responses, but also can also induce effects on parts of the brain associated with reasoning processes of system 2. Janata and Barucha (2002, p. 121-135) found that exposure to polyphonic music results in brain activity that recruits general attention and working memory units. A similar view was held by Peretz and Zatorre (2005, p. 89-114), who argued that different aspects of music require different levels and systems of cognitive processing. Their research also suggests that increasing the complexity of the music stimuli, i.e. by increasing the amount of instruments in the track, results in a higher level of cognitive processing.

Behavioural economics research using music has tended to focus on how exposure to a music stimulus can affect consumption behavior (Areni & Kim, 1993, p. 336-340; Miliman, 1986, p.286-289; Matilla & Wirtz, 2001, p. 273-289; Yalch & Spangenberg, 1990, p. 55-63). This paper will extend the use of music in behavioural economics to observe the way in which exposure to a music stimulus can affect performance under economic incentives.

2.3. Behavioural economics literature related to performance and cognition The main contribution from performance related literature that is essential to this paper is Ariely et al’s (2009, p. 451-469) paper ‘Large stakes and big mistakes’. In the second experiment of their paper, they intentionally select two games, which require differing levels of cognitive capacity. The two games are a key-pressing task, which requires no cognitive effort and is automatic, and an arithmetic task, which requires cognitive effort to perform simple additions. By doing this, it enables them to look at how a variable, in this case arousal from the incentive level, affects performance of tasks which employ differing cognitive systems. Using the framework

(9)

proposed by Kahneman (2003, p. 1451), the key-pressing task of Ariely would fall into the cognitive processes of system 1 (intuition), whereas the arithmetic task demands the higher order processing capacity of system 2 (reason).

Aside from the study of Ariely et al, research studying the focus of attention are also relevant to this paper. Easterbrook (1959) argued that increased emotion can enhance motivation for tasks that do not require creative insight. The increase in motivation can help the agent to narrow their focus of attention to a smaller variety of dimensions. This efficiency enhancement can aid agents in performance in various types of cognitive game.

Research looking at the effects of moving from controlled to automatic thinking are also relevant for this paper. Various studies show that moving from automatic to controlled cognitive processes can have a detrimental effect on performance (Langer & Imber, 1979; Camerer Loewenstein and Drazen, 2005). Ariely (2001, p. 453) uses the example of how a golfer who is overthinking can perform worse than when he is performing more automatically.

Research on the effects of skill level on performance also play a role in this paper. Kahneman (2003, p. 1450) proposes that prolonged training results in increased skill, and this makes higher order cognitive tasks become more automatic and effortless. He uses Simon and Chase’s (1973) the example of a chess-player who can instantaneously see the perfect move in a chess game.

2.4. Key theory for the remainder of the paper

In light of the success of Ariely et al. (2009, p. 451-469) in their paper on performance, the experiment of this paper utilised aspects of their experimental design. Primarily, this was the focus on variations in the performance level for different kinds of cognitive task. Rather than varying the incentive level however, the experiment focused upon the effects of music stimuli on agent ability to perform tasks that require different cognitive systems.

In addition to this, the cognitive framework proposed by Kahneman (2003, p. 1451), containing perception, intuition (system 1) and reasoning (system 2) as the three channels, will be referred to throughout the remainder of the paper.

(10)

The unique direction of this paper is to expand the work in behavioural economics on the interaction of music and cognition, in order to develop an understanding of the interaction of a music stimulus and performance under economic incentives. The experiment used to study this interaction is explained in the following section.

(11)

3. Methodology 3.1. Experimental Design 3.1.a. Sample description, conditions and payment

Forty-one participants took part in the experiment across three days in July 2017. Participants were recruited by email or word of mouth to participate in an economic experiment studying the effects of music. They were invited to attend the experiment at a given time, and were requested to bring a music album of their own choice to the experiment. Of the 41 participants, 17 were female and 24 male. 31 participants had at least a bachelor degree, 8 were soon to graduate and 2 had no university education. Annual income of the group ranged from the ‘0-10 000’ euro bracket to incomes ‘over 40 000’ euros.

