Systematic Review and Meta-Analysis: Do White Noise and Pink Noise Help With Attention in Attention-Deficit/Hyperactivity Disorder?

Abstract

Objective:

Public interest in the potential benefits of white/pink/brown noise for attention-deficit/hyperactivity disorder (ADHD) has recently mushroomed. White noise contains all frequencies of noise and sounds like static; pink and brown noise have more power in the lower frequencies and may sound respectively like rain or a waterfall. This meta-analysis evaluated these effects on laboratory tasks in individuals with ADHD or elevated ADHD symptoms.

Method:

Eligible studies reported on participants with diagnosis/elevated symptoms of ADHD who were assessed in a randomized trial using laboratory tasks intended to measure aspects of attention or academic work involving attention or executive function while exposed to white/pink/brown noise and compared to a low/no noise condition. Two authors independently reviewed and screened studies for eligibility. A random-effects meta-analysis was conducted, with preplanned moderator analyses of age, diagnostic status, and task type. Publication bias was evaluated. The GRADE tool was used to assess certainty of the evidence. Sensitivity analyses were conducted to evaluate robustness.

Results:

Studies of children and college age young adults with ADHD/ADHD symptoms (k=13, N=335) yielded a small but statistically significant benefit of white/pink noise on task performance (g=0.249; 95% CI=0.135,0.363; p<0.0001). No studies of brown noise were identified. Heterogeneity was minimal and moderators were non-significant; results survived sensitivity tests and no publication bias was identified. In non-ADHD comparison groups (k=11; N=335), white/pink noise had a negative effect (g=−0.212, 95% CI=−0.355,−0.069, p=0.0036).

Conclusion:

White/pink noise provides a small benefit on laboratory attention tasks for individuals with ADHD or high ADHD symptoms, but not for non-ADHD individuals. Discussion addresses theoretical implications, cautions, risks, and limitations.

Study preregistration information:

White noise for ADHD: A systematic review and meta-analysis; https://www.crd.york.ac.uk/prospero; CRD42023393992.

INTRODUCTION

Some 10% of children aged 3-17 have been identified as having attention-deficit/hyperactivity disorder (ADHD).¹ Current ADHD treatment includes medication and behavioral therapy, often in combination. While standard treatments for ADHD can be highly effective, they face the hurdles of (a) side effects,² (b) lack of compliance and discontinuation;^3,4 and (c) lack of access, due to limited health behavioral health services or limited medication supply.^5,6 Seeking to address the need for novel ADHD treatments that are safe, effective, accessible, and enable long-term compliance, over 50 drugs, device and behavioral protocols are currently being evaluated in registered clinical trials (www.clinicaltrials.gov). Furthermore, consumer interest in complementary and integrative treatments remains high.⁷

Recently, substantial public interest has focused on the potential benefits of white/pink/brown noise exposure for enhancing task performance for ADHD individuals, as represented by tens of millions of social media views on personal improvement websites and several articles in national press outlets in the past year.^8-11 “White” refers to noise that conveys no information (“random noise”) and is distinguished by a lack of patterning, with equal intensity of different frequencies (i.e., a flat power spectrum; Figure 1). Pink (1/f) and brown noise are related but with decreasing spectral power density as frequency increases; pink and brown noise differ in the rate of decrease in power as frequency increases.¹²

Figure 1: Time Series Plots of 60 Seconds of White, Pink, and Brown Noise, Respectively.

Note: To hear an example of each type of noise, see: https://sciencenotes.org/colors-of-noise-white-pink-brown-and-more/

Note: Panels represent temporal plots of one minute of white, pink, and brown noise, respectively, as generated using the tuneR package in R. The y-axis is the amplitude limit of each wave, rescaled from −1 to 1. The plots illustrate that white noise has equal intensity at all sound frequencies. Pink and brown noise have greater intensity at different (lower) frequencies. White noise is sometimes described as sounding like radio static, pink noise like heavy rain, and brown noise like ocean waves or rolling thunder.

