Pre-change stimulus duration modulates acoustic change complex amplitude without affecting latency in normal-hearing adults

The acoustic change complex (ACC), a cortical auditory evoked potential, reflects the brain's ability to detect changes in an ongoing sound. This study investigates the influence of pre-change stimulus duration on the ACC in normal-hearing adults. Twenty normal-hearing adults (10 men, 10 women) participated. ACC N1-P2 responses were recorded using a 1,000 Hz base tone that increased by 12% in frequency with pre-change durations ranging from 1,000 to 5,000 ms. Linear mixed models were used to evaluate effects of pre-change duration and sex on ACC amplitude and latency. ACC N1-P2 amplitudes significantly increased with longer pre-change durations, with 5,000-ms stimuli yielding amplitudes 2.3 times larger compared to those evoked by a duration of 1,000 ms. Female participants showed significantly larger ACC N1-P2 amplitudes and shorter N1 latencies compared to male participants across all durations. Longer pre-change stimulus durations increased ACC N1-P2 amplitudes while latencies remained stable. Durations ≥3,000 ms produced consistently robust responses, but duration choice should balance response strength with testing efficiency. Further work is needed to determine how these effects generalize across populations and clinical settings.

Auditory evoked potentials (AEPs) can be applied to evaluate hearing abilities across the auditory pathway: brainstem regions, midbrain, and auditory cortices in particular. Due to their objectivity and reproducibility, AEPs are well-suited for objective assessment of hearing function (Edmonds, 2008). For decades, it has been acknowledged that cortical auditory evoked potentials (CAEPs) hold promise as objective measurements linked to speech perception (Eggermont and Ponton, 2002). However, relations between speech perception and onset CAEP morphologies, elicited in reaction to the onset of a pure tone, have been inconsistent among studies. Either weak to moderate correlations (r = −0.29–0.43) (Lammers et al., 2015; Vonck et al., 2022) or insignificant relationships (p = 0.09–0.990) (Billings et al., 2009) were found. CAEPs evoked to speech stimuli have been shown to correlate with and predict speech perception in noise, with measures such as N1 amplitude and latency reflecting neural encoding of the signal-to-noise ratio (SNR) (Billings et al., 2013, 2015). These responses primarily capture early sensory encoding, while behavioral speech perception is additionally influenced by higher-order processes. Studies using speech-evoked CAEPs further show that onset responses reflect how effectively the auditory system separates speech from background noise. In normal-hearing listeners, performance is mainly associated with suppression of noise-evoked responses, whereas in cochlear implant users it is linked to enhanced responses to the speech signal, indicating different underlying neural mechanisms. Moreover, these relationships are task-dependent, with CAEPs more strongly predicting single-word than sentence-level speech perception (Berger et al., 2023; Kim et al., 2021).

The acoustic change complex (ACC) is a cortical pre-attentive auditory evoked potential that reflects the brain's ability to detect changes in an ongoing sound (Martin and Boothroyd, 1999) and is said to be a more promising auditory evoked potential to predict speech perception abilities compared to the onset CAEP (Meehan et al., 2025; Schellekens et al., 2025; Tremblay et al., 2003; Vonck et al., 2022). The ACC is applicable and robust in normal-hearing populations but also in hearing loss and cochlear implant populations (Friesen and Tremblay, 2006; He et al., 2015; Kang et al., 2025). Depending on stimulus characteristics and recording parameters, ACC responses to changes in continuous stimuli can exhibit relatively large amplitudes and prolonged N1 latencies compared to onset-evoked CAEPs, suggesting the involvement of neural populations responding specifically to acoustic change in addition to those responding to stimulus onset (Jones, 2003; Vonck et al., 2022). This can be explained by the different characteristics of the ACC compared to the CAEP as the ACC is influenced by factors such as magnitude, range and detection of change (Vonck et al., 2019), whereas the CAEP is primarily influenced by sound level (Adler and Adler, 1991). The ACC has been documented to exhibit responses to alterations in speech stimuli, as well as variations in intensity, frequency, duration, or spatial positioning (Vonck et al., 2019).

