Research: Noise, Cognitive Function, and Worker Productivity

Title: Noise, Cognitive Function, and Worker Productivity
Authors: Joshua T. Dean
Access the original paper here
Listen to a deep-dive podcast:

Paper summary

This document explores the impact of noise on worker productivity in a real-world setting. Through experiments with textile workers, it demonstrates that increased noise levels reduce output, suggesting that noise significantly impairs cognitive function, which is crucial for such tasks. Furthermore, the research finds that workers appear unaware of noise’s detrimental effect on their productivity, as their willingness to pay for quieter conditions is not influenced by whether their pay depends on performance. This highlights a potential need for policy interventions to address the negative economic consequences of noise pollution, especially in rapidly urbanizing developing regions.

If teachers are to remember one thing from this study, it should be…

Noise can impair cognitive functions, such as attention and working memory, which are crucial for task management and performance, and individuals (like the workers in this study) may not be aware of how significantly noise affects their performance

Paper Deep Dive

Define any technical terms used in the paper

Based on the sources provided, here are some of the technical terms used in the paper, along with their definitions or explanations found within the text:

Working memory: A component of cognitive function involving the ability to store and manipulate information in memory and continuously update information. These skills are needed for tasks like sewing to keep track of how elements fit together and one’s place in the task. Assessed using the Reverse Corsi Block and N-Back tasks.

Abatement: Actions taken to reduce noise levels. This is relevant to firm decisions about managing workplace noise.

Ad valorem tax: A type of tax mentioned in the theoretical discussion of efficiency implications in Appendix E, typically based on value [E268]. (Not explicitly defined in the main text or appendix).

Attention: A component of cognitive function (also called executive function), described as the ability to direct one’s attention. It is crucial for tasks like sewing, requiring focus on multiple cues simultaneously. Specific aspects assessed include sustaining focus (Psychomotor Vigilance Task) and ignoring distractions (d2 task).

Bayesians: Refers to individuals who update their beliefs based on new information. The “failure to notice” model mentioned in the study describes individuals as Bayesians with a two-level hierarchical belief structure.

Becker et al. (1964) procedure (BDM): A method for eliciting willingness to pay (WTP) in an incentive-compatible way, where respondents state their maximum WTP for a good, and a random price is drawn [103, A191]. If the price is below their WTP, they purchase the good at the random price. A modified version was used in the study to elicit WTP for quiet working conditions [103, A191].

Clustered Standard Errors: A statistical adjustment made when calculating standard errors in regressions to account for the fact that observations within certain groups (in this study, rooms within a session) may not be independent. This is important because the treatment was randomized at the room x session level.

Cognitive function (also called executive function): Defined as all of the general-purpose abilities involved in task management. This includes key skills such as the ability to direct one’s attention, manipulate information in memory, and switch between tasks. These abilities are considered critical for many types of work and are correlated with success in various areas, including school.

Common factor analysis (CFA): A statistical method used to create an index from multiple cognitive test results. It assumes that correlations between measures are due to a single latent variable (cognitive function) and constructs factors that explain the most covariance. The first factor from this analysis was the preferred index of cognitive function.

Compensating differential: An economic concept discussed in the theoretical Appendix E, typically referring to a wage difference that arises to compensate workers for undesirable job characteristics [E259]. The study investigates the related idea of workers being willing to pay for (accept lower wages for) quiet working conditions.

dB (Decibel): A unit for measuring noise level. The study notes that a 10 dB increase is perceived by the human ear as twice as loud. Noise levels in regressions are often reported in 10s of decibels for interpretability.

Effort task (Placebo): A task designed to require inputs like motor control and effort but minimal cognitive function (alternating pressing ‘a’ and ‘b’ keys). It was used as a placebo to help identify that noise’s impact was primarily through cognitive function, not other mechanisms like reduced effort.

Externalities: Used in the context of noise pollution, which is described as an externality. An externality is a cost or benefit that is not reflected in the price of a good or service, affecting a third party not directly involved in the transaction (e.g., the cost of noise to someone living or working nearby) (Definition inferred from context, not explicitly stated in sources).

Fixed effects: Statistical terms (e.g., session, person, room, wage fixed effects) included in regressions to control for unobserved factors that are constant for a particular session, person, room, or wage condition. They help isolate the effect of the variable of interest (noise/treatment).

Flat rate: A form of compensation that provides a fixed payment per session, often calibrated to yield approximately the same total pay as piece rate conditions. Used alongside piece rates to assess willingness to pay for quiet under different incentive structures.

Hierarchical linear model: A statistical model used to estimate individual-level effects (like the impact of noise on a person’s productivity) more precisely by “partial pooling” information across individuals. It’s analogous to methods used to evaluate teacher “value added”.

