The btrmt. lectures started as a way to teach this stuff to anyone who’d listen—me, a microphone, and one idea at a time, after a day of lecturing. But not everyone likes to listen, so here’s the cleaned-up version for the readers. Rather hear it? Here’s the episode. And the article it grew out of is Eerie coincidences aren’t that eerie.
Is your phone listening to you?
I think a lot of people are worried that our devices are listening to us. At least among the people I hang out with—and maybe that says more about me than about the phenomenon. But I think a fairly common opinion, or suspicion at least, is that social media apps and other apps on our phones are picking up our conversations to sell us ads.
And I think we think this because it’s pretty surprising how often an advert appears on Amazon, or in the Instagram feed, that reflects a recent conversation. I’ll speak about something completely random, or something I think is random, something I would never normally talk about. And lo and behold, within a couple of days, I’m shown an ad for precisely that random thing.
People try to explain this away by saying these algorithms are fantastic at learning about you. So fantastic, in fact, that they predict the things you want and the things you talk about. It’s not that they’re listening to you. It’s that they’re predicting that not only do you want this stuff, you’re likely to talk about it.
What’s changed since I wrote the article
When I first wrote an article on eerie coincidences, from what I could tell, this basically wasn’t happening. Now my perspective has changed a little. I was essentially sceptical of the idea, but we’ve since seen examples of companies doing this, and maybe I shouldn’t have been so laissez-faire about it. Of course these companies would if they could, right?
I’ll link to the particular article I’m thinking of, from 404 Media, which reported on a media group that advertised “active listening” ad targeting. As a consequence, they were removed from the Google Partner Program.
And the thing that brought it to my attention at all: somebody shared with me a paper from Northeastern that looked at something like 17,000 Android apps. They didn’t find any evidence that these apps were secretly enabling the microphone to send audio. But they did find apps sending screenshots and screen recordings to third parties. So this is definitely happening.
But the reason I was sceptical is something I still want to talk about. And that is that eerie coincidences are probably just coincidences that the mind makes seem more important than they actually are. That’s what I want to talk about today. Some eerie coincidences are suspicious, but probably most are just you getting tricked by your perceptions. So let me explain how.
A paranoid person’s phone
I’m a pretty paranoid person, I would say. To give you an example, I use an operating system called GrapheneOS on my Android phone. I do that because I’m worried about security and privacy—though I don’t really know what that means. So I thought I’d go through the fairly difficult rigmarole of getting it installed on my phone, rather than hand over all my data to Google on the stock operating system.
Or, as another example: I don’t know if you know this, but on your Android phone you can have different users, like different users on a computer, as if multiple people were using the same phone. And you can use this feature to sandbox apps. So I have my Instagram in a separate user profile, to keep it away from the other apps on my phone. Again, not because I’m worried about anything specific. Just because I’m generally paranoid about stuff spying on me.
And I know this is pretty weird, because I have a mate who works at Google, and he saw me doing this one time. He was like, “I’d love to talk to you, because I was just having a conversation with another Google colleague about who actually uses user profiles on the phone.” So, pretty paranoid, I think.
And even then, I will still notice oddly specific ads that seem absurdly relevant to recent conversations, and to nothing else in my behaviour that I can detect. That’s the thing that really made me think they’re probably not listening to what I’m saying. Probably this is something else. Now, again, we have evidence that some people are trying to listen, so this isn’t always true. But let me give you examples of other eerie coincidences that probably aren’t that eerie.
We only notice the mistakes
When I mark student work—essays, or the students I’m supervising—I will almost always only comment on the things that need correcting. Much to my students’ dismay, if what they’ve written is good, I almost never stop to make a note of it. It’s such a problem, I think, for students who expect more emotional intelligence from their supervisors, that I have to say it up front, and in emails when I send feedback: a lack of commentary means it’s good.
But I don’t think this is unusual. I think you probably do this too, right? Maybe you’re reading a book, or watching a TV show, and you’re completely absorbed in the thing until you notice a mistake. A typo in the book, or the captions on the TV show are wrong or weird. And all of a sudden you’re jolted back into yourself. You’re no longer absorbed in the story. You’re paying attention to the error.
Interruptions and the predicting brain
I think this experience is on the same continuum as the “is my phone listening to me?” experience. Because it’s a feature of our brain, our nervous system, to come alive—to come especially active—when it notices interruptions.
The interruption theory of emotion is one of my favourite theories in psychology, and I’ll link an article in the show notes for more on it. The core idea is that interruptions to our experience—interruptions to our expectations—are the thing that make us stop, pay attention, and have emotions in response to the world.
