All Blog

Human Reaction Time: Average Speed and App Latency

10 min read

Aug 24, 2026

TL;DR

The average human reaction time is about 250 milliseconds for a simple visual stimulus, although results vary by test, stimulus, attention, age, and hardware. Auditory and tactile responses are often faster. Elite performers may approach 100–120 ms in controlled situations, while results below 100 ms frequently involve anticipation rather than a completed conscious response. For application developers, this does not mean every system needs 13 ms latency; it means latency, jitter, rendering, and interaction design must fit the task.

How Quickly Humans Process Input—and What Reaction Time Means

Reaction time is the interval between the appearance of a stimulus and the beginning of a person's physical response. For example, a reaction-time test may measure the delay between a light changing color and the user pressing a button.

That reaction time definition sounds simple, but the result combines several steps:

A sensory organ detects the stimulus.
The nervous system transmits the signal.
The brain recognizes and interprets it.
The person selects a response.
Muscles begin the required movement.

The clock normally stops when the movement starts, not when it finishes. This distinction separates reaction time from movement time. Pressing a button after seeing a light may involve a 240 ms reaction followed by another short interval to complete the press.

It is also important not to confuse perception with reaction. Research has shown that people can identify the general meaning of certain images shown for only 13 ms, as summarized by MIT News. That finding concerns rapid visual processing. It does not mean a person can perceive an event, decide what to do, and execute a physical response in 13 ms.

Network latency is different again. A 40 ms network delay can become only one component of the total time between a remote action and a visible or audible response. Encoding, transport, buffering, decoding, display refresh, and human reaction all contribute to the experience.

Practical Human Reaction-Time Benchmarks

If you are asking, “What is the average human reaction time?”, the most useful general benchmark is approximately 250 ms for a simple visual reaction test.

Typical controlled-test ranges are:

Visual stimulus: commonly around 200–300 ms
Auditory stimulus: often around 150–250 ms
Tactile stimulus: often around 150–250 ms
Fast trained responses: approximately 100–200 ms, depending on the task
Complex choice responses: usually slower than simple reactions

These are reference ranges, not universal biological limits. A browser-based test also measures the latency of the screen, browser, operating system, mouse or touchscreen, and test code. Comparing results from different devices can therefore be misleading.

The fastest human reaction time is often described as roughly 100–120 ms for highly trained people under tightly controlled conditions. Results substantially below 100 ms should be treated cautiously because the participant may have anticipated the stimulus.

What is a good reaction time? For a simple online visual test:

Under 300 ms: within a common adult range
Around 200–250 ms: good for many people
Under 200 ms: notably fast
Near 100–120 ms: exceptional and highly dependent on testing conditions

A single score is less informative than repeated results. Use several trials, discard obvious accidental clicks, and compare the median rather than the fastest attempt.

Three Main Types of Reaction Time

Simple reaction time involves one known stimulus and one known response. A light appears, and the participant presses a key. This format usually produces the fastest scores.

Recognition reaction time requires the participant to respond only to a target stimulus. For example, press when the circle turns green but not when it turns red.

Choice reaction time presents multiple possible stimuli and responses. A driver deciding whether to brake, steer, or accelerate is performing a choice response. More alternatives and more complex decisions generally increase the delay.

This helps explain why a 250 ms visual-test result cannot predict exactly how quickly someone will react in traffic, a competitive game, or a live auction. Real tasks contain uncertainty, context, and consequences that a simple test removes.

Classic human-computer interaction guidance also distinguishes among system-response intervals. Feedback near 100 ms can feel immediate, while longer delays become progressively more noticeable and disruptive. These are experience-design guidelines, however, not hard limits that apply equally to every application.

What Changes a Person’s Reaction Time?