Instrumental Variables (IV): A statistical technique used to estimate a causal relationship when direct measurement is difficult. In this study, the indicator for being in a “treated room” (where engine noise was added) was used as an instrument for the actual “Noise Level” to estimate its causal impact on productivity and cognitive function. A split-sample IV was used to estimate the hypothetical “return to cognitive function”.

Inverse hyperbolic sine transformation: A mathematical transformation applied to outcome variables that are skewed and have zero values (like pocket counts). It allows interpretation of coefficients similar to a log transformation for larger values while handling zeros.

Piece rate: A form of compensation where workers are paid based on their output, specifically the number of perfect pockets produced in the first experiment and points earned from cognitive tests in the second experiment. This compensation structure was used to test whether workers were aware of noise’s productive impact.

Placebo: See Effort task.

Principal component analysis (PCA): Another statistical method mentioned for aggregating test scores into an index. It estimates components that explain the most variance in the observed test scores. Used as a robustness check alongside common factor analysis.

Private information: In the theoretical context of Appendix E, this refers to information known by the worker (like their sensitivity to noise’s productive impact) but not by the firm [E262].

Randomized experiments: The research design used in the study. This involves randomly assigning participants to different conditions (e.g., exposure to noise or quiet) to estimate the causal effect of the treatment.

Randomization inference (Fisher p-values): A method for calculating p-values by simulating the randomization procedure many times and comparing the observed results to the distribution of results under simulated random assignments. Used as a robustness check for statistical significance.

Residual claimant: In the theoretical context of Appendix E, a worker paid by piece rate is the residual claimant of changes in their productivity, meaning they directly benefit from increased output [E266].

Standard deviation (σ): A measure of the variability or dispersion of a dataset. Used as a standard unit to report the size of effects on cognitive function indices.

Statistical significance: Indicated by p-values (e.g., 0.008, 0.035) or asterisks next to coefficient estimates (* for 10% significance, ** for 5%, *** for 1%) [e.g., 169]. These indicate the likelihood of observing the results if there were no true effect.

Take it or Leave it (TIOLI): A method for eliciting demand where a good is offered at a single, randomly chosen price [A193]. Used in a related study (Berkouwer and Dean, 2019) as a validation of the BDM procedure used in this paper [A193].

Willingness to pay (WTP): The maximum price a person is willing to pay for a good or service. In this study, it was elicited for the opportunity to work in quiet conditions. It was used to assess whether individuals were aware of the productive benefits of quiet by seeing if WTP was higher when compensation was based on performance.

What does this paper add to the current field of research?

Based on the provided sources, this paper makes several key contributions to the field of research:

Provides Causal Evidence on the Reduced-Form Impact of Noise on Worker Productivity in a Real-World Setting: While noise is known to be ubiquitous, especially in low and middle-income settings, and cognitive science research suggests it might reduce productivity, there was limited causal evidence demonstrating or quantifying its importance for productivity, particularly in real-work settings. This paper uses a randomized experiment in a textile training course in Kenya to estimate that a 10 dB increase in noise (perceived as twice as loud) reduces output by approximately 5%. This offers valuable quantification of the economic implications of excessive noise exposure.
Studies Cognitive Function as a Mechanism for Noise’s Impact on Productivity: The paper investigates how noise affects productivity. Decades of laboratory work have shown noise can impair cognitive function, and cognitive function is considered critical for many types of work and is correlated with success. However, prior work examining stimuli associated with poverty affecting productivity either didn’t measure cognitive function quantitatively or the stimuli affected productivity through multiple channels. This paper bridges this gap by showing that the same noise change that reduced productivity in the first experiment also impaired cognitive function (performance on a cognitive test index by 0.07 standard deviations).
Evaluates the Importance of Cognitive Function Relative to Other Inputs (like Effort): To confirm that noise primarily impacts productivity via cognitive function and not other mechanisms (like reduced effort or motor control), the study included a placebo effort task in the second experiment. This task required effort and motor control but minimal cognitive function. The results showed noise had no impact on performance on the effort task. Combined with the productivity and cognitive function results, this strongly suggests cognitive function is an important input to productivity. The paper even calculates a hypothetical “return to cognitive function” using a split-sample IV, suggesting a one-standard-deviation change in cognitive function could change productivity by approximately 79%.
Assesses Whether Individuals are Aware of Noise’s Productive Impact and Willingness to Adapt: A crucial aspect for understanding the real-world impact of environmental impediments is whether individuals notice and adapt to them. While the disutility from noise is studied, whether individuals are aware of its productive impact was largely unknown. The paper directly addresses this by eliciting participants’ willingness to pay (WTP) for quiet working conditions while randomly varying their compensation structure (piece rate vs. flat rate). The core finding is that individuals’ willingness to pay for quiet was unaffected by whether their pay depended on their performance. This suggests they are not aware that quiet conditions would increase their performance pay.
Provides Evidence Consistent with a “Failure to Notice” Mechanism: The paper explores why workers neglect noise’s productive impact. The lack of responsiveness isn’t driven by individuals least affected by noise, nor is it resolved by simply prompting them to think about the productive effects. Instead, responses are consistent with a “failure to notice” model. Individuals had some ability to predict their overall output but were unable to predict the impact of noise. They also seemed aware of their lack of understanding and were unwilling to bet money on their stated beliefs about noise’s impact. This suggests individuals may fail to attend to certain variables (like noise’s productive effect) and don’t attempt to infer their impact, even if they have some general understanding of their own ability.
Highlights the Efficiency Implications of Worker Neglect: The paper argues that worker-level neglect creates the potential for efficiency losses, even if firms are aware of the average impact of noise. If workers neglect the productive impact of noise, they won’t sort efficiently into noisy or quiet firms based on their sensitivity, undermining mechanisms like piece rates designed to achieve efficient allocation based on heterogeneous productivity responses. The study finds essentially no correlation between workers’ disutility from noise (measured by WTP under flat rate) and their estimated productive value of quiet, suggesting sorting based on annoyance wouldn’t correct for productivity losses. This highlights that understanding how workers attend to different aspects (disutility vs. productivity) of workplace characteristics is crucial for assessing efficiency implications.

