Go/no-go test: read stops and response times

What a go/no-go round asks you to do

Press Space, or the answer pad, for the green lamp with a dot. Do nothing for the orange lamp with a bar. The shape carries the instruction as well as the color. During the dark gap, wait for the next signal rather than predicting when it will appear.

Six practice signals come first and do not count towards the result. The scored go/no-go round has 80 signals at levels 1 and 2, and 100 at level 3. The signals to hold back on number 20, 16 and 15 respectively. The response window also becomes shorter: 800 ms, 500 ms and 400 ms.

These settings matter. A round with rarer no-go signals and a shorter response window is not the same task with a different label. Keep the level and input method the same when comparing results.

Four numbers that need each other

• Answered in time: go signals given a valid response. In the arrow-based stop-signal mode, the response must also point the right way.
• Correct stops: signals on which you correctly made no response.
• False presses: responses on signals that required you to hold back.
• Missed go signals: signals that needed an answer but did not receive a valid one.

Median go time is calculated from correctly answered go signals. It leaves incorrect and missing responses out of the timing, so a quick median can sit beside a poor completion count. Read the counts before deciding what the time means.

Doing nothing for the whole round would produce many correct stops but no go answers. Pressing at every signal would do the reverse. Neither is following the rule. Likewise, answering more slowly may avoid false presses while missing go deadlines. There is no single number here that makes those trade-offs disappear.

This version sets aside a trial if its press arrives before the signal or within 150 ms of it, and reports it separately. That is a scoring rule for this practice, not a universal boundary of human reaction time. A late press cannot rescue a missed go signal; on a hold signal, a press just after the lamp goes out still counts against holding back.

Stop signal is a different round

In Stop signal, every trial begins with a left or right arrow. Answer in that direction unless an orange stop signal appears afterwards. Go/no-go tells you which response rule applies at the start. Stop signal sometimes asks you to cancel a response already being prepared.

The longer round contains 192 arrows in four blocks of 48, with a stop signal on a quarter of them. Each arrow allows 700 ms at every level. The stop delay starts at 250 ms, lengthens by 50 ms after a successful stop and shortens after a failed stop, within the practice's limits. It adjusts the task rather than rewarding you with an easier next trial.

Where the research fits

Go/no-go is a laboratory task format, not a single fixed questionnaire. Nieuwenhuis and colleagues (2003) varied the frequency of go and no-go trials while recording electrical brain responses. Their results challenged a simple interpretation of one signal as a pure marker of inhibition. Frequency and response conflict were part of the explanation too. A browser error count is certainly not a reading of that brain signal.

Logan, Cowan and Davis (1984) studied stopping in reaction-time experiments and modelled a race between going and stopping. If going finishes first, a response escapes; if stopping finishes first, no response is made. The model allows a stopping time to be estimated even though a successful stop produces no button press to time directly.

The 2019 stop-signal consensus guide by Verbruggen and colleagues recommends an integration estimate, checks on failed-stop response times, and care when response rates approach either extreme. It also warns that enough trials for group comparisons may still be insufficient for reliable individual estimates.

Why stop-signal time may say “Not measured”

This practice estimates stop-signal time from the go-time distribution and the delays actually presented. It does not show the estimate if there are too few go responses, if the proportion of stop signals pressed through is outside 25 to 75 percent, or if the failed-stop responses are not faster on average than the go responses. Other invalid combinations are also refused.

A missing estimate is not a score of zero and not a judgment about you. The observed counts remain useful. Even when an estimate is shown, this browser practice does not establish a laboratory-quality individual measurement. Its 48 scheduled stop trials, your device and the way you approached the round all belong in that interpretation.

How to approach a round

Use the six practice signals to learn the controls and the two shapes. Then answer go signals promptly while keeping the hold rule in mind. In Stop signal, do not wait deliberately to see whether a stop cue appears; the task asks you to begin the go response and cancel it if needed.

Afterwards, look for what changed together. A faster median with more false presses describes a different balance, not an unqualified better result. Keep keyboard and touchscreen rounds separate, and compare the same mode and level. Finished rounds can be saved locally; ending early saves no completed result.

The Hold the press lesson explains the two rules and the trade-off. Open the Go/no-go practice to try them. For responding without a no-go choice, see the reaction time guide.