Thirteen of the reviewed studies in Table 1 included middle-aged to older adults with moderate or mild-to-moderate hearing loss. Liang et al. (2020) recruited moderate-to-severe hearing loss in younger to middle-aged adults, and Paul et al. (2025), recruited younger and older adults with only high-frequency hearing loss. Eight studies examined neural activity during passive visual stimulation, and seven included active behavioral tasks. Eleven studies used electroencephalography (EEG) to measure visual evoked potentials or neural oscillations. Four studies measured hemodynamic responses and functional connectivity using functional magnetic resonance imaging (fMRI) or functional near-infrared spectroscopy (fNIRS).

The visual evoked potential (VEP) was the most common technique used to measure cross-modal plasticity in humans with partial hearing loss. The VEP complex consists of ordered components labeled P1, N1, P2, N2, etc., that represent the maxima and minima of phase-locked voltage fluctuations along the visual cortical processing pathway. VEPs are characterized by their amplitude and latency. When measured by scalp EEG, VEP components reflect a composite of multiple underlying cortical generators that sum at the surface sensors (Arroyo et al., 1997; Freunberger et al., 2007). Seven of eleven EEG studies reported correlations between hearing loss and VEPs. Several studies claim that visual cross-modal plasticity occurs when hearing loss coincides with larger VEP amplitudes and shorter VEP latencies, which purportedly indicate more sensitive or efficient visual processing. Shown in Table 2, this configuration of results was observed in 16 cases (i.e., results of an analysis). However, study outcomes were highly inconsistent. The opposite effect, where smaller amplitudes or longer latencies occurred with increasing hearing loss, were reported in 8 instances, and 24 sets of analysis across VEP components were not significant. Effects on the P2 component were the least consistent in direction and spatial location, while effects on the P1 and amplitude of N1 were more consistent.

Multiple factors potentially explain inconsistent correlations between VEP components and hearing loss. VEPs were evoked by several types of stimuli, including visual speech (Aguiar et al., 2025), motion (Aguiar et al., 2025; Campbell and Sharma, 2014, 2020; Glick and Sharma, 2020), written words (Paul et al., 2025) shapes (Loughrey et al., 2023; Zhu et al., 2024), reversing checkerboards (Seol et al., 2024) and photographs (Liang et al., 2020), which could change VEP morphology because they are processing along separate anatomical pathways (Harter et al., 1982). Some studies included an active task, and attention and memory demands are known to modulate VEP amplitudes and latencies (Adam and Collins, 1978; Eason, 1981). Participant-level factors, such as age degree of hearing ability, and cognitive function (Glick and Sharma, 2020) also add to variability to the data and could affect correlations between hearing sensitivity measures and VEP features. It is also possible that the association between hearing loss and neural activity indexed by VEPs is not sufficiently robust for evaluating claims about cross-modal plasticity. Beyond VEP analysis, both Shende et al. (2024) and Aguiar et al. (2025) analyzed time-frequency representations of EEG data and found increased theta or alpha oscillation synchronization occurring 100–200 ms after the onset of visual stimulation.

Neural activation and functional connectivity recorded with fMRI was less frequently used to measure cross-modal responses. Rosemann and Thiel (2018) found increased activation in frontal areas to visual speech in participants with hearing loss, and frontal activation in medial, middle, and inferior frontal gyrus was correlated with an increased visual bias in an audiovisual speech task. However, there was no evidence for visual cross-modal recruitment of auditory cortex. Two other fMRI studies did not find effects of hearing loss on neural activation to visual images or visual motion (Puschmann and Thiel, 2017; Rosemann and Thiel, 2020).

All but two studies reviewed in Table 1 used cross-sectional and correlational designs. The longitudinal study conducted by Glick and Sharma (2020) measured VEPs in older adults with age-related hearing loss before and after 6 months of hearing aid use. VEP P1, N1, and P2 latencies over right temporal cortex were shorter in the hearing loss group compared to a control group prior to provision of hearing aids, but latencies returned to levels comparable to typical hearing adults following the intervention. This suggests that visual cross-modal plasticity can be reversed by hearing treatment. In a complementary study, Mai et al. (2024) used fNIRS to measure hemodynamic responses to visual checkerboard stimulation in people with hearing loss, before and after 4 weeks of speech-in-noise training. There were no effects on auditory cortical activation, but training decreased functional connectivity between auditory, frontal, parietal and temporo-parietal areas, with this effect disappearing after 4 weeks. The decrease in frontotemporal connectivity suggested reduced visual cross-modal recruitment because of increased auditory experience.

