Cognitive representations of multilevel buildings: two- or three-dimensional?

Cognitive representations of multilevel buildings: two- or three-dimensional? | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Cognitive representations of multilevel buildings: two- or three-dimensional? OTMAR BOCK This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6350898/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 31 Aug, 2025 Read the published version in Experimental Brain Research → Version 1 posted You are reading this latest preprint version Abstract Previous research suggests that conventional multilevel buildings are cognitively represented as a stack of horizontal planes, whereas an atrium-shaped architecture which allows easy visual access to other levels is represented as a volume. The present study investigated whether easy physical access to other levels also promotes a volumetric representation. Participants were examined in a virtual 3D grid maze in which they could access higher and lower levels at each intersection . During a learning phase, they were transported through the maze across twelve intersection, each featuring a unique object. In the subsequent test phase, they were asked to indicate the location of these objects on a schematic drawing of the maze. Response accuracy in the test phase was similar when the twelve visited objects were arranged in a horizontal plane and when they were laid out in a volume. In accordance with earlier reasoning, this suggests that easy physical access to other levels indeed can facilitate a volumetric cognitive representation of multilevel architectures. Additional findings suggest that this representation emerged gradually rather than abruptly like a sudden insight, and that transport through the maze without left and right turns facilitated the internal representation, probably by reducing the associated cognitive load. cognitive map mental representation multistorey architecture spatial navigation spatial orientation human Figures Figure 1 Figure 2 Figure 3 Introduction To navigate landscapes, cities or buildings, we often rely on cognitive representations of the environment, often called “cognitive maps” (Tolman, 1948; Nadel, 2013). These maps enable us not only to follow familiar routes but also to detect shortcuts, plan detours around roadblocks, reach familiar destinations from a new starting point, and even devise entirely novel routes. Cognitive maps are not scaled replicas of the real world. Rather, behavioral and neurophysiological evidence suggests that these representations can be fragmented, distorted, biased by assumptions, organized within multiple reference frames, and sometimes lack metric information (McDonald & Pellegrino, 1993; Peer et al., 2021). Cognitive maps are formed gradually with practice (Kim & Bock, 2021; Zhang et al., 2014), at a rate that varies widely between individuals (Ishikawa & Montello, 2006), depending on factors such as emotional and personality traits (Pazzaglia et al., 2018), stress, sex and age ( Hegarty et al., 2022). Although the world around us is three-dimensional, cognitive maps are not necessarily three-dimensional as well. Just as flat road maps and GPS screens allow us to successfully navigate even hilly terrains, and information boards depicting a building as a stack of horizontal planes allow us to find our destination, two-dimensional cognitive maps could also be sufficient for successful wayfinding. Experimental evidence indeed suggests that multilevel buildings are cognitively represented as a stack of horizontal planes. When navigating such a building, participants typically first ascended to the correct level before proceeding horizontally toward their destination (Hölscher et al., 2006; Buechner et al., 2007); this strategy allowed them to reach their goal more quickly and with fewer errors than along other routes (Hölscher et al., 2006). Furthermore, participants can recall the location of previously encountered objects more accurately when those objects are on their own current level in the building rather than on a different level (Luo et al., 2020; Zwergal et al., 2016). Taken together, these findings suggest that multilevel buildings are cognitively represented as a stack of horizontal planes, so that locating an object on another level requires an extra processing step – integrating information across multiple planes – which can introduce errors (Jeffrey et al., 2013). Conversely, there also is experimental evidence for volumetric rather than stacked representations of multilevel buildings. Thus, Lu and Ye (2019) found that in an atrium-shaped shopping mall, same-level and different-level objects were located with similar accuracy. Those authors attributed their discrepant finding to the unusual architecture of the mall. Unlike conventional architectures, where opaque floors and ceilings prevent vision of other levels, the atrium-shaped mall allowed vision of all levels from a single vantage point. According to this interpretation, limited visual access to other levels supports stacked representations, while easy visual access to other levels facilitates volumetric representations. Indirect support for this interpretation comes from a study without architectural context (Hinterecker et al., 2018), in which a target array was displayed in either a horizontal or a vertical plane. This study reported that targets in either plane could be located with similar accuracy. This again suggests that, if visual access is unobstructed, vertical representations of space need not be more error-prone than horizontal ones. Conventional architectures limit not only visual , but also physical access to other levels. While we can move freely within a given level, movement between levels is restricted to one or a few stairwells, lifts, or escalators available in a building. The architecture used by Lu and Ye (2019) allowed easy visual access to other levels but limited physical access in a similar way as in conventional buildings, hence it left open whether easy physical access also facilitates volumetric representations. Two other studies used architectures that are suitable for answering this question. They designed virtual buildings which, conversely to the shopping mall of Lu and Ye (2019), limited visual access to other levels as in conventional buildings but allowed easy physical access to other levels by means of lifts (Thibault et al., 2013) or ladders (Dollé et al., 2015) placed in each room. One group of participants was transported through the virtual building level by level (first the bottom, then the middle and finally the top), whereas another group progressed vertically through the building column by column (moving up one column, down the next, and finally up the third). Both groups saw a unique object in each room. Subsequently, both groups were shown the same rooms in mixed order, and were asked to identify the object in one of the neighboring rooms. Although the two studies yielded an interesting set of findings [1] , they did not resolve whether the buildings were represented volumetrically or rather as a stack of horizontal layers. This is because both groups were transported through the same building, only along different routes, and both groups experienced within-level transport as well as between-level transport (the level-by-level group was transported up a column to the next level, and the column-by-column group was transported sideways along a level to the next column). Thus, both groups received comparable information and had to form a cognitive map of the same building. The present study was undertaken to determine whether easy physical access to other levels facilitates a volumetric cognitive representation of multilevel architectures. Drawing on the work of Thibault et al., (2013) and Dollé et al. (2015), transport up and down a level was possible at each object encountered, whereas vision of other levels was limited by opaque floors and ceilings. Unlike that earlier work, however, participants were tested with two distinct spatial layouts. In one, twelve objects were arranged within a 3D volume; in the other, the same number of objects was arranged in a horizontal plane. Horizontal transport was accomplished without turns, such that participants’ orientation relative to the environment remained fixed. This was to avoid the extra computational burden of rotational transformations between the egocentric reference frame in which the environment is viewed, and the allocentric (world-centered) reference frame in which cognitive maps are anchored (Richardson et al., 1999; Chrastil, 2013). Thus, instead of turning left or right, participants were transported sideways, and instead of turning left or right again, they were transported backwards. The visual experience was as if looking through the side window, or respectively through the rear window, of a moving car or bus. Objects were located at intersections from which six corridors of identical dimensions and décor departed (left, right, forward, back, up, down), thereby eliminating the horizontal/vertical asymmetry in ease of access which existed in both earlier studies. A within-participant design rather than a group comparison was employed to increase statistical power. If the 3D layout is represented as a stack of horizontal planes, objects should be located with poorer accuracy in the 3D layout compared to the horizontal layout, in analogy to the findings from conventional buildings. However, if the 3D layout is represented volumetrically, accuracy should be comparable in both layouts, in analogy to the findings from an atrium-shaped mall. Two experiments with somewhat different procedures were conducted to ensure the robustness of findings. One experiment additionally assessed the learning curve, and the other additionally evaluated the impact of left and right turns on participants’ accuracy. Experiment 1 Methods Participants The sample size required to compare participants’ performance in two layouts was calculated using G*power (Faul et al., 2007). For a within-factor with two levels, f = 0.25, a = 0.05 and ß = 0.9 a sample size of n = 44 was yielded. Note that statistical power actually was better than ß = 0.9 since each participant was examined three times per layout (see below), thus providing three rather than one data