Auditory Discomfort and Visual Sensitivity

Abstract

The field of visual discomfort is fast growing, identifying parameters of visual stimuli that evoke discomfort, their effect on the visual system, and their heightened impact in clinical populations. In contrast, there is little work on auditory discomfort. First, we reviewed the current findings in the field of auditory discomfort and report a preliminary study investigating the effects of uncomfortable sounds on near infrared spectroscopy (NIRS) responses from auditory cortex. Nine participants listened to pure 1kHz tones that were modulated by a 2Hz or a 16Hz sinusoidal wave that made the tone sound as though it wobbled. The oxyhaemoglobin response to the uncomfortable 16Hz sound was larger and slower compared to the relatively more comfortable 2Hz sound. This suggests cortical excitability to uncomfortable sounds, similar to the heightened response reported in the visual modality. Second, we assessed the relationship between visual and auditory sensitivity in a large sample (N=669) of participants who completed the Pattern Glare Test (PGT), the Amsterdam Misophonia Scale, the Hyperacusis Questionnaire, and the Functional Hearing Questionnaire (FHQ). While all measures of auditory sensitivity were significantly correlated with the number of illusions seen in the PGT (.1>r_s<.3), the relationship with misophonia was significantly weaker than other comparisons. This highlights that while those who are visually sensitive are more likely to exhibit increased auditory sensitivity, the measures that focus on pain (PGT, hyperacusis, FHQ) may be dissociable from those that rely on emotive discomfort (misophonia). Together, this suggests similar mechanisms underlying auditory and visual discomfort.

Part 1: Individual Differences in Auditory Sensitivity

It is becoming increasingly known that some individuals experience increased sensitivity to sensory information. For example, lying down in a dark quiet room during a migraine attack to reduce the pain from light and noise, or an individual with photosensitive epilepsy avoiding flickering lights to prevent a seizure (Fisher et al., 2005). In fact, sensory sensitivities are being recognized as becoming a prevalent symptom in many clinical diagnoses including autism (Danesh et al., 2021; Simmons et al., 2009), schizophrenia (Haigh et al., 2016; Javitt & Sweet, 2015), traumatic brain injury (Callahan et al., 2018; Thielen et al., 2023), and stroke (Thielen et al., 2023). However, what is less known is that sensory sensitivities can range from mild to severe in the general population in the absence of a clinical condition. These symptoms or experiences can manifest as differences between individuals in how uncomfortable the same stimulus is perceived to be.

As with visual discomfort, ways of categorizing auditory sensitivity have appeared that encapsulate different types of auditory discomfort. There are three common types of auditory sensitivity: phonophobia, hyperacusis, and misophonia. Phonophobia refers to a fear or aversion from sounds that are otherwise innocuous and is often reported as a clinical symptom. Hyperacusis is another type of auditory sensitivity that is defined by feelings of pain from otherwise innocuous sounds, or that the sounds are perceived as being louder than they are. Because of this definition, hyperacusis and phonophobia are commonly confused (Asha’ari et al., 2010). Misophonia is another type of auditory sensitivity. It is defined as feelings of anger or being annoyed at specific sounds such as finger tapping, fidgeting, or chewing, sometimes by a specific person rather than as a general sensitivity to all chewing sounds (Swedo et al., 2022). The emotional component in misophonia is essential in distinguishing it from hyperacusis.

Certain clinical conditions are more at-risk for developing auditory sensitivities and this tends to be labelled as phonophobia. For example, the diagnosis of migraine includes phonophobia as a key symptom (IHS, 2018) and individuals with migraine report lower discomfort thresholds to increasing volume at a range of frequencies (100–12,000Hz (Mourgela et al., 2023)). Individuals who experience pattern glare (report seeing visual illusions in striped patterns) report greater discomfort to sounds including frequency and amplitude modulated sweeps (Haigh et al., 2022). Both migraine and pattern glare are typically linked to cortical hyper-excitability and the discomfort is thought to reflect a homeostatic mechanism to avoid stimuli that cause heightened cortical activity (Wilkins et al., 2021). Individuals with Williams syndrome have lower Loudness Discomfort Levels and increased likelihood of phonophobia and hyperacusis compared to non-clinical individuals (Silva et al., 2021). Again, this has been linked to cortical hyper-excitability (Zarchi et al., 2015). Individuals at their first-episode of psychosis rate sounds such as frequency sweeps (increasing from 50–8000Hz) modulated by a sine wave as being more uncomfortable to listen to than non-clinical individuals (de Bustamante Simas, da Silva, et al., 2022; de Bustamante Simas, dos Santos, et al., 2022). However, songs played in reverse were most uncomfortable to individuals with panic disorder, whereas first-episode psychosis and non-clinical individuals were relatively unaffected (de Bustamante Simas, da Silva, et al., 2022). Panic disorder is often characterized by a hyper-responsive limbic system to sounds (Brinkmann et al., 2017) highlighting an overlap between painful and emotive sounds that may need to be distinguished to better isolate underlying mechanisms. Psychosis, on the other hand, is typically linked to cortical hypo-excitability (Randeniya et al., 2018) although patients often report subjective experiences of sensory flooding suggesting being over-whelmed by the sensory environment (Bunney et al., 1999; McGhie & Chapman, 1969; Torrens et al., 2023). Together, this suggests a link between perturbed sensory processing and auditory discomfort. Clinically sensitivity in the auditory domain tends tend to be referred to as phonophobia, and because they are linked to a clinical condition, phonophobia is not often examined in isolation. This makes phonophobia prevalence difficult to estimate.

