Intermittent Reinforcement: Why You Cannot Stop Chasing Someone Who Hurts You

Intermittent reinforcement is when someone treats you well only some of the time. Not always. Not never. Just often enough to keep you hoping. Your brain reads that pattern as a reason to keep trying, which is why the most confusing relationships are often the hardest ones to leave. Here's what the pattern is, why it works on smart people, and how to get out from under it.

Let me say the uncomfortable part first. If you're stuck on someone who is warm one week and distant the next, the problem isn't that you love them too much. The problem is that you're on a schedule. And that schedule was built to be sticky.

What Is Intermittent Reinforcement?

The idea comes out of learning research. Give an animal a reward every single time it presses a lever and it learns fast. Stop the reward, and it gives up fast too. But give the reward only some of the time, at random, and something odd happens. The animal keeps pressing long after the rewards stop.

That gap has a name. Researchers call it the partial reinforcement extinction effect. Work summarised in the American Psychological Association teaching materials on conditioning and learning describes the core finding plainly. Behaviour kept going by an on and off reward fades much more slowly than behaviour rewarded every time.

Read that again, but about your relationship. The inconsistency didn't weaken your attachment. It strengthened it.

This isn't a metaphor that got stretched too far. C. B. Ferster and B. F. Skinner mapped it in Schedules of Reinforcement in 1957, a 741 page book built on more than 900 cumulative record figures from experiments run at Harvard between 1949 and 1955. Of the four schedules they tested, the variable ratio schedule, where the reward lands after an unpredictable number of tries, produced both the highest response rates and the behaviour that was hardest to extinguish. That's the same schedule a slot machine runs on. It's also, unfortunately, the one a hot and cold partner runs on.

Do Humans Actually Behave This Way, or Just Lab Animals?

Fair pushback. Pigeons and rats aren't people, and a lot of pop psychology jumps straight from a 1950s lever box to your text messages without ever checking whether the leap holds.

It holds. Robin Hogarth and Marie Claire Villeval ran the test on humans with real money. Their study, circulated first as IZA discussion paper 5103 and later published in the Journal of Economic Behavior and Organization in 2014 (volume 100, pages 1 to 19), paid people to work through a real effort task and let them quit any time they liked. Three groups. One got paid every single period. One got paid one period in three on a schedule they could see coming. One got paid one period in three at random.

Then the researchers switched the money off without warning.

The continuously paid group walked almost straight away. No pay, no point, done. Both intermittent groups kept going, and put in more total effort, long after the reward had stopped completely.

Now move that into a kitchen at eleven at night. Someone who was warm every day and then goes cold is easy to read as a change, so you react to it. Someone who was warm one week in three has already taught you that a quiet stretch means nothing. So you wait it out. You've waited before. It paid off before.

That's not a character flaw. It's the same result Hogarth and Villeval pulled out of people who had nothing at stake but a small cash payment.

One more detail from that study, because it's the part people skip. The random schedule didn't just make people persist. It made their effort erratic, and it left them worse off overall than the steady schedule did. Persisting and winning are not the same thing.

Why Does Random Kindness Hook You Harder Than Steady Kindness?

Here's the part that stings. A partner who is kind every day is easy to leave. Not because you want to leave. Because you're not anxious. You know what tomorrow looks like. Your brain isn't running background calculations about them.

A partner who is kind twice a month is a puzzle. And your brain hates an unsolved puzzle. So it does what brains do. It looks for the pattern. Was it because you didn't text first? Because you wore that shirt? Because you brought up the thing at dinner?

You start running experiments. Each experiment is another reason to stay a bit longer and find out.

There's a second thing going on. When rewards are rare, they feel bigger. A normal apology from a reliable person is just an apology. The same apology from someone who never gives them feels enormous. You're not measuring their behaviour against a fair standard. You're measuring it against their worst week.

That's how the bar drops. Slowly, and without you agreeing to it.