Moreover, the amplitude of the ACC appears to be influenced by the duration of the auditory stimulus, more specifically the pre-change duration (i.e., the duration of the auditory stimulus before the change occurs to evoke an ACC response). Most studies investigating the effect of pre-change stimulus duration on the ACC have used pre-change durations of 200–3,000 ms (Shalaby et al., 2022; Vonck et al., 2021) of which 400 ms has been commonly used up until now (Schellekens et al., 2025). A decade ago, (Ganapathy et al. 2013) investigated this relationship using relatively short pre-change durations (50–150 ms) with large frequency changes (1,000–2,000 Hz, 100% increase) in pure tones (Ganapathy et al., 2013). They found that increasing pre-transition duration resulted in increased ACC N1-latencies but no significant amplitude changes within this limited duration range. More recently, (Schellekens et al. 2025) investigated the effect of varying pre-change durations of 250, 500, 1,000, 2,000, and 3,000 ms on the ACC (Schellekens et al., 2025). The stimulus they used had a base tone of 1,000 Hz that changed to a 10% higher frequency. Unlike (Ganapathy et al. 2013), their results showed an increase of the ACC N1-P2 amplitude with increasing pre-change stimulus durations.

Besides tonal stimuli containing acoustic changes, the ACC can also be evoked by speech stimuli. To our knowledge, only (Ganapathy et al. 2013) investigated the effect of pre-change stimulus duration on the ACC using the consonant-vowel (CV) speech syllable /sa/, with /s/ being the pre-change stimulus and /a/ the change stimulus. With pre-change durations ranging from 50 to 150 ms, they found that with increasing the pre-change duration, the ACC N1 latency increased as well. However, they did not find a significant effect on N1 and P2 amplitudes. It should be acknowledged that the choice of speech syllable (and pre-change duration) to evoke the ACC response is important as some syllables might elicit overlapping ACC responses. According to (Tremblay et al. 2003) it is possible that the ACC is not always recorded because of overlapping components due to short pre-change durations.

A study by (Vonck et al. 2022) used a 3,000 ms pure tone, followed by a frequency modulation sweep to a target frequency of +12% to elicit ACC responses (Vonck et al., 2022). Using this stimulation paradigm, clear ACC responses with good SNRs could be recorded and a strong correlation was found between the ACC amplitude and latency and the speech reception threshold (SRT) (r = 0.74–0.87). Patients with longer ACC N1 latencies have worse speech perception scores than patients with comparable hearing thresholds and shorter N1 latencies. This highlights the potential of the ACC as a tool in the objective assessment of speech perception (Tremblay et al., 2003). Moreover, the ACC is language-independent, making it a useful measurement for people with insufficient language proficiency, for patients with cognitive impairment, for children or in cases of suspected malingering (Martinez et al., 2013; van Heteren et al., 2022). However, the long pre-change duration (3,000 ms) increases study duration limiting the potential clinical use of their presented prediction model. In the present study, we used the same stimulus type as (Vonck et al. 2022) with varying base tone durations, investigating the effect of pre-change duration on the ACC N1-P2 amplitude and N1 latency. Specifically for this type of stimulus, (Vonck et al. 2022) reported that increasing the base tone duration from 1,000 ms to 3,000 ms already yields 30–50% larger amplitudes (Vonck et al., 2019). This effect is due to neural adaptation with less neuronal activity after 3,000 ms compared to 1,000 ms. In other words, auditory neurons have more time to adapt to a longer duration of the base frequency, allowing more neurons to react to the frequency change (Wang et al., 2005).

Based on previous research, we anticipated that shorter pre-change durations (< 3,000 ms) would yield smaller ACC N1-P2 amplitudes, whereas longer durations (>3,000 ms) would produce larger ACC N1-P2 amplitudes. However, we expected the ACC amplitude to reach a plateau at one of the longer durations, consistent with adaptation and response saturation in auditory cortex (Bartlett and Wang, 2005; Wang et al., 2005). The aim of this study was to determine whether comparable ACC N1-P2 amplitudes and N1 latency values could be achieved with shorter stimuli while maintaining distinct SNRs, with the goal of reducing overall test duration with the stimulus type presented by (Vonck et al. 2022). If their prediction proves suitable for clinical use, employing shorter stimuli would be advantageous, as it would reduce the overall testing time and improve patient experience.

