As described by McIsaac et al. (2015), DT refers to the concurrent performance of two tasks that can be executed independently and measured separately, having distinct goals. DT is typically categorized into three types, based on the nature of the tasks being performed: motor-motor, cognitive-cognitive, and cognitive-motor dual-task (Piqueres-Juan et al., 2021). Motor-motor and cognitive-cognitive DT involves two tasks performed concurrently, whether motor (e.g., as walking while playing an instrument) or cognitive (e.g., counting backward while categorizing figures), respectively. CMDT involves one motor task and one cognitive task performed concurrently, such as balancing while engaging in working memory tasks. While all three configurations appear frequently in daily contexts, recent research has predominantly focused on CMDT, which represents a critical domain for understanding how cognitive functions affect motor performance and vice versa (Li et al., 2018; Postigo-Alonso et al., 2019). CMDT requires sophisticated integration of multiple complex skills, including strategic planning, task implementation, performance monitoring, working memory allocation, and attentional modulation. However, the cognitive capacity to allocate resources across multiple tasks is inherently limited, resulting in performance decrements that manifest as reduced performance quality and increased task completion time, regardless of individual competencies (MacPherson, 2018). These issues become even more notable with aging, where older adults require significantly more attentional resources to complete tasks due to both cognitive and motor decline (García-López et al., 2023; Tuena et al., 2023; Versi et al., 2022), as supported by neuroimaging evidence revealing age-related structural and functional changes, particularly in frontal regions, that impact higher-level cognitive functions (Andrews-Hanna et al., 2007; Grady, 2012; Raz et al., 2005). These cognitive decrements correlate with impaired mobility parameters, including gait velocity reduction and fall risk (Yogev-Seligmann et al., 2008), potentially compromising functional independence (Wajda et al., 2017). In pathological conditions, cognitive-motor interference exceeds isolated domains (Li et al., 2018; Wollesen and Voelcker-Rehage, 2014). For instance, disorders related to pathological aging (e.g., Parkinson's disease and Mild Cognitive Impairment) demonstrate particularly pronounced DT costs, with patients exhibiting disproportionate performance deterioration compared to healthy age-matched controls due to compromised neural networks and impaired automaticity of motor and cognitive processes (Al-Yahya et al., 2011; Yogev-Seligmann et al., 2008).

Given these clinical implications, an accurate assessment of CMDT performance becomes crucial. Typically, assessment design integrates both motor detections (e.g., spatiotemporal gait parameters during standardized walking tasks) and cognitive challenges under single and DT conditions. Established protocols include the cognitive Timed Up and Go test (cognitive-TUG), where participants perform the standard Timed Up and Go test (Podsiadlo and Richardson, 1991) while engaging in cognitive tasks such as counting backward, generating animal names, or spelling given words backward (Kasiukiewicz et al., 2021; Shumway-Cook et al., 2000). Another widely used protocol is the Walking While Talking test, involving walking at a preferred pace during concurrent cognitive tasks of varying complexity (Yogev-Seligmann et al., 2008). Performance variations in both domains (i.e., DT cost)—manifested as the difference in performance between single-task and DT conditions—provide quantitative indices of cognitive-motor interference. The assessment of CMDT offers valuable insights for the detection of emerging cognitive-motor problems and facilitates timely individualized intervention (Mancioppi et al., 2021). This assessment foundation translates into therapeutic applications, as CMDT performance demonstrates significant trainability through targeted interventions.

The CMDT training, integrating simultaneous performance of cognitive tasks (e.g., verbal fluency, memory exercises) with motor activities (walking, balance exercises) (Fritz et al., 2015; Piqueres-Juan et al., 2021; Tuena et al., 2023), has proven as a valuable approach for enhancing cognitive and motor functions across healthy aging and pathological populations (Wiśniowska et al., 2023; Wongcharoen et al., 2017; García-López et al., 2023; Li et al., 2020; Tait et al., 2017; Versi et al., 2022). Despite literature providing several methods to deliver CMDT (i.e., sequential or simultaneous task performance), simultaneous training, performing motor and cognitive tasks at the same time, seems to be the most effective in improving balance, gait, and reducing fall risks in older adults with chronic conditions (Spanò et al., 2022). The effectiveness of these interventions is attributed to the activation and strengthening of compensatory neural mechanisms, where alternative neural circuits develop to support cognitive-motor function despite age-related changes (Miura et al., 2024; Park and Reuter-Lorenz, 2009). However, current CMDT training approaches face several significant limitations that restrict their clinical translation and effectiveness, such as the optimal dose-response. This parameter remains undefined, with training protocol varying widely from 2–3 sessions per week lasting 30–60 min over a period of 4–12 weeks (Forte et al., 2023; Petrigna et al., 2025). Additionally, these limitations extend to CMDT evaluation methodologies, where there is no consensus on an optimal evaluation method, with protocols varying considerably across studies in terms of task selection and outcome measures, limiting meaningful comparison of results and clinical applicability (Leone et al., 2015). Most CMDT protocols also remain anchored to traditional methods with minimal incorporation of advanced technologies that could enhance engagement, precision, and objective monitoring. Moreover, actual CMDT options may fail to capture real abilities in an ecological context because the stimuli often do not reflect real-life activities.