point for each level of layout. Forty-four participants were recruited through the internet platform Prolific and tested remotely on the internet platform Gorilla. Inclusion criteria were age range 20 – 40, use of a desktop PC, and fluency in English (since instructions were in English). Participants were 31.0 ±8.5 years old, 20 were female and 24 male, 33 held a university or college degree, ten a high school degree, and one had other education). This experiment is part of a larger research program pre-approved by the author’s institutional Ethics Commission (No. 062/2020). It has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent to participate and to have the data published anonymously was obtained from each participants before testing began. Setup and Procedures A virtual 3D maze was constructed with the architectural design software pCon.planner. Six corridors departed from each intersection, running leftward, rightward, forward, backward, upward and downward (see Fig. 1a). All corridors looked exactly the same, as they were created by copy-and-paste in pCon.planner. Two identical pictures of a bouquet faced each other at the center of each corridor to accentuate the gravitational vertical. Transport through the maze was pre-recorded with pCon.planner’s camera, keeping camera direction relative to the maze constant at all times. Camera speed was set so that transport from one intersection to the next took 2 seconds. The experiment consisted of practice, learning and test phases. In practice phases, participants were transported horizontally forward across three intersections. At each intersection, they were stopped for 3 s while a unique object was displayed, floating in the intersection (butterfly in Fig. 1a). After completing the trip, they saw a schematic drawing of the layout they had travelled, with discs and lines representing intersections and corridors, respectively (Fig. 1 b). One of the three objects they had visited was displayed as well, and they had to click, using their mouse, the disk where that object had been located. If the response was wrong, they had to try again until the response was correct. This procedure was repeated for all three objects. After that, the practice phase was repeated with vertical transport across three intersections. This ensured that all participants successfully used the schematic-drawing test to locate horizontally and vertically arranged objects. In the learning phases, participants were transported across twelve intersections arranged either in 3D (cf. Fig. 1c) or in a horizontal plane (cf. Fig. 1d). Again, transport took 2 s, and a unique object was displayed for 3 s during each stop at an intersection. Objects were cartoon images of an animal, a plant, a man-made or a natural structure. Before a trip through the maze began, they were shown a schematic drawing of layout, direction of transport, start point and end point (cf. Fig. 2 c & d). During the trip, they were shown the drawing again at the 3 d , 6 th and 9 th intersection (i.e., after completing a forward or backward segment of the trip). The drawings shown at intersections highlighted the participants’ momentary position in yellow, to help them stay oriented. These drawings were presented on the right wall ahead, after the object disappeared but before the next transport began. Before a learning phase began, participants were instructed to memorize the spatial location of each object since they would be asked about those locations later. They also were told that memorizing the temporal sequence of objects would be of no help since they would be tested in a different order. In the test phases, the schematic drawing from the preceding learning phase was shown again, but without red annotations. One of the objects visited during the learning phase was displayed alongside the drawing, and participants had to click the disc representing that object’s location. If a correct response was given within 12 s (time established in pilot tests), the next object was presented. Otherwise, an error message appeared for 3 s before the next object was displayed. Each of the twelve objects was presented once in a mixed order, and then once again in a different mixed order, never repeating the order experienced during the learning phase. Participants were instructed not to interrupt the test phase by other activities or rest breaks, and not to use external aids such as paper and pencil. The recorded timestamps confirmed that participants progressed swiftly through the experiment, without delays that would indicate looking up written notes, mental rehearsal, fatigue or mind wandering. Both in the 3D and the horizontal layout, the first learning phase consisted of two trips. The first trip was along the route shown in Fig.2 and the second was in the reverse direction, to discourage participants from memorizing the serial order of objects. After both trips were completed, a test phase followed. Then came the second learning phase (one trip, direction as in Fig. 1), followed by the second test phase, the third learning phase (one trip, direction opposite to Fug. 1), and the third test phase. Schematic drawing in the maze were presented for 4.5 s during the first learning phase, and for 3 s during the second and third learning phase. The order in which objects were presented during the test phase differed between the first, second and third repetition of this phase, to ensure that participants could not memorize a sequence of correct responses. Each participant was examined once in the 3D layout and once in the horizontal layout, using a different set of twelve objects in each layout. The assignment of object sets to layouts, and the temporal order of layouts, were both counterbalanced across participants. Data analysis Performance was quantified as accuracy, the proportion of correct responses in each repetition of a test phase. It ranged from 0 (all 24 mouse clicks wrong) to 1 (all mouse clicks correct). Accuracy served as the dependent variable in a mixed-effects model with Sex (female, male), Layout (horizontal, 3D) and Repetition (1, 2, 3) and as fixed effects, and Participant ID as random effect. Contrasts were calculated for Repetition 2 compared to Repetition 1, and for Repetition 3 compared to Repetition 2. Because of pronounced floor and ceiling effects (see Fig. 1), a logistic rather than a linear regression model was employed, since logistic regression is not sensitive to the distribution of residuals. Accordingly, accuracy was treated as an ordinal-scaled variable. The analysis was performed using the R function clmm (package ordinal, version 2023.12-4.1). Results Figure 2 illustrates that response accuracy varied widely, exhibiting ceiling effects (i.e., accuracy = 1) as well as floor effects (i.e., accuracy at or below the chance level of 2 / 24 = 0.08). This is consistent with previous findings on the high interindividual variability of spatial representations (Ishikawa & Montello, 2006 ; Pazzaglia et al., 2018 ; Hegarty et al., 2022 ). In total, 36.4% of data points in Fig. 2 were floor or ceiling effects, which prompted us to carry out logistic rather than linear regression. The outcome of this analysis is shown in Table 1 . Table 1 Outcome of logistic mixed effects model for accuracy in Experiment 1. effect estimate S.E. z p Sex (male vs. female) -0.622 0.720 -0.864 0.388 Layout (3D vs. horizontal) -0.444 0.243 -1.829 0.067 Repetition (2 vs. 1) 1.952 0.301 6.477 < 0.001 Repetition (3 vs. 2) 1.699 0.324 5.239 < 0.001 Note. Estimates are based on contrasts. For example, the two estimates for Repetition indicate that accuracy was 1.952 log units higher in repetition 2 compared to repetition 1, and was 1.699 log units higher in repetition 3 compared to repetition 2. According to Table 1 , Layout was not significant but significance was yielded for Repetition, with accuracy increasing from one repetition to the next. Since ceiling effects might have blurred any differences between layouts, particularly for repetition 2 and even more so for repetition 3, the analysis wars repeated with only the data from repetition 1. For this a reduced model was calculated, with only the terms Sex and Layout. In this exploratory analysis, Layout was again not significant (estimate = -0.244, S.E.= 0.388; z = -0.623; p = 0.527). Experiment 2 Methods To verify the robustness of findings, and to compare accuracy when participants are transported during the learning phase without versus with left and right turns, 36 new participants were recruited through the same platform and with the same inclusion criteria as for Exp. 1. They were 30.9 ± 8.9 years old, 15 were female and 21 were male; 9 held a high school degree, 26 a university or college degree, and 1 had other education. This experiment is again part of the research program pre-approved by the author’s institutional Ethics Commission (No. 062/2020), and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent to participate and to have the data published anonymously was obtained from each participants before testing began. Again, all provided their informed consent before testing began. Procedures differed from Exp. 1 in only two ways. First, both layouts were administered to each participant twice, once as in Exp. 1 and once with left and right turns in the learning phase. In the latter mode of transport, visual experience during horizontal movement resembled walking through corridors in real life, and during vertical movement it resembled riding a glass elevator. As a second difference to Exp. 1, only one repetition was administered for each combination of layout and transport mode, in order to avoid excessive fatigue. A different set of twelve objects was assigned to each combination of layout and transport mode. This assignment, as well as the order of layouts and transport modes, was counterbalanced across participants. Accuracy during the test phase was analyzed using a mixed-effects model with the fixed effects Sex, Layout and Transport (no turns, turns), and with Participant ID as random effect. Results Figure 3 illustrates that, as in Exp. 1, accuracy spanned the entire range from 0 to 1. Although the incidence of floor and ceiling effects was now reduced to 6.9%, the residuals of a linear mixed-effects model were not normally distributed (Shapiro test: p = 0.022), so