Various visual tests have been conducted in AD patients to assess contrast sensitivity, motion detection, visual field topography, visual acuity, and color vision (Figure 1). Although there are some discrepancies regarding some specific symptoms across different studies, it is generally agreed that some visual functions are impaired in AD patients compared to age-matched healthy controls (Bublak et al., 2011). Meanwhile, higher-order visual functions, such as contrast sensitivity and conscious moving object detection, appear to be more affected compared to basic visual functions such as visual acuity. See Table 1 for the prevalence.

Impaired central auditory processing has been found at early and mid-phases of AD (Strouse et al., 1995), often preceding the onset of clinical dementia (Gates et al., 2002). Meanwhile, hearing loss has been proposed as a significant risk factor for the development of AD (Griffiths et al., 2020), and the use of hearing aid shows some preventive effect (Livingston et al., 2020). Even for subclinical hearing loss, a study of 6,451 participants observed that cognitive function decreases with every 10 dB reduction in hearing (Golub et al., 2020). See Table 1 for the prevalence.

Alzheimer’s disease pathology, including amyloid plaques, neurofibrillary tangles and neurodegeneration, has been observed in the human occipital cortex (Braak et al., 1989; Armstrong et al., 1990; McKee et al., 2006; Crutch et al., 2012). Regarding the laminar distribution, some studies have demonstrated that layer (L)2/3 has the highest density of Aβ plaques (Duyckaerts et al., 1986; Braak et al., 1989; Armstrong, 2012), while others have reported high-density deposition L4 (Rafalowska et al., 1988), and L5 (Lewis et al., 1987). In early onset AD cases, the senile plaques are more frequently observed in the upper layers of the occipital cortex, and a change of visually evoked responses is suggested to be used as a diagnostic marker for early onset AD (Armstrong et al., 1990). Meanwhile, it has been consistently reported that the visual association area has a higher number of Aβ plaques compared to the primary visual area (Armstrong, 2012).

The neurofibrillary tangle (NFT) formation has also been observed in the visual cortical areas of AD brains (Lewis et al., 1987; McKee et al., 2006), although with a lower prevalence as compared to Aβ (27 vs. 72% of AD brains) (Armstrong et al., 1990). The NFT formation shows the highest density in L2/3 (Armstrong, 2012) and L5 (Lewis et al., 1987), and is significantly more abundant in visual association areas than the primary visual area (Lewis et al., 1987; Braak et al., 1989; Armstrong et al., 1990; Giannakopoulos et al., 2000; Armstrong, 2012). The difference is reported to be as high as 20-fold (Lewis et al., 1987). Another study (Leuba et al., 2001) has reported that the NFT formation is almost absent in the primary visual area in mild cognitive impairment (MCI), which is considered a precursor to AD.

Besides Aβ deposition and NFT formation, other pathological changes have also been reported in the visual cortex of AD brains. Extensive spine loss has been observed, together with pathology of apical and horizontal dendrites of neurons (Mavroudis et al., 2011). Cell loss has been reported with divergent vulnerability (Geneviève and Rudolf, 1994), although to a much lower degree compared to regions such as prefrontal and temporal association areas (Frisoni et al., 2007). Loss of SMI32-immunoreactive pyramidal neurons has been found in L4 of the primary visual cortex and L3 of the secondary visual cortex, while inhibitory interneurons, including parvalbumin, calretinin, and calbindin immunoreactive neurons, do not show any significant decrease in primary or secondary visual cortex (Leuba et al., 1998). Degeneration of cholinergic fibers in the visual cortex is not conclusive, as no deficit (Geula and Mesulam, 1996) and large depletion (Beach et al., 1992) have both been reported. Furthermore, gliosis occurs in the visual cortex, being the heaviest in primary visual cortex layers 2, 3, and 4 (Beach et al., 1988), and is potentially correlated with the plaque deposition (Geneviève and Rudolf, 1994). Neuropil threads without NFT formation have also been observed in L5 of the visual cortex (Braak et al., 1989). Moreover, inhibitory synapse loss is observed in the occipital lobe of AD patients, potentially contributing to the disruption of excitatory/inhibitory balance (Kurucu et al., 2022).