During the history of neuropsychology, attention has shifted from traditional paper-and-pencil methods to computer-based programs, offering multimodal and customizable instruments (Hu et al., 2021). Contemporary approaches fundamentally maintain the same theoretical foundations as conventional training, with the primary innovation being the integration of gamification elements and interactive technologies that enhance engagement and motivation (Bieryla and Dold, 2013; Eggenberger et al., 2016; Hu et al., 2021; Nouchi et al., 2012; Rendon et al., 2012). Among these technological advances, VR emerges as a particularly transformative tool for CMDT, offering a unique advantage in creating realistic tasks while enabling multisensory bodily interactions (Riva et al., 2018). The effectiveness of VR depends on several key features: the sense of presence within the environment (the sensation of “being there”), agency (active participation), and immersion (sensory fidelity level) (Cipresso et al., 2018; Riva et al., 2020; Slater, 2009). Based on these features, the literature conceptualized VR along a continuum of levels (Borghesi et al., 2022): fully immersive systems (e.g., Cave environments with body tracking, head-mounted displays) offering complete isolation from the real world and direct virtual interaction; semi-immersive systems (e.g., large-screen projections with tracking capabilities) providing moderate immersion; and non-immersive systems (e.g., traditional screens with motion sensors, console games with motion controllers) delivering a degree of immersion and interaction beyond traditional screens yet fall short of fully immersive experiences.

Whatever its form, VR offers significant advantages to address limitations in traditional CMDT: reproducible, ecologically valid evaluation environments, such as virtual supermarkets, home environments (Da Costa et al., 2022; Raspelli et al., 2011), or virtual towns (Da Costa et al., 2022) that can be deployed across different clinical settings and aging populations. This reproducibility provides the infrastructure necessary for standardization, which requires researchers to converge on shared protocols beyond technical implementation. VR systems enable automated, objective performance tracking that captures both cognitive and motor parameters simultaneously, reducing measurement variability and facilitating precise DT cost calculations (Tuena et al., 2023; Wei et al., 2025). Moreover, VR enhances cost and temporal efficiencies, safety protocols, and accessibility (Wei et al., 2025) while creating engaging scenarios that promote transfer to daily activities within controlled laboratory conditions (Yun et al., 2023). Finally, when training is delivered via VR-based interventions, it demonstrates significant improvements in outcomes compared to traditional training methods (Wei et al., 2025), indicating not only methodological advantages but also enhanced clinical effectiveness. Nevertheless, VR implementation faces important practical challenges, particularly for older populations. Cybersickness (i.e., nausea, dizziness, disorientation, and general discomfort, depending on system characteristics) may potentially pose a risk to older adults due to age-related vestibular and visual changes (Rebenitsch and Owen, 2016). Hardware cost and technical complexity can limit accessibility for resource-constrained settings and home-based interventions (Wei et al., 2025). Technological literacy barriers may require extended familiarization and ongoing support, while safety concerns include fall risk during immersive experiences for individuals with balance impairments. Usability issues may further compromise feasibility (Realdon et al., 2019; Tuena et al., 2023). These considerations necessitate careful system selection and user-centered design that prioritize safety, comfort, and accessibility alongside technical capabilities.

