Go/no-go task vs stop-signal task: how the two differ
A go/no-go task shows one stimulus at a time and asks you to respond to some and hold back on others. A stop-signal task asks for a response on every trial, then on some trials sends a signal to cancel it. The first tests a decision made when the stimulus appears. The second tests stopping a response already begun.
The game above is Go / No-Go. Its sibling is Stop Signal. Both are built on these two laboratory tasks, and the numbers below come from their code.
What is a go/no-go task?
A go/no-go task presents a single stimulus per trial, and the stimulus itself tells you what to do. One kind means respond. Any other kind means do nothing.
The design is old. Franciscus Donders used it in the reaction time experiments he ran in the 1860s. He compared three methods. In method a, a light comes on and you press a button. In method b, there are two lights and two buttons, and you press the one that matches. In method c, you press the button when the target light comes on and do nothing when the other light does. Method c is a go/no-go task.
Two kinds of error are possible: responding to a no-go stimulus and failing to respond to a go stimulus.
What is the stop-signal task?
The stop-signal task gives you a go task on every trial and, on a minority of trials, follows the go stimulus with a stop signal that tells you to withhold the response. The time between the go stimulus and the stop signal is called the stop-signal delay.
Logan and Cowan gave the task its standard theory in 1984, in a Psychological Review paper titled “On the ability to inhibit thought and action: A theory of an act of control”. They described performance as a race between two processes. A go process starts when the go stimulus appears, and a stop process starts when the stop signal appears. Whichever finishes first decides whether you respond.
The model matters because the speed of stopping cannot be measured directly. A successful stop produces no response to time. A 2019 consensus guide in eLife, written by Verbruggen and a large group of stop-signal researchers, puts it this way: unlike the latency of go responses, the latency of response inhibition cannot be observed directly. It has to be estimated, and the estimate is called the stop-signal reaction time.
The same guide gives the settings most laboratories use. It recommends stop signals on 25 percent of trials. It describes a tracking procedure in which the stop-signal delay gets longer after each successful stop and shorter after each failed one, and notes that many studies move it in steps of 50 milliseconds. Tracked this way, a person ends up responding on about half of the stop trials. The guide also says participants should be told to respond quickly and not to wait for the stop signal.
How do the two tasks differ?
They differ in when you learn that you should not respond. In go/no-go you know as soon as the stimulus appears. In the stop-signal task you find out after you have started.
| Go/no-go task | Stop-signal task | |
|---|---|---|
| Signals per trial | One | One, or two on stop trials |
| When you learn not to respond | When the stimulus appears | After a delay |
| What a stimulus means | The same thing every time | The same arrow can be a go trial or a stop trial |
| Main errors | Responding to a no-go, missing a go | Responding on a stop trial |
| What is adjusted | Usually nothing | The stop-signal delay |
The third row has a consequence that Verbruggen and Logan pointed out in a 2008 review. In go/no-go, a given stimulus is paired with stopping every time it appears. In the stop-signal task, stimuli are paired with going and stopping inconsistently. They wrote that automatic inhibition is more likely to develop in the go/no-go task, where the link between a stimulus and stopping can be learned.
How hard a go/no-go task makes it to hold back also depends on its settings. Wessel reported in 2018 that only fast go/no-go tasks with rare no-go trials reliably produced the motor activity that then has to be stopped. In slow versions with equal numbers of go and no-go trials, he measured about 75 percent less inhibition-related brain activity. His survey of the literature found that about 40 percent of go/no-go studies used equal numbers of the two trial types.
How does our Go / No-Go game work?
One shape at a time appears, and you tap the pad for a blue circle and do nothing for anything else. A round lasts 30 seconds. Each blue circle tapped while it is on screen is one point.
These are the rules in the code:
- The first two shapes are always blue circles.
- After that, each shape has a 31 percent chance of being a no-go, with two exceptions. The shape after a no-go is always a blue circle, and after six blue circles in a row the next shape is always a no-go.
- The pause between shapes is empty and lasts between 280 and 460 milliseconds.
- The pad answers on touch-down. A tap during the pause counts as a mistake unless it lands within 220 milliseconds of the last shape leaving, in which case it is ignored as a late answer.
We generated 5,000 rounds with a short program to see what share of shapes those rules produce. Across 600,000 shapes, 25.4 percent were no-go trials. The longest run of blue circles was six and no round had two no-go shapes in a row.
That puts the game on the fast, rare no-go side of the range Wessel described: one no-go in four, and a shape on screen for less than a second.
