Learning from navigation, and tasks assessing its accuracy: The role of visuospatial abilities and wayfinding inclinations

Introduction

Navigation is an essential everyday activity, in which we experience space from a personal point of view, based on sensorimotor information about our position in space, self-to-object distances, and self-motion. This information enables us to learn a sequence of landmarks, turns, and changes of direction, and to memorize a set of place-action associations. It is a complex ability involving sensory cues, computational mechanisms and spatial representations (Wolbers & Hegarty, 2010). We can distinguish between a locomotion component, involving our body movements coordinated with local and proximal surroundings, and a wayfinding component, which entails an efficient goal-directed and planned movement through an environment (Montello, 2005). Learning from navigation can be done using virtual environments (VE) as a good approximation of the real world (Richardson et al., 1999). The resulting mental representations, or cognitive maps (Tolman, 1948), can be assessed by asking participants to recall environmental information using measures such as those based on estimating distances or drawing a sketch map, or to move within an environment, repeat a previously-seen route or find new ones (e.g., Denis, 2018; Lhuillier & Gyselinck, 2020).

The present study examined the environmental information acquired from VE navigation, using different tasks to assess the resulting mental representation, and how its accuracy related to visuospatial individual difference factors.

One interest of research in spatial cognition lies in identifying factors capable of optimizing the accuracy of our environmental representations. These factors involve individual differences in spatial thinking (Hegarty & Waller, 2005), including our cognitive abilities (objectively measurable) and our inclinations (self-reported environment-related preferences). Below, we describe the main visuospatial abilities and inclinations that have been considered in relation to navigation learning accuracy.

Visuospatial cognitive abilities can be defined as skills needed to generate, retain and transform abstract visual images (Lohman, 1988). They comprise a set of distinct skills (see Hegarty & Waller, 2005; Uttal et al., 2013), including the abilities needed to make transformations, as in mental rotation. This mental turning of objects is measured with the Mental Rotations Test (MRT; Vandenberg & Kuse, 1978), which involves identifying 3D objects in rotated views. Another distinct but related type of rotation (Hegarty & Waller, 2004), which we could call subject rotation, involves asking a person to imagine adopting a different position in space, and seeing their surroundings from a new perspective. The ability to do so can be measured with the Perspective-Taking Task (PTT; Hegarty & Waller, 2004), in which respondents imagine adopting new positions within a configuration of objects. Both these rotation skills have been round related to accuracy in learning from navigation (using VEs; Allen et al., 1996; Fields & Shelton, 2006; Kozhevnikov et al., 2006; Lokka & Çöltekin, 2019; Moffat et al., 1998; Münzer & Stahl, 2011; Ruginski et al., 2019; Weisberg et al., 2014). When both rotations (object rotation and perspective taking) are considered at the same time, they can be treated as a single latent factor (using a Structural Equation Modeling approach), which has been found to affect spatial task performance (as in map-related and pointing tasks; Hegarty et al., 2006; Meneghetti et al., 2014). They can also be considered separately, and may both go in the same direction in predicting accuracy in spatial task performance, when tracing a route, for instance (Moffat et al., 1998). The effect of one may sometimes prevail, however, on recall task performance (map-related and pointing tasks) involving object rotation (Meneghetti et al., 2016; Weisberg et al., 2014), or perspective taking (Allen et al., 1996; Fields & Shelton, 2006; Muffato et al., 2020; Ruginski et al., 2019).

Some studies on navigation considered not only rotation abilities but also visuospatial working memory (VSWM), which is the ability to retain and process spatial information. While rotation is considered a higher-order cognitive ability, and sometimes part of spatial intelligence (Martínez et al., 2011), VSWM is a basic mechanism involved during on line processing (e.g., examined with dual-task paradigm; Hund, 2016; Labate et al., 2014). It is also an individual capacity to process information sequentially (Hegarty et al., 2006; Meneghetti et al, 2014, 2016; Münzer & Stahl, 2011; Pazzaglia et al., 2018), or simultaneously (Fields & Shelton, 2006; Lokka & Çöltekin, 2019; Muffato et al., 2019, 2020). While few studies considered VSWM alone (Muffato et al., 2019), some researchers considered both VSWM and rotation abilities. These abilities can be pooled into a single latent factor capable of affecting navigation performance (Hegarty et al., 2006; Pazzaglia et al., 2018). When visuospatial (rotation) tasks are kept separate from VSWM tasks, the influence of the latter may be less prevalent (Fields & Shelton, 2006; Münzer & Stahl, 2011), or work on a different level, as a mediator between variables (Meneghetti et al, 2014, 2016; Muffato et al., 2020).

Overall, reviewing the above literature suggests that visuospatial abilities (VSWM and rotation) can be considered together as a single visuospatial (or spatial) factor that contributes to supporting accuracy in learning from navigation in tasks that require navigating again within the same environment (as in route tracing and shortcut finding (Pazzaglia et al., 2018)) or using other measures as in pointing and map-related tasks (Hegarty et al., 2006; Meneghetti et al., 2014).

