Associations of Mobility and Frailty Risk with Wayfinding Ability in Aging

Abstract

Older adults frequently report difficulties with community wayfinding, which may be exacerbated by age-related mobility decline; however, the intersection of these processes to support navigation as a fundamental behavior for independent functioning is poorly understood. Beyond impaired locomotion, pre-clinical mobility decline among middle-aged and older adults is associated with cognitive decline that may affect wayfinding. The present study examined associations of mobility, grip strength as an indicator of frailty risk, multi-domain cognition, and wayfinding performance on a seated virtual navigation paradigm, along with self-reported strategy preference, in a sample of 74 healthy, cognitively normal (MMSE; M = 28.76, SD = 0.95) middle-aged and older adults (aged 53 – 89 years; M = 70.82, SD = 8; 70.3% female). Worse mobility was associated with poorer memory for environmental details. This association was partially mediated by domain-specific cognition, and processing speed had the largest unique contribution. Weaker grip strength—reflecting increased frailty risk—was associated with worse navigation efficiency in locating the hidden platform during the task, and this effect was largely independent of other cognitive correlates. Mobility and frailty risk were associated with self-reported navigation strategy preference in the real world. Individual differences in visuospatial immediate recall and delayed associative recall contributed to this association, although mediation was not supported. Taken together, mobility and frailty risk were differentially associated with wayfinding deficits in aging, which was partially shared with domain-specific cognition. Continued study can explore mobility and frailty as modifiable risk factors to maintain wayfinding ability and promote independence into late life.

Introduction

The global older adult population is growing, with the number of adults aged 65 and older projected to exceed 1.5 billion by 2050 (Padeiro et al., 2023). This demographic shift has generated extensive concern about age-related changes in cognitive and mobility functions, which are vulnerable to decline beginning in middle age (Demnitz et al., 2017; Handing et al., 2023). Although the relation between cognitive and mobility decline is somewhat nuanced, evidence from studies of older adults suggests that mobility challenges often precede cognitive decline (Oveisgharan et al., 2024; Tolea & Galvin, 2016). Earlier and more subtle mobility-related changes often emerge around age 50 and can progress to more overt decline as individuals get older (Cai et al., 2022; Hayek et al., 2025). Common deficits include an age-related slowing of gait speed, impaired gait adaptation, reduced balance and postural control, and diminished motor planning abilities (Bytyçi & Henein, 2021; Freiberger et al., 2020; Osoba et al., 2019). When these changes are combined with age-related declines in processing speed, declarative memory, and executive functions (Demnitz et al., 2016, 2017; Hooghiemstra et al., 2017), these deficits can significantly impact community participation and the ability to maintain a sense of independence in late life. Spatial navigation is a fundamental everyday behavior that relies on the outlined mobility and cognitive correlates; as such, it is susceptible to decline in advanced age, and severe impairments have been consistently associated with diminished independence and increased dementia risk (Coughlan et al., 2018; Laczó et al., 2022; Verghese et al., 2017). Yet, the role of mobility in age-related differences in navigation ability is understudied and has been examined almost exclusively in the context of enacted movement, which has left a critical gap in understanding how mobility interacts with underlying cognitive processes necessary for successful navigation.

Spatial navigation requires the integration of complex cognitive wayfinding processes and intentional movement towards a goal location (Chargo et al., 2026; Montello & Sas, 2006). Wayfinding processes often reference a cognitive representation of the environment (a “cognitive map”) that is developed through experience, enables spatial orientation, and supports the learning, planning, and adaptation of travel routes (Dalton et al., 2019; Epstein et al., 2017; Jamshidi et al., 2020). While few studies have examined earlier changes in wayfinding, the available literature suggests that middle-aged adults demonstrate challenges in acquiring new spatial information (Muffato et al., 2022; Yu et al., 2020). Similarly, older adults demonstrate difficulties in the ability to encode environmental details (i.e., landmark cues, or places) during initial exploration and require greater time to learn travel routes (Head & Isom, 2010; Hilton et al., 2021; Zhong & Moffat, 2016). A failure to encode environmental details may contribute to the creation of an incomplete or inaccurate cognitive representation of the environment, and this can contribute to difficulties when planning travel routes.

With advancing age, individuals demonstrate inefficiencies in route travel, often requiring greater time to reach a goal destination (Daugherty et al., 2016). Navigation efficiency doesn’t appear to always be a priority, as both middle-aged and older adults will favor habitual or familiar paths of longer distance over choosing a more direct, but novel, route (Adamo et al., 2012; Daugherty & Raz, 2017; Driscoll et al., 2003; Harris & Wolbers, 2012). The transition from mid- to late life has also been associated with differences in navigation strategies, where there tends to be a shift from flexible, allocentric strategies that prioritize spatial relationships between landmarks and locations, to more rigid, body-centered egocentric and procedural strategies that generally rely on fixed sequences of turns to reach a destination (Gazova et al., 2013; Harris et al., 2012; Lithfous et al., 2014; Rodgers et al., 2012; Zhong & Moffat, 2018). Current accounts of navigation attribute these age-related deficits to declines in related cognitive processes, largely overlooking the role of mobility, including the awareness of mobility constraints and its potential to prospectively shape strategy choice and execution (Chargo et al., 2026).

Decreased wayfinding ability may be closely related to functional mobility decline (Chargo et al., 2026). Common age-related mobility deficits such as slowed gait speed and poor balance can hinder safe and efficient navigation towards a goal destination and may limit an individual’s willingness or ability to explore their environment (Osoba et al., 2019; Xie et al., 2016). Moreover, mobility decline has been known to contribute to heightened concern or fear of falling that can limit exploration and subsequent learning of the surrounding environment (Ramanoël et al., 2022; Rogge et al., 2021). Age-related reductions in gait speed are associated with decline in efficiency during route travel while ambulating, as well as impaired spatial learning and encoding of environmental details, which was hypothesized to be due to individuals allocating greater cognitive resources for mobility management (Agathos et al., 2020). These findings do align with a cognitive-motor interference and dual-task framework for understanding navigation behaviors (Hill & Ekstrom, 2025).

However, mobility extends beyond locomotion alone and there is a gap in understanding its cognitive correlates that may further account for age-related deficits in wayfinding ability. While our group has shown that clinical mobility impairments in persons with Multiple Sclerosis correlate with subsequent recall of environmental details on a map reproduction task following a seated virtual navigation task (Chargo et al., 2024), these ideas have yet to be examined in healthy aging where pre-clinical mobility decline is expected to emerge.