There's brain data underneath this. Christian Fiorillo, Philippe Tobler and Wolfram Schultz recorded dopamine neurons directly and published the results in Science in 2003 (volume 299, pages 1898 to 1902). They found a signal that tracks uncertainty rather than reward, and it peaks when the chance of a reward sits at 50 percent. Not at 100 percent. At a coin flip. You can read the paper hosted by the University of Cambridge, where Schultz runs his lab. Put plainly, your brain is built to fire hardest when it genuinely can't call the outcome. A partner who is reliably warm gives you nothing to compute. A partner who might be warm gives you a puzzle your dopamine system is designed to chase.

Does the Same Reward Stop Working Once It Becomes Predictable?

Close to it, and there's a human brain scan study that shows it about as cleanly as this kind of research ever shows anything.

The Fiorillo work above recorded neurons in monkeys. So the obvious question is whether people work the same way. David Zald and colleagues at Vanderbilt ran the test on humans and published it in the Journal of Neuroscience in 2004 (volume 24, issue 17, pages 4105 to 4112). Healthy volunteers picked cards for real money inside a PET scanner while the researchers tracked dopamine in the striatum.

Here's the design detail that makes it worth your time. There were three conditions. In one, a dollar showed up unpredictably. In another, people were paid on every fourth card, so they knew exactly when it was coming. The third was a control with no money at all. The first two paid out at the same rate. Same cash, same frequency. The only thing that changed was whether you could see it coming.

The unpredictable condition moved dopamine in the striatum. The predictable one produced no significant change. Same money, and the brain barely registered it as an event.

Now carry that back into your relationship, because it explains something that probably bothers you.

You've likely had the experience of someone treating you well, consistently, and feeling almost nothing. No spark. No pull. And you concluded there was no chemistry, or that you're broken, or that you only want people who don't want you.

Look at what the scanner is doing instead. The steady person isn't giving you less. They're giving you the every fourth card version. Fully predictable, and so your reward system stops flagging it. The unreliable one is running the other condition, and the machinery lights up every time. That difference in how it feels isn't a readout of how good the two people are for you. It's a readout of how hard they are to predict.

So is unpredictable simply more dopamine?

No, and the study is more interesting than that. The unpredictable condition didn't just crank everything up. Dopamine transmission rose in one part of the striatum, the caudate, and fell in a neighbouring part, the putamen. The researchers had predicted the increase from the animal work. The suppression surprised them.

So the honest version is not that uncertainty floods you with reward chemistry. It's that uncertainty makes your brain do something complicated that predictability doesn't do at all. This was also a small PET study from 2004, and one scan is not a settled fact about your relationship.

Luke Clark and Martin Zack, whose review is cited further down, draw a distinction that's useful for keeping this straight. Reward variability is the objective thing, the actual pattern of when something arrives. Uncertainty is your subjective experience of it. The pattern is out there in their behaviour. The uncertainty is the thing you're living inside, and it's the part that hurts.

Does This Only Happen in Romantic Relationships?

No, and that's worth knowing, because it takes some of the shame out of it. The same schedule runs a parent who is affectionate on good days and unreachable on bad ones. It runs the boss whose praise arrives once a quarter and keeps a whole team over performing. It runs friendships where you're the one who always reaches out.

It also runs most of your phone. Luke Clark at the University of British Columbia and Martin Zack at the Centre for Addiction and Mental Health published a review in Addictive Behaviors in 2023 (volume 140, article 107626) arguing that reward variability is a design feature, not an accident. They trace the same variability that drives gambling into gaming, shopping, social media and other digital products, and they make a point that lands hard in this context. Products get more compelling when the reward is both frequent enough to keep you engaged and uncertain enough that you can't predict it. You can read the record on PubMed, run by the US National Library of Medicine.

So when you catch yourself checking whether they replied, notice what your hand is doing. It's the same motion as a pull on a lever. That's not an insult. It's just what an uncertain reward does to a nervous system, and slot machine designers have been paid very well to understand it.