Human reaction time changes from moment to moment. The most important influences include:

Stimulus type: Auditory and tactile cues can produce faster reactions than visual cues because each sensory pathway processes input differently.
Stimulus intensity: A bright flash or loud sound is usually detected faster than a weak, ambiguous signal.
Response complexity: Pressing one button is faster than interpreting several options and selecting the correct action.
Expectation: Knowing when and where a stimulus will appear can improve the result, although too much predictability may turn reaction into anticipation.
Attention: Distraction and divided attention add processing time.
Fatigue and sleep: Tired users tend to respond more slowly and inconsistently.
Age: Reaction time generally develops through childhood, can be strongest in early adulthood, and may slow later in life.
Practice: Training improves stimulus recognition and response selection, particularly when the task is repeated consistently.
Physical and emotional state: Stress may sharpen a simple response in some situations but impair judgment and choice performance in others.
Device latency: Display refresh rate, input polling, Bluetooth accessories, browser scheduling, and touch processing can all alter an online score.

This last factor matters when evaluating a reaction-time app. If one user tests on a high-refresh-rate monitor with a wired mouse and another uses an older phone, their scores do not represent physiology alone.

For fairer comparisons, keep the device, browser, input method, posture, and testing environment constant. Measure enough trials to identify consistency as well as speed.

The Processes Behind a Complete Human Response

A measured response is the end of a chain, not one instantaneous reflex. The total can be understood as four broad phases.

Sensory acquisition: Light reaches the retina, sound reaches the inner ear, or pressure activates receptors in the skin.

Neural transmission and perception: Signals travel through the nervous system and become available for recognition.

Interpretation and response selection: The brain applies context, decides whether the stimulus matters, and selects an action.

Motor initiation: The nervous system sends an instruction to the relevant muscles, which begin moving.

A simple visual test minimizes interpretation and response selection. Real-world tasks do not. During a video conference, for example, a participant must hear a statement, understand its meaning, decide whether to speak, and then begin responding. Network delay sits on top of those human processes.

From Sensory Input to Eye, Brain, and Muscle Response

The frequently quoted 13 ms figure describes the brain's ability to extract meaning from a rapidly presented image. It is evidence of very fast visual processing, but it is not the fastest complete human reaction time.

A full voluntary response generally requires at least around 100 ms even under exceptional conditions. Most everyday responses take longer because the person must distinguish the stimulus, apply context, and avoid making the wrong move.

That distinction prevents two common errors:

Treating visual-recognition speed as a physical-response benchmark.
Assuming all application latency above 13 ms damages performance equally.

A user may not consciously identify a small increase in delay, yet repeated or variable delays can still affect timing. Conversely, a stable 80 ms delay may be perfectly acceptable for one interaction and inadequate for another. The task determines the threshold.

How Unwanted Latency Affects Human Performance

Latency does not replace human reaction time; it adds to the interaction loop. In a remote application, a simplified loop looks like this:

Remote event → capture → encode → network → decode → render → user perception → decision → physical response

If a remote event needs 150 ms to reach the screen and the user needs 250 ms to react, the earliest response may begin roughly 400 ms after the original event. Return-path delay can extend the complete closed loop further.

The impact depends on three properties:

Absolute delay: How long information takes to arrive
Jitter: How much that delay changes from moment to moment
Loss or corruption: Whether information arrives late, damaged, or not at all

Consistency can be as important as the lowest possible number. People can adapt to a modest fixed delay more easily than unpredictable timing. Jitter forces media applications to buffer more data, increasing delay to preserve smooth playback.

For voice communication, the ITU-T G.114 recommendation provides widely used guidance for one-way transmission time. Interactive audio becomes more difficult as delay increases because speakers interrupt each other, pause unnecessarily, or lose conversational rhythm.

Developers should therefore avoid reducing human perception to one universal threshold. Optimize the complete experience: capture delay, network routing, media processing, rendering, and feedback all matter.

What These Benchmarks Mean for Real-Time Developers

Different real-time applications need different latency budgets. The right question is not “Can humans detect one millisecond?” but “At what point does delay change behavior, fairness, comprehension, or control in this task?”

Online Gaming

Turn-based games may tolerate hundreds of milliseconds because players do not expect an immediate continuous response. Fast action games are less forgiving: delayed movement, voice cues, or hit confirmation can affect control and perceived fairness.

Genre matters. A 200 ms delay may be inconvenient in a strategy game but decisive in a competitive shooter. Developers should measure input-to-action and motion-to-photon delay rather than reporting network ping alone.