Participants

Twenty adults (10 men, 10 women) with normal hearing, aged between 21 and 37 years (mean age = 26.4, SD = 4.3) participated in the study. The mean ages of the men and women did not significantly differ (26.2 and 26.5 years, respectively). All participants underwent pure-tone audiometry testing (125 Hz−8,000 Hz) to confirm normal hearing with pure tone thresholds below 20 dB HL for all frequencies. The best hearing ear was determined by the pure tone average [PTA (in dB HL)] over 500, 1,000, 2,000, and 4,000 Hz. This ear received auditory stimulation during the ACC recording. In the event of identical pure tone averages for both ears, the subjectively best ear was used. Participants with any neurological disorders or prior musical training were excluded from participating.

This prospective study was conducted at the Antwerp University Hospital (UZA) in Belgium. The study, designed and conducted according to the Declaration of (World Medical Association 2013), was approved by the Ethical Committee of the Antwerp University Hospital (UZA) (reference number B3002022000172). Written informed consent was obtained from all participants prior to the study procedure during the following recruitment period: 14/06/2023 to 19/01/2024.

ACC stimuli and recording procedure

The stimulus consisted of a 1,000 Hz base frequency that swept to a 12% higher frequency over 3 ms, resulting in a target frequency of 1,120 Hz (Figure 1A). This frequency modulation was implemented as a continuous logarithmic sweep, corresponding to a change of 0.163 octaves at a sweep rate of 54.5 octaves per second. The duration of the base tone varied which resulted in different pre-change stimulus durations. The durations that were used in this study were 1,000 ms, 1,500 ms, 2,000 ms, 3,000 ms, 4,000 ms, and 5,000 ms. The higher frequency tone, presented after the sweep, had a duration of 300 ms. A 200 ms interstimulus interval was adopted. The stimuli had a rise-fall time of 5 ms. All stimuli were presented in a randomized block order to correct for attention-related bias.

Stimuli were generated using MATLAB (version 24.2, R2024b, The Mathworks, Inc.) and presented monaurally to the better ear through RadioEar DD45 supra-aural headphones at a level of 75 dB SPL, using Presentation ® software (Version 23.0, Neurobehavioral Systems, Inc., Berkeley, CA, www.neurobs.com). Spectrograms of the stimuli can be found in Supplementary Figure S1.

Recordings were performed inside a sound-attenuated booth. Participants were seated in a comfortable reclining chair positioned 1.5 meters from a 26” screen on which a silent, captioned movie played. This viewing distance ensured that participants could keep eye movements to a minimum, even while reading subtitles. Before each recording session, participants received detailed instructions to minimize muscle and eye movements and to fixate on the center of the screen, thereby reducing potential interference from muscle and eye movement artifacts. Electrophysiological responses were acquired using Ag/AgCl electrodes placed according to the 10–20 system and recorded with a Nihon Kohden Neuropack X1 Evoked Potential system. Minimal electrode placement was used with the active electrode placed at the vertex (Cz) of the skull, the reference electrode positioned on the contralateral mastoid (A1/A2), and the ground electrode on the lower forehead (Fpz). Responses exceeding a threshold of 100 μV, due to eye movements, blinks or muscle artifacts were rejected. On average, this resulted in the rejection of 1–10% of trials per subject. Electrode impedance was rigorously maintained below 5 kΩ for each electrode. The electrode signals were sampled at a rate of 2,048 Hz and filtered within the range of 0.01 to 100 Hz while recording. A notch filter of 50 Hz was applied during recordings well. Responses were acquired in a time window of 1,000 ms, encompassing a 100 ms pre-stimulus period. In this time window, the trigger to record the ACC response was sent at 0 ms. For each recording, 100 accepted sweeps were acquired and waveforms were averaged. The complete recording procedure had a duration of approximately 1 h.

