Mild cognitive impairment (MCI) is a syndrome considered as a prodromal stage of dementia characterized by a decline in cognitive cognition in one or more domains, although daily functioning may remain unaffected [1]. It is estimated that 10%–20% of people over the age of 65 years are diagnosed with MCI [2]. Some researchers suggest that MCI tends to progress toward a clinical diagnosis of Alzheimer’s disease (AD), with an annual conversion incidence rate of approximately 30% [3], with a lifetime risk ranging from 60%–90% [4]. The main clinical manifestations of MCI include memory loss, along with diminished attention, executive function, language function, visuospatial ability, and numeracy [5]. Research indicates that attention deficits predict cognitive decline in the early stages of probable AD and MCI [6]. Attention deficits in MCI patients may result in severe daily activity dysfunction [7].

Research has identified various attention functions, such as alertness, selective attention, and divided attention, which overlap with the domain of executive functions [8]. Numerous tests exist for the evaluation of attention performance at a behavioral level [9]. Thus, the Alertness, Go/No-go, Incompatibility, and Divided Attention subtests from the Test of Attention Performance (TAP, version 2.3, Psytest, Herzogenrath, North Rhine-Westphalia, Germany) appear to capture three essential and independent components of attention function in both healthy and disease states. On a neuroanatomical level, attention is linked to two primary networks, the dorsal attention network (DAN), which is involved in top-down selection of stimuli and responses, and the ventral attention network (VAN), which predominantly directs attention to salient events and might act as a “circuit breaker” for the DAN [10]. One study also considered the frontoparietal network (FPN) and cingulo-opercular network (CON) as extended components of the attention network, which are associated with executive attention [8]. The two executive attention networks act relatively independently in producing top-down control: the CON is hypothesized to maintain task-sets while the FPN is thought to facilitate rapid adjustments in adaptive control [11].

An analytic review [12] identified divided attention as a strong predictor for the progression from MCI to AD. Divided attention refers to the capacity to concentrate on multiple tasks or stimuli simultaneously, commonly termed multi-tasking or dual tasking [13]. Studies have shown that divided attention impacts multiple aspects of individuals with AD, including difficulties in simultaneity of focus on two or more related stimuli, impairments in motor tasks such as gait and postural stability, and increased risk of falls in daily activities [14, 15]. Rodda et al. [16] observed that divided attention might be the earliest aspect of attentional control impacted in early stage AD. Functional magnetic resonance imaging (fMRI) [17] and positron emission tomography (PET) studies [18] have reported that individuals with amnestic MCI (aMCI) exhibit impaired performance and decreased activation in the left prefrontal cortex when compared with normal control (NC) subjects. Alertness, a non-selective attention component, denotes a general readiness state that enhances stimulus processing and response initiation [19], fundamental to all cognitive and attentional performances [20], with the alertness network relying on frontal and parietal regions as well as the thalamus.

Selective attention [21] is the capability to filter out irrelevant stimuli and concentrate on a specific stimulus or task. It broadly covers component processes such as spatial and object selection in visual search, along with switching and response inhibition. Selective attention deteriorates linearly with AD progression [22]. A previous study suggested that patients with aMCI have subtle deficiencies in visual search (orienting) task, but their conflict resolution is similar to that of NCs [23], suggesting a potential dysfunction in selective attention. Van Dam et al. [24] conducted an fMRI study on aMCI patients and found that behavioral deficits in executive control of attention (such as enhanced conflict effects) coexist with corresponding neural deficits (reduced activation in the prefrontal and anterior cingulate cortex). Executive control of attention plays a role in complex mental tasks by monitoring and resolving conflicts, acting as a potential top-down influence on selective attention. Selective attention involves cortical networks spanning multiple regions, although their exact localization is unclear.

Although deficits in attention and executive control of attention are often among the first deficits observed following memory loss in prodromal AD, their neural underpinnings have not yet been fully elucidated. Combining behavioral studies of attention mechanisms with advanced technologies, such as functional neuroimaging, may aid in accurately identifying the pathophysiology underlying deficits linked to MCI. Resting-state fMRI (rs-fMRI) [25] is highly sensitive in detecting abnormalities in neuronal activity and function in neurodegenerative diseases. rs-fMRI measures the amplitude of the blood-oxygen level-dependent (BOLD) signal, reflecting the energy expenditure and intensity of neuronal activity. The amplitude of low-frequency fluctuation (ALFF) is a data-driven rs-fMRI-based indicator that measures the total power of the BOLD signal within the 0.01 Hz to 0.1 Hz range, revealing abnormalities in spontaneous neural activity in brain regions. The ALFF indicates changes in neural activity associated with AD spectrum disorders, aiding discrimination between healthy individuals, MCI, and AD patients [26].