Participants performed the experiment individually, which involved three different tasks. They performed the 3 tasks under 3 different conditions. The three conditions under which the task was played were: silence, control music and participant-selected music. The sequence of conditions was randomized for each participant. Two different kinds of music stimuli were used. Control music refers to music selected by the experimenter, from which each individual was exposed to the same music. Given that more complex tracks are associated with an increase in brain activity, an album considered highly complex (in terms of the range of sounds and instrumentation used), was selected as the control in order to ensure strong activation of system 2 upon exposure. An album that was expected to be relatively unknown was also selected, in order to prevent differences in familiarity between subjects. The title of the album referred to was Holly Herndon’s experimental electronic album Platform, from 2014. Participant-selected music refers to the album participants brought along themselves to the experiment, this varied for each participant.

Participants were paid for performance in the experiment. Payment for each task was conditional upon performance. Participants were told that, at the close of the experiment, the performance one of the rounds of one of the participants would be randomly selected for payment. The magnitude of the payment depended upon the success of their individual performance in each task. For each of the three tasks,

(12)

participants could earn up to 25 euros. Thus it was possible for a participant to earn up to 75 euros in the experiment.

3.1.b. The Tasks

The three tasks were separated into two categories: low cognitive demand and high cognitive demand tasks. One task was of a low cognitive demand level, and two of a higher cognitive demand level. This was done to observe how performance is affected differently for tasks which require different forms of cognition. In addition, as argued by Ariely et al. (2009, p. 460) the tasks chosen were selected due to their familiarity with tasks that participants may experience in the workplace (simple addition, memorising, typing).

Due to the multitude of reasoning behaviour that can take place in system 2, two higher order cognitive tasks were included in the experiment to expand the range of system 2 functions that could be studied. Though the study of more types of cognitive task was preferred, this was limited due to time restrictions for the experiment.

The three tasks were adapted from Ariely et al’s (2009, p. 451-469) paper on incentives and performance. The way in which participants earnt points was also formed given the performance levels of that experiment.

The low cognitive demand task was to type an alternation of two keys into the keyboard as many times as possible within 2 minutes. For each 20 correct combinations typed, participants received a reward of one point. It was possible to earn up to 25 points for this task.

The first high cognitive demand task was an arithmetic task. The task was to find the two numbers in a matrix of 12 numbers, which summed to 10. The numbers were all three digits and two decimal places. After answering a matrix, participants could move onto the next one. For each correct matrix solved, participants received 2.5 points. There were 10 matrices. It was thus possible to earn up to 25 points for this task. Participants had 2 minutes to perform the task.

The second high-cognitive demand task was a memory task. A series of three-digit numbers was revealed to the participants. At a random point, the series was stopped

(13)

and the participants were made to recall the last 3 numbers. For each correct recollection, a participant earnt 5 points. There were 5 series to be shown in total. It was thus possible to earn up to 25 points for this task.

3.2. Statistical analysis

Once all the data was collected, a statistical analysis was performed using the software STATA. This included two regression analyses, which were performed in order to perform statistical tests on the differences in performance level across the three treatments. The first regression was used to test if performance increased or decreased under exposure to either music stimuli, relative to the silent treatment. The second regression was used to test differences in performance level for participant selected music relative to control music. This is further explained in the results section.

3.3. Hypotheses

Using the cognitive framework postulated by Kahneman (2003, p. 1450-1452), predictions were made regarding the effects of music on the performance of the three games. It was believed that the introduction of a music stimulus would affect both the lower order cognitive system (system 1) and the higher order cognitive system (system 2).

As success in the key-pressing task requires no higher order cognitive effort (except monitoring) and is automatic (system 1), it can be inferred that performance on this task depends only upon effort exerted. Hence the increased use of system 2 by the introduction of a music stimulus does not affect the agent’s ability to perform the task. On the other hand, it is possible that the emotional impact of the music stimulus increases enjoyment, which causes the agent to increase effort on the task (Blood and Zatorre, 2001, p. 11818-11823). This is with the exception of the case that under silence, effort in the task is already at a maximum, in which case it would be expected that performance under exposure to music remains the same.