We first examined evidence for visual cross-modal recruitment of auditory cortex in partial hearing loss. Research in animal models shows that visual recruitment occurs in higher-order, non-primary auditory cortex (Lomber et al., 2010; Meredith et al., 2011). Here, neurons are innervated by pre-existing “heteromodal” visual or somatosensory synaptic inputs that typically modulate auditory processing in an inhibitory fashion and affect the phase of ongoing neuronal oscillations (Kayser et al., 2008; Morrill and Hasenstaub, 2018). Kral and Sharma (2023) propose that deafness reduces input to higher-order cortical neurons, and homeostatic mechanisms cause heteromodal inputs to strengthen and transition from a modulatory to a driving role. This raises visual activity levels in higher-order auditory regions and results in cross-modal recruitment. Neural activity recorded from a rat model of noise-induced hearing loss also suggests that cross-modal recruitment can occur with partial hearing decline. Noise-exposed animals show an increase in the number of neurons responding to visual and audiovisual stimulation in higher-order auditory and multisensory cortex compared to typical hearing control rats (Schormans et al., 2017; Schormans and Allman, 2024). These data also show evidence for a transitional shift in the functional border between auditory and visual areas across sensory cortex, where auditory-responding regions recede and visual-responding regions expand. The expansion of visual-responding areas across sensory cortex with cross-modal recruitment might be expected to arise in human neuroimaging research as increased activity in auditory temporal regions anterior to visual cortex.

VEPs measured by scalp EEG cannot definitively resolve neural sources because volume conduction through skin and other tissues distorts the distribution of neuroelectric potentials. Underlying neural generators can only be approximated by using spatial filtering and modeling of tissue conductance. Of seven EEG studies that employed source analysis, two found that hearing loss was associated with increased right-sided or bilateral temporal cortex activation, consistent with evidence of cross-mod recruitment (Aguiar et al., 2025; Paul et al., 2025). Four additional studies using source analysis to resolve VEP generators found descriptive evidence for activation of temporal cortical areas (Campbell and Sharma, 2014, 2020; Glick and Sharma, 2020; Liang et al., 2020), but these studies did not statistically compare these activations across hearing loss and typical hearing groups. One additional study did not find statistically significant differences in auditory cortical source activation between hearing loss and typical hearing groups (Stropahl and Debener, 2017). Notably, no fMRI or fNIRS study in this review showed evidence for cross-modal recruitment. It is possible that visual cross-modal recruitment does not alter metabolic demands in auditory or multi-sensory regions, but this claim cannot be inferred from null results.

Taken together, only a few EEG studies showed statistical evidence for cross-modal recruitment in adult humans with partial hearing loss, while others only showed qualitative differences in activation. Note that source analysis in EEG studies does not rule out the possibility for “leakage” of intra-modal activity into the source activity. If the observed changes in VEP components with hearing loss result from cross-modal plasticity, then it is likely that intra-modal plasticity plays a substantial role. Resting-state fMRI analysis have shown increased resting-state functional connectivity in visual subnetworks in groups with hearing loss (Ponticorvo et al., 2022) and increased connectivity between visual and cognitive areas (Lian et al., 2025). These studies were not included in our main review because no visual stimulation was used, but they suggest that modulation of VEPs by hearing loss can arise through intra-modal compensation.

The heteromodal inputs into non-primary auditory cortex through which visual cross-modal recruitment is proposed to develop are top-down modulatory connections originating high-level multimodal regions implicated in cognition (Ghazanfar and Schroeder, 2006). With sensory deprivation, these top-down hetero-modal connections are proposed to upregulate through homeostatic compensatory plasticity and become the driving input, allowing these regions to perform their functional roles (Kral and Sharma, 2023). Engagement of cognitive neuromodulatory systems such as the basal forebrain cholinergic system are top-down and experience-dependent and have also been shown to gate cortical sensory plasticity after sensory loss (Ramanathan et al., 2009). In humans with deafness, visual cognitive tasks activate temporal cortex, providing evidence for top-down involvement in cross-modal plasticity (Cardin et al., 2018). In partial hearing loss, recruitment of frontal brain regions has been observed during auditory speech tasks and is hypothesized to reflect a compensatory increase in resources for cognitive functions such as effort (Pichora-Fuller et al., 2016) or working memory (Rönnberg et al., 2010). Cross-modal plasticity may also draw from compensatory cognitive resources.

