Signs the relationship is over: signal vs noise
Not every doubt means the end. Tell the 4 real signals apart from emotional noise using Rusbult's EVLN model and machine-learning evidence on desire.
“I’ve been asking myself for months whether this is already over — and I still don’t know.” If that sentence sounds familiar, this piece is about that. About how to sort the real signals from the emotional noise, without turning every doubt into a diagnosis.
When Rusbult, Johnson, and Morrow published the EVLN model in 1986 — Exit, Voice, Loyalty, Neglect — they named four responses to an unsatisfying relationship. DOI: 10.1177/001872678603900103. “Is it over?” becomes answerable when you know which of the four is operating — and for how long.
This piece walks through the four signals that actually carry weight and separates them from the noise: fear of being alone, comparison on social media, passing guilt, external crises. The idea is not to decide for you. It is to give you a map so that you can decide, with the lens on what matters rather than what scares.
1. The trap of confusing noise with signal
Most doubts that reach a therapist’s office are not signs that the relationship is over. They are noise that feels like signal because it arrives with huge emotional intensity: a big fight, a comparison on social media, a quiet week, a sentence that came out wrong. All of that can feel like confirmation that “it is done,” when they are fluctuations of a living system.
The problem: noise and signal use the same channel — body, discomfort, silences. That is why the question cannot be answered by intuition alone; it is answered by separating one from the other with observable markers.
The number matters because many of those doubts resolve themselves once what triggered them passes. Others do not. The difference between the ones that resolve and the ones that do not is usually a single one: a signal has pattern. Noise has the shape of an event.
2. The EVLN model — the map that changed the question
Before 1986, theory assumed that people respond to dissatisfaction basically with two moves: stay or leave. Rusbult, Johnson, and Morrow showed that view was too narrow. They proposed that the human response to dissatisfaction has four branches, not two.
| EVLN response | What it is | What it implies over time |
|---|---|---|
| Exit | Ending the relationship in a more or less explicit way | Resolves dissatisfaction from outside; requires a support network |
| Voice | Talking, negotiating, trying to make the relationship change | Activates change if there is reciprocity; runs dry if there isn't |
| Loyalty | Passively waiting for the situation to improve on its own | Sustains the status quo; carries accumulated emotional cost |
| Neglect | Stopping investment in the bond — letting it rot | Deteriorates the relationship through inattention; often precedes exit |
The clinical question is not “do I leave?”. It is: which of the four am I in? And most importantly: am I in voice or have I already moved to neglect? Voice implies something could still change; neglect is a passive response where the person has stopped investing and probably does not even notice. DOI: 10.1177/001872678603900103
How to know whether you are in voice or have already shifted to neglect
An operational check: look at the last four serious conversations you had with your partner. Was there a clear intention to change something specific? Was there a concrete agreement, however small? Did anyone follow through? If yes, voice is still functioning. If no — if the conversations repeated without agreement, or if the agreements did not hold — you are probably already in neglect territory, even if no one has named it. The grey zone between voice and neglect is where doubt becomes chronic without the decision being formalised.
3. The four signals that actually weigh — with substance
There are four signals that, when they appear together or sustain over time, carry real clinical weight. They are not the only ones, but they are the most robust in the evidence reviewed. If all four are present, the signal is solid. If only one shows up, it is probably noise. If two or three sustain for months, they are worth examining with care.
| Signal | What you observe | How it sustains |
|---|---|---|
| 1. Sustained damage | Broken sleep, baseline anxiety, loss of interest in personal projects | Months of progression with no identifiable external cause |
| 2. Voice depleted | Serious conversations with agreements that do not hold over time | Pattern repeats every few months despite new conversations |
| 3. Load asymmetry | One person subsidises all logistics, emotional life, child-rearing, shared project | If the other person stopped contributing, the system collapses |
| 4. Values clash | Children, location, money, way of living — central axes incompatible | Years of talking about the same clash without positions moving closer |
3.1 · Sustained damage to your health or personal baseline
The first signal is the most concrete: body and mood have been paying the bill for months — relief when they are away, tension when they return. One bad month is noise; six months is data.