a logistic regression model was applied. Table 2 again shows no statistical significance for Layout, but shows significance for Transport: accuracy after a learning phase with turns was reliably lower than that after a learning phase without turns (i.e., negative estimate in the last row of Table 2 ). Table 2 Outcome of logistic mixed effects model for accuracy in Experiment 2 effect estimate S.E. z p Sex (male vs. female) 0.132 0.771 0.172 0.864 Layout (3D vs. horizontal) -0.374 0.302 -1.238 0.216 Transport (turns vs. no turns) -0.844 0.309 -2.730 0.006 For a unified analysis, data from the matching conditions in both experiments were combined. These were the first repetition in Exp. 1, and transport without turns in Exp. 2. A mixed-effect model was calculated with accuracy as the dependent variable, Sex, Layout and Experiment as fixed effects, and Participant ID as random effect. The term Experiment was included since participants’ prior testing practice was somewhat different in the two experiments (i.e., 2 * 3 = 6 conditions in Exp. 1 but 2 * 2 = 4 conditions in Exp. 2). Again, the residuals of a linear model were not normally distributed (Shapiro test: p < 0.001), so a logistic model was used. Table 3 shows that in the combined data, Layout still did not reach statistical significance. Table 3 Outcome of logistic mixed effects model for accuracy in the combined data from Experiments 1 and 2 effect estimate S.E. z p Sex (male vs. female) -0.051 0.502 -0.102 0.918 Layout (3D vs. horizontal) -0.551 0.293 -1.882 0.060 Experiment (1 vs. 2) 0.216 0.501 0.431 0.666 The mean and standard deviation of accuracy in the combined data were 0.663 ± 0.284 for the 3D layout, and 0.705 ± 0.238 for the horizontal layout. These are only rough estimates of the data distribution since accuracy scores were not normally distributed (Kolmogorov-Smirnov test with Lilliefors correction: p > 0.001). With the same caveat, the effect size of Layout was approximated as Cohen’s d, yielding d = 0.160. Thus, the Layout term accounted for only a negligible proportion of the variance in accuracy. Discussion The present work examined whether easy physical access to other levels in a multilevel architecture promotes a volumetric cognitive representation – similar to that observed in a building with easy visual access to other levels (Lu and Ye, 2019 ) - or rather supports a representation as a stack of horizontal planes – consistent with findings for conventional buildings (Buechner et al., 2007 ; Hölscher et al., 2006 ; Luo et al., 2020; Zwergal et al., 2016 ). In line with the reasoning in these earlier studies and in Jeffrey et al., (2013), poorer localization of objects in a 3D layout compared to a horizontal layout would be interpreted as evidence for a stacked representation, whereas similar localization in both layouts would suggest a volumetric representation. Two experiments using somewhat different procedures were conducted with a total of 80 participants. In both experiments, the accuracy of object localization did not differ significantly between the 3D and the horizontal layout. When the matching conditions from both experiments were combined to base the analysis on data from all 80 participants, accuracy the difference between layouts remained non-significant. Cohen’s d, calculated as a rough estimate of the quantitative difference between the 3D and horizontal layouts, was 0.160. This compares well to the value of 0.093 which can be calculated from published data on object localization in a building with easy visual access to other levels (Lu and Ye, 2019 ), but not as well to the values of 0.75 and d = 1.86 which can be calculated from published data on object localization in conventional buildings (Luo et al., 2010 ; Zwergal et al., 2016 ). This pattern of findings supports the view that easy physical access, like easy visual access, promoted the formation of volumetric rather than stacked cognitive maps. Exp. 1 additionally confirmed that cognitive maps develop gradually through repeated practice rather than emerging abruptly through sudden insight. This finding, consistent with earlier work (Bock et al., 2024 ; Boone et al., 2018 ; Wiener et al., 2012 ), implies that the formation of an internal representation is a cumulative process in which spatial information is gradually consolidated. It would be interesting to find out whether this consolidation occurs faster near the external boundaries of the explored space than it its center, given the available evidence for a fundamental role of boundaries in spatial cognition (review in Lee, 2017 ). Exp. 2 additionally documented that cognitive representations in both layouts were more accurate when participants were transported without turns, even though transport with turns is more common in everyday life. The advantage of no-turn transport likely reflects lower computational costs, as no rotational transformations between egocentric and allocentric reference framers were required. Similarly, earlier research found that task realism can degrade performance, possibly due to its higher cognitive load (Engström et al., 2017 ; Tian et al., 2022 ). Notably, however, other studies have reported that task realism enhances rather than degrades performance, attributing this to the benefits of familiarity (Bock & Hagemann, 2010 ; Godden & Baddeley, 1975 ; Verhaeghen et al., 2012 ). It therefore appears that everyday-like experimental designs can either improve or impair performance, depending on situational factors. One limitation of this study is that it compares data from 80 participants with earlier findings based on smaller samples: 24 (Zwergal et al., 2016 ), 31 (Lu & Ye, 2019 ) or 40 participants (Luo et al., 2010 ). Due to their smaller sample size, these earlier data may be less reliable given the aforementioned substantial variability. Another limitation is that the present findings may not generalize to buildings with more complex floor plans, varied décor, and helpful or distracting architectural elements such as office doors, stairwells, signage, and obstacles (Arthur, 1992 ; Lynch, 1960 ). They also may not apply to guided walks involving active locomotion, to unguided exploration, or to task motivations beyond participating in an experiment on spatial orientation. In conclusion, our results suggest that easy physical access to other levels facilitates the formation of volumetric cognitive maps. If future research confirms the generalizability of this finding, a practical implication would be that in buildings with functionally distinct floors (e.g., hospitals, department stores, major train stations, airports), better physical access to other levels via additional staircases, escalators and lifts could not only improve users’ convenience and safety but also promote their holistic, three-dimensional representation of the building’s functional configuration. Declarations Funding: This work was financially supported by the Marga und Walter Boll Stiftung, grant 210-05.01-21. The sponsor had no other role except for funding. Author contributions: all CRediT roles by the author Declaration of interests: The author has no relevant financial or non-financial interests to disclose. Ethics : This study is part of a larger research program pre-approved by the author’s institutional Ethics Commission (No. 062/2020). It has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Consent to participate: obtained from all participants before testing began. Consent to publish: obtained from all participants before testing began. Data and materials availability: Raw data and the software code to run the experiments can be obtained from the author upon reasonable request, but the software company will charge for use of the software code. References Arthur P, Passini, R (1992) Wayfinding: people, signs, and architecture. McGraw-Hill, Columbus OH Bock O, Hagemann A (2010) An experimental paradigm to compare motor performance under laboratory and under everyday-like conditions. Journal of Neuroscience Methods 193:24–28. https://doi.org/10.1016/j.jneumeth.2010.08.005 Bock O, Huang J-Y, Onur OA, Memmert D (2024) The structure of cognitive strategies for wayfinding decisions. Psychological Research 88:476–486. Boone AP, Gong X, Hegarty M (2018) Sex differences in navigation strategy and efficiency. Memory & Cognition 46:909–922. https://doi.org/10.3758/s13421-018-0811-y Buechner S, Hölscher C, Strube G (2007) Path choice heuristics for navigation related to mental representations of a building. Proceedings of the European Cognitive Science Conference. Pp 504–509. Dollé L, Droulez J, Bennequin D, Berthoz A, Thibault G (2015) How the learning path and the very structure of a multifloored environment influence human spatial memory. Advances in Cognitive Psychology 11:156–162. Engström J, Markkula G, Victor T, Merat N (2017) Effects of Cognitive Load on Driving Performance: The Cognitive Control Hypothesis. Human Factors: The Journal of the Human Factors and Ergonomics Society 59:734–764. https://doi.org/10.1177/0018720817690639 Faul F, Erdfelder E, Lang A-G, Buchner A (2007) G* Power 3. Behavior Research Methods 39:175–191. doi.org/10.3758/BF03193146 Godden DR, Baddeley AD (1975) Context‐dependent memory in two natural environments: On land and underwater. British Journal of Psychology 66:325–331. https://doi.org/10.1111/j.2044-8295.1975.tb01468.x Hegarty M, He C, Boone AP, Yu S, Jacobs EG, Chrastil ER (2022) Understanding Differences in Wayfinding Strategies. Topics in Cognitive Science 10:102-119. https://doi.org/10.1111/tops.12592 Hinterecker T, Leroy C, Zhao M, Butz Mv, Bülthoff HH, Meilinger T (2018) No advantage for remembering horizontal over vertical spatial locations learned from a single viewpoint. Memory & Cognition 46:158–171. https://doi.org/10.3758/s13421-017-0753-9 Hölscher C, Meilinger T, Vrachliotis G, Brösamle M, Knauff M (2006) Up the down staircase: Wayfinding strategies in multi-level buildings. Journal of Environmental Psychology 26:284–299. https://doi.org/10.1016/j.jenvp.2006.09.002 Ishikawa T, Montello DR (2006) Spatial knowledge acquisition from direct experience in the environment. Cognitive Psychology 52:93–129. doi.org/10.1016/j.cogpsych.2005.08.003 Jeffery KJ, Jovalekic A, Verriotis M, Hayman R (2013) Navigating in a three-dimensional world. Behavioral and Brain Sciences 