The new behavior generator is an NLP technique for building a response before you need it: you watch yourself performing the behaviour you want on a mental screen, edit that film until it convinces you, then step inside it and run it from the inside. It was developed by Robert Dilts as part of his work on strategies, and it is the NLP format closest to what athletes and musicians call mental rehearsal.
It is used for behaviour you can already describe but cannot yet produce under pressure: speaking up in a meeting after the moment has passed, staying level when a teenager pushes back, giving feedback without softening it into nothing. The technique adds no motivation. It adds a rehearsed route, so the moment arrives with something already in it.
The short version
| What it is | A rehearsal format for installing a specific new behaviour before the situation arrives |
| Who created it | Robert Dilts, building on the NLP work on modelling and strategies |
| What it is for | One behaviour in one context, not a personality trait |
| How long it takes | 10 to 15 minutes once you know the sequence |
| What you need | A clear trigger moment, and a model to borrow from if you cannot picture yourself doing it |
| Where it fails | Vague goals, a skipped ecology check, and a rehearsal never attached to a real cue |
Why would rehearsing something in your head change what you do?
Because mental practice produces measurable performance gains, and because most behaviour under pressure is automatic rather than chosen. If the only version of a difficult conversation you have ever run is the one where you go quiet, that is the version your nervous system reaches for.
The best evidence on this comes from outside NLP. A meta analysis by James Driskell, Carolyn Copper and Aidan Moran, published in the Journal of Applied Psychology in 1994, pooled decades of studies on mental practice and found a reliable positive effect on performance. Three details from that work matter here. The effect is smaller than actual physical practice. It is stronger for tasks with a large cognitive component, which is most workplace and relationship behaviour. And it fades over weeks unless it is refreshed, which is why one rehearsal is rarely enough.
The second strand is observational learning, mapped by Albert Bandura from the 1960s onwards: people acquire complex behaviour by watching others perform it, not only by being rewarded for their own attempts. Borrowing a model in step three of this technique leans directly on that.
Be clear about what this does and does not prove. Both bodies of research concern mental practice and modelling in general. Neither tested the new behavior generator as a named format, and no controlled trial of it exists. What you can reasonably say is that the technique is assembled from ingredients that have held up elsewhere.
How do you run the new behavior generator step by step?
In six steps, and the order matters more than the polish. Give it a quiet ten minutes and a real situation coming up in your calendar, not a hypothetical one.
- Name the behaviour and its moment. Not “be more assertive”. Something like “when my colleague talks over me in the Monday review, I finish my sentence in a steady voice”. If you cannot name the second at which it starts, you have a wish rather than a target.
- Watch yourself do it, from outside. Put a screen a few metres in front of you and see yourself, from the back or the side, performing the new behaviour in that exact setting. Staying outside the picture at this stage is deliberate: it lets you judge the performance without the old feeling flooding back.
- Borrow a model if the screen stays blank. Put someone who does this well into the picture instead, real or fictional, and watch them handle the same moment. Then swap their face and body for yours, keeping the timing and the posture. This is the step people skip, and it is the one that rescues the technique when imagination refuses to cooperate.
- Edit until it is credible. Slow it down, change the words, lower the voice, take out the part where you over explain. Run it again. You are looking for a version you would actually believe if you watched a colleague do it, not an idealised one.
- Step into the film and run it from inside. Now associate: see through your own eyes, hear your own voice, feel the posture. This is where the rehearsal stops being a plan and starts being an experience. Run it two or three times.
- Check the ecology, then future pace it. Ask whether anything in you objects to behaving this way, and take objections seriously rather than overriding them. Then attach the rehearsal to its real cue: picture the actual trigger, the colleague speaking over you, and let the new behaviour follow it. Without this last move you own a nice film and no reflex.
Steps two and five are not interchangeable, and running only one of them is the most common way people water the technique down. The outside view gives you editorial distance. The inside view gives your body the reference. Both, in that order. The closing move is the one covered in our glossary entry on future pacing, and the objection check is the ecology check that runs through every well built NLP intervention.
How is it different from modelling and the swish pattern?