Visual association area has been suggested as one of the earliest brain regions showing pathological changes (senile plagues and NFT) in AD. The early display of pathology in the occipital cortex or its cortical thinning may be a potential biomarker for AD (Hwang et al., 2021). Studies using fMRI have suggested that higher-order visual association areas exhibit functional deficits earlier than the lower-order primary visual area (Huang et al., 2021). The early manifestation of AD pathology in the association cortex highlights its vulnerability (McKee et al., 2006). In comparison, the primary visual cortex is relatively spared (Brun and Gustafson, 1976; Cronin-Golomb et al., 1991; Iseri et al., 2006; Jack et al., 2008; Grothe et al., 2017). Nevertheless, an increase in the power of gamma oscillations, which are often associated with cognitive processes (Murty et al., 2018), has been observed in the primary visual cortex of AD patients during a visuospatial processing task (Wiesman et al., 2021b), suggesting functional deficits.

While there have been plenty of studies showing the presence of AD pathology in the visual cortex, how it contributes to the visual impairments remains largely unclear. Occipital atrophy has been associated with the visual hallucination (Holroyd et al., 2000) and the visual field deficit in AD (Armstrong, 1996). It has also been proposed that Aβ deposition leads to hyperactivity of the visual cortex, disrupting orientation tuning, functional connectivity and visual memory (Grienberger et al., 2012; L’Esperance et al., 2023; Niraula et al., 2023). A diminished cross-modal effect by activation of visual cortex observed in AD patients during a visual cue dependent task further suggests altered interactions between visual cortex and other cortical areas (Drzezga et al., 2005). Understanding how visual functional connectivity is altered may allow for specific impairments such as visuospatial dysfunction to serve as a potential biomarker for mild cognitive impairment (Yamasaki et al., 2012) and facilitate the development of non-invasive therapeutics (Martorell et al., 2019).

Pathological changes in auditory pathways, including the auditory cortex, have been reported in AD patients (Sinha et al., 1993). In general, the changes in the auditory cortex share some similarities with the visual cortex. First, Aβ plaques are prevalent, as they have been reported in up to 89% of AD brains (Armstrong et al., 1990). Layers 2 and 3 show the highest density of Aβ deposition (Duyckaerts et al., 1986; Rafalowska et al., 1988). The associative area exhibits more severe pathological changes than the primary area (Esiri et al., 1986). The Aβ oligomer, closely correlated with the plaque deposition, has also been observed in the auditory cortex (Savioz et al., 2016). However, the pathological difference between primary and associative cortices may not correlate with clinical symptoms, as shown by a study comparing auditory performance associated with the primary auditory cortex (sound localization and perception of complex tones) and that associated with auditory association cortex (phoneme discrimination and tonal memory) (Kurylo et al., 1993).

Tau pathology is observed in the auditory cortex, being more severe in the associative area than the primary area (Esiri et al., 1986; Lewis et al., 1987; Arnold et al., 1991). Interestingly, chondroitin sulfate proteoglycans show high-level expression in the primary auditory cortex, but at lower levels in association areas, leading to a hypothesis that higher levels of these extracellular matrix proteins may be associated with less tau pathology (Brückner et al., 1999).

Neuronal loss is also a prominent pathological marker in the temporal cortex. Significantly decreased gray matter density has been found in the auditory association cortex but not the primary auditory cortex of AD patients (Aylward et al., 2020). Late-onset AD results in much greater atrophy of the temporal lobe, especially in the superior temporal gyrus, than early onset AD (Frisoni et al., 2007). A study using Golgi staining in the auditory cortex of AD brains has found that layer 1 Cajal–Rezis cells exhibit significant cell loss, together with dendritic loss and distortion (Baloyannis et al., 2007). Regarding the interneuron, it has been suggested that somatostatin-positive neurons decrease in the temporal cortex in AD, but not parvalbumin-positive neurons (Waller et al., 2020). The choline acetyltransferase activity, which is indicative of cholinergic neuronal transmission, shows a significant decrease in the auditory cortex (Esiri et al., 1990).

Gliosis has been observed in the auditory cortex of AD brains (Styren et al., 1998). In the lateral temporal cortex, the astrocytic glial fibrillary acidic protein (GFAP) increases with AD severity and correlates with amyloid and tau pathology (Buchanan et al., 2020), suggesting that gliosis could be a robust indicator of disease progression.

The dysfunction of several neurotransmitter systems has been demonstrated in the AD temporal cortex. Reduction of serotonin (5-HT) 2A receptors has been found to correlate with the decline in cognitive performance (Lai et al., 2005). Deficits of nicotinic receptors have been reported, although no agreement has been reached regarding the specific subunits that are impaired (Guan et al., 2001; Martin-Ruiz et al., 2002). GABA-related transcriptions, including the receptors and GABA-synthesizing enzymes, are downregulated in the temporal cortex (Govindpani et al., 2020), and the GABAA receptor α5 subunit is found to decrease based on immunostaining (Kwakowsky et al., 2018). The glutamate transporter-1 (GLT-1) in astrocytes, which is responsible for removing excessive glutamate, is significantly reduced (Hoshi et al., 2018), suggesting that glutamate neurotoxicity may play an important role in the neurodegeneration of AD. Moreover, postsynaptic density protein 95 (PSD-95), but not presynaptic synaptophysin, decreases progressively in the temporal cortex across the Braak stages of AD (Buchanan et al., 2020).