While individual differences in auditory sensitivity can to some extent be explained by a psychiatric or neurological condition, increased sensitivity can occur in the general/non-clinical population. Misophonia appears to impact around between 12% (individuals living in Ankara, Turkey; (Kılıç et al., 2021)) to around 20% (undergraduate students in the United States and China; (Wu et al., 2014; Zhou et al., 2017); in Iran; (Yektatalab et al., 2022), although ‘severe’ misophonia only impacts 6% of the population (Jakubovski et al., 2022; Zhou et al., 2017).

Hyperacusis appears to be less prevalent in the general population than misophonia (0.2–17.2%; (Båsjö et al., 2016; Hannula et al., 2011; Ren et al., 2021)), with 17% showing severe hyperacusis (Aazh et al., 2018). However, the prevalence increases in those in certain occupations such as musicians (67%; (Lüders et al., 2016); 40% in drummers; (Halevi-Katz et al., 2015)) and teachers (67%; (Meuer & Hiller, 2015)), often co-occurring with tinnitus. This highlights potential environmental factors and hearing damage that might increase the likelihood of experiencing hyperacusis. Incidence increases further still when examining neurological conditions. Incidence in Williams syndrome ranges from 4.5% (Levitin et al., 2005) to 95% (Klein et al., 1990) and in autism ranges from 18% (Rosenhall et al., 1999) to 63% (small study of 20 participants by (Khalfa et al., 2004)) and 69% (Danesh et al., 2015). One emerging theory is that hyperacusis is linked to enhanced auditory gain (see review by (Henry et al., 2022)) and that sound therapy (the use of sound to reduce the effects of tinnitus) focused on reducing auditory gain may help to alleviate hyperacusis.

What is unclear is whether individuals who have increased auditory sensitivity are more likely to have increased visual sensitivity too. To address this, we conducted a survey from a large sample to identify the relationships between auditory sensitivity (specifically hyperacusis and misophonia) and visual sensitivity (pattern glare). If auditory and visual sensitivity are likely to occur in the same individuals then this suggests a common cross-sensory mechanism, where increased sensitivity in one modality is likely to result in increased sensitivity in another modality.

Study of Co-Occurring Auditory and Visual Sensitivity

To identify the relationship between visual sensitivity (pattern glare) and different measures of auditory sensitivity, we asked participants to complete self-report questionnaires on their auditory behaviors. Specifically, we focused on hyperacusis and misophonia, along with a functional hearing questionnaire. We then assessed the relationship between scores on these standardized questionnaires with the number of illusions they saw in the Pattern Glare test (Wilkins & Evans, 2001) as a measure of visual sensitivity. We chose to focus on the PGT as it does not directly ask about visual discomfort but captures individual differences in visual sensitivity. If sensory sensitivity impacts several sensory modalities, then we might expect that those individuals who experience many illusions in the Pattern Glare Test (high pattern glare) to also show increased measures of hyperacusis and misophonia, and for their auditory sensitivities to impact their daily lives.