Different starting points. The new behavior generator assumes you know what you want to do and cannot yet do it. Modelling assumes someone else can do something and you want the structure behind it. The swish pattern assumes a cue reliably sets off a behaviour you do not want.
| New behavior generator | Modelling | Swish pattern | |
|---|---|---|---|
| Starting point | You know the behaviour you want | Someone else performs it well | A trigger fires an unwanted behaviour |
| What you work with | A rehearsed film of yourself | Another person’s strategy, beliefs and physiology | The submodalities of two images |
| Typical time | 10 to 15 minutes | Hours to weeks | Around 5 minutes |
| Main risk | Rehearsing a version you will never use | Copying the surface and missing the structure | Running it too slowly to build a reflex |
In practice the two often pair up. If a habit keeps hijacking the moment, clear the cue first with the swish pattern, which we walk through in detail in how to use the swish pattern for habits, then rehearse the replacement with this technique. Removing a behaviour without installing one leaves an empty second, and empty seconds get filled by whatever used to live there.
Where does the new behavior generator fall short?
In three predictable places, and knowing them saves you from concluding that the whole thing is theatre.
First, when the goal is a trait. “More confident”, “less reactive” and “a better listener” cannot be filmed, so there is nothing to rehearse. Convert them into a moment and an action, or the technique has no material to work on.
Second, when you rehearse a stranger. If the edited version of you speaks in a register you have never used, some part of you will decline to perform it when the moment arrives, and you will read that refusal as proof that NLP does not work. The fix is to lower the ambition of the film until it is plainly you, on a good day, rather than a different person.
Third, when the behaviour is not really the problem. If avoiding a conversation protects you from something you have not looked at, rehearsal alone will not move it. That is a job for a technique that asks what the avoidance is doing for you, not a job for a better film.
One more limit worth naming plainly. Rehearsal is a self management tool. Where the difficulty is anxiety, panic, phobia or the aftermath of trauma, this complements professional care and does not replace it, and a technique that asks you to relive the trigger repeatedly is not the right first move without support.
What I see when people run this in training
Two things, almost every time. The first is that people edit the film to make themselves impressive rather than credible, and the impressive version is the one that collapses on contact with a real Monday. When they cut it back until it looks ordinary, it survives. The second is that the future pacing step gets treated as a formality, a quick nod before opening the eyes. Run properly it takes a minute and it is the difference between a pleasant visualisation and a behaviour that shows up on its own. The technique is not demanding. It is just easy to do at eighty per cent, and at eighty per cent it does nothing.
Carolin Mallmann, Licensed Trainer of The Society of NLP, Lisbon
The new behavior generator is one of the formats practised in our seven day NLP Practitioner certification in Lisbon, where you run it on something happening in your actual week, with someone watching who can tell you when the film stopped looking like you.
Frequently asked questions
Who created the new behavior generator?
Robert Dilts, one of the developers of NLP who worked on strategies, logical levels and modelling. The format grew out of the early NLP interest in how people build new skills by observing and rehearsing rather than by being instructed. It has been taught in Practitioner trainings in various versions since the 1980s.
How is this different from simply visualising success?
Visualising success usually means picturing the outcome: the applause, the signed contract, the calm afterwards. The new behavior generator rehearses the process instead, second by second, including the awkward opening line. Research on mental practice consistently finds process rehearsal more useful than outcome fantasy, and some studies suggest that dwelling on the outcome alone can reduce the effort people put in.
Do I need clear mental pictures for this to work?
No. Some people see detailed images, others get a vague sense of shape, and a minority see nothing at all. If pictures are faint, run the technique through sound and feeling instead: hear the tone of voice you want, feel the posture and the pace of your breathing. The rehearsal needs a representation, not a high definition one.
How many times should I run it?
Two or three repetitions in the first sitting, then a short refresh before the situation itself. The meta analytic finding that mental practice effects decay over weeks argues against treating it as a one off. Thirty seconds of the associated version, run in the corridor before the meeting, is usually enough to bring it back.
Can I use it for something I have never done before?
Yes, and that is precisely what the borrowed model step is for. Watch someone who does it well, keep their timing and structure, then substitute yourself. What you cannot do is rehearse a skill you do not yet possess, such as a language or a technical procedure. Rehearsal shapes how you deploy an ability. It does not manufacture one.
Is the new behavior generator the same as future pacing?
No. Future pacing is a single move that links a change to the situation where it needs to appear, and it is used at the end of many different NLP techniques. The new behavior generator is a complete six step format that happens to finish with future pacing. Think of one as the last step of the other.
Keep exploring
- How to use the swish pattern for habits, for the technique that clears the old cue before you rehearse the replacement.
- Six step reframing in NLP, if the behaviour you are trying to replace keeps returning because it is quietly doing a job.
Carolin Mallmann is a Licensed Trainer of The Society of NLP, trained by Richard Bandler. She has trained more than 300 people and runs English language NLP trainings in Lisbon in groups of no more than 10.