Recent systematic reviews examined CMDT interventions assisted with technologies, focusing either on aging population with chronic diseases (Tuena et al., 2023) or on healthy older adults (Wei et al., 2025). However, these reviews have focused predominantly on training applications, with CMDT assessment methodologies receiving comparatively less systematic attention despite their crucial role in designing tailored interventions (Choi et al., 2025; Tuena et al., 2023; Wei et al., 2025). Furthermore, literature also lacks a systematic comparison across different levels of VR immersion (non-immersive, semi-immersive, and fully immersive) within the CMDT context. Moreover, emerging technological innovation (such as 360° media), which has shown promise in other cognitive assessment domains (Mancuso et al., 2024), may offer an additional solution for CMDT applications. Although their current use in this context remains unclear. The present scoping review addresses these gaps by mapping both assessment and intervention applications across the full aging spectrum (≥65), with particular attention to: (i) characterizing the spectrum of VR immersion levels currently employed (non-immersive, semi-immersive, fully immersive) and their relationship with ecological validity, (ii) providing a comprehensive view integrating both assessment and training applications of VR-based CMDT, and (iii) identifying which technological approaches are currently utilized and which promising innovations remain unexplored, as well as methodological inconsistencies that limit clinical translation.

This scoping review was conducted on April 17, 2023 (and updated on January 21, 2025) and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines (Tricco et al., 2018). No protocol was registered in advance, consistent with the exploratory nature of scoping reviews (Munn et al., 2018; Tricco et al., 2018). We performed a computer-based search in three databases: PubMed, Web of Science, and Embase. These databases were selected to provide multidisciplinary coverage relevant to VR-based interventions in aging populations. Web of Science indexes content across psychology, engineering, and rehabilitation sciences; Embase provides comprehensive coverage of biomedical and allied health journals; and PubMed ensures thorough capture of medical and clinical research. The literature review methodology incorporated two distinct search strings, each with a specific focus. The first tackled assessment methodologies, while the second addressed rehabilitation strategies:

The term “patient” was strategically included alongside aging-related keywords to encompass physiological aging and older clinical populations, aligning with established conventions in scientific literature. Similarly, we included terminology related to 360° media in our search strategy, although this represents an emerging frontier in this field. This inclusive approach ensured comprehensive coverage of literature, preventing the omission of relevant research.

The search strings were applied to title and abstract fields in PubMed and Embase, and to the Topic field (which includes title, abstract, author keywords, and Keywords Plus) in Web of Science. Search strategies were adapted to each database's specific syntax while maintaining semantic equivalence: in PubMed, we used field tags [Title/Abstract]; in Embase, we applied.tw. (text word) for title, abstract, and keyword fields; in Web of Science, we searched the Topic field, which includes title, abstract, author keywords, and Keywords.

The following hierarchy of eligibility criteria was adopted for two-stage screening: title/abstract and full-text:

1. Articles in English;

2. Original research articles reporting empirical data from experimental, quasi-experimental, or observational studies. Reviews, meta-analyses, systematic reviews, editorials, commentaries, conceptual articles, case studies, and short papers were excluded;

3. Articles that defined their intervention/assessment method as a CMDT (with methods either thoroughly described or properly referenced);

4. Articles with use of VR during cognitive-motor dual task;

5. Articles with average age of the population 65 years or older.

Studies were excluded if they: (i) described exclusively technical aspects or VR prototypes without reporting any data from human participants performing CMDT; (ii) employed VR in a purely descriptive or proof-of-concept manner without any quantitative or qualitative measurements of CMDT performance, cognitive outcomes, motor outcomes, or user experience; (iii) focused solely on system development or technical validation without assessing behavioral or functional measures during DT conditions.

The selected articles were assigned to four independent researchers (JP, FB, VM, and SC), who worked in pairs to select the data using a standardized form. We included an additional researcher (FBr) who collected the data for every study and then verified its accuracy and completeness. Conflicts were resolved by consensus within each pair or by involvement of the fifth author when necessary. Each pair screened all records by title and abstract. Articles meeting initial criteria advanced to full-text assessment, where the same paired-reviewer process was applied. Reasons for exclusion at both stages of screening were documented according to the five selection criteria outlined above.

Consistent with scoping review methodology (Munn et al., 2018), we did not conduct a formal quality assessment or risk of bias appraisal, as we aimed to map the breadth and characteristics of VR-based CMDT approaches rather than to synthesize intervention effectiveness.