Bidding and Auctions

Auction users need prompt updates, but raw speed is only part of trust. The system must also provide authoritative timestamps, deterministic ordering, clear countdown behavior, and a fair policy for bids received near closing.

A visually fast interface that displays bids in the wrong order is worse than one with a small, consistent delay. Keep the server authoritative, synchronize state, and tell users when an action is accepted rather than relying only on optimistic UI feedback.

Real-Time Collaboration

In online classrooms, meetings, and collaborative workspaces, delays affect conversational turn-taking and attention. Choppy audio can be more disruptive than slightly reduced video quality because participants cannot follow instructions or recognize when another person has finished speaking.

The practical priority is usually to preserve intelligible audio, manage weak networks, and keep delay stable. Video resolution can adapt when bandwidth falls. The media layer should also expose quality data so the application can distinguish a user problem from a network or device problem.

This is where purpose-built infrastructure can reduce implementation risk. Tencent RTC Engine provides a cross-platform real-time audio and video core for calls, conferences, live streaming, and custom interactive experiences. Its global routing, weak-network resilience, monitoring, and media-processing capabilities address the part of the latency chain that application teams would otherwise need to build and operate themselves.

How to Build an App Like This with Tencent RTC

For a latency-sensitive classroom, meeting, coaching session, or other interactive audio/video experience, use Tencent RTC Engine through trtc-sdk-v5. The objective is not to make human reaction instantaneous. It is to minimize avoidable media delay and maintain a stable experience when network conditions change.

A practical architecture contains:

Web or native clients: Capture audio/video, publish local media, and render remote streams.
Application server: Authenticates users and generates short-lived userSig credentials. Secrets must never be embedded in browser code.
Tencent RTC Engine: Handles room-based real-time media transport, routing, weak-network adaptation, and stream delivery.
Application telemetry: Records join time, freezes, packet loss, and user-visible response milestones.

Follow the Tencent RTC Engine overview to select the appropriate SDK and understand its room and media model.

A focused web implementation can follow these steps:

Create a Tencent RTC application and obtain its sdkAppId.
Generate userSig securely on your application server.
Install trtc-sdk-v5 and create one client instance per session.
Register remote-stream events before entering the room.
Enter the room using a stable room ID and unique user ID.
Start local audio and video; this captures and publishes the local media.
Render available remote video streams and stop them when users leave.
Test on realistic mobile and weak-network conditions, not only local Wi-Fi.

import TRTC from "trtc-sdk-v5";

const trtc = TRTC.create();

trtc.on(TRTC.EVENT.REMOTE_VIDEO_AVAILABLE, async ({ userId, streamType }) => {
  await trtc.startRemoteVideo({
    userId,
    streamType,
    view: `remote-video-${userId}`
  });
});

trtc.on(TRTC.EVENT.REMOTE_VIDEO_UNAVAILABLE, ({ userId, streamType }) => {
  trtc.stopRemoteVideo({ userId, streamType });
});

await trtc.enterRoom({
  sdkAppId: 1400000000,
  userId: "participant-42",
  userSig: userSigFromYourServer,
  roomId: 9001,
  scene: "rtc"
});

// Capture and publish local media after permission is granted.
await trtc.startLocalAudio();
await trtc.startLocalVideo({ view: "local-video" });

Do not evaluate the implementation using room-entry time alone. Measure the intervals users actually experience: action to remote playback, speech to remote audio, freeze duration, and recovery after bandwidth loss. Separate network transport time from browser rendering and human response time so optimization work targets the real bottleneck.

FAQ

What is the average human reaction time?

A simple visual reaction averages roughly 250 ms. Auditory and tactile responses are often faster, while choice-based tasks are usually slower.

What is considered a good reaction time?

Around 200–250 ms is a good result for many adults on a simple visual test. Compare repeated trials on the same device because hardware and browser latency affect the score.

Is 13 ms the fastest possible human reaction time?

No. Thirteen milliseconds refers to rapid visual information processing, not a complete physical response. Exceptional full reactions are more commonly associated with approximately 100–120 ms under controlled conditions.