H1: For those tasks that require low cognitive effort (system 1), the introduction of a music stimulus either maintains or increases performance.

(14)

Given research that exposure to music utilises the higher-order systems of working memory and attention, for those tasks that require high cognitive effort (system 2), it was expected that the introduction of a music stimulus would limit the cognitive capacity available to perform reasoning tasks (Janata & Barucha, 2002, p. 121-135; Peretz & Zatorre, 2005, p. 89-114). Hence it was expected that performance on the arithmetic and memory task would decline under exposure to a music stimulus.

H2: For those tasks that require high cognitive effort (system 2), the introduction of a music stimulus decreases performance.

Due to research suggesting that the emotional response to music is dependent upon its familiarity, it was expected that participant-selected music would have a more significant emotional response than control music (North and Hargreaves, 1995, p. 77-93). Consequently, this enhanced emotion would result in a greater application of effort for the (automatic) key-pressing task (Easterbrook, 1959). Hence the change to performance level for lower level cognitive tasks would be even greater under participant selected music than under control music.

H3: Performance on the task requiring system 1 is enhanced more by participant-selected music than by control music.

Given research that self-selected music is more likely to induce an emotional response, it was believed that this emotional response is also likely to induce more activity in system 2 (North and Hargreaves, 1995, p. 77-93). In addition, exposure to familiar music is more likely to induce associations, which are likely to affect system 2, whereas novel music stimuli are not able to induce associations (Konecni, 2008, p. 120). Consequently, it was believed that self-selected music would cause further harm to the performance of higher-order cognitive tasks than control music, due to the increase in activity of system 2 processes.

H4: Performance on high cognitive demand tasks (system 2) is worse under participant-selected music than control music.

(15)

4. Results

4.1. Summary Statistics

Using the rationale of Ariely et al’s (2009, p. 55) experiment, there are two ways to look at the dependent variables of the three tasks in this experiment. Firstly, one can observe the performance levels in raw scores. This is presented in table 1.

Comparison across the games however, is simplified by looking at performance in the games as a percentage of the maximum possible score. As participants received points in a linear manner, this measure is identical to looking at performance as the proportion of maximal earnings. The results in this format are displayed in table 2. The trends across the two methods are the same regardless of the presentation mode. However, in order to simplify comparisons across tasks, the remainder of the paper will focus on performance as a percentage of the maximal score.

Consequently, the trends of the experiment are depicted in figure 1. From this figure, one can see that performance increased for the key-pressing task for both music stimuli, relative to silence. Performance worsened for the arithmetic task for both stimuli relative to silence. For the memory task, performance only made a significant decline under exposure to control music. This was not observed for exposure to participant-selected music.

4.2. Regressions

Looking at the trends is not sufficient to draw conclusions about the differences in performance level in the various treatments. In order to do so, two regression analyses were formed in order to perform statistical tests on the differences between the data. The first regression was used to test if performance increased or decreased under exposure to either music stimuli, relative to the silent treatment. The second regression was used to test differences in performance level for participant-selected music relative to control music.

The outcome of the first linear regression (with robust standard errors) is presented in table 3. This uses the dependent measure of performance, and the independent variables of dummies for control and participant-selected music. The results are separated by task.

(16)

TABLE 1

Raw scores by task and treatment

Performance Task Mean performance as a raw score (S.D.)

Silence Control Participant

Key Pressing 258.0 284.9 281.4 (37.7) (32.3) (37.0) Arithmetic 5.3 4.1 3.6 (1.7) (1.2) (1.5) Memory 3.6 2.7 3.4 (1.0) (1.1) (0.9)

Notes: The interpretations of the scores varies by task: (i) Key pressing, number of correct

combinations entered. (ii) Arithmetic, number of matrices correctly solved in 2 minutes. (iii) Memory task, number of correctly recited series. 41 participants performed each of the three tasks.