The personal baseline also includes what feels like yours: your capacity to decide, your creativity, your energy for your own projects. If something eroded steadily since you have been in this relationship, and you cannot attribute it to an identifiable external cause, the signal is yours. And it is worth hearing.
3.2 · Repeated pattern after honest conversations
The second signal has to do with voice. There were conversations — serious, honest, in calm moments — and something moved for a few weeks before sliding back. One cycle is not failure. Three cycles without sustained change is the model’s definition of exhausted voice.
When voice runs dry what remains is loyalty or neglect. Rusbult documented waiting as the most common response in couples with years invested — but passive waiting is not neutral: while nobody talks, decay continues.
3.3 · The relationship only works because one person does all the work
The third signal is asymmetric. One person carries the emotional, logistical, sexual, financial, child-rearing, shared-project load. The other person is present, but does not contribute at the same level. If this has been going on for years, what was installed is not a partnership: it is a system where one person subsidises the other’s life without real reciprocity.
This signal is the hardest to see from inside, because it is usually normalised over years. A useful way to detect it: if you stopped doing everything you do for this relationship, what would happen in a month? If the honest answer is “it would collapse,” the signal is asymmetric. It does not need to be dramatic to be real.
3.4 · Your fundamental values clash on big decisions
The fourth signal is the most structural. Children, location, money, way of living. If on any of these central axes your clash with your partner is sustained and conversation does not bring positions closer, there is an incompatibility of values. Not personality. Not quirks. Values.
Values incompatibility does not resolve with more conversation or more years together. It sustains or it does not. If you have been talking about the same clash for years without it moving closer, the conversation is a loop. Loops do not generate change — they generate tiredness. This signal, combined with any of the other three, is the most solid part of the picture.
Why the four signals usually appear combined
It is rare that only one of the four shows up. Most often they arrive together, in different intensities. A person with sustained damage to their body has probably also tried voice without result. A person with a values clash on big decisions has probably been carrying more weight than their share for years. The signals feed each other: the body deteriorates because the voice was not heard; the voice is not heard because values do not adjust; values do not adjust because no one has the energy for a conversation that no longer lands. That is why the map is read as a system, not as a checklist.
4. The noise — what does NOT decide
If the four signals carry weight, everything else falls into the category of noise until evidence says otherwise.
| Noise | Why it appears | Why it does NOT decide |
|---|---|---|
| Fear of being alone | The "alternative" fantasised from inside the relationship | Fantasised solitude is not a real alternative; it is a fear |
| Comparison with social-media couples | Exposure bias: only the highlight reel is visible | Nobody posts their grey Tuesdays |
| Guilt for doubting | The cultural script that "true love does not waver" | Doubting is evaluation, not moral diagnosis |
| Passing external crisis | Work, financial, health, move, bereavement | It has an identifiable start and usually resolves |
| A big fight | Intense and specific emotional activation | One fight is not a pattern; a pattern of fights is |
Noise does not get removed; it gets recognised. Recognising it separates a mature decision from a tired reaction.
Joel, Eastwick, and Finkel applied machine-learning models in 2017 to data from thousands of people to predict romantic outcomes. The uncomfortable finding: what people self-report predicts worse than what is measured with structure — internal perceptions of “a clear signal” are largely noise. DOI: 10.1177/0956797617714580.
5. The passing crisis — the one most often mistaken for an ending
Of all noises, the passing crisis is the hardest. An external crisis — work, grief, moving, illness — shuts down conversation and desire just like a dead relationship. The difference sits in markers: identifiable external cause, onset after that cause, partial response to stress relief.
The difference between a passing crisis and a real signal lies in four markers. It has an identifiable start. It is transient — has an expected resolution. It affects the system equally. And it ends when the external factor improves — if the relationship returns to its course when the crisis ends, it was noise.
If what you feel began at the same time as a specific crisis, and persists after the crisis has passed, the suspicion changes. What looked like noise starts to look like signal. Then it is worth applying the lens of the four signals.