Beyond limitations in enacted locomotion, pre-clinical decline in mobility in mid- to late life is associated with decline in cognitive processes that support wayfinding ability. Visuospatial working memory is often associated with measures of route efficiency in wayfinding, and it is particularly important for the adaptation of travel routes in response to a changing environment (Blacker et al., 2017; Escamilla et al., 2020; Weisberg & Newcombe, 2018). Declarative and associative memory functions support the ability to encode and recall relevant details that are needed for the creation and use of a cognitive map of the environment (Bestgen et al., 2017; Daugherty & Raz, 2017; Ekstrom & Ranganath, 2018; Manns & Eichenbaum, 2009; Schiller et al., 2015). Furthermore, common age-related decline in processing speed may contribute to impaired integration of visuo-spatial information and decision making for navigation in novel or dynamic environments (Bryden et al., 2023; Salthouse, 1996; Wei et al., 2020). Importantly, these same cognitive abilities are correlated with age-related mobility decline. Age-related slowing of gait speed and the emergence of balance-related challenges have been consistently reported to correlate with a decline in global cognition, as well as in domains of processing speed, declarative memory, working memory, and executive functioning (Demnitz et al., 2016; Hooghiemstra et al., 2017; Kueper et al., 2017; Tabbarah et al., 2002; Tolea & Galvin, 2016). Given that these cognitive processes are strong correlates of both wayfinding and mobility – and are vulnerable to decline with age – this shared cognitive architecture offers a plausible mechanism that may help explain how mobility decline is associated with wayfinding impairments beyond enacted movement alone.

In addition to typical aging, frailty is among the most common risk factors for cognitive and functional decline in community-dwelling older adults and can double the risk for dementia (Borges et al., 2019; Grande et al., 2019). Frailty is a clinical phenotype characterized by having at least three of the following criteria: slowness, low level physical activity, weakness, self-reported exhaustion, and unintentional weight loss (Chen et al., 2014; Fried et al., 2001). In initial stages of the phenotype, individuals with pre-clinical frailty may exhibit fewer than three indicators, typically without clinical impairment that would interfere with daily life (Fried et al., 2001; Veld et al., 2015). Grip strength, which provides an objective measure of muscle weakness, is one of the most reliable indicators of frailty that has been consistently associated with mobility decline, including a reduction in gait speed and poor postural control (Alves et al., 2020; Jiang et al., 2024; Shinohara et al., 2021; Wang et al., 2025). Moreover, individuals with indicators of frailty have an increased prevalence of cognitive impairment as compared to those without (Buchman et al., 2007; Furtado et al., 2016; Halil et al., 2015; Karanth et al., 2024). Studies report not only a global cognitive decline among pre-clinically frail and frail adults, but also domain specific declines in declarative and working memory functions that are relevant to wayfinding (Furtado et al., 2018; Karanth et al., 2024; Li et al., 2024). Older adults with pre-clinical frailty risk are likely to exhibit a unique profile of cognitive and mobility decline that may have an impact on their wayfinding ability and subsequent memory of environmental landmark details.

Taken together, the present study seeks to address the outlined gaps by examining associations of wayfinding in a seated virtual navigation paradigm with independent assessments of mobility function and cognition in a sample of community-dwelling middle-aged and older adults. A seated paradigm was employed to allow study of cognitive correlates to mobility without the confound of enacted movement, thereby isolating potential shared cognitive mechanisms linking mobility and wayfinding. Based upon the existing literature, the following hypotheses were tested:

(1) Decreased mobility and higher frailty risk will be associated with worse navigation efficiency as evidenced by greater time traveled to reach the goal platform in the virtual environment, and worse recall of environmental details following the task.

(2) The association between mobility decline and wayfinding ability will be partially shared with a set of cognitive processes: processing speed, visuospatial immediate recall, and associative memory function.

Materials and Methods

Transparency and Openness

We report how we determined our sample, any data exclusion, all manipulations, and all measures in this study. All data, analysis code and research materials are available on written request to the corresponding authors. Data were analyzed using IBM SPSS Statistics, version 30.0.0.0 and MPlus (v 8.1; Muthén and Muthén). This study’s design and hypotheses, and its analysis were not pre-registered.

Participants

A sample of 74 healthy middle-aged and older adults from the larger longitudinal Detroit Aging Brain Study (DABS) were recruited for participation. DABS includes a community-based cohort of healthy adults, ages 18–90 years at enrollment, with repeated longitudinal assessments spanning 20 years and counting. On previous longitudinal occasions, the sample had self-reported health history and daily lifestyle activities, and blood-based biomarkers of vascular and metabolic health risk collected, which were used to identify persons with potential risk for age-related mobility decline or pre-clinical frailty for recruitment to the current study. Individuals aged 50 years and older were selected for participation to increase the likelihood of observing pre-clinical differences in both cognition and mobility. Of the enrolled 74 participants, 49 had completed a similar virtual navigation paradigm during prior study years, and 22 were task-naïve. Those with prior experience were not significantly different from task-naïve individuals in outcome measures of interest (environmental recall: t(69) = 1.39, p = 0.17; navigation efficiency: t(71) = - 1.89, p = 0.06); therefore, hypotheses were tested with all participants as a single group. Participants were aged 53–89 years (M = 70.82, SD = 8.0) and, on average, had attained a college-level education (M = 16.09 years education, SD = 2.18). The age distribution of the sample is presented in Figure 1. All participants were community-dwelling and ambulated without the use of an assistive device at the time of the study. See Table 1 for a detailed demographic profile of the sample.

Figure 1.

Age distribution of the study sample. Age is displayed in 5-year intervals. Bars represent the number of participants within each age group.

Table 1.

Description of Sample Demographics

Variable	Descriptive Statistic
Sample Size	74
Female, n (%)	52 (70.3%)
Caucasian, n (%)	60 (81.1%)
Age (M ± SD, years)	70.82 ± 8.0
MMSE (M ± SD)	28.76 ± 0.95
CES-D (M ± SD)	4.65 ± 3.94
Grip Strength, pounds (M ± SD)	60.73 ± 21.29
T25FW-Forward, seconds (M ± SD)	6.58 ± 1.21
MiniBEST, total (M ± SD)	21.85 ± 2.66
SDMT, # correct (M ± SD)	52.05 ± 8.89
BVMT-R, total immediate recall (M ± SD)	51.87 ± 9.60
VAL, delayed recall (M ± SD)	35.39 ± 9.62

Note. Sample means and standard deviations are reported. MMSE – Mini-Mental State Examination; CES-D – Center for Epidemiologic Study-Depression Scale; Grip strength (as an indicator for frailty risk) is reported in pounds of grip force; T25FW - Timed 25-Foot Walk, completion time in seconds; MiniBEST – Mini Balance Evaluation Systems Test, total score; SDMT – Symbol Digit Modalities Test, number of correct responses; BVMT-R – Brief Visuospatial Memory Test-Revised, total immediate recall T-scores; VAL – Visual Auditory Learning, delayed recall (number of correctly identified symbol-word pairs).

Additional eligibility criteria required that participants be native English speakers or have acquired English by age 7 with at least a high school education. Participants had no history of cardiovascular or endocrine disease, cancer, neurological or psychiatric conditions. Medicated, controlled hypertension was allowed and infrequent in the reported sample (n = 22, 29.7%). Participants were excluded if they endorsed the use of centrally acting prescription medication, had a history of learning disability, mobility impairment that interfered with the ability to perform activities of daily living, or reported having a head injury with loss of consciousness in excess of 5 minutes. Additionally, all participants were screened for probable dementia with the Mini-Mental State Examination (MMSE) (Folstein et al., 1975); a score of < 26 excluded individuals from participation. Depressive symptoms were assessed with the Center for Epidemiologic Study-Depression Scale (CES-D) (Radloff, 1977); a score ≥ 16 resulted in exclusion from participation. All participants reported right-hand dominance (Edinburgh Handedness Questionnaire; (Oldfield, 1971) and were screened to have normal or corrected-to-normal hearing, and vision of at least 20/50.