In summary, the paper provides novel causal evidence on the magnitude of noise’s impact on productivity in a real work setting, pinpoints cognitive function as the primary mechanism, and demonstrates that workers neglect the productive effects of noise, suggesting potential inefficiencies and a need for policy intervention rather than relying solely on individual adaptation or market mechanisms.

What are the characteristics of the participants in the study?

Based on the sources provided, the participants in this study have the following characteristics:

28.7% reported being more annoyed by noise than others.

Recruitment Source and Background: Participants for both experiments were recruited from groups of manual laborers who gather at the gates of nearby textile factories hoping to be hired for a day’s work outside of Nairobi, Kenya. This population is accustomed to significant levels of noise, often working in large, noisy factories or being exposed to frequent road noise. They were recruited to participate in a textile training course, suggesting they were interested in or already involved in textile production or similar manual labor. The choice of this population was partly because they could potentially use the skills learned for future employment, helping the experience approximate typical working conditions. The sample is also noted as being demographically similar to many poor communities where the importance of cognitive function is of interest.

Number of Participants: The first experiment involved 128 participants. The second experiment used 213 individuals from the same population.

Demographic Profile (Combined Sample): While there are slight differences between the samples for Experiment 1 and Experiment 2, they are noted as being “relatively similar on demographic terms”. Across both experiments (a total of 341 participants), the sample had:

A mean age of 27.11 years.

56.6% were female.

62.5% had completed high school or more education.

They reported a mean typical daily wage of 597.0 Ksh.

They reported working a mean of 1.592 days in the previous week.

What are the key implications for teachers in the classroom?

Based on the sources provided and our conversation, while the paper primarily focuses on the impact of noise on worker productivity in a textile training course, its findings regarding the link between noise, cognitive function, and individual awareness have significant implications for teachers and the classroom environment, particularly in settings with high ambient noise.

Here are the key implications for teachers and the classroom:

Noise Impairs Student Cognitive Function, Hindering Learning: The paper demonstrates that increased noise impairs cognitive function, specifically task management abilities like attention and working memory. These cognitive skills are also considered critical for success in school. Just as workers’ productivity decreases due to impaired cognitive function in noise, it is strongly implied that students’ ability to focus, process information, and learn in a classroom will be hindered by noise, which directly taxes these same cognitive resources. The paper mentions that further research is needed on other noise sources, explicitly including “your child overhearing your neighbor’s television while trying to study,” connecting the research to a learning context.
The Impact is Likely Due to Impaired Cognitive Processing, Not Reduced Effort: The study specifically ruled out reduced effort or motor control as the mechanism for productivity loss by using a placebo effort task unaffected by noise. This suggests that noise isn’t simply making students “try less,” but is rather making it cognitively harder for them to perform learning tasks, regardless of their effort level.
Students May Be Unaware of Noise’s Negative Impact on Their Learning: A central finding is that workers neglected the productive impact of noise and were unwilling to pay more for quiet working conditions even when their pay depended on performance. This was consistent with a “failure to notice” model, where individuals couldn’t accurately predict the impact of noise on their output. This suggests that students might similarly be unaware of how much noise is affecting their cognitive function and academic performance. They might not attribute difficulties in focusing or learning to noise, making them unlikely to advocate for quieter conditions or adapt their study habits effectively.
The Responsibility for Addressing Noise Falls More Heavily on Educators and Institutions: Because students may not notice the productive impact of noise on their learning and therefore won’t take actions to mitigate it themselves, teachers and school administrators need to take the lead in recognizing and addressing noise as an impediment to learning. This could involve classroom management strategies to reduce internal noise, addressing external noise sources where possible, or potentially educating students on the importance of quiet for cognitive tasks like studying. The paper’s broader recommendation for policy intervention over individual adaptation also extends to educational settings, suggesting systemic approaches may be necessary.
Noise Levels Common in Many Environments Can Have Measurable Effects: The study found that a 10 dB increase in noise, equivalent to going from a dishwasher to a vacuum cleaner, significantly reduced productivity and impaired cognitive function. Noise levels in urban and industrial areas, including potentially near schools, can reach much higher, even dangerous, levels. This highlights that even moderate, common levels of noise can have a detrimental effect.

