Navigation in VR versus navigation in the real world

As pointed out by Taube et al. (2013), it is important to be aware that navigation in VR is not the same as navigation in the real world. In many VR setups (Fig. 1), sensory input is mainly provided via the visual system, and this input can differ from the input provided by other senses or the motor system. During passive navigation or in brain-imaging experiments, only optic flow induces the perception of movement through space whereas the participant's body remains static. This discrepancy between body-based and visual cues may lead to differences in the encoding of VR versus real-world environments, and the potential for transfer of spatial knowledge from one modality to the other might be limited. Ruddle et al. (2011b), for example, showed that participants who physically walk in VR are better at following and subsequently retracing a route than participants who only make physical rotations and use a joystick to translate (see Ruddle et al., 2011a, for a similar finding on allocentric processing). Coutrot et al. (2018 preprint) compared user data from the abovementioned SHQ mobile video game app with navigational performance in a real-world environment. A positive correlation between navigating in VR and the real world was found, with an advantage for male over female participants, which was more pronounced in VR. Although the authors did not assess the general VR experience of their participants, one might speculate whether this increased gender difference is related to differing degrees of VR and gaming experience in males versus females. There is indeed evidence that video game experience is related to navigational performance in desktop and immersive VR setups but not in real-world environments (Richardson et al., 2011). Importantly, gender differences in performance were largely attenuated in this study when video game experience was included as a covariate in the analysis. Murias et al. (2016) extended these findings by showing that experience in playing video games with navigational components but not just the experience in using game controls improved navigational performance in terms of the utilization of different navigational strategies (allocentric versus egocentric). Thus, with increasing experience in playing games such as SHQ, female players are likely to catch up with their male counterparts in the context of navigation in VR.

One should also keep in mind that participants might adapt their navigational strategy and, for example, exploit different sources of information differently depending on the task context, even if overall navigation performance is comparable between VR and real-world navigation. For example, it has recently been shown that participants rely more on distinct features (i.e. landmarks) than on the geometrical layout when asked to reorient themselves in a virtual room using an HMD in combination with a manual wheelchair as compared with navigation in a real room (Kimura et al., 2017).

In contrast to traditional navigation paradigms using static stimuli or videos, however, VR is able to create a more realistic sense of ‘being there’ than ever before. The degree of immersion in VR can be enhanced in multiple ways; for example, by integrating the participant's body into the virtual environment. Phenomena such as the rubber hand illusion (i.e. integrating external objects into the representation of the own body by synchronous visuotactile stimulation) illustrate that body representations can be manipulated quite easily in humans (see also Slater and Sanchez-Vives, 2016; Spanlang et al., 2014). Thus, through achieving synchrony between different sensations (visual–proprioceptive: virtual body is presented where it should be, visual–motor: virtual body moves in the same way as own body, visual–tactile: synchronous touch), a feeling of embodiment can be created that facilitates performance in contexts where participants do not see their own body; for example, when they wear a HMD or during brain imaging. Research in younger adults shows that the experience of having a body in VR or the sense of a virtual self facilitates performance in tasks requiring, for example, mental rotation or perspective taking (Pan and Hamilton, 2018).

Applicability of VR to study spatial navigation across different animal species

To date, there are very few studies comparing the neural processing of navigating in VR versus the real world in animals. However, in animal cognition too, VR offers major benefits over classical testing in ‘real-world’ conditions (e.g. mazes). For example, in rodents, the vast majority of studies on place cells, grid cells and other cells implicated in spatial coding have been conducted in small-scale or vista environments (i.e. circular or rectangular boxes). While many important insights have been gained from these studies, navigation in the real world occurs in much more complex environments that cannot be apprehended from a single vantage point (Wolbers and Wiener, 2014). Given that the mechanisms that govern vista- versus environmental-scale navigation only partially overlap, it is important to move beyond traditionally used experimental paradigms. In this context, VR setups (Fig. 2) offer the unique opportunity to overcome these limitations and to study navigational behavior and the underlying neural mechanisms in realistic, large-scale environments.

Moreover, in many rodent VR setups, the animal is placed on a free-moving ball with its head or body fixed while visual input is provided over a screen that covers its entire field of view (Fig. 2A; Minderer et al., 2016). In this way, precise control of the sensory cues that carry relevant information for navigation is possible and each cue's contribution to the animal's actions can be dissociated. For example, Campbell et al. (2018) used VR to systematically manipulate the gain of self-motion and landmark information while head-fixed mice traversed a linear track for rewards and recorded activity from different cell types in the MTL (i.e. grid, border and speed cells). The manipulation was implemented by changing the ratio of the rotation of the ball the mice were placed on to the speed of the optic flow. Using tetrode recordings, the results show that border cells responded to landmark information, whereas the other cell types responded to both cue types and were sensitive to changes in the relationship between the movement of the ball and the visual translation on the screen, a finding that would not have been possible without VR.