In addition to visuospatial abilities, personal inclinations may contribute to our ability to mentally represent an environment accurately. Such inclinations are generally assessed with questionnaires measuring wayfinding attitudes and preferences, asking people to imagine being in an environment (e.g., a car park or a building) and to think how they feel and behave in such situations. These wayfinding inclinations, or visuospatial self-assessments (Meneghetti et al., 2014, 2020) include perceived sense of direction (Hegarty et al., 2002), preferred environment representation mode (as survey-like or route-like; Lawton, 1994; Pazzaglia & Meneghetti, 2017), pleasure in exploring places (Meneghetti et al., 2014), and spatial anxiety (Lawton, 1994). Some studies focusing on the effects of wayfinding attitudes and preferences on navigation performance demonstrated the effects of: a high spatial anxiety; a good orientation strategy based on a configured environment (in pointing (Lawton, 1996)); a strong sense of direction (in route tracing (Hund & Nazarczuk, 2009), shortcut finding, pointing and map-related tasks (Labate et al., 2014)); and high pleasure in exploring (in shortcut finding (Pazzaglia et al., 2017)). Interest in elucidating people's wayfinding inclinations is increasing. Studies have shown that wayfinding inclinations are related (De Beni et al., 2014), and can be grouped into a single factor or divided into positive (such as perceived sense of direction and pleasure in exploring) and negative (spatial anxiety and preference for moving in known places) inclinations (Meneghetti et al., 2014, 2020).

When wayfinding inclinations and visuospatial abilities were considered at the same time, they both supported accuracy in learning from navigation (Meneghetti et al., 2014; Muffato et al., 2019; Münzer & Stahl, 2011; Weisberg et al., 2014) - with few exceptions (Fields & Shelton, 2006, detected no influence of people's wayfinding inclinations). Visuospatial abilities have a more evident role than wayfinding inclinations, however, in terms of the values expressing the relations at least (Hegarty et al., 2006; Pazzaglia et al., 2018).

In short, the above-mentioned literature identifies visuospatial abilities and wayfinding inclinations as separate but related factors, and both contribute to environment learning (Hegarty et al., 2006; Pazzaglia et al., 2018). That said, the variety of recall tasks and individual measures used in different combinations is partly responsible for the heterogeneity of the reported results. Some studies focused on single recall tasks, such as pointing (Fields & Shelton, 2006) or shortcut finding (Pazzaglia et al., 2017). Others used various tasks, such as pointing and map-related tasks (Fields & Shelton, 2006; Hegarty et al., 2006; Meneghetti et al., 2014), navigating again in the same environment, with route tracing and shortcut finding (Münzer & Stahl, 2011; Pazzaglia et al., 2018) or combinations of the above (Meneghetti et al., 2016; Muffato et al., 2019, 2020). Performance in these tasks was also related to different individual visuospatial measures. Some studies treated visuospatial abilities as distinct measures (Meneghetti et al., 2016; Ruginski et al., 2019), and others grouped them into a single factor (Hegarty et al., 2006; Meneghetti et al., 2014; Pazzaglia et al., 2018). The contribution of self-assessed wayfinding inclinations varied (partly because different self-reported measures were used), and the combined effects of both visuospatial measures (inclinations and abilities) has generally been less thoroughly examined in relation to environment learning in navigational settings.

The effects of a set of several individual visuospatial abilities and inclinations on accuracy in learning from navigation need to be assessed more systematically.

In the available literature on learning from navigation (based on guided tours along a path in a VE), accuracy was assessed using various recall tasks, but few studies considered multiple recall tasks at the same time. Some studies examined recall task accuracy in relation to individual difference measures, focusing on visuospatial abilities or wayfinding inclinations, but very few considered multiple individual difference measures at once. The aim of the present study was therefore to gain a comprehensive picture of the part played by individual visuospatial abilities and wayfinding inclinations, identifying the best factor solution, and their relationship with several tasks measuring recall of a previously-learned environment in navigation-like condition.

An ad hoc study was designed where a large sample of participants saw a path through a desktop VE from a first-person point of view. A VE was used to control the encoding (for the content presented) and output measures (e.g., Denis, 2018; Lhuillier & Gyselinck, 2020). Specifically, the desktop VE was preferred as it maintains a realistic sense of distance, and can elicit representations similar to real environments or immersive VEs (Waller et al., 2004). It has been widely used in previous studies on the relation between environment learning and individual difference measures (see Table S2 for a review; supplementary material).

After the tour of the desktop VE, we assessed participants’ environment learning accuracy with a series of recall tasks: route retracing (reproducing the previously-seen path); shortcut finding (identifying the shortest route); and landmark locating (extracting locations of landmarks in a layout from a mental representation).

We then administered several visuospatial measures of individual differences that previous literature had examined (albeit singly) and found related to accuracy in environment learning (from navigation): visuospatial cognitive abilities, in terms of VSWM (with the Corsi blocks task), rotation (object rotation and perspective taking), and self-reported wayfinding inclinations (sense of direction, preferences for environment representation, pleasure in exploring places or preference for moving in known places, and spatial anxiety).