TABLE 2

Percentage of maximal performance by task and treatment

Performance Task Mean percentage of maximal earnings (S.D)

Silence Control Participant

Key Pressing 0.74 0.81 0.80 (0.10) (0.09) (0.11) Arithmetic 0.53 0.41 0.36 (0.17) (0.12) (0.15) Memory 0.72 0.54 0.68 (0.20) (0.22) (0.18)

Notes: Percentage of maximal performance was calculated in the following way for each

task: (i) Key pressing, raw score divided by 350. (ii) Arithmetic, raw score divided by 10. (iii) Memory task, raw score divided by 5. 41 participants performed each of the three tasks.

(17)

Notes: The vertical axis measures the mean performance, measured using the score as a

percentage of maximal performance. The horizontal axis indicates the type of task: key-pressing, arithmetic, memory. For each task, the three treatments are displayed in the following order: Silent treatment, Control-music treatment, Participant-music treatment.

These results support the first hypothesis, that exposure to a music stimulus increases performance for tasks which are undemanding of cognitive capacity. For the key-pressing task, positive differences in performance level were significant for both the control-music treatment and the participant-selected music treatment, at the 5% and 10% level respectively.

Support for the second hypothesis, that exposure to a music stimulus decreases performance in high cognitive tasks, is varied. For the arithmetic task, in line with the hypothesis, performance declined in both treatments. These differences were significant at the 5% level for control music and the 1% level for participant-selected music. However, for the memory task, a significant (at the 1% level) decline in performance was only observed for the control music treatment. Under participant-selected music, there was no significant difference in the performance level.

This can be summarised as follows. While control music limited reasoning ability of systems associated with arithmetic tasks and memory tasks, participant selected music only limited the reasoning ability of systems associated with arithmetic tasks but not those of memory tasks. This may be explained by the concept of familiarity.

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Key Pressing Arithmetic Memory

Me an Pe rfo rm an ce FIGURE 1

Mean performance by task and treatment

(18)

Because those tracks are more familiar to participants, they are less likely to use up the available working memory capacity. Whereas for arithmetic tasks, a more controlled reasoning process is required, the functioning of this process is limited by both music stimuli.

The comparison of participant-selected music relative to control music is presented in the regression with robust standard errors in table 4. This uses performance for each task as the dependent variable, and the independent variable is the dummy for participant-selected music. The results are separated by task.

These results did not support our third hypothesis that under low cognitive demand tasks, performance is increased more under participant music than under control music. There were no significant differences observed between control and participant music for the key-pressing task.

One possible explanation for this is that performance was already at a near maximum under either of the music stimuli. This prevents participants from being able to achieve a higher performance level when they are undertaking a more intense emotional response under participant-selected music.

The results only gave limited support for the fourth hypothesis, that performance for high demand cognitive task is worsened further under participant music. Only the performance in the memory task held a significant difference under participant music relative to control music. Performance under participant-selected music was improved, this was significant at the 5% level.

This may again follow the aforementioned reasoning that participants found it easier to retain working memory capacity when exposed to music that is more familiar.

(19)

TABLE 3

Linear regression results by task

Performance Dummies Coefficient

(robust S.E.)

Key

Pressing _Arithmetic _Memory

Coefficient Control music 0.0768** -0.1146** -0.1805*** (0.0221) (0.0324) (0.0469) Participant music 0.0662* -0.1634*** -0.0429 (0.2358) (0.0351) (0.0430) Constant 0.7373 0.5268 0.7220 (0.1681) (0.2637) (0.0319) Observations 41 41 41 R-squared 0.1024 0.173 0.1278

Notes: The dependent variable is mean performance in each game (calculated using fraction of

maximal performance), the independent variables are dummies for the two types of music stimulus used: experimenter-selected and participant-selected. Robust S.E.s are reported. Significant differences are marked in the following way: (p < 0.1) is marked *. (p < 0.05) is marked **. (p < 0.01) is marked ***.

TABLE 4

Linear regression results comparing control music relative to participants music

Performance Dummies Coefficient

(robust S.E.)

