You are following your partner from the kitchen to the hallway, still talking, because the conversation cannot end like this. They have gone quiet. A hand on the doorframe. A sudden interest in the laundry. The more you press, the less you get, and the less you get, the harder you press. Researchers have a name for this loop, and a finding that changes who gets blamed: it is called demand and withdraw, and the roles have less to do with who you are than with whose change is on the table.
What the researchers actually counted
Demand is the pressing side of the loop: requests, complaints, criticism, the attempt to get a change discussed and settled now. Withdraw is the retreating side: going quiet, changing the subject, leaving the room, agreeing to anything just to end the conversation. Either move makes sense on its own. Together, they form a pattern in which each person gives the other more of exactly what they were trying to escape.
In 2014, Paul Schrodt, Paul Witt, and Jenna Shimkowski pulled together 74 studies of this pattern, covering 14,255 people, and published the synthesis in Communication Monographs. The pattern tracked with worse outcomes across the board, and the link was strongest for how the relationship itself felt and how well the couple communicated. The detail that matters most is the symmetry. The pattern carried roughly the same cost whether it ran in one direction or the other. A chasing partner and a retreating partner, in either arrangement, predicted a relationship both people rated lower.
The pattern is the problem. Neither person in it is.
That finding cuts against the usual telling, in which one person cares too much and the other cares too little. The studies describe something less personal and more mechanical. Pursuit reads as pressure to the person receiving it, so they protect themselves by retreating. Retreat reads as indifference to the person watching it, so they pursue harder to get a response at all. Each side experiences the other as the cause. The research says the loop is the cause.
The swap test
The cleanest evidence comes from an older study that changed the topic and watched the roles move. In 1990, Andrew Christensen and Christopher Heavey asked 31 couples to hold two conflict discussions, published in the Journal of Personality and Social Psychology. In one discussion, the topic was a change one partner wanted. In the other, the topic was a change the other partner wanted.
Both partners behaved differently depending on whose topic it was. Whoever wanted the change was more likely to press for it. Whoever was being asked to change was more likely to withdraw. The researchers also found an average tilt that should be stated plainly: across the discussions, the husbands withdrew more overall, while the wives were not more demanding overall. An average tilt is real. It is also not the whole story, because the same person who retreated on one topic pressed on the next.
This is why the pattern survives every attempt to assign it to a personality. The chaser in the kitchen conversation about money may be the one who goes quiet when the topic turns to how little time the couple spends together. Swap the topic and the roles often swap with it. The question worth asking is not who the difficult one is. It is who wants the change this time, and whether that change has any other way to be heard.
What to do with this
Name the topic before you judge the behavior. A useful first sentence is a plain one: this is my change to ask for, or this is your change to ask for. That sentence does not settle anything. It stops the loop from pretending to be a verdict on either person, and it gives the partner being asked to change a role other than defendant.
Separate a pause from a disappearance. Leaving a heated conversation can be wise. Leaving it with no return time is what turns a pause into withdrawal. A stated return, later tonight at nine, or tomorrow after breakfast, keeps the change on the table without keeping the pressure on. The change still gets discussed. The retreat stops reading as a refusal.
And keep one change per conversation. The loop tightens fastest when a single request arrives carrying every past version of itself. One change, discussed to an actual end, teaches both people that pressing is not the only way to be heard and that quiet is not the only way to be safe.
The chase is rarely a character flaw. It is usually a change that has nowhere else to go. Give the change a time and a hearing, and the chase and the retreat both lose their job.