Here's why that reframe helps. If you've been telling yourself you're weak for not being able to walk away, you're competing against a mechanism that entire industries have engineered on purpose. Willpower was never the thing being tested.

What Does It Look Like in a Real Relationship?

It rarely looks like abuse from the outside. It mostly looks like a person who is hard to read. See how many of these land.

• The good weeks are genuinely good. They're present, funny, warm. You think the hard part is over.
• Then the temperature drops for no reason you can name. Short replies. Busy. Tired. Nothing you can point at.
• You start managing yourself. You read your own messages twice before sending. You wait a bit so you don't seem needy.
• Big gestures arrive right when you're about to give up. A long apology. A trip. A serious talk about the future.
• Your friends have stopped asking. They ran out of ways to say the same thing.
• You describe them in two halves. There's the real them, and then there's how they have been lately. You have been saying lately for two years.
• Relief feels like love. When the cold spell ends, the rush you feel is huge. That rush isn't closeness. It's the alarm switching off.

That last one deserves a minute. If your best feeling in a relationship is the moment the anxiety stops, you're not measuring happiness. You're measuring the distance between fear and normal. A steady relationship doesn't have that spike, and at first it can feel boring. It's not boring. It's what safe feels like when you're not used to it.

You may also notice this pattern in dating that never quite becomes a relationship. If that sounds familiar, our guide on the psychology of situationships covers the same engine running in a different car.

Why Do You Hold On Tighter When It Gets Worse?

Most people expect the opposite. Treat someone badly and they leave. That's not what happens. As the good moments get rarer, the chasing often gets stronger. There are a few reasons.

You have already paid. Months or years of effort sit behind you. Leaving means calling all of it a loss. So you stay, partly to protect the investment. This isn't weakness. It's a very human way of counting.

You have proof it can be good. This is the cruel one. You're not chasing a fantasy. You met the warm version. They were real. So you're not being naive when you believe they exist. You're just wrong about how often you get to have them.

Your nervous system got loud. Living on alert changes your baseline. Calm starts to feel like something is wrong. Our guide on nervous system regulation goes deeper on why your body does this and what settles it.

You started blaming yourself. Because if the cause is you, then the fix is you, and that feels better than accepting you can't control it. Self blame is a way of keeping hope. It's also exhausting.

Some people who run this pattern do it knowingly. A 2016 study by Back and colleagues in the Journal of Personality and Social Psychology, listed on PubMed, split narcissism into two parts. One part draws people in through charm and admiration seeking. The other pushes people away by putting them down. The researchers found this explains a real puzzle. The same person can be great at starting relationships and bad at keeping them. If you've wondered how someone could be so magnetic and so cold, that's your answer. Both are them.

Not everyone who runs hot and cold is a narcissist, and that label is thrown around far too easily. But if the quieter version fits better, our guide on covert narcissism is a more careful look.

What Does the Research Say About Intermittency and Abuse?

Start with the scale, because people living inside this pattern usually assume they're an unusual case. The Centers for Disease Control and Prevention runs the National Intimate Partner and Sexual Violence Survey, and its 2016 to 2017 round surveyed 15,152 women. It found that 49.4 percent of women have experienced psychological aggression by an intimate partner in their lifetime, and 47.3 percent have experienced contact sexual violence, physical violence or stalking by one. Those figures are set out in a National Academies of Sciences, Engineering, and Medicine report hosted by the National Library of Medicine. Roughly half. Whatever else you are, you're not the odd one out.

This is where the lever pressing research stops being an analogy. Donald Dutton and Susan Painter proposed a theory they called traumatic bonding, and the claim was specific. Strong emotional attachments form out of two ingredients, not one. You need a power imbalance, and you need the bad treatment to come and go rather than stay constant.