It should be noted that many of the studies have not specifically targeted the auditory cortex, but rather the larger temporal cortex (Armstrong et al., 1990; Waller et al., 2020). It is possible that some of the observations of neuronal loss and neurotransmission disruptions might be in fact in regions beyond the auditory cortex. Moreover, the relationship of AD pathology in auditory cortex to hearing dysfunction is still under debate.

Amyloid beta and tau pathology has been observed in the olfactory bulb of individuals with AD as well as MCI (Kovács et al., 1999; Tsuboi et al., 2003; Attems and Jellinger, 2006; Albers et al., 2015). AD patients also show prominent atrophy in the primary olfactory cortex (Al-Otaibi et al., 2020). Meanwhile, fMRI studies have reported decreased activation of the primary olfactory cortex in AD patients passively presented with odors, while some other studies have found increased activation in early AD (Meadowcroft et al., 2019). It has been proposed that AD-related pathology begins within olfaction-related structures and subsequently spreads to additional brain regions and ultimately encompasses multiple areas of the brain (Braak and Braak, 1991; Claire, 2019).

Amyloid and tau pathology has also been observed in the somatosensory cortex in AD, and the pathological changes are more pronounced in the association area than the primary region (Arnold et al., 1991). On the other hand, cholinergic fibers remain relatively preserved in the somatosensory cortex of AD brains (Geula and Mesulam, 1996). Moreover, abnormal magnetoencephalography responses to somatosensory stimulation have been found in the primary somatosensory cortex in MCI (Stephen et al., 2010).

Overall, the cortical pathological changes share some similarities across different sensory modalities. The changes are in general more severe in sensory association areas than the primary sensory cortices, with superficial cortical layers exhibiting the most pronounced changes. This spatial bias of the pathological progression might possibly explain why the basic sensory functions (e.g., visual acuity, auditory frequency discrimination) are largely spared, while higher sensory functions (e.g., visual contrast sensitivity, hearing in a relatively noisy environment) exhibit deficits with early onsets.

Visual behavior has been examined in the 5xFAD mouse, which overexpresses human amyloid precursor protein (APP) and presenilin 1 (PS1), harboring five familial AD mutations and rapidly develops amyloid pathology. A study using an optomotor test, which measures the head movement following the direction of drifting gratings at various spatial frequencies, has found impaired visual behavior suggesting impaired contrast sensitivity at 6 months of age (Zhang et al., 2021). In addition, visual evoked potential (VEP) recording in the visual cortex has revealed abnormalities in visual acuity in young 5xFAD mice (Criscuolo et al., 2018). In rTg4510 mice, which exhibit tauopathy, the visually evoked pupil dilation was completely disrupted (Papanikolaou et al., 2022). Other visual related behavioral changes would require a further comprehensive characterization.

Hearing loss has been reported in the APP/PS1 mouse, which expresses a chimeric mouse/human amyloid precursor protein and a mutant human presenilin 1. As assessed by auditory brainstem response (ABR), hearing loss first appears at high frequency ranges as early as 2 months old and then extends to low frequency ranges. By 3–4 months of age, hearing loss is observed within the whole-frequency range (Liu et al., 2020).

In the 5xFAD mouse, it has been suggested that central gain increases preceding the reduction of ABR amplitudes (Na et al., 2023). The acoustic startle response (ASR) starts to decline at the age of 3–4 months, and ABR threshold increases at 13–14 months of age (O’Leary et al., 2017). The latency of ASR is prolonged as well (Story et al., 2019). In the TgCRND8 (overexpressing mutant human APP) mouse, the ASR consistently increases starting from 10-weeks old (McCool et al., 2003).

In the prepulse inhibition (PPI) test, APP/PS1 and TgCRND8 mice show a reduction in PPI at 7- and 22-month-old, respectively (McCool et al., 2003; Wang et al., 2012). However, in the P301S (mutant tau protein) mouse model, PPI is enhanced (Takeuchi et al., 2011).

