Relevant literature was identified by systematically searching the electronic databases Pubmed, Web of Science, CINAHL, and EMBASE. Systematic searches were conducted from the inception of the respective database up until September 2021. The search strategy was developed and refined according to the PICOS scheme (P = healthy older adults, I = training of spatial navigation, C = control groups, O = motor or cognitive or spatial ability outcomes, S = study design), advised by a librarian. All search results were exported into Mendeley (version 1803). After the removal of duplicates, the remaining citations were uploaded to Rayyan QCRI (Ouzzani et al., 2016), a free-of-charge web application that allows collaboration on systematic reviews. Two reviewers (MF, CM) independently screened the titles and abstracts of all results for eligibility according to the inclusion and exclusion criteria. In case the information provided in the title and abstract was insufficient to decide whether an article should be included or excluded, the full-text version of these articles was retrieved and screened for eligibility. Afterward, full-text versions of all potentially relevant studies were obtained and assessed for inclusion by both reviewers. Disagreements were resolved by discussion and consensus. The PRISMA Flow Diagram (Moher et al., 2011) shows the results of the study selection process and documents the reasons for study exclusion (Figure 1) and an example of the coherent search string for the final search on each database is depicted in Appendix 1. For PubMed and CINAHL the results were limited to humans by using the filter function. Grey literature was searched through Google Scholar. Next to that, reference lists of identified articles and review articles in the associated research field were hand-searched to detect additional relevant publications that were not identified by the databases.

Both reviewers documented their data charting process with a modified version of the ‘Data collection form for intervention reviews: RCTs and non-RCTs’. Applied items were key study characteristics like the author(s), year of publication, study design, the aim of the study, population characteristics, intervention characteristics, control interventions, follow-ups, drop-outs, assessed outcomes, and self-reported limitations.

Additionally, information was collected about the content of spatial navigation interventions, training modalities, spatial ability-related outcome measures and outcomes. It should be noted, that only behavioral spatial ability-related outcome measures are reported and discussed, disregarding additional reported data (e.g., imaging data). Data were listed as a mean and standard deviation (SD) if reported. To capture all relevant information, the form was refined in an iterative process and data were updated continuously. Extracted data of both researchers were compared and discrepancies were resolved by discussion. According to the research questions, data about study design, study aim, population and intervention characteristics, spatial ability-related outcome measures and outcomes were used to identify relevant studies and to summarize the applied spatial ability-related assessments and training effects on spatial navigation abilities.

Included studies met the following criteria: (I) published in the English language, (II) accessible as a full-text version, (III) at least one group of healthy older adults included with a (IV) mean age of 65 years or older, (V) three or more spatial navigation training sessions conducted and (VI) at least one spatial ability-related outcome reported.

Studies were excluded if they were (I) only cross-sectional or (II) books, reviews, editorials, single case studies, conference abstracts, or interviews. A sample of solely (III) pathologically impaired participants also led to exclusion as well as (IV) the absence of spatial outcomes.

A quality assessment was performed according to Kmet et al. (2004). This critical appraisal was performed to depict the methodological quality of the included articles and to discriminate the meaningfulness of derived evidence. The original quality assessment contains 14 criteria, of which 11 were applicable to the included studies (cf. criteria 1–11 in Table 1). For each criterion, a maximum of two points was given by two independent reviewers, depending on the degree to which the criterion was met (“yes” = 2, “partial” = 1, “no” = 0, “x” = n/a). Discrepancies were resolved by discussion. For each article, a total score was summed up across relevant items and divided by the total possible score reduced by the number of not applicable criteria times two [i.e., if 10 criteria were fully met but one was not applicable, the score was 20/(22–1*2) = 1]. Due to the limited literature available on this topic, the quality assessment was not used as an inclusion criterion but as a measure of study quality. Referring to Curry et al. (2021) and Lannes et al. (2021) studies with a score of 0.80 and above were considered high quality, whereas scores between 0.70 and 0.79 were defined as good quality. Studies scoring between 0.50 and 0.69 were classified as moderate quality. Studies below the score of 0.50 were classified as poor.

