Love or Anxious Attachment? How to Tell the Difference

The two states share a surface, and that is the whole problem

Intrusive thoughts about one person. Checking your phone more than you would like to admit — a pattern that shows up in the phone-log data as well as the self-reports, and that the research on texting habits traces back to the same place. A day that reorganises itself around whether they have replied. Both love and attachment anxiety produce all of that, which is why asking yourself how strongly you feel it gets you nowhere.

This is not a modern observation. When Hazan and Shaver (1987) translated Bowlby's infant attachment categories into adult romance, they asked adults to describe their most important romantic relationship. The group whose answers matched the anxious-ambivalent pattern described love in terms of extremes of emotion, obsessive preoccupation, a desire for union, and a longing for reciprocation. Read those descriptions cold and they sound like an unusually passionate account of being in love. That overlap is not a failure of the participants' self-knowledge. It is the finding.

So the useful question is not how much you feel. It is what the feeling is organised around — the person, or the uncertainty about the person. The rest of this article is a set of places where those two come apart in ways you can actually observe rather than infer.

If you want a rough read on where you sit before going further, our attachment style test scores the dimension in about two minutes and is explicit about what a score like that cannot tell you.

What the research measures is a dial, not a diagnosis

Popular writing treats anxious attachment as a category you either belong to or do not. The measurement literature has not worked that way for nearly thirty years.

Brennan, Clark and Shaver (1998) ran a large factor analysis over essentially every adult attachment self-report measure that existed at the time. Underneath all of them they found two continuous dimensions: attachment-related anxiety and attachment-related avoidance. In R. Chris Fraley's summary of the resulting instrument, anxiety is the extent to which people are insecure rather than secure about the availability and responsiveness of romantic partners, and avoidance is the extent to which they are uncomfortable being close to others rather than secure depending on them. That instrument, the Experiences in Close Relationships scale, is still the standard tool.

Two things follow, and both matter more than they sound.

First, there is no line to fall on the wrong side of. A moderately high anxiety score is ordinary, not a condition. Second, and this is the part that gets skipped: an attachment questionnaire measures attachment. It does not measure love. No self-report scale — ours very much included — can tell you whether what you feel for a specific person is real. What it can tell you is what you tend to do when closeness feels uncertain, which is a different and more answerable question. We say the same thing on our editorial policy page, and we would rather repeat it here than let a two-minute quiz carry weight it cannot hold.

Your body keeps a steadier record than your interpretation

The most useful evidence here is physiological, precisely because it does not depend on you labelling your own feelings correctly.

On the anxious side, Pietromonaco and Powers (2015), reviewing this literature, report that most studies have found anxiously attached individuals show elevated cortisol in response to a stressor and take longer to recover afterwards, and that anxiously attached married people produce higher daily cortisol overall. Not a spike during a fight — a baseline that sits higher and comes down more slowly.

On the secure side, Coan, Schaefer and Davidson (2006) put sixteen married women in an fMRI scanner under threat of electric shock. Holding their husband's hand produced a broad reduction in activation across the systems that carry threat response. Holding a stranger's hand reduced it much less. And the size of the effect tracked the quality of the marriage: the better the relationship, the less the threat network fired while the partner was holding on. Calm in one specific person's presence is a measurable event, not a poetic way of saying the relationship has gone flat.

The honest caveat, because this is where popular writing overreaches. These are small samples and the pattern is not uniform. In the dating-couple cortisol study that opened this line of work, Powers and colleagues (2006) found the sharpest reactivity in particular attachment pairings rather than in anxious individuals across the board, and later work found that spouses' attachment combinations predicted the hormonal pattern better than either partner's score alone. No cortisol study is a verdict on your relationship. Take the direction of the evidence, not the precision.

The practical version: over a month, in this person's presence, does your baseline settle or climb? Not whether you feel excited — whether your shoulders come down. Your own baseline is a more consistent narrator than your interpretation of it. That is a claim about your body, though, not about anyone else's: reading another person's nonverbal behaviour turns out to be a much weaker instrument, and what body language actually predicts goes through how weak.

Uncertainty genuinely raises the intensity, which is exactly the trap

There is a real experimental finding here, and it is worth stating precisely, because the sloppy version of it is everywhere.

Whitchurch, Wilson and Gilbert (2011) showed college women the profiles of four male students who had supposedly already seen their profiles. Some were told those men had rated them highly. Some were told the ratings were average. A third group was told the men had rated them either highly or average — without being told which. That uncertain group ended up the most attracted of all, more than the group told they were liked a lot. The mechanism the authors proposed is unglamorous: not knowing makes you think about someone more, and thinking about someone more raises how attracted you report being.

What that study is: a clean demonstration that the mechanism exists. What it is not: a law. One paper, one sample of undergraduate women, an artificial task, and it measures rated attraction over minutes rather than love over months.

We are flagging it carefully because the famous version of this idea stands on much weaker ground than its fame suggests. Dutton and Aron's 1974 suspension-bridge study is still cited across this genre as proof that fear gets mistaken for love; a published reanalysis argues the original data do not support that conclusion, and later work on arousal and attraction finds effects that are short-lived and heavily context-dependent. It is one of several studies in this field that everyone quotes and almost nobody has read.

Why any of it matters for the question you came with: if your intensity climbs when you cannot tell where you stand, and flattens out when you can, then some part of that intensity is a function of the ambiguity rather than of the person. That is the engine running quietly underneath most of what people describe as a situationship, and it is why the least stable relationships so often get remembered as the most passionate ones. Geographic distance produces a milder and better-measured version of the same effect, and what happens when it ends is the subject of what distance does to attraction.