Study procedures were approved by Wayne State University’s Institutional Review Board (IRB-23-05-5799). All participants provided written and informed consent prior to the start of data collection.

Testing Procedures

All testing procedures were completed in a single measurement occasion. Participants completed a comprehensive battery of cognitive tests, including the assessment of wayfinding ability, followed by a series of clinical mobility tests that were administered to evaluate walking and balance performance. Various self-report questionnaires were also administered to capture data on participants’ real-world wayfinding strategy preference. The cognitive assessments included as a part of this study were not previously administered to participants in DABS, which had an independent cognitive battery.

Assessments of Wayfinding

Virtual Morris Water Maze.

Participants completed a virtual adaptation of the Morris Water Maze (vMWM) task, based on one described by Moffat and Resnick (Moffat & Resnick, 2002). The task has been reported before (Chargo et al., 2024) and summarized here briefly. The virtual environment was viewed on a 24-inch desktop computer screen from a first-person perspective, where participants remained seated throughout the duration of the task. Participants were instructed to use their dominant right hand to control a joystick in order to navigate through the virtual environment at a fixed speed. Movement could be ceased at any time and movement in the backward direction was not permitted.

Practice.

Participants were first introduced to a practice pool environment to familiarize themselves with movement through the virtual environment. Five visible platforms identified by cubes labeled A – E were presented in the practice pool. Participants were instructed to cross the visible platforms in alphabetical order. Successful completion of the practice trial confirmed their ability to manipulate the joystick effectively.

Learning Trials.

The on-screen virtual environment consisted of a circular, water-filled arena that was situated within a larger room that contained several distinct objects (a grove of trees, a fountain, a stack of tires, a group of three pillars, and a set of vases) located around the pool’s perimeter, in addition to two unique wall features (indented corner and diagonal ivy covered) that remained unchanged throughout the duration of the task. See Figure 2A for a depiction of the vMWM environment. A goal platform was located in the center of one of the pool quadrants and hidden beneath the water’s surface. Participants were instructed to locate the platform as quickly and accurately as possible. The hidden platform remained in the same location across trials and could be found by referencing its position relative to the surrounding cues. Participants began each trial in one of five starting locations that were placed in the three pool quadrants that did not contain the platform and were equidistant to the platform. Starting positions were counterbalanced to present all five starting locations each three times, for a total of 15 learning trials. Facing direction at the start of each trial was randomized. Once participants crossed over the hidden platform, it became visible, movement was stopped and a sound played to indicate successful completion of the trial. If the platform was not located within 2 minutes, the trial was automatically terminated. Efficiency in locating the platform was measured as the total time traveled (in seconds) from the starting location to the first intersection of the hidden platform on successful trials. For each participant, the time required to locate the hidden platform across all successful learning trials was averaged and used in the reported analyses.

Figure 2.

(a) First-person perspective showing two of the landmark cues that surround the pool. (b) Sample map replication with freely recalled landmark cues and the marked platform location.

Visible Platform Trial.

To conclude administration of the vMWM, participants completed a joystick control trial in which the platform was visible and marked by four red flags around its perimeter. Participants were instructed to navigate to the center of the platform as quickly as possible. Time to reach the visible platform (in seconds) was recorded as an index of joystick control.

Recall of Environmental Details.

Following completion of the vMWM, participants were provided a blank piece of paper and instructed to draw an overhead view of the virtual environment, incorporating as much detail as possible and notating the location of the hidden platform with an “X.” See Figure 2B for a sample map reproduction. Memory of the environment was evaluated based on the number of cues included (i.e., objects located around pool perimeter, presence of outer wall) and their correct placement relative to the marked platform. A higher score indicated better free recall of environmental details.

Real-World Navigation Strategy Preference

Navigation Strategy Questionnaire (NSQ).

Participants completed the NSQ to determine individual preference in wayfinding strategies used during real-world navigation (Zhong & Kozhevnikov, 2016). The NSQ is a 44-item self-report questionnaire in which respondents rate their agreement with statements that align with different real-world strategy use: procedural (15 items; e.g., “If I need to return to my origin, it is easier for me to retrace my route than to find a new shortcut”), egocentric spatial updating (17 items; e.g., “It is easy for me to estimate the distance and direction between my moving body and the landmarks I have passed by on the route”), and survey-based or allocentric (12 items; e.g., “My mental representation of the route I traversed is analogous to a schematic map [e.g., floor plan, blue-print, metro map] rather than a first-person perspective of routes and landmarks”). All items were rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree) and sub-scale scores for each strategy were computed by averaging individual item responses, with higher values reflecting a greater preference for that strategy type.

Multi-Domain Cognitive Assessment

Symbol Digit Modalities Test (SDMT).

The SDMT is a timed assessment of cognitive processing speed (Strober et al., 2022). Participants completed the oral version of the assessment to minimize motor demands associated with standard written administration. At the top of the assessment form, a key is presented with digits 1–9 which are each paired with a corresponding geometric symbol. The remainder of the form consists of a pseudorandomized sequence of symbols, and participants are given 90 seconds to verbally identify the digit that corresponds with the presented symbol as quickly and accurately as possible while the administrator records their responses. Performance was scored as the total number of correct responses completed within the allotted time – a higher value was reflective of better processing speed.

Brief Visuospatial Memory Test-Revised (BMVT-R).

Visuospatial learning and memory for immediate recall was assessed with the BVMT-R (Benedict et al., 1996). Participants were first presented with a set of 6 shapes arranged in a 2 × 3 grid layout for 10 seconds and were then asked to draw the shapes in their correct location, and as accurately as possible. These procedures were repeated across 3 learning trials, and scores from each trial were summed to create a total raw score out of 36 (maximum of 12 points per trial). Standardized administration of the task also includes a delayed recall component that is administered after 25-minutes; however, this study only used immediate recall scores. Raw immediate recall scores were converted to T-scores following standard administration procedures, and these T-scores were used in all subsequent analyses.

Visual Auditory Learning (VAL).

VAL, a subtest of the Woodcock-Johnson Test of Cognitive Abilities, was administered to assess associative memory function (Dombrowski et al., 2025). Participants were presented with a series of unfamiliar visual symbols, each paired with an orally presented word (i.e., a unique symbol paired with the word “dog”). Following the initial learning phase, participants were instructed to translate sequences of visual symbols into their corresponding words based on the arbitrary symbol-word associative pairings. Corrective feedback was provided during the initial learning phase. Immediate recall performance was scored based on the number of errors made; however, these values were not analyzed in the present study. Following a 30 – 40 minute delay period, participants were presented the symbols again and asked to recall the corresponding words. Delayed recall performance was scored based on the number of correct symbol-word associations, where higher scores were reflective of better delayed associative memory.

Mobility Assessment

Timed 25-Foot Walk (T25FW).

The T25FW was used to assess mobility function broadly (Kieseier & Pozzilli, 2012). Participants completed the T25FW at a self-selected comfortable pace in the forward (FW) and backward (BW) direction along a 25-foot walkway while maintaining safety. For all BW trials, participants were instructed to direct their gaze ahead instead of looking down at their feet. Participants completed two trials in both directions, and completion times were recorded in seconds. An average was then computed separately for FW and BW. Greater completion time was reflective of worse general mobility (i.e., slowed gait). To ensure safety, all participants wore a gait belt and were closely monitored by trained research personnel throughout the assessment.

