The Psychology of Jealousy: The Feeling, the Behavior, and the Line Between Them

Jealousy has a working definition, and it is not the one in common use

The definition most of the research still runs on is Gordon White's, from 1981: jealousy is a complex of thoughts, feelings and actions that follows a threat to the existence or the quality of a relationship, where the threat comes from the perception of a real or potential attraction between one's partner and a rival. The phrase doing the most work in that sentence is the one people skip. White wrote that the rival may be imaginary. The machinery does not need a real third person to start running; it needs you to perceive one.

That is why jealousy cannot be evaluated by how strongly it is felt. The feeling is a response to a perception, and perceptions vary in accuracy. A strong reaction to something that is actually happening and a strong reaction to something assembled out of a delayed reply are the same feeling from the inside and completely different situations from the outside.

The second thing worth fixing early is the confusion with envy, because the two get used interchangeably and they are not the same emotion. Parrott and Smith ran two experiments on this in 1993 — one asking people to recall a personal experience of one or the other, one using stories in which the conditions for each were manipulated separately — and found the experiences came apart cleanly. Envy was characterized by inferiority, longing, resentment, and disapproval of one's own feeling. Jealousy was characterized by fear of loss, distrust, anxiety and anger.

This matters practically rather than semantically. Fear of losing something you have and longing for something you do not have call for different responses, and a person who has misnamed which one they are in will reliably take the wrong action — usually directing at a partner something that was never about the partner. More pieces in this vein under Mindset.

What you feel and what you do are separate variables

The most useful move the research literature made was to stop measuring jealousy as one thing. Laura Guerrero's 1998 study of 144 people in ongoing romantic relationships measured jealousy experience and jealousy expression separately, and the two behaved differently depending on attachment style.

People with a negative model of themselves reported more cognitive jealousy — the suspicious, rehearsing, what-does-that-mean kind of thinking — than people with a positive self-model. People with a negative model of others reported feeling fear less intensely, used less relationship-maintaining behavior, and engaged in more avoidance and denial. Preoccupied participants reported more surveillance behavior specifically. Dismissively attached participants reported less fear.

Read that as a set of dissociations rather than a set of types. Two people can report the same intensity of jealousy and do entirely different things with it: one goes quiet, one checks a phone, one raises it directly, one blames a stranger they have never met. The feeling did not determine the behavior. Something else did.

Which is why the question people usually arrive with — is my jealousy normal, is it too much — is close to unanswerable as asked. There is no calibrated scale for how much of a feeling a situation deserves, and the honest answer to how much jealousy is normal is that the range is enormous and self-report is a poor instrument. The variable that can actually be examined is what the jealousy produces. That is observable, it is describable to another person, and it is the thing that either damages a relationship or does not.

If you want a rough read on which attachment pattern you bring to this, our attachment style test scores it in about two minutes and states plainly what a questionnaire cannot tell you. The related distinction between intense attachment and love is worked through in love or anxious attachment.

The line is control, and past it this stops being a relationship question

There is a set of behaviors that gets described as jealousy and is more accurately described as control: requiring passwords or location sharing, reading messages, vetting who a partner may see, treating ordinary friendships or a workplace as a betrayal, demanding an account of hours that were not unusual, and making contact with the partner's family or friends expensive enough that it stops happening. These are not a stronger version of caring. They are coercive control, and the research treats them as their own variable rather than as a point on the jealousy scale.

Keilholtz and colleagues' 2023 meta-analysis of relationship dynamics and emotional intimate partner violence found perpetration significantly associated with a cluster: perpetration of controlling behavior, victimization of controlling behavior, jealousy, the demand-withdraw pattern, relationship distress, the perpetrator's power in the relationship, anxious attachment, and traditional gender roles. Jealousy appears in that list, but it appears alongside control rather than instead of it — which is the more accurate way to hold it. Jealousy on its own is not a diagnosis of anything. The controlling behavior is what carries the weight.

The reason to name this early rather than at the end is Campbell and colleagues' 11-city case-control study from 2003, which compared 220 intimate partner femicide victims, through proxy interviews, with 343 abused control women. Among the risk factors that survived multivariate analysis were the perpetrator's access to a gun, a previous threat with a weapon, and estrangement — specifically, estrangement from a controlling partner. Stalking, forced sex and abuse during pregnancy were significant at the bivariate level.