Participants

845 undergraduate students from the University of Nevada, Reno completed an online Qualtrics survey about their sensory sensitivities. The survey was run between April 2022 and March 2023. Seven participants reported having a diagnosis of epilepsy or a history of seizures and were not allowed to continue with the survey. A further 169 participants reported a neurological or psychiatric condition, or were on medication(s) for anxiety, depression, or mood disorders and were excluded from analysis. The remaining 669 participants were at least 18 years old and completed the following questionnaires (mean age 20.1 years old, SD=8.0 years; 465 females, 187 males, 15 non-binary, 2 preferred not to say). A post-hoc power analysis showed that with 669 participants, a two tailed correlational analysis with 5% error rate can detect correlations as small as r_s=.108 with 80% power (Faul et al., 2007, 2009). All procedures were approved by the Institutional Review Board at the University of Nevada, Reno (#1333057) and were in accordance with the Declaration of Helsinki.

Materials and Procedure

Participants completed a misophonia questionnaire (A-MISO-S; (Schröder et al., 2013), a hyperacusis questionnaire (Khalfa et al., 2002), and a functional hearing questionnaire (Saunders et al., 2005).

The misophonia questionnaire provides an explanation of what a misophonic sound is (“For people with misophonia, everyday sounds like chewing, crunching, or breathing can cause intense negative emotions such as anger, rage, fear or disgust. The condition can have a highly detrimental impact on daily life but is often poorly understood, largely because sounds that trigger it (chewing, crunching, etc.) are easily ignored by most people.”) and asks participants to respond along a 5-point scale how much or how frequently these situations occur based on this definition of misophonia. For example, “How much distress do the misophonic sounds cause you? (In most cases, distress is equated with irritation, anger, or disgust. Only rate the emotion that seems triggered by misophonic sounds, not generalized irritation or irritation associated with other conditions.)” And participants had the following response options: “None” (scores 0), “Mild, occasional irritation/distress” (scored 1), “Moderate, disturbing irritation/anger/disgust, but still manageable” (scored 2), “Severe, very disturbing irritation/anger/disgust” (scored 3), or “Extreme, near constant and disturbing anger/disgust” (scored 4). Scores were summed across the six items for each participant and so could range from 0–24. The use of the A-MISO-S has been independently validated and there is good internal consistency between the individual items (Naylor et al., 2021). The A-MISO-S has been commended as being a brief and well-used in the literature, despite some limitations (Rosenthal et al., 2021).

The hyperacusis questionnaire asks participants to rate how much they experience a described situation, for example “Has anyone you know ever told you that you tolerate noise or certain kinds of sound badly?” The participant can choose between “Yes, a lot” (scored as a 3), “Yes quite a lot” (scored 2), “A little” (scored 1), and “No” (scored 0). The scores across 14 statements (one question was removed for similarity with the FHQ) were summed resulting in a range of scores from 0–42. The hyperacusis questionnaire has been validated and translated into numerous languages, including Turkish, Italian, Japanese, Arabic, and Persian (Altın et al., 2023; Fioretti et al., 2015; Kashani et al., 2022; Oishi et al., 2017; Shabana et al., 2011; Yilmaz et al., 2017).

The functional hearing questionnaire uses the same rating scale as the hyperacusis questionnaire and asks about general behaviors such as, “Do you ever use earplugs or earmuffs to reduce your noise perception (Do not consider the use of hearing protection during abnormally high noise exposure situations)?” The summed responses across the eight statements (one was removed for similarity with the Hyperacusis Questionnaire) were summed resulting in a range of scores from 0–24.

Responses were scored so that a higher score reflected greater misophonia, hyperacusis, or worse functional hearing ability.

Participants then completed an adjusted version of the Pattern Glare Test (Wilkins & Evans, 2001). A single achromatic square-wave grating pattern was presented on the screen for 5 seconds before being removed. As this survey was completed remotely online, we had no control over the screen participants used to complete the survey. Therefore, the spatial frequency and contrast of the pattern could not be controlled for and having comparison spatial frequency patterns would not have been useful. This method has been successfully used previously to identify individual differences in visual sensitivity (Torrens et al., 2024). In addition, subsequent analyses of Qualtrics’ resizing of the pattern on laptops, tablets, and phones demonstrated that the spatial frequency of the pattern only varied between 2.9–3.5cpd (assuming typical viewing distances (Torrens et al., 2024)). Participants were then asked if they experienced any of the following: pain/discomfort, shadowy shapes amongst the lines, shimmering of the lines, flickering, colors, blue, bending of the lines, nausea/dizziness, or unease. Participants could select as many options as they wanted, could select ‘Other’ so that they had the option to describe their own sensation, or could select ‘None’. The number of options selected was then added to provide a pattern glare score. Due to the uncontrolled nature of the online setup, we could not specify the spatial frequency of the pattern presented in the Pattern Glare Test.