Mini Balance Evaluation Systems Test (MiniBEST).

The MiniBEST is a standardized clinical assessment used to evaluate balance performance across four sub-domains: anticipatory balance, reactive postural control, sensory orientation, and dynamic gait (Yingyongyudha et al., 2016). The assessment includes 14 performance-based tasks, each rated on a 3-point ordinal scale (0–2), with a total possible score of 28. For the purpose of this study, only MiniBEST total scores were used as a comprehensive measure of dynamic balance performance—a higher score indicated better balance.

Grip Strength

Manual grip strength, an indicator of frailty (i.e., weakness), was measured using a Jamar hydraulic hand dynamometer (Huang et al., 2022). Participants were instructed to remain seated with their feet kept flat on the ground. The elbow was flexed at approximately 90° and held close to the body during testing. Three trials were completed for each hand (administration alternated between right and left hands), and an average grip force in pounds was then computed. Higher values are suggestive of greater grip strength, and therefore lower frailty risk.

Statistical Analyses

All preliminary data screening procedures and correlation analyses were conducted using SPSS (V 30.0.0.0). Univariate distributions were found to be approximately normal with the exception of navigation efficiency, for which values were log-transformed to alleviate skew. Five univariate outliers were detected (z > |3.29|), however, no individual case was identified as a multivariate outlier (Mahalanobis distance; critical χ² (18) = 42.31, α = 0.001). Statistical models were evaluated with and without univariate outlier cases, and comparable results were observed; thus, findings are interpreted and reported for the complete sample to prioritize external validity and statistical power.

Descriptions of study variables were made by correlation and regression models. Bivariate Pearson correlations were first conducted to examine associations among participant’s age, mobility and balance performance, frailty risk, cognition, wayfinding ability (recall of environmental details, navigation efficiency: average time), and real-world navigation strategy preference. Two separate models of multiple regression were then tested to identify unique predictors of navigation efficiency and environmental recall. Age was entered first as a covariate, followed by candidate predictors that included FW completion time on the T25FW, balance performance (MiniBEST), and grip strength as an indicator of frailty risk. Because FW and BW completion times were highly correlated, BW was not included in the model so to reduce multicollinearity. Performance on the joystick control trial (i.e., time to reach the visible platform) was included as a covariate in the model predicting navigation efficiency. Rank ordered comparisons of standardized regression coefficients (β) were considered to determine the unique contribution of each predictor. All significance testing was evaluated at α = 0.05.

For the study hypotheses, tests of mediation were specified in structural equation models (SEM) in MPlus (v 8.1; Muthén and Muthén). The latent construct of mobility was reflective of three indicators: FW completion time on the T25FW which was set with a loading factor of −1 for local identification, BW completion time on the T25FW, and balance performance (MiniBEST), which had freely estimated factor loadings. FW and BW completion time residuals were allowed to correlate provided there is a degree of shared measurement error due to similarity in assessment protocols, which if not estimated independently, could affect hypothesis test estimates in the model. First, a model was constructed to test the effect of grip strength and latent mobility predicting navigation efficiency via three intercorrelated cognitive mediators: processing speed (SDMT), visuospatial immediate recall (BVMT-R), and delayed associative memory (VAL). Potential differences in joystick control was included as a covariate of navigation efficiency and was correlated with grip strength and latent mobility to account for effects of general motor control. Second, a similar model tested the effect of grip strength and latent mobility as predictors of environmental recall via the same three cognitive mediators. The cognitive mediators are functional domains that have been established correlates of navigation performance and subsequent memory of the environment. As such, the purpose of this model was not to examine their predictive utility, but rather to determine which of the processes confer the effect of grip strength and mobility under conditions where enacted movement is not directly involved in task performance. Any evidence of indirect effects would suggest for shared cognitive processes with mobility.

As an alternative account, we tested models of general cognition that may mediate mobility effects on wayfinding outcomes. In the alternative models, cognition was modeled as a latent construct reflective of SDMT, BVMT-R, and VAL as three indicators. The purpose of these models was to determine whether the association between mobility, frailty risk and wayfinding reflects variance shared across the relevant cognitive domains, as opposed to domain-specific effects that are tested in the primary models. SDMT was set with a factor loading of 1 for local identification, and all other factor loadings were freely estimated. In addition to objective model fit, comparative model fit was evaluated by Akaike Information Criterion (AIC) and sample-size adjusted Bayesian Information Criterion (BIC) relative to the hypothesized models. As a secondary assessment, model R² and indirect effect significance was compared between the nested models.

For the second hypothesis, grip strength and mobility limitations were tested as potential correlates of real-world navigation strategy preference through a shared cognitive mechanism. A separate model was specified that was similar to the first hypothesis tests. Preference for procedural, egocentric and survey navigation strategies (continuous scores) were outcome measures, predicted by grip strength and latent mobility via correlated cognitive mediators. Men in the sample reported a stronger preference for survey navigation strategies (t(72) = −2.980, p < .01); as such, sex was included as a covariate of survey navigation strategy preference.

Across all models, grip strength and latent mobility were correlated. Sex was included as a covariate of grip strength given that men had, on average, stronger grip than women in the sample: (t(72) = −6.24, p < 0.001). Although the sample spanned a relatively broad age range, the distribution was centered around older adulthood, with most participants clustered around the seventh decade of life. Given this distribution and its non-significant contribution in the initial tests with multiple regression, age was excluded as a covariate in all SEM tests.

All loadings are reported as unstandardized, and all path coefficients are reported as unstandardized b-weights, which allowed for interpretation on their original scales. Model fit was determined using standard indices (Raykov & Marcoulides, 2012): chi-square (χ²) non-significance (α = 0.05), root mean square error of approximation (RMSEA) below 0.08, comparative fit index (CFI) exceeding 0.90, and a standardized root mean residual (SRMR) below 0.10. Evidence of mediation was assessed by indirect effect (MacKinnon et al., 2007) with Sobel z test significance (α = 0.05) and bias-corrected bootstrapped (10,000 draws) 95% confidence intervals (BS 95% CI) that did not overlap zero (Hayes & Scharkow, 2013). Effect composition was described as percentage direct and indirect effects to the cumulative effect.

Results

Poor Balance and Frailty Risk were Associated with Worse Navigation Efficiency and Recall of Environmental Details

Correlation analyses revealed that worse mobility, as indicated by slower completion time on the T25FW in the forward and backward direction, and weaker grip strength (reflecting greater frailty risk) were both associated with less efficient navigation in the virtual environment (T25FW – FW: r = 0.27, p = 0.02, T25FW – BW: r = 0.35, p < 0.01. Grip Strength: r = −0.44, p < 0.001), whereas no significant bivariate correlations were observed with recall of environmental details (See Table 2). Individuals with poor balance, as indicated by lower scores on the MiniBEST, remembered fewer details about the virtual environment (r = 0.28, p = 0.02) and were less efficient in navigating to the hidden platform (r = −0.33, p < 0.01). See Table 2 for correlations among all study variables.

Table 2.

