A common assumption about abusive relationships is that if things aren’t bad constantly, the relationship can’t really be that bad. People outside the relationship ask why someone would stay with a partner who hurts them, as though the hurt were the whole of the experience. Psychologist Lenore Walker’s research on battered women suggested that assumption misses what actually keeps people attached. In 1979, Walker published The Battered Woman, based on interviews with 1,500 women, and described a pattern now widely known as the cycle of abuse: not a constant state, but three repeating phases. First comes tension building, where small conflicts and criticism accumulate and the person on the receiving end starts managing their partner’s moods to avoid a blowup. Second is an acute incident, the explosion the tension was building toward. Third is what Walker called the honeymoon phase: apology, remorse, affection, sometimes gifts, and a version of the partner that looks like the person the relationship began with. Then the tension starts building again.
The honeymoon phase is the part of the theory that explains what outsiders often find hardest to understand. It is not that the person being hurt fails to notice the incident. It is that the incident is followed by a real, felt period of calm and even love, which makes the relationship feel intermittently good rather than uniformly bad, and it is that intermittent goodness, not naivety, that Walker’s model identifies as the reason people stay.
Walker’s cycle theory has a close relative in later research that helps explain the mechanism behind it. Psychologists Donald Dutton and Susan Painter proposed a theory of traumatic bonding in the early 1980s and tested it empirically in a 1993 study published in the journal Violence and Victims. Dutton and Painter found that two conditions, together, produce unusually strong emotional attachment in an abusive relationship: a power imbalance, where one person is dominated by the other, and intermittent treatment, where good and bad periods alternate unpredictably rather than following a steady pattern. Their research drew on a broader body of behavioral psychology showing that rewards delivered on an unpredictable schedule create stronger, more persistent attachment than rewards delivered consistently. Applied to a relationship, a partner who is affectionate unpredictably, rather than reliably, can end up more psychologically difficult to leave than one who is affectionate all the time, because the uncertainty itself keeps a person invested in the hope that the good version will return.
It is worth being precise about what this research does and doesn’t establish. Walker’s original cycle theory has drawn real criticism since 1979. Later researchers studying domestic violence have found that not every abusive relationship follows the three-phase pattern cleanly. Some relationships involve abuse from both partners, which the original model doesn’t account for. Some involve controlling behavior that isn’t physically violent and doesn’t fit neatly into “tension” and “incident” categories. And critics have pointed out that Walker’s original sample, drawn heavily from women in prisons, jails, and mental health facilities, may not represent the full range of abusive relationships. The honeymoon phase itself often fades as a relationship goes on, with the calm periods growing shorter or disappearing entirely while the tension and incidents continue, which is a real limitation on how literally the three-part cycle should be read as a fixed timeline. The theory is best understood as a description of a common pattern in some abusive relationships, not a checklist that every relationship must match to count as abusive.
The scale of the underlying problem is well documented. The CDC’s National Intimate Partner and Sexual Violence Survey, collected between September 2023 and September 2024 from more than 15,000 adults, found that more than one in three women and roughly one in six men in the United States have experienced contact sexual violence, physical violence, or stalking by an intimate partner at some point in their lives, and nearly one in three women have experienced psychological aggression by a partner. Those numbers describe a common experience, not a rare one, which is part of why the underlying psychology, why people stay and how the bond is reinforced, has been a serious subject of research for more than four decades rather than a niche one.
None of this research is about assigning blame to the person being hurt for staying, and none of it claims that leaving is simple. What it does offer is a more accurate picture than the outside assumption that abuse is constant and that staying means not noticing it. The tension, the incident, and the calm afterward are not separate from the bond, they are what the bond is built from. Recognizing the pattern, in a friend’s relationship or one’s own, is less about spotting cruelty and more about noticing the shape of the whole cycle, including the parts that feel like love.
If this describes a relationship in your life, the fact that there are good days does not cancel out the frightening or controlling ones. Affection after an incident is not proof that the incident did not matter, and an apology is only meaningful when it is accompanied by sustained accountability and lasting change. A relationship does not need to be harmful every day to be harmful.
For someone inside this pattern, the safest first step may not be an immediate confrontation or departure. It may be speaking privately with a trusted person, documenting what has been happening, or contacting a domestic violence service to understand the options available. Leaving can be the period of greatest danger in an abusive relationship, so decisions about when and how to do it are best made with appropriate support and a safety plan.
The central lesson of the research is simple: the loving periods are not evidence that the abuse is imaginary. They may be one of the reasons the relationship has become so difficult to leave.