Key

Pressing Arithmetic Memory

Coefficient Participant music -0.0102 -0.0488 0.1366** (0.0219) (0.0298) (0.0448) Constant 0.8141 0.4122 0.5415 (0.0144) (0.0189) (0.0344) Observations 41 41 41 R-squared 0.0027 0.0324 0.1041

Notes: The dependent variable is mean performance in each game (calculated using fraction of

maximal performance), the independent variable is the dummy for participant-selected music. Robust S.E.s are reported. Significant differences are marked in the following way: (p < 0.1) is marked *. (p < 0.05) is marked **. (p < 0.01) is marked ***.

(20)

4.3. Limitations

It is possible that the experiments in the game did not offer sufficient variation in potential performance level to test variations between treatments. For example, it may be that in the key-pressing task, participants had already optimized their performance in the game under control music. Thus, despite further cognitive stimulation and emotional activity, it was not possible for them to perform the game at any higher level. This theory would not hold for the arithmetic and memory task however, as performance generally declined for both music stimulus treatments. In addition to this, the two high demand cognitive tasks did not allow for much variation in performance. For the arithmetic task, it was only possible to achieve a performance level between 1 and 10 correct matrices. For the memory task, only 5 series were used. This could have been strengthened by allowing participants to perform tasks for longer and increasing the number of matrices or series used. By having more variation in results, more accuracy concerning the difference in performance level across the treatments could be taken.

A further limitation of the results is that only two forms of demanding cognitive tasks were used, whilst there are a far large number of reasoning processes that could be included for study. For example, a creativity task and a visualization task may enable conclusions to be expanded to cognitive processes affecting problem solving and insight. Creativity is one area that prior research showed can be negatively affected by the effects of certain stimuli on the cognitive process (Easterbrook, 1959). Reasoning behaviours such as creativity and insight skills are also commonly used in the workplace environment.

The experiment also suffers from a selection bias. As participants were found by word of mouth and email amongst social circles in Amsterdam, it is likely that they are not a representative sample of all workers in a workplace.

It is also possible that they possess arithmetic and memory skills above the average, given their relatively high-level and recent exposure to education. This would have an effect on whether cognitive processes were automatic or controlled for many of the participants (Langer & Imber, 1979; Camerer, Loewenstein and Drazen, 2005). If, for example, students who had undertaken much numerical study were well

(21)

trained in arithmetic, this implies that the arithmetic task may have actually been relatively automatic (a system 1 process), rather than a controlled process of system 2. This would make it less likely that performance is worsened by a stimulus which uses up part of the capacity of system 2. This could have been explored using a small test to calculate the skill level of individuals in the experiment for memory and arithmetic tasks.

Given the fairly high income of many of the participants, the monetary incentive used in the experiment may also be insufficient to encourage maximum effort by every participant. Unlike the paper of Ariely et al. (2009, p. 454) in which they were fortunate enough to use low-income earners in order to improve the strength of their incentives, this experiment was limited by budgetary restrictions to performing on people located in Amsterdam. As a consequence, monthly income amongst the sample was relatively high compared to the incentive. Improving the saliency of the decisions of participants in the experiment may help to reduce variability in the performance level.

(22)

5. Discussion and Conclusions 5.1. Summary of main conclusions

There are three main conclusions that can be drawn from the experiment.

Firstly, in line with the first hypothesis, exposure to a music stimulus increases performance for tasks that require a low level of cognitive demand (only effort). This supports research suggesting that music can induce emotion, which in turn can have positive effects on tasks by which performance is driven solely by effort (Blood and Zatorre, 2001, p. 11818-11823).

Secondly, in partial support of the second hypothesis, exposure to a music stimulus generally worsens performance for highly demanding cognitive tasks. With the exception of the participant-selected music treatment for the memory task, performance declined significantly in every other situation. This is in line with research suggesting that exposure to a music stimulus activates cognitive systems associated with higher-order system 2 processes (Janata & Barucha, 2002, p. 121-135; Peretz & Zatorre, 2005, p. 89-114).

Differences in the effects on the memory task relative to other tasks could be driven by the effects of familiarity on how well participants are able to maintain available cognitive capacity for different kinds of task, though this requires further exploration. These results do suggest however, that beliefs concerning the effects of a music stimulus upon performance in highly demanding cognitive tasks should not be oversimplified. This is hardly surprising given the multitude of ways in which cognitive systems can be affected by both the perception channel and the reasoning channel, especially when performing different kinds of cognitive task (Janata & Barucha, 2002, p. 121-135).