Although the explosion of interest on social media is recent, the idea that external environmental stimulation may help with ADHD is longstanding in the literature.^13-15 Despite well-known studies on event rates,^16-18 the literature on white noise stimulation as the manipulation of choice has remained small and never summarized. In that regard as well, the relevant experimental literature on ADHD has focused on white noise, but the mathematical similarity to pink and brown noise suggests that the white noise literature is highly relevant to the public interest in pink, brown, or other random noise.

The purpose of this meta-analysis was to empirically evaluate the effect of white/pink/brown noise effects on task performance in individuals with ADHD or associated attention problems. Although randomized trials have been conducted of this question, sample sizes have been too small to be convincing as individual studies; thus, a formal empirical summary of the literature is needed.

METHOD

This meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and A MeaSurement Tool to Assess systematic Reviews (AMSTAR-2) guidelines (Supplement 1 and Supplement 2, available online). The protocol for the review was written prospectively and indexed in the PROSPERO database before the review began (CRD2023393992).

Search strategy

The literature search included the PubMed/MEDLINE, PsychInfo, Scopus, Cochrane Central Register of Controlled Trials, and Google Scholar databases. The PubMed search strategy included terms and MeSH headings related to attention, ADHD, and white/pink/brown noise and was modified for use in other databases (Supplement 3, available online). The reference lists of all included studies were searched for other relevant studies, and Scopus was used to citation-search each included study. Preprint servers including PsyArXiv and the Open Science Framework were searched for registered protocols. ClinicalTrials.gov was searched for research in progress, and study authors contacted to ask about study status and published results. The Cochrane Database of Systematic Reviews and the PROSPERO database were searched for relevant reviews, completed or in progress. Only studies available in the English language were included due to lack of resources for translation. For grey literature, Google Scholar, Google, PubMed, and PsychInfo were searched for abstracts, conference proceedings, theses, dissertations, or unpublished manuscripts. Grey literature was included if it otherwise met inclusion criteria. No limits on publication dates were applied to the literature search, and all eligible studies published through November 1, 2023, were included.

Study Selection

Inclusion criteria for the review were as follows:

1. Study population: participants must have reported a previous diagnosis of ADHD before the study, be diagnosed with ADHD as part of the study, or have elevated symptoms of ADHD at the time of the study by self-, parent-, or teacher-report questionnaire.

2. Study type: studies were required to include laboratory measures intended to require attention, executive functioning; or academic tasks presumed by authors of the study to involve attention, concentration, or executive function.

3. Intervention type: studies included interventions of white, pink, or brown noise.

4. Comparison group: studies were not required to have a non-ADHD comparison group but could, so long as an ADHD/ADHD symptom group was included. Studies limited only to typical developing individuals were excluded.

5. Blinding of condition was not required and was not attempted in any studies. All included studies reported randomized assignment or statistically examined order effects.

Study Inclusion and Data Collection

After completion of the literature search, all identified manuscripts were uploaded into the Rayyan software and de-duplicated.²⁰ Study screening was done by two authors (AB, MK), independently in duplicate. Manuscripts which met inclusion criteria for the review were retrieved in full and their complete text assessed independently by two authors (AB, MK). Any disagreements were resolved by discussion or by consulting the first author. Reasons for study exclusion were documented to report in a PRISMA flowchart. Data items extracted from each study are listed in Supplement 4, available online. If a study was eligible for inclusion, but did not include data sufficient for meta-analysis, up to three attempts were made to contact study authors for data.

Types of Task Measures

Eligible outcome measures were laboratory tests intended to require attention, memory, executive function, or academic effort. Measures are listed in Table 1 (and described in detail in Table S1, available online) and included the Go/No Go test, Spanboard test, and the Connor’s Kiddie Continuous Performance Test, as well as verbal memory and recall assessments, such as the verbal episodic memory test and sentence recall test. Measures were classified as attention/executive function/attention versus “other” by the first author for moderator analysis.