Bivariate Correlations Among Study Variables

		Bivariate Correlations
Measure	Mean (SD)	Age	FW	BW	Mini-BEST	Grip	SDMT	BVMT	VAL	Env Recall	Nav Effic	Joy Cntrl	NSQ - Proc	NSQ - Ego	NSQ-Surv
Age	70.82 (8.00)	--
T25FW - FW	6.58 (1.21)	.375 **	--
T25FW – BW	9.18 (2.84)	.621 **	.665 **	--
MiniBEST	21.85 (2.66)	−.495**	−.506**	−.499**	--
Grip Strength	60.73 (21.30)	−.245*	−.225	−.325**	.243 *	--
SDMT	52.05 (8.90)	−.325**	−.346**	−.346**	.426 **	−.039	--
BVMT-R	51.86 (9.60)	−.250*	−.295*	−.237*	.419 **	.125	.425 **	--
VAL	35.39 (9.62)	−.161	−.308**	−.242*	.354 **	−.038	.516 **	.613 **	--
Environment Recall	7.08 (2.40)	−.092	−.046	−.087	.280 *	.123	.345 **	.309 **	.338 **	--
Navigation Efficiency	48.01 (16.88)	.377 **	.272 *	.345 **	−.329**	−.441**	−.301**	−.260*	−.182	−.397**	--
Joystick Control	12.87 (5.44)	.454 **	.468 **	.376 **	−.401**	−.290*	−.340**	−.306*	−.132	−.170	.505 **	--
NSQ - Procedural	3.62 (0.53)	.026	.096	−.086	−.262*	−.007	−.156	−.023	−.128	−.177	.312 **	.194	--
NSQ - Egocentric	2.95 (0.73)	−.006	.096	.110	−.008	.217	.014	.118	−.185	.181	−.196	−.087	−.047	--
NSQ - Survey	2.97 (0.75)	.054	−.029	−.041	.107	.307 **	−.011	.232 *	.049	.096	−.205	−.161	−.118	.435 **	--

Note. Pearson (r) correlations are reported for the bivariate relation between age, mobility as indexed by time to complete the Timed 25 Foot Walk (T25FW) in the forward (FW) and backward (BW) direction, balance performance (MiniBEST – Mini Balance Evaluation Systems Test), grip strength measured in lbs., processing speed (SDMT – Symbol Digit Modalities Test), visuospatial immediate recall (BVMT – Brief Visuospatial Memory Test-Revised), associative memory (VAL – Visual Auditory Learning, delayed recall), free recall of environmental details (Env Recall), navigation efficiency as indexed by average time to locate the hidden platform in seconds (Nav Effic), time to complete the joystick control trial in seconds (Joy Cntrl), and procedural (Proc), egocentric (Ego), and survey (Surv) navigation strategy preference (NSQ – Navigation Strategy Questionnaire). Bolded values emphasize statistically significant correlations.

^*p < .05,

^**p < .001.

Based on the observed correlations, two multiple regression models were constructed and tested time to complete the T25FW (forward), balance performance (MiniBEST), and grip strength as predictors of navigation efficiency and environmental recall. The model predicting navigation efficiency was significant (F_{5, 61} = 9.31, p < 0.001). Controlling for possible differences in joystick control (b = 0.01, p = 0.01), individuals with greater frailty risk took longer to locate the hidden platform within the virtual environment (b = −0.003, sr = 0.35, p < 0.001). Accounting for this effect, age (b = 0.004, sr = 0.13, p = 0.19), general mobility function (b = −0.02, sr = 0.08, p = 0.38) and balance performance (b = −0.01, sr = 0.09, p = 0.33) did not significantly predict efficient navigation. The model predicting environmental recall (F_{4, 66} = 1.71, p = 0.16) identified balance performance as a significant predictor (b = 0.31, sr = 0.27, p = 0.02), and age (b = 0.02, sr = 0.04, p = 0.72), general mobility function (b = 0.22, sr = 0.09, p = 0.41) and frailty risk (b = 0.01, sr = 0.07, p = 0.56) were not. Comprehensive rank-ordered results of the multiple regression analyses are provided in Table 3.

Table 3.

Summary Table of Rank-Ordered Predictors of Navigation Efficiency and Environmental Recall

	Navigation Efficiency: Average Time (s)					Environmental Recall
Predictor	b	β	t	sr 2	p	b	β	t	sr 2	p
Age	0.004	0.161	1.324	0.016	0.190	0.015	0.049	0.354	0.002	0.724
T25FW: Comfortable FW	−0.016	−0.109	−0.884	0.007	0.380	0.224	0.113	0.831	0.009	0.409
MiniBEST	−0.008	−0.124	−0.989	0.009	0.326	0.313	0.338	2.328	0.075	0.023
Grip Strength	−0.003	−0.369	−3.623	0.122	< 0.001	0.008	0.072	0.586	0.005	0.560
Joystick Control Trial	0.012	0.335	2.772	0.071	0.007	-	-	-	-	-

Note. Bolded values were statistically significant unique effects. Regression parameters are reported with squared semi-partial correlations (sr²).

Frailty Risk was Associated with Navigation Efficiency Independent of Cognition

Following initial descriptions of the associations among mobility and grip strength with wayfinding outcomes, hypotheses were tested in SEM. The first hypothesized model tested grip strength and latent mobility predicting navigation efficiency via three intercorrelated cognitive mediators: processing speed (SDMT), visuospatial immediate recall (BVMT-R) and associative memory (VAL, delayed recall) (See Figure 3). The model had excellent fit: χ²(22) = 22.726, p = 0.417; RMSEA = 0.021, CFI = 0.997, and SRMR = 0.048. The model accounted for 37.6% of the variance in navigation efficiency (R² = 0.376, p < 0. 001).

Figure 3.

A diagram of the structural equation model with results reported as unstandardized coefficients. Depicted is the outcome navigation efficiency that is predicted by grip strength and mobility via three cognitive measures: SDMT, BVMT-R, and VAL. All cognitive measures were allowed to intercorrelate in the analysis. Mobility is modeled as a latent construct reflective of three indicators: FW, BW, MiniBEST. FW and BW, as well as grip strength and mobility were allowed to correlate, represented by the curved double-headed arrows. Grip strength and mobility were also allowed to correlate with performance on the joystick control trial but is not illustrated in the model. Sex was included as a covariate. * p < 0.05, ** p < 0.01.

The latent mobility factor was reflective of three indicators with the following standardized loadings: FW (T25FW – FW = −0.683, p < 0.001), BW (T25FW – BW = −0.644, p < 0.001), and balance performance (MiniBEST = 0.726, p < 0.001). The total effect of grip strength on navigation efficiency was significant (-0.026, p = 0.003), whereas the total effect of mobility was not (-0.031, p = 0.376). There was no evidence of mediation for either predictor as indicated by non-significant cumulative indirect effects of grip strength (indirect = 0.004, z = 1.05, p = 0.29; BS 95% CI: −0.006, 0.031) and mobility (indirect = −0.027, z = −1.08, p = 0.28; BS 95% CI: −0.158, 0.059). Interestingly, only 10.54% of the total effect of grip strength on navigation efficiency was attributed to the indirect path via specific cognitive deficits, with the direct effect accounting for 89.46%. By comparison, 84.31% of the total effect of mobility was attributed to the indirect path, while the direct effect only accounted for 15.69% (See Table 4). Notably, when considering the unique contributions of each of the cognitive mediators, processing speed (SDMT) was the main contributor to the overall indirect effect for both grip strength and mobility, with no unique contributions noted for visuospatial immediate recall (BVMT) and delayed associative memory (VAL), which may be attributed to the real-time processing demands of the navigation task. (Table 4).

