Attachment
On backpacks, paneer, parents and half-lives.
26 October 2026
We humans can get attached to pretty much anything and anyone. It can be a stuffed toy just lying on your bed. It can simply be clothes, or a backpack. I really don’t like changing my backpack very often. It can also be an animal. Attachment doesn’t even need to be with some “thing”. It can be with your routine, or with which side of the bed you sleep on. I haven’t changed my side of the bed for a long time, and I’m not planning to either. Some people are quite malleable about their routine, and some will have an emotional breakdown if their routine doesn’t go as planned.
You can also get attached to songs, to the feeling they bring to your mind, hitting those right receptors every time you listen to them. And of course we get attached to many people in our lifetime. First our parents and siblings, then the friends we make in school, college and office, and also people we meet online. Sometimes we don’t even recognise that we are attached until that person is absent. Perhaps that’s the curse of attachment, that you feel it the most when you are not with that person. It’s also difficult to tell whether we are attached to a person or to the joy we feel when we are with them. The two overlap a bit.
The brain builds a model
Attachment to a person is pretty different. A person is dynamic, and that’s why this attachment can affect our brain so much. It doesn’t happen in a moment. It happens over time. In some way, our brain restructures itself over time. When we are talking to someone, meeting them or seeing them daily, we get to know little little details about them. When do they wake up? When do they sleep? What do they usually eat? We process this either consciously or subconsciously. Our brain likes data a lot. It takes all this data and builds a “predictive” model, which tries to predict every little thing about the person. What will they eat today? (Oh, they ate paneer yesterday, so there’s a good chance they won’t eat paneer today.)
Say V is what my brain predicts, for example how much they will talk to me today, and r is what actually happens, both from 0 (not at all) to 1 (a lot). The surprise is the gap between the two:
Every day, the brain moves its prediction a little towards what actually happened.
When the prediction breaks
Earlier I told you how some people have an emotional breakdown when their routine doesn’t go as planned. A bit of that can be explained by the predictive nature of their minds. Whenever things don’t go as predicted, they feel some sort of pain.
On day 6 my brain has barely learned the pattern, so a quiet day is a small surprise. On day 45 the prediction is close to 1, and the same quiet day is the biggest surprise on the chart. But does the surprise correlate with hurt? No. Prediction alone doesn’t explain this.
We can try to explain the pain and joy of attachment using neurotransmitters like dopamine, oxytocin and vasopressin. These chemical systems influence how much significance the brain assigns to social experiences, how it learns from them, and how it responds to the availability or loss of someone we are attached to.
The link between dopamine and prediction goes back to the 1990s. Wolfram Schultz found that dopamine neurons in monkeys fire for an unexpected reward, stay quiet for an expected one, and dip below normal when an expected reward doesn’t come [1]. This is exactly δ, measured in a brain.
So the hurt needs a second factor, the emotional significance E: how much this person matters to my brain. The surprise only hurts as much as E lets it:
A big surprise from someone with a small E barely hurts. A small surprise from someone with a big E can hurt a lot.
Time is not the whole story
There is probably variation between brains in the flux of chemicals released for a given social experience, and we don’t know if all brains react the same way to a given flux. But combined, the flux and the brain’s reaction to it form the key part of the feedback loop of the prediction model. You might be able to predict everything about a colleague you have been seeing daily in office for years, when they eat, when they go home, and still not be bothered much if they resign. On the other hand, you can get intensely attached to someone you met only a few months ago, based on the sensitivity of your brain and the flux of neurotransmitters released on exposure. In other words, the colleague has a small E, and the new person a big one.
From presence to security
The role of time in attachment is very interesting. Being with someone for a long time doesn’t directly correlate with how intense the attachment is. In fact, take the example of our parents. When we were children we were intensely attached to them, and their absence used to make us cry. Then we grew up, went to college, moved to a different city to work, and now we can continue our daily work without missing them a lot. Does that mean the attachment with them faded away? No. The joy of seeing them every day has faded, but the attachment hasn’t, so it was never just the joy we were attached to. It was them. With time the attachment evolved. It became a sense of security for the brain, and it’s not just dopamine release now. We don’t predict their actions anymore, but any thought of something happening to our parents can cause a massive breakdown.
Attachment can remain extremely strong even when the need for someone’s constant presence disappears. The prediction model may gradually shift from predicting someone’s behaviour to predicting their availability and reliability. With time, an intense attachment stops meaning the intense feeling of meeting them again and again, and it probably turns into a sense of security. A stronger sense of security can sometimes make physical separation easier, not harder. So the amount of distress caused by someone’s absence isn’t necessarily proportional to the strength of the attachment.
The same thing happens in a romantic relationship. Think of the surprise bars from Figure 1. As we move towards security, the surprise bar for a day of not talking actually decays. In the first few months, a day without talking is a big negative bar. A few years in, the brain doesn’t predict “they will talk to me today” anymore. It predicts “they are there if I need them”, and a quiet day doesn’t break that. So the bar shrinks towards zero, even though the bond keeps getting stronger.
Attachment is also not just about the distress it causes in someone’s absence. Take a friend from school. You were very attached to them back then, and eventually you both took separate paths. You can still hold a significant attachment to them. You will still think about their overall well-being, but without any distress.
De-attachment is a different dimension
We have discussed attachment a decent amount, so now de-attachment. At first thought, we tend to think de-attachment is just the reverse of attachment. But is it?
I like to think that de-attachment is a separate dimension. Call the state of your mind just before you meet someone you will like in the future S0, and the state when you are deeply attached to them S1. Now suppose, for some unfortunate reason, you can no longer contact them. Will your brain go back to an S0-like state? As in, does the bond completely reverse? I think the bond may not disappear. What weakens with time is our dependence on that bond. The feedback loop fades, and so does the impact of that bond on us, but that doesn’t mean the bond is gone. This process is de-attachment. It takes you to a new state, call it S2, which is not S0. So de-attachment might be orthogonal to attachment.
Psychologists who study learning have seen something similar. When an expectation stops coming true, the old learning isn’t erased. The brain learns “not anymore” on top of it, and the old learning can come back later [2].
Every event has a half-life
I like to say this quote: “every event has a half-life”. The pain we feel on day 0 is not the same as what we feel on day 100. Its effect fades away, and who knows, it might follow the same trend as radioactive decay, with some rate β:
With a half-life of 30 days, the pain on day 100 is about a tenth of day 0. It never quite reaches zero.
Of course, not all de-attachment is caused by a sudden change in dynamics. Sometimes life gets busy, exposure fades slowly over time, and eventually the significance of that bond fades away too.
Clinging is not attachment
When we say attachment is not good, I think we are often referring to the hurt that comes from clinging, possessiveness, or dependence, rather than attachment itself. We tend to overlook the sense of security that attachment can provide. The same bond that makes us vulnerable to someone’s absence can also make us feel secure simply knowing they are there.