The third conclusion that can be drawn is that, contrary to expectation, differences in performance under participant-selected music relative to control music may be less extreme than expected. There is limited support for hypotheses three and four. Though there was a significant difference observed for the memory task, performance was generally not significantly different under the two treatments. It is conceded, however, that similarities in the performance level for the two tasks may be driven by the lack of possible variation in the performance levels of each task.

(23)

Further understanding of what is driving the difference in performance under participant music for the memory task requires further investigation. Developing a more complex understanding of which reasoning systems are employed by different kinds of task may help to understand why only the performance on the memory task improves under participant-selected music. As mentioned, it could be that exposure to familiar music affects the working capacity of that specific system required for memory functions less than exposure to novel music.

5.2. Summary of main limitations

There were a number of limitations that were raised in the results section for the experiment, that will now be summarised.

The experiment could be improved by allowing for more variation in performance level for all three kinds of task. This would prevent participants from reaching a maximum performance level too easily, which makes comparisons between treatments more difficult. In addition, more variation in the performance levels possible for the arithmetic and memory task would enable more accuracy regarding the difference in performance level between treatments.

As well as allowing for more variation in the task, increasing the number of different types of cognitive task used in the experiment would help to separate the effects of a music stimulus on a wider variety of cognitive functions.

In addition to this, removing the selection bias in the sample would improve the experiment. This would also prevent the risk of over-trained or over-skilled participants performing seemingly controlled cognitive tasks in an automatic manner. This problem weakens inferences concerning the interaction of system 2 tasks and music stimuli.

Furthermore, increasing the saliency of the monetary incentives, either by using a sample of lower income earners, or by increasing the incentive levels themselves would enhance how seriously participants valued the outcome of their decisions.

(24)

5.3. Steps for future research

Concerning future research on this topic. Stronger conclusions could be made using the same experiment by improving the strength of economic incentives and increasing the variability of possible performance levels in the task.

In addition, extending the experiment to include closer assessment of the impact of familiarity on memory tasks may help to uncover the differences in results for the memory task in regression 4.

The conclusions of the paper are not generalizable to all tasks. Using a wider range of cognitive task in future experiments would be helpful in uncovering which kinds of reasoning in system 2 are affected by exposure to a music stimulus. This would allow for more conclusions concerning different types of task representative of those taking place in performance-related situations (such as the workplace).

In addition to this, further neuroscience and psychological developments making a more deliberate effort to untangle the ways in which system 2 reasoning processes occur (for example, memory processes compared to arithmetic processes), could allow for a better understanding of the differences in performance levels for different kinds of cognitive task under exposure to music.

(25)

6. Appendix

Experiment instructions Introduction

Welcome to this experiment. Please pay careful attention to these instructions.

It is possible to earn money in the course of this experiment. The amount earnt is dependent upon your performance across 3 tasks. Your potential earnings are denoted in points.

The experiment is conducted in 3 rounds. In each part, you will perform 3 different kinds of task.

At the conclusion of the experiment, one participant is randomly chosen. The performance of this participant in one of the rounds will be randomly selected for payment. Each point earnt in that round corresponds to €1.00. In each task it is possible to earn up to 25 points. Thus it is possible to earn up to €75.00. Each participant and each round are equally likely to be selected for payment.

Task description

In this experiment you will perform 4 different kinds of tasks. Your earning depends upon your individual performance only and not on the performance of other participants. The three different kinds of tasks are a key pressing task, arithmetic task and memory task. The tasks are explained below. You will be given an opportunity to do a practice run for each task after reading this section.

Key pressing task

For this task, you are given a combination of 2 letters. The task is to type the combination into the keyboard as many times as possible within 2 minutes.

For every 20 correct combinations that are typed you receive 1 point. You can earn a maximum of 25 points for this task.

Arithmetic task

For this task, you are given matrices which contain 9 numbers. The task is to find the 2 numbers in the matrix which sum to 10. After circling the 2 numbers, you can move onto the next matrix. You have 1 minute to perform this task.