Table 1:

Characteristics of Included Studies

		ADHD group			Non-ADHD group
Study	Nation	N	% Male	Mean age	N	% Male	Mean age	Age range	Design	Measures
Allen 2018 24	Australia	28	64	9.0	--	--	--	7-14	within	Visual search task
Baijot 2016 25	Belgium	13	62	8.5	17	46	9.2	7-12	within	Go/No Go
Batho 2020 26	Canada	52	65	15.5	--	--	--	14-16	between	Reading comprehension accuracy; time spent reading; writing accuracy; writing fluency
Chen 2022 27	Taiwan	13	100	8.3	13	100	8.4	6-10	within	Digit Span task
Helps 2014 28	UK	35	69	9.4	54	24	9.6	8-10	within	Spanboard task; Go/No-Go task; verbal episodic memory task; verbal recognition task
Kaundart 2016 29a	US	38	30b	NR	60	30b	NR	NR	within	Sentence recall test
Lin 2022 30	Taiwan	52	67	5.5	52	67	5.4	4-6	within	Conners Kiddie Continuous Performance Test-2nd edition
Metin 2016 31	Belgium	25	88	10.2	28	57	10.5	8-12	within	Standard choice delay task; adjusting choice delay task
Roye 2017 32a	US	17	29	20.8	18	33	18.7	NR	within	Antisaccade task; operation span task; symmetry span task; Rey auditory verbal learning task
Söderlund 2007 33	Sweden	21	100	11.2	21	100	11.2	9-13	within	Self-performed mini-task; verbal task
Söderlund 2010 34	Norway	10	50	11.7	41	49	11.7	11-12	within	To-be-remembered task
Söderlund 2012 35a	Sweden	11	64b	8.3b	11	64b	8.3b	7-10	within	Spanboard
Söderlund 2016 36	Sweden	20	80	12.9	20	55	13.9	NR	within	Verbal episodic memory task; Spanboard

Note: ADHD= attention-deficit/hyperactivity disorder. NR= not reported. UK= United Kingdom. US= United States.

^aNot peer reviewed.

^bData reported for combined ADHD and comparison group only.

Strategy for Data Aynthesis

Data extracted from each study included means, standard deviations (SD), test statistics (i.e., t-tests), measures of effect (i.e., d), samples sizes, standard error (SE), and p-values as available. The effects of noise interventions on ADHD symptoms or cognitive outcomes were synthesized and expressed as the bias-corrected standardized difference called Hedges’ g. Of most importance was the within-subject effect of white/pink/brown noise versus other noise conditions on task performance in individuals with ADHD or attention problems. Because pre-post correlations were universally unavailable, they were conservatively set at r =0.50 for primary analyses (the default for the meta-analytic software used).

Data Analysis

The R software and the software Comprehensive Meta-Analysis(CMA)^® v. 2.2.064 were used for the meta-analyses.^{21, 22} A random effects meta-analytic model was selected a priori due to uncertainty about whether population effect size would be the same across this group of studies that involved different countries and task combinations.²¹ When multiple task outcomes were included in the same report, those effects were averaged using published methods.²¹ That is, we averaged the effect sizes for the different tasks and computed the variance with the formula for the variance of a composite of correlated scores (reference #21, p. 228). Briefly, effect sizes within a study are denoted as Y₁, Y², etc. In the simple case of 2 effects, the mean is Y=1/2 (y₁+y₂). The variance of the mean is V_Y=1/4 (V_Y1+V_Y2+2r√V_Y1√V_Y2) where r is the correlation between the outcomes. This formula is extended for as many effects as are included. Effect sizes were then computed based on available information obtained from publications or from authors, typically means and standard deviations, or else test parameters and sample size. Heterogeneity was evaluated with the I² statistic.