Despite the growing body of research on attention in MCI, the greatest focus has been on a single aspect of attention, such as divided attention, selective attention, or alertness. To the authors’ knowledge, no study has simultaneously investigated all three aspects of attention in MCI patients. Moreover, previous conflicting results have highlighted the influence of task paradigms and condition severity, which has contributed to the current confusion of attention deficits in MCI patients. Thus, in the present study, we administered four standard paradigm subtasks of the TAP to measure aspects of attention in MCI patients. Finally, to explore neural alteration in MCI patients, we used z-scored ALFF (zALFF) to investigate changes in the attention network and aimed to address the following questions: (i) are all subtypes of attention equally impacted in MCI patients and which attention subtypes are most affected? (ii) how is the attention network altered in patients with MCI? and (iii) what is the correlation between attention deficit and the zALFF value?

Divided attention is an early detectable deficit in aMCI that can be correlated with fMRI changes [17]. This is the first study to detect a divided attention deficit in MCI that is significantly associated with the left precuneus.

Normal aging leads to decreased efficiency in intrinsic and phasic alertness, causing older adults to perform more poorly than younger individuals on related tasks [33]. Alertness deficits can substantially impact behavioral and cognitive impairments associated with MCI [34]. The current study identified a significant difference in the omissions of alertness without a warning tone between the groups, suggesting that individuals with MCI showed a reduced intrinsic alertness response compared with healthy elderly individuals. Specifically, MCI individuals were unable to maintain a preparatory response over an extended period of time. In contrast, no significant difference in phasic alertness was found (i.e., briefly focusing attention on an anticipated event) between the groups. Although there was no significant difference in reaction time for alertness, the MCI group exhibited diminished intrinsic and phasic alertness when compared with the NC group. This study supports the conclusion that reduced information processing speed is a key cognitive deficit in MCI. In the Go/No-go test, the “2 out of 5” task addresses response selection problems. This task was selected to emphasize higher memory demands, which resulted in significantly increased response latency, thereby suppressing immediate impulsive response [35]. The findings indicate that there was no significant difference in response selection between the two groups. However, whether patients with MCI exhibit more inhibitory control problems, another important aspect of selective attention, requires further investigation. The incompatibility task results align with the results of previous studies [23], indicating that inhibition control was not impaired in the MCI group when compared with the NC group. However, a study by Perry et al. [21] observed that patients with mild AD (mean MMSE = 20.4) exhibited deficits in sustained, divided, and selective attention tasks. In the early stages of the disease (mean MMSE = 26.1), patients performed relatively well on sustained and divided attention tasks, but showed deficits in selective attention tasks including response selection and inhibition, auditory target selection, and attentional switching speed. The differing findings may be related to disease severity and assessment paradigms, and further research is required to determine whether selective attention is affected in patients with MCI. Divided attention, an increasingly important clinical aspect of attention in MCI, refers to the ability to allocate cognitive resources simultaneously across multiple targets or processes. According to the capacity model, the simultaneous performance of two tasks requires a division of resources and performance can be improved by exerting a higher level of effort. The key criterion for the successful division of attention in this procedure is the omissions, as reaction times are of secondary importance. The number of visual omissions and the total number of omissions were statistically different between the two groups, suggesting a considerable limitation in attention capacity. Consistent with previous studies [17, 36], the present study confirmed that individuals with MCI have a significant deficit in divided attention. It is noteworthy that visual stimuli were most frequently omitted. Further clarification is needed to determine whether the high rate of visual omissions is entirely attributable to reduced capacity or whether it also reflects an impairment of visual processing.

Neuroimaging studies indicate that the frontal and parietal cortices play a crucial role in producing attention-related regulatory signals that feed back to the visual cortex. The attention network comprises two primary components: higher-order nodes, such as the frontal and parietal cortices, which regulate spatial attention and lower-order nodes, such as the sensory cortex, where attention-related sensory processing changes are implemented [37]. Current literature indicates that the brain regions containing pathological Tau accumulation overlap significantly with areas of the visual network (VIN) [38], particularly the occipital and lingual gyri, which are stable brain regions within the VIN [39]. The occipital lobe is the primary processing area for visual stimuli in the brain and mainly responsible for object recognition, including the functional properties of objects. The middle occipital gyrus is related to spatial perception around the human body [40]. In MCI patients, reduced activation has been mainly observed in the FPN, whereas AD patients exhibit more hypoactivation in the VIN, and inefficient VIN activation may disrupt higher cognitive processing in AD patients [41]. The current study observed increased zALFF values in the left middle occipital gyrus in the MCI group when compared with the NC group. A meta-analysis [31] revealed that patients with aMCI exhibited higher ALFFs in the right lingual gyrus, left middle occipital gyrus, left hippocampus, and left inferior temporal gyrus when compared with NC groups. Concordant with prior findings, we identified that there were functional impairments and compensatory mechanisms in the left middle occipital gyrus. We also found that zALFF values increased in the left superior orbitofrontal gyrus and right orbitofrontal inferior gyrus in the MCI group. These two brain regions are key parts of the prefrontal cortex and are typically involved in cognitive functions including attention, inhibitory control, habit formation and working, and spatial or long-term memory [42]. A previous rs-fMRI study in AD has identified abnormal functional connectivity in the prefrontal cortex, a region believed to exert top-down control by connecting various brain regions to facilitate complex cognitive processes [43]. According to prior research [44], increased zALFF values in these brain regions might reflect neuronal compensation and/or amyloid-$\beta$-induced excitoxicity, suggesting that more neural resources are recruited to sustain task performance in individuals with MCI.