For every correct matrix you can earn 2.5 points. There are 10 matrices. You can earn a maximum of 25 points for this task.

Memory task

For this task, you will be shown a series of numbers. At a random point, the series will be stopped and you must attempt to recall the last 3 numbers.

For each time the last 3 numbers are recalled correctly, you earn 5 points. There are 5 series to be shown. You can earn a maximum of 25 points for this task.

(26)

Practice Round

You will now be given the opportunity to perform a practice of each of the tasks. Your performance in this round does not have an impact on your potential earnings.

For every 20 correct combinations that are typed you receive 1 point. You can earn a maximum of 25 points for this task.

The task begins when the experimenter reveals the combination of letters for you to type. After 2 minutes the task will end.

Arithmetic task

For this task, you are given 10 matrices which contain 9 numbers. The task is to find the 2 numbers in the matrix which sum to 10. After circling the 2 numbers, you can move onto the next matrix. You have 1 minute to perform this task.

For every correct matrix you can earn 2.5 points. There are 10 matrices. You can earn a maximum of 25 points for this task.

The task begins when the experimenter reveals the first matrix. After 1 minute the task will end.

Memory task

For each time the last 3 numbers are recalled correctly, you earn 5 points. There are 5 series to be shown. You can earn a maximum of 25 points for this task.

The task begins when the experimenter reveals the series. The task will end after the 5th_series.

Round: silence

You will now be given the opportunity to perform each of the tasks in silence. Your performance in this round contributes to your potential earnings.

(27)

Arithmetic task

Memory task

Round: Control music

You will now be given the opportunity to perform each of the tasks again.

In this round you will perform the task whilst listening to preselected by the experimenter

music through the headphones.

Your performance in this round contributes to your potential earnings.

(28)

Arithmetic task

Memory task

Round: Participant-selected Music

You will now be given the opportunity to perform each of the tasks again.

In this round you will perform the task whilst listening to music of your own choice through the headphones.

Your performance in this round contributes directly to your potential earnings.

(29)

Arithmetic task

Memory task

(30)

7. Bibliography

Areni, C. S. and Kim, D. (1993). The influence of background music on shopping behavior: classical versus top-forty music in a wine store. Advances in

Consumer Research, 20, 336-340

Ariely, D., Gneezy, U., Loewenstein, G. and Mazar, N. (2009). Large Stakes and Big Mistakes. The Review of Economic Studies, 76(2), 451-469.

Atkinson, G., Wilson, D., & Eubank, M. (2004). Effects of music on work-rate

distribution during a cycling time trial. International Journal of Sports Medicine, 25(08), 611-615.

Berkowitz, L. (1993). Towards a General Theory of Anger and Emotional

Aggression: Implications of the Cognitive Neoassociationistic Perspective for the Analysis of Anger and Other Emotions. Advances in Social Cognition, (6), 1–46.

Blood, A. J. and Zatorre, R. J. (2001). Intensely Pleasurable Responses to Music Correlate with Activity in Brain Regions Implicated in Reward and Emotion.

Proceedings of the National Academy of Sciences of the United States of America, 98(20), 11818-11823.

Camerer, C. F, Loewenstein, G. and Drazen P. (2005). Neuroeconomics: How Neuroscience Can Inform Economics. Journal of Economic Literature. 43, 9-64.

Easterbrook, J. A. (1959). The Effect of Emotion on Cue Utilization and the Organization of Behavior. Psychological Review, 66 (3), 183-201.

Epstein, S. (1994). Integration of the cognitive and the psychodynamic unconscious.

American Psychologist, 49, 709–24.

Evans, J. B. T. (1984). Heuristic and analytic processes in reasoning. British Journal

of Psychology, 75, 451–68.

(31)

Garbarino, E. C. and Edell J. A. (1997). Cognitive Effort, Affect, and Choice. Journal

of Consumer Research, 24, 147–158.

Gilbert, D. T. (1989). Thinking Lightly About Others: Automatic Components of the Social Inference Process. Unintended thought. 189–211.

