Skip to content
Lizely

Spatial Memory Game

Remember one object in an original 3D coordinate scene, rotate the view, and derive its new perspective position across five exact rounds.

Privacy: your files never leave your device. All processing happens locally in your browser.

How to use

  1. 1.Press Reveal scene and study the marked object across the X, Y, and Z depth rails; take as long as you need.
  2. 2.Press Hide and rotate view only after memorizing the position. The objects disappear and the published yaw and pitch are applied.
  3. 3.Choose the option card where that same world point projects in the rotated view. Clear five rounds for exactly 1,000 points.

About Spatial Memory Game

Spatial Memory Game is a five-round recreational puzzle built around actual three-dimensional coordinates. Each round begins with a covered scene rather than an already visible answer. Press Reveal scene to inspect three original objects placed at different horizontal, vertical, and depth positions. One object is surrounded by a dashed marker and named as the item to remember. There is no countdown and no speed bonus, so you decide when the position is clear.

The scene uses three world axes. X represents left and right, Y represents height, and Z represents depth. A perspective floor grid and three labeled axis rails make those relationships visible without WebGL or a hardware sensor. The orb, prism, and ring use distinct silhouettes, but the task is not to remember a color or identify a shape. The important information is where the marked object sits relative to the origin, the other objects, and the depth grid.

After studying the marked object, press Hide and rotate view. The objects disappear before the recall choices are shown. The camera then adopts the round's published yaw and pitch. Yaw turns the scene around its vertical axis, while pitch tilts the view. Four option cards show the same rotated coordinate frame with a crosshair at a different projected screen position. Choose the card that places the remembered world point in the correct location after that view change.

This makes the game materially different from Number Memory Test, Visual Memory Test, and Position Memory Game. Number Memory Test asks for a hidden digit string. Visual Memory Test asks players to recreate marked cells on a fixed two-dimensional grid. Position Memory Game asks players to return named objects to their original cells. Spatial Memory Game does not reproduce the original screen position at all. It rotates a remembered point in three dimensions and asks where perspective projection sends that point in a new view.

The underlying geometry is deterministic and local. Every scene stores three literal world points. To create the recall view, the game first applies a yaw rotation around the Y axis. It then applies a pitch rotation around the X axis. Those two rigid rotations preserve the point's distance from the origin. A perspective step converts the rotated X and Y coordinates into screen coordinates, using rotated Z as depth. Therefore two points with the same X and Y but different Z values can appear at different screen positions.

The five scene files, fifteen object positions, view angles, and twenty candidate positions are original Lizely product fixtures. They are not copied from a standardized memory task, examination, textbook, psychological instrument, medical protocol, external game, or real-world coordinate dataset. The rotation convention is anchored to Carnegie Mellon's published 3D axis-matrix material, while Microsoft Learn and MDN independently anchor the depth-dependent perspective behavior. The focal length of 240, camera distance of 320, 320 by 220 screen, angles, and world points remain authored display and game parameters rather than external facts. Production stores no correct option index and gives no option a semantic answer label. For each round, the game projects the marked object's literal world coordinate, compares that result with the four literal candidate positions, and requires exactly one geometric match.

The interface keeps Reveal and Hide explicit. A covered scene does not accept an answer, and Hide cannot run before Reveal. During recall, pointer and touch users can press any option card. Keyboard users can Tab into the option group, use the Arrow keys across its two-by-two layout, and press Enter or Space. Roving focus keeps the actual focused button, visible focus ring, and logical cursor aligned. After a user activates Reveal, focus moves to Hide; after Hide, it moves to Option A; after a correct non-final answer, it returns to Reveal; and completion or deadlock moves focus to the readable terminal status. Initial page load never steals focus. The marked reveal object exposes its exact X, Y, and Z world coordinates, while each visible candidate exposes its exact screen x value out of 320 and screen y value out of 220. These descriptions provide an exact semantic equivalent without marking correctness. No correct marker, answer class, hidden answer text, data attribute, or answer-bearing ARIA description appears before selection.