Reassurance that does not stay put

There is a behaviour with its own research literature that maps almost exactly onto what people mean when they ask this question. It is called excessive reassurance seeking: persistently asking for assurance that you are lovable and wanted, regardless of whether that assurance has already been given.

Shaver, Schachner and Mikulincer (2005) studied it in couples across two samples. Reassurance seeking was related to depression — but the relationship was explained by attachment anxiety. In other words it is not a standalone habit or a personality quirk. It is the behavioural signature of the anxious dimension doing exactly what the theory says it does: turning the volume of the attachment system up rather than down.

Two findings from that same work are worth sitting with, because they are unwelcome. Partners did notice each other's reassurance seeking. And the link between attachment anxiety and depression was not explained by how either partner rated the relationship, while relationship quality tracked avoidance more than anxiety. Read together, that says something uncomfortable and useful: a partner supplying more reassurance is not the variable that resolves this.

The observable version is the half-life. Reassurance that answers a specific worry and then rests is one thing. Reassurance that is received warmly, believed for an afternoon and gone by evening is a different thing, and the difference is not in the quality of what your partner said. That same half-life is the most useful test there is for separating an accurate jealousy signal from a false alarm, which is one of the distinctions worked through in the psychology of jealousy. If you keep landing in the same loop across different people, what repeats is worth looking at directly — and so is the question of what is producing the repetition, which is the whole subject of why smart people keep dating the wrong people.

Intensity that lasts is not the same as intensity that consumes

The most direct empirical answer to whether calm means the love is gone comes from a review rather than a single study.

Acevedo and Aron (2009) pooled 25 studies covering 6,070 people in short- and long-term relationships, and did something the popular conversation rarely does: they separated romantic love — intensity, engagement, sexual interest — from the obsessive component, meaning preoccupation, uncertainty and anxiety about the partner. The two came apart cleanly. Romantic love without the obsessive component was associated with satisfaction, wellbeing and self-esteem in long-term relationships. The obsessive component showed up in short relationships and did not carry over into long ones.

So the equation of obsession with depth is, on the available evidence, backwards over any long horizon. Intensity survives. Obsession does not, and where it persists it is not doing the work people credit it with.

This is a review of correlational studies, so it cannot establish that obsession causes anything. What it does establish is that the two components are separable and that they behave differently over time — which is enough to retire the idea that a settled relationship is a diluted one.

A question that operationalises it: would this feeling still be here if I were completely certain of them? If the answer is yes, the feeling is not running on the uncertainty. If the honest answer is that it would deflate, that is worth knowing, and it is not the same as discovering the relationship is worthless. Steady behaviour that keeps showing up when nothing is at stake is what the early signs of someone actually falling for you look like, and it is unspectacular by design. The research that separates the two kinds of intensity, and finds that the consuming one is the component that reliably fades, is covered in why the honeymoon phase ends.

Both can be true at once, and usually are

The framing in this article's title is a little dishonest, and we would rather say so than pretend otherwise. Love or anxious attachment implies a fork in the road. Dimensions do not work that way.

You can love someone and score high on attachment anxiety. The anxiety does not disprove the love, and the love does not neutralise the anxiety. Both run at once, and in early relationships they almost always do, because early relationships are genuinely uncertain and the anxious dimension is a response to uncertainty about availability. Nothing has gone wrong when both are present.

What does follow is a limit on what you can conclude. Your own activation is not evidence about your partner. It does not prove they are pulling away, and it does not prove they are not. Anxiety inflates ambiguous signals in one direction the same way avoidance discounts them in the other, and neither distortion announces itself as a distortion from the inside. If you notice the same shape recurring across different partners, the constant in that equation is you — which reads harshly and is actually the good news, because it is the part you have any leverage over.

That is the whole reason we keep a Mindset section rather than only writing about what other people are doing, and it is why the numbers on undefined relationships are worth reading as much for what nobody has measured as for what two surveys did.

What moves the dial, and what is not an attachment problem at all

Attachment orientation is not fixed. Chopik, Edelstein and Grimm (2019) pooled three longitudinal samples — 628 people, measured across ages 13 to 72 — and estimated how these dimensions move over a lifetime. Attachment anxiety stays relatively high through adolescence and declines from there; avoidance drifts down slightly across the whole lifespan. That is mean-level change across a population, not a promise about any individual, and the movement is slow. But it moves, which is more than the this-is-just-who-I-am framing allows.

On what actually shifts it, we are going to be thinner than most articles in this genre, because the evidence is thinner. Therapy has support. Sustained time in a relationship where responses are consistent and non-escalating has support. A five-step protocol does not, so we are not going to invent one.

One boundary, and it is the most important paragraph on this page. Some of what gets filed under my anxious attachment is not attachment. If your partner monitors your phone or your location, decides who you are allowed to see, controls money or transport, threatens you, or punishes you with silence or rage until you comply — or if you have quietly reorganised your own behaviour around avoiding their reaction — that is coercive control. It is a recognised pattern of abuse, and attachment theory is the wrong lens for it: it hands you a vocabulary in which the problem is your nervous system rather than their conduct, and that framing keeps people in these relationships for years. Find A Helpline, linked in the sources below, lists free and confidential services by country; in the United States, 988 reaches the Suicide and Crisis Lifeline. Please use them ahead of anything on this site.

And for the ordinary case, where nobody is being controlled and the question is genuinely just hard: a score is the start of a conversation, not an answer. Our attachment style test says so on its own page, and we would rather it be useful once than reassuring forever. The question that usually comes next — whether the dial moves at all, how far, and over what period — is set out with the numbers in what the longitudinal evidence says about attachment styles changing. The place the dial is easiest to read is the hour after an argument, which is covered in what the repair research shows about reconnecting after a fight. If what you notice there is mostly the urge to settle it before the evening ends, the conflict style test measures that pull on its own, separately from attachment.