How long a shape stays up depends on a step number, which is your level plus one for every eight shapes shown in the round. Your level starts at 0, rises after two rounds in a row with at least 80 percent accuracy and falls after a round under 50 percent.
| Step | Shape stays up | No-go shapes |
|---|---|---|
| 0 | 900 ms | Red square only |
| 1 | 865 ms | Red square only |
| 2 | 830 ms | Red square only |
| 3 | 795 ms | Any of five |
| 5 | 725 ms | Any of five |
| 7 | 655 ms | Any of five |
| 9 | 585 ms | Any of five |
| 11 and up | 520 ms | Any of five |
The display time drops by 35 milliseconds per step. From step 3 the no-go shape can be a red square, a green triangle, a yellow diamond, a purple star or a black cross.
How does our Stop Signal game work?
An arrow appears and you tap left or right to match it before a deadline. On two arrows in every block of eight, a red stop sign replaces the arrow after a delay, and the right move is to tap nothing until the deadline passes. A round lasts 40 seconds.
Two stop trials in eight is the 25 percent share that the consensus guide recommends. Two stop trials never come back to back, and a round opens with two plain arrows. The empty pause before each arrow varies between 360 and 600 milliseconds so that its arrival cannot be timed.
The delay between the arrow and the stop sign follows the same tracking rule the guide describes. Each stop you hold makes the next sign 50 milliseconds later. Each stop trial you tap on makes the next one 50 milliseconds earlier. The delay never goes below 60 milliseconds or above the deadline minus 350 milliseconds, and it starts fresh each round.
| Level | Deadline for an answer | Delay at start of round | Longest delay allowed |
|---|---|---|---|
| 0 | 1,100 ms | 180 ms | 750 ms |
| 2 | 1,020 ms | 200 ms | 670 ms |
| 4 | 940 ms | 220 ms | 590 ms |
| 6 | 860 ms | 240 ms | 510 ms |
| 8 | 780 ms | 240 ms | 430 ms |
| 10 and up | 700 ms | 240 ms | 350 ms |
The deadline shortens by 40 milliseconds per level and the starting delay grows by 10 milliseconds per level up to level 6.
One difference from the laboratory task: a tap on a stop trial counts as a failed stop even if it lands before the sign has appeared. The game then shows the sign so you can see what the trial was.
What follows from these rules when you play?
In Go / No-Go, the sequence rules can be used.
- After you hold back on a shape, the next one is a blue circle, so you can tap as soon as it appears.
- If you have tapped six blue circles in a row, the seventh shape is not a blue circle.
- Keep your finger off the pad during pauses. The 220 millisecond allowance only covers a tap that was aimed at a shape and arrived late.
In Stop Signal, waiting does not pay.
- An arrow left unanswered at the deadline is a mistake, and three arrows in four are plain ones.
- If you delay every answer to see if a sign comes, you will hold the early stops. Each one pushes the next sign 50 milliseconds later, until the sign arrives after the moment you have been waiting for.
- The sign always leaves at least 350 milliseconds before the deadline. A stop is held by not tapping for that remaining time.
A score in either game is a count of correct responses in one round. It is not a measurement of you, and it is not the stop-signal reaction time from the research, which needs a controlled setup to estimate.
For related reading, see what is a normal reaction time and what the research shows about brain training games. Arrow Focus and Opposite Day are two more games about holding back a ready answer.
Common questions
What is a go/no-go task?
It is a task where one stimulus appears at a time. You respond as fast as you can to the go stimulus and make no response to any other. Mistakes are responses to a no-go stimulus and missed go stimuli.
What is the difference between a go/no-go task and a stop-signal task?
In a go/no-go task you know from the moment the stimulus appears if you should respond. In a stop-signal task every trial starts as a go trial, and on a minority of trials a stop signal arrives shortly afterward, so a response that is already under way has to be cancelled.
What is stop-signal reaction time?
It is an estimate of how long the stopping process takes in a stop-signal task. It cannot be observed directly, because a successful stop produces no response to time, so it is calculated from the go reaction times and the share of stop trials on which the person responded.
Sources
- Radboud University, Donders Institute: Biography of F.C. Donders (the a, b and c reaction time methods)
- Verbruggen, F., & Logan, G. D. (2008). Response inhibition in the stop-signal paradigm. Trends in Cognitive Sciences, 12(11), 418-424.
- 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.
- Wessel, J. R. (2018). Prepotent motor activity and inhibitory control demands in different variants of the go/no-go paradigm. Psychophysiology, 55(3), e12871.