6. Comparison with social media and guilt — the noise that costs the most
There is a noise that did not exist twenty years ago: comparison with idealised couples on social media. Nobody posts their month of awkward silence. Comparing your inside with their edited outside produces a false failure signal.
Guilt appears mostly when someone doubts. Doubting is not doing something wrong; doubting is evaluating. What can be a signal is the opposite: if guilt has disappeared — if staying relieves you more than leaving would — that absence is a stronger marker than guilt without reasons for it.
8. Exercise — the four questions to separate signal from noise
A simple exercise, to do honestly and without rush. It does not need a conversation with your partner. It can be done on a sheet of paper, in an afternoon, or across several afternoons. The idea is to pass each noise candidate through the lens of the four signals, one by one. If it passes the test, it is signal. If it does not, it is noise.
- Question 1 — Sustained damage: Is there anything in your health, sleep, energy, mood, or capacity to decide that has eroded in a sustained way since you have been in this relationship, and that is not explained by another cause? If yes, mark it. If not, next.
- Question 2 — Voice depleted: Were there serious conversations where you talked about what wasn't working, where there were agreements, and where the agreements did not hold over time? Has this pattern gone on for more than a few months? If yes, mark it. If not, next.
- Question 3 — Load asymmetry: If you stopped doing everything you do for this relationship — logistics, emotional, sexual, financial, project — would the system keep functioning on its own? If the honest answer is no, mark it. If yes, next.
- Question 4 — Values clash: Are there central axes of your life — children, location, money, shared project, way of living — where your partner's position and yours are sustainedly incompatible, and conversation does not bring them closer? If yes, mark it.
- Reading the map: if you marked all four, the signal is solid. If three, there is an important signal. If two, it is worth examining with care. If one or none, you are most likely facing noise — and it is worth checking whether an external crisis or an emotional peak is loading the moment.
Why this exercise is better done on paper, not in your head
When the evaluation is done only in your head, the noise sneaks in with the same authority as the signal. The tired body, the insomnia, the emotional peak, the sentence from a fight — everything feels the same. Writing it by hand or in a phone note, in four lines, one signal per line, with dates if you have them, helps separate the moment from the pattern. What is written holds up when the inner voice shouts. And it lets you come back in a week to check whether what you wrote still holds — or whether what looked like signal dissolved once the emotional peak dropped.
9. When the answer is Exit — and when it is not
The EVLN model does not say Exit is always the right answer. It says it is one of four valid responses, each with distinct consequences.
Exit is the right response when all four signals are sustainedly present, when voice has run dry, when the load is asymmetric, when values clash structurally. Exit is also the right response when there is violence — physical, sexual, psychological, financial. There the question turns into how to leave safely.
Exit is not the right response mid-external-crisis: decisions taken at the bottom of a stress curve carry documented bias. Voice with a date is — and if no movement follows, the signal stands clean.
Joel et al. (2017) reinforce from another angle: internal perceptions (“I feel nothing anymore”) poorly predicted actual breakups. What felt like certainty was partly mood.
10. Voice — when there is still a chance
If after mapping the four signals you find one or two present, but voice has not run dry — there is still possible conversation, still reciprocity available — voice remains a valid response.
Voice, at its best, has three components. First, a specific conversation about what concretely is not working. Second, a concrete, small, measurable agreement, with a short deadline to review. Third, a real willingness from both sides to sustain the agreement — not only promising it, but coming back to it when it fails.
If after one cycle of voice with those three components there is no movement, the response starts to stop being voice. But voice deserves to be tried at least once with structure before the decision is formalised. Especially if what you feel as “signal” is more like fear of being alone or comparison with other couples, which are noise.
11. The case — someone who saw a signal where there was only noise
To show how the map works in reverse, a case. We will call her M. M had been doubting her relationship for six months. On the surface, the relationship was functioning: amicable cohabitation, acceptable sex life, shared projects. Inside, M had a growing sense that it was already over.