The gap detection test has revealed temporal processing impairments in 5xFAD mice. The gap detection deficit is evident as early as approximately 2 months old and worsens over time. Interestingly, the impairment exhibits sex dimorphism, being more prominent and earlier (2-month) in males than females (Kaylegian et al., 2019; Weible et al., 2020). Meanwhile, both the neuronal responses to silent gaps in continuous white noise and the spontaneous firing in the auditory cortex are progressively reduced (Weible et al., 2020).

Together, these identified hearing deficits are suggested to serve as a potential early non-invasive detection biomarker in AD mouse models, since they exhibit an onset much earlier than cognitive impairments such as spatial learning deficits (Liu et al., 2020). Further research is necessary to explore the underlying neural mechanisms and determine whether the findings can be translated to human AD cases.

Amyloid plaques have been observed in the visual cortex of 5xFAD and APP/PS1 mice, including the primary and secondary visual cortex (Whitesell et al., 2019; Zhang et al., 2021; Tsui et al., 2022). In the 5xFAD mouse, Aβ pathology starts from layer 5 of the cortex, due to the usage of Thy1 as the promotor, at 2 months old (Oakley et al., 2006). At 9 months old, Aβ deposition can be observed in layers 4, 5, and 6 (Tsui et al., 2022). Gliosis surrounding the Aβ plaques has been found as well (Soula et al., 2023). In the rTg4510 transgenic mouse, which is a tauopathy model, the neurofibrillary tangles emerge in the visual cortex between 2 and 4 months of age. Meanwhile, short- and long-term visual plasticity, as assessed by local field potential responses to a repeated visual stimulus, are disrupted at both early (5-month) and late (8-month) stages of tauopathy (Papanikolaou et al., 2020). These results indicate that tau pathology can also affect intrinsic cortical plasticity. A relatively comprehensive investigation of the post-weaning developmental stages in the 5xFAD mouse has shown that APP overloading occurs in L5 pyramidal neurons of the primary visual cortex during the critical period (4–5 weeks) for visual cortical plasticity (Chen et al., 2022).

Functional changes in the visual cortex have also been investigated. Using in vivo two photon calcium imaging, an increase in spontaneous activity and a reduction in visual and motor triggered signals in V1 have been observed in APP/PS1 mouse (Liebscher et al., 2016). In hAPP mouse, hyperactivity in visual cortex has also been observed, likely attributed to the increased excitatory-inhibitory synapse ratio (Niraula et al., 2023). Meanwhile, in the hAPP mouse, the visually evoked activity increased, which is different from the observation in APP/PS1 mouse (Niraula et al., 2023). In rTg4510 transgenic mouse, which develops tauopathy, the visual plasticity were found disrupted at early stages in V1 (Papanikolaou et al., 2022).

Similar to visual cortex, in the 5xFAD mouse, amyloid plaques are found in auditory cortex, including the primary and ventral (secondary) auditory cortex (Tsui et al., 2022; Weible and Wehr, 2022). At 9 months old, Aβ deposition is observed in layers 4, 5, and 6 (Tsui et al., 2022). It is worth noting that Aβ deposition profiles are variable between different animal models. For example, in APP/PS1 mice, Aβ plaques are observed more in the dorsal auditory cortex (Whitesell et al., 2019).

In the 3xTg-AD mouse, which is a triple transgenic model containing three mutations associated with familial Alzheimer’s disease, it has been found that hearing loss induced by noise exposure before the disease phenotype is manifested causes persistent synaptic and morphological alterations in the auditory cortex (Paciello et al., 2021). This is associated with earlier increased tau phosphorylation, neuroinflammation, and redox imbalance. The results suggest that hearing loss could potentially accelerate the neurodegeneration onset.

In the olfactory system of the APP/PS1 model, Aβ deposition has been observed in the olfactory epithelium, olfactory bulb and olfactory cortex (Wu et al., 2013). In the Tg2576 mouse that overexpresses a mutant form of APP with the Swedish mutation, non-fibrillar Aβ deposition has been found within the olfactory bulb at 3 months of age, earlier than the deposition within any other brain region, and found to be in correlation with olfactory deficits (Wesson et al., 2010). This suggests that non-fibrillar rather than fibrillar Aβ-related mechanisms might contribute to early olfactory perceptual loss in AD. However, it shall be noted that, the Tg2576 mouse model exhibits non-fibrillar Aβ deposition not only in the olfactory bulb (Wesson et al., 2010) but also in the hippocampus (Alcantara-Gonzalez et al., 2021) as early as 2–3 months of age.