The online survey took around 20 minutes to complete and participants earned course credit for taking part.

Results

The number of illusions seen in the Pattern Glare Test were correlated against the summed score from the misophonia questionnaire (Figure 1A), the summed score from the hyperacusis questionnaire (Figure 1B), and the summed score from the functional hearing questionnaire (Figure 1C). Spearman’s correlations were used as all measures were heavily positively skewed. The strength of the correlations were then compared to one another using a Fisher-Z-transformation (Lenhard & Lenhard, 2014) to identify which measures of auditory sensitivity are most highly related to visual sensitivity. Because of the multiple comparisons being conducted, Bonferroni correction was used to adjust the p-value for significance (5% / 6 tests = p<.008).

Figure 1.

Relationship between the number of illusions seen in the Pattern Glare Test and (A) total misophonia score, (B) total hyperacusis score, and (C) total score on the Functional Hearing Questionnaire (FHQ). All correlations were significant.

Spearman’s correlations showed a significant positive correlation between the number of illusions seen in the Pattern Glare Test and the summed A-MISO-A score (r_s(667)=.11, p=.003), summed Hyperacusis Questionnaire score (r_s(667)=.29, p<.001), and summed FHQ score (r_s(667)=.25, p<.001). Together, this suggests that those who experience more pattern glare also experience greater auditory sensitivity.

Interestingly, when the strength of the correlations were compared, the relationship between pattern glare and misophonia was significantly weaker than the relationship with hyperacusis (z=3.43, p<.001) and with the FHQ (z=2.65, p=.008). There was no significant difference between the strength of the correlations between pattern glare and hyperacusis and with the FHQ (z=0.79, p=.431).

Interim Discussion

These results suggest that those who are visually sensitive are also likely to experience auditory sensitivity. This is particularly the case for those who experience hyperacusis, which is commonly defined as pain from innocuous stimuli, which is more akin to the visual stress and pattern glare symptoms reported by those who are visually sensitive. Together this suggests a multisensory sensitivity rather than a modality specific abnormality.

The relationships between the measures were small, which may partly be because the PGT does not directly ask about symptoms or behaviors related to visual discomfort but utilizes a proxy measure in the number of illusions seen to an uncomfortable pattern. The misophonia, hyperacusis, and FHQ scales all focus on specific behaviors linked to auditory discomfort which may be weakening the relationships with PGT. However, using a proxy measure of visual discomfort also serves as a strength of this study as it avoids the participant matching visual and auditory-related behaviors when responding.

One limitation to note is that the spatial frequency of the pattern from the Pattern Glare Test likely varied across participants as we could not control the device used and the parameters of the screen. This variability likely increased the noise in the data and suggests that the relationships identified are underestimated and may be stronger under more controlled settings. This may be another cause for the small correlation coefficients, despite the large sample size. However, there is evidence that the perception of illusions in the Pattern Glare Test persist across a range of viewing distances and so may not affect the reports collected using Qualtrics (Monger et al., 2016). In addition, a previous assessment of the changes in spatial frequency depending on screen type (phone, tablet, laptop) showed that the pattern was still presented by Qualtrics within the mid-range frequency band (between 2.9–3.5cpd; assuming typical screen distances (Torrens et al., 2024)). The benefit of conducting a remote study is that we were able to collect data from a large sample, offsetting any potential effects of the noisy measure.

The question then arises whether alleviating the sensitivity in one modality would positively improve the sensitivity in another modality too. This could be due to a common mechanism that is driving the discomfort in all sensory modalities, and that reducing strain on the system in one modality could be enough to reduce the strain in other modalities. For example, the cortical hyper-response has been linked to failures within local inhibitory mechanisms to contain the excitation (Palmer et al., 2000; Penacchio et al., 2023). If inhibitory mechanisms are struggling to contain excitation in multiple cortical areas, then the discomfort in one modality may be linked to the discomfort in another. Alternatively, improving the sensitivity in one modality may help to make the sensitivity in other modalities more bearable and less overwhelming. For example, participants experiencing suboptimal heat temperatures report reduced tolerance to noise (Pellerin & Candas, 2003). This may not be due to a single mechanism of discomfort that is diverting resources across modalities but may reflect a decreased cognitive tolerance of discomfort. With the increasing use of tinted ophthalmic lenses to reduce visual discomfort (Wilkins & Evans, 2022), the impact of the lenses on auditory discomfort can easily be tested.