Seventeen studies examined VR-based CMDT assessment, predominantly targeting community-dwelling older adults with subclinical conditions rather than clinical conditions. Specifically, the majority focused on older adults with hearing impairment (Downey et al., 2023; Mohanathas et al., 2024; Tung Lau et al., 2016; Wu et al., 2014), subjective cognitive decline (Carr et al., 2019), elevated risk of falls (Gaspar et al., 2013; Zhou et al., 2019). Only two studies included clinical populations with memory impairment (Cook et al., 2013) and Parkinson's disease (Rosenfeldt et al., 2024). Three studies included community-dwelling older adults with health conditions Carr et al., (2020); Francis et al., (2015); Lewis et al., (2023). This distribution suggests VR-based CMDT assessment is currently positioned as a preventive screening tool rather than a diagnostic instrument. Indeed, the emphasis on sensory deficits and fall risk may indicate that CMDT performance is considered a sensitive early marker of functional decline in otherwise healthy aging.

Authors demonstrated a strong preference for immersive VR configurations (12/17 studies). Eight studies employed fully immersive systems such as the CAVE Automatic Virtual Environment system (CAVE) was utilized by Ehlers et al. (2017) and Nagamatsu et al. (2011), Street LAB system (Carr et al., 2019, 2020; Downey et al., 2023; Mohanathas et al., 2024; Tung Lau et al., 2016), and Head-Mounted Display (HMD) (Lewis et al., 2023; Rosenfeldt et al., 2024; Souza Silva et al., 2020). Semi-immersive systems occurred in six studies utilizing either simulator driving platforms (Cook et al., 2013; Gaspar et al., 2013; Wu et al., 2014) or instrumented treadmills with projection screens (Francis et al., 2015; Mack et al., 2023; Stojan et al., 2023). Moreover, assessment protocols overwhelmingly favored realistic environmental situations. The most common paradigm simulating urban navigation with traffic (Carr et al., 2019, 2020; Downey et al., 2023; Ehlers et al., 2017; Mack et al., 2023; Mohanathas et al., 2024; Nagamatsu et al., 2011; Stojan et al., 2023; Tung Lau et al., 2016). These environments provided virtual street-crossing scenarios where participants had to safely cross streets while conversing with hands-free phones or a 6-lane street intersection in Toronto, where participants engaged in street crossing tasks while performing concurrent cognitive activities such as word recognition or dichotic listening tasks. Five studies simulates daily activities such as grocery shopping Lewis et al., (2023); Rosenfeldt et al., (2024), subway navigation Souza Silva et al., (2020), and driving scenarios Cook et al., (2013); Gaspar et al., (2013); Wu et al., (2014). One study employed stylized, game-like stimuli using a forest with wasps, prioritizing experimental control over realism. These data suggest that the field has converged on functional simulation as the goal standard for assessment, with the street crossing as a canonical task.

Task configurations showed various motor-cognitive combinations. Tasks used in nine of the 17 studies could be classified as “walking while” tasks Ehlers et al., (2017); Francis et al., (2015); Lewis et al., (2023); Mack et al., (2023); Nagamatsu et al., (2011); Rosenfeldt et al., (2024); Souza Silva et al., (2020); Stojan et al., (2023); Tung Lau et al., (2016). These paradigms combined a simple walking task as the motor component, usually performed on an instrumented treadmill, with a concurrent cognitive task, including conversing with a hands-free phone Ehlers et al., (2017); Nagamatsu et al., (2011), words recognition Tung Lau et al., (2016), counting backwards Francis et al., (2015), serial threes and color word Stroop task Stojan et al., (2023), reading and memorizing text messages Souza Silva et al., (2020), and virtual shopping tasks Lewis et al., (2023); Mack et al., (2023); Rosenfeldt et al., (2024). The remaining eight studies used alternative motor-cognitive combinations. Five studies adopted standing, postural, or balance tasks as the motor component paired with various cognitive challenges, such as listening tasks Carr et al., (2019), (2020); Mohanathas et al., (2024), serial subtraction tasks de Rond et al., (2021), auditory tasks Downey et al., (2023). Three studies employed driving simulation as the motor component: Cook et al. (2013) associated it with story listening, rehearsing, and recalling; Wu et al. (2014) with speech recognition, while Gaspar et al. (2013) with a 1-back task. “Walking while” paradigm appears as a standardized framework that facilitates comparison across studies and aligns with traditional assessments like the TUG and WWTT.