M went through the four-signal map. Sustained damage: no. Voice depleted: not really. They had not even had the pending conversation. Asymmetry: something there, but not structural. Values clash: no, they shared the central axes.
What M discovered when making the map was that her “signal” came from noise. A comparison with a friend who had just gotten married. A diffuse fear of missing out. A tense week with a small conversation left unfinished. When put under the lens, none passed the four-signal test.
What M decided was to stay a bit longer, open the pending conversation, see what happened. The conversation did not go as expected, but it happened. And the sense of finality faded when the doubt stopped carrying itself alone. M did not stay out of endurance. She stayed out of choice, after ruling out the noise.
12. The case — someone who saw noise where the signal was clear
The reverse case also exists. We will call her C. C had carried for years a sense that something was off, but kept dismissing it as noise. Work crises. Financial stress. A move that did not go well. Raising small children. Every time the doubt appeared, C told herself: “it is the stage, it is the moment, it will pass.”
C went through the map. Sustained damage: yes — years of poor sleep, rising baseline anxiety, no energy for personal projects. Voice depleted: yes — the topic had come up at different stages, with different agreements, none sustained for more than a few weeks. Asymmetry: yes — carrying nearly all the emotional, logistical, and child-rearing load. Values clash: yes, on two central axes.
Four out of four. The signal was solid. What C had been calling noise for years was, in reality, a complete clinical picture. What C did afterwards was take a decision with structure, not at emotional peak. And discovered that the decision had been forming for years — what was happening was that she had been disguising it as noise to avoid looking at it.
13. When there is violence — the question stops being that
So far we have talked about relationships where the conflict is corrosive but not dangerous. There is one case in which the map loses clinical meaning as we have set it up: when there is violence. Physical, sexual, psychological, financial. Or any combination.
If you are in a relationship where there is violence — past or present, explicit or subtle, isolated or repeated — the decision to stay is not yours in the sense this piece frames. Violence changes all four EVLN responses completely. It destroys voice — speaking becomes dangerous. It turns loyalty into a control mechanism. And it often pushes the person being harmed toward silence and toward the feeling that there is no way out.
If this part describes you, what comes next is not deciding between staying and leaving. It is reaching out for professional help immediately and activating a safety network. The crisis lines are at the end of this piece.
14. Summary — what to do with this piece
If you made it this far without doing the exercise, it is worth doing: four columns — signal, since when, what you tried, what changed. The “since when” column separates a passing crisis from an installed pattern.
If you need to decide with a bit more structure but do not yet want to talk to anyone, the hub’s map can give you a first map. Rusbult, Johnson, and Morrow’s (1986) paper is available open access. And if what you are living includes violence, physical fear, or any situation where your safety is at risk, what is urgent is talking to someone who can help you today.
15. Caja de crisis
⚠️ If what you are living includes violence, physical fear, or any situation where your safety is at risk, the conversation this piece proposes can wait. What is urgent is talking to someone who can help you today.
| Country | Line | Type |
|---|---|---|
| 🇺🇸 United States | 988 | Suicide & Crisis Lifeline (free, 24h) |
| 🇬🇧 United Kingdom | 116 123 | Samaritans (free, 24h) |
| 🇨🇦 Canada | 1-833-456-4566 | Crisis Services Canada (free, 24h) |
| 🇨🇴 Colombia | 123 | Emergency line (free, 24h) |
| 🇨🇴 Colombia | 106 | Psychosocial guidance line (free, 24h) |
| 🇲🇽 Mexico | 5255 2598 1212 | Línea de la Vida (free, 24h) |
| 🇪🇸 Spain | 024 | Teléfono de la Esperanza |
| 🌍 International | findahelpline.com | Global directory |
If there is violence in the relationship, the question is not “do I stay or leave?” — it is safety, and these lines are the door. Nobody deserves to live in fear. Asking for help is not weakness; it is the most concrete move you can make today.
If you prefer professional support with a clinical perspective, the team at RDK Terapia offers initial orientation sessions that can help you map the four EVLN signals with professional support.