Data collection and analysis

Amplitudes and latencies of both the ACC and CAEP were manually determined by marking the N1 peak as the most negative peak between 70 and 130 ms or 600 and 700 ms and the P2 peak as the most positive peak between 150 and 250 ms or 700 and 800 ms for the ACC responses and onset CAEP responses, respectively. The noise floor of the EEG recording was quantified by calculating the root mean square (RMS) of the pre-change interval (−100 to 0 ms) using MATLAB version 24.2 (R2024b, The Mathworks, Inc.). This analysis yielded an average baseline EEG noise level of 0.85 μV (approximately 1 μV). The presence of the ACC N1-P2 amplitude was determined upon visual inspection. In case of uncertainty, another researcher was consulted (LJ). We used an ACC N1-P2 amplitude criterion of 3 μV, representing three times the measured noise floor, in case the amplitude was not clear for both researchers. Regardless of presence, the amplitude values of all ACC N1-P2 responses in all participants were included for statistical analyses. The ACC N1-P2 peak-to-peak response was chosen as outcome measure as this has been frequently used in previous research on this specific response (Kirby and Brown, 2015; McGuire et al., 2021; van Heteren et al., 2022; Vonck et al., 2022).

Statistical analyses were completed using JMP ® Pro software (JMP ® Pro, Version 17.0.0, 2021 SAS Institute Inc., Cary, NC, 1989–2024). Linear mixed models were designed to explore the effect of pre-change stimulus duration on the ACC and CAEP amplitude and latency. Both the ACC and CAEP amplitude and latency were chosen as the outcome variables. Pre-change stimulus duration and sex were added as fixed factors, and the interaction Stimulus duration * Sex was added to the models since previous research about (sub)cortical responses reported sex differences (Delhez et al., 2020; Jerger and Hall, 1980; McFadden, 2009; Solanki and Mehta, 2015). Subject was added as a random intercept. Post-hoc pairwise comparisons with the Tukey method were used to determine which differences between stimulus durations were significant (Supplementary Table S1–S4). To account for eight statistical comparisons (six pre-change stimulus durations and effects of sex), a Bonferroni correction was applied, resulting in a corrected significance threshold of p = 0.00625. Unless otherwise specified, reported p -values are uncorrected.

Success rate when applying a 3 μV amplitude cut-off

Only for pre-change durations of 3,000, 4,000, and 5,000 ms, all 20 subjects elicited clear and reproducible ACC responses with the characteristic N1-P2 waveform morphology. The success rates for eliciting ACC N1-P2 responses of at least 3 μV in all participants were 70%, 85%, 85%, 100%, 100%, and 100% for pre-change durations of 1,000, 1,500, 2,000, 3,000, 4,000, and 5,000 ms respectively. In the female population, the success rates were 80% for a pre-change duration of 1,000 ms and 100% for all other durations. Success rates in the male population were 60%, 70%, 70%, 90%, 100%, and 100% for durations of 1,000, 1,500, 2,000, 3,000, 4,000, and 5,000 ms, respectively.

The effects of pre-change stimulus duration on ACC amplitude and latency

The linear mixed model showed a significant effect of pre-change stimulus duration on the ACC N1-P2 amplitude, illustrating that the ACC amplitude increased with longer duration of the stimulus before the frequency change [ F (5, 88) = 21.65, p < 0.0001, η p 2 = 0.55, Figure 2A]. The mean ACC N1-P2 amplitude at 5,000 ms was 2.34 times larger than the mean amplitude at 1,000 ms, resulting in a higher SNR. The increasing ACC N1-P2 amplitude with pre-change stimulus duration originated from an increase in N1 amplitude [ F (5, 88) = 23.38, p < 0.0001, η p 2 = 0.57, Figure 3]. The P2 amplitude did not differ between pre-change durations [ F (5, 88) = 0.50, p = 0.7772, η p 2 = 0.03]. The mean N1 amplitude at 5,000 ms was 3.15 times larger than the mean N1 amplitude at 1,000 ms. Residual noise levels did not significantly differ between pre-change stimulus durations [ F (5, 87) = 1.01, p = 0.4143, η p 2 = 0.05].

