Ambivalence in Motivational Interviewing

Ambivalence is the state of wanting two incompatible things at the same time: to change, and to stay as you are. Motivational Interviewing treats it as the normal condition of a person facing a hard change, not as denial, poor insight or a missing motivation that the clinician has to supply. The practitioner’s job is not to break the deadlock by arguing for the change side, because arguing for one half of a conflict reliably recruits the other. It is to help the person hear themselves work it out.

That last point is where most of the practical difficulty sits, and where a widely taught tool turns out to do the opposite of what it promises.

Why ambivalence is the normal state

Someone who feels no pull in either direction is not ambivalent. They are either not considering the change at all, or they have already decided. Ambivalence sits between those, and its presence usually means the person is closer to change than the person who feels nothing.

This is worth holding onto, because ambivalence is easy to read as obstruction. A client who says they want to stop and then explains why they cannot is not being evasive. They are reporting both halves of an accurate internal picture. Miller and Rollnick give this its own chapter in the fourth edition of Motivational Interviewing (2023), on responding to sustain talk and discord, rather than treating either as an obstacle to get past before the real work starts.

What ambivalence sounds like

Ambivalence is a state. Change talk and sustain talk are how that state surfaces in speech. Keeping those separate makes sessions easier to read, because you stop trying to diagnose a person and start listening to sentences.

Most of the time both halves arrive in a single breath, joined by “but”:

“I know I should cut down. But it’s the only thing that switches my head off at night.”

Note the order. The change half came first and the sustain half came last, which is the usual pattern, and the half a person says last is the one they hear themselves commit to. That is the whole reason a double-sided reflection ends on the change side rather than the other way round.

Is it ambivalence, or something else?

Four different things look like ambivalence from the outside and need different responses. Getting this wrong is the most common reason a session stalls.

What you are seeing	What it usually is	What it needs
”I want to, and I don’t”	Ambivalence	Evoke both sides, reflect, end on the change side
The client pushing back against you	Discord in the relationship	Repair the alliance before anything else
”That’s not really what I came here about”	A focus problem, not a motivation one	Renegotiate the agenda
”I’d like to, I just know I can’t”	Low confidence, not low desire	The confidence ruler, and ability change talk
Flat, no pull either way	Not yet contemplating, or already decided	Do not manufacture a conflict

The second row is the one that costs most. Discord is dissonance between the client and the practitioner; ambivalence is a conflict inside the client. Evoking harder in the face of discord makes it worse, because you are adding pressure to a relationship that is already signalling too much of it. Agenda mapping and the rulers handle rows three and four.

Why a pros and cons list can backfire

The standard tool taught for ambivalence is decisional balance: work through the advantages and disadvantages of changing and of staying the same, weighted evenly, without steering. It is intuitive, it is on most MI handouts, and it is the technique several of the highest-ranking pages on this subject still recommend as the default.

The evidence does not support using it that way.

Miller and Rose set out the case in Behavioural and Cognitive Psychotherapy in 2015, comparing decisional balance against MI’s evocation of change talk as two different responses to the same client state. Their conclusion is that decisional balance is “both theoretically and empirically contraindicated with ambivalent people” when the goal is change. The mechanism is not mysterious. Asking someone to articulate the case for the status quo, carefully and at length, gives them practice in saying it.

The supporting findings are specific. Nenkov and Gollwitzer found in 2012 that among people who had not yet decided, decisional balance procedures significantly decreased goal commitment. Carey and colleagues added decisional balance to a motivational intervention with student drinkers in 2006 and found no effect against controls, noting that one unintended result was to remind students of what they got out of drinking. Magill and colleagues, analysing session content against outcomes, found that a focus on ambivalence predicted worse drinking results for outpatients.

So when is it right? When you have genuinely chosen not to favour a direction. Miller and Rose call this equipoise, and their example is a clinician laying out treatment options without advocating for one. That is an honest and appropriate use. The error is reaching for a neutral tool while privately hoping it produces a decision, which is most of the time it gets used.

The rule is short. If you want to see change, evoke. If you have decided neutrality is the right stance, decisional balance is the tool that matches it.

