GPS use negatively affects environmental learning through spatial transformation abilities

Abstract

Research has established that GPS use negatively affects environmental learning and navigation in laboratory studies. Furthermore, the ability to mentally rotate objects and imagine locations from other perspectives (both known as spatial transformations) is positively related to environmental learning. Using previously validated spatial transformation and environmental learning tasks, the current study assessed a theoretical model where long-term GPS use is associated with worse mental rotation and perspective-taking spatial transformation abilities, which then predicts decreased ability to learn novel environments. We expected this prediction to hold even after controlling for self-reported navigation ability, which is also associated with better spatial transformation and environmental learning capabilities. We found that mental rotation and perspective-taking ability fully account for the effect of GPS use on learning of a virtual environment. This relationship remained after controlling for existing navigation ability. Specifically, GPS use is negatively associated with perspective-taking indirectly through mental rotation; we propose that GPS use affects the transformation ability common to mental rotation and perspective-taking.

Full text

Running head: GPS USE AND SPATIAL ABILITIES 1
GPS use negatively affects environmental learning through spatial transformation abilities
Ian T. Ruginski1,2
Sarah H. Creem-Regehr1
Jeanine K. Stefanucci1
Elizabeth Cashdan2
1 University of Utah, Department of Psychology
2 University of Utah, Department of Anthropology
Acknowledgments. We thank Taylor Larsen, Miley Nguyen, and Jenn Isenhour for their
help collecting data. Thank you to the National Science Foundation, whose funding helped support
this work under grant number 1329091.
Declarations of interest: none.
Address correspondence to Ian T. Ruginski, 380 S. 1530 E., Room 502, Department of
Psychology, University of Utah, Salt Lake City, UT 84112. Email: Ian.Ruginski@gmail.com

GPS USE AND SPATIAL ABILITIES 2
ABSTRACT
Research has established that GPS use negatively affects environmental learning and
navigation in laboratory studies. Furthermore, the ability to mentally rotate objects and imagine
locations from other perspectives (both known as spatial transformations) is positively related to
environmental learning. Using previously validated spatial transformation and environmental
learning tasks, the current study assessed a theoretical model where long-term GPS use is
associated with worse mental rotation and perspective-taking spatial transformation abilities,
which then predicts decreased ability to learn novel environments. We expected this prediction to
hold even after controlling for self-reported navigation ability, which is also associated with better
spatial transformation and environmental learning capabilities. We found that mental rotation and
perspective-taking ability fully account for the effect of GPS use on learning of a virtual
environment. This relationship remained after controlling for existing navigation ability.
Specifically, GPS use is negatively associated with perspective-taking indirectly through mental
rotation; we propose that GPS use affects the transformation ability common to mental rotation
and perspective-taking.
Key Words: spatial cognition; navigation; GPS use; spatial abilities; geospatial technology;
environmental learning

GPS USE AND SPATIAL ABILITIES 3
1. Introduction
GPS use is an aspect of navigation experience that has recently become common in
industrialized societies, yet little is known about the relationship between lifetime variation in
GPS use and spatial abilities. Laboratory studies have shown that GPS use reduces navigation
efficiency and accuracy (Gardony, Brunyé, & Taylor, 2015; Ishikawa, Fujiwara, Imai, & Okabe,
2008). Similarly, GPS use impairs spatial memory for environments when used to aid navigation
(Hejtmánek, Oravcová, Motýl, Horáček, & Fajnerová, 2018; Münzer, Zimmer, Schwalm, Baus,
& Aslan, 2006; Parush, Ahuvia, & Erev, 2007; Willis, Holscher, Wilberz, & Li, 2009; c.f.
Sönmez & Önder, 2019). Much of this work argues that GPS use disrupts environmental learning
via attentional and/or working memory mechanisms. For example, navigators who use GPS may
pay more attention to the device than virtual environments (Hejtmánek et al., 2018), have
difficulty learning due to divided attention (Gardony et al., 2015), or do not encode environments
into spatial working memory (Münzer et al., 2006). This research has established that GPS use
negatively affects environmental learning using many different methodologies, such as eye-
tracking (e.g. Hejtmánek et al., 2018) and comparison of distracting auditory and visual cues
(e.g. Münzer et al., 2006) in both real and virtual environments. However, the cognitive
mechanisms by which everyday GPS use (as opposed to most of the previous work that
manipulated GPS use in a lab setting) adversely affects environmental learning are still unclear.
To examine this question, we measured GPS use, visual environmental learning (i.e., navigation
without body-based cues) in a virtual landscape, and two spatial abilities that might mediate this
relationship: mental rotation and perspective-taking.
Mental rotation (Vandenberg & Kuse, 1978) and perspective-taking (Kozhevnikov &
Hegarty, 2001) are two widely studied cognitive processes that involve different types of

GPS USE AND SPATIAL ABILITIES 4
imagined spatial transformations. Perspective-taking can be defined as one’s ability to imagine
oneself in another position in the environment (involving an egocentric spatial transformation),
while mental rotation can be defined as one’s ability to imagine an object in another position in
the environment, thereby involving an object-based spatial transformation (Newcombe &
Shipley, 2015). These two spatial abilities are of relevance to navigation for a few reasons. First,
these abilities are expected to play a role in environmental learning, as environmental learning
involves continual updating of one’s location in the environment and the spatial relationship of
oneself to landmarks in the environment, as well as the spatial relationship between landmarks in
the environment (Wolbers & Hegarty, 2010, Wolbers & Wiener, 2014). Similarly, tests of spatial
memory for learned environments often require imagining oneself in locations previously visited
from a different perspective than is currently viewable in the environment (egocentric
transformation), or from an object-centered viewpoint such as marking locations on a map.
Previous work has shown that both mental rotation and perspective-taking abilities correlate with
measures of environmental learning, such as pointing to previously visited landmarks (Fields &
Shelton, 2006; Kozhevnikov, Motes, Rasch, & Blajenkova, 2006; Muffato, Toffalini,
Meneghetti, Carbone, & De Beni, 2017). Mental rotation also predicts real-world orienteering
ability (Malinowski, 2001; Silverman et al., 2000), as well as higher levels of geographic
knowledge for the locally traversed environment (Dabbs, Chang, Strong, & Milun, 1998).
Further, these spatial transformation abilities are strongly predictive of environmental
learning, even after controlling for self-reported sense of direction and verbal intelligence
(Hegarty, Montello, Richardson, Ishikawa, & Lovelace, 2006). In a meta-analysis, seven studies
with a total of 662 participants reported an unweighted average correlation of .26 between small-
scale spatial abilities (such as perspective-taking and the embedded figures test) and spatial