The finding that matters for anyone reading this in a bad situation is the one about leaving. The period around separation from a controlling partner is not a safe interval; it is the elevated-risk one. That is exactly why safety planning is done with people trained to do it and not from an article. Find A Helpline, in the sources below, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day.

These are correlational and case-control findings, and none of them says that a jealous partner becomes a dangerous one. What they say is narrower and still worth acting on: control is the variable that predicts harm, control is visible from outside, and it is not something to negotiate about.

Attachment predicts the shape of jealousy, not only the volume

Bowlby's framework, set out for a clinical readership in A Secure Base in 1988, treats the attachment system as a behavioral system that switches on when proximity to an attachment figure is threatened. Jealousy is close to a textbook activation of it: a perceived threat to a bond, producing a push to restore proximity. That framing explains why jealousy feels disproportionate to its trigger — the system is built to fire early and cheaply, on the logic that a false alarm costs less than a missed one.

What Sharpsteen and Kirkpatrick added, in 1997, was that attachment style predicts the qualitative form the jealousy takes and not merely how much of it there is. In their data, anxious participants were relatively more likely to resist expressing their anger. Avoidant participants were especially likely to turn their anger and blame on the interloper — the third party, who in many cases had done nothing. Securely attached participants were especially likely to express anger toward the partner and to maintain the relationship, which sounds worse and is in fact the version that keeps the problem in the room where it can be solved.

A larger and much more recent study points the same way with different instruments. Richter, Schlegel, Thomas and Troche surveyed 847 people in 2022 and modeled Big Five traits and attachment dimensions together as predictors of romantic jealousy. Higher neuroticism, lower agreeableness and lower openness predicted higher jealousy; attachment anxiety partly mediated the neuroticism effect and fully mediated the agreeableness effect. The relationships did not differ by gender, relationship status, or whether the person had experienced infidelity.

Every one of those studies is cross-sectional self-report, which means they describe associations across a population and cannot tell you what is happening in one relationship, including yours. They are also not a verdict on anyone's character: attachment patterns are learned and they move, slowly and partially, in the way the longitudinal evidence on whether attachment styles change sets out. What they are good for is recognizing that the intensity of the feeling is partly a property of the person having it, which is a more tractable problem than it first sounds — and a considerably better one than the alternative explanation, which is that your partner keeps doing something wrong. Why smart people keep dating the wrong people covers the same machinery from the partner-selection side.

Checking manufactures the uncertainty it is trying to settle

The most common behavioral response to jealousy now is looking something up. There is a small research literature on this and it is unusually consistent.

Muise, Christofides and Desmarais surveyed 308 undergraduates in 2009 and found that increased Facebook use significantly predicted Facebook-related jealousy in a hierarchical regression that controlled for individual, personality and relationship factors. Their proposed explanation was a feedback loop: the platform exposes people to information about a partner that is often ambiguous and that they would not otherwise have had, and that new information drives further use.

Elphinston and Noller, in 2011, found the same relationship with the mechanism specified — what they called Facebook intrusion was linked to relationship dissatisfaction via jealous cognitions and surveillance behaviors. The monitoring was not a symptom sitting beside the dissatisfaction. It was on the path to it. Tokunaga gave the behavior its research name the same year: interpersonal electronic surveillance. A more recent survey of 346 people traces one route into it that has nothing to do with a rival — being ignored for a partner's phone lowered perceived responsiveness, which raised anxiety, which predicted checking; the authors are careful that the model accounts for only about five per cent of the variance, and the phone-use piece reports it with that caveat attached.

The caveats are real and worth stating. All of this is cross-sectional, most of it is undergraduate samples, and the platform studied is now well over a decade older with different mechanics. Nothing here establishes direction: jealous people plausibly just check more. But the two readings converge on the same practical point, which is that checking does not deliver what it promises. A social feed is made of fragments without context — a photograph with no explanation, a name with no history, a timestamp with no account of the evening. Ambiguity is the raw material, and ambiguity is precisely what jealousy metabolizes.