Another point to consider is that the relationship between pattern glare and hyperacusis was significantly stronger compared to misophonia. This may suggest that there are clusters of sensory sensitivity: one that features pain and discomfort and another that more prominently features an emotive response such as disgust. In the visual domain, there is a condition called trypophobia which is defined as a severe aversion to holes, for example, when looking at lotus pods (Cole & Wilkins, 2013). The aversion is believed to be linked to skin lesions and has similar emotive reactions as misophonia. If there are subcategories of sensory sensitivities that can be defined by pain or by an emotive response, then there may be (slightly) different neural mechanisms underlying the sensitivity. Better definitions of the different types of sensory sensitivity will help reveal the causes of the sensitivity and, in turn, options for relieving the sensitivity.

Another component for investigations into auditory sensitivity is to identify what kinds of auditory stimuli are uncomfortable to listen to, why this may be the case, and to investigate the mechanisms underlying the discomfort, how they are similar, or how they differ from the mechanisms associated with visual discomfort.

Part 2: What We Know About Auditory Discomfort So Far

In general, there is little known about what auditory stimuli evoke discomfort and why. One of the most consistent findings is that stimuli that are natural, for example, animal sounds and the wind through leaves, are preferred over unnatural sounds such as engines and alarms (Axelsson et al., 2010). Natural sounds have specific pitch and loudness properties, which generally predict human auditory psychophysical thresholds (Monson et al., 2013), suggesting that our auditory system has evolved to be maximally sensitive to sounds that we are likely to encounter in nature. This provides a testable hypothesis where stimuli that are unnatural are more likely to be uncomfortable to listen to because they are different from what our auditory system has evolved to process.

The difficulty with this approach is that natural stimuli are remarkably complex and not all unnatural stimuli are uncomfortable. For example, compare white noise to a siren. The siren is often uncomfortable (by design) but white noise can be calming (Capezuti et al., 2022) and yet both are artificially created and bear little resemblance to any sounds in nature. In addition, music is artificial - generated by man-made instruments to produce notes and harmonies that cannot be found in nature - and yet is often considered to be pleasant, even if the music evokes an emotion. There is some evidence that the human auditory system is more receptive to harmonious sounds compared to inharmonious sounds (Bregman, 1994; Chan et al., 2023; Itoh et al., 2003; Tramo et al., 2001) and that only some of the auditory tuning may be due to cultural exposure (Athanasopoulos et al., 2021; Lumaca et al., 2018); for a full discussion see (Peretz, 2006). There is, however, evidence that music of some form has existed for millenia, and so it is possible that the auditory system has evolved to encode music in a similar way that it has evolved to encode natural sounds. Either way, this then muddies the theory that the auditory system has a preference for natural sounds, unlike what is seen in the visual system.

Therefore, shifting the emphasis from categories of stimuli that are uncomfortable to identifying the parameters of stimuli that drive discomfort may be more fruitful. Recent efforts to identify specific parameters that drive auditory discomfort have suggested that carrier frequencies between 2.5–5.5kHz and modulation frequencies between 1–16 Hz are the most likely to evoke discomfort (Kumar et al., 2008). Furthermore, knowing the source of the sound can make the sound even more unpleasant (Reuter & Oehler, 2011) suggesting a semantic component to auditory discomfort. We recently found that discomfort increases linearly with higher carrier frequencies in amplitude modulated sounds and that higher central frequencies within frequency sweeps increases discomfort. However, the range of frequencies within the sweeps did not reliably affect discomfort (Haigh et al., 2022). In other words, the overall perceived pitch of the sound had an effect on discomfort.

We also found that increasing the modulation frequency of tones decreased discomfort and could make high carrier frequencies comparatively comfortable to listen to. For example, increasing the modulation frequency increased the discomfort for low carrier frequencies such as 0.5kHz and 1kHz, whereas modulation frequency had less of an effect for high carrier frequencies such as 4Hz and 8Hz. Finally, we found that the higher the number of frequency sweeps presented within a second also increased discomfort (Haigh et al., 2022). The increased repetition may be causing an effect similar to that of increasing visual flicker, where increasing the flicker frequency (up to 16Hz) is uncomfortable to view (Wilkins, 1995; Wilkins et al., 1989). Flicker rates faster than 16Hz tend to be less uncomfortable. It is possible that increasing the sweep repetition rate may at some point also become less uncomfortable, similar to visual flicker.