Hammond, K. R. (1996). Human judgment and social policy. Oxford University

Press.

Hoch, S. J. and Loewenstein, G. F. (1991). Time Inconsistent Preferences and Consumer Self-Control. Journal of Consumer Research, 17, 492–507. Janata, P. Tillman, B. and Bharucha, (2002). Listening to polyphonic music recruits

domain-general attention and working memory circuits. Cognitive, Affective, &

Behavioral Neuroscience, 2 (2), 121-140.

Johnson-Laird, P. N. (1983). Mental models: Towards a cognitive science of language, inference and consciousness. Harvard University Press.

Karageorghis, C. I., & Terry, P. C. (1997). The psychophysical effects of music in sport and exercise: A review. Journal of Sport Behavior, 20(1), 54.

Klein, G. (1998). Sources of power: How people make decisions. MIT Press. Konecni, V. J. (2008). Does Music Induce Emotion? A Theoretical and

Methodological Analysis. Psychology of Aesthetics, Creativity, and the Arts, 2(2), 115–129.

Langer, E. J. and Imber, L. G. (1979). When Practice Makes Imperfect: The Debilitating Effects of Overlearning. Journal of Personality and Social

Psychology, 37 (11), 2014-2024.

Ledoux, Joseph E. (1987), Emotion, Handbook of Physiology, Section 1: The

Nervous System: Higher Functions of the Brain, (5), 419–460.

Lesiuk, T. (2005). The effect of music listening on work performance. Psychology of music, 33(2), 173-191.

Levinson, S. C. (1995). Interactional biases in human thinking. Social intelligence and interaction, Cambridge University Press.

(32)

Loewenstein, G. F. (1996). Out of Control: Visceral Influences on Behavior.

Organizational Behavior and Human Decision Processes, 65, 272–292.

Luce, M. F. (1998). Choosing to Avoid: Coping with Negative Emotion-Laden Consumer Decisions. Journal of Consumer Research, 24, 409–433. Luce, J., Bettman, R. and Payne, J. W. (1997). Choice Processing in Emotionally

Difficult Decisions. Journal of Experimental Psychology: Learning, Memory

and Cognition, 23, 384–405.

Mattilaa A. S. and Wirtz, J. (2001). Congruency of scent and music as a driver of in-store evaluations and behaviour, Journal of Retailing, 77, 273–289.

Milliman, R. E. (1986). The Influence of Background Music on the Behavior of Restaurant Patrons Source, Journal of Consumer Research, 13(2), 286-289.

North, A. C. and Hargreaves, D. J. (1995). Subjective complexity, familiarity, and liking for popular music, Psychomusicology, 14, 77-93.

Pashler, H. E. (1998). The Psychology of Attention. Cambridge, MIT Press.

Peretz, I and Zatorre, R. J. (2005). Brain organization for music processing. Annual review of psychology, 56, 89–114.

Posner, M. I. & Snyder, C. R. R. (1975). Attention and cognitive control. Information

processing and cognition.

Pollock, J. L. (1991). OSCAR: A general theory of rationality. Philosophy and AI: Essays at the interface, MIT Press.

Reber, A. S. (1993). Implicit learning and tacit knowledge. Oxford University Press. Shiffrin, R. M. & Schneider, W. (1977). Controlled and automatic human information

processing: II. Perceptual learning, automatic attending, and a general theory.

Psychological Review, 84, 127–90.

Shiv, B. and Fedorikhin. A. (1999). Heart and Mind in Conflict: The Interplay of Affect and Cognition in Consumer Decision Making. Journal of Consumer Research, 26(3), 278–92.

(33)

Simon, H. A. and Chase, W. G. (1973). Skill in Chess. American Scientist, 61(4), 394– 403.

Sloman, S. A. (1996). The empirical case for two systems of reasoning.

Psychological Bulletin 119, 3–22.

Stanovich, K. E. and West, R. F. (2000). Individual Differences in Reasoning:

Implications for the Rationality Debate. Behavioral and Brain Sciences, 23(5), 645–65.

Yalch, R. and Spangenberg, E. (1990). The Effects of Store Music on Shopping