36:523–543. https://doi.org/10.1017/S0140525X12002476 Kim K, Bock O (2021) Acquisition of landmark, route, and survey knowledge in a wayfinding task: in stages or in parallel? Psychological Research 85(5): 2098–2106. https://doi.org/10.1007/s00426-020-01384-3 Lee S A (2017) The boundary-based view of spatial cognition: a synthesis. Current Opinion in Behavioral Sciences 16:58–65. https://doi.org/10.1016/j.cobeha.2017.03.006 Lu Y, & Ye Y (2019) Can people memorize multilevel building as volumetric map? A study of multilevel atrium building. Environment and Planning B: Urban Analytics and City Science 46:225–242. https://doi.org/10.1177/2399808317705659 Luo Z, Luo W, Wickens CD, Chen I-M (2010). Spatial learning in a virtual multilevel building: Evaluating three exocentric view aids. International Journal of Human-Computer Studies 68(10):746–759. https://doi.org/10.1016/j.ijhcs.2010.06.004 Lynch K (1960) The image of the city. MIT Press, Cambridge MA. McDonald TP, Pellegrino JW (1993) Psychological Perspectives on Spatial Cognition. Advances in Psychology 96:47–82. https://doi.org/10.1016/S0166-4115(08)60039-4 Nadel L (2013) Cognitive Maps. In Waller D, Nadel L (Eds.), Handbook of Spatial Cognition, (pp. 155–172). American Psychological Association. https://doi.org/10.1037/13936-000 Pazzaglia F, Meneghetti C, Ronconi L (2018) Tracing a Route and Finding a Shortcut: The Working Memory, Motivational, and Personality Factors Involved. Frontiers in Human Neuroscience 12. https://doi.org/10.3389/fnhum.2018.00225 Peer M, Brunec IK, Newcombe NS, Epstein RA (2021) Structuring Knowledge with Cognitive Maps and Cognitive Graphs. Trends in Cognitive Sciences 25:37–54. https://doi.org/10.1016/j.tics.2020.10.004 Thibault G, Pasqualotto A, Vidal M, Droulez J, Berthoz A (2013) How does horizontal and vertical navigation influence spatial memory of multifloored environments? Attention, Perception, & Psychophysics 75:10–15. https://doi.org/10.3758/s13414-012-0405-x Tian K, Markkula ., Wei C, Sadraei E, Hirose T, Merat N, Romano R (2022) Impacts of visual and cognitive distractions and time pressure on pedestrian crossing behaviour: A simulator study. Accident Analysis & Prevention, 174:106770. https://doi.org/10.1016/j.aap.2022.106770 Tolman EC (1948) Cognitive maps in rats and men. Psychological Review 55:189–208. https://doi.org/10.1037/h0061626 Verhaeghen P, Martin M, Sędek G (2012) Reconnecting cognition in the lab and cognition in real life: The role of compensatory social and motivational factors in explaining how cognition ages in the wild. Aging, Neuropsychology, and Cognition, 19:1–12. https://doi.org/10.1080/13825585.2011.645009 Wiener JM, Kmecova H, Condappa O (2012) Route repetition and route retracing. Frontiers in Aging Neuroscience 4. https://doi.org/10.3389/fnagi.2012.00007 Zhang H, Zherdeva K, Ekstrom AD (2014) Different “routes” to a cognitive map: dissociable forms of spatial knowledge derived from route and cartographic map learning. Memory & Cognition, 42:1106–1117. https://doi.org/10.3758/s13421-014-0418-x Zwergal A, Schöberl F, Xiong G, Pradhan C, Covic A, Werner P, Trapp C, Bartenstein P, la Fougère C, Jahn K, Dieterich M, Brandt T (2016) Anisotropy of Human Horizontal and Vertical Navigation in Real Space. Cerebral Cortex 26:4392–4404. https://doi.org/10.1093/cercor/bhv213) Footnotes The two studies yielded inconsistent results when comparing response accuracy in the two groups, and when comparing response accuracy for same-level objects to that for different-level objects. However, both studies consistently observed higher accuracy for objects located along the previously experienced direction of transport rather than perpendicular to it. The latter finding suggests that locating objects by serial order memory was more efficient than locating them by means of a cognitive map. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 31 Aug, 2025 Read the published version in Experimental Brain Research → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6350898","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":443441970,"identity":"e7a41f64-54f9-4479-a85e-e240bfe98329","order_by":0,"name":"OTMAR BOCK","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACAwYGZoYPbAwJDAxsQG4BEB8gQgvjDLgWAyK1MPOQpMWcvfmxsU1ZXZ58A1vigw8Gdxj4jjfg12LZc8w4Oefc4WKDA2yHDWcYPGOQPEPAGoMbOcyHc9sOJG6Qf94mzWNwGCiSQIQWy7a6xPkN7G3Sf0Ba7j8grCWZsY05seEA2zFpBrAt+HWA/WLYc+5w4oYDbMmGPQaHeSTPEHAYKMQkfpSBHMZm+OBHxWE5vuMHCFiDDnhIVD8KRsEoGAWjABsAAPBuReiBVhuaAAAAAElFTkSuQmCC","orcid":"","institution":"German Sport University Cologne","correspondingAuthor":true,"prefix":"","firstName":"OTMAR","middleName":"","lastName":"BOCK","suffix":""}],"badges":[],"createdAt":"2025-04-01 08:08:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6350898/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6350898/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00221-025-07136-2","type":"published","date":"2025-08-31T15:58:21+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":82077566,"identity":"399aceca-529b-46fa-a81b-89d09c555ff8","added_by":"auto","created_at":"2025-05-06 14:04:38","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":103156,"visible":true,"origin":"","legend":"\u003cp\u003eMaze interior, as seen by the participants during the learning phase (a), and schematic drawings shown to participants during the practice phase (b), during the learning phase with a three-dimensional layout (b), and during the learning phase with a horizontal layout (c). Arrows and text in red indicate the direction of transport\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6350898/v1/430dbc7294c8e596398321f1.jpg"},{"id":82079407,"identity":"8f907484-8ced-4928-b53b-8db8cd7de2d3","added_by":"auto","created_at":"2025-05-06 14:12:38","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":76176,"visible":true,"origin":"","legend":"\u003cp\u003eBox plot of response accuracy in both layouts and all three repetitions of Exp. 1. Data points are raw scores, with each point corresponding to the accuracy of one person\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6350898/v1/c14f74c3740b7ead55cb107f.jpg"},{"id":82079406,"identity":"e69f0f93-5c04-4b89-964b-af34f3713392","added_by":"auto","created_at":"2025-05-06 14:12:38","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":68201,"visible":true,"origin":"","legend":"\u003cp\u003eBox plot of response accuracy in both layouts and both transport modes of Exp. 2. Data points are raw scores, with each point corresponding to the accuracy of one person\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6350898/v1/e40275240c0ccfe42b6629c9.jpg"},{"id":90345549,"identity":"cf4d1d0a-7d5a-4fcf-b9f5-19bb0133009c","added_by":"auto","created_at":"2025-09-01 16:10:32","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":703374,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6350898/v1/abce04cc-d96e-4db4-8d94-79f37c871668.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Cognitive representations of multilevel buildings: two- or three-dimensional?","fulltext":[{"header":"Introduction","content":"\u003cp\u003eTo navigate landscapes, cities or buildings, we often rely on cognitive representations of the environment, often called “cognitive maps” (Tolman, 1948; Nadel, 2013). These maps enable us not only to follow familiar routes but also to detect shortcuts, plan detours around roadblocks, reach familiar destinations from a new starting point, and even devise entirely novel routes.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCognitive maps are not scaled replicas of the real world. Rather, behavioral and neurophysiological evidence suggests that these representations can be fragmented, distorted, biased by assumptions, organized within multiple reference frames, and sometimes lack metric information (McDonald \u0026amp; Pellegrino, 1993; Peer et al., 2021). Cognitive maps are formed gradually with practice (Kim \u0026amp; Bock, 2021; Zhang et al., 2014), at a rate that varies widely between individuals (Ishikawa \u0026amp; Montello, 2006), depending on factors such as emotional and personality traits (Pazzaglia et al., 2018), stress, sex and age ( Hegarty et al., 2022).\u003c/p\u003e\n\u003cp\u003eAlthough the world around us is three-dimensional, cognitive maps are not necessarily three-dimensional as well. Just as flat road maps and GPS screens allow us to successfully navigate even hilly terrains, and information boards depicting a building as a stack of horizontal planes allow us to find our destination, two-dimensional cognitive maps could also be sufficient for successful wayfinding.\u003c/p\u003e\n\u003cp\u003eExperimental evidence indeed suggests that multilevel buildings are cognitively represented as a stack of horizontal planes. When navigating such a building, participants typically first ascended to the correct level before proceeding horizontally toward their destination (Hölscher et al., 2006; Buechner et al., 2007); this strategy allowed them to reach their goal more quickly and with fewer errors than along other routes (Hölscher et al., 2006). Furthermore, participants can recall the location of previously encountered objects more accurately when those objects are on their own current level in the building rather than on \u0026nbsp;a different level (Luo et al., 2020; Zwergal et al., 2016). Taken together, these findings suggest that multilevel buildings are cognitively represented as a stack of horizontal planes, so that locating an object on another level requires an extra processing step – integrating information across multiple planes – which can introduce errors (Jeffrey et al., 2013).\u003c/p\u003e\n\u003cp\u003eConversely, there also is experimental evidence for volumetric rather than stacked representations of multilevel buildings.\u0026nbsp;Thus, Lu and Ye (2019) found that in an atrium-shaped shopping mall, same-level and different-level objects were located with similar accuracy.