Moderator Analyses

Studies were divided into preplanned subgroups for meta-analysis by age (preschoolers/children/adults; age was also treated as a continuous measure in meta-regression), diagnostic status (formal versus no formal diagnosis of ADHD; ADHD/high attention problems versus typically developing), and type of outcome measure used (attention/executive function versus other types of tasks with lower attentional demand). Multiple post-hoc sensitivity analyses were undertaken to evaluate robustness of results.

Publication Bias

I² was used as a measure of heterogeneity and publication bias was assessed using funnel plots and Duval and Tweedie’s trim and fill method.²³

Certainty Assessment (GRADE)

The Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) tool was used to assess the quality of evidence for each meta-analyzed outcome (GradePRO Guidance Development Tool; www.gradepro.org). Evidence from randomized, controlled trials are assigned a rating of “high quality,” yet downgraded for serious concerns about risk of bias, imprecision, indirectness, inconsistency, or publication bias.

RESULTS

Study Selection

The screening identified 753 records through literature searches and hand-searching the reference lists of primary studies, with 601 unique studies remaining after duplicates were removed. After title and abstract screening in Rayyan, 578 studies were excluded. Twenty-three full-text articles were assessed for inclusion and 10 excluded with reasons noted in the PRISMA diagram (Figure 2) and Table S2, available online. Thirteen articles on white (k=12) and pink noise (k=1) were retained in the meta-analysis; no studies on brown noise were identified.^24-36 A 14^th study that otherwise met inclusion criteria only used a non-attentional outcome which assessed hearing thresholds in decibels and yielded an extreme outlier effect size and was therefore excluded.³⁷ Eleven studies included pediatric participants (one with preschoolers), and two studies were in adult college students, both unpublished master’s theses (Table 1). Characteristics of studies excluded after reading full text are noted in Table S2; demographic details of participants in included studies are provided in Table S3; funding sources of included studies are listed in Table S4, available online. Of the 11 studies in children, ten were published in peer-reviewed journals and one was published in a non-peer-reviewed book chapter. Nine studies included participants with formal diagnoses of ADHD, while four featured individuals without formal diagnoses but with elevated symptoms of ADHD per parent-, teacher- or self-report. Eight studies included at least some participants who were currently medicated for ADHD but withheld medication the day of testing; the remaining three studies included some or all participants who were medicated during the study visit.

Figure 2: PRISMA) Diagram.

Primary Analyses of ADHD/High Attention Problem Groups

Main Results.

The primary meta-analysis of children/adults with ADHD or with elevated ADHD symptoms included k=13 studies with N=335 participants and 41 effect sizes in a random-effects model (Figure 3). It yielded a statistically significant non-zero overall effect (g=0.249; 95% CI=0.135,0.363; p<0.0001). Heterogeneity across studies was trivial (Q₍₁₂₎=5.713; I²<0.01; p=0.93). Table 2 and Figure 3 summarize the results. Although experts caution against choosing a fixed versus random effects model based on post-hoc heterogeneity findings,²¹ for completeness and in view of the near-complete lack of between-study heterogeneity, we note that the fixed effect model would yield essentially identical results (g=0.249, SE=0.058). As expected based on the low heterogeneity, none of the planned moderator analyses were statistically significant (Supplement 5, available online). For the cross-task comparison, tasks did not easily fall into task types. For interested readers, Table S1 provides a more detailed description of each task in each study, and Table S5, available online, provides results for each individual task within each study.

Figure 3: Forest Plot of Primary Model Results for Individuals With Attention-Deficit/Hyperactivity Disorder (ADHD) or Elevated Attention Problems.

Note: Size of square proportional to study N/weight; lines represent 95% CI. Final row diamond encompasses 95% CI.