In the visual system, uncomfortable stimuli produce a hyper-response in the visual cortex, particularly in those who are visually sensitive (Coutts et al., 2012; Haigh et al., 2019; Huang et al., 2003). This suggests a homeostatic response to protect the system from working too hard to process certain (unnatural) stimuli (Wilkins et al., 2021). Currently, the impact of uncomfortable auditory stimuli on the auditory system is unknown. However, it is reasonable to predict that a similar homeostatic mechanism would exist for the auditory system, where uncomfortable sounds would produce a hyper-response in the auditory cortex. Therefore, we conducted a preliminary study to track the amplitude of the cortical response in the auditory cortices to uncomfortable and relatively more comfortable sounds.

Study of Haemodynamic Responses to an Uncomfortable Auditory Stimulus

To assess whether there is any evidence that uncomfortable auditory stimuli evoke a larger cortical response, we conducted a preliminary study measuring the oxy- and deoxyhaemoglobin response over primary auditory cortices in response to pure tones that were modulated by a 2Hz or a 16Hz sinusoidal wave. According to our recent findings, the 16Hz sound was much more uncomfortable to listen to than the 2Hz sound (Haigh et al., 2022). Therefore, we would predict that the more uncomfortable 16Hz sound would elicit a larger haemodynamic response than the 2Hz sound.

Methods

Participants

Nine participants took part (mean age 21 years old, SD=1.9 years; 6 females, 3 males) took part in a 45 minute near infrared spectroscopy (NIRS) study assessing auditory responses to amplitude modulated sounds. A post-hoc power analysis showed that with nine participants, 5% alpha and 80% power, we can detect large effect size (Cohen’s d>1; calculated using G*Power (Faul et al., 2007, 2009)). All participants were older than 18 years old, did not report having a neurological or psychiatric condition, nor were on medications for anxiety, depression, or mood disorders, and did not have epilepsy or a history of seizures. Participants gave their written consent to take part. All procedures were approved by the Institutional Review Board at the University of Nevada, Reno (#1333057) and were in accordance with the Declaration of Helsinki.

Stimuli

All stimuli were created and presented in MATLAB and presented over Etymotic ER2 insert earphones. Two types of sounds were presented. Both were 1kHz pure tones presented for 2s that were modulated in amplitude at 100% by a 2Hz or a 16Hz sinusoidal wave that made the tone sound as though it were wobbling. The same sound was presented 6 times with 0.5s in between each sound presentation for 14.5s in total. The multiple presentations were to reduce the auditory cortex from adapting to the sound whilst maintaining a long presentation block to maximize the haemodynamic response elicited. This was then followed by 25s of silence before the next block of sounds was presented. The order of 2Hz or 16Hz sounds was randomized. The long breaks in between trials were to allow the sluggish metabolic response to return to baseline before the next block began. We have used this method of stimulus presentation before to successfully assess differences in haemodynamic responses as a function of stimulus parameters (Haigh et al., 2013, 2015).

Participants were asked to focus on a grey screen that contained a black central fixation cross while the sounds were presented. During the breaks between sound blocks, the cross occasionally (25% of the time) turned white for 0.1s. Participants were asked to press the spacebar when they saw the cross turn white to ensure that they were awake during the study. If they did not respond or took longer than 2s to respond, then the cross turned red to signal to the participant that they missed the target.

Equipment and Data Analysis

An 8-channel NIRSport 2 near infrared spectroscopy system from NIRx sampling at 10.17Hz was used to measure the change in oxy- and deoxyhaemoglobin response over left and right auditory cortex. Two emitters and two detectors were placed over TP7 and TP8 in a grid. Data were recorded using Aurora from NIRx.

Data were bandpass filtered between 0.1–0.02Hz to remove slow potentials and heartbeat from the signal. Data from 10s before the onset of the block of stimuli to 35s after the onset of the first sound in the block were extracted and baseline correct (the average response 10s before the onset of the stimulus to the start of the stimulus block was subtracted from the rest of the epoch). To ensure that only responsive channels were included in the analysis, we calculated the average of the response after the start of the block and divided it by the noise in the baseline. Any epochs that did not show an increase in the oxyhaemoglobin response or a decrease in the deoxyhaemoglobin response in any of the 8 channels were removed from the analysis. This ensured that there was a response for each block from at least 2 channels.

The remaining responses were averaged together to produce a single response per participant. The average response from the start of the block (0–5s after the onset of the stimulus block) was subtracted from the average response between 10–20s after the start of the stimulus block (the peak of the response) and compared across the 2Hz and 16Hz conditions using a paired t-test.