As shown in Figure 1, the systematic search within the databases PubMed, Web of Science, CINAHL and EMBASE identified 2,927 articles for consideration. Three hundred-five duplicates were removed. After title and abstract screening, a total of 60 full-text articles were included for further examination. The main reasons for exclusion were for example participants, that did not meet the inclusion criteria for age, pathological impaired samples without a healthy control group, or the absence of spatial navigation interventions or spatial ability-related outcomes. Two of the studies were dropped due to publication in Italian. In the end, 10 articles were incorporated into the qualitative synthesis.

Table 2 gives a comprehensive overview of the general study characteristics of the included articles. Table 3 provides a deeper insight into the interventions performed, and their training modalities as well as the spatial ability-related assessments used. It further briefly presents the findings concerning the applied spatial ability-related measures.

Summarizing all records, a total of 1,003 adults participated in the studies (cf. Table 2), this includes younger or pathologically impaired comparison groups. Participants’ age ranged from 20 to 95 years, however, the older age group that was further assessed had a mean age of 65 years and above. Overall, 516 women and 487 men were included in the research activities. Three hundred-sixty-eight of these participants were healthy older adults receiving a spatial navigation-related intervention, while the others either had pathologic impairments or had been younger, healthy control participants. One study included men only (Lövdén et al., 2012) and two studies included considerably more female participants (Mitolo et al., 2017; Serino et al., 2017), the gender distribution in the remaining studies was approximately even.

Studies included in this systematic scoping review used heterogeneous designs and differed significantly in methodological quality (Table 1). Six studies were rated as high quality while one study was of good quality. Three studies were classified as moderate quality, none of the studies were considered of poor quality. It should be noted that the studies with the lowest scores omitted to provide sufficient and structured descriptions of their randomization processes, thereby reducing reproducibility. For two of them, this may be a concern of outdated reporting structures since they were published in the 1980s (Willis and Schaie, 1986; Schaie et al., 1987). In summary, the overall quality of the study pool is moderate to high.

The included studies (N = 10) differed in the interventions provided as well as the training methods (cf. Table 3). Whereas one study implemented a group intervention (Mitolo et al., 2017), all remaining studies performed one-on-one trainings. Of those, two studies trained their participants at home, one guided by a trainer (Willis and Schaie, 1986), the other one unaccompanied (Whitlock et al., 2012). All other studies provided training to their participants in a laboratory setting. In terms of training methods, one study implemented iPad-based mini-games (Binder et al., 2016), another study instructed participants to play an original computer game (Whitlock et al., 2012). In four studies participants performed active and passive movements in virtual environments (Lövdén et al., 2012; Wiener et al., 2012; Kober et al., 2013; Serino et al., 2017), one study conducted real-world behavioral training (Mitolo et al., 2017), one trained picture-based (Nemmi et al., 2017), and another study proceeded real-world spatial activities as well as cognitive spatial activities (Willis and Schaie, 1986). Schaie et al. (1987) did not provide detailed information about the type of training implemented. Moreover, training modalities differed across the studies, i.e., training periods ranged from one to 16 weeks with the number of sessions extending from five to 50 sessions in total, while the overall training duration varied from 100 min to 50 h with individual training sessions extending from 20 to 60 min. Two studies did not describe their training modalities in sufficient detail (Wiener et al., 2012; Nemmi et al., 2017). Finally, training frequency varied from one to seven sessions per week.