\u0026nbsp;Those authors attributed their discrepant finding to the unusual architecture of the mall. Unlike conventional architectures, where opaque floors and ceilings prevent vision of other levels, the atrium-shaped mall allowed vision of all levels from a single vantage point. According to this interpretation, limited visual access to other levels supports stacked representations, while easy visual access to other levels facilitates volumetric representations. Indirect support for this interpretation comes from a study without architectural context (Hinterecker et al., 2018), in which a target array was displayed in either a horizontal or a vertical plane. This study reported that targets in either plane could be located with similar accuracy. This again suggests that, if visual access is unobstructed, vertical representations of space need not be more error-prone than horizontal ones.\u003c/p\u003e\n\u003cp\u003eConventional architectures limit not only \u003cem\u003evisual\u003c/em\u003e, but also \u003cem\u003ephysical\u003c/em\u003e access to other levels. While we can move freely within a given level, movement between levels is restricted to one or a few stairwells, lifts, or escalators available in a building. The architecture used by Lu and Ye (2019) allowed easy visual access to other levels but limited physical access in a similar way as in conventional buildings, hence it left open whether easy physical access also facilitates volumetric representations.\u003c/p\u003e\n\u003cp\u003eTwo other studies used architectures that are suitable for answering this question. They designed virtual buildings which, conversely to the shopping mall of Lu and Ye (2019), limited \u003cem\u003evisual\u003c/em\u003e access to other levels as in conventional buildings but allowed easy \u003cem\u003ephysical\u003c/em\u003e access to other levels by means of lifts (Thibault et al., 2013) or ladders (Dollé et al., 2015) placed in each room. One group of participants was transported through the virtual building level by level (first the bottom, then the middle and finally the top), whereas another group progressed vertically through the building column by column (moving up one column, down the next, and finally up the third). Both groups saw a unique object in each room. Subsequently, both groups were shown the same rooms in mixed order, and were asked to identify the object in one of the neighboring rooms. Although the two studies yielded an interesting set of findings\u003csup\u003e[1]\u003c/sup\u003e, they did not resolve whether the buildings were represented volumetrically or rather as a stack of horizontal layers. This is because both groups were transported through the same building, only along different routes, and both groups experienced within-level transport as well as between-level transport (the level-by-level group was transported up a \u003cem\u003ecolumn\u003c/em\u003e to the next level, and the column-by-column group was transported sideways along a \u003cem\u003elevel\u003c/em\u003e to the next column). Thus, both groups received comparable information and had to form a cognitive map of the same building.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe present study was undertaken to determine whether easy physical access to other levels facilitates a volumetric cognitive representation of multilevel architectures. Drawing on the work of Thibault et al., (2013) and Dollé et al. (2015), transport up and down a level was possible at each object encountered, whereas vision of other levels was limited by opaque floors and ceilings. Unlike that earlier work, however,\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eparticipants were tested with two distinct spatial layouts. In one, twelve objects were arranged within a 3D volume; in the other, the same number of objects was arranged in a horizontal plane.\u003c/li\u003e\n \u003cli\u003eHorizontal transport was accomplished without turns, such that participants’ orientation relative to the environment remained fixed. This was to avoid the extra computational burden of rotational transformations between the egocentric reference frame in which the environment is viewed, and the allocentric (world-centered) reference frame in which cognitive maps are anchored (Richardson et al., 1999; Chrastil, 2013). Thus, instead of turning left or right, participants were transported sideways, and instead of turning left or right again, they were transported backwards. The visual experience was as if looking through the side window, or respectively through the rear window, of a moving car or bus.\u003c/li\u003e\n \u003cli\u003eObjects were located at intersections from which six corridors of identical dimensions and décor departed (left, right, forward, back, up, down), thereby eliminating the horizontal/vertical asymmetry in ease of access which existed in both earlier studies.\u003c/li\u003e\n \u003cli\u003eA within-participant design rather than a group comparison was employed to increase statistical power.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eIf the 3D layout is represented as a stack of horizontal planes, objects should be located with poorer accuracy in the 3D layout compared to the horizontal layout, in analogy to the findings from conventional buildings. However, if the 3D layout is represented volumetrically, accuracy should be comparable in both layouts, in analogy to the findings from an atrium-shaped mall. Two experiments with somewhat different procedures were conducted to ensure the robustness of findings. One experiment additionally assessed the learning curve, and the other additionally evaluated the impact of left and right turns on participants’ accuracy.\u003c/p\u003e\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n"},{"header":"Experiment 1","content":"\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\u003cp\u003e\u003cem\u003eParticipants\u003c/em\u003e\u003c/p\u003e\u003cp\u003eThe sample size required to compare participants’ performance in two layouts was calculated using G*power (Faul et al., 2007). For a within-factor with two levels, f = 0.25, a = 0.05 and ß = 0.9 a sample size of n = 44 was yielded. Note that statistical power actually was better than ß = 0.9 since each participant was examined three times per layout (see below), thus providing three rather than one data point for each level of layout. \u0026nbsp;\u003c/p\u003e\u003cp\u003eForty-four participants were recruited through the internet platform Prolific and tested remotely on the internet platform Gorilla. Inclusion criteria were age range 20 – 40, use of a desktop PC, and fluency in English (since instructions were in English). Participants were 31.0 ±8.5 years old, 20 were female and 24 male, 33 held a university or college degree, ten a high school degree, and one had other education).\u003c/p\u003e\u003cp\u003eThis experiment is part of a larger research program pre-approved by the author’s institutional Ethics Commission (No. 062/2020). It has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent to participate and to have the data published anonymously was obtained from each participants before testing began.\u003c/p\u003e\u003cp\u003e\u003cem\u003eSetup and Procedures\u003c/em\u003e\u003c/p\u003e\u003cp\u003eA virtual 3D maze was constructed with the architectural design software pCon.planner. Six corridors departed from each intersection, running leftward, rightward, forward, backward, upward and downward (see Fig. 1a). All corridors looked exactly the same, as they were created by copy-and-paste in pCon.planner. Two identical pictures of a bouquet faced each other at the center of each corridor to accentuate the gravitational vertical. Transport through the maze was pre-recorded with pCon.planner’s camera, keeping camera direction relative to the maze constant at all times. Camera speed was set so that transport from one intersection to the next took 2 seconds.\u003c/p\u003e\u003cp\u003eThe experiment consisted of practice, learning and test phases. In practice phases, participants were transported horizontally forward across three intersections. At each intersection, they were stopped for 3 s while a unique object was displayed, floating in the intersection (butterfly in Fig. 1a). After completing the trip, they saw a schematic drawing of the layout they had travelled, with discs and lines representing intersections and corridors, respectively (Fig. 1 b). One of the three objects they had visited was displayed as well, and they had to click, using their mouse, the disk where that object had been located. If the response was wrong, they had to try again until the response was correct. This procedure was repeated for all three objects. After that, the practice phase was repeated with vertical transport across three intersections. This ensured that all participants successfully used the schematic-drawing test to locate horizontally and vertically arranged objects.\u003c/p\u003e\u003cp\u003eIn the learning phases, participants were transported across twelve intersections arranged either in 3D (cf. \u0026nbsp;Fig. 1c) or in a horizontal plane (cf. Fig. 1d). Again, transport took 2 s, and a unique object was displayed for 3 s during each stop at an intersection. Objects were cartoon images of an animal, a plant, a man-made or a natural structure. Before a trip through the maze began, they were shown a schematic drawing of layout, direction of transport, start point and end point (cf. Fig. 2 c \u0026amp; d). During the trip, they were shown the drawing again at the 3\u003csup\u003ed\u003c/sup\u003e, 6\u003csup\u003eth\u003c/sup\u003e and 9\u003csup\u003eth\u003c/sup\u003e intersection (i.e., after completing a forward or backward segment of the trip). The drawings shown at intersections highlighted the participants’ momentary position in yellow, to help them stay oriented. These drawings were \u0026nbsp;presented on the right wall ahead, after the object disappeared but before the next transport began. Before a learning phase began, participants were instructed to memorize the spatial location of each object since they would be asked about those locations later. They also were told that memorizing the temporal sequence of objects would be of no help since they would be tested in a different order.