In conditions of head or body fixation, however, vestibular input is minimized and consequently is in conflict with the visual input provided by the VR system, which may alter the way the animal encodes the environment. Aghajan et al. (2015) showed that the spatial selectivity and stability of place cells in area CA1 of the hippocampus are largely reduced when body-fixed rats engage in random foraging in a 2D virtual environment as compared with real-world navigation. Moreover, only 40% of the measured cells were activated in VR. Interestingly, spatial selectivity was enhanced in VR when rats traveled to fixed locations to receive a reward; that is, when locomotion cues were more spatially informative and repeatedly paired with distant visual cues. Acharya et al. (2016) found that CA1 place cells that are modulated by the head direction of the animal during random foraging in a real-world environment show a similar modulation in VR, despite the absence of vestibular cues in this setting. Thus, despite impaired spatial selectivity, visual cues appear to be sufficient for generating directional selectivity in hippocampal place cells in rodents.

Together, these studies illustrate how VR can be applied to gain novel insights into the neural mechanisms of spatial navigation in rodents. However, there are important differences between rodent and primate spatial navigation (see Zhao, 2018, for a recent review). In humans, body-based cues may not always be critical for efficient navigation (e.g. Waller and Greenauer, 2007). A greater reliance on visual cues in humans and non-human primates during navigation may lead to differences between the representation of VR in comparison to real-world environments across species (Ekstrom, 2015). In rodents, the proportion of photoreceptors on the retina is substantially lower and they have limited binocular overlap and are more dependent on head movements to cover their visual surroundings (Huberman and Niell, 2011). The primate visual system, in contrast, is able to extract spatial information over larger distances, and some spatial cell types observed in the primate MTL are primarily tuned to visual information (Ekstrom et al., 2003). Moreover, Killian et al. (2012) found evidence for grid cells in the entorhinal cortex of non-human primates during visual exploration of distant space, a finding that was subsequently replicated and extended in humans by means of functional magnetic resonance imaging (fMRI), suggesting that the representation of the visual search space in 2D may drive the signal (Julian et al., 2018; Nau et al., 2018b). These findings suggest that primates, in contrast to rodents, may entertain multiple representations of space in parallel when lying in the scanner or when navigating in VR, resulting in a larger overlap between the processes that operate during navigation in VR versus the real world. Thus, there is growing evidence that visual exploration of space shares important features with the actual exploration in the primate brain (Nau et al., 2018a).

Challenges when using VR to study older age groups

Despite the substantial benefits that VR offers, it is important to be aware of several challenges when using this approach with older age groups. First, older adults are often less experienced than younger adults with new technologies such as VR, which may result in stress and anxiety when they are asked to use them (Barnard et al., 2013). This may become further pronounced in experimental settings when they are eager to perform well in order to make a good impression. Most of today's older adults grew up in an analog world and may have never used the Internet, played videogames or walked in a 3D virtual environment. Their understanding of very basic digital and technological concepts such as using icons to navigate a website and peripherals for navigation in VR might consequently be limited (see Castilla et al., 2013, for the evaluation of a multi-application system targeted at older adults based on Internet technology and VR). Including only older adults with VR experience in studies using VR paradigms, in contrast, may result in a selection bias, making it difficult to generalize the results to the whole aging population. Whether older adults are inclined to adopt new technologies depends on factors such as past experiences with technology, their self-perception (i.e. ‘I am too old to learn new things’), the intention to learn, the characteristics of the interface and the availability of emotional and technical support (Barnard et al., 2013).

Second, cybersickness may occur if sensory inputs from the real world are in conflict with those from VR. Cybersickness is similar to motion sickness, including symptoms of nausea, headache and eye strain, and has been linked to individual factors (age, gender, VR experience), characteristics of the display system (e.g. screen size, resolution), and task design (e.g. the speed of movement and rotation, the smoothness of control, the predictability of events; Renkewitz and Alexander, 2007). Older adults seem to be more susceptible to cybersickness than younger adults (Arns and Cerney, 2005). Liu (2014) showed that scene rotation speed and the duration of exposure in VR contributes to the experience of cybersickness in older adults during passive navigation in a virtual environment displayed on a standard 2D computer screen. No adverse side effects were reported in a sample of older adults when VR was used to trigger autobiographical memories by presenting scenes on a large 320×240 cm screen including 3D sound and finger tracking that allowed in-place navigation and interaction with objects (Benoit et al., 2015). Thus, besides age, additional factors such as general VR experience and the specifics of the task at hand may mediate older adults' susceptibility to cybersickness. The less experienced they are together with experimental manipulations that, for example, are less realistic or that they do not expect, the more cybersickness they may experience.

