Components of Display Latency
Total delay before a stimulus appears on screen is the sum of multiple components. First, rendering delay: time from browser issuing draw commands to GPU writing to framebuffer, typically 1-5ms. Second, scanout delay: time for framebuffer contents to transfer to the monitor, dependent on refresh rate. At 60Hz, one frame equals 16.7ms; if a stimulus is generated mid-frame, up to 16.7ms additional delay occurs. At 144Hz this reduces to maximum 6.9ms, at 240Hz to 4.2ms. Third, panel response time: time for LCD pixels to reach target luminance; 4-8ms for IPS panels, 1-3ms for TN, under 0.1ms for OLED. Combined, a 60Hz IPS monitor can produce up to 25-30ms display latency. A 240Hz OLED monitor compresses this to 5-10ms. This 15-20ms difference directly reflects in reaction time test scores.
Input Delay and Polling Rate
Time from mouse click or keypress to PC recognition also affects measurements. USB device polling rate determines how frequently the PC checks device state. At 125Hz (standard), maximum 8ms input delay occurs; at 1000Hz (gaming devices), maximum 1ms. Bluetooth connections add further delay, potentially 10-30ms additional. Keyboard switch mechanisms also matter; mechanical switches with shallower actuation points (depth where contacts close) produce faster signal generation from physical press. Standard membrane keyboards have 3-4mm actuation distance, while gaming mechanical keyboards offer 1-1.5mm. This difference translates to 5-15ms when converted to press speed. Touchscreens have the largest delay, requiring 25-70ms from touch detection to signal processing.
Software Latency and Browser Impact
In web-based reaction time tests, the browser's JavaScript execution environment becomes an additional delay source. requestAnimationFrame timing precision varies by browser and OS; Chrome on Windows achieves approximately 1ms precision, but some environments produce 4-16ms jitter. Garbage collection (GC) timing can cause spikes of tens of milliseconds. Additionally, OS-level compositors (Windows DWM, macOS WindowServer) may perform framebuffering, inserting one frame of additional delay. These software delays can total 5-20ms. Bench tests employ designs minimizing these delays (high-precision timers, reduced GC pressure), but complete elimination is impossible. The key insight is that repeating measurements in the same environment causes hardware/software delays to cancel as constants, making only human performance changes trackable.
Cautions When Comparing Scores Across Environments
Directly comparing reaction time scores measured on different devices is scientifically inappropriate. Between a 60Hz monitor with Bluetooth keyboard and a 240Hz monitor with 1000Hz wired mouse, hardware delay alone can differ by 30-50ms. This represents 15-25% of reaction time, translating to 20-30 percentile points. The same person with identical ability can appear to move from 'average' to 'top 20%' purely from environmental differences. Since Bench's percentile display is calculated from all users' measurements, users with high-performance setups are statistically advantaged. Recognizing this limitation, scores should be interpreted as 'changes over time within the same environment,' with absolute percentile values treated as approximate references only.
Environment Settings to Maximize Measurement Precision
Environment settings to maximize reaction time test measurement precision, in priority order. First, use wired input devices. Bluetooth delay variation is the largest noise source for measurements. Second, close unnecessary background processes. Video playback, file sync, and virus scans particularly consume CPU/GPU resources, causing frame drops. Third, minimize browser tabs. Each tab consumes memory and CPU, increasing GC frequency. Fourth, enable monitor response speed mode (overdrive). However, excessive overdrive causes overshoot (inverse ghosting), so moderate settings are recommended. Fifth, disable OS animation effects. Windows transparency effects and macOS animations increase compositor load. With these settings unified, testing under identical conditions each time is the prerequisite for tracking true changes in cognitive performance.