\u003c/p\u003e\u003cp\u003eIn the test phases, the schematic drawing from the preceding learning phase was shown again, but without red annotations. One of the objects visited during the learning phase was displayed alongside the drawing, and participants had to click the disc representing that object’s location. If a correct response was given within 12 s (time established in pilot tests), the next object was presented. Otherwise, an error message appeared for 3 s before the next object was displayed. Each of the twelve objects was presented once in a mixed order, and then once again in a different mixed order, never repeating the order experienced during the learning phase. Participants were instructed not to interrupt the test phase by other activities or rest breaks, and not to use external aids such as paper and pencil. The recorded timestamps confirmed that participants progressed swiftly through the experiment, without delays that would indicate looking up written notes, mental rehearsal, fatigue or mind wandering.\u003c/p\u003e\u003cp\u003eBoth in the 3D and the horizontal layout, the first learning phase consisted of two trips. The first trip was along the route shown in Fig.2 and the second was in the reverse direction, to discourage participants from memorizing the serial order of objects. After both trips were completed, a test phase followed. Then came the second learning phase (one trip, direction as in Fig. 1), followed by the second test phase, the third learning phase (one trip, direction opposite to Fug. 1), and the third test phase. Schematic drawing in the maze were presented for 4.5 s during the first learning phase, and for 3 s during the second and third learning phase. The order in which objects were presented during the test phase differed between the first, second and third repetition of this phase, to ensure that participants could not memorize a sequence of correct responses.\u003c/p\u003e\u003cp\u003eEach participant was examined once in the 3D layout and once in the horizontal layout, using a different set of twelve objects in each layout. The assignment of object sets to layouts, and the temporal order of layouts, were both counterbalanced across participants.\u003c/p\u003e\u003cp\u003e\u003cem\u003eData analysis\u003c/em\u003e\u003c/p\u003e\u003cp\u003ePerformance was quantified as accuracy, the proportion of correct responses in each repetition of a test phase. It ranged from 0 (all 24 mouse clicks wrong) to 1 (all mouse clicks correct). Accuracy served as the dependent variable in a mixed-effects model with Sex (female, male), Layout (horizontal, 3D) and Repetition (1, 2, 3) and as fixed effects, and Participant ID as random effect. Contrasts were calculated for Repetition 2 compared to Repetition 1, and for Repetition 3 compared to Repetition 2.\u003c/p\u003e\u003cp\u003eBecause of pronounced floor and ceiling effects (see Fig. 1), a logistic rather than a linear regression model was employed, since logistic regression is not sensitive to the distribution of residuals. Accordingly, accuracy was treated as an ordinal-scaled variable. The analysis was performed using the R function \u003cem\u003eclmm\u003c/em\u003e (package \u003cem\u003eordinal,\u003c/em\u003e version 2023.12-4.1).\u003c/p\u003e\n\u003ch3\u003eResults\u003c/h3\u003e\n\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates that response accuracy varied widely, exhibiting ceiling effects (i.e., accuracy = 1) as well as floor effects (i.e., accuracy at or below the chance level of 2 / 24 = 0.08). This is consistent with previous findings on the high interindividual variability of spatial representations (Ishikawa \u0026amp; Montello, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Pazzaglia et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Hegarty et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In total, 36.4% of data points in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e were floor or ceiling effects, which prompted us to carry out logistic rather than linear regression. The outcome of this analysis is shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOutcome of logistic mixed effects model for accuracy in Experiment 1.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eeffect\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eestimate\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS.E.\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ez\u003c/p\u003e \u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (male vs. female)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.622\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.864\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.388\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLayout (3D vs. horizontal)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.444\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.243\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-1.829\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.067\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRepetition (2 vs. 1)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.952\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.301\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e6.477\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt; 0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRepetition (3 vs. 2)\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.699\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.324\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.239\u003c/p\u003e \u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt; 0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote. Estimates are based on contrasts. For example, the two estimates for Repetition indicate that accuracy was 1.952 log units higher in repetition 2 compared to repetition 1, and was 1.699 log units higher in repetition 3 compared to repetition 2.\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e \u003cp\u003e\u003c/p\u003e \u003cp\u003eAccording to Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Layout was not significant but significance was yielded for Repetition, with accuracy increasing from one repetition to the next. Since ceiling effects might have blurred any differences between layouts, particularly for repetition 2 and even more so for repetition 3, the analysis wars repeated with only the data from repetition 1. For this a reduced model was calculated, with only the terms Sex and Layout. In this exploratory analysis, Layout was again not significant (estimate = -0.244, S.E.= 0.388; z = -0.623; p = 0.527).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Experiment 2","content":"\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eTo verify the robustness of findings, and to compare accuracy when participants are transported during the learning phase without versus with left and right turns, 36 new participants were recruited through the same platform and with the same inclusion criteria as for Exp. 1. They were 30.9 ± 8.9 years old, 15 were female and 21 were male; 9 held a high school degree, 26 a university or college degree, and 1 had other education. This experiment is again part of the research program pre-approved by the author’s institutional Ethics Commission (No. 062/2020), and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Informed consent to participate and to have the data published anonymously was obtained from each participants before testing began.\u003c/p\u003e\u003cp\u003e Again, all provided their informed consent before testing began.\u003c/p\u003e\u003cp\u003eProcedures differed from Exp. 1 in only two ways. First, both layouts were administered to each participant twice, once as in Exp. 1 and once with left and right turns in the learning phase. In the latter mode of transport, visual experience during horizontal movement resembled walking through corridors in real life, and during vertical movement it resembled riding a glass elevator. As a second difference to Exp. 1, only one repetition was administered for each combination of layout and transport mode, in order to avoid excessive fatigue.\u003c/p\u003e\u003cp\u003eA different set of twelve objects was assigned to each combination of layout and transport mode. This assignment, as well as the order of layouts and transport modes, was counterbalanced across participants.\u003c/p\u003e\u003cp\u003eAccuracy during the test phase was analyzed using a mixed-effects model with the fixed effects Sex, Layout and Transport (no turns, turns), and with Participant ID as random effect.\u003c/p\u003e\n\u003ch3\u003eResults\u003c/h3\u003e\n\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates that, as in Exp. 1, accuracy spanned the entire range from 0 to 1. Although the incidence of floor and ceiling effects was now reduced to 6.9%, the residuals of a linear mixed-effects model were not normally distributed (Shapiro test: p\u0026thinsp;=\u0026thinsp;0.022), so a logistic regression model was applied. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e again shows no statistical significance for Layout, but shows significance for Transport: accuracy after a learning phase with turns was reliably \u003cem\u003elower\u003c/em\u003e than that after a learning phase without turns (i.e., negative estimate in the last row of Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOutcome of logistic mixed effects model for accuracy in Experiment 2\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eeffect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eestimate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS.E.\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ez\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (male vs. female)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.132\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.771\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.172\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.864\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLayout (3D vs. horizontal)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.374\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.302\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-1.238\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.216\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTransport (turns vs. no turns)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.844\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.309\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-2.730\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFor a unified analysis, data from the matching conditions in both experiments were combined. These were the first repetition in Exp. 1, and transport without turns in Exp. 2. A mixed-effect model was calculated with accuracy as the dependent variable, Sex, Layout and Experiment as fixed effects, and Participant ID as random effect. The term Experiment was included since participants\u0026rsquo; prior testing practice was somewhat different in the two experiments (i.e., 2 * 3\u0026thinsp;=\u0026thinsp;6 conditions in Exp. 1 but 2 * 2\u0026thinsp;=\u0026thinsp;4 conditions in Exp. 2). Again, the residuals of a linear model were not normally distributed (Shapiro test: p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), so a logistic model was used. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows that in the combined data, Layout still did not reach statistical significance.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOutcome of logistic mixed effects model for accuracy in the combined data from Experiments 1 and 2\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eeffect\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eestimate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS.E.\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ez\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (male vs. female)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.051\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.502\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.102\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.918\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLayout (3D vs. horizontal)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.551\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.293\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-1.882\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.060\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExperiment (1 vs. 2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.216\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.501\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.431\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.666\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe mean and standard deviation of accuracy in the combined data were 0.663\u0026thinsp;\u0026plusmn;\u0026thinsp;0.284 for the 3D layout, and 0.705\u0026thinsp;\u0026plusmn;\u0026thinsp;0.238 for the horizontal layout. These are only rough estimates of the data distribution since accuracy scores were not normally distributed (Kolmogorov-Smirnov test with Lilliefors correction: p\u0026thinsp;\u0026gt;\u0026thinsp;0.001). With the same caveat, the effect size of Layout was approximated as Cohen\u0026rsquo;s d, yielding d\u0026thinsp;=\u0026thinsp;0.160. Thus, the Layout term accounted for only a negligible proportion of the variance in accuracy.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe present work examined whether easy physical access to other levels in a multilevel architecture promotes a volumetric cognitive representation \u0026ndash; similar to that observed in a building with easy \u003cem\u003evisual\u003c/em\u003e access to other levels (Lu and Ye, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) - or rather supports a representation as a stack of horizontal planes \u0026ndash; consistent with findings for conventional buildings (Buechner et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; H\u0026ouml;lscher et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Luo et al., 2020; Zwergal et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). In line with the reasoning in these earlier studies and in Jeffrey et al., (2013), poorer localization of objects in a 3D layout compared to a horizontal layout would be interpreted as evidence for a stacked representation, whereas similar localization in both layouts would suggest a volumetric representation.\u003c/p\u003e \u003cp\u003eTwo experiments using somewhat different procedures were conducted with a total of 80 participants. In both experiments, the accuracy of object localization did not differ significantly between the 3D and the horizontal layout. When the matching conditions from both experiments were combined to base the analysis on data from all 80 participants, accuracy the difference between layouts remained non-significant. Cohen\u0026rsquo;s d, calculated as a rough estimate of the quantitative difference between the 3D and horizontal layouts, was 0.160. This compares well to the value of 0.093 which can be calculated from published data on object localization in a building with easy visual access to other levels (Lu and Ye, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), but not as well to the values of 0.75 and d\u0026thinsp;=\u0026thinsp;1.86 which can be calculated from published data on object localization in conventional buildings (Luo et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Zwergal et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). This pattern of findings supports the view that easy physical access, like easy visual access, promoted the formation of volumetric rather than stacked cognitive maps.\u003c/p\u003e \u003cp\u003eExp. 1 additionally confirmed that cognitive maps develop gradually through repeated practice rather than emerging abruptly through sudden insight. This finding, consistent with earlier work (Bock et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Boone et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Wiener et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), implies that the formation of an internal representation is a cumulative process in which spatial information is gradually consolidated. It would be interesting to find out whether this consolidation occurs faster near the external boundaries of the explored space than it its center, given the available evidence for a fundamental role of boundaries in spatial cognition (review in Lee, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eExp. 2 additionally documented that cognitive representations in both layouts were more accurate when participants were transported without turns, even though transport with turns is more common in everyday life. The advantage of no-turn transport likely reflects lower computational costs, as no rotational transformations between egocentric and allocentric reference framers were required. Similarly, earlier research found that task realism can degrade performance, possibly due to its higher cognitive load (Engstr\u0026ouml;m et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Tian et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Notably, however, other studies have reported that task realism enhances rather than degrades performance, attributing this to the benefits of familiarity (Bock \u0026amp; Hagemann, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Godden \u0026amp; Baddeley, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e1975\u003c/span\u003e; Verhaeghen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). It therefore appears that everyday-like experimental designs can either improve or impair performance, depending on situational factors.\u003c/p\u003e \u003cp\u003eOne limitation of this study is that it compares data from 80 participants with earlier findings based on smaller samples: 24 (Zwergal et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), 31 (Lu \u0026amp; Ye, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) or 40 participants (Luo et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Due to their smaller sample size, these earlier data may be less reliable given the aforementioned substantial variability. Another limitation is that the present findings may not generalize to buildings with more complex floor plans, varied d\u0026eacute;cor, and helpful or distracting architectural elements such as office doors, stairwells, signage, and obstacles (Arthur, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1992\u003c/span\u003e; Lynch, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e1960\u003c/span\u003e). They also may not apply to guided walks involving active locomotion, to unguided exploration, or to task motivations beyond participating in an experiment on spatial orientation.\u003c/p\u003e \u003cp\u003eIn conclusion, our results suggest that easy physical access to other levels facilitates the formation of volumetric cognitive maps. If future research confirms the generalizability of this finding, a practical implication would be that in buildings with functionally distinct floors (e.g., hospitals, department stores, major train stations, airports), better physical access to other levels via additional staircases, escalators and lifts could not only improve users\u0026rsquo; convenience and safety but also promote their holistic, three-dimensional representation of the building\u0026rsquo;s functional configuration.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This work was financially supported by the Marga und Walter Boll Stiftung, grant 210-05.01-21. The sponsor had no other role except for funding.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u003c/strong\u003e all CRediT roles by the author\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interests:\u003c/strong\u003e The author has no relevant financial or non-financial interests to disclose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics\u003c/strong\u003e: This study is part of a larger research program pre-approved by the author\u0026rsquo;s institutional Ethics Commission (No. 062/2020). It has therefore been performed in accordance with the ethical standards laid down in the\u0026nbsp;1964 Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate:\u003c/strong\u003e obtained from all participants before testing began.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to publish:\u003c/strong\u003e obtained from all participants before testing began.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and materials availability:\u003c/strong\u003e Raw data and the software code to run the experiments can be obtained from the author upon reasonable request, but the software company will charge for use of the software code.