All experimental designs applied at least one form of spatial outcome measure with most studies utilizing more than one spatial ability-related assessment (cf. Table 3). These assessments recorded (visuospatial) short-term memory, variations of the Corsi Block Test (Kober et al., 2013; Binder et al., 2016; Mitolo et al., 2017; Serino et al., 2017), the City Map Path Learning Test (Binder et al., 2016), Pathway Span Task, Jigsaw Puzzle Test (Mitolo et al., 2017), the Spatial 2-back (Lövdén et al., 2012) or the Digit Span Test (Serino et al., 2017). Visual perception and visual memory were rated with the Benton Visual Retention Test and the Visual Pursuit Test (Kober et al., 2013). Moreover, spatial perception and mental rotation abilities were examined utilizing the 3D Spatial Orientation Test (Binder et al., 2016), Primary Mental Abilities Space, Object Rotation (Willis and Schaie, 1986; Schaie et al., 1987), Alphanumeric Rotation and the Mental Rotation Test (Lövdén et al., 2012; Whitlock et al., 2012), the Spatial Imagination Test (Kober et al., 2013) or Cube Comparison (Willis and Schaie, 1986). Further, Lövdén et al. (2012) and Kober et al. (2013) assessed VR route-finding performance or navigation performance tasks to evaluate route learning and route knowledge, while Wiener et al. (2012) and Nemmi et al. (2017) implemented intervention-specific route and direction tasks. Alternatively, the Route Learning Test and the Landmark Sequence Test were used to measure these spatial outcomes. Spatial orientation was assessed by the Object Perspective Tests (Whitlock et al., 2012) and the Guilford-Zimmerman Test (Lövdén et al., 2012). The Route Memory Test, the Location Memory Test and Object Position Memory were applied to test episodic memory (Lövdén et al., 2012). Moreover, executive function and spatial visualization were assessed using the Frontal Assessment Battery (Serino et al., 2017) and the Paper Folding Test (Whitlock et al., 2012), respectively. Mitolo et al. (2017) further applied various spatial questionnaires like the Sense of Direction and Spatial Representations Questionnaire, the Attitude to Environmental Tasks Questionnaire, the Spatial Anxiety Scale, and the Spatial Self-Efficacy Questionnaire. It should be noted that the majority of the reported assessments only partially depict spatial aspects of the complex cognitive process of spatial navigation. Therefore, they might reflect spatial abilities that contribute to spatial navigation to a certain degree, but are not sufficient to map this complex cognitive ability in its entirety.

A total number of 10 studies met the applied inclusion criteria and were subjected to a quality assessment. The quality of the included studies was moderate to high. Trials included a total number of 1,003 participants, 368 of those being healthy older adults and therefore target group of this scoping review. Overall, the study outcomes indicate that spatial abilities can be improved in healthy older adults by means of the designed training interventions. Six of nine studies reporting pre-post data found significant improvements for at least one of their spatial ability-related outcomes (Willis and Schaie, 1986; Lövdén et al., 2012; Wiener et al., 2012; Kober et al., 2013; Mitolo et al., 2017; Nemmi et al., 2017), however, applying a methodological variety of training protocols in terms of type of intervention, training periods and training durations. Aiming to identify the most relevant approaches to achieve spatial ability improvements in healthy older adults, several research gaps in the current literature were identified. While the available research is of moderate to high quality, the total number of published studies is low. Nevertheless, there is need for additional methodologically solid research including larger sample sizes, addressing specifically healthy older adults.

The first aim of this scoping review was to summarize the characteristics of existing spatial navigation interventions to enhance spatial abilities in healthy older adults. A total number of 10 different interventions could be identified. Overall, the implemented training methods varied widely, active and passive movement approaches in virtual environments (Lövdén et al., 2012; Wiener et al., 2012; Kober et al., 2013; Serino et al., 2017) were used most frequently, followed by gamification approaches (Whitlock et al., 2012; Binder et al., 2016), as well as picture-based, cognitive, and real world-approaches (Willis and Schaie, 1986; Mitolo et al., 2017; Nemmi et al., 2017). A recent meta-analysis of combined cognitive-motor interventions revealed positive effects on global cognition in older adults (Wollesen et al., 2020). Despite these promising findings only two of the studies included in this review applied this dual-task type of training protocol (Lövdén et al., 2012; Mitolo et al., 2017) and improved spatial ability-related outcomes successfully. However, the extent of the motor components concurrent to the cognitive training was not stated clearly by Mitolo et al. (2017).

This variety continued for the applied training periods (1–16 weeks), the overall training duration (100 min–50 h) and the duration of training sessions (20–60 min).