The linear mixed model revealed that the ACC N1-P2 amplitude differed between male and female participants [ F (1, 18) = 13.92, p = 0.0015, η p 2 = 0.44, Figure 2B]. At each pre-change duration, the ACC amplitude was larger in women compared to men. More specifically, at every pre-change stimulus duration, the ACC N1-P2 amplitude was 1.5–2 times larger in female participants than in male participants. There was a significant interaction between pre-change stimulus duration and sex [ F (5, 88) = 5.10, p = 0.0004, η p 2 = 0.22]. With increasing durations, the increase in ACC N1-P2 amplitude is greater in female compared to male participants. The ACC N1-P2 amplitude appeared to reach saturation at 5,000 ms in male participants, but not in female participants. The residual noise levels did not significantly vary between males and females [ F (1, 18) = 2.31, p = 0.1460, η p 2 = 0.11], nor was there a significant interaction between pre-change stimulus duration and sex (F (5, 87) = 1.39, p = 0.2348, η p 2 = 0.07).

The ACC N1 latency did not vary with pre-change stimulus duration [ F (5, 88) = 1.83, p = 0.1159, η p 2 = 0.09, Figure 4A]. The ACC N1 latency was shorter in women compared to men in our study population [ F (1, 18) = 12.22, p = 0.0026, η p 2 = 0.40, Figure 4B]. The interaction Stimulus duration * Sex was not significant for this variable [ F (5, 88) = 1.91, p = 0.1012, η p 2 = 0.10]. Both male and female participants showed similar ACC N1 latencies in response to different pre-change stimulus durations, even though women had shorter N1 latencies overall.

The effects of pre-change stimulus duration on the onset CAEP amplitude and latency

The onset CAEP, a smaller N1-P2 waveform that occurred around 100 ms after the onset of the following stimulus was recorded in the same time window (Figure 1B). For all pre-change stimulus durations, all 20 subjects elicited clear and reproducible responses. This confirmed the presence of the obligatory onset response in all subjects.

After correcting for multiple comparisons, the onset CAEP N1-P2 amplitude did not change with pre-change stimulus duration of the base frequency stimulus [ F (5, 88) = 3.21, p = 0.0103, η p 2 = 0.15, Figure 5A].

Additionally, the N1-P2 amplitude of the onset CAEP did not significantly differ between male and female subjects [ F (1, 18) = 3.65, p = 0.0722, η p 2 = 0.17, Figure 5B], while the interaction Stimulus duration * Sex was significant [ F (5, 88) = 5.83, p = 0.0001, η p 2 = 0.25]. Specifically, while the amplitude in male subjects remained relatively constant, the onset CAEP N1-P2 amplitude in female subjects was larger at the longer pre-change durations of 3,000, 4,000, and 5,000 ms.

Similarly to the ACC, the N1 latency of the onset CAEP did not vary with pre-change stimulus duration [ F (5, 88) = 1.30, p = 0.2720, η p 2 = 0.07] (Figure 5C). There were no significant sex effects for this variable [ F (1, 18) = 2.91, p = 0.1050, η p 2 = 0.14] (Figure 5D). These findings indicate that the obligatory onset CAEP was not affected by variations in duration of the previous frequency change eliciting the ACC or by the N1-P2 amplitude of the preceding ACC.

This study examined how pre-change stimulus duration influences the ACC N1-P2 response in normal-hearing adults using a 12% frequency change paradigm. Across 1,000–5,000 ms durations, ACC N1-P2 amplitudes increased steadily with longer pre-change durations, driven primarily by changes in the N1 component. These results refine existing findings by demonstrating that the effect of stimulus duration is robust. Our findings illustrate the trade-off between response robustness and testing time that clinicians and researchers must consider, especially if a language-independent prediction model like that of (Vonck et al. 2022) would be used clinically. Additionally, we have found that the ACC N1-P2 amplitude is larger and the ACC N1 latency is shorter in females compared to males.