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eArthur P, Passini, R (1992) Wayfinding: people, signs, and architecture. McGraw-Hill, Columbus OH\u003c/li\u003e\n\u003cli\u003eBock O, Hagemann A (2010) An experimental paradigm to compare motor performance under laboratory and under everyday-like conditions. Journal of Neuroscience Methods 193:24\u0026ndash;28. https://doi.org/10.1016/j.jneumeth.2010.08.005\u003c/li\u003e\n\u003cli\u003eBock O, Huang J-Y, Onur OA, Memmert D (2024) The structure of cognitive strategies for wayfinding decisions. Psychological Research 88:476\u0026ndash;486. \u003c/li\u003e\n\u003cli\u003eBoone AP, Gong X, Hegarty M (2018) Sex differences in navigation strategy and efficiency. Memory \u0026amp; Cognition 46:909\u0026ndash;922. https://doi.org/10.3758/s13421-018-0811-y\u003c/li\u003e\n\u003cli\u003eBuechner S, H\u0026ouml;lscher C, Strube G (2007) Path choice heuristics for navigation related to mental representations of a building. Proceedings of the European Cognitive Science Conference. Pp 504\u0026ndash;509.\u003c/li\u003e\n\u003cli\u003eDoll\u0026eacute; L, Droulez J, Bennequin D, Berthoz A, Thibault G (2015) How the learning path and the very structure of a multifloored environment influence human spatial memory. Advances in Cognitive Psychology 11:156\u0026ndash;162.\u003c/li\u003e\n\u003cli\u003eEngstr\u0026ouml;m J, Markkula G, Victor T, Merat N (2017) Effects of Cognitive Load on Driving Performance: The Cognitive Control Hypothesis. Human Factors: The Journal of the Human Factors and Ergonomics Society 59:734\u0026ndash;764. https://doi.org/10.1177/0018720817690639\u003c/li\u003e\n\u003cli\u003eFaul F, Erdfelder E, Lang A-G, Buchner A (2007) G* Power 3. Behavior Research Methods 39:175\u0026ndash;191. doi.org/10.3758/BF03193146\u003c/li\u003e\n\u003cli\u003eGodden DR, Baddeley AD (1975) Context‐dependent memory in two natural environments: On land and underwater. British Journal of Psychology 66:325\u0026ndash;331. https://doi.org/10.1111/j.2044-8295.1975.tb01468.x\u003c/li\u003e\n\u003cli\u003eHegarty M, He C, Boone AP, Yu S, Jacobs EG, Chrastil ER (2022) Understanding Differences in Wayfinding Strategies. Topics in Cognitive Science 10:102-119. https://doi.org/10.1111/tops.12592\u003c/li\u003e\n\u003cli\u003eHinterecker T, Leroy C, Zhao M, Butz Mv, B\u0026uuml;lthoff HH, Meilinger T (2018) No advantage for remembering horizontal over vertical spatial locations learned from a single viewpoint. Memory \u0026amp; Cognition 46:158\u0026ndash;171. https://doi.org/10.3758/s13421-017-0753-9\u003c/li\u003e\n\u003cli\u003eH\u0026ouml;lscher C, Meilinger T, Vrachliotis G, Br\u0026ouml;samle M, Knauff M (2006) Up the down staircase: Wayfinding strategies in multi-level buildings. Journal of Environmental Psychology 26:284\u0026ndash;299. https://doi.org/10.1016/j.jenvp.2006.09.002\u003c/li\u003e\n\u003cli\u003eIshikawa T, Montello DR (2006) Spatial knowledge acquisition from direct experience in the environment. Cognitive Psychology 52:93\u0026ndash;129. doi.org/10.1016/j.cogpsych.2005.08.003\u003c/li\u003e\n\u003cli\u003eJeffery KJ, Jovalekic A, Verriotis M, Hayman R (2013) Navigating in a three-dimensional world. Behavioral and Brain Sciences 36:523\u0026ndash;543. https://doi.org/10.1017/S0140525X12002476\u003c/li\u003e\n\u003cli\u003eKim K, Bock O (2021) Acquisition of landmark, route, and survey knowledge in a wayfinding task: in stages or in parallel? Psychological Research 85(5): 2098\u0026ndash;2106. https://doi.org/10.1007/s00426-020-01384-3 \u003c/li\u003e\n\u003cli\u003eLee S A (2017) The boundary-based view of spatial cognition: a synthesis. Current Opinion in Behavioral Sciences 16:58\u0026ndash;65. https://doi.org/10.1016/j.cobeha.2017.03.006\u003c/li\u003e\n\u003cli\u003eLu Y, \u0026amp; Ye Y (2019) Can people memorize multilevel building as volumetric map? A study of multilevel atrium building. Environment and Planning B: Urban Analytics and City Science 46:225\u0026ndash;242. https://doi.org/10.1177/2399808317705659 \u003c/li\u003e\n\u003cli\u003eLuo Z, Luo W, Wickens CD, Chen I-M (2010). Spatial learning in a virtual multilevel building: Evaluating three exocentric view aids. International Journal of Human-Computer Studies 68(10):746\u0026ndash;759. https://doi.org/10.1016/j.ijhcs.2010.06.004\u003c/li\u003e\n\u003cli\u003eLynch K (1960) The image of the city. MIT Press, Cambridge MA.\u003c/li\u003e\n\u003cli\u003eMcDonald TP, Pellegrino JW (1993) Psychological Perspectives on Spatial Cognition. Advances in Psychology 96:47\u0026ndash;82. https://doi.org/10.1016/S0166-4115(08)60039-4\u003c/li\u003e\n\u003cli\u003eNadel L (2013) Cognitive Maps. In Waller D, Nadel L (Eds.), Handbook of Spatial Cognition, (pp. 155\u0026ndash;172). American Psychological Association. https://doi.org/10.1037/13936-000\u003c/li\u003e\n\u003cli\u003ePazzaglia F, Meneghetti C, Ronconi L (2018) Tracing a Route and Finding a Shortcut: The Working Memory, Motivational, and Personality Factors Involved. Frontiers in Human Neuroscience 12. https://doi.org/10.3389/fnhum.2018.00225\u003c/li\u003e\n\u003cli\u003ePeer M, Brunec IK, Newcombe NS, Epstein RA (2021) Structuring Knowledge with Cognitive Maps and Cognitive Graphs. Trends in Cognitive Sciences 25:37\u0026ndash;54. https://doi.org/10.1016/j.tics.2020.10.004\u003c/li\u003e\n\u003cli\u003eThibault G, Pasqualotto A, Vidal M, Droulez J, Berthoz A (2013) How does horizontal and vertical navigation influence spatial memory of multifloored environments? Attention, Perception, \u0026amp; Psychophysics 75:10\u0026ndash;15. https://doi.org/10.3758/s13414-012-0405-x\u003c/li\u003e\n\u003cli\u003eTian K, Markkula ., Wei C, Sadraei E, Hirose T, Merat N, Romano R (2022) Impacts of visual and cognitive distractions and time pressure on pedestrian crossing behaviour: A simulator study. Accident Analysis \u0026amp; Prevention, 174:106770. https://doi.org/10.1016/j.aap.2022.106770\u003c/li\u003e\n\u003cli\u003eTolman EC (1948) Cognitive maps in rats and men. Psychological Review 55:189\u0026ndash;208. https://doi.org/10.1037/h0061626\u003c/li\u003e\n\u003cli\u003eVerhaeghen P, Martin M, Sędek G (2012) Reconnecting cognition in the lab and cognition in real life: The role of compensatory social and motivational factors in explaining how cognition ages in the wild. Aging, Neuropsychology, and Cognition, 19:1\u0026ndash;12. https://doi.org/10.1080/13825585.2011.645009\u003c/li\u003e\n\u003cli\u003eWiener JM, Kmecova H, Condappa O (2012) Route repetition and route retracing. Frontiers in Aging Neuroscience 4. https://doi.org/10.3389/fnagi.2012.00007\u003c/li\u003e\n\u003cli\u003eZhang H, Zherdeva K, Ekstrom AD (2014) Different \u0026ldquo;routes\u0026rdquo; to a cognitive map: dissociable forms of spatial knowledge derived from route and cartographic map learning. Memory \u0026amp; Cognition, 42:1106\u0026ndash;1117. https://doi.org/10.3758/s13421-014-0418-x\u003c/li\u003e\n\u003cli\u003eZwergal A, Sch\u0026ouml;berl F, Xiong G, Pradhan C, Covic A, Werner P, Trapp C, Bartenstein P, la Foug\u0026egrave;re C, Jahn K, Dieterich M, Brandt T (2016) Anisotropy of Human Horizontal and Vertical Navigation in Real Space. Cerebral Cortex 26:4392\u0026ndash;4404. https://doi.org/10.1093/cercor/bhv213) \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e The two studies yielded inconsistent results when comparing response accuracy in the two groups, and when comparing response accuracy for same-level objects to that for different-level objects. However, both studies consistently observed higher accuracy for objects located along the previously experienced direction of transport rather than perpendicular to it. The latter finding suggests that locating objects by serial order memory was more efficient than locating them by means of a cognitive map.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"cognitive map, mental representation, multistorey architecture, spatial navigation, spatial orientation, human","lastPublishedDoi":"10.21203/rs.3.rs-6350898/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6350898/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePrevious research suggests that conventional multilevel buildings are cognitively represented as a stack of horizontal planes, whereas an atrium-shaped architecture which allows easy visual access to other levels is represented as a volume. The present study investigated whether easy \u003cem\u003ephysical\u003c/em\u003e access to other levels also promotes a volumetric representation. Participants were examined in a virtual 3D grid maze in which they could access higher and lower levels \u003cem\u003eat each intersection\u003c/em\u003e. During a learning phase, they were transported through the maze across twelve intersection, each featuring a unique object. In the subsequent test phase, they were asked to indicate the location of these objects on a schematic drawing of the maze. Response accuracy in the test phase was similar when the twelve visited objects were arranged in a horizontal plane and when they were laid out in a volume. In accordance with earlier reasoning, this suggests that easy \u003cem\u003ephysical\u003c/em\u003e access to other levels indeed can facilitate a volumetric cognitive representation of multilevel architectures. Additional findings suggest that this representation emerged gradually rather than abruptly like a sudden insight, and that transport through the maze without left and right turns facilitated the internal representation, probably by reducing the associated cognitive load.\u003c/p\u003e","manuscriptTitle":"Cognitive representations of multilevel buildings: two- or three-dimensional?","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-06 14:04:33","doi":"10.21203/rs.3.rs-6350898/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a79025ed-4d6b-420f-a272-5dfb9df54059","owner":[],"postedDate":"May 6th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-09-01T16:08:22+00:00","versionOfRecord":{"articleIdentity":"rs-6350898","link":"https://doi.org/10.1007/s00221-025-07136-2","journal":{"identity":"experimental-brain-research","isVorOnly":false,"title":"Experimental Brain Research"},"publishedOn":"2025-08-31 15:58:21","publishedOnDateReadable":"August 31st, 2025"},"versionCreatedAt":"2025-05-06 14:04:33","video":"","vorDoi":"10.1007/s00221-025-07136-2","vorDoiUrl":"https://doi.org/10.1007/s00221-025-07136-2","workflowStages":[]},"version":"v1","identity":"rs-6350898","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6350898","identity":"rs-6350898","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}