ConductScience VR
Geometric arenas in virtual reality
People learn which corner of an empty virtual arena hides a goal, then search for it again after the arena changes shape or they are moved outside its walls.
ConductScience Editorial · Published November 15, 2026 · 6 references
- Task
- Find a goal hidden in a right-angled corner, where only the arena’s shape marks the place.1
- Rectangle to kite
- Adults trained in a virtual rectangle searched the matching corner of a kite.2
- Outside the walls
- Moved outside the arena, adults searched longest near the outside of the goal corner.1
- Children
- From about 10.5 years, children searched in the right place when only the corner’s walls, or only the whole shape, could guide them.3
What geometric arenas test
Geometric arenas test how people use the shape of a space to find a place that nothing else marks.1,3 A person might remember the whole boundary, called a global-shape representation, or only the stretch of wall that marks the goal, a local-shape representation.3 Each test takes the goal away and records how long a person searches near the corners.4,3 The change of arena between training and test decides which of the two representations can lead back to the goal.3
The question goes back to experiments with rats by Cheng in 1986.4 Rats learned to find food buried in one corner of a rectangular arena, with a different feature in each corner.4,1 With the features removed, the rats still searched the correct corner and the one diagonally opposite, which has the same shape, more often than chance would predict.4,1 Cheng proposed that animals hold the overall shape of a space in a dedicated geometric module, which other cues cannot influence.1
Pearce and colleagues later pointed out that the rats need not have learned the overall shape at all.4 They could have learned only the corner where the food was, for example one with a short wall to the left of a long wall.4 In a rectangle, the corner diagonally opposite has the same local shape, so the whole shape and the corner alone send a rat to the same two places.1,3
A kite built from the rectangle’s four walls separates the two.3 Its overall shape is different, so a memory of the whole rectangle gives no reason to prefer any of its corners.3 It keeps two right-angled corners, one with a short wall to the left of a long wall and one with the short wall on the right.4,1 People who learned the goal corner’s walls should search the kite corner that matches them, while people who learned the whole shape should favor neither right-angled corner.4
Moving people outside the arena works the other way, keeping the whole shape and changing how the goal corner looks.1 From outside, the goal corner shows a short wall to the right of a long wall, and its walls meet at 270 degrees instead of 90.1,3 A memory of the whole shape still leads to the outside of the goal corner, while the trained arrangement of walls now matches the other right-angled corner.1
| Design | Training and test | What can lead to the goal |
|---|---|---|
| Same arena | Trained and tested inside the same kite3 | The corner’s walls and the whole shape alike3 |
| Transfer to a new shape | Trained in a rectangle and tested in a kite made of the same walls, or the reverse4,3 | Only the corner’s walls3 |
| Inside to outside | Trained inside a kite, tested from outside it1 | Only the whole shape1 |
| Cross-shaped arena | Trained inside a cross, tested from outside it1 | The whole shape and the corner’s walls, each to a different corner1 |
How a session runs
In the studies by Buckley and colleagues, people see the arena in first person on a desktop screen and move with the arrow keys.4,1,3 The up and down keys move a person forward and back, and the left and right keys turn them.4 In the later studies, people were no more than 50 cm from the screen, and a small table and chair set it at each child’s eye height.1,3 The children’s version showed a 45-degree field of view.3
The rectangle and the kite share the same four walls, two of 9 m and two of 27 m.4 These sizes take the walking speed in the arena as 2 m per second, similar to walking in the real world, and give each arena a perimeter of 72 m.4 The kite’s other two corners measure 143.14 and 36.86 degrees, and the walls are about 2.5 m high.4 The floor has a grass texture and the sky is black, so only the shape of the walls shows where the goal is.4,1 The goal is an invisible square in one right-angled corner, 1.08 m wide for adults and 2 m wide for children.4,3 The rectangle holds two goals, one in each of its two matching corners, while the kite holds one.4
Practice moving in a separate arena
Adults first moved around an eight-sided practice arena for two 30-second trials.4 For children, the experimenter showed two trials in a square practice arena before the child ran one.3 The instructions give the goal a story, such as a hidden wireless signal for adults or a hiding ghost for children.1,3
Train with the goal hidden in a right-angled corner
Training runs 16 trials, each starting halfway between the kite’s acute and obtuse corners and facing a random direction.1,3 In the rectangle, each trial starts from the center.4 A trial ends only when the person finds the goal, and a message on the screen tells them so.4 A flag appears at the goal if they have not found it after 120 seconds, or after 60 seconds in the children’s version.1,3 Each trial’s time and distance to the goal are recorded, and people get faster as training goes on.4,3 In the children’s cross experiment, training was cut from 16 trials to 8 to limit fatigue.3
Test with the goal removed
After training, the person gets a test trial with the goal removed, without being told.3 A test inside an arena lasts 60 seconds, a test outside the kite 120 seconds, and a test outside the larger cross 240 seconds.4,1,3 Outside the kite, the test starts 3.15 m from the middle of one wall, facing it.1
Score the time spent near the corners
Inside an arena, square zones 3.24 m on a side mark the right-angled corners.4 Outside, L-shaped zones with sides of 6.48 m and 3.24 m wrap around the right-angled corners.1 The score is the time in the correct zone as a share of the time in the correct and incorrect zones, so chance is 50%.1,3 With the four zones of the cross test, chance is 25%.1 Buckley and colleagues also recorded which corner adults entered first, following Pearce and colleagues’ rat study.4
What the task has shown in people
Adults moved from a rectangle to a kite