Another important finding of this study was that zALFF values were reduced in the left middle cingulate gyrus (Brodmann Area 24) and left precuneus, which are key brain regions of the FPN. An imaging study has implicated broad areas of the FPN as contributors to attentional control across various forms of top-down selection [45]. The cingulate cortex is involved in inhibition control, behavioral correction, and emotional processing regulation. Increased functional connectivity between the hippocampus and cingulate gyrus has been detected prior to MCI [46]. The middle cingulate cortex is associated with higher-order functions including attention allocation, reward, decision-making, morality and ethics, impulse control (e.g., error detection and correction), and emotion processing. The middle cingulate cortex is connected to the prefrontal cortex, parietal lobe, and motor system, serving as a hub for “top-down and bottom-up attention” primarily for response selection and feedback-driven decision making. A study on ALFF and fractional ALFF within the Alzheimer Spectrum [26] revealed a significant decreasing trend of values across several brain regions, including the right MCC, bilateral inferior cerebellum lobe, and bilateral precuneus in ALFF/fALFF. Cera et al. [47] discovered that cognitive decline in MCI impacts the global functional connectivity of the cingulate cortex, particularly the dorsal posterior MCC, with a significant difference in the functional connectivity pattern when compared with healthy elderly individuals. A neuroimaging study [48] indicates that the MCC is crucial for cognition and attention, becoming active during cognitively demanding tasks that require stimulus response selection in the presence of competing streams of information such as divided attention and work-memory tasks. The precuneus, located in the posterior parietal cortex, is involved in a wide range of highly integrated tasks such as visuo-spatial imagery, episodic memory retrieval, and self-processing operations [49]. Lower functional connectivity has been observed between the precuneus and retrosplenial cortex in patients with subjective cognitive decline [50]. A meta-regression analysis has shown that the increased severity of cognitive impairment in aMCI patients was associated with greater decreases in ALFFs in the cuneus/precuneus cortices [31]. Teipel et al. [51] reported reduced cortical activation in cerebellar and posterior cortical areas, such as the parietal cortex, precuneus, posterior lateral temporal cortex, and parahippocampal gyrus, during working memory performance in AD/MCI patients when compared with healthy older adults. Brain activation patterns in healthy older adults have also revealed that divided attention is associated with substantial activation within the precuneus and posterior parietal regions [52]. Functional neuroimaging studies have shown that episodic memory processes not only require medial temporal lobe activity, but also attention functions [53]. This is consistent with the results reported here, suggesting that attention is crucial in episodic memory processes. Additionally, decreased zALFF values in the left precuneus were significantly negatively correlated with divided attention. These findings suggest that divided attention might be significantly enhanced in patients with MCI by modulating the function of the left precuneus. Further, we found decreased zALFF values in the right rolandic operculum of the MCI group. The rolandic operculum, situated at the junction of the frontal-parietal lobes, potentially acts as a key hub for information transfer between the frontal, parietal, and temporal lobes in bottom-up attention modulation. These changes may directly imply reduced spontaneous neural activity in these regions, which may be due to neurophysiological processes, indicating a functional deficit or downregulation of excitability.

The findings suggest that patients with MCI exhibit relatively normal performance on selective attention and phasic alertness tasks, but show a reduced capacity for divided attention and intrinsic alertness tasks. The decrement in capacity for divided attention in MCI patients might be associated with abnormalities in spontaneous neural activity in the left precuneus. This study offers novel insights into the neural basis of divided attention in MCI patients, offering a valuable reference for understanding attention alteration and alterations in the attention network, which might contribute to the early diagnosis and prevention of MCI.

The datasets used and/or analyzed for this study are available on request to the corresponding authors.

JL, YZ, YWY designed the research study and wrote the manuscript. JJZ, WLC, HYX collected data. YPX and QRX analyzed data. SXL and JH conceptualized and designed the study and give final approval of the version to be published. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The studies involving human participants was approved by the Ethics Committee of the Third People’s Hospital Affiliated to Fujian University of Traditional Chinese Medicine (Ethic Approval Number: 2023-kl-017). The study was carried out in accordance with the guidelines of the Declaration of Helsinki, and all participants had signed informed consent.

This research was funded by the Natural Science Foundation of Fujian Province (2020J01754) and School Management Project of Fujian University of Traditional Chinese Medicine, Program Number: X2024004 and X2024003.

The authors declare no conflict of interest.