Table 4.

Effect Composition of Cognitive Mediation Models Predicting Navigation Efficiency and Environmental Recall

Effect Composition	Navigation Efficiency		Environmental Recall
	Grip Strength	Mobility	Grip Strength	Mobility
Total Effect	−0.026, p < 0.01	−0.031, p = 0.38	0.037, p = 0.80	0.660, p = 0.20
Indirect Effect	0.004, p = 0.29	−0.027, p = 0.28	−0.151, p = 0.11	1.064, p = 0.04
% Direct Effect	89.46%	15.69%	55.48%	27.65%
% Cumulative Indirect Effect	10.54%	84.31%	44.52%	72.35%
Processing Speed: SDMT	10.54%	84.31%	25.64%	37.31%
Visuospatial Immediate Recall: BVMT	0.00%	0.00%	2.70%	11.86%
Associative Memory: VAL	0.00%	0.00%	16.18%	23.18%

Note. Calculations of total and indirect effects were made from unstandardized coefficients, with reported z-test significance. Effect decomposition of direct and indirect effects were calculated relative to the cumulative effect. SDMT – Symbol Digit Modalities Test; BVMT – Brief Visuospatial Memory Test-Revised; VAL – Visual Auditory Learning, delayed recall.

Domain-Specific Cognition Mediated the Association Between Mobility and Environmental Recall

The second hypothesized model tested grip strength and mobility as predictors of environmental recall via the same three cognitive mediators: SDMT, BVMT-R, and VAL (See Fig. 4). The model had excellent fit as indicated by a non-significant chi-square (χ²(16) = 19.567, p = 0.241), RMSEA = 0.055, CFI = 0.983, and SRMR = 0.050; and accounted for 17.8% of the variance in recall of environmental details (R² = 0.178, p = 0. 034).

Figure 4.

A diagram of the structural equation model with results reported as unstandardized coefficients. Depicted is the outcome environmental recall that is predicted by grip strength and mobility via three cognitive measures: SDMT, BVMT-R, and VAL. All cognitive measures were allowed to intercorrelate in the analysis. Mobility is modeled as a latent construct reflective of three indicators: FW, BW, MiniBEST. FW and BW, as well as grip strength and mobility were allowed to correlate, represented by the curved double-headed arrows. Sex was included as a covariate. * p < 0.05, ** p < 0.01.

The total effects of grip strength and mobility on recall of environmental details were not significant (Grip total = 0.037, p = 0.80; Mobility total = 0.660, p = 0.20). 44.52% of the total effect of grip strength on recall of environmental details was attributed to the indirect path via specific cognitive deficits, while the direct effect accounted for 55.48%. On the other hand, 72.35% of the total effect of mobility on recall of environmental details was attributed to the indirect path, while the direct effect accounted for 27.65%. While there was no evidence of mediation for the effect of grip strength on environmental recall (indirect = −0.151, z = −1.61, p = 0.11; BS 95% CI: −1.528, 0.033), a significant cumulative indirect effect was observed for mobility (indirect = 1.064, z = 2.06, p = 0.04; BS 95% CI: −0.258, 9.239), indicating some support for domain-specific cognitive processes to partially mediate the relation between mobility function and subsequent memory of environmental details. Effect decomposition indicated that processing speed (SDMT) had the largest unique contribution across both predictors, accounting for 25.64% and 37.31% of the cumulative indirect effect for grip strength and mobility, respectively (See Table 4).

Alternative Hypothesized Model of a General Cognitive Effect

A plausible alternative account of the observed effects could reflect a general cognitive deficit, provided correlations among the tested cognitive mediators. To test this hypothesis, two alternative models were constructed in which a latent general cognition construct was substituted in as the mediator of grip strength and mobility on wayfinding performance measures. The latent general cognition was reflective of three indicators with the following standardized loadings: processing speed (SDMT = 0.679, p < 0.001), visuospatial immediate recall (BVMT-R = 0.725, p < 0.001), and delayed associative memory (VAL = 0.778, p < 0.001). Both general cognition models demonstrated acceptable and comparable fit to the data and explained a similar proportion of variance in navigation efficiency and recall of environmental details.

The model predicting navigation efficiency exhibited good fit (χ²(28) = 32.689 p = 0.247; RMSEA = 0.048; CFI = 0.981; SRMR = 0.058); and accounted for 37.8% of the variance in average time to locate the hidden platform (R² = 0.378, p < 0.001), nearly identical to that of the model with distinct cognitive mediators (R² = 0.376, p < 0.001). Similarly, the model predicting environmental recall had good fit as indicated by a non-significant chi-square (χ²(22) = 27.742, p = 0.184), RMSEA = 0.059, CFI = 0.973, and SRMR = 0.056; and accounted for 24.0% of the variance in the recall of environmental details (R² = 0.240, p = 0.068), as compared to the 17.8% in the original hypothesized model of competing cognitive mediation. Because both general cognition models produced acceptable fit – though a comparison of AIC and BIC comparative fit indices suggest slightly worse fit (Efficiency: AIC = 3088.91, BIC = 3183.36; Environmental recall: AIC = 3067.44, BIC = 3150.38) as compared to the original models (Efficiency: AIC = 3086.87, BIC = 3167.51; Environmental recall: AIC = 3063.61, BIC = 3132.73)—they cannot be rejected on the basis of fit alone. Yet, there were no significant cumulative indirect effects from grip strength and mobility predictors to navigation efficiency and recall of environmental details (all p’s > 0.05), providing little evidence to support a general cognitive effect. Moreover, by collapsing the three cognitive indicators into a single latent construct of general cognition, we lose some specificity for interpretation, creating a challenge to understand the unique cognitive mechanisms that support an association between frailty risk, mobility function and wayfinding performance. Therefore, the alternative models were not interpreted further.

Association of Mobility and Frailty Risk with Real-World Navigation Strategy Preference

Modest correlations were observed between better balance (MiniBEST) and lesser preference for procedural strategy, and stronger grip (lesser frailty risk) with greater preference for survey strategy (Table 2). The SEM model that included mobility, grip strength and cognitive correlates predicting strategy preference scores demonstrated good overall fit (χ²(21) = 27.463, p = 0.156; RMSEA = 0.064; CFI = 0.972; SRMR = 0.049) and explained a modest proportion of variance: 7.0% in procedural (R² = 0.070, p = 0.349), 20.6% in egocentric (R² = 0.206, p = 0.036), and 23.6% (R² = 0.236, p = 0.011) in survey preference. Although there was no evidence of mediation across any path, as indicated by non-significant indirect effects (all p’s > 0.05), the effect decomposition was calculated to better characterize the shared contributions of cognitive correlates (See Table 5).

Table 5.