GPS USE AND SPATIAL ABILITIES 5
knowledge acquisition in virtual environments (Hegarty & Waller, 2005). However, there is also
contradictory evidence as to the role and relative importance of mental rotation and perspective-
taking in environmental learning during navigation. Kozhevnikov et al. (2006) found that
perspective-taking uniquely predicted environmental learning above and beyond mental rotation
ability, even though the two transformation abilities were highly correlated. The self-to-object
spatial updating processes unique to perspective-taking were considered essential for encoding
environments effectively. Others have tested mediation models, arguing that the imagined spatial
transformation processes in mental rotation support navigation indirectly through better imagined
perspective-taking skills. Results have been mixed; Hegarty et al. (2006) found that perspective-
taking does not mediate the effect of spatial abilities like mental rotation on environmental
learning, whereas Allen, Kirasic, Dobson, Long, and Beck (1996) found that perspective-taking
mediates the effect of mental rotation on environmental learning. Including both of these spatial
transformation tasks will allow us to address these previously conflicting results (see Figure 1).
The primary goal of the current study is twofold. First, we sought to answer the question
of whether and how everyday GPS use affects different types of spatial transformation and
environmental learning abilities, even after accounting for existing navigation ability. Second,
we sought to clarify the relations between spatial transformation and environmental learning
abilities. To address these questions, we asked individuals to complete two types of imagined
transformation tasks and a virtual environmental learning task. GPS use and navigation ability
(Hegarty, Richardson, Montello, Lovelace, & Subbiah, 2002) were each assessed by self-report.
While using GPS may have the immediate benefits of accurately arriving at a given
location, we suggest that those who navigate often with GPS perform imagined transformations in

GPS USE AND SPATIAL ABILITIES 6
Figure 1. Conceptual model demonstrating the theory driving the research questions of the
current study. Positive (+) and negative (-) signs above each path coefficient indicate the
predicted direction of association for each variable pair relationship. GPS use is expected to be
associated with lesser spatial transformation skills. It is expected that diminished transformation
skills will then be associated with an impaired ability to learn newly encountered virtual
environments (environmental learning). The dashed line indicates a direct path that we expect to
be accounted for by the spatial transformation abilities mediators. Navigation ability is
considered as a potential confounding factor to the effects of GPS usage on spatial abilities. It is
unknown whether the relationship between mental rotation and environmental learning is
mediated by perspective-taking (grey path), which is assessed in the current study by comparing
competing models.
navigation to a lesser extent, negatively affecting their long-term spatial transformation abilities
(see Figure 1). In contrast, those who navigate without GPS spend more time honing these spatial
skills and benefit from the “desirable difficulties” (Bjork, 1994) of navigating on their own. Based
on previous work (e.g., Hegarty et al., 2006; Muffato et al., 2017), we expected that the increased
spatial transformation skills across participants would translate to better ability to learn and
remember newly encountered environments, as measured by higher performance on validated
virtual environmental learning tasks in the laboratory (e.g., Weisberg, Schinazi, Newcombe,

GPS USE AND SPATIAL ABILITIES 7
Shipley, & Epstein, 2014). Importantly, the study evaluated the hypothesis that GPS use is
associated with virtual environmental learning through mental rotation and perspective-taking
even after controlling for navigation ability, which is known to influence spatial abilities and
navigation (Hegarty et al., 2006; Pazzaglia & Taylor, 2007; Weisberg et al., 2014).
To our knowledge, only one study has assessed the effects of navigation ability, mental
rotation ability, and GPS use in everyday life on the ability to learn environments. Ishikawa (2018)
found that navigation ability, mental rotation ability, and experience with GPS contributed
independently to wayfinding in a real environment. We view our work as complementary to
Ishikawa (2018), with three important distinctions. First, individuals were not assisted with tools
for wayfinding (maps or GPS) during the experiment as in Ishikawa’s (2018) study. Second,
individuals learned a virtual environment, rather than a real world environment. Third, we included
an additional measure of egocentric spatial transformation ability (perspective-taking,
Kozhevnikov & Hegarty, 2001), which is known to be predictive of environmental learning
(Kozhevnikov et al., 2006), but dissociable from mental rotation-based spatial transformations
(Zacks, Mires, Tversky, & Hazeltine, 2000; Hegarty et al., 2006). To better understand which
spatial transformation processes are affected by GPS use, our study also sought to evaluate whether
GPS use affects perspective taking only, mental rotation only, or both skills simultaneously.
Overall, this work addresses whether the growing use of GPS technologies could affect spatial
cognitive abilities above and beyond existing navigation skill, potentially laying the groundwork
for future work testing interventions that could affect spatial abilities throughout the lifespan.

GPS USE AND SPATIAL ABILITIES 8
2. Material and Methods
2.1 Participants & Sample Size
Participants were 201 students attending the University of Utah aged 18 to 47, with a
mean age of 21.3 (SD = 4.5). 113 identified as female, 86 as male, and 2 as androgynous. One
participant did not follow directions for the perspective-taking task. This participant’s
perspective taking data were considered missing in analyses (see section 3.1.1 for more on
handling of missing data). A sample size of 201 was used to approach Kline’s (2015)
recommended minimum 10:1 parameter:sample size ratio for adequate power, as there will be 21
free parameters in the most complex model evaluated. In addition, the sample size was
sufficiently powered (greater than .8, which occurs at a sample size of 148 when using bias-
corrected bootstrapping) to detect a modest (mean τ′ = .067) indirect effect (Fritz & MacKinnon,
2007).
2.2 Experimental Design and Measures
2.2.1 Virtual SILCton
The virtual SILCton task was adopted from Weisberg et al. (2014) to assess
environmental learning. Using a keyboard and mouse to navigate, participants first learned two
routes through a virtual college campus environment (see Figure 2).
Each route featured four landmarks, which were college campus buildings located at
Temple University (e.g., Batty House, an admissions building). Participants followed red arrows
indicating a route and were not allowed to diverge from the paved roads in the environment.
Floating diamonds indicated that an important landmark was nearby and each of these landmarks
was labeled with a clearly visible sign. After learning routes with landmarks, participants learned
two routes that connected the landmarked routes. Then participants completed pointing, distance

GPS USE AND SPATIAL ABILITIES 9
estimation, and cognitive map tasks, which were used as dependent measures to compose the latent
environmental learning factor (Hegarty et al., 2006).
Figure 2. Overhead view of virtual SILCton environment. Two routes learned in separate parts of
the environment are depicted, each featuring four landmarks (A1 to A2 and B1 to B2; solid lines).
After learning routes featuring landmarks, participants learned two routes connecting the
previously explored portions of the environment (C1 to C2 and D1 to D2; dashed lines).
Participants never saw this view of the environment. However, for the cognitive map-building
task, participants placed landmarks on a blank canvas (no roads featured) from this perspective.
Figure adopted from Weisberg et al. (2014).
The pointing task was completed from an egocentric perspective in front of each landmark
in the environment. Participants controlled a crosshair appearing in the center of the screen using
the mouse and clicked to record their response. 28 trials involved pointing from a landmark to
another landmark that was encountered on the other route traversed in the environment and were
averaged to create the between-route pointing variable (angular degrees of error). 28 trials involved