Lew and colleagues trained adults in a virtual rectangle and found robust transfer to the matching corner of a kite.2 They used two designs, one modeled on the rat task and one on studies of young children.2 Buckley and colleagues repeated the result from a rectangle to a kite and from a kite to a rectangle.4 People spent more time at the test corner whose walls matched the goal corner in training, and at least 75% in each group went there first.4 The authors read this as evidence against a strategy of heading for one end of a single wall.4
Young children in real arenas
Lew and colleagues ran the same transfer with children of 2.5 to 3.5 years in real arenas, while the adults had used virtual ones.2 The children showed no transfer of search to the kite.2 They still searched above chance in the rectangle they trained in, and in a study that used the kite alone.2 Lew and colleagues suggest that the overall shape may help people regain their heading, and that matching a corner within a new shape may be an advanced developmental achievement.2
Corner shape and wall color
Buckley and colleagues found that a corner’s shape and the color of the walls compete as cues, in a way that fits associative theories of spatial learning.4 Blocking means that a cue learned first keeps a cue added later from being learned.4 In one experiment, people first found the goal from wall colors in a square, then trained in a rectangle with the same colors.4 In a test kite they spent no more time at the correct corner than at the wrong one, while a control group preferred the correct corner.4 In a second experiment, learning the corner’s shape first blocked later learning about wall colors.4 The authors note that this two-way blocking is hard to explain with theories in which people encode the whole shape.4
Searching from outside the walls
Buckley and colleagues trained adults inside a virtual kite, then tested them from outside its walls.1 Counting only time in the two corner zones, they spent 76% at the goal corner when the test looked like training, against 50% by chance.1 They spent 73% there when the floor, walls and sky changed between training and test.1 Buckley and colleagues argue that local walls or a remembered view cannot easily explain this, and that a memory of the whole shape can.1
Outside a cross-shaped arena, a corner with the goal corner’s walls and right angle lies near the middle of the arena, away from where the goal was.1 Of their time in the four zones, adults spent 62% at the outside of the goal corner, against 25% by chance, and only 7% at the matching corner near the middle.1 Together with earlier work, Buckley and colleagues suggest that people encode both the corner’s walls and the whole shape.1
School-age children
Buckley and colleagues ran four experiments with 69 to 72 children each, aged 6 to 12, trained in a virtual arena and tested once.3 Trained and tested in the same kite, children searched the goal corner above chance from 6 years.3 When the test let only the corner’s walls or only the whole shape lead to the goal, children searched above chance from about 10.5 years.3 Outside the cross, where the two representations conflict, children followed the whole shape above chance from about 8.5 years, younger than when the whole shape alone had to guide them.3 The authors suggest that the two abilities develop in parallel, and caution against reading 10.5 years as a sharp turning point.3 Like the young children in Lew and colleagues’ study, 6-year-olds succeeded when the arena kept its shape but not when it changed.3
Wall features and shape
Kelly and Bischof tested how adults orient in enclosed virtual rooms with no landmarks, where only the room’s shape and features on its walls could guide them.5 With the features on the walls themselves, men and women both encoded the shape and the features.5 Earlier studies with separate landmarks had found that, for women, the landmarks overshadowed the shape.5 People relied on the features more than on the shape, though their first experience of a room changed how they weighed the two.5 A goal was easier to find near a place that stood out by its shape or its features.5 Kelly and Bischof conclude that adults use a flexible strategy to encode space.5
How it compares with the rat arena tasks
The transfer from rectangle to kite follows a rat water maze design by Pearce and colleagues, and gave comparable results.4 Rats learned to find a submerged platform in one corner of an arena with two long and two short sides.6 They were then trained in a new arena with the same kinds of sides and a different overall shape, such as a kite after a rectangle.6,4 The platform was easier to find in the corner that matched the old one than in its mirror image.6 Pearce and colleagues suggest that the rats located the platform by local cues, without using the arena’s overall shape.6 Lesions of the hippocampus impaired the rats in these arenas, and the authors suggest that it plays a role in navigating by those cues.6
Blocking of shape learning by wall colors has also been reported in rats in water mazes.4 In both rats and people, the hippocampus has been linked to navigation by the shape of the boundaries.4
Rats and people differed in one result.4 In the kite, Pearce and colleagues found no significant difference between rats’ first visits to the correct right-angled corner and to the acute corner.4 People in the virtual kite preferred only the correct right-angled corner.4 Buckley and colleagues suggest that the rats’ visits to the acute corner may reflect an unconditioned preference for small, dark corners.4
Buckley and colleagues note that, unlike the rat studies, theirs used virtual arenas and gave people instructions at set points.1 A desktop screen also takes away the vestibular, proprioceptive and somesthetic input that walking through a real room provides.4,3 The authors point out that rats are denied some of these cues too, because researchers place them into the apparatus.3 Studies of animals and children usually record the first corner chosen, while Buckley and colleagues score time in corner zones, the only suitable measure outside the walls.3
Why the studies used virtual arenas
Buckley and colleagues chose virtual arenas to control strictly which strategies children could use.3 Children could not see over the walls, and differences in height, angle of view and travel speed between younger and older children were controlled.3 Buckley and colleagues add that a virtual arena is not limited by how much laboratory space is available.3
Build virtual geometric arenas with ConductScience VR
We write the task software in house and build each set of arenas to the protocol you plan to run.

