The Science of Ambivalence: Why Staying and Leaving Both Feel Right

If you have spent months — or years — oscillating between "we can fix this" and "I have to leave," you may have started to suspect there is something wrong with you. There isn't. What you are experiencing is the predictable output of several well-studied psychological systems working on a single hard problem at the same time. Understanding the mechanics does not make the decision easier, but it can stop you from pathologizing yourself for having it.

What ambivalence actually is

Ambivalence in the psychological sense is not confusion. It is the simultaneous presence of genuinely opposing forces, each of which you endorse. The wish to stay and the wish to leave are both real; so are the reasons behind each. That is why pure logic does not resolve ambivalence — you are not missing information so much as trying to weigh two different kinds of value against each other.

Researchers studying decisions like this describe three systems that tend to be active at once. When they pull in different directions, what you feel is ambivalence. When they converge, what you feel is clarity. Most of the time they do not converge quickly.

System one: cognitive dissonance

Leon Festinger's 1957 theory of cognitive dissonance argued that the mind works to keep its beliefs, values, and behaviors internally consistent. When they drift out of alignment — when you believe you are a committed spouse but are beginning to feel you may need to leave — the gap itself produces a specific psychological discomfort. The mind will work, often without your awareness, to close the gap.

Two closure strategies are common and usually both appear in the same person.

• Upgrade the marriage. You catch yourself remembering the good times more vividly, discounting complaints that felt serious last week, or noticing evidence that the relationship is really not so bad.
• Downgrade the marriage. The opposite: you notice every flaw more sharply, you build a case, you find the marriage increasingly intolerable on the very days you suspected you might stay.

If you have had both of those days in the same week, you are not being inconsistent. You are watching dissonance-reduction try to resolve in two directions. It is the same system on both days, not two different selves.

System two: the Investment Model

Caryl Rusbult's Investment Model, developed through the 1980s and 90s, predicts commitment in close relationships from three variables: satisfaction (how rewarding the relationship feels), quality of alternatives (how viable life outside it looks), and investment size (how much you have put in — years, children, shared history, intertwined finances, social identity).

The model explains a pattern that surprises almost everyone who finds themselves inside it. People routinely stay in unsatisfying relationships — not because they are weak, but because high investment and low perceived alternatives produce genuine commitment, even in the absence of satisfaction. The three variables move somewhat independently. A marriage can be low on satisfaction and still high on commitment, and that combination is not irrational; it is what the model predicts.

Ambivalence is what the Investment Model looks like from the inside, when the three variables are in tension. Satisfaction has dropped. Alternatives have begun to look more viable than they used to. But investment is enormous and the commitment it generates is not a feeling you can simply override.

System three: loss aversion

Kahneman and Tversky's prospect theory, and the decades of research that followed, established one of the most reliable findings in behavioral economics: losses are felt roughly twice as intensely as equivalent gains. The prospect of losing your marriage, your house, your nightly routine with your children, your sense of a shared future — those losses have psychological weight that is not matched by the symmetrical gains (autonomy, peace, a better life later) even when, on paper, the gains would be larger.

Loss aversion is not a bug. It is the reason you hesitate before doing something irreversible. But it does have a predictable effect on this particular decision: it biases the internal accounting toward staying even when the stay case is genuinely weaker than the leave case. People in ambivalence often describe knowing, in their bones, that the loss of leaving would hit harder than the gain — and also knowing that this asymmetry does not mean leaving is the wrong choice.

The asymmetry in who leaves

Michael Rosenfeld's Stanford research, drawing on the How Couples Meet and Stay Together longitudinal panel, reported an asymmetry that is robust across the North American data: women initiate roughly 70% of divorces in heterosexual marriages. The same asymmetry does not appear in same-sex relationships, where breakup initiation is approximately even. Rosenfeld interprets the heterosexual finding as reflecting the higher costs women tend to carry inside the institution of marriage — housework, emotional labor, career compromise — rather than any general gender difference in commitment.

For purposes of ambivalence, the relevant finding is not the statistic itself but what it implies about time. The women who eventually initiate usually describe a long runway — years, often — of ambivalence before the decision consolidates. Their marriages did not end on the day they filed. They ended in a long accumulation of small unresolved moments that only later looked like a decision.

High-distress and low-distress divorces look different

Paul Amato and Bryndl Hohmann-Marriott's 2007 analysis distinguished divorces that followed visibly high-distress marriages from divorces that followed marriages that had looked, from the outside, reasonably functional. The low-distress group — sometimes called "good-enough" marriages that end anyway — reported ambivalence more consistently and longer. They also reported worse post-divorce adjustment on average. That is not an argument that they should have stayed. It is a reminder that ambivalence does not vanish on the day the decision is made; it is often the strongest predictor of how long after the decision you will still be asking whether you made the right one.

The high-distress group, by contrast, often described the decision itself as a relief. Ambivalence had burned off earlier. The post-divorce adjustment in that group was, on average, better — not because leaving is easy but because the thing being left was clearly worse than the leaving.

What to do with ambivalence while you have it

A few practices, drawn loosely from decision-science and couples-therapy literatures, tend to help people make ambivalence workable instead of unbearable.

1. Name the three systems. When you catch yourself oscillating, it helps to ask — am I upgrading the marriage right now, or downgrading it? Is this the investment pull, or the alternatives pull? Is this loss aversion? Labeling the force usually loosens its grip.
2. Write down both cases at full strength. On paper, separately, make the strongest honest case for staying and the strongest honest case for leaving. Most people, doing this for the first time, find that one side is thinner than they thought — or, revealingly, that both are thicker than they thought.
3. Separate the 'should I leave' question from the 'if I left, could I live with it' question. They are different questions and answering them separately tends to reduce the dissonance pressure.
4. Notice what you avoid. The topics you cannot bring yourself to think about — a specific conversation with a child, a specific financial calculation, a specific truth about your own behavior — are often the most informative. Ambivalence almost always includes avoidance.
5. Give it a time horizon, not a deadline. Rather than "I will decide by March," try "I will revisit this question with fresh evidence in three months, and I will use that time to gather it." Deadlines force dissonance-reduction. Horizons allow genuine data to accumulate.

The bottom line

You are not broken because you cannot decide. You are trying to solve, with a single answer, a problem that decades of research describe as having several moving parts that do not naturally align on the same timeline. The research does not tell you what to do. It tells you that the state you are in is a recognized state, that it has structure, and that the patient, honest accumulation of evidence — rather than the premature imposition of a verdict — is the approach most associated with decisions people do not later reverse.

Common questions