In the 5xFAD mouse, amyloid plaques can be observed in the somatosensory cortex (Ali et al., 2019), including primary and secondary somatosensory cortex. At 9 months old, Aβ deposition is observed in layers 4, 5, and 6 of the somatosensory cortex (Tsui et al., 2022). Plaques are more and larger with increasing cortical depths (Ali et al., 2019). There is significant spine loss on basal dendrites of neurons in somatosensory cortices of 6-month-old females (Crowe and Ellis-Davies, 2014). Regarding the interneuron, parvalbumin neurons show a significant loss at 6–9 months old, which is more prominent in the deeper layers (Ali et al., 2019). In the barrel cortex of APP/PS1 mice, Aβ pathology exhibits some spatial specificity, with plaques more concentrated in septal areas than barrels (Beker et al., 2012). Using voltage-sensitive dye imaging, a study has found abnormal sensory responses evoked by whisker deflections in the barrel cortex: both the amplitude and spatial spread of the responses are larger in transgenic than in control mice (Maatuf et al., 2016).

Hyperactivity in both neurons and astrocytes brain has been documented in AD brains (Busche et al., 2008; Kuchibhotla et al., 2009; Wright et al., 2013; Busche and Konnerth, 2015; Targa Dias Anastacio et al., 2022), and the hyperactive neurons are found spatially close to amyloid beta deposition in the double transgenic APP23xPS45 mouse (Busche et al., 2008). Moreover, it has been proposed that hyperactivity could induce synaptic dysfunction and loss at early stages of AD (Busche et al., 2008). Regarding sensory processing, hyperactivity increases background neural activity and thus reduces signal-to-noise ratio of sensory responses, which may impair the neural circuit’s ability to accurately distinguish relevant sensory information from irrelevant background noise.

Reduced neuronal activity and decreased metabolism have also been observed in AD pathology (Silverman et al., 2001; Busche et al., 2008). A synaptic failure theory is proposed to explain this functional deficit (Selkoe, 2002). Soluble tau has been suggested to be able to suppress neuronal activity in an in vivo two-photon Ca²⁺ imaging study (Busche et al., 2019). In sensory systems, reduced neuronal activity may directly result in functional impairment. For example, the reduced visually evoked activity in V1 in APP/PS1 mice (Liebscher et al., 2016) is associated with visuomotor integration deficits in AD.

The synaptic loss is highly correlated with functional deficits in AD (Terry et al., 1991) and is proposed as an early event (Masliah et al., 1990). Microglia may be a major player through synapse pruning (Rajendran and Paolicelli, 2018). Oligomeric A-beta species may impair the function and structure of synapses as well (Sheng et al., 2012). The synaptic loss can also lead to diminished functional connectivity between cortical neurons and cortical areas. This may underlie various sensory processing deficits observed in AD.

Inhibitory neurons are particularly susceptible to AD (Petrache et al., 2019), leading to a decrease in the inhibitory tone. In APP23 mice, application of diazepam reduces the activity of hyperactive neurons, suggesting that reduced inhibition underlies hyperactivity (Busche et al., 2008). The reduced inhibition results in hyperexcitability of excitatory neurons, which may not only interfere with normal sensory signal processing but also exacerbate neuronal degeneration. Inhibitory neurons themselves may exhibit different activity changes. In the APP/PS1 mouse, cortical L2/3 somatostatin neurons in proximity to amyloid beta show increased activity, while parvalbumin neurons show decreased activity (Algamal et al., 2022). In the hippocampus, it has been shown Aβ-induced hyperexcitability of parvalbumin neurons leads to memory impairment although no changes are observed yet in intrinsic properties of pyramidal neurons (Hijazi et al., 2020). Thus, alterations in excitatory-inhibitory balance may greatly affect sensory processing in the cortex (Wood et al., 2017).

While primary cortices appear to be relatively spared from amyloid plaques and NFT formation, soluble amyloid-beta and tau have been proposed to be able to induce hyperactivity (Kopeikina et al., 2012; Hector and Brouillette, 2021), adding another layer to the complex pathogenesis of AD.

The death of neurons within sensory cortices can result in the disintegration of circuits necessary for normal sensory information processing, contributing to sensory deficits.

Peripheral deficits can result in a reduction of sensory information relayed to cortical neurons, further exacerbating the sensory deficit.

It has also been proposed that pathological changes in other central regions may contribute to sensory deficits. For instance, deficits in the CA1 region of the hippocampus have been suggested to contribute to deficits in the acoustic startle response in the Tg4-42 mouse (Sichler et al., 2019).

In AD patients, cross-model suppression is progressively impaired (Drzezga et al., 2005), which may contribute to the dysfunction of sensory processing.