The most significant methodological weakness emerged in DT cost calculations. The statistical comparison approach was the most used (10/17 studies). It was used inferential statistics (ANOVA, linear mixed models) to compare single-task vs. DT conditions, without computing DT cost ratio Carr et al., (2019), (2020); Cook et al., (2013); Francis et al., (2015); Gaspar et al., (2013); Mack et al., (2023); Mohanathas et al., (2024); Nagamatsu et al., (2011); Tung Lau et al., (2016); Wu et al., (2014). On the other hand, DT cost ratio was calculated in 7 of the 14 studies. They computed proportional or absolute change metrics; however, they used non-standardized formulas to quantify the magnitude of cognitive-motor interference Carr et al., (2020); Downey et al., (2023); Ehlers et al., (2017); Lewis et al., (2023); Rosenfeldt et al., (2024); Souza Silva et al., (2020); Tung Lau et al., (2016). Computational formulas varied, with some using percentage change calculations and others using absolute difference scores, though several studies did not explicitly specify their method. These approaches are mathematically and conceptually non-interchangeable Leone et al., (2015), limiting cross-study comparison and consensus of standardized metrics.

Four studies examined VR-based CMDT training. This limited body of research contrasts with the 17 assessment studies revealing a substantial imbalance in the field.

The research spanned populations, including healthy older adults Nayak et al., (2021); Zukowski et al., (2022), and adults with and without hearing loss Bruce et al., (2019). Unlike assessment studies that focused predominantly on subclinical populations, training research included both preventive applications and therapeutic interventions. This might reflect the field's exploratory phase.

Authors demonstrated a technological preference for non-immersive VR systems (2/4 studies). Particularly, they employed a treadmill with a frontal flat screen displaying computer games (Nayak et al., 2021). Activities were controlled with an inertial mouse tied to the participant's head during the interaction with stylized stimuli (rectangular objects, circular targets from commercial Big Fish Games). Bruce et al. (2019) used an iPad to display interactive cognitive exercises (i.e., discrimination task) while patients performed various physical exercises (i.e., aerobic training) (Bruce et al., 2019). Semi-immersive systems occurred in one study using a treadmill with a large projection screen displaying the virtual environment directly in front of the participants (Zukowski et al., 2022). Participants gathered pizza ingredients while avoiding obstacles in a virtual Italian village street with a wooden cart avatar controlled by body movements. Hassandra et al. (2021) provided a fully immersive HMD configuration with the VRADA: participants cycled while HMD simulated an outdoor natural virtual environment (with a forest path, trees, grass, and mountains) where participants had to solve math quizzes. Training protocol favored abstract stimuli and simplified environments, with two of the four studies employing stylized stimuli from commercial game objects such as geometric shapes (e.g., rectangles and circles), fruits, and vehicles on plain backgrounds (Bruce et al., 2019; Nayak et al., 2021). On the other hand, two studies replicate an Italian village street with simplified graphics (Zukowski et al., 2022) and a forest path with mathematical overlays (Hassandra et al., 2021). No training study replicated the high fidelity of real environments used in assessment protocols.

Moreover, training protocols presented substantial variability in terms of duration and frequency. Two studies conducted a single-session training of 20 min (Hassandra et al., 2021; Zukowski et al., 2022) as a feasibility test and efficacy evaluation, respectively. In contrast, extended programs lasting 6 (Bruce et al., 2019) or 12 weeks (Nayak et al., 2021) with 2 weekly sessions of 30–60 min, respectively. The field lacks consensus on the minimal effective dosage. Single-session studies demonstrated acute effects on gait and cognition (Zukowski et al., 2022), while extended programs showed improvements in motor functions and working memory (Bruce et al., 2019; Nayak et al., 2021). However, without standardized protocols or dose-response analyses, optimal training parameters remain undefined. No studies included follow-up evaluations.

Task configurations showed various CMDT. Tasks used in two of the four studies could be classified as “walking-based” paradigms. These paradigms combined a walking task as the motor component, performed on a treadmill, with a concurrent cognitive task, including visual search (Nayak et al., 2021; Zukowski et al., 2022). One study employed a “cycling-based” paradigm with a stationary bike and mathematical calculation tasks. Bruce et al. (2019) combined aerobic training with an iPad-based discrimination task.

We noted that studies evaluated two primary domains: intervention efficacy (measuring motor functions, cognitive abilities, and DT performance), and user experience, including feasibility, acceptability, and tolerability (Hassandra et al., 2021), and adherence rates (Nayak et al., 2021). Results revealed positive findings across interventions that improved walking speed, dual task accuracy (Zukowski et al., 2022), motor functions, and working memory (Bruce et al., 2019) with high acceptability ratings and high compliance among older adults (Hassandra et al., 2021; Nayak et al., 2021).