GPS USE AND SPATIAL ABILITIES 10
pointing from a landmark to another landmark that was encountered on the same route and were
averaged to create the within-route pointing variable (angular degrees of error).
For the distance estimation task, individuals adjusted sliders for the distances between each
landmark in the environment. Individuals were instructed that “the top slider bar shows the true
distance between the two buildings specified. Drag the remaining six slider bars to indicate the
distance between the other pairs. The longest distance in each set (page) will fill the entire slider.”
These instructions provided a relative “measuring stick” to complete distance estimations. In total,
participants completed 48 trials (eight landmarks with six estimations per landmark). All trials
were averaged prior to analyses and are in the metric of “Unity meters”, which approximate one
real-world meter.
For the cognitive map task, individuals were shown a blank rectangle “map” and provided
with 2-D representations of each landmark encountered in the environment from a birds-eye view
perspective. If participants hovered their mouse over the 2-D representation of a landmark, they
were provided with an image of that landmark as it was seen in the environment (from an
egocentric perspective). Participants placed all eight landmarks on the map for a total of eight
trials. Variance captured by individuals’ cognitive maps (r2) was then calculated using bi-
dimensional regression (Friedman & Kohler, 2003). Participants were allowed as much time as
needed to complete the pointing, distance estimation, and cognitive map tasks.
2.2.2 Vandenberg & Kuse Mental Rotation Task
We utilized a computerized mental rotation task developed by Weisberg et al. (2014).
Individuals had 3 minutes to complete each 10-trial section of the test. For a single trial,
individuals were asked to select two of four images that matched a target image by imagining the
rotation of the blocks. Scoring used a psychometric framework, such that individuals were

GPS USE AND SPATIAL ABILITIES 11
rewarded for true positives and true negatives (+2), but penalized for false positives and false
negatives (-2). Non-answers were scored a 0. Because two answers were correct on each trial and
there were 20 total trials, the maximum possible score was +80 and the minimum was -80.
2.2.3 Kozhevnikov & Hegarty Perspective-taking Task
To measure perspective-taking, we utilized a paper and pencil version of the spatial
orientation task (SOT) as detailed in Kozhevnikov and Hegarty (2001). The test consists of 12
items where individuals are asked to perform imagined pointing to different objects on a two-
dimensional map-like array of objects. Individuals were allotted five minutes to answer all 12
questions, and all tests were scored by hand using a protractor. If individuals did not have
sufficient time to answer a question, a score of 90 degrees of error was assigned for that
question, consistent with chance performance. All errors that exceeded 180 degrees were
subtracted from 360 prior to analysis to determine the smallest deviation possible, as the angular
deviation between two angles cannot exceed 180 degrees.
2.2.4 GPS & Survey Questions
The GPS question asked, “About how often do you use a GPS for navigation when
traveling?” Individuals responded on a 5-point Likert scale with one indicating never, two
indicating rarely, three indicating sometimes, four indicating often, and five indicating always.
Individuals were also asked to answer questions about mobility, or travel frequency. One
item assessing daily travel asked, “On days when you work or go to school, about how many
different places do you go to, on a typical day? (do not count as "different" other buildings or
rooms within the same workplace, school, or marketplace).” This question was scaled from one
to seven, with each Likert scale answer corresponding to zero places, one place, two places, three

GPS USE AND SPATIAL ABILITIES 12
places, four places, five to six places, and seven or more places, respectively. Another item
assessing monthly travel frequency asked, “In the month when you traveled the most, about how
many different cities, towns, or villages did you spend the night in?” A final question assessing
number of places visited in Utah (adopted from Padilla et al., 2017) was asked to measure
longer-term, lifetime mobility. These questions were then coded into total distance traveled to
places visited in Utah (calculated as the distance from the center of the University of Utah’s
campus to each place visited). The places visited in Utah question was only completed by a
subsample (n = 154) of the study due to procedural errors.
The Santa Barbara Sense of Direction questionnaire (SBSOD; Hegarty et al., 2002) was
also given, and data on basic demographic variables were collected.
2.3 Materials
The experiment was run on a Dell computer with four Intel Core i7-4770 3.40 GHz processors,
16 GB of RAM, and Windows 7 as its operating system. The computer was connected to a
1920x1200 resolution, 24-inch Dell LCD monitor. The display was updated at a frame rate of 60
Hz.
2.4 Procedure
Prior to conducting research, the University of Utah Institutional Review Board (IRB)
reviewed and approved the current study as adhering to ethical guidelines. Upon arriving to the
study, participants provided informed consent via an IRB-approved consent form. Participants
first completed the Weisberg et al. (2014) virtual SILCton learning and memory task
(approximately 25-45 minutes, described in section 2.2.1). Participants then completed a digital

GPS USE AND SPATIAL ABILITIES 13
version of the Vandenberg and Kuse (1978) mental rotation task, the Santa Barbara Spatial
Orientation perspective-taking task (Kozhevnikov & Hegarty, 2001), and a self-report survey
(described in section 2.2.4), in that order.
3. Results
3.1 Data Analysis
Prior to analyses, linear regressions were conducted to determine multicollinearity
between the perspective-taking and mental rotation spatial transformation variables for the
environmental learning outcomes. Although correlation was high between mental rotation and
perspective-taking (r = -.56, p < .001), the variance inflation factor between mental rotation and
perspective-taking did not exceed 5 (VIF = 1.39), and thus was considered non-problematic for
further analyses (Cohen, Cohen, West, & Aiken, 2013).
Because the GPS use variable was ordinal, robust maximum likelihood (MLR) was used
for structural equation model estimation given the appropriateness of MLR for ordinal data with
five or more categories (Rhemtulla, Brosseau-Liard, & Savalei, 2012). All indirect effects
reported, however, require bootstrapping to account for non-normality of indirect effect
distributions, which uses maximum likelihood estimation (Preacher & Hayes, 2008). Missing
data was relatively sparse
1
, and one individual’s perspective-taking data was not scored due to a
1
Multiple steps were taken to assess and handle missing data. First, coverage (proportion of univariate pairwise
missingness; Newsom, 2015) was inspected, and we found that data was mostly non-missing, with many coverage
levels between variables around .99 and the lowest coverage .955. We attempted to assess correlations between
observed data and missingness, but did not find any significant associations between observed variables (mental
rotation, perspective-taking, environmental learning measures) and missingness. This was likely due to a lack of
variability, as there were mostly one to two missing values per variable, with seven at most. Due to this, we made
the assumption that the data were missing at random (without statistical evidence of associations between observed
values and missingness), and given limited missing data, full information maximum likelihood (FIML) was
implemented to handle missing data in all structural equation models that did not implement bootstrapping
(Newsom, 2015). Bootstrapped models utilized list-wise deletion due to limitations of the lavaan version 0.6-3

