Virtual environments used in training.

A: Screenshots of six different outdoor environments that used in training. These are also images used in both the behavioral and fMRI testing. B: Structure of the virtual maze used in the training phase. First, participants followed the path to the outdoor scene, and performed an object detection task at a designated location. In the navigable condition, participants were able to navigate forward after the task to reach the destination. In the non-navigable condition, an invisible wall was placed in front of the designated location. When the participant tried to navigate forward, they would not be able to as there was an invisible wall. They would have to turn back and use another path to reach the destination.

Behavioral testing showed navigational experience affect visual scene representation.

A: The top panel shows the visual same/different task paradigm used in the testing. The second and third panel demonstrates an example of condition assignment. In the analysis, only trials with different response were divided into matched navigability and mismatched navigability condition. When a participants respond ‘different’ to a scene pair, a slower response time is expected in the matched condition compared to the mismatched condition if the navigational experience affect visual scene representation. B: Matched navigability condition (Mean=685ms, 95% CI: [637, 732]) showed a slower response time compared to the mismatched condition (Mean=712ms, 95% CI: [662, 762]) in the experimental group but not in the control group. These results suggest navigational experience affect visual scene representation even in visual task that does not involve navigation. ** indicates p<0.01.

Navigational experience information can be decoded from PPA.

A: Participants performed a same/different color dot task while viewing scene images in the fMRI scanner. A SVM classifier is trained to classify whether the neural pattern of visually controlled scenes are distinguishable based on navigable experience or not. A leave-one-trial-out cross validation approach is used to test the performance of the classifier. B: Scene-selective region PPA shows an above chance classification accuracy suggesting its representation contains navigational experience information. * indicates p<0.05.