A few months ago, I learned that I probably can’t feel the emotions signalled by oxytocin, the "love hormone". This raises lots of interesting questions - what things I do and don’t feel, how the world looks different from an oxytocin-less perspective, how a lack of oxytocin changes one’s values or goals, etc. But it would be putting the cart before the horse to dive into those questions without first walking through how I learned this about myself and the evidence, so that everybody has an appropriate level of confidence in the underlying assumption.

The Evidence Which Privileged The Hypothesis

It started with investigating a confusion. Lots of the supposedly-happy relationships around me looked pretty awful to my eye, so why the heck were people (apparently) happy with them? What on earth was making these relationships net positive, let alone good?

I wrote a few LessWrong pieces on that confusion, and eventually Caleb Biddulph responded with a hypothesis: perhaps I don’t actually feel much of the thing most commonly called “(companionate) love”, and have therefore been confusing it with something else which I do feel. Caleb also spelled out the physical sensations he experiences with love, and sure enough… his description was not at all familiar to me.

I asked a few other people to describe what companionate love feels like, physically. Sure enough, it did not sound like anything I ever remember feeling.

I had previously asked some people with relationships which seemed bad to me what the major sources of value were from their relationships. Various pointers to “intimacy” topped the list. If that whole intimacy thing was a feeling which I couldn’t experience… as opposed to a cluster of practical benefits, as I’d previously conceptualized it… that sure would explain why these supposedly-happy relationships around me looked pretty awful to my eye!

My ex (of a 10 year relationship) had explicitly hypothesized from time to time that I just didn’t feel love like normal people do. I don’t think either of us had taken that hypothesis completely seriously, but…

Looking back on my childhood, it was clear for a long time that I didn’t form bonds the usual way. I didn’t react the usual way to the deaths of pets. I was eager to get away from my parents at a younger age than normal.

It just made a whole lot of sense.

And physiologically, the obvious guess for what would cause a lack of companionate love was a problem with oxytocin signalling.

Background: Oxytocin

This section is my current gestalt understanding of oxytocin. Take it with a grain of salt.

Oxytocin is often called “the love hormone”. Specifically companionate love - there’s a different hormone (vasopressin) associated e.g. having a crush, new relationship energy, limerance, etc (all of which I do feel). Early work on oxytocin found it released in mothers during breastfeeding, triggering and reinforcing the mother-child emotional bond. Over the years, it’s been associated with lots of other flavors of companionate love.

My current best guess is that oxytocin is the main hormonal signal underlying anything people describe as a feeling of “deep connection”. This includes standard examples of companionate love, like e.g. the love one feels for family. But (I claim) it also includes things like:

• Weaker versions of “deep connection”, like the feelings induced by intentional relating exercises.
• The feeling of connection one gets when deeply empathizing with another.
• Feelings of deep connection to one’s community, God, or the universe.

The Genetic Evidence

The natural next step was to get my whole genome sequenced, and check my oxytocin and oxytocin receptor genes. I checked the receptor first - it’s a much bigger gene, so a more likely place for a breaking mutation to appear.

Sure enough, there was a single base pair deletion 42 amino acids in to the open reading frame (ORF) of the 389 amino acid protein. That induces a frameshift error, totally messing up the entire rest of the protein. And I did do some basic sanity checks - the sequencing had plenty of depth (i.e. it probably wasn’t noise), and other genes I spot checked did not have any frameshift-inducing mutation.

But that’s not yet a full story. Humans have two copies of each gene, and only one copy had that particular frameshift error. The frameshift error would mess up the protein in a way which triggers nonsense-mediated decay, so the mutated sequence shouldn’t be transcribed much. So by itself, the frameshift mutation should just leave me with 2x less oxytocin receptors than usual (which is usually not a huge deal for signal function), or maybe even closer to normal if there’s any feedback control on receptor density. Upshot of all that: in order for oxytocin signalling to be very broken, there would have to also be some function-breaking mutations on my other copy of the receptor gene.

And there were some other mutations (substitutions, nothing as obvious as a frameshift), a couple of which were predicted by alphafold to be pretty deleterious to the protein’s function…

… but unfortunately today’s standard sequencing technology doesn’t let me know which copy of a gene a mutation is on. We sequence by chopping DNA up into little chunks, sequencing the little chunks, then stitching it all together computationally. But since two copies of the same gene have mostly the same sequence, the stitching step can’t tell which copy a mutation is on, just that it’s on one of them.

The shortest way around this is to get ones’ parents’ genomes also sequenced. If one subset of my mutations appears in one parent, and another subset appears in my other parent… well then, I know that the one subset is on one copy, and the other subset on the other copy (with high probability).

So I got my parents’ genomes sequenced. One of them had the frameshift-inducing mutation, as expected. The other had a few substitutions which I share. Alas, that parent's substitutions… were also shared by my other parent^[1]! Which means I can’t fully nail things down with the available information: I don’t know for sure whether the substitutions I have were on the same copy as the frameshift and I have another healthy copy, or if they were on the other copy from the other parent.

That’s my current state of knowledge.

Recap of the key pieces:

• One copy of my oxytocin receptor gene is definitely completely broken.
• There’s a couple substitutions which would likely break function, but I don’t know for sure which copy they’re on.
• Even if the second copy does have the likely-function-breaking substitutions, I don’t know whether the result is a complete absence of oxytocin signalling or just very weak oxytocin signalling.

Combined with the evidence that made me privilege this whole hypothesis in the first place, I’m pretty confident that my oxytocin signalling is either very weak or entirely absent. But I am relying at least partially on the less-legible evidence which made me privilege the hypothesis; the genetic evidence alone is damn strong evidence in favor of the hypothesis but not fully conclusive on its own.

1. ^{^}

   At this point I wondered if I was using a dubious reference genome, and finding “substitutions” relative to a reference genome which was itself nonstandard. I asked an LLM and its answer was basically “no”, the reference genome is a consensus genome.