GPS USE AND SPATIAL ABILITIES 14
failure to follow instructions. Prior to analyses, all error score variables (pointing angular error,
distance estimation error, perspective-taking angular error) were reverse coded to make higher
values correspond to better ability in our models.
3.2 Preliminary Analyses & Descriptive Statistics
Prior to running structural equation models, we determined if mediation was viable and
if GPS use uniquely predicted spatial abilities scores above and beyond the daily travel, monthly
travel, and distance to places visited in Utah questions (see Tables 1 and 2 for complete zero-
order correlations and descriptive statistics, respectively). Controlling for all travel frequency
measures, preliminary analyses revealed a direct effect of GPS use on environmental learning
ability (B = -0.22, β = -0.19, SE = 0.09, p = 0.01), mental rotation ability (B = -7.32, β = -0.30,
SE = 1.85, p < .001), and perspective-taking (B = -6.01, β = -0.24, SE = 1.86, p < .001). Further,
neither daily travel (B = 0.02, β = 0.02, SE = 0.05, p = .78), monthly travel (B = -0.001, β = -
0.001, SE = 0.04, p = .98), or total distance to places visited in Utah (B = -0.001, β = -0.05, SE =
0.0001, p = 0.52) questions predicted GPS use. These results suggest that GPS use was not
strongly influenced by travel frequency, and that GPS use was uniquely important to spatial
transformation abilities and environmental learning regardless of travel frequency.
program (Oberski, 2014), but did not appear to significantly deviate from the models that implemented FIML, likely
due to limited, nonsystematic missingness.

GPS USE AND SPATIAL ABILITIES 15
Table 1
Univariate correlations between study measures
Note. All variables have been transformed so that higher scores indicate more ability, use, or travel. *
indicates p < .05. ** indicates p < .01.
Table 2
Descriptive statistics for all study measures
1
2
3
4
5
6
7
8
9
10
1. Navigation Ability
–
2. GPS Use
-.36**
–
3. Mental Rotation
.21**
-.32**
–
4. Perspective Taking
.27**
-.26**
.53**
–
5. Within Pointing
.34**
-.20**
.23**
.26**
–
6. Between Pointing
.25**
-.08
.12
.14*
.55**
–
7. Distance
.26**
-.21**
.24**
.29**
.58**
.47**
–
8. Cog Map r2
.20*
-.16*
.20*
.23**
.56**
.50**
.59**
–
9. Daily Travel
.10
-.08
-.03
-.04
-.14*
-.03
-.01
-.10
–
10. Monthly Travel
.15*
.00
.02
.08
.03
.02
-.03
-.04
.05
–
11. Utah Distance
.12
- .05
.14
.05
.07
-.03
-.01
-.05
.09
.17*
Measure
Mean
SD
Skewness
Kurtosis
Navigation Ability
3.8
1.0
-0.3
2.8
GPS Use
3.4
0.9
-0.2
2.7
Mental Rotation
30.5
21.8
0.03
2.6
Perspective Taking
-35.0
23.0
0.9
2.9
Within Pointing
-27.8
12.2
0.4
2.8
Between Pointing
-45.9
12.8
-0.6
2.7
Distance
-134.4
52.5
1.3
5.6
Cog Map r2
0.49
0.26
0.1
1.9
Daily Travel
3.5
1.5
0.4
2.8
Monthly Travel
3.5
1.7
1.3
4.2
Utah Distance
1576
956
0.3
2.2

GPS USE AND SPATIAL ABILITIES 16
3.3 Theoretical Models of GPS use and Spatial Abilities
3.3.1 Environmental Learning Latent Factor Measurement Model
A structural equation model was first fit to the measurement portion of the model that
was expected based on past research: a latent visual environmental learning factor (Hegarty et
al., 2006). This factor was composed of four measures, including between- and within-route
pointing errors (Weisberg et al., 2014), a cognitive mapping measure (r2 measured by bi-
dimensional regression; Friedman & Kohler, 2003), and a distance estimation measure.
Maximum likelihood estimation was used for this model as these outcomes were roughly
normally distributed. As measures varied in terms of scale, a factor variance identification
approach was used to identify the model, where the variance of the latent visual environmental
learning factor was fixed to 1 (Newsom, 2015).
Model fit was relatively good across multiple indices (RMSEA = 0.058, CFI = 0.995, TLI
= 0.985, SRMR = 0.016, χ2 = 3.34(2), p = 0.19). Between-route pointing error (B = 8.83, β = 0.69,
SE = 0.87, p < .001), within-route pointing error (B = 9.24, β = 0.76, SE = 0.81, p < .001), r2
variance captured by cognitive maps (B = -0.21, β = -0.78, SE = 0.02, p < .001), and distance
estimation error (B = 37.17, β = 0.71, SE = 3.54, p < .001) all loaded onto the latent factor. Given
that all observed variables loaded onto the latent environmental learning factor and model fit was
good, this factor was retained as an outcome for the full theoretical models tested in sections
3.3.2 and 3.3.3.
3.3.2 Mediation Model with no Relationship Between Mental Rotation and Perspective-taking
We found that model fit was below adequate according to multiple indices (Robust
RMSEA = 0.133, Robust CFI = .875, robust TLI = 0.767, SRMR = 0.074, χ2 = 72.03 (15), p <

GPS USE AND SPATIAL ABILITIES 17
.001). Given good fit from the measurement portion of the model reported in section 3.3.1, poor
fit suggests that the majority of misfit originates from the structural (regression) portion of this
model. Results of the model are presented in Figure 3.
Figure 3. Model 1, in which no relationship was allowed to exist between mental rotation
and perspective-taking (this regression path was constrained to zero during model estimation).
Standardized betas are presented for each regression path. **indicates p < .01, * indicates p <
.05. Significant paths (p < .05) are bolded for emphasis.
3.3.3 Alternative Mediation Model with a Direct Relationship Between Mental Rotation and
Perspective-taking
The previous model was compared with a model that allowed mental rotation to predict
perspective-taking. The following results are for the model estimated using robust maximum
likelihood estimation (see Figure 4). Model fit was good overall according to multiple indices
(Robust RMSEA = 0.038, Robust CFI = 0.991, Robust TLI = 0.981, SRMR = 0.032, χ2 =
18.20(14), p = 0.19).