The ACC amplitude increases with pre-change stimulus duration

The N1 component is mainly driven by sensory processing, acting as a marker for sound detection and responding to attention-grabbing features, rather than reflecting subjective perception or discrimination abilities (Joos et al., 2014).

Auditory neurons exhibit adaptation to frequent, repetitive stimuli, yet they sustain their responsiveness to stimuli possessing different physical characteristics (Pérez-González and Malmierca, 2014). This represents a unique form of processing that could contribute to the detection of change and deviation. When presented with a longer auditory stimulus, auditory neurons have more time to adapt and recover from the previous stimulus which means that a sudden frequency change results in a greater N1 response, i.e., a higher ACC N1-P2 amplitude. Apart from adaptation, the increasing ACC N1-P2 and relatively stable CAEP N1-P2 responses with longer pre-change stimulus durations indicate that a specific population of neurons may respond more to stimulus changes than to onsets (Vonck et al., 2022). The longer ACC N1-latency compared to the CAEP N1-latency in this, and previous studies further support the involvement of two distinct neuron populations (Liang et al., 2016; Vonck et al., 2022). However, Liang et al. observed that individuals with larger ACC N1-P2 responses often exhibit larger CAEP N1-P2 responses, suggesting that both could also share a common neural population (Liang et al., 2016). Moreover, an interstimulus interval allows for dishabituation and more neural recovery, which might also contribute to larger ACC N1-P2 responses (Vonck et al., 2019). This contrasts with shorter stimuli, where auditory neurons have less time to adapt to a particular tone. When the stimulus changes, the response will not be as large, resulting in a smaller ACC N1-P2 amplitude. However, the P2 component does not significantly change between pre-change durations, suggesting this response is less sensitive to temporal aspects of stimulus presentation. Instead, the P2 component is believed to play a more cognitive role, reflecting attention-modulated processes necessary for performing auditory discrimination tasks (Crowley and Colrain, 2004; Liang et al., 2016).

Implications for clinical ACC applications

Both long and short stimulus durations have their advantages and disadvantages. Using a longer pre-change duration yields a larger ACC N1-P2 amplitude which results in a better SNR. We do not know if the amplitude keeps rising with longer pre-change stimulus durations than 5,000 ms, although we believe that a plateau should occur eventually. On the other hand, shorter pre-change durations would reduce total test duration, improving potential clinical applications of ACC recordings in order to potentially predict hearing and speech abilities. The effect of pre-change stimulus duration on the ACC has been investigated before but there is a difference between our stimulus and theirs. (Ganapathy et al. 2013) investigated the minimum pre-change duration of both tonal and speech stimuli (Ganapathy et al., 2013). Their pre-transition tone had a frequency of 1,000 Hz that swept to 2,000 Hz (+100%) after a pre-transition duration of 50–150 ms, increasing duration in 10 ms steps. The sweep between both base and target frequency had a duration of 20 ms. Although they varied the pre-transition duration, the total stimulus duration was fixed at 350 ms. Additionally, they used an inter-stimulus interval of 800 ms. A similar and more recent study by (Schellekens et al. 2025) investigated pre-change stimulus durations of 250, 500, 1,000, 2,000, and 3,000 ms and their effect on the ACC (Schellekens et al., 2025). The frequency change of the stimulus used in this study was 1,000–1,100 (+10%) with a 5 ms logarithmic increase. The target frequency had a duration of 495 ms and the inter-stimulus interval had a value of 500 ms. Both our stimuli and those of (Schellekens et al. 2025) are characterized by rather subtle frequency changes (12% and 10%, respectively) compared to (Ganapathy et al. 2013), which may be more representative of sound in daily life as the ability to detect modulations in ongoing sound is important to detect environmental changes, distinguish vowels from consonants in continuous speech and to acknowledge pitch differences in music (Vonck et al., 2019). It needs to be addressed that our stimulus has a more subtle frequency change compared to (Ganapathy et al. 2013), and a shorter inter-stimulus interval compared to both (Ganapathy et al. 2013) and (Schellekens et al. 2025) It would be indeed possible to evoke ACCs with shorter stimuli given a greater frequency difference and a longer inter-stimulus interval which causes more neurons to fire to a greater acoustic change and allows more neurons to recover, respectively. However, use of large frequency changes may not reflect the neural processing dynamics involved with smaller, more relevant frequency changes. The stimulus paradigm used in this study relates to the difference limen for frequency (DLF) which is said to correlate with pitch, a fundamental perceptual attribute in the perception of speech (Gong et al., 2013). Therefore, a paradigm like the one presented in this study would be interesting to use in the clinical assessment of frequency discrimination and speech perception (McGuire et al., 2021).