Table 2:

Results of Primary Meta-Analysis of ADHD

Study 1st author, year	Design	N	Hedges' g	SE	95% CI		z-value	p
					Lower	Upper
Allen 2018 24	Within	28	0.030	0.189	−0.340	0.400	0.160	0.873
Baijot 2016 25	Within	13	0.062	0.260	−0.447	0.572	0.239	0.811
Batho 2020 26	Between	52	0.013	0.340	−0.653	0.679	0.039	0.969
Chen 2022 27	Within	13	0.546	0.281	−0.005	1.096	1.944	0.052
Helps 2014 28	Within	35	0.301	0.170	−0.032	0.635	1.770	0.077
Kaundart 2016 29	Within	38	0.295	0.141	0.018	0.571	2.091	0.037
Lin 2022 30	Within	52	0.273	0.162	−0.044	0.591	1.688	0.092
Metin 2016 31	Within	25	0.073	0.194	−0.308	0.453	0.374	0.708
Roye 2017 32	Within	17	0.230	0.236	−0.231	0.692	0.977	0.328
Söderlund 2007 33	Within	21	0.463	0.230	0.013	0.913	2.016	0.044
Söderlund 2010 34	Within	10	0.244	0.215	−0.177	0.664	1.135	0.256
Söderlund 2012 35	Within	11	0.532	0.313	−0.081	1.145	1.702	0.089
Söderlund 2016 36	Within	20	0.284	0.285	−0.274	0.842	0.997	0.319
Overall		335	0.249	0.058	0.135	0.363	4.282	<0.0001

Sensitivity Analyses

Sensitivity analyses suggested results were robust. Effects were nearly identical in peer-reviewed (k=10, g=0.243) and non-peer-reviewed studies (k=3; g=0.249; Q₍₁₎=0.002, p=0.96). Leaving out any one study yielded an effect size range from 0.234 to 0.272, all statistically significant (Table S6, available online). Varying the cross-condition correlation from 0.00 to 0.90 did not meaningfully affect results (Table S7, available online). Six of the studies involved the same first or senior author (G. Söderlund).^{25, 28,33-36} Excluding these six studies did not affect results (remaining k=7, g=0.228; 95% CI=0.097, 0.359; nor did excluding studies with G. Söderlund as first author (remaining k=9, g=0.217, SE=0.065, p=0.0008). Comparing the three studies with participants who were medicated during testing to the remainder who were not medicated yielded a seemingly notable difference in effect size (qualitatively larger for unmedicated groups) but the number of studies yielded too little power to detect a statistically non-zero difference (Table S8, available online).

Reporting Bias.

Funnel plots and the trim-and-fill method indicated a lack of publication bias with no studies imputed and an identical effect size (Figure S1, available online).

Certainty of Evidence.

Using the GRADE criteria, the confidence in the effect size of white/pink noise to improve attention or executive function in individuals with ADHD/symptoms of ADHD was moderate (Table S9, available online). The quality of the evidence was downgraded for possible risk of bias largely due to the inability to blind participants and study staff to the intervention, as well as incomplete outcome reporting. Because many studies reported results without the necessary information to compute a within-subject effect of white noise on ADHD symptoms, authors were contacted and generously provided additional data for us to compute effect sizes. However, this meant that covariates were not able to be considered in effect size calculation (Table S10, available online).

Results for Non-ADHD Individuals

Most studies included a non-ADHD or “normal attention” group. We therefore examined their results as well (full results in Table S11 and Figure S2, available online). The random-effects meta-analysis (k=11, N=335) yielded a statistically significant negative effect of white/pink noise on task performance (g=−0.206; SE=0.063; 95% CI: −0.330, −0.081; p=0.001) again with minimal heterogeneity (Q₍₁₀₎=12.4, p=0.026) and no significant moderator effects. As is evident from the point estimates, standard errors, and confidence intervals, the negative effect in the non-ADHD/low attention problems group is statistically different than positive effect in the ADHD/high attention problems group (Q₍₁₎=28.0, p<0.0001; for sensitivity analyses for non-ADHD individuals, see the relevant supplemental information cited earlier, available online). Funnel plots indicated minimal publication bias.

