Written and maintained by CASRAI Editorial Board
Published
On this page
The go/no-go task is a response-inhibition paradigm in which participants respond quickly to frequent “go” stimuli and withhold the response on infrequent “no-go” stimuli. It is scored mainly by the commission error rate (responses on no-go trials), the omission error rate (missed go trials) and the reaction time on correct go trials. Because the go response is made prepotent, a commission error is read as a failure of restraint.
What is the go/no-go task?
In the go/no-go task (GNG) participants respond to one class of stimuli (go) and withhold the response to another class (no-go). Response inhibition is defined as the ability to suppress a prepotent motor response, and commission errors on no-go trials are a frequent index of it, often treated as a marker of impulsive action. Omission errors are the opposite failure: no response on a go trial. Both error types and the go reaction time are recorded for every participant.
Design choices matter more than the formula. In a 213-student study the stimuli lasted 250 ms, 70% of trials were go and 30% no-go, and four blocks of 120 trials differed only in inter-stimulus interval (ISI: 400, 600, 800 or 1000 ms), after 10 practice trials with equal go and no-go numbers. Shorter ISIs produced more commission errors, and at 400 ms also more go omissions and more error variance.
Go/no-go task vs stop-signal task
The go/no-go task measures restraint: refraining from a response that has not yet been launched. The stop-signal task measures cancellation: stopping a response that is already under way, summarised as the stop-signal reaction time (SSRT). The consensus guide on the stop-signal task notes that stop-signal trials where the go and stop signals appear simultaneously resemble no-go trials, with trigger failures analogous to false alarms on no-go trials, and that such trials are thought to measure restraint rather than cancellation.
Stop-signal design numbers worth knowing
The same consensus guide gives concrete recommendations. Use a standard two-choice reaction-time go task, because very simple go tasks make the go and stop processes overlap and violate the race model. Present stop signals on a minority of trials, with 25% recommended for standard studies. Use a tracking procedure in which the stop-signal delay (SSD) rises after each successful stop and falls after each failed stop, so that the probability of responding given a stop signal converges on about 0.50; a 50 ms step (three refreshes at 60 Hz) is suggested. Plan at least 50 stop trials for group comparisons (about 200 trials in total at 25% stop signals, roughly 7 to 10 minutes), because 25 stop trials are rarely reliable. The data table below sets these values beside the go/no-go design reported above.
How the go/no-go task is scored
Each participant yields four counts: correct go responses (hits), omissions (go trials with no response), commission errors (no-go trials with a response, also called false alarms) and correct withholds. From them:
- Commission error rate = commission errors ÷ number of no-go trials.
- Omission error rate = omissions ÷ number of go trials.
- Mean go RT = mean reaction time of correct go responses only.
Every symbol is a count of trials in one cell of the design: the denominator of the commission rate is the number of no-go trials, not the total number of trials. Report both rates because a participant who simply responds less often lowers commission errors by raising omissions.
Worked example with the arithmetic
The numbers below are an illustrative calculation, not data from a study. A participant completes 200 trials with 70% go and 30% no-go, which is 140 go trials and 60 no-go trials. They respond to 133 go trials and withhold on 7 (omissions). They respond on 9 no-go trials (commissions) and withhold on 51.
- Commission error rate = 9 ÷ 60 = 0.150, or 15.0%.
- Omission error rate = 7 ÷ 140 = 0.050, or 5.0%.
- Check: 133 + 7 = 140 go trials, and 9 + 51 = 60 no-go trials.
- Mean go RT over the 133 correct go responses: if the RTs sum to 54,929 ms, the mean is 54,929 ÷ 133 = 413 ms.
Worked example: stop-signal reaction time by the integration method
For the stop-signal task, the consensus guide recommends the integration method with go omissions replaced by the maximum RT. With 200 go trials and a probability of responding given a stop signal of 0.45, the nth go RT is the 90th fastest (200 × 0.45 = 90). SSRT = nth RT − mean SSD. If, hypothetically, the 90th fastest go RT is 480 ms and the mean SSD is 250 ms, SSRT = 480 − 250 = 230 ms. The simpler mean method (mean go RT minus mean SSD) is biased and less reliable, so it is not recommended.
Interpretation: reading the scores
| Pattern | Reading |
|---|---|
| Higher commission rate, normal omission rate | More failures to withhold, the pattern read as poorer response inhibition (impulsive action). |
| Low commission rate with high omission rate | Possible overly cautious responding rather than good inhibition; omissions need reporting alongside commissions. |
| Commission errors rising with a very short ISI | The ISI itself drives errors. In the 213-participant study, commission errors at extremely short ISIs (400 ms) were judged possibly inappropriate for assessing inhibition. |
| Faster go RT with more commission errors | A speed-accuracy trade-off is possible; inspect go RT alongside errors. |
Assumptions and common mistakes
- Treating the rate as a pure inhibition score. Commission errors depend on the go/no-go ratio, stimulus duration and ISI, so scores are comparable only across identical task settings.