What to do instead

Evocation puts the argument for change in the client’s mouth rather than yours. In practice that means asking the questions that invite the change half, then making sure it does not evaporate:

Ask for the change side specifically. “What are the reasons you’d want to?” gets a different session from “let’s look at both sides.”
Reflect both halves, and finish on change. Acknowledging only the change side sounds like you were not listening. Acknowledging only the sustain side leaves them there.
Use the confidence ruler when the block is ability rather than desire. People who want to change and believe they cannot need a different conversation.
Let it stay unresolved. Ambivalence resolves over sessions, not in the last five minutes of one, and pushing for a decision is how you get a commitment nobody keeps.

The MI Spirit does the underlying work here. If partnership and acceptance are genuinely in place, the client can voice the sustain half without it becoming a position they have to defend.

What goes wrong

Five failure modes account for most of what a supervisor hears on tape.

Taking the change side. The righting reflex, and the most common one. You supply the argument for change, and the client, hearing only half their experience represented, supplies the other half.

Treating ambivalence as a stage to clear. It gets framed as a gate before the real work. It is the work.

Manufacturing it. If a client is not ambivalent, evoking a conflict they do not have is a way of creating sustain talk out of nothing. There is a live research question here: Ewing and colleagues, writing in Addiction in 2016, asked whether ambivalence is even a prerequisite for adolescents to benefit from MI, and found that young people who were not ambivalent still responded.

Reading discord as ambivalence. Covered above, and worth re-checking whenever a session feels like hard work.

Balancing too evenly, for too long. This is decisional balance leaking back in conversationally, without the handout. Every reflection gives equal airtime to both sides, the session feels admirably neutral, and the client leaves exactly where they arrived.

How to practise working with ambivalence

Reading about the righting reflex is not the same as catching yourself mid-sentence. The skill is noticing, live, that you have started building the case for change, and stopping. That is a reps problem, and it is what the MI Practice Lab is for: you hold a voice conversation with an AI client who is genuinely ambivalent rather than compliant, and afterwards you get back your change talk and sustain talk markers, an OARS-tagged transcript, and the Critical Moments where the client’s ambivalence shifted and your response either helped or hardened it.

It does not replace supervision. It gives you somewhere to make the mistake first.

Frequently Asked Questions

What is ambivalence in Motivational Interviewing?

Ambivalence is simultaneously wanting and not wanting the same change, and in MI it is treated as the normal state of someone considering a difficult decision rather than as denial or a lack of motivation. It surfaces in speech as change talk and sustain talk, usually in the same sentence and usually joined by “but”. The MI response is to evoke and reflect both sides rather than argue for the change side.

Is ambivalence the same as resistance?

No, and MI stopped using the word “resistance” in the third edition partly because it blurred two different things. Ambivalence is a conflict inside the client about the behaviour, and it shows up as sustain talk. Discord is friction between the client and the practitioner, and it usually means the clinician has pushed too hard. They look similar in the room but need opposite responses: reflect the first, repair the second.

Should I use a pros and cons list to resolve ambivalence?

Usually not, if your goal is change. Miller and Rose concluded in 2015 that decisional balance is contraindicated with ambivalent people when the aim is to help them change, because asking someone to build the case for the status quo strengthens it. Nenkov and Gollwitzer found it significantly reduced goal commitment among people who had not yet decided. Decisional balance is appropriate when you have deliberately chosen to stay neutral and not favour any outcome.

Why does arguing for change make ambivalence worse?

Because ambivalence has two sides and a conversation tends to distribute them. If you take the change side, the client is left holding the other, and people are persuaded far more by what they hear themselves say than by what they are told. This is the righting reflex, and it is the reason MI puts so much weight on evoking the client’s own arguments rather than supplying better ones.

What if my client is not ambivalent at all?

Then do not create it. A flat response usually means the person is not yet considering the change or has already decided, and both call for something other than evoking a conflict. Ewing and colleagues found in 2016 that adolescents who showed no ambivalence still responded well to MI, which suggests ambivalence is a common starting point rather than a required one.

Want to practise catching the righting reflex before it costs you a session? The MI Practice Lab puts you in front of a genuinely ambivalent AI client, then shows you the Critical Moments where your response hardened the sustain talk instead of shifting it. Start a free trial: 5 minutes, no card required.