- The arena shapes, wall lengths and heights, wall colors and floor textures, and the goal corner and its size
- Training trials, test trials with the goal removed, transfers to a new shape and moves from inside to outside the walls, with your start points and time limits
- Each trial’s path, time and distance to the goal, time in each corner zone and the first corner entered, saved for analysis
- A desktop screen with a keyboard or joystick, or a headset
- The software alone, or a package with the headset and computer
We also build teaching versions, so students can test each other on the arena tasks in class.
Related pages
Rodent behavior tests rebuilt in virtual reality for people, and how we build them.
Our water maze pool, in sizes for mice and rats.
A platform hidden in a virtual pool, found from cues around the room.
Map-based and route-based navigation in a virtual star-shaped maze.
Questions
- Why is the arena shaped like a kite?
- A kite made from a rectangle’s walls changes the overall shape but keeps right-angled corners with the same pairs of walls.1,3 So a search at the matching kite corner, after training in a rectangle, fits a memory of the corner’s walls and is hard to explain by the whole shape.4
- Do people remember the whole shape or only the goal corner?
- Adults can use either one.2,1 They carry a corner’s walls over to a new shape, and from outside an arena, where the corner’s walls are reversed, they still follow the whole shape.2,4,1 When the two point to different corners outside a cross, adults mostly follow the whole shape.1
- What does the task measure?
- It measures how long people search near the corners once the goal is removed.4,1,3 The score is the share of that time spent at the correct corner, and some studies also record the first corner a person enters.4,1 Training adds the time and distance to the goal on each trial.3
- From what age do children use the shape of an arena?
- Children searched the right corner of an unchanged virtual arena from 6 years, the youngest age tested.3 Using only the corner’s walls or only the whole shape came later, from about 10.5 years.3 In real arenas, children of 2.5 to 3.5 years did not carry a corner over from a rectangle to a kite.2
References
- Buckley MG, Holden LJ, Spicer SG, et al. Crossing boundaries: Global reorientation following transfer from the inside to the outside of an arena. Journal of Experimental Psychology: Animal Learning and Cognition. 2019. doi:10.1037/xan0000206
- Lew AR, Usherwood B, Fragkioudaki F, et al. Transfer of spatial search between environments in human adults and young children (Homo sapiens): implications for representation of local geometry by spatial systems. Developmental Psychobiology. 2014. doi:10.1002/dev.21109
- Buckley MG, Holden LJ, Smith AD, et al. The developmental trajectories of children’s reorientation to global and local properties of environmental geometry. Journal of Experimental Psychology: General. 2024. doi:10.1037/xge0001265
- Buckley MG, Smith AD, Haselgrove M. Blocking spatial navigation across environments that have a different shape. Journal of Experimental Psychology: Animal Learning and Cognition. 2016. doi:10.1037/xan0000084
- Kelly DM, Bischof WF. Orienting in virtual environments: How are surface features and environmental geometry weighted in an orientation task?. Cognition. 2008. doi:10.1016/j.cognition.2008.07.012
- Pearce JM, Good MA, Jones PM, et al. Transfer of spatial behavior between different environments: implications for theories of spatial learning and for the role of the hippocampus in spatial learning. Journal of Experimental Psychology: Animal Behavior Processes. 2004. doi:10.1037/0097-7403.30.2.135
More virtual reality tests
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
ConductScience VR