The data analysis was conducted in two steps. First, we defined the factorial model that better groups the visuospatial measures using a confirmatory factor analysis (CFA) approach. Based on the literature, two different models were tested. Referring to Hegarty et al. (2006), and Pazzaglia et al. (2018), we first tested a more parsimonious model, distinguishing between two visuospatial factors: abilities and inclinations. Then, as suggested by other studies (Meneghetti et al., 2014, 2020), we also tested a threefold model that distinguished between positive and negative factors of visuospatial inclinations as well. In a second step, we tested and estimated the effects of visuospatial factors on accuracy in learning from navigation tasks using multivariate generalized linear models. We used this approach because there were multiple dependent variables (route retracing, shortcut finding, landmark locating), and they were not necessarily normally distributed.

A Bayesian framework was used for all data modeling because it enables previous knowledge to be considered as quantitative information directly in the analysis. Such previous knowledge is formalized as “informed priors” (e.g., McElreath, 2016). In the Bayesian framework, “prior” and “posterior” distributions are key concepts. They are probability distributions. Prior information can come from previous literature and meta-analyses (as in our case), or even from plausible expectations. The “prior” thus indicates the available knowledge about the possible values of a given parameter (e.g., an effect size, a regression coefficient) before new data are observed. For example, an informed prior distribution may indicate that the value of a regression coefficient will most probably range between 0.10 and 0.40 (e.g., with 95% confidence), with a normal distribution of probability within that range (and therefore with the single most likely value in the middle, at 0.25). A prior can also be uninformed (as software often set by default), such as a flat uniform distribution spanning the domain of real numbers (this expresses a skeptical stance, and it can be used to prevent estimates from being affected by any previous knowledge). The “posterior” indicates the same probability distribution after it has changed in the light of new evidence (i.e., new data) being collected and analyzed, so it reflects both prior knowledge and new evidence. In a Bayesian framework, we can consider the discrepancies between prior and posterior as indicating the extent to which prior knowledge has been updated after considering new evidence. This is important for several reasons. First, it contributes to shifting the focus from statistical significance to quantitative reasoning about effect sizes, as advocated even in the American Psychological Association Publication Manual since its 6th edition (American Psychological Association, 2010). Second, it improves the precision of the values estimated because previous knowledge is added directly to the information yielded by newly-collected data. Third, it encourages the consideration of any discrepancies between old (prior) and new (posterior) evidence, which may alert us to underlying procedural or theoretical issues, or even to biases in the way prior evidence was gathered. Considering all the above aspects, we estimated the effects of individual difference measures (with the factorial model that grouped them best) on task accuracy (route retracing, shortcut finding and landmark locating) after learning from navigation, considering and formalizing prior evidence in the literature (see McElreath (2016); Kruschke & Liddell (2018) for more information on Bayesian approach to data analysis).

Section snippets

Participants

The study involved 292 undergraduate students (262 females) ranging from 18 to 28 years old (M = 20.74, SD = 2.93). All participants were volunteers recruited by word of mouth. Their vision was normal or corrected to normal, and they reported no history of medical, neurological or psychiatric disorders. All participants were informed about the aims of the study, and gave their written informed consent as required by the university's research ethics committee. One female participant was excluded

Results

Bivariate correlations and descriptive statistics are provided in Table S1 (see Supplementary material, Part 1). It should be noted that 21% of participants were successful in the shortcut finding task (identifying the best shortcut in 38.33% of cases, while the other 61.67% include minor deviations); the other 79% of participants failed on this task. For the landmark locating task the mean was 1.56 (SD: 1.44).

Discussion

The present study aimed to examine the relationship between individual differences in visuospatial thinking (visuospatial abilities and wayfinding inclinations) and accuracy in recalling a path learned in navigation-like (desktop-based) condition, measured with several tasks (route retracing, shortcut finding, and landmark locating tasks). The relations between the measures of individual differences and the tasks measuring environment learning accuracy were examined using not only the data

Conclusion

To conclude, this study contributes to expanding our understanding of the individual differences in visuospatial measures that concur in optimizing environment learning from navigation (using a desktop VE). Adopting a Bayesian framework for the data analysis (a novel approach in the field of spatial cognition) enabled us to consider information from previous literature and obtain more robust estimates. Overall, our findings clearly show the utility of considering both visuospatial abilities and

Author statement

Chiara Meneghetti: Conceptualization; Data curation; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Writing – original draft; Writing – review & editing.

Laura Miola: Conceptualization; Data curation; Formal analysis; Software; Visualization; Writing – original draft; Writing – review & editing.

Enrico Toffalini: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Software; Validation; Visualization; Writing – original draft;

Funding

This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics

The study was approved by the Ethical Committee for Psychological Research at the University of Padova (No. 3428). All participants were informed about the purposes of the study and gave their written informed consent in accordance with the Declaration of Helsinki (General Assembly of the World Medical Association, 2014).

Declaration of competing interest

None.

Acknowledgments

The present work was conducted as part of the Dipartimenti di Eccellenza research program (art.1, commi 314–337 legge 232/2016), supported by a grant from MIUR to the Department of General Psychology, University of Padua.

Learning from navigation, and tasks assessing its accuracy: The role of visuospatial abilities and wayfinding inclinations (original) (raw)