The timing of the peak of the responses were also compared. For the oxyhaemoglobin response, the time it took for the response to peak (between 10–20s after the start of the stimulus block) for each condition (2Hz and 16Hz) was calculated for each participant. For the deoxyhaemoglobin response, the time it took for the response to reach its minimum (within the same time period) for each condition for each participant was calculated. Again, paired t-tests were used to calculate any differences between condition.

Cohen’s d effect sizes were calculated as a measure of the magnitude of the effects of the sounds on the NIRS signals. Pearson’s correlations were conducted to compare responses over oxygenated and deoxygenated signals and to compare the peak amplitude with the timing of the peak response.

Results

Oxyhaemoglobin responses were significantly larger in amplitude to the 16Hz sound (mean=0.000407μM, SD=0.0003μM) compared to the 2Hz sound (mean=0.000149μM, SD=0.0002μM; t(8)=2.41, p=.043; d=1.06; Figure 2A). There were no significant differences in the amplitude of the deoxyhaemoglobin response between the two sounds (16Hz: mean=−0.000105μM, SD=0.0009μM; 2Hz: mean = −0.000076μM, SD=0.00005μM; d=0.44; t(8)=1.34, p=.217; Figure 2B). Participants who produced large oxyhaemoglobin responses were more likely to produce large deoxyhaemoglobin responses (r(16)=−.58, p=.012; Figure 3A).

Figure 2.

(A) The oxyhaemoglobin response to the 2Hz (blue) and the 16Hz (pink) sounds. The amplitude of the oxyhaemoglobin response was significantly larger to the 16Hz sound than the 2Hz sound. (B) The deoxyhaemoglobin response to the 2Hz (blue) and the 16Hz (pink) sounds. There was no significant difference between the amplitude of the responses to the sounds. The grey bars show the timing of the baseline and the peak responses used to calculate the change in amplitude.

Figure 3.

(A) Significant relationship between oxy and deoxyhaemoglobin response amplitudes. (B) Significant relationship between timing of peak oxy and deoxyhaemoglobin responses. (C) Relationship between amplitude and timing in oxy (black) and deoxyhaemoglobin (grey) responses (neither are significant).

Oxyhaemoglobin responses also peaked significantly later to the 16Hz sound (mean=17.2s, SD=1.7s) compared to the 2Hz sound (mean=14.3s, SD=2.5s; t(8)=3.8, p=.005; d=1.36). Deoxyhaemoglobin responses showed peak differences that were approaching significant (t(8)=2.2, p=.060; d=1.20) where responses to the 16Hz sound were also slightly slower (mean=17.5s, SD=1.6s) than to the 2Hz sound (mean=15.2s, SD=2.2s). Participants who produced slower oxyhaemoglobin responses were more likely to produce slower deoxyhaemoglobin responses (r(16)=.53, p=.022; Figure 2B). However, there was no significant relationship between peak amplitude of the responses and the timing of the peak response (oxy: r(16)=.17, p=.505; deoxy: r(16)=.09, p=.713; Figure 2C).

Discussion and Future Research

These preliminary NIRS data suggest that uncomfortable auditory stimuli may elicit a larger cortical response compared to sounds that are less uncomfortable. This was only evident in the oxyhaemoglobin response, which typically has the higher signal-to noise (Strangman et al., 2002). However, there were some differences in the timings of the peak oxy- and deoxyhaemoglobin responses, with the relatively more comfortable 2Hz sound peaking earlier than the uncomfortable 16Hz sound. The larger peak response amplitudes could have taken more time to peak (Coutts et al., 2012), which would explain the larger slower response to the uncomfortable 16Hz sound, but there was no significant relationship between the two measures in our sample. This suggests that the amplitude and the timing of the peak response could reflect different components of auditory discomfort.

If the larger cortical response to the 16Hz sound is related to the discomfort, then this suggests a similar homeostatic mechanism as the visual system, where the discomfort is a homeostatic response to avoid stimuli that cause the sensory cortex to work hard (Haigh et al., 2013; Wilkins et al., 2021). However, some of the key properties of auditory stimuli that increase discomfort, for example, pitch, are processed much earlier in the auditory system, at the level of the cochlear and the brainstem, compared to the visual system’s encoding of uncomfortable stripes or colors, which is mostly post-retinal. Therefore, assessing the effects of discomfort on the whole auditory pathway is essential for accurate estimates of which system is first impacted by uncomfortable stimuli.