From an exercise science point of view, the effect of cognitive-motor interventions is highly dependent on the training control and the conduction of the training load (e.g., progression, intensity; Herold et al., 2019; Wollesen et al., 2020). Improvements in cognitive-motor performance were shown following a minimum overall duration of 330 min of progressive training (Wollesen and Voelcker-Rehage, 2014). This duration was reached in two of the included studies (Lövdén et al., 2012; Mitolo et al., 2017), both showing improvements in spatial ability-related outcomes. In contrast, Kober et al. (2013) found significant improvements in implementing cognitive training in virtual reality with a total training duration of 100 min.

The second aim was to give an overview of spatial ability-related measures to evaluate spatial navigation intervention effects in healthy older adults.

It became evident that the applied measures varied widely from established standardized tests (e.g., variations of the Corsi Block and Mental Rotation Test) addressing cognitive functions (e.g., visual–spatial short-term memory, mental rotation) to task-specific tests (e.g., route learning, survey learning) and spatial ability-related questionnaires. Six of the included studies applied standardized tests (Willis and Schaie, 1986; Schaie et al., 1987; Whitlock et al., 2012; Binder et al., 2016; Mitolo et al., 2017; Serino et al., 2017), while two of them used task-specific assessments (Wiener et al., 2012; Nemmi et al., 2017). It has to be assumed that spatial navigation is a complex multisensory process (Wolbers and Hegarty, 2010). However, a clear definition of spatial navigation and related cognitive processes is missing within the included studies. This hampers the attempt to link spatial cognitive tests with spatial navigation abilities.

Kober et al. (2013) as well as Lövdén et al. (2012) utilized a combination of standardized and task-specific tests. When evaluated by task-specific measurements (Wiener et al., 2012) or a combination of task-specific and standardized measurements (Lövdén et al., 2012; Kober et al., 2013), VR-based interventions led to significant improvements in at least one outcome measure. In contrast, when evaluated by standardized tests only, the VR-intervention failed to show significant improvements regarding spatial navigation abilities (Serino et al., 2017). Two gamified interventions made use of established outcome measures (Whitlock et al., 2012; Binder et al., 2016), both failing to reveal significant improvements. However, Binder and colleagues reported statistical trends of improvement for all their outcome measures. This might be due to the fact that Binder et al. (2016) developed iPad-based mini-games to specifically improve cognitive functions, while Whitlock et al. (2012) used a pre-existing computer game which had been developed without this purpose (World of Warcraft).

The studies performing cognitive training interventions using realistic scenarios (Nemmi et al., 2017: screenshots of existing crossroads; Mitolo et al., 2017: living areas in nursing homes), used either standardized measures (e.g., Forward Corsi Block Test; Mitolo et al., 2017) or task-specific measures (Route Task & Survey Task; Nemmi et al., 2017), both finding significant improvements in at least some of their spatial ability-related outcomes. In the study of Willis and Schaie (1986), a conceptual cognitive training and established assessments were conducted, and significant improvements were observed.

This overview depicts a broad variety of applied spatial ability-related outcome measures in relation to the different intervention approaches. The applied assessments depict partial aspects of the complex cognitive ability called spatial navigation, which needs to be considered when interpreting outcomes in terms of intervention effects. The majority of assessments capture small-scale spatial abilities only, which mainly cover two-and three-dimensional spatial visualization and spatial relations. Few of the used assessments survey large-scale spatial abilities, which refer to the concrete ability to orientate and navigate in complex large-scale environments (Yuan et al., 2019).

This does not allow for drawing conclusions on the eligibility of selected spatial ability-related outcome measures. Overall, it can be assumed that lab-based, standardized assessments of spatial abilities are suitable tools in basic research as they can depict specific cognitive aspects. At the same time, task-specific large-scale assessments might be more suitable to observe intervention effects since they illustrate the complex aspects of spatial navigation performance rather than underlying cognitive mechanisms. Furthermore, these tools could also be considered to reflect the transfer of the measured spatial ability to everyday spatial navigation more accurately.