GPS USE AND SPATIAL ABILITIES 18
Figure 4. Model 2, in which mental rotation ability was allowed to predict perspective-
taking ability. All other paths in Model 1 were also included in this model. **indicates p < .01, *
indicates p < .05. Significant paths (p < .05) are bolded for emphasis.
We compared the model reported in section 3.3.2 and the current model using a Satorra-
Bentler adjusted chi-square difference test for models estimated using robust maximum
likelihood estimation (Satorra & Bentler, 2001). We found that the more saturated model
estimating a regression path between mental rotation and perspective-taking fit the data better
than the less saturated model with this parameter fixed to 0 (χ2 difference = 61.96(1), p < .001).
This suggests that the more saturated model where perspective-taking mediates the direct effect
of mental rotation on environmental learning more accurately captures the relationship between
GPS use, spatial transformation skills, and environmental learning ability.
To evaluate evidence of mediation, indirect effects were computed using bootstrapped
bias corrected 95% confidence intervals in R using lavaan version 0.6-3. This method was
selected to account for non-normality and subsequent bias in bootstrapped indirect effects (see
MacKinnon, Lockwood, & Williams, 2004). In addition, we included the potential confounding
variable of navigation ability as measured by SBSOD in our structural equation model.
Accounting for this confound fulfills assumptions in mediation analysis and reduces bias in

GPS USE AND SPATIAL ABILITIES 19
estimated direct and indirect effects (see VanderWeele, 2016, for more on assumptions of
mediation). In total, five indirect effects were bootstrapped (see Table 3).
Table 3
Estimates of bootstrapped indirect effects testing for mediation
Indirect Effect
B
β
SE
Lower CI
Upper CI
GPS → MRT → EL
-0.03
-0.03
0.03
-0.12
0.02
GPS → PT → EL
-0.02
-0.02
0.02
-0.09
0.01
GPS → MRT → PT
-2.91
-0.12
0.99
-4.99
-1.08
MRT → PT → EL
0.01
0.09
0.002
0.001
0.010
GPS → MRT → PT → EL
-0.03
-0.02
0.02
-0.08
-0.01
Note. Bootstrapped bias-corrected 95% confidence intervals were used to determine if an indirect effect
was present. If the 95% confidence interval does not include zero, this suggests evidence of an indirect
effect. If the indirect effect column is bolded, the direct effect of the first predictor was completely
accounted for, consistent with full mediation. Indirect effects with plain text indicate no evidence of
mediation (abbreviations: PT, perspective-taking; MRT, mental rotation; EL, environmental learning). B
refers to unstandardized effect coefficients, while β refers to standardized effect coefficients.
The first effect tested whether mental rotation ability alone mediated the direct effect of
GPS use on environmental learning ability. The second tested whether perspective-taking alone
mediated the direct effect of GPS use on environmental learning. We found that neither
perspective-taking nor mental mediated the direct effect of GPS use on environmental learning
ability alone.
However, we found evidence of two-mediator mediation, where mental rotation and
perspective-taking spatial transformation abilities together mediate the effect of GPS use on
environmental learning ability, even after controlling for the confound of navigation ability. This
two-mediator mediation was consistent with two single-mediator indirect effects also present in
the model. Mental rotation ability fully mediated the direct effect of GPS use on perspective-

GPS USE AND SPATIAL ABILITIES 20
taking, and perspective-taking fully mediated the effect of mental rotation on environmental
learning.
Second, in order to further determine how GPS use and navigation ability differed in their
effects on spatial transformation and environmental learning abilities, we explicitly compared
effects in the model using bootstrapped difference tests (see Table 4). Navigation ability
accounted for more variance in perspective-taking ability than GPS use did as assessed both
indirectly through mental rotation ability and directly. Further, navigation ability was more
strongly associated with environmental learning than GPS use was, as assessed indirectly
through mental rotation and perspective taking ability. However, GPS use contributed more to
mental rotation ability than navigation ability did.
Table 4
Differences between pathways tested using bootstrapping
Note. The effects of navigation ability and GPS use columns explicitly show the paths being compared. A
bolded column indicates that that effect on the named outcome was greater than the opposing column.
Both direct and indirect effects were compared (abbreviations: PT, perspective-taking; MRT, mental
rotation; EL, environmental learning). B refers to unstandardized effect coefficients, while β refers to
standardized effect coefficients.
Effects of
navigation ability
Effects of GPS use
B
β
SE
p-value
SBSOD → PT
GPS → MRT → PT
6.83
0.29
1.61
< .001
SBSOD → PT
GPS → PT
5.57
0.24
1.99
0.005
SBSOD → MRT
GPS → MRT
-8.76
-0.29
1.89
< .001
SBSOD → EL
GPS → MRT → PT → EL
0.39
0.34
0.1
< .001

GPS USE AND SPATIAL ABILITIES 21
3.4 Gender Differences
Though not a focus of our study, we also tested for gender differences across study
measures in the case that these would be of interest to other spatial abilities researchers
considering past gender differences in spatial abilities (e.g. Voyer, Voyer, & Bryden, 1995).
Results are reported in Table 5, along with a measure of effect size (Cohen’s d, Cohen, 1988).
We observed gender differences showing a male advantage in navigation ability, mental rotation,
perspective-taking, within-route pointing, and distance estimation, with the largest gender
difference being in mental rotation (p < .001, d = 0.79). Men also reported using GPS devices
less than women (p = .004, d = 0.39).
Table 5
Gender differences in study measures
Note. All tests were independent samples t-tests (two-tailed) assuming unequal variances amongst groups.
** indicates p < .001. * indicates p < .01. All variables have been transformed so that higher scores
indicate more ability, use, or travel. All differences are women’s mean minus men’s mean. Two
individuals self-identifying as androgynous were not included in these analyses.
Variable
Men (n = 86)
Women (n = 113)
Mean
Difference
Mean Difference
95% CI
Cohen’s d
Mean SD
Mean SD
Navigation Ability
4.1 0.9
3.6 1.0
-0.54**
-0.81, -0.27
0.11, 0.63
-21.9, -10.2
-8.9, -20.9
-2.1, -8.6
-5.1, 2.3
-39.9, -12.1
-0.13, 0.03
-0.55, 0.30
-0.27, 0.68
-525.1, 99.2
0.56
GPS Use
3.2 0.9
3.5 0.9
0.36*
0.39
Mental Rotation
39.5 21.4
23.5 19.6
-16.0**
0.79
Perspective Taking
-26.4 17.9
-41.4 21.4
-14.9**
0.68
Within Pointing
-24.9 10.4
-30.3 13.0
-5.4**
0.45
Between Pointing
-45.3 13.2
-46.7 12.6
-1.4
0.11
Distance
-120.0 42.4
-146.0 57.0
-26.0**
0.22
Cog Map r2
0.52 0.26
0.47 0.27
-0.05
0.19
Daily Travel
3.6 1.4
3.4 1.6
-.13
0.09
Monthly Travel
3.5 1.7
3.3 1.7
.21
0.12
Utah Distance
1697 1029
1484 902
-212.9
0.22