The observable version of this: notice how long the relief lasts after a checking episode. If the honest answer is twenty minutes, the behavior is not gathering information. It is regulating a feeling, badly, and it will need to run again shortly. The same behaviour aimed at an ex rather than at a partner is one of the clearer markers that a breakup is still running, which is where the signs you have moved on and the signs you have not starts. Sustained ambiguity is expensive in its own right — the same dynamic that makes situationships corrosive — and no amount of scrolling resolves it. More on how thin channels distort intention under Texting, and on why text runs flat and negative rather than randomly wrong.

The claim about men and women that you have read everywhere is contested

One finding about jealousy has escaped into general circulation more than any other: that men are more distressed by sexual infidelity and women by emotional infidelity. It comes from Buss, Larsen, Westen and Semmelroth in 1992, and it is worth knowing what it actually rests on, because it gets cited as though it were settled.

The result came from a forced-choice question — which of these two would upset you more — and the sex difference is reliable when you ask it that way. What is disputed is whether the forced-choice format is measuring an evolved response or creating one.

DeSteno, Bartlett, Braverman and Salovey published two studies in 2002 arguing the latter. The sex difference appeared only under forced choice; on every other measure they used, no sex differences were found, and both men and women reported greater jealousy in response to sexual infidelity. In a second study, the forced-choice difference disappeared under cognitive load — that is, when participants were prevented from deliberating. Their conclusion was that the effect is an artifact of an effortful decision strategy induced by the response format.

Harris reviewed the whole evidence base in 2003, across self-report, psychophysiological measures, domestic violence and morbid jealousy cases, and found only one hypothetical measure consistent with the innate-module hypothesis, contradicted by a range of others including people's reported reactions to real infidelity they had actually experienced. A meta-analysis of jealousy-inspired homicides that corrected for murder base rates found no evidence that jealousy disproportionately motivates men to kill.

And the argument did not end there: Sagarin published a comment in the same journal in 2005 defending evolved sex differences, and work since has reported the difference using continuous as well as forced-choice measures. So the honest summary is three-part. The forced-choice result replicates. What it means is genuinely unsettled. And treating "men mind one thing, women mind another" as an established fact about a particular couple goes far past anything the data support.

This is not a pedantic point. That claim is routinely used to license monitoring — it is just how men are, it is just how women are — and it is nowhere near strong enough to license anything. We take the same posture toward other widely repeated dating statistics in the real numbers behind modern dating and what has actually been measured about situationships.

Sometimes the signal is accurate, and there is no clean test

There is a failure mode in the other direction, and the popular attachment writing produces a lot of it: treating every jealous reaction as one's own pathology, to be managed internally and never raised. Sometimes the perception is correct. White's definition allows the rival to be imaginary; it does not require it.

Nothing sorts these reliably in the moment, and any article claiming otherwise is overselling. What is available is a longer time window and a few questions whose answers are observable rather than inferred.

Does it track something that changed, and could you describe that change to a neutral third party without editorializing — a shift in how evenings are spent, a topic that has become off-limits, a pattern of accounts that do not line up? Or does it fire on ambiguity as such: a delayed reply, an unfamiliar name, a photograph?

Does it survive an answer? A concern that is about information ends when the information arrives. A concern that is about your own alarm system accepts the reassurance, holds for a while, and restarts at the next ambiguous input. Guerrero's finding is the anchor here: cognitive jealousy — the suspicious, looping kind — was predicted by a negative model of the self, not by anything the partner had done.

Does it travel? If the same feeling with the same texture arrived in earlier relationships with different people, the constant across those relationships is you. That is not an accusation and it is not a dismissal of the current situation. It is a narrower and more workable problem than the alternative.

None of the three is diagnostic and all three can be true at once. What they are useful for is refusing both of the easy conclusions — that the feeling proves something, and that the feeling proves nothing. Reading ambiguous behavior accurately is hard in general; the research on telling flirting from friendliness and the body language cues that hold up both cover why single cues are such poor evidence.

What to do with the feeling, and what the evidence for it actually is

The first move with reasonable support behind it is naming the emotion. Lieberman and colleagues, in a 2007 imaging study, found that affect labeling — putting a feeling into words — reduced amygdala response to negative emotional images relative to other kinds of encoding, and increased activity in right ventrolateral prefrontal cortex, with the two inversely correlated.