They tested it. Their study, published in Violence and Victims in 1993 (volume 8, issue 2, pages 105 to 120) and indexed by the National Library of Medicine on PubMed, assessed 75 women who had recently left abusive relationships, then followed up again six months later. They measured attachment to the former partner, self esteem, and trauma symptoms, against things like the intermittency of the maltreatment, the power differential, and how much the woman had invested.

The finding that matters for you is this. How often the abuse happened mattered less than how unpredictably it happened. Intermittency was the stronger predictor of how attached these women still felt after leaving, and it was still doing work six months on.

Sit with that for a second. A relationship that was consistently bad would have been easier to leave behind. The good stretches were not the redeeming part. They were the part that made it stick.

That's also the honest answer to a question people ask themselves at three in the morning. If it was really that bad, why do I miss them? Because the mechanism that built the attachment was never measuring how bad it was. It was measuring how unpredictable it was. Our guide on trauma bonding walks through what that attachment looks like from the inside, and why it can outlast the relationship by years.

One caveat, because the evidence isn't all one direction. This is correlational research on people who had already left, so it can't prove intermittency caused the attachment rather than travelling alongside it. And as noted at the end of this article, the extinction effect that looks so clean in animal studies doesn't always reproduce in applied clinical work. The pattern is real and well documented. It just isn't a law of physics.

How Is This Different From a Normal Rough Patch?

Fair question. Every relationship has bad months. Someone loses a job. Someone gets sick. People go quiet. That's not what this is.

Here's how to tell them apart.

• A rough patch has a cause you both agree on. This pattern has a cause you've to guess at, alone.
• In a rough patch, they know they have been distant. In this pattern, raising it makes you the problem.
• A rough patch ends. This one cycles. Good stretch, cold stretch, repeat.
• In a rough patch you both feel bad. Here, only one of you seems to be tracking it.
• After a rough patch you feel closer. After a cold spell, you feel grateful. Those aren't the same feeling.

Gratitude is the tell. If a relationship keeps putting you in a position where normal treatment feels like a gift, something is off, no matter how much you love them.

How Do You Break the Hold It Has on You?

You don't break it by understanding it. You already understand it or you wouldn't be reading this. You break it by changing what you do.

1. Stop running tests. Every time you pull back to see if they chase, or send one more message to see if they answer, you restart the clock. The test is the behaviour keeping you stuck.
2. Write down the cold weeks while they're happening. Not for them. For you. Warm days rewrite your memory fast. A dated note doesn't.
3. Count in months, not moments. Ask how many of the last twelve weeks felt good. Not how good the best day was. The average is the relationship. The peak is the bait.
4. Expect it to get worse before better. When you stop responding, the pull spikes hard. You will feel panicky and certain you made a mistake. That feeling is the pattern fighting back. It's not evidence.
5. Cut the drip. Half contact keeps the schedule alive. Occasional check ins are the schedule. If you want out, you need a real gap.
6. Get people back around you. This pattern shrinks your world quietly. Tell one person the truth, including the embarrassing parts.
7. Work on the ground you stand on. This is slower work, and it matters most. Our guide on building self worth covers why the bar drops in the first place, and how to raise it back.

One more thing. You don't need them to admit anything. Waiting for the moment they finally see what they did is its own trap, and it can cost you years. You can leave without the apology. Most people do.

When Should You Get Outside Help?

Reach out if you can't stop even though you want to. If your sleep, work, or health has slipped. If you've started hiding the relationship from people who love you. If you feel unsafe in any way at all.

None of those mean you're broken. They mean the pattern is doing what it's designed to do, and doing it well. A therapist can help you see the shape of it in real time, which is much harder to do alone when you're inside it. If you want structured tools to work through between sessions, PositivePsychology.com has practitioner grade worksheets on boundaries, self trust, and letting go.

And if the cold periods have included threats, control of your money, or anything that frightened you, please contact a local domestic abuse line. Intermittent reinforcement is very often part of a bigger pattern. Our guide on trauma bonding explains what that bigger pattern looks like.