GPS USE AND SPATIAL ABILITIES 22
4. Discussion
Broadly, the current study sought to determine whether and how an emerging and
pervasive navigational aid — GPS — affects spatial abilities. Prior work has established that
GPS use in the laboratory adversely affects environmental learning and wayfinding outcomes
(Hejtmánek et al., 2018; Gardony et al., 2015; Ishikawa et al., 2008). However, it was unclear
whether long-term GPS use in everyday life would affect virtual environmental learning in the
absence of a direct manipulation. We hypothesized that long-term GPS use would indirectly and
negatively affect environmental learning ability through decreased object-based (mental rotation)
and egocentric (perspective-taking) spatial transformation abilities, which are known to be
associated with environmental learning outcomes (Fields & Shelton, 2006; Kozhevnikov et al.,
2006; Muffato et al., 2017).
We found evidence of two-mediator mediation consistent with our hypotheses, where
mental rotation and perspective-taking fully mediated the effect of GPS use on environmental
learning of a virtual environment. Critically, this relationship remained even while controlling
for self-reported navigation ability, suggesting that the effect of GPS use on environmental
learning cannot be fully explained by navigation ability (consistent with Ishikawa, 2018).
Overall, we view the primary contribution of the work as establishing that GPS use in everyday
life affects environmental learning indirectly via differences in spatial transformation processes
such as mental rotation and perspective-taking. Our results can only be generalized to visual
environmental learning without the proprioceptive and vestibular cues provided through walking
and head movements (Chrastil & Warren, 2012), though visual environmental learning is
substantially related to real-world environmental learning (Hegarty et al., 2006).

GPS USE AND SPATIAL ABILITIES 23
One important limitation of our study is that GPS use was assessed using a single, self-
reported question. Although there are likely advantages to using a finer grained measure of GPS
use (such as a measure that accounts for hours of use, e.g. Ishikawa, 2019), we did our best to
account for potential confounding factors. In preliminary analyses (section 3.2), the effect of
GPS use on spatial transformation and environmental learning outcomes remained regardless of
individuals’ travel frequency, suggesting that GPS use is uniquely associated with spatial
outcomes. Nonetheless, the GPS use question may indirectly measure another construct, such as
spatial ability or confidence. We anticipated this, and found that the effect of GPS use remained
even after accounting for self-reported sense of direction across all study variables. Despite
taking these steps, we still view the use of a single, self-reported GPS use measure as a
limitation. Future work would benefit from implementing a more thorough GPS use measure that
1) disentangles reason for GPS use and 2) implements multiple measurements of GPS use. In
addition, it is important to note that we cannot make causal claims as to the nature of the
relationship between GPS use, spatial transformation abilities, and environmental learning,
because GPS use was not manipulated over time. Note that a model which switches the ordering
of GPS use and spatial abilities (where mental rotation and perspective taking predict GPS-use,
which in turn predicts environmental learning ability) has the exact same fit indices as model 2
(Robust RMSEA = 0.038, Robust CFI = 0.991, Robust TLI = 0.981, SRMR = 0.032, χ2 =
18.20(14), p = 0.19). This alternative model can be viewed in OSF supplementary materials and
is demonstrative as to why we did not test models with different orderings, as well as why we
cannot claim causality in our study. It could be that more GPS use decreases spatial
transformation abilities or that decreases in spatial transformation abilities lead to increased GPS
use. GPS use may also decrease environmental learning indirectly through alternative paths not

GPS USE AND SPATIAL ABILITIES 24
measured in the current study, such as navigational style (Richter, Dara-Abrams, & Raubal,
2010) or human gaze behavior (Brügger, Richter, & Fabrikant, 2019). However, the current
results are an informative first step in broadly establishing a negative association between
everyday GPS use and environmental learning while also identifying spatial transformation
abilities as a potential mediating factor in this relationship.
In past work that manipulated GPS use in the laboratory, researchers have proposed a few
alternative explanations as to why GPS use negatively affects environmental learning. These
accounts most often claim that GPS use affects environmental learning through changes in
spatial attention and/or working memory. For example, navigators pay more attention to the GPS
device than the environment (Hejtmánek et al., 2018) or divide their attention between device
and environment (Gardony et al., 2015; Willis et al., 2009), impairing learning. Others have
argued that using GPS replaces the need for navigators to actively encode the environment into
spatial working memory (Münzer et al., 2006; Parush et al., 2007). We view our results as
complementary to these accounts, but also providing a different level of explanation. We suggest
that when individuals reduce their attention or reliance on environmental features during
navigation, the need for imagined spatial transformations also is reduced. Continuous use of GPS
may decrease the spatial transformation abilities that would have supported environmental
learning without the technology. Spatial transformation abilities may also share other spatial
cognitive processes that relate to environmental learning (e.g., spatial working memory; Muffato
et al., 2017), suggesting directions for future work that combine examination of effects of GPS
use, working memory, and spatial transformations on environmental learning.
Our research also builds on Hegarty et al.’s (2006) model of the positive relationship
between spatial transformation and environmental learning abilities, which is supported by

GPS USE AND SPATIAL ABILITIES 25
previous experimental work (Fields & Shelton, 2006; Kozhevnikov et al., 2006; Muffato et al.,
2017). Specifically, in showing that perspective-taking fully mediates the direct effect of mental
rotation-like spatial abilities on environmental learning, our result provides a conceptual
replication of Allen et al. (1996), but not Hegarty et al. (2006)
2
. This result suggests that
perspective-taking plays a critical mediating role between object-based spatial transformations
like mental rotation and acquisition of spatial knowledge from novel environments. Self-reported
navigation ability also did not predict mental rotation ability in our model, consistent with past
work (Hegarty et al., 2006, Hegarty et al., 2002). We also replicated previously observed gender
differences in spatial transformation abilities (Voyer, Voyer, & Bryden, 1995; Hegarty et al.,
2006) and environmental learning (Weisberg et al., 2014).
Furthermore, we found that individuals of lower navigation ability use GPS more often
than those of higher ability (β = -0.38). Navigation ability (β = 0.30) also contributed
significantly to environmental learning, as expected based on past work in both real (Burte &
Montello, 2017) and virtual (Pazzaglia & Taylor, 2007; Weisberg et al., 2014; Weisberg &
Newcombe, 2016) environments. Despite these two results, however, we still observed that GPS
use indirectly affects environmental learning above and beyond the strong effect of navigation
ability. So what is it about GPS use that contributes to decreased spatial transformation abilities
and, in turn, to worse environmental learning? Interestingly, we found that GPS use was more
negatively associated with mental rotation ability (β = -0.26) than with perspective-taking ability
2
The spatial abilities measured in the current study only conceptually, but not directly, test a replication of Hegarty
et al. (2006) and Allen et al.’s (1996) studies. Different constructs were utilized to test their mediation models. In
Allen et al.’s study, latency of perspective-taking was used as the mediator. A different spatial abilities latent factor
was also used as the focal predictor, comprised of surface development, cube comparison, map planning, hidden
figures, and Gestalt completion. In Hegarty et al.’s study, a latent spatial ability factor was used as the focal
predictor, which contained mental rotation as one observed variable, but also two other related spatial abilities:
embedded figures and the arrow span task.