The third study aim was to identify effective spatial navigation training interventions for healthy older adults. Effect sizes generally provide statistical evidence for comparing the range of intervention effects across studies. Six of nine studies reported statistically significant improvements in at least one of the spatial ability-related outcomes post training. Only one of these studies (Mitolo et al., 2017) reported effect sizes reflecting training gains for spatial measures in healthy older adults. Mitolo et al. (2017) calculated effect sizes using Cohen’s d and found a range from 0.1 up to 1.3, reflecting small to large intervention effects for the spatial measures (Cohen, 1992), which significantly improved after training. Some of the other studies also examined effect sizes comparing intervention and control group, but not pre-and posttest data for each group. Since no comparable training effects can be derived from these data, future studies should report the related effect sizes to enable a cross-study evaluation of training effects.

To classify the relevance of specific training approaches, it is also important to evaluate their transfer effects on everyday functioning (Weng et al., 2019). Transfer effects can be divided in near vs. far transfer effects (post-training improvements of the trained tasks vs. improvements in cognitive function and untrained cognitive tasks or performance in everyday situations, respectively, Barnett and Ceci, 2002). Improvements were detected for both forms of interventions addressing cognitive functions like fluid intelligence (Baltes et al., 1989), working memory (Strobach and Huestegge, 2017), or attention (Cassavaugh and Kramer, 2009). Two of the included studies generated near transfer outcomes (Wiener et al., 2012; Nemmi et al., 2017) conducting task-specific assessments (e.g., route learning in a virtual environment), while four of them collected rather far transfer outcomes referring to cognitive functions by standardized measurements (Whitlock et al., 2012; Binder et al., 2016; Mitolo et al., 2017; Serino et al., 2017), but not everyday life spatial navigation performance. Furthermore, Lövdén et al. (2012) and Kober et al. (2013) evaluated both near and far transfer effects, although they also did not cover the everyday life spatial navigation performance. For the studies performed by Schaie et al. (1987) as well as Willis and Schaie (1986) an assignment of transfer effects was inconclusive, as the intervention contents were not clearly described. Overall, none of the included studies applied assessments examining everyday spatial navigation performance. This renders the evaluation of intervention effects in terms of spatial navigation improvements in everyday life impossible.

Nonetheless, the encouraging findings of this scoping review are particularly valuable when assuming an association of efficient spatial abilities and life space-mobility. Since the constriction of life space increases the risk of frailty (Xue et al., 2008), nursing home admission (Sheppard et al., 2013), cognitive decline (Crowe et al., 2008) and social isolation (Taylor et al., 2019), it seems all the more sensible to counteract such life space reduction by promoting and maintaining spatial navigation abilities. Whether these findings are transferable to other population groups needs to be investigated.

The main strength of this review is its structured and thorough methodological approach in line with the PRISMA Extension for Scoping Reviews (PRISMA-ScR; Tricco et al., 2018). As the literature search was performed in a structured, and reproducible manner, it can be assumed that nearly all relevant literature in this underexplored research field has been identified. This allows a comprehensive overview of the available evidence as well as the accentuation of gaps in knowledge. In addition, a quality assessment of the included studies was performed to classify their methodological quality as well as the significance of derived evidence.

Despite conducting a thorough search of the literature, only 10 studies met our inclusion criteria. This may be due to a variety of limiting factors. Even though the search strategy was inclusive, it cannot be ruled out that our choice of keywords and search strings impeded the detection of available literature. Further, only articles published in English were included, which might have led to the exclusion of additional relevant evidence in this emerging field of research. It appears that only little research has investigated this topic for this specific population group applying a structured study design including a pre-and posttest as well as at least three spatial navigation-related intervention sessions. Altogether, sample sizes of healthy older adults involved in the included studies were relatively low as this target group frequently served as control-or comparison group for individuals with pathological impairments or different age groups. Overall, the 10 included studies presented with a strong heterogeneity regarding their study designs, population characteristics, interventions, training modalities and outcome measures. This significantly hampered comparability of their findings. In line with the scope of this review, results were thus presented in a descriptive manner. Moreover, there is a need for diagnostically informative, comparable outcome measures with high reliability and validity that convey a meaningful picture of intervention effects.