- Dividing by the wrong denominator. Use no-go trials for commissions and go trials for omissions.
- Including errors in the go RT. Compute go RT from correct go trials only.
- Calling a go/no-go result an SSRT. SSRT belongs to the stop-signal task. In that task, the consensus guide advises checking independence (compare mean RT of unsuccessful stop trials with mean go RT) and not estimating SSRT when the probability of responding given a signal departs substantially from 0.50 or go omissions are high.
- Silent exclusions. Report any excluded participants or trials and the criteria.
The task belongs to the same family of speeded choice paradigms as the flanker task, the lexical decision task and the attentional blink, and its reaction-time handling resembles semantic priming studies.
How to report the go/no-go task (APA example)
State the design (number of trials, go/no-go ratio, stimulus duration, ISI), the dependent variables and the exclusions, then the descriptive values. A template using the example above:
The task comprised 200 trials (70% go, 30% no-go; stimulus duration 250 ms). The participant made commission errors on 15.0% of no-go trials (9 of 60) and omission errors on 5.0% of go trials (7 of 140); mean go RT was 413 ms.
For group analyses, give the mean and standard deviation of each rate per condition, followed by the test statistic in APA format, for example t(df) = value, p = value, with an effect size.
How to calculate it in Excel, R or Python
Each snippet computes the three scores from the example counts.
Excel
Commission rate: =B2/B3 (B2 = commission errors, B3 = no-go trials) Omission rate: =B4/B5 (B4 = omissions, B5 = go trials) Mean go RT: =AVERAGEIFS(D:D, C:C, "go", E:E, "correct")
R
commission_rate = 9 / 60
omission_rate = 7 / 140
# trials: data frame with columns type ("go"/"nogo"), correct (TRUE/FALSE), rt
go = trials[trials$type == "go", ]
mean_go_rt = mean(go$rt[go$correct])
Python
commission_rate = 9 / 60 # 0.15 omission_rate = 7 / 140 # 0.05 go_rts = [rt for t, ok, rt in trials if t == "go" and ok] mean_go_rt = sum(go_rts) / len(go_rts)
Frequently asked questions
What does a commission error mean in a go/no-go task?
A commission error is a response on a no-go trial. It is the main index of failed response inhibition, or impulsive action, in the task.
What is the difference between a go/no-go task and a stop signal task?
The go/no-go task measures restraint, withholding a response before it starts. The stop-signal task measures cancellation of an ongoing response and yields SSRT, estimated with a tracking procedure on the stop-signal delay.
What go/no-go ratio is typical?
The ISI study above used 70% go and 30% no-go trials. Ratios change error rates, so report the ratio you used.
Does the inter-stimulus interval change the result?
Yes. In that study shorter ISIs raised commission errors, and the 400 ms condition also raised omissions and error variance.
Sources
- Verbruggen, F., et al. (2019). A consensus guide to capturing the ability to inhibit actions and impulsive behaviors in the stop-signal task. eLife, 8, e46323. doi.org/10.7554/eLife.46323 (read 2026-11-15).
- Do shorter inter-stimulus intervals in the go/no-go task enable better assessment of response inhibition? Scandinavian Journal of Psychology, PMC8048576. pmc.ncbi.nlm.nih.gov/articles/PMC8048576 (read 2026-11-15).
Data: Go/no-go and stop-signal task design parameters reported in two open-access sources
Design values for the go/no-go task (213-participant ISI study) and recommended stop-signal task settings (consensus guide), compiled from two openly readable papers.
| Design parameter | Task | Value | Unit | Source |
|---|---|---|---|---|
| Proportion of go trials | Go/no-go (study) | 70 | percent | Scand J Psychol study, PMC8048576 |
| Stimulus duration | Go/no-go (study) | 250 | ms | Scand J Psychol study, PMC8048576 |
| Longest ISI condition | Go/no-go (study) | 1000 | ms | Scand J Psychol study, PMC8048576 |
| Shortest ISI condition | Go/no-go (study) | 400 | ms | Scand J Psychol study, PMC8048576 |
| Trials per block | Go/no-go (study) | 120 | trials | Scand J Psychol study, PMC8048576 |
| Practice trials | Go/no-go (study) | 10 | trials | Scand J Psychol study, PMC8048576 |
| Proportion of stop-signal trials | Stop-signal (recommended) | 25 | percent | Verbruggen et al. 2019, eLife |
| Tracking step size for SSD | Stop-signal (recommended) | 50 | ms | Verbruggen et al. 2019, eLife |
| Target probability of responding given stop signal | Stop-signal (recommended) | 0.5 | probability | Verbruggen et al. 2019, eLife |
| Minimum stop trials for group comparisons | Stop-signal (recommended) | 50 | trials | Verbruggen et al. 2019, eLife |
Sources
- A consensus guide to capturing the ability to inhibit actions and impulsive behaviors in the stop-signal task (accessed 2026-11-15)
- Do shorter inter-stimulus intervals in the go/no-go task enable better assessment of response inhibition? (accessed 2026-11-15)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