Each correct projection awards exactly 200 points. Five correct rounds produce the fixed maximum of 1,000 points. One distinct incorrect option can be repaired in the same round. Repeating that unchanged wrong option does not add another error, because the failure signature combines the round and display position. A second different wrong option closes the run. Completion and deadlock freeze Reveal, Hide, cursor movement, and answer submission. Restart always returns to the first covered scene with zero score and no stored failures.

Independent tests use ten externally anchored literal cases for identity, Y-axis and X-axis basis rotations, ordered yaw-then-pitch composition, the authored principal point, neutral-depth scale, and positive-versus-negative depth behavior. A second literal list covers the five authored round targets, views, expected projected points, and expected display positions, while a separate matrix oracle recalculates every round rather than reading the production result. Tests also prove that rotation preserves three-dimensional radius, every fixture keeps a positive perspective denominator, every round has four unique candidates and exactly one match, all cursor boundaries behave correctly, duplicate failures are ignored, a first miss remains repairable, terminal actions freeze, and the full route reaches exactly 1,000.

All play happens in the browser with React and SVG. The game does not upload coordinates, request an account, call an API, use WebGL, inspect device orientation, access a camera, read motion hardware, generate random scenes, or add a package. If local storage is available, the shared game shell remembers only the best completed score on that device. Blocking storage removes that convenience without changing the geometry or answer.

Spatial Memory Game is entertainment, not a measurement. It does not diagnose memory loss, neurological conditions, vision, learning needs, dementia, injury, attention, intelligence, spatial ability, school readiness, or occupational fitness. The five fixed authored scenes are not normed, calibrated, adaptive, or comparable between people. Display size, input method, familiarity with perspective drawings, and ordinary distraction can affect a play session without supporting any conclusion about health or ability.

Methodology & sources

Author five local scenes with three literal XYZ object positions apiece, one visible marked object, one literal yaw-pitch recall view, and four literal candidate screen positions. Reveal is explicit; Hide is legal only after Reveal and removes every object before recall. Use the disclosed right-handed active convention: rotate each target with a Y-axis yaw matrix followed by an X-axis pitch matrix, then project with screenX = 160 + rotatedX * 240 / (320 - rotatedZ) and screenY = 110 - rotatedY * 240 / (320 - rotatedZ). Carnegie Mellon anchors the axis-matrix convention, Microsoft Learn anchors camera-distance perspective scaling, and MDN independently cross-checks positive-versus-negative depth behavior. The focal length 240, camera distance 320, 320 by 220 screen, angles, world points, and candidate points are authored display or game parameters. Compare candidates within 0.001 screen unit and require exactly one derived match; production stores neither a correct index nor a semantic option label. Ten source-anchored literal golden cases pin the matrix signs, composition order, center, scale, and depth direction; a separate oracle then recalculates all five authored round fixtures. State tests cover explicit Covered-Reveal-Recall phases, exact 200-point progression to 1,000, same-error deduplication, first-error repair, second-distinct-error deadlock, every two-by-two keyboard cursor boundary, invalid action atomicity, Restart equality, and terminal freezing. Component controls provide pointer activation, roving keyboard focus, exact non-answer screen x/y descriptions for visible candidates, exact XYZ descriptions for the marked reveal object, and no pre-submit semantic answer attribute. User-triggered transitions hand focus from Reveal to Hide, Hide to Option A, each correct non-final answer to Reveal, and terminal states to their readable status without stealing focus on initial load. The game is a local SVG recreation with no random source, WebGL, hardware, external scene dataset, diagnostic claim, new dependency, or network call.

Frequently asked questions

How does the 3D view change work?
Each round applies a real yaw rotation, then a pitch rotation, and finally a perspective projection that uses the rotated Z coordinate as depth.
How is this different from the other memory games?
It does not repeat digits, recreate highlighted 2D cells, or return objects to original slots. You remember one world point and locate its new projection after a 3D camera rotation.
Does the score measure spatial memory or cognitive ability?
No. This is a fixed recreational puzzle, not a standardized, educational, medical, neurological, diagnostic, intelligence, or ability assessment.

Related tools

Mini Games guides

View all