GPS USE AND SPATIAL ABILITIES 26
(β = -0.09), and GPS use only affected perspective-taking indirectly through mental rotation (β =
-0.12). These differences in effects point to a diverging influence of GPS use dependent on type
(object-based vs. egocentric) of spatial transformation.
GPS use might relate to mental rotation more than perspective-taking due to the different
spatial processes involved in each type of ability. Both require imagined transformations, but
perspective-taking also requires spatial updating of one’s position relative to other spatial
reference points. Spatial updating is a process involved in perspective-taking that relates to path
integration and ability to effectively learn environments (Wolbers & Hegarty, 2010). It involves
maintenance of one’s position in relation to landmarks in the environment. Spatial updating
dissociates perspective-taking from mental rotation despite sharing an imagined transformation
process (Hegarty & Waller, 2004; Kozhevnikov et al., 2006). In our study, we found evidence
that mental rotation fully explains the effect of GPS use on perspective taking. This result
suggests that GPS use affects the common spatial transformation process underlying both mental
rotation and perspective-taking. One possible mechanism is that imagined spatial transformations
underlie the formation of allocentric (viewpoint independent) representations. If GPS use
eliminates the need for allocentric representations because of the explicit route-based
information that it provides, then this transformation process is less likely to be regularly used.
Our study focuses on the general effect of everyday GPS use on spatial transformation
and subsequent environmental learning abilities. A useful direction of future work could evaluate
which aspects of GPS use contribute to degraded spatial outcomes, as GPS devices can be used
in many different ways to support navigation. For instance, the GPS can be used as an allocentric
map (without turn-by-turn directions) versus an egocentric directional guide, or for other
purposes such as to track traffic trends for route selection. Schwering, Krukar, Li, Anacta, and

GPS USE AND SPATIAL ABILITIES 27
Fuest (2017) have suggested that GPS devices negatively contribute to environmental learning
ability due to their tendency to provide route-specific information. They propose that GPS
devices would better support environmental learning by providing information that aids
orientation to global space and landmarks, rather than turn-by-turn directions requiring little
spatial attention (see also Brunyé, Gardony, Holmes, & Taylor, 2018; Münzer, Fehringer, &
Kühl, 2016). Changing the information provided by GPS devices would help answer the question
of whether providing different spatial information decreases the likelihood of negative
associations of GPS use with environmental learning and transformation abilities. When
globally-oriented, individuals may be more likely to actively incorporate spatial transformation
processes during spatial learning. On the other hand, GPS devices may contribute to deficits in
spatial transformation and/or environmental learning abilities regardless of the information they
provide. Future work should test this possibility by manipulating the types of information GPS
devices provide over time, as well as frequency of device use.
In sum, our work suggests that GPS exerts its negative influence on spatial cognitive
abilities in the long-term, building on work that has shown its negative effects on environmental
learning in the short-term. Most noteworthy is that 1) GPS independently relates to spatial
transformation and environmental learning abilities even after accounting for the fact that
individuals of lower navigation ability use GPS devices more often, 2) GPS use indirectly relates
to environmental learning through decreased spatial transformation abilities, and 3) GPS use
relates to mental rotation, but not perspective-taking, and only affects perspective-taking
indirectly through mental rotation. The latter result suggests that GPS use is associated with the
common spatial transformation process underlying both mental rotation and perspective-taking.
The relationship between decreased mental transformation abilities and increased use of GPS

GPS USE AND SPATIAL ABILITIES 28
may be a consequence of reduced attention or encoding of one’s environment, consistent with
previously shown decrements in navigation tasks when GPS use has been manipulated.

Running head: GPS USE AND SPATIAL ABILITIES 29
References
Allen, G. L., Kirasic, K. C., Dobson, S. H., Long, R. G., & Beck, S. (1996). Predicting
environmental learning from spatial abilities: An indirect route. Intelligence, 22(3), 327-
355.
Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings.
In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing about knowing (pp.
185–205). Cambridge, MA: MIT Press.
Brügger, A., Richter, K. F., & Fabrikant, S. I. (2019). How does navigation system behavior
influence human behavior?. Cognitive research: principles and implications, 4(1), 5.
Brunyé, T. T., Gardony, A. L., Holmes, A., & Taylor, H. A. (2018). Spatial decision dynamics
during wayfinding: intersections prompt the decision-making process. Cognitive
Research: Principles and Implications, 3(1), 13.
Burte, H., & Montello, D. R. (2017). How sense-of-direction and learning intentionality relate to
spatial knowledge acquisition in the environment. Cognitive Research: Principles and
Implications, 2(1), 18.
Chrastil, E. R., & Warren, W. H. (2012). Active and passive contributions to spatial
learning. Psychonomic Bulletin & Review, 19(1), 1-23.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences.
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2013). Applied multiple
regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Lawrence
Erlbaum Associates, Inc.
Dabbs, J. M., Chang, E. L., Strong, R. A., & Milun, R. (1998). Spatial ability, navigation
strategy, and geographic knowledge among men and women. Evolution and
Human Behavior, 19(2), 89-98.
Fields, A. W., & Shelton, A. L. (2006). Individual skill differences and large-scale
environmental learning. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 32(3), 506.
Friedman, A., & Kohler, B. (2003). Bidimensional regression: assessing the configural similarity
and accuracy of cognitive maps and other two-dimensional data sets. Psychological
Methods, 8(4), 468.
Fritz, M. S., & MacKinnon, D. P. (2007). Required sample size to detect the mediated effect.
Psychological Science, 18(3), 233-239.

