Multiple Object Tracking Practice

What multiple object tracking asks you to do

This trainer separates identification from tracking. At the start of a round, a subset of otherwise identical balls is highlighted. The highlight disappears before motion begins. From that moment onward there is no visual label that tells you which balls are targets. Your job is to maintain several object identities while targets and distractors change position together. When motion stops, the same objects become selectable and you reconstruct the target set from attention and short-term visual memory.

Why the paths are deterministic instead of improvised every frame

Each committed source contains normalized starting positions and motion vectors. The runtime evaluates every object from a single normalized path parameter, so the same seed reconstructs the same path exactly. A generator samples each source across its complete path and rejects candidates whenever two object centers approach closer than the frozen separation contract. This avoids a common ambiguity in moving-dot games: two dots visually merging so closely that identity becomes guesswork instead of tracking.

Speed changes playback time, not path geometry

The path is defined against normalized progress from zero to one. A 0.8× round and a 1.2× round therefore visit the same sequence of positions; the faster round simply completes that sequence in less clock time. This design is important for verification. Increasing speed does not create a new unverified collision that was absent at the default speed. It also makes a Clean best-speed record easier to interpret because the geometric problem remains constant while temporal demand changes.

Easy, Standard and Challenge change attention load

Easy uses six moving objects and asks you to track two. Standard uses eight objects and three targets. Challenge uses ten objects and four targets. The number of distractors rises at the same time as target count, so harder rounds require maintaining more identities inside a denser field. Challenge also uses a mild shared speed-surge profile during playback. All objects share that normalized timing profile, preserving the verified relative path while making the motion rhythm less uniform.

How fixed and Adaptive speed differ

Fixed Practice offers 0.8×, 1.0× and 1.2× playback. Adaptive starts from a browser-local speed for the selected level. An exact round increases that speed by 0.05×; a non-exact round lowers it by 0.05×, within a conservative range. The adaptive number is a local task setting, not a population percentile. It is useful because it keeps the next round near your current success boundary without inventing a global brain score.

Exact means every target and no distractor

After motion stops you must select exactly the requested number of objects. The result reports how many true targets you recovered and how many selected objects were distractors. An Exact round requires every target and zero false picks. This makes partial success visible instead of collapsing everything into a binary score. Over time the local history can show whether you are usually losing one target, overconfidently substituting a distractor, or recovering the whole target set.

Peek is intentionally marked Assisted

During motion, Peek briefly reveals the target set again. That can be useful when learning how to distribute attention, but it removes part of the memory burden. The round is therefore marked Assisted as soon as Peek is used. Assisted rounds still record useful hit and false-pick data, but they do not replace the Clean best-speed record. Clean means the targets were identified once at the beginning and not revealed again before selection.

A better tracking strategy than chasing one ball at a time

Following a single target with your eyes can make the others disappear from attention. A more stable strategy is to keep your gaze near the center of the field and represent the targets as a changing configuration: a triangle, line, cluster or set of relative gaps. As that shape stretches and folds, update the whole configuration instead of repeatedly jumping your gaze between individual balls. This is also why increasing target count changes the task so sharply.

Why selection time is reported separately

The moving phase has a fixed playback duration for its speed setting, so a total round timer would mainly report a number chosen by the game. The trainer instead reports selection time from the moment motion stops until you submit your target set. Selection time is secondary to accuracy, but it can reveal hesitation after tracking. Because mouse, touch and keyboard input differ, comparisons are most meaningful on the same device and control method.

Generated locally and verified independently

The frozen source bank contains 90 project-generated trajectories, split evenly across the three levels. Runtime materialization applies symmetry transforms, object-order permutations and deterministic target combinations to create many replay configurations without downloading a remote level pack. A separate verifier reimplements path reflection and minimum-separation checks, validates all 90 sources, samples 6,000 Practice configurations and enumerates the frozen 2026 Daily schedule.

What the local statistics mean

DailyBrainArc stores attempts, exact rounds, Clean exact rounds, target hits, false picks, local adaptive speed, Clean best speed by level, recent rounds and a specialist Daily streak. These are direct summaries of your own browser sessions. They do not claim to measure a medical condition, intelligence or real-world attention capacity. Screen size, browser scheduling, pointer method and fatigue can all affect performance.

How this differs from Reaction Time, Aim Trainer and Schulte Table

Reaction Time measures how quickly you respond to one event. Aim Trainer combines target acquisition with pointer movement. Schulte Table asks you to scan a static field in a known sequence. Multiple Object Tracking is different because several identities must remain active while the whole visual field changes continuously. That makes it a genuine product gap rather than another skin over an existing speed task.

Where to start

Begin with Easy at 1.0× and aim for repeated Exact Clean rounds. Move to Standard before increasing speed if two-target tracking feels comfortable. Use 0.8× when the object count itself is overwhelming, and use Adaptive when you want the browser to nudge temporal difficulty around your current success boundary. If you use Peek, treat that round as study and focus on how your internal target configuration changed just before you lost one.