In essence, the paper suggests that noise is an underappreciated “environmental cognitive impediment” that likely affects students’ ability to learn in the classroom, and crucially, students may not even realize it is happening or understand its impact. This places the onus on educators and educational institutions to proactively create quieter learning environments.

Why might teachers exercise caution before applying these findings in their classroom?

Based on the sources and our conversation, while the paper provides valuable insights into the effects of noise, teachers should exercise caution before directly applying these specific findings in their classrooms for several reasons:

Different Tasks and Cognitive Demands: The study’s primary measure of productivity was sewing pockets in a textile training course. This task requires specific cognitive functions like sustained attention and fine motor control guided by cognitive processing. Classroom learning involves a much wider range of tasks (reading, writing, listening, problem-solving, discussion) that may rely on different combinations of cognitive skills, and their susceptibility to noise might vary [Implied by 44, 45, 65]. The paper notes that the effects might differ for “a less cognitively demanding task”.
Different Participant Population: The study participants were adult manual laborers accustomed to significant levels of noise, often working in factories or near busy roads. Students, depending on age and background, may have different levels of cognitive development, prior noise exposure, and adaptation to noise. The findings from this specific demographic might not directly translate to a student population.
Different Noise Characteristics: The study generated engine noise designed to mimic traffic and industrial noise – primarily non-informational sounds. Classroom environments often involve noise with informational content, such as chatter from other students or external sounds like construction. The paper acknowledges that “different sources of noise pollution vary in predictability and informational content” and explicitly states that “Further research is needed to understand the effects of other common sources, such as your child overhearing your neighbor’s television while trying to study”, highlighting that classroom noise is a different context.
Specific Cognitive Mechanisms Studied: The paper focused on noise impairing “task management skills like attention and working memory”. While these are important for learning, the study didn’t delve into the effects of noise on all aspects of cognitive function relevant to learning, such as long-term memory encoding, comprehension of complex information, or abstract reasoning [Implied by 10, 81].
Study Duration and Outcome Focus: The main productivity experiment occurred over two weeks, and the cognitive experiment over two days, with a focus on immediate task performance and contemporaneous effects. Learning is a long-term process of acquiring and retaining knowledge and skills. The paper’s findings on short-term task output and cognitive test performance may not fully capture the impact of noise on cumulative learning and academic achievement over months or years [Implied by 51, 70, 71].
Ethical Noise Level Constraints: For ethical reasons, the noise levels in the experiment were kept below the hazardous levels common in some workplaces, staying below 80 dB(A) for extended periods. While the tested levels were significant (dishwasher vs. vacuum cleaner), actual classroom noise levels can vary widely and might fall outside the tested range or involve different patterns of exposure, making direct quantitative application challenging [Implied by 40, 41].

In summary, while the study strongly suggests that noise is an important environmental factor that can impede cognitive function and productivity – a finding highly relevant to the learning environment – the specific context, participant group, task, and noise characteristics differ from a typical classroom. These differences mean that while the principle of noise affecting learning is supported, the precise magnitude, mechanism, and types of effects found in this study may not be directly generalizable to students and academic tasks without further context-specific research.

What is a single quote that summarises the key findings from the paper?

An increase of 10 dB reduces productivity… The same noise change impairs cognitive function but not effort task performance. This illustrates how environments associated with poverty can affect economic outcomes by impairing cognitive function… Individuals’ willingness to pay does not depend on the wage structure; suggesting that they are not aware that quiet working conditions would increase their performance pay. This cautions against the ability of workers to appropriately adapt to the impacts of noise and suggests dealing with environmental cognitive impediments may require policy intervention