GPS USE AND SPATIAL ABILITIES 30
Gardony, A. L., Brunyé, T. T., & Taylor, H. A. (2015). Navigational aids and spatial memory
impairment: the role of divided attention. Spatial Cognition & Computation, 15(4), 246-
284.
Hegarty, M., Richardson, A. E., Montello, D. R., Lovelace, K., & Subbiah, I. (2002).
Development of a self-report measure of environmental spatial
ability. Intelligence, 30(5), 425-447.
Hegarty, M., & Waller, D. (2004). A dissociation between mental rotation and perspective-taking
spatial abilities. Intelligence, 32(2), 175-191.
Hegarty, M., & Waller, D. (2005). Individual differences in spatial abilities. In P. Shah,
A. Miyake (Eds.), The Cambridge Handbook of Visuospatial Thinking,
Cambridge, UK: Cambridge University Press, 121–169.
Hegarty, M., Montello, D. R., Richardson, A. E., Ishikawa, T., & Lovelace, K. (2006).
Spatial abilities at different scales: Individual differences in aptitude-test
performance and spatial-layout learning. Intelligence, 34(2), 151-176.
Hejtmánek, L., Oravcová, I., Motýl, J., Horáček, J., & Fajnerová, I. (2018). Spatial knowledge
impairment after GPS guided navigation: Eye-tracking study in a virtual
town. International Journal of Human-Computer Studies, 116, 15-24.
Ishikawa, T., Fujiwara, H., Imai, O., & Okabe, A. (2008). Wayfinding with a GPS-based
mobile navigation system: A comparison with maps and direct experience.
Journal of Environmental Psychology, 28(1), 74-82.
Ishikawa, T. (2018). Satellite Navigation and Geospatial Awareness: Long-Term Effects of
Using Navigation Tools on Wayfinding and Spatial Orientation. The Professional
Geographer, 1-13.
Kline, R. B. (2015). Principles and practice of structural equation modeling. New York,
NY: Guilford Press.
Kozhevnikov, M., & Hegarty, M. (2001). A dissociation between object manipulation
spatial ability and spatial orientation ability. Memory & Cognition, 29(5), 745-
756.
Kozhevnikov, M., Motes, M. A., Rasch, B., & Blajenkova, O. (2006). Perspective-taking
vs. mental rotation transformations and how they predict spatial navigation
performance. Applied Cognitive Psychology, 20(3), 397-417.
MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect
effect: Distribution of the product and resampling methods. Multivariate Behavioral
Research, 39(1), 99-128.

GPS USE AND SPATIAL ABILITIES 31
Malinowski, J. C. (2001). Mental rotation and real-world wayfinding. Perceptual and
Motor Skills, 92(1), 19-30.
Muffato, V., Toffalini, E., Meneghetti, C., Carbone, E., & De Beni, R. (2017). Individual
visuo-spatial factors and familiar environment knowledge: A structural equation
modeling analysis. Personality and Individual Differences, 113, 96-102.
Münzer, S., Zimmer, H. D., Schwalm, M., Baus, J., & Aslan, I. (2006). Computer-assisted
navigation and the acquisition of route and survey knowledge. Journal of Environmental
Psychology, 26(4), 300-308.
Münzer, S., Fehringer, B. C., & Kühl, T. (2016). Validation of a 3-factor structure of
spatial strategies and relations to possession and usage of navigational aids.
Journal of Environmental Psychology, 47, 66-78.
Newcombe, N. S., & Shipley, T. F. (2015). Thinking about spatial thinking: New
typology, new assessments. In Studying visual and spatial reasoning for design
creativity (pp. 179-192). Dordrecht, Germany: Springer Publishing.
Newsom, J. T. (2015). Longitudinal structural equation modeling: A comprehensive
introduction. New York, NY: Routledge.
Oberski, D. L. (2014). Lavaan. survey: An R package for complex survey analysis of
structural equation models. Journal of Statistical Software, 57(1), 1-27.
Parush, A., Ahuvia, S., & Erev, I. (2007, September). Degradation in spatial knowledge
acquisition when using automatic navigation systems. In International Conference on
Spatial Information Theory (pp. 238-254). Springer, Berlin, Heidelberg.
Pazzaglia, F., & Taylor, H. A. (2007). Perspective, instruction, and cognitive style in spatial
representation of a virtual environment. Spatial Cognition and Computation, 7(4), 349-
364.
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and
comparing indirect effects in multiple mediator models. Behavior Research Methods,
40(3), 879-891.
Rhemtulla, M., Brosseau-Liard, P. É., & Savalei, V. (2012). When can categorical variables be
treated as continuous? A comparison of robust continuous and categorical SEM
estimation methods under suboptimal conditions. Psychological methods, 17(3), 354.
Richter, K. F., Dara-Abrams, D., & Raubal, M. (2010, September). Navigating and learning with
location based services: A user-centric design. In Proceedings of the 7th International
Symposium on LBS and Telecartography (pp. 261-276).
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for

GPS USE AND SPATIAL ABILITIES 32
moment structure analysis. Psychometrika, 66(4), 507-514.
Schwering, A., Krukar, J., Li, R., Anacta, V. J., & Fuest, S. (2017). Wayfinding through
orientation. Spatial Cognition & Computation, 17(4), 273-303.
Silverman, I., Choi, J., Mackewn, A., Fisher, M., Moro, J., & Olshansky, E. (2000). Evolved
mechanisms underlying wayfinding: Further studies on the hunter-gatherer theory of
spatial sex differences. Evolution and Human Behavior, 21(3), 201-213.
Sönmez, B. E., & Önder, D. E. (2019). The influence of GPS-based navigation systems on
perception and image formation: A case study in urban environments. Cities, 86, 102-
112.
Vandenberg, S. G., & Kuse, A. R. (1978). Mental rotations, a group test of three-
dimensional spatial visualization. Perceptual and Motor Skills, 47(2), 599-604.
VanderWeele, T. J. (2016). Mediation analysis: a practitioner's guide. Annual Review of Public
Health, 37, 17-32.
Voyer, D., Voyer, S., & Bryden, M. P. (1995). Magnitude of sex differences in spatial abilities: a
meta-analysis and consideration of critical variables. Psychological bulletin, 117(2), 250.
Weisberg, S. M., Schinazi, V. R., Newcombe, N. S., Shipley, T. F., & Epstein, R. A.
(2014). Variations in cognitive maps: Understanding individual differences in
navigation. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 40(3), 669.
Weisberg, S. M., & Newcombe, N. S. (2016). How do (some) people make a cognitive
map? Routes, places, and working memory. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 42(5), 768.
Willis, K. S., Hölscher, C., Wilbertz, G., & Li, C. (2009). A comparison of spatial
knowledge acquisition with maps and mobile maps. Computers, Environment and
Urban Systems, 33(2), 100-110.
Wolbers, T., & Hegarty, M. (2010). What determines our navigational abilities?. Trends in
Cognitive Sciences, 14(3), 138-146.
Wolbers, T., & Wiener, J. M. (2014). Challenges for identifying the neural mechanisms
that support spatial navigation: the impact of spatial scale. Frontiers in Human
Neuroscience, 8, 571.
Zacks, J. M., Mires, J., Tversky, B., & Hazeltine, E. (2000). Mental spatial transformations of
objects and perspective. Spatial Cognition and Computation, 2(4), 315-332.