DISCUSSION

To our knowledge, this is the first formal meta-analysis of studies evaluating the effect of white/pink noise on laboratory-assessed cognitive performance in individuals with ADHD. The results provide empirical evidence that white/pink noise (one study of pink noise³⁰) modestly improves task performance in children, adolescents, and young adults with ADHD or elevated attentional problems. Effects of other types of colored noise, including brown noise, in ADHD have not yet been empirically assessed. The effect size was smaller than medication effects (average 0.7 or above), but similar to many other complementary and integrative interventions for ADHD in the 0.2 to 0.3 range.^7,38

These findings may be useful for individuals and clinicians considering use of white/pink noise to improve attentional focus for individuals with ADHD or with high levels of ADHD symptoms or attention problems, with the caveat (as expanded on below) that additional high-quality studies with adequate blinding and full reporting of data are needed before firm conclusions can be drawn. Brown noise has not been studied, although conceptually white noise and brown noise operate at the same power frequency such that the effects seen here may well extend to brown noise. However, studies to evaluate individual differences in benefit from white versus pink versus brown noise would be helpful.

Before clinical guidelines can be adopted, however, several crucial considerations pertain: (a) studies lacked adequate control conditions in many instances and blinding is difficult for this intervention, so a Hawthorne effect cannot be ruled out; (b) limited data reporting prevented consideration of relevant covariates in many cases; (c) although the task batteries employed in some studies could cause fatigue effects, perhaps leading to under-estimates of potential benefit, randomization and lack of order effects in the studies provides some assurance against that possibility; (d) the range of populations studied was limited leaving generalizability in question. In fact, regarding this last point, information on racial, ethnic, and other sample characteristics was generally lacking. Results should be considered in light of the clear need for better general understanding of the way ADHD may implicate different mechanisms and/or different clinical needs across different ethnic, racial, and cultural groups and populations.^{39, 40}

Crucially, only brief interventions were employed (several minutes) and at variable decibel levels. The optimal decibel level (or, perhaps more importantly, the minimally effective decibel level) for regular use remains unclear. This is crucial because sustained exposure to even moderate decibel levels can harm hearing or hearing development,⁴¹ and because use of headphones for entertainment at damaging volumes among young people is a widespread problem.⁴² Caution and clarity are both needed regarding decibel level and safety of sustained use and adequate breaks.

Finally, the small number of studies precluded a powerful test of whether effects are larger with more severe ADHD symptoms. A recent large study suggested that cognitive and other potentially mechanistic correlates of ADHD are more robust with a more stringent definition of ADHD.⁴³ Because effects were negative in non-ADHD individuals, it will be critical to determine the clinical indicators for likely white/pink/brown noise benefits. Thus, further studies to ascertain effect sizes, clinical best practices, safety limitations, and age-appropriate levels of use would be useful.

Indeed, the intensive public interest and the data here, representing, a potential proof-of-concept, should mandate greater interest from funding agencies and scientists in this intervention to further clarify its utility and obtain the information necessary for clinical guidelines. Further demonstration of clinical utility and proper safety guidelines could lead to rapid development of clinical recommendations given that noise intervention would adhere to the SECS (Safe, Easy, Cheap, and Sensible) rule of clinical adoption:⁴⁴ white/pink/brown noise is relatively safe, easy to implement, accessible, can be done inexpensively at home, does not preclude other interventions, and could be used in conjunction with other interventions (e.g. medications) with low concern for side effects.

Theoretical Implications

Because of the wide range of tasks used, despite variability of results across tasks within studies, data were insufficient to confirm which kinds of tasks benefit or to confirm a particular mechanism of effect. Clarifying tasks that do and do not benefit and defining mechanism would be very useful both for clinical guidance and for theory.