Third, as outlined above, successfully navigating in a virtual environment requires a certain degree of immersion in VR, which can be facilitated by the experience of having a body in VR. However, there is evidence that older adults are actually less embodied than younger adults (Costello and Bloesch, 2017; Diersch et al., 2016, 2013; Kuehn et al., 2018). Presenting an avatar or body parts might consequently support task understanding but not necessarily the degree of immersion older adults experience in VR. Determining the specific circumstances and contexts (e.g. reference frames or navigational strategies) for which the presence of a body or an avatar increases immersion in VR through embodiment and consequently supports navigational learning is therefore an important goal for future research.

Fourth, the presentation of additional information such as an avatar or a high degree of interaction in VR can be distracting and can in turn negatively affect older adults' navigational performance as a result of age-related changes in attentional control. Age-related deficits in the allocation of attention to relevant information while inhibiting irrelevant information has been associated with performance declines in a variety of context-dependent tasks (Gazzaley et al., 2005; Lustig et al., 2007). In line with this, Merriman et al. (2018) found that the presence of moving avatars in a virtual environment resulted in performance decreases during spatial learning in older but not younger adults compared with navigation in an empty environment. This effect was specific to the presence of conspecifics and no such performance declines were observed when moving objects were embedded in the scene. Measuring different aspects of episodic memory encoding in younger and older adults, Sauzéon et al. (2015) showed that active versus passive exploration of a virtual room reduced false object recognition in younger adults but increased it in older adults, which was linked to their executive functioning abilities.

Fifth, during active navigation using an HMD or a treadmill, age-related changes in sensorimotor control (i.e. difficulties in balance and gait, higher movement variability) may require older adults to prioritize sensorimotor processing over cognitive processing in order to reduce the risk of falling in multiple-task situations (Schäfer et al., 2006). In line with this, Lövdén et al. (2005) showed that older but not younger adults' navigational performance during walking improves if a walking aid (i.e. holding on to a handrail while walking on a treadmill) is provided. Virtual guidance during walking in VR on a treadmill has also been found to attenuate age-related differences in navigational performance and walking variability (Schellenbach et al., 2010b).

How to address age-related challenges when using VR

Some of the problems mentioned in the previous section, particularly those linked to the first two challenges, are presumably cohort specific and might be alleviated in future generations as they become more and more used to the integration of digital devices and VR in their everyday life. In the meantime, we recommend a number of measures researchers should consider to ensure that performance data obtained from older adults are not confounded by limited understanding of and suboptimal interaction with the particular VR setup (see Table 1 for a summary). First of all, it is important to provide sufficient training in VR before testing (e.g. until a pre-defined performance criterion is reached). For example, Schellenbach et al. (2010a) showed that 20 min of training to walk on a treadmill in virtual environments is sufficient to reach stable walking patterns in older and younger adults. Providing support in the form of walking aids or virtual guidance may free up cognitive resources required to focus on navigational task demands. The virtual environment, the input devices/controllers and the task at hand should be introduced step by step in the training phase to reduce the participants' stress level due to too much information presented at the same time. The experimenter may additionally demonstrate the task at the beginning of the experiment. We further suggest simplifying the experimental setup as much as possible (e.g. using button boxes instead of joysticks if applicable) and ensuring that the movement and rotation speed in VR is quite low (e.g. close to the biological walking speed of around 1.6 m s⁻¹). Creating the impression of a physical presence in the virtual environment, while avoiding presentation of too much distracting information, can further facilitate older adults' performance. For example, the interactive nature of VR can be stressed by including a virtual experimenter who directly interacts with the participant and practices the task with them. Finally, the participants' level of computer literacy can be assessed beforehand by means of questionnaires such as the computer literacy scale (CLS; Sengpiel and Dittberner, 2008). Their level of cybersickness can be determined with the Simulation Sickness Questionnaire (Kennedy et al., 1993). Participants who are not used to computers or whose risk of developing cybersickness is very high, might consequently be excluded from testing. For behavioral studies, instead of using standard desktop PCs, mobile devices (e.g. tablets) might be easier to operate for older adults because they are intuitively usable (Barnard et al., 2013).