Cold Takes

You can't unhear it

31 July 2026 · Everything here is sourced, and what didn't hold up is listed at the bottom next to what did.

I write music. Not as a career, and I'd never claim to be a professional recording artist, but I will call myself an artist, because of the hours that go into the writing. And it's been enough to ruin something for me, though I've been trying to work out whether ruined is even the right word for it.

Here's the thing I keep going back and forth on. When you're listening to a song, how much of it is actually the words? The perspective, what the person is saying, where their head was at. And how much of it is just the feeling, where you hear it and it does something to you and the lyrics are almost beside the point.

I genuinely don't know. And I've started to wonder whether I'm even a reliable witness any more, because since I started writing I can't hear anything without also hearing how it was put together. Which might just be a selfish way to listen. So I went looking for whether anyone had actually studied this.

They have, and the answer turns out to be more interesting than either side of the argument I thought I was having.

The feeling gets there first

There's an experiment from this year where fifty people listened to song excerpts, then had to say whether a word flashed up on a screen was a real word or nonsense. Some of those words were connected to the music of what they'd just heard. Some were connected to the lyrics. Whichever channel is doing the heavy lifting should make its words easier to process.

Music won. About 637 milliseconds for music-linked words against 655 for lyric-linked ones. That gap is eighteen thousandths of a second, which is small enough that I'd call it a direction rather than a knockout.

I want to be careful with that, because it isn't the whole picture. Other research on mood finds lyrics pulling just as much weight as the music does, and sad words over a bright beat will still take you down. The version I've landed on is that the music moves you and the lyrics tell you what to do with having been moved. They're two different jobs, and which one matters more depends on what you needed the song for that day.

But it depends whose ear it lands in

This is the bit that actually answered my question, and it isn't about the song at all.

Researchers put eighteen trained musicians and eighteen non-musicians in a scanner and played them the same piece of music, then watched which parts of the brain kept lighting up. The two groups did not listen to the same song the same way. They weren't really using the same equipment.

The non-musicians leaned on the network tied to personal memory and self-referential thinking, which the researchers describe as a reliance on autobiographical and emotional processing. In plain terms, the untrained ear runs a song through where was I when I first heard this, and who was I with.

The musicians didn't do that at all. They lit up frontal regions, reward, and the areas that tie hearing to movement. That ear runs a song through how is this built, and could I have done that.

So the whole bars-or-vibe argument isn't really two camps disagreeing about music. It's two camps having genuinely different experiences and using the same word for both.

What it costs

If you write at all, even badly and even occasionally, you've already started sliding from one group into the other. I don't think that's an upgrade, though. I think it's a trade.

What you lose is the version where a song is just weather that happens to you. Once you know how a verse gets built you can't fully stop seeing the joins, and that memory route goes quiet, the one where a song is really a person or a summer. Nobody warns you about that part, and you don't get it back.

What you get instead is that things start impressing you that used to just wash over you, and not because they got better. It's because you finally know what they cost to make. I'll hear something now and the thought is genuinely damn, how did you even think of that, and that thought is only available to me because I've sat there failing to think of it myself.

Which brings me to choruses

I cannot write a chorus to save my life. Verses I can grind out, but a hook either shows up or it doesn't, and when it doesn't there is absolutely nothing you can do about it. So I have enormous respect for people who can, and I always assumed that was just my own hang-up talking.

It isn't, though. Someone measured it. They took four pop songs and cut each one into twenty pieces, isolating single layers so that one line or one figure could be judged on its own, then had people rate how much each piece stood out and stuck. Three things predicted a high score: being from the topline rather than the backing, being from the chorus, and containing what musicologists call a compound hook.

So the hardest part to write is also the part carrying the whole song. Which means that not being able to write hooks isn't a hole in my judgement of them, it's the reason my judgement is any good. I'm the one who knows how far that shot actually was.

So which is it

I don't think it's a real question, and not because it comes down to taste. It's because the question assumes there's one listener, and there isn't. There's the you that hears a song and remembers a person, and there's the you that hears it and counts the bars. Both of them are you. They just don't get handed the same record.

And if you're stuck between the two, I don't think that's confusion. I think it's what it feels like to be someone who both listens and writes, and it might be the most honest place to be standing in the whole argument.

The receipts

What the evidence supports, what's shaky, and what I believed going in that turned out not to be backed. No Cold Take runs without this block.

Holds up

• Music primes meaning faster than lyrics, and lyrics can dampen emotional response. Lexical decision task, 50 listeners; ~637ms vs ~655ms for music-linked vs lyric-linked targets. The paper also reports lyrics reducing the intensity and positivity of emotional ratings. — Frontiers in Psychology, 2026
• Musicians and non-musicians engage different brain networks listening to the same music. Non-musicians showed greater default mode network engagement, which the authors read as autobiographical and affective processing; musicians showed more frontal, reward and sensorimotor engagement. — Frontiers in Neuroscience, 2025
• Toplines, choruses and compound hooks are rated more memorable and salient. Four pop songs, twenty isolated excerpts each, rated by listeners for salience and memorability. — Byron, Rushworth & Stewart, Music Perception, Feb 2025

Shaky, so don't build on it

• The musician study is small, and it isn't about rap at all. Eighteen people per group, and one piece of music, a Piazzolla tango. The authors flag the sample size themselves and note their analysis method is sensitive to a parameter they had to pick arbitrarily. The best explanation I can find for how I hear rap was run on a tango.
• The priming study is narrow and says so. Fifty native French speakers, one genre, one language, one composer. An 18ms effect out of that is a lead, not a law.
• "Lyrics matter as much as melody for mood" comes from a different literature. It's a real counterweight, but it isn't measuring the same thing as the priming work, so the two don't cleanly cancel out.

Didn't survive checking

• "Rap listeners pay more attention to lyrics than listeners of other genres." I assumed this was true and went looking for the study. I couldn't find a clean comparison. The nearest thing is self-reported survey work with no cross-genre control. It might well be true. It doesn't appear to be demonstrated, so I haven't claimed it above.