Optimizing stimulus parameters while maintaining response quality is essential for clinical implementation. The ACC N1-P2 amplitude was not always easily detected in every participant when using the 1,000, 1,500, and 2,000 ms stimulus durations. With stimulus durations of 3,000 ms and longer, the ACC became detectable in every subject, although in the female population ACC N1-P2 amplitudes were already clearly distinguishable from a duration of 1,500 ms onwards. To apply a prediction model such as the one proposed by (Vonck et al. 2022), a high SNR would be recommended for clinical implementation. Clinicians must be stringent when determining whether an ACC response is present. For this reason, using a criterion such as a minimum amplitude of 3 μV is appropriate, as noise levels can vary between patients and across sessions. In our study, when hypothetically applying this 3 μV threshold, ACC responses were consistently observed only from a pre-change duration of 3,000 ms onwards. While previous work by (Vonck et al. 2019) systematically investigated the effects of rate, magnitude, and direction on ACC responses, the present study adds (pre-change) stimulus duration as a crucial parameter for model optimization. Although the general effects of pre-change duration on ACC responses have been reported in the literature, confirmation with this specific stimulus type was necessary before shortened protocols for, e.g., speech prediction models can be implemented.

Reduced latency and larger amplitude of the ACC in women compared to men

Our study found that the ACC N1 latency was shorter, and its N1-P2 amplitude larger in women compared to men, which complements findings in literature generally showing better auditory function in women than in men. It suggests that especially young pre-menopausal females have better hearing sensitivity to spectral change than males (Aloufi et al., 2023). For the ACC N1 latency this effect was consistent at a pre-change stimulus duration of 1,000 ms. For the ACC N1-P2 amplitude this effect was highly consistent over two out of six recordings, i.e., 4,000 ms and 5,000 ms. Studies have reported that, on average, women have greater hearing sensitivity than men which has already been demonstrated with pure-tone audiometry, auditory brainstem response (ABR), and CAEP measurements (Delhez et al., 2020; Jerger and Hall, 1980; McFadden, 2009; Solanki and Mehta, 2015). ABR data of women tend to show earlier latencies compared to men (Jerger and Hall, 1980) and women show greater sensitivity across all audiometric test frequencies (0.25–8 kHz) (Chung et al., 1983). Although it should be mentioned that sex differences in ABR might be due to difference in head size (Aoyagi et al., 1990; Hall, 2007).