Three studies in Table 1 showed evidence for frontal activation when visual stimulation was used. One used fMRI to show increased frontal activation when participants viewed visual sentences (Rosemann and Thiel, 2018), and two showed descriptively higher frontal activation sourced from VEPs evoked by a motion stimulus (Campbell and Sharma, 2020; Glick and Sharma, 2020). Another study did not find increased frontal activity in hearing loss when using a visual moving dot stimulus but did find higher resting state connectivity between visual and frontal regions (Puschmann and Thiel, 2017). In 10 adults with partial hearing loss, Mai et al. (2024) used fNIRS to show that speech-in-noise training reduced frontotemporal connectivity during visual stimulation. This suggested that increased auditory activity levels attenuated cross-modal activity patterns. One final fMRI study that presented faces and images in a working memory task found no differences in frontal activation between hearing loss and control groups (Rosemann and Thiel, 2020). Therefore, with some exception, these studies agree with claims that cross-modal plasticity arises through top-down connections and implicates frontal cognitive brain networks.

Some studies in Table 1 were not conducted to test for cross-modal plasticity and were instead designed to identify evidence of cognitive decline. Hearing loss has been reliably shown to increase the risk of dementia and cognitive impairment (Deal et al., 2017; Lin et al., 2011, 2023) although no causal relationship has been established. One hypothesis suggests that hearing loss increases the demand on cognitive resources, which may include strategies to increase reliance on visual perception. However, long-term demands of cognitive resource use may degrade cognitive flexibility and function, resulting in or exacerbating cognitive impairment and dementia (Slade et al., 2020). Some EEG studies in Table 1 used visual stimulation in attention and memory tasks to measure cognitive ability, because auditory stimulation would be affected by hearing loss. Zhu et al. (2024) found worse visual attention performance correlated with smaller cortical N2 responses in older adults with partial hearing loss, consistent with cognitive decline. Loughrey et al. (2023) reported smaller amplitudes for the P3 and late positive potential in a group of adults with hearing loss, but no difference was found for performance on a working memory task. Loughrey et al. (2023) also found that the earlier cortical potentials P1 and N1 were higher in amplitude in the hearing loss group, which was interpreted as cross-modal plasticity. These data may suggest that cross-modal plasticity can occur with cognitive-decline, but the absence of behavioral differences casts doubt on this interpretation.

Clarifying evidence was provided by Glick and Sharma (2020) who conducted a longitudinal study to show that several months of hearing aid use reversed EEG evidence of cross-modal plasticity. The authors also tracked how cognitive function changed before and after the intervention. At baseline, VEP evidence of cross-modal plasticity in participants with hearing loss was correlated with worse performance on multiple cognitive tasks including processing speed and working memory. After 6 months of hearing aid use and the reversal cross-modal plasticity, performance on the cognitive tasks improved. One interpretation is that cross-modal plasticity exhausted cognitive reserves and lowered cognitive performance, which could be alleviated with hearing restoration. However, it is also possible that cognitive effects and cross-modal VEP plasticity are independent and were indirectly related through their correlation to hearing loss.

Both cognitive decline and partial hearing loss are more common in older age, but little is known about age effects on cross-modal plasticity. Studies in Table 1 often include older adults, but do not commonly attempt to model age-related effects separately from hearing loss effects. EEG studies show that VEP latencies increase with age, especially for the P1 response (Alain et al., 2022; Price et al., 2017; Shaw and Cant, 1980), but VEP latencies are expected to decrease with visual cross-modal plasticity due to increased visual efficiency (Campbell and Sharma, 2014). To address this, both Aguiar et al. (2025) and Paul et al. (2025) examined effects of participant age, respectively, using visual speech or visual text. Interestingly, both studies found statistically independent and inverse effects of age and hearing loss on the visual P2 response. Aguiar et al. (2025) found shorter latencies with hearing loss and longer latencies with age, whereas Paul et al. (2025) found the opposite: shorter latencies with age, and longer latencies with hearing loss. As mentioned, stimulus differences and task requirements could account for the inconsistent direction of these correlations, but these studies nonetheless show evidence for diverging effects of age and cross-modal plasticity on visual processing.

Cross-modal recruitment is believed to be responsible for enhanced perceptual abilities found in deaf or blind populations. Improvement in perceptual ability is thought to be constrained to functions that are shared between senses. This is because cross-modal recruitment develops through pre-existing neural connections, allowing sensory cortical areas retain their functional role when switching their response preference from a deprived sense to a remaining sense, which overall supports “supra-modal” perception of objects or events (Kral and Sharma, 2023). Deprived regions do not take on new functions that are specific to remaining senses. In agreement, visual “unimodal” functions, such as color and contrast perception and foveal acuity, and contrast, do not appear to improve with deafness [reviewed in Alencar et al. (2018)] as there is no auditory analogue. On the other hand, visual object motion detection and peripheral object tracking, which are both supported by auditory and visual systems, are enhanced in deaf humans (Codina et al., 2011; Shiell et al., 2014).

