Manipulating the visibility of barriers to improve spatial navigation efficiency and cognitive mapping.

Previous studies from psychology, neuroscience and geography showed that environmental barriers fragment the representation of the environment, reduce spatial navigation efficiency, distort distance estimation and make spatial updating difficult. Despite these negative effects, limited research has examined how to overcome barriers and if individual differences mediate their causes and potential interventions. We hypothesize that the reduced visibility caused by barriers plays a major role in accumulating error in spatial updating and encoding spatial relationships. We tested this using virtual navigation to grant participants 'X-ray' vision during environment encoding (i.e., barriers become translucent) and quantifying cognitive mapping benefits of counteracting fragmented visibility. We found that compared to the participants trained with naturalistic environment visibility, participants trained in the translucent environment had better performance in wayfinding and pointing tasks, which are theorized to measure navigation efficiency and cognitive mapping. Interestingly, these benefits were only observed in participants with high self-report sense of direction. Together, our results provide important insight into (1) how perceptual barrier effects manifest, even when physical fragmentation of space is held constant, (2) establish a novel intervention that can improve spatial learning, and (3) provide evidence that individual differences modulate perceptual barrier effects and the efficacy of such interventions.


Procedure for part and whole conditions
As described in the main text, the procedure differences among all experimental conditions (whole_opaque, whole_translucent, part_opaque and part_translucent) were only in the training phase and the descriptions of the testing phase can be found in the main text. Therefore, we only describe training phase of the whole and part conditions here.

Whole conditions.
Participants were presented with a virtual town with all buildings fully textured. Participants were first provided a list of the names of the target storefronts (9) and five of which were marked with a check mark. Participants were told to find the storefronts with a check mark first, in any order, but we emphasized that all 9 target storefronts were equally important. The five checkmarked storefronts in the whole conditions were to match the five storefronts presented in the first training session of the part conditions. The first training session ended when at least six minutes had passed and participants found all the five check-marked storefronts (indicated by hitting the event trigger poles). If participants used less than six minutes to find all the checkmarked storefronts, they were encouraged to find the remaining storefronts (4) until the sixminute time limit was reached; if participants could not find all the check-marked storefronts in six minutes, the first session would not end until all of the five were found. When the first session ended, participants performed a pointing task. The second training session started when participants finished the five pointing trials.
In the second training session, participants found the remaining targeted storefronts in any order and were told to keep exploring to learn the spatial layout of this environment if there was any time left. The second training session ended when at least six minutes had passed and all the remaining storefronts were found. Participants then performed the 13 consecutive pointing task trials.

Part conditions.
The training scheme in this condition was identical to the whole conditions with the following changes: a) In the first session, the name list of the storefronts provided to participants only contained the five (out of 9) check-marked storefronts from the whole conditions. In the second session, the name list contained the remaining four target storefronts. b) In the first session, only the buildings that contained the five to-be-found storefronts were textured. The other buildings with the four storefronts to be learned in the second session were not textured (Figures 1b + d; also see video demo [https://osf.io/3bcxs]). In the second session, the untextured buildings from the first session became textured, and all but one textured building in the first session were untextured ( Figure 2). The one shared textured building between session 1 and 2 encouraged participants to remember that both sessions involved the same environment. This mimicked the two sessions training experience of the whole conditions, except that participants were not able to see the (currently) irrelevant storefronts in each of the two training sessions while still being able to perceive the full geometry of the environment. The pointing tasks in the part conditions, probing within and between phase location associations, were identical to those in the whole conditions.  The means and standard deviations (in parenthesis) of the self-report information and training behaviors in the opaque and translucent conditions. No significant differences were found on any of the metrics (ps > .12). ANCOVA was used to control for all the metrics presented in this table. After controlling for these factors, participants in the translucent conditions still outperformed their counterparts in the opaque conditions in the first wayfinding task (F(1,36) = 14.15, p = .001), the third wayfinding task (F(1,36) = 4.63, p = .038), the first pointing task in the testing phase ( Figure 3  -.146 -.619*** Notes: The correlation coefficients between SOD and performance across the opaque and translucent conditions* p < .05, ** p < .01, *** p < .001. WOwhole, opaque condition. POpart, opaque condition. WTwhole, translucent condition. PTpart, translucent condition. The partial correlation coefficients between SOD and performance when gender effect was controlled for across experimental conditions. SOD and wayfindingthe partial correlation coefficient between SOD and the performance of the first wayfinding task. SOD and pointingthe partial correlation coefficient between SOD and the performance of the second pointing task in the testing phase. * p < .05, ** p < .01. WOwhole, opaque condition. POpart, opaque condition. WTwhole, translucent condition. PTpart, translucent condition. Figure S1. Wayfinding performance for high and low SOD groups in the penetrable environment. Opaque conditions were a combination of WO and PO conditions. Translucent conditions were a combination of WT and PT conditions. Error bars are ± 1 SEM estimated from data within conditions. n.s. non significant.

Figure S2.
Wayfinding and pointing task performance for high and low SOD groups, separated by experimental condition. WOwhole, opaque condition. POpart, opaque condition. WTwhole, translucent condition. PTpart, translucent condition.
6. When you are in your city do you naturally individuate cardinal points, that is do you find easily where north, south, east, and west are?
1 (not at all) 2 3 4 5 (very much) 7. Someone is describing for you the route to reach an unfamiliar place. Do you prefer: a. to make an image of the route?
1 (not at all) 2 3 4 5 (very much) b. to remember the description verbally?
1 (not at all) 2 3 4 5 (very much) 8. In a complex building (store, museum) do you think spontaneously and easily about your direction in relation to the general structure of the building and the external environment?