Limitations and directions for future research

A few limitations of this study and directions for future research should be addressed. Although the study was based on a relatively small sample of 20 participants, the observed effects were large, indicating that the available sample provided sufficient sensitivity for detecting our primary outcomes. The main effect of Stimulus Duration on ACC N1-P2 amplitude was strong, with F (5, 88) = 21.65, η p 2 ≈ 0.55 (Cohen's f ≈ 1.11). The Duration * Sex interaction was also sizable [ F (5, 88) = 5.10; η p 2 ≈ 0.22, f ≈ 0.53], and the main effect of Sex was large [F (5, 88) = 13.92; η p 2 ≈ 0.44, f ≈ 0.89]. These large to very large effect-size estimates support the adequacy of 20 participants for detecting both within- and between-subject effects as well as their interaction in repeated-measures/mixed-model framework. Furthermore, the robustness of our design aligns with previous ACC research in normal-hearing adults, which has reported clear amplitude increases with longer pre-change durations using similar sample sizes (Ganapathy et al., 2013; Schellekens et al., 2025), reinforcing the practical sufficiency of our cohort size. Given this, a post-hoc power analysis based on the present data yields an estimated power of 0.99 (F -test, repeated-measures ANOVA, within-between interaction; f = 0.53, α = 0.05, total sample size = 20, two groups, six measurements, correlation among repeated measures = 0.5, non-sphericity correction ε = 1). Nevertheless, studies with larger sample sizes would be valuable to assess the stability and generalizability of the present findings.

Another potential limitation relates to the use of a 3-μV amplitude threshold for determining ACC presence. Although smaller responses may be visually detectable when noise levels are low, clinical EEG recordings often vary considerably depending on the patient population, clinical context, and recording environment. For this reason, using an amplitude criterion set at three times the noise floor may provide a pragmatic reference for distinguishing ACC responses under noisier conditions.

Because the aim of this study was not to compare ACC and CAEP responses directly, a separate onset CAEP recording was not acquired. Instead, we relied on the onset CAEP N1-P2 response embedded within the ACC recording window. Future work should include a dedicated onset CAEP measurement to enable direct comparison of onset- and change-related cortical responses.

Additionally, only the clinically better ear was assessed for each participant. Given that ear asymmetries may differ by sex and age, this choice may have influenced the observed sex differences (Jerger and Hall, 1980; McFadden et al., 2010). Previous studies have reported sex-dependent ear effects, particularly at higher frequencies. Future studies should include bilateral ACC recordings to determine whether ear-specific factors contribute to amplitude differences observed between men and women.

Moreover, future work should also explore the cortical generators and spatial characteristics of ACC responses. Only one EEG study to date has examined sex differences in source localization of higher-order auditory responses such as MMN, P300, and N400 (Tsolaki et al., 2015). Extending such source-level analyses to the ACC N1-P2 could provide important insights into underlying neural mechanisms and could further elucidate sex-related differences in cortical auditory processing, especially given the ACC's potential clinical role in predicting speech perception outcomes.

Finally, as current research on this topic has primarily focused on normal-hearing populations, the applicability of these findings to individuals with hearing loss and cochlear implant users remains to be determined. Previous work has shown that hearing loss is associated with reduced ACC amplitudes and prolonged latencies, suggesting that the detectability of ACC responses may be diminished in these populations (Meehan et al., 2025). Consequently, if a pre-change duration of 1,000 ms yields detectable responses in a majority of normal-hearing listeners, it is reasonable to expect lower response detectability under similar conditions in hearing-impaired listeners.

Pre-change durations, such as 3,000 ms, have already been employed successfully in hearing-impaired and CI populations, likely facilitating sufficient neural encoding of the ongoing stimulus prior to the acoustic change. However, it remains unclear whether shorter durations using similar stimulus parameters would be adequate to evoke reliable ACC responses in these groups. Some evidence suggests this may be possible. (McGuire et al. 2021) successfully elicited ACC responses in CI users using only a 1,000-ms stimulus with a frequency change occurring midway. Notably, the magnitude of the frequency change in that study was substantially larger (70%) than those used in other paradigms (e.g., 12%), which likely facilitated response detection. Indeed, larger acoustic changes are known to enhance ACC response robustness, particularly in populations with reduced auditory resolution (Vonck et al., 2019).

Longer pre-change stimulus durations produced larger ACC amplitudes in normal-hearing adults, while latencies remained stable. These findings show how duration shapes ACC morphology for a specific type of frequency change stimulus. While durations of 3,000 ms and longer yielded large, visible responses in all participants under our recording conditions, the choice of duration in clinical or research contexts should balance response robustness with testing efficiency. Further work is required to determine how these duration effects translate across patient populations, stimulus types, and clinical environments.