What Else Do People Ask About Intermittent Reinforcement?

What is intermittent reinforcement in simple terms?

It means the good treatment shows up sometimes, not every time. You never know which version of the person you'll get. Because the reward is random, your brain keeps trying. Steady kindness is easy to walk away from. Random kindness is very hard to walk away from. That's the whole trick, and it works on almost everyone.

Why is a hot and cold person so hard to get over?

Because you're not getting over a person. You're getting off a schedule. Your brain learned that waiting pays off eventually. Research on partial reinforcement shows that behaviour rewarded only some of the time takes much longer to fade than behaviour rewarded every time. So the on and off treatment made the pull stronger, not weaker.

Do people do this on purpose?

Some do. Some don't. Plenty of people run hot and cold because they're avoidant, overwhelmed, or repeating what they grew up with. The effect on you is the same either way. You still lose sleep, still check your phone, still shrink yourself. Intent matters for how you feel about them. It doesn't change what the pattern is doing to you.

How do I stop chasing someone who runs hot and cold?

Stop testing whether the good version will come back. Every test resets the clock. Cut the contact that feeds the loop, write down the cold weeks so the warm days can't erase them, and expect to feel worse before you feel better. That spike of panic when you stop is normal. It's the pattern fighting to keep you, not proof that you should stay.

Is intermittent reinforcement the same as a trauma bond?

They're close but not identical. Intermittent reinforcement is the mechanism, the on and off reward schedule. A trauma bond is the attachment that mechanism builds over time, usually when the cold periods involve real harm. Think of it this way. Intermittent reinforcement is the engine. The trauma bond is the car it drives.

Sources: American Psychological Association, supplemental technical materials for Essentials of Conditioning and Learning, chapter 10 (apa.org). Ferster, C. B., and Skinner, B. F. (1957), Schedules of Reinforcement, Appleton-Century-Crofts. Fiorillo, C. D., Tobler, P. N., and Schultz, W. (2003), "Discrete coding of reward probability and uncertainty by dopamine neurons," Science 299, 1898 to 1902 (pdn.cam.ac.uk). Zald, D. H., et al. (2004), "Dopamine transmission in the human striatum during monetary reward tasks," Journal of Neuroscience 24(17), 4105 to 4112 (jneurosci.org). Back, M. D., et al. (2016), "Narcissism and romantic relationships: The differential impact of narcissistic admiration and rivalry," Journal of Personality and Social Psychology (pubmed.ncbi.nlm.nih.gov/27560609). Dutton, D. G., and Painter, S. (1993), "Emotional attachments in abusive relationships: a test of traumatic bonding theory," Violence and Victims 8(2), 105 to 120 (pubmed.ncbi.nlm.nih.gov/8193053). Hogarth, R. M., and Villeval, M. C. (2010), "Intermittent reinforcement and the persistence of behavior: experimental evidence," IZA discussion paper 5103 (iza.org), published in revised form in the Journal of Economic Behavior and Organization (2014) 100, 1 to 19. National Academies of Sciences, Engineering, and Medicine (2024), Essential Health Care Services Addressing Intimate Partner Violence, reporting CDC National Intimate Partner and Sexual Violence Survey 2016/2017 figures (ncbi.nlm.nih.gov/books/NBK605464). Clark, L., and Zack, M. (2023), "Engineered highs: reward variability and frequency as potential prerequisites of behavioural addiction," Addictive Behaviors 140, 107626 (pubmed.ncbi.nlm.nih.gov/36701907). Worth noting for balance: Lerman, Iwata, Shore and Kahng (1996), Journal of Applied Behavior Analysis 29, 153 to 171, tested intermittent versus continuous schedules in three adults in a clinical setting and found no significant difference in how quickly the behaviour extinguished, so the lab effect does not reproduce in every applied context.

For self-reflection purposes only. Not a substitute for professional mental health support. If you feel unsafe, please contact a local domestic abuse or crisis helpline.