Effect Composition of Cognitive Mediation Model Predicting Navigation Strategy Preference

Effect Composition	Procedural		Egocentric		Survey
	Grip Strength	Mobility	Grip Strength	Mobility	Grip Strength	Mobility
Total Effect	0.019, p = 0.54	−0.162, p = 0.15	0.095, p = 0.03	−0.161, p = 0.25	0.024, p = 0.67	0.052, p = 0.72
Indirect Effect	0.004, p = 0.78	0.020, p = 0.85	−0.002, p = 0.93	0.155, p = 0.31	−0.007, p = 0.77	0.179, p = 0.22
% Direct Effect	57.4%	60.5%	60.5%	36.6%	72.0%	37.0%
% Cumulative Indirect Effect	42.6%	39.5%	39.5%	63.4%	28.0%	63.0%
Processing Speed: SDMT	8.8%	5.0%	14.3%	17.4%	5.3%	4.4%
Visuospatial Immediate recall : BVMT	12.9%	22.8%	6.1%	23.1%	22.7%	58.6%
Associative Memory: VAL	20.9%	11.7%	19.1%	22.9%	0.0%	0.0%

Note. Calculations of total and indirect effects were made from unstandardized coefficients, with reported z-test significance. Effect decomposition of direct and indirect effects were calculated relative to the cumulative effect. SDMT – Symbol Digit Modalities Test; BVMT – Brief Visuospatial Memory Test-Revised; VAL – Visual Auditory Learning, delayed recall.

There was a significant total effect of grip strength on egocentric navigation strategy preference (Grip total = 0.095, p = 0.03). Only 39.5% of the total effect of grip strength was attributed to the indirect path via specific cognitive differences, with the direct effect accounting for 60.5%. Moreover, for both procedural and egocentric navigation strategy preference, the effect of grip strength and mobility were shared mostly with visuospatial immediate recall (BVMT-R) and delayed associative memory (VAL) processes. The effect of grip strength and mobility on survey navigation strategy preference was predominantly shared with visuospatial immediate recall, accounting for 22.7% and 58.6% of the cumulative indirect effect for grip strength and mobility, respectively (Table 5). Together, these findings suggest that mobility and grip strength as an indicator of frailty risk are associated with differences in wayfinding performance, and to a lesser degree real-world navigation strategy preference, and these effects are partially shared with cognitive processes that are vulnerable in aging.

Discussion

The present study aimed to investigate associations among mobility, frailty, multi-domain cognition and wayfinding ability in a sample of middle-aged and older adults. The use of a seated virtual navigation task and self-reported strategy preference for real world navigation provided a well-rounded investigation of mobility correlates to wayfinding that goes beyond locomotion alone. Our prior research has demonstrated associations between independent assessments of general mobility function (T25FW completion time, balance performance) and wayfinding ability in a clinical sample of Multiple Sclerosis (Chargo et al., 2024; Chargo, Takla, et al., 2025), and the current report in healthy middle-aged and older adults replicates a similar pattern in aging. We further extended the evidence by examining unique contributions of a shared set of cognitive processes that may underlie the association of mobility and wayfinding in a seated, virtual navigation task.

Accounting for potential differences in joystick control, weaker grip strength—reflecting increased frailty risk—was associated with worse navigation efficiency during the seated virtual task, and this effect was largely independent of cognitive correlates. Poorer balance was associated with worse free recall of environmental details, and as part of a general mobility assessment, this effect was largely shared with slower processing speed. Mobility and frailty risk indicators were also modestly associated with real-world preference for procedural, egocentric and survey navigation strategies, which had some overlap with individual differences in visuospatial immediate recall and delayed associative recall. The findings discussed here provide compelling initial evidence that support a role for mobility to modify aspects of wayfinding through cognitive processes vulnerable in aging, independently of and in addition to pre-clinical frailty risk.

Despite observed correlations among grip strength as an indicator of frailty risk, mobility function (T25FW completion time, balance) and navigation efficiency, a multiple regression model found that only grip strength was uniquely significant, above all other predictors including joystick control. Frailty has been broadly recognized as a systemic risk factor for age-related cognitive and mobility decline (Borges et al., 2019; Grande et al., 2019; Jiang et al., 2024; Shinohara et al., 2021). While few empirical studies have examined frailty-mobility associations in the context of everyday navigation, a recent study with institutionalized older adults found that slower completion times on the Floor Maze Test were associated with increased odds of developing a clinical frailty diagnosis, and this effect was independent of gait speed (Souza et al., 2023). The present study examined the reverse relation (i.e., frailty risk as the predictor) and extend the literature to identify a robust link between frailty risk and wayfinding performance among otherwise healthy middle-aged and older adults. The effect of weaker grip strength on navigation efficiency was largely direct, with a smaller shared effect with cognition—an overlap driven exclusively by processing speed.

Taken together, pre-clinical frailty may signal generalized risk to complex cognitive-mobility processes that real-time wayfinding relies upon, making the navigation efficiency measure sensitive to its effect. Frailty risk may reflect early stages of risk for neurodegeneration—such as reduced structural integrity of navigation-relevant brain regions, including the medial temporal lobe (MTL), or diminished white-matter integrity (Cipolli et al., 2024; Lin et al., 2025). Frailty is associated with activity curtailment among older adults (Ellmers et al., 2023), which may be associated with diminished tendency for exploration of novel environments that could lead to decline in wayfinding ability. However, the current study found a prominent effect of frailty on virtual navigation efficiency and not recall of environmental details that is dependent on exploration of the open environment. Additional study is warranted to further understand the nature of this effect. Nonetheless, the association of grip strength with performance on the task, statistically independent of joystick control, is a novel contribution to the literature on age-related differences in wayfinding and highlights a modifiable risk factor to consider.

Independent of frailty risk indicated by grip strength, mobility was associated with poor memory of environmental details. Consistent with our previous reports in a sample of persons with Multiple Sclerosis (Chargo, Takla, et al., 2025), we found that poorer balance was associated with fewer details recalled from the virtual environment. Maintaining balance and postural control, particularly in novel or unpredictable environments, draws on a range of higher-level cognitive abilities: executive functions, sensorimotor integration, attentional control, and even elements of motor planning (Caetano et al., 2017; Small et al., 2021; Woollacott & Shumway-Cook, 2002). Common age-related declines in these abilities can make it extremely difficult to adapt navigation behaviors in response to environmental changes. Individuals who exhibit poorer balance tend to experience greater difficulties when planning travel routes and navigating complex environments (Martín-Pozuelo et al., 2024), while balance-focused training can enhance performance on virtual navigation assessments that require the creation and use of a mental representation of the environment (Merriman et al., 2022). Although dynamic balance was not directly relevant to performance on the seated virtual navigation task, the associations observed in the current report suggest a contribution of complex cognitive-mobility processing to wayfinding.