Causal evidence for a relationship between cross-modal plasticity and enhanced visual ability comes from the congenitally deaf cat. In the deaf cat auditory cortex, Lomber et al. (2010) found cross-modal recruitment in higher-order areas including the posterior auditory field (PAF) and dorsal zone (DZ). In hearing cats, these areas are, respectively, implicated in auditory localization and auditory motion detection. Congenitally deaf cats also show evidence for enhanced visual localization and visual motion detection. Lomber et al. (2010) used cooling loops to temporarily deactivate PAF and DZ. PAF deactivation abolished deaf cats’ visual localization advantage but not visual motion detection, and DZ deactivation abolished visual motion detection but not visual localization. This double dissociation provides strong causal evidence that cross-modal plasticity supports enhanced visual ability. Auditory cortical areas maintained their “supra-modal” role (object localization in PAF, and object motion in DZ) but utilized visual information delivered through heteromodal inputs to PAF and DZ to perform their function. A follow-up study in congenitally deaf cats confirmed this finding, showing that the auditory field of the anterior ectosylvian sulcus (FAES), a region normally involved in acoustic orienting in hearing cats, is visually reorganized in deaf cats. Temporary FAES deactivation using cooling loops caused visual orienting deficits in deaf cats, again showing a retention of function in cross-modally affected regions (Meredith et al., 2011).

However, no studies in our review showed enhanced visual function or a relationship between visual neural activity and behavioral performance on perceptual or cognitive tasks in partial adult-onset hearing loss (Table 1). In fact, Zhu et al. (2024) found worse visual attention in participants with poorer auditory function, which was interpreted as evidence of cognitive decline. While our review cannot corroborate the claim that cross-modal plasticity affects visual-only abilities in partial hearing loss, behavioral research shows evidence for stronger visual influences during audiovisual perception. Compared to people with typical hearing, those with hearing loss are more affected by visual distractors than auditory distractors (Puschmann et al., 2014), are more likely to experience illusory McGurk fusions of auditory and visual phonemes (Rosemann and Thiel, 2018; Stropahl and Debener, 2017) and show larger visual enhancements of neural speech tracking (Puschmann et al., 2019). In the brain, increased visual activation correlates to the susceptibility of experiencing McGurk illusions (Rosemann and Thiel, 2018), and partial hearing loss is associated with increased connectivity between auditory and visual areas during audiovisual stimulation (Puschmann and Thiel, 2017).

Together, these studies suggest that cross-modal plasticity in partial hearing loss is more likely to result in altered visual modulation of auditory function, rather than altered visual function per se. At the neural level, visual heteromodal inputs to higher-order auditory cortical neurons may increase in strength but remain modulatory (or a mixture of modulatory and driving) in partial hearing loss because of residual auditory activity. When auditory deprivation rises to deafness, the absence of auditory input permits heteromodal inputs to fully drive auditory cortical neurons (Kral and Sharma, 2023). Visual modulation of auditory function may also occur in multisensory cortex in partial hearing loss, as shown in animal models (Schormans et al., 2017; Schormans and Allman, 2024).

It does not appear that cross-modal plasticity in partial hearing loss is associated with enhanced audiovisual abilities as is found in deafness. For example, some studies show reduced or no change in audiovisual perception in mild-to-moderate hearing loss compared to typical-hearing controls (Gieseler et al., 2018; Tye-Murray et al., 2007). In addition, research showing more frequently experienced McGurk illusion in partial hearing loss (Rosemann and Thiel, 2018; Stropahl and Debener, 2017) does not mean the visual influence is beneficial for accurate perception, as is sometimes inferred. Visual events are in fact more distracting (Puschmann et al., 2014). In rats, noise-induced hearing loss impaired perception of timing for audiovisual events (Schormans and Allman, 2019). Therefore, the putative increase in visual modulation of auditory perception with partial hearing loss is not beneficial per se. It is also possible that cross-modal plasticity benefits perceptual ability only if these abilities strengthen through experience. In other words, the visual benefits consequent of cross-modal plasticity in partial hearing loss are not spontaneous, exposure and training are needed to improve target functions. For example, adults with mild-to-moderate hearing loss who purposefully rely on visual information as a coping strategy may accrue a visual benefit over time through cross-modal plasticity, but those without long-term experience will not.