Codependency arrived from the addiction-recovery movement in the 1980s and never left. It is one of the most widely used words in relationship writing — a cycle, a pattern, a personality, a thing you can be diagnosed with by a quiz.
It is also, as a research construct, in poor condition. And there is a neighbouring construct that is not, which describes something narrower and more useful.
The definitions do not agree with each other
The problem is not that codependency has been studied and found wanting. It is that it has never been pinned down well enough to study properly.
There is no universally accepted definition, no agreed set of behaviours that distinguishes a codependent person from anyone else, and no consensus on diagnostic criteria. Reviewers looking across the widely cited definitions find recurring themes — an external focus, self-sacrifice, attempting to control other people, suppressing one’s own emotions — but themes across disagreeing definitions are not the same thing as a definition.
The measurement literature tells the story on its own. There is a Spann–Fischer scale, a Holyoake Codependency Index, and a 2012 paper titled, without irony, “Development and validation of a revised measure of codependency.” A field that keeps rebuilding its ruler is a field that does not yet agree what it is measuring.
What “poorly defined” does and does not mean
It does not mean the experiences described are imaginary. People do organise their lives around someone else’s crisis, do lose track of their own preferences, do find that the relationship’s stability depends on their continuing to absorb something. All of that is real and common.
What it means is that the label bundles several different things — a partner’s addiction, a family role learned in childhood, an anxious attachment style, a personality disposition — into one word, and a word that covers all of them cannot tell you which one you have. That is a problem for treatment, not just for research.
The construct next door is defined and measured
Working separately and from a different tradition, Vicki Helgeson and Heidi Fritz set out something narrower: unmitigated communion, which they define as a focus on others to the exclusion of the self.
The crucial move is the distinction they drew. Caring about other people — communion — is not the problem and is associated with good outcomes. Unmitigated communion is correlated with it and is not the same thing, and across four studies they showed it is distinguished by three specific features.
- A negative view of the self. Not modesty and not selflessness — a genuinely poor self-appraisal sitting underneath the caring.
- Turning to others for self-evaluative information. Using other people as the instrument for finding out whether you are all right.
- Psychological distress. Which the other two are not merely associated with, but appear to produce.
And they went further than describing it. They tested the pathway, and found that relying on others for self-esteem leads to overinvolvement with others and neglect of the self — and that those two, in turn, account for the link between unmitigated communion and distress.
Where the word came from, and why that shaped it
The origin explains a lot of the trouble.
Codependency emerged in the 1980s from the addiction-recovery movement, where it named something specific: the spouse or family member whose behaviour had adapted around a drinker. That was a real observation about a real situation, and the word did useful work in it.
What happened next is that it expanded. Within a decade it covered any relationship in which someone was too accommodating, then any pattern of putting others first, then — in its widest usage — most of ordinary caring behaviour, with checklists broad enough that almost anyone answers yes.
A construct that grows to cover everything stops distinguishing anything, which is exactly what the measurement literature found when it tried to build a scale. The word’s usefulness and its imprecision came from the same expansion.
Why the narrower construct is more useful
That mechanism is the reason this matters practically rather than academically.
“Codependency” tells you that you are too involved with someone. Unmitigated communion tells you which part of that is doing the damage: not the caring, and not even the involvement on its own, but the fact that your read on your own worth is being taken from someone else, which produces the overinvolvement and the self-neglect downstream.
The caring is not the fault
Communion — a positive caring orientation towards others — and unmitigated communion are correlated but distinct. The advice to care less is aimed at the part the research finds benign; the part that carries the distress is where self-evaluation is being sourced from.
Which changes what a person is meant to do. “Set boundaries” and “stop enabling” both target the behaviour. The measured pathway starts a step earlier, at where you are getting your information about yourself from.
What the popular framing gets wrong even where it is pointing at something real
Two things follow that the recovery-derived version tends not to say.
The first is that this is a disposition and a pattern, not a diagnosis. It is measured on a continuum, everyone is somewhere on it, and where you sit can differ between relationships. “Am I codependent?” is the wrong shape of question in a way that “how much of my sense of how I am doing comes from one person’s state?” is not.
The second is that the cycle language does a specific kind of harm. Describing someone as caught in a cycle implies the engine is inside them, which lands badly on people whose situation is largely being generated by another person’s behaviour. A partner’s addiction, a parent’s illness, a relative’s instability — those are external facts, and a framework that locates the whole problem in the person absorbing them is not neutral.
Try this
For a week, when you notice yourself checking how someone else is, add one question: what would I need to know to decide whether I am all right today, if their mood were not available as evidence?
Most people find they have very little to answer with, which is the finding rather than a failure. The mechanism in the research is self-evaluation sourced from another person; noticing the absence of any other source is the first place the pathway is visible.
You are not trying to care less. Caring is the part the evidence exonerates.
The parts of this that are genuinely hard
The unmitigated communion research is cross-sectional and largely self-reported, and its mediation model — self-esteem reliance producing overinvolvement and self-neglect, which produce distress — is a statistical pathway rather than a demonstrated causal chain. It is better specified than codependency. It is not proven.
There is also a real cost to being told that a familiar word is not a solid construct. For many people, “codependency” was the term that first made a confusing situation legible, and Al-Anon and similar communities have been genuinely useful to people the formal literature has served poorly. Saying the construct is ill-defined is a statement about the research, not about whether anyone was helped.
And where the other person’s behaviour is the primary driver — active addiction, coercive control, untreated illness — none of this is the relevant frame. Something that looks like overinvolvement can be an accurate response to danger, and the useful thing at that point is not a personality construct but practical support.
The bottom line
Codependency has no agreed definition, no settled criteria and a succession of competing scales, which is why forty years of it has produced so little that can be acted on.
The construct beside it is sharper. Unmitigated communion is a focus on others to the exclusion of the self, it is distinct from ordinary caring, and its measured route to distress runs through taking your read on yourself from somebody else. That is a narrower target than a cycle, and unlike a cycle it points at something you can actually go and look at.
Sources (4)
Every factual claim in this article comes from one of these sources.
Keep exploring
Continue reading
Deeper into what you just read.
Explore more
Adjacent ground, if this was useful.
One thing worth knowing
A surprising finding. The evidence behind it. And what it means for your life. BetterYou+ sends one useful idea at a time—nothing to keep up with.