In regard to theory of mechanism, these findings are broadly consistent with a hypo-arousal or optimal-arousal theory of ADHD.¹³ Indeed, the hypothesis that attention and cognition (and performance in general) benefit from optimal arousal has been discussed in psychology for well over a century.⁴⁵ These including proposals for both over-arousal and over-response to stimuli⁴⁶ and under-arousal theories of extraversion and dysregulatory psychopathology.^{14,15, 47, 48} Cortical under-arousal¹³ in ADHD itself has been investigated using task-based manipulations of arousal, including faster versus slower inter-stimulus intervals;^16,49 computationally-derived cognitive parameters (e.g., d’ or the drift rate parameter from the diffusion model^16,50); and electroencephalogram (EEG) measures of cortical arousal via comparison of the ratio of slow to fast wave cortical activity, albeit with somewhat mixed support.^50-54 Because arousal operates in an inverted U-shaped pattern (too much or too little degrades performance), patterns of response to arousal interventions should differ between ADHD and typically-developing groups.

Typically-developing individuals who are already functioning at an optimal level of arousal should perform worse in the context of noise because increasing arousal pushes them past their optimal state— consistent with results observed here.

Next Steps

Overall, although not demonstrative of hypo-arousal theory of ADHD, these findings are consistent with theory and should encourage more well-developed mechanistic studies, including studies of neural activation patterns.⁵⁵ Söderlund and colleagues’ recent extension of arousal theory specifically hypothesizes disrupted stochastic resonance (a process by which random noise can amplify signal in complex systems) as being the mechanisms via which arousal dysregulation may contribute to cognitive impairments. They called their approach the moderate brain arousal theory of ADHD.³³ Because such noise is hypothesized to facilitate stimulus detection, enhance learning, and facilitate environmental adaptation,^56-58 it may be speculated that the addition of external noise in the environment may modify neural firing in a way that facilitates cognitive performance for these individuals.

As already noted, the range of tasks across studies makes it difficult to ascertain which cognitive functions are potentially being supported by white noise or whether the mechanism of action is actually arousal regulation, as well as whether the mechanism is via stochastic resonance, specifically. Studies comparing white or brown noise to other environmental stimulation (e.g., brighter light, other kinds of noise) are still needed to strengthen theoretical inference and clinical guidance. Larger studies focused on testing the range of effect across task type and with varying types of auditory stimulation (including other colors of noise but also non-random pure tones that could directly address theories of stochastic resonance) will be necessary to properly test theories mechanistically. As well, additional studies are needed to confirm whether effects a result of uniform response or heterogenous within-group responses and potential variation in response at different points in development.⁵⁹ Arousal dysregulation has been identified in ADHD from childhood into adulthood,^60,61 but the vast majority of studies have focused on children, leaving developmental effects uncertain.

This review focused on task performance; we did not find sufficient studies for a meta-analysis concerning effects on behavior or ADHD symptoms in general. Thus, studies of real-world changes in symptoms or impairment (modeled on clinical trials over time) would be helpful. Further, the literature reviewed here did not enable us to evaluate how white/pink noise may provide benefit or interference for ADHD or non-ADHD youth under different perceptual or task contexts or under different white/pink noise conditions per se. (For more discussion, see ^{62, 63}).

In conclusion, white and pink noise provide a small average benefit on laboratory tasks for children and college age adults with ADHD or else high ADHD symptoms. In the studies selected here, white/pink noise did benefit for individuals without ADHD or with average levels of inattention symptoms, and even have small negative effects. Findings appear consistent with an optimal arousal theory of ADHD but more sophisticated mechanistic studies are needed to test that theory further. Clinical guidelines would require more data before clear guidance could be provided, yet such data should be attainable. Cautions and limitations include need for additional high quality evidence, modest effect size, lack of data on sustained use, insufficient data on variation in sex or race/ethnicity, and risk of hearing damage if noise levels are excessive over a period of time.