This is one of the first studies known to focus on the effect of uncomfortable sounds on auditory functioning. Interestingly, there seems to be the opposite effect of musical sounds on auditory functioning – consonant/pleasant sounds elicit larger responses (N100 event-related potentials) from primary auditory cortex than dissonant/unpleasant sounds, when measured using electroencephalography (EEG)(Itoh et al., 2003). We have chosen to focus on non-musical stimuli so far when exploring auditory discomfort, but this discrepancy in characterizing early auditory responses depending on whether the sounds are musical or not emphasizes the need to unravel the effects of musical preference from discomfort on auditory functioning.

When reflecting on the individual differences in auditory sensitivity, if auditory discomfort is driven by an over-response within the auditory pathway, as demonstrated by our preliminary NIRS data, then the heightened neural response should be even larger in those who are auditory sensitive, for example, those who experience hyperacusis. This is seen in visual responses to uncomfortable images in individuals who are visually sensitive. For example, individuals with migraine have larger metabolic responses in visual cortex compared to control participants, and the hyper-response is reduced with the use of tinted ophthalmic lenses that reduce the discomfort (Coutts et al., 2012; Huang et al., 2003, 2011). In our recent behavioral study of auditory sensitivity, individuals with high pattern glare (typically considered to be a visual sensitivity) reported greater auditory discomfort, and so if individuals with high pattern glare/greater visual sensitivity show even larger neural responses to uncomfortable auditory stimuli than those with low sensitivity, then this further supports the theory that the neural hyper-response could be a domain-general mechanism to avoid uncomfortable sensory stimuli.

In terms of what might be driving the hyper-response, computational modeling of the neural response to visual stimuli suggests that the system uses sparse encoding to perform optimally. Uncomfortable visual stimuli cannot be encoded sparsely, increasing the neural activity and potentially producing feelings of discomfort as a homeostatic mechanism that avoids metabolically expensive stimulation (Hibbard & O’Hare, 2015; Penacchio et al., 2023). It is possible that uncomfortable auditory stimuli produce the same problems and cannot be sparsely encoded. This assumes that the discomfort is driven by cortical mechanisms which are able to employ these computations. However, the sensitivity to pitch specifically could indicate that earlier parts of the auditory pathway such as the cochlear or brainstem are responsible. Assessment of musical perception of “roughness” and consonance/dissonance has pointed to activity in the auditory nerve as reflecting the perception of pleasantness in music (for a summary, see (Tramo et al., 2001)). However, modelling of auditory processing in early auditory areas, including the brainstem, fit sparse encoding models (deep learning model (Zhang et al., 2019); inferior colliculus in bats (González-Palomares et al., 2021); and in cats (Chen et al., 2012)), suggesting that uncomfortable sounds could also be disrupting optimal encoding. Either way, identifying the mechanisms underlying auditory discomfort could help inform the mechanisms that drive visual discomfort.

It is important to consider other variables that are known to affect sensory discomfort and may reveal key mechanisms that could drive an increase in domain-general cortical excitability. While our studies excluded participants with a psychiatric or neurological diagnosis, subclinical traits of anxiety, depression, schizotypy (schizophrenia), and autism are known to vary in the non-clinical population and these traits have been independently linked to sensory perturbations (for example, see (Green & Ben-Sasson, 2010; Hui et al., 2023; Liss et al., 2006; Torrens et al., 2024). As mentioned previously, types of auditory discomfort such as hyperacusis and misophonia are commonly reported symptoms in a number of neurological and psychiatric conditions (Henry et al., 2022; Siepsiak et al., 2022) and subclinical depression, obsessive compulsive disorder, and stress have been linked to misophonia and hyperacusis in the general population (Almadani et al., 2024; Couth et al., 2020; Manohar et al., 2023). These individual differences likely caused some of the variability evident in Figure 3. Therefore, accounting for trait, and potentially state, measures of subclinical symptoms could help identify the underlying neural causes of sensory discomfort.

In summary, there are many questions remaining regarding what causes auditory discomfort. Starting with what is known about visual discomfort and finding the auditory analog seems to be a sensible place to start, while being cognizant that many features of auditory stimuli are processed earlier and faster than are processed in the visual system. In an ideal world, we would be able to predict how much discomfort an auditory stimulus would evoke, similar to having the ability to predict the likelihood of a visual media clip inducing a photoparoxysmal response in individuals with photosensitive epilepsy (Kasteleijn-Nolst Trenité et al., 1999). If we are able to predict negative responses in this way in the auditory system too, then we might be able to optimize our auditory environment to reduce the discomfort and stress experienced, particularly by those who are sensitive.