Be precise about what that does and does not establish. It was a laboratory study using emotional images, not jealous partners in real relationships, and it does not show that naming jealousy makes it go away. The narrower claim it supports is that describing an emotion recruits regulatory machinery that simply having the emotion does not. That is a reason to say to yourself, specifically, that you are feeling jealous and about what — it is not a technique, and it is not a cure.

The second move is the conversation, and the form matters more than the content. The version that reliably fails is interrogation: where were you, who is that, why did you laugh like that. It leaves the other person only two available moves, defending or conceding, and neither produces information — a defense tells you nothing, and a concession extracted under pressure tells you less.

The version worth trying separates three things that interrogation collapses into one: what happened, what you felt, and what you would like. When this happened I felt jealous; I think it is partly about that; could we do this. It is not a script for getting a result out of someone, and it should not be sold as one. Its only advantage is that it does not destroy the conversation in the first sentence. Holding the gap between feeling something and acting on it is most of what emotional maturity looks like from the inside, and it is a practice rather than a trait.

This is the part of the literature Gottman's observational work speaks to. Why Marriages Succeed or Fail, from 1994, is the accessible statement of the coding scheme his lab built by watching real couples argue, and of the escalating pattern — criticism, contempt, defensiveness, stonewalling — that it identified. The underlying longitudinal work with Levenson in 1992 found that specific interaction behaviors and physiological arousal, rather than what the couple was arguing about, tracked which marriages later dissolved. What that same body of observational work supports doing instead — including what the famous five-to-one ratio actually counted — is set out in the research on couples who last. One caveat that the popular retellings drop: the widely quoted prediction-accuracy percentages have been criticized for having been derived and reported on the same samples, which inflates them. The descriptive coding is the durable contribution — it is also the structure our conflict style test borrows, scoring criticism, contempt and defensiveness apart from stonewalling; the headline numbers are not, and the same caveat applies to the divorce-prediction percentages discussed in what the repair research shows about reconnecting after a fight. On what escalation looks like from the beginning, see the first-date mistakes that quietly kill attraction.

Two things that are not a plan

The first is provoking it deliberately. Fleischmann, Spitzberg, Andersen and Roesch studied jealousy induction in 2005 and mapped it: two classes of goal, which they labeled relational rewards and relational revenge; three classes of tactic, being relational distancing, a flirtation facade, and displaying relational alternatives; and three classes of partner response — aggressive, withdrawal, and relational compensation.

Read that third list again, because it is the argument. The move is usually made hoping for relational compensation, the response where the partner moves closer. Two of the three documented outcomes are not that, and one of them is aggression. Kaufman-Parks, Longmore, Giordano and Manning, working with data on 892 young adults from the Toledo Adolescent Relationships Study, found jealousy induction associated with experiences of partner violence after accounting for family background, relationship and sociodemographic factors, and found jealousy-inducing behavior itself associated with verbal conflict, a partner's infidelity and controlling behaviors, and childhood experiences of parent-child physical aggression. These are correlations in a young-adult sample, not a causal chain. They are still enough to take the tactic off the table.

There is a plainer objection that does not need a citation. Deliberately inducing jealousy means manufacturing distress in another person in order to extract information about how much they care. If what you want is to know where you stand, a direct question is cheaper, faster, and gives an answer that means something — which is the same conclusion the evidence reaches about staying in an undefined situation.

The second thing that is not a plan is waiting out jealousy that has stopped being proportionate at all. Seeman's 2016 review of pathological jealousy describes a recognized clinical presentation: delusional jealousy is roughly equally prevalent in men and women, more common in older adults, and its antecedents can be neurologic, drug-related or psychological — most often preceded by low self-esteem and marked dependence on a partner. It can be triggered by a partner's behavior and then maintained by reasoning biases, and it poses real danger to the patient, the partner and the imagined rival, sometimes requiring involuntary hospitalization. Seeman is also candid that treatment effectiveness does not yet have a firm evidence base.

The marker distinguishing that from ordinary jealousy is not intensity either. It is that the belief has become fixed and does not move in response to evidence, and that it is organizing behavior. That is a clinician's territory, reached through a doctor, and not something a relationship conversation is built to hold.