Of the models testing mediation, there was evidence for an indirect effect of latent mobility on environmental recall via distinct cognitive correlates. The majority of the effect was shared with processing speed and, to a lesser extent, delayed associative memory. The ability to remember details about the environment is supported by declarative memory function – which is important for the encoding, storage, and retrieval of contextual details necessary for constructing a coherent environment representation – or cognitive map (Daugherty & Raz, 2017; Ekstrom & Ranganath, 2018; Schiller et al., 2015). However, the ability to efficiently encode that information in real time while navigating may depend more on processing speed. Because the vMWM assessment has 2-min timed trials, there is a speed component for the participant to encode and integrate details about the platform location and relevant landmark cues. Therefore, an age-related slowing of processing speed may constrain efficient information uptake, ultimately resulting in less detailed representations of the virtual environment at subsequent recall. Processing speed is among the earliest and most consistent cognitive abilities to decline with age and has been associated with age-related mobility declines (Demnitz et al., 2016, 2017; Park & Festini, 2017; Salthouse, 1992; Salthouse, 2010). Although the study sample demonstrated relatively high SDMT performance, consistent with administration of the oral version of the test (Strober et al., 2020), age-related deficits in processing speed were still evident. Slowing of processing speed underlies broader age-related decrements across multiple domains of cognition (Salthouse, 1994, 1996) – as such, mobility effects mediated by processing speed may be relevant to declines in cognitive outcomes, in general, with pronounced effects in early stages of decline for challenging and dynamic processes, such as wayfinding.

Given the real-world significance of mobility deficits and frailty risk, it is plausible that the effects we observed in a virtual navigation task may be related to individual differences in navigation strategy preference. Advanced age is associated with a shift in preference toward proceduralized navigation strategy – which is hypothesized to be a compensatory adaptation due to declines in cognition (Gazova et al., 2013; Rodgers et al., 2012; Zhong & Moffat, 2018). Within the proposed cognitive-mobility-navigation framework, age-related mobility decline may prospectively shape the selection and subsequent use of navigation strategies (Chargo et al., 2026). In the current sample, there were small correlations between worse balance and higher preference for procedural strategy, and between greater grip strength and higher preference for survey strategy. Given the evidence for worse mobility, and worse balance in particular, predicting poor environmental recall, the greater preference for procedural strategy and lesser preference for survey strategy may reflect an impoverished cognitive map. A similar observation can be made of the relatively smaller contribution of BVMT immediate recall to the tests of indirect effects, in which processing speed and associative memory recall were larger. Episodic memory is a canonical correlate of navigation, in particular memory for environmental details that would be represented in the cognitive map (Burgess et al., 2002; Ekstrom & Ranganath, 2018). In study of age-related differences, this effect has been commonly reported in designs that contrast young and older adults (Chargo, Dahle, et al., 2025; Head & Isom, 2010; Moffat et al., 2006), and therefore the age restriction of this sample may have resulted in a smaller effect size. Another plausible explanation is that immediate episodic recall isn’t a strong correlate shared with mobility-related differences in wayfinding among middle-aged and older adults.

This is consistent with the previously reported age-related shift away from allocentric navigation that relies on declarative memory systems towards proceduralized strategies (Iachini et al., 2009; Zhong & Moffat, 2018). Survey and allocentric strategies are considered effortful and more cognitively demanding than procedural strategies (Wiener et al., 2020), and therefore the age-related shift may in part be due to limited cognitive resources. Here we have expanded the evidence to also include a role for mobility, even when locomotion is not directly relevant to the task. One plausible explanation is that an individual is aware of their mobility constraints as a form of navigational affordance, which leads to strategy preference and selection of environmental details when learning new environments (Chargo et al., 2026). The vMWM task employed in the current study included five counter-balanced starting locations with randomly generated facing directions, which decreased the likelihood of learning the platform location with a procedural strategy and favorably weighted the use of survey and allocentric strategies. Therefore, middle-aged and older adults with worse balance and worse grip strength may have performed worse in the vMWM, in part due to these real-world strategy preferences. Further research is necessary to understand the role of mobility and frailty in the real-time learning process and the modification by navigation strategy preference using experimental paradigms (Chargo et al., 2026).

Taken together, the present findings provide compelling initial evidence for a role of mobility in wayfinding that is, in part, shared with differences in cognitive abilities that are vulnerable in advanced age. The reported results highlight that mobility’s effect on wayfinding ability is largely related to impaired environmental recall that is shared with slowed processing speed, whereas frailty risk indicated by grip strength may be associated with a general deficit in navigation efficiency. However, based on our findings, cognition may only represent one piece of a rather complex story. It’s plausible that perceptual or affective factors – such as sustained attention or reduced confidence in mobility (Peeters et al., 2019; Peterson et al., 2007) – fundamentally impact how individuals approach exploring a novel environment, their preferred navigation strategy, and their execution of route travel. Investigation of these ideas may be an immediate and meaningful target for future research, offering a more comprehensive understanding of how mobility-related factors shape real-world navigation in aging, and by extension promote maintenance of functional independence into late life.

Limitations

The study results should be considered for strengths and limitations. First, the cross-sectional design of this study, while informative, does not support inference of causal mechanisms and prevents testing of age-related change (Raz & Lindenberger, 2011). The observed age-related differences reported in this study are assumed to reflect underlying aging trajectories that begin earlier in life, with larger longitudinal studies needed to further clarify the temporal ordering of age-related changes in mobility, cognition and wayfinding ability. Second, the vMWM is a well-validated paradigm for laboratory-based assessment of wayfinding; however, its ecological validity is inherently limited compared to real-world settings (Hamilton et al., 2002). Nevertheless, by using a seated virtual task we were able to minimize the direct influence of walking or balance performance, thereby isolating the contribution of mobility-related factors to the cognitive processing that supports navigation. Although this study contributes to the literature on the cognitive components of mobility that would support wayfinding, it provides a constrained account that does not capture proprioception or other aspects of enacted movement as highlighted in other research (Agathos et al., 2020; Hill et al., 2024; Hill & Ekstrom, 2025). Future studies will extend this work to incorporate both ambulatory virtual and real-world navigation paradigms to further clarify a role for mobility and its cognitive processes in wayfinding. Third, performance on the vMWM prioritizes the use of allocentric navigation strategies, as its design naturally minimizes reliance on proceduralized and egocentric strategies. As such, individuals who favor these strategies may exhibit greater deficits related to strategy preference, and future experimental study is warranted to examine this further. Finally, we found limited evidence for cognitive correlates to wayfinding in this sample, which is not consistent with the broader literature (Blacker et al., 2017; Weisberg & Newcombe, 2016). The cognitive battery administered was drawn from a well-established translational framework commonly used in clinical populations, ensuring methodological relevance and comparability with our other studies examining cognitive-mobility associations. However, these assessments may not have been as sensitive in relatively healthy middle-aged and older adults. Efficient navigation depends on the continuous and flexible engagement of cognitive resources, including working memory span, dynamic executive functions, and attentional control that were not included in the current study (Alessia et al., 2022; Chargo et al., 2026; Wei et al., 2020). Additional study with a more extensive cognitive battery is needed to continue to characterize the specific cognitive architecture underlying the mobility-wayfinding association reported here.

Conclusions

To our knowledge, the findings discussed in the current study are the first to demonstrate associations between independent assessments of mobility and frailty risk with performance on a seated virtual navigation paradigm in a sample of healthy middle-aged and older adults. The reported results underscore that pre-clinical mobility decline may be a potential early indicator of cognitive vulnerability that extends to complex everyday behaviors such as spatial navigation. Incorporating assessments of wayfinding into clinical or community-based evaluations may therefore offer a sensitive and ecologically relevant means of identifying early risk for cognitive-mobility decline and informing targeted interventions aimed at promoting functional independence into life.