Since that theory in the ’80s, a lot of brain science has found itself circling a similar concept. The most recent instantiation of it, if you’re keen to Google it—and I’ll give some links as well—is the predictive coding stuff from, say, Lisa Feldman Barrett, or the active inference stuff from the notoriously obscure writings of Karl Friston.
Fundamentally, though, our nervous systems appear to have developed in large part to map actions to the environment. To map the statistical structure of the world and your behaviour in it, and make sure that whatever you perceive automatically produces adaptive actions in response. And this function of the nervous system has been around since long before brains were a thing. Before we had a central nervous system, with a brain on top of the spinal cord, we just had the peripheral nervous system. The basic nervous web.
Cheapskates on autopilot
But of course, it doesn’t make sense to do this mapping of perception to action online, all the time—every moment of the day, your neural system actively working to connect your perceptions to the best possible responses. Usually the best possible responses are highly prescribed, over-determined, and the body can manage them passively.
What I mean is that our environments aren’t really very varied. They can be very complex, and human behaviour is extraordinary in many ways. But there’s a reason so much of our conversation around growth and wellness concentrates on routines and habits. The environment might be very complex, but we’ve spent our whole lives working out the best paths through it. And the vast majority of what we do is done on autopilot.
We’re cheapskates. We encode the best mappings of perception to action, and then we let them run in the background until something happens that makes us stop and go, “Oh, we need to update this map. Something’s wrong here.”
So this is why I won’t make a note when my students are writing good stuff. I’m absorbed in what they’re doing. I’m not being interrupted; it’s a coherent argument. Until some strange sentence or illogical statement jolts me out of that experience, I’m not going to notice. And it’s the same with you and your book or your TV show. It’s flowing smoothly through the brain, meeting all your expectations about how books should go, or the captions on a TV show, until you hit an error or some event violates your expectations. Then you’re jolted out. You’re interrupted. You wake up and pay attention in a different way.
Now, all of that might seem a tangent from what started off being about whether our phones are listening to us. But I think it’s contiguous with that experience, because this interruptive feature of the brain tells us something. If the world is aligned with our expectations, we shouldn’t notice. When things are expected, we should essentially ignore them—not be interrupted. We should only really notice things when they interrupt our experience.
Peanut butter toast
So let me give you an example to bring us back to the point. I love peanut butter toast. I eat heaps of it. And if it weren’t for a twenty-year-old MythBusters episode that still lives in my head to this day, I would have thought that whenever I drop my peanut butter toast, it lands peanut butter side down.
If you test this, what you’ll find is that it’s just as likely to fall peanut butter side up as down. And the reason you probably think one happens more than the other is: why would you notice when it falls peanut butter side up? There’s nothing wrong. You just start eating it.
The only reason I think it falls peanut butter side down is that I only get interrupted when it lands upside down. If it’s peanut butter side up, I can just pick it up and start eating. I don’t need to get the cloth and wipe the floor.
The ads you never notice
So basically, I think this is more or less what’s happening with the listening phone thing. We get inundated with ads all day, every day, and you never notice when the ads are random and unrelated to conversations you’ve had recently. You’re good at ignoring useless ads, useless information. You only notice the really strange ones—the ones that are just generally peculiar, or the ad that’s strangely topical given what you’ve been talking about with your mates. But the proportion of ads you notice, against the ads you can’t remember, is probably super, super tiny.
That’s my speculation, anyway. Maybe somebody could do an undergrad research project on it. But probably most unusual coincidences are this.
So let me close out with a couple of implications, and then we’ll wrap up, and hopefully this will end up being a super short podcast.
Interesting isn’t the same as true
I could just leave the lecture there. Probably most unusual, eerie, suspicious coincidences aren’t unusual, eerie or suspicious. They’re just a matter of you not noticing the counterfactuals. You’re not comparing the relative likelihood, because you can’t remember all the times it went normally. Your entire nervous architecture is tuned to detect and recognise interruptions to your expectations, not the things that are contiguous with them.
And I will essentially leave it there. But I will point out that I very often talk about what’s known as the sociology of the interesting, and I’ll see if I can link to a couple of places where I talk about it more. Essentially the idea is that whether something is true doesn’t seem to matter much for whether we find it interesting. For us to find something interesting, it needs to violate our weakly held beliefs. We need something like a hot take on something we don’t care very much about.
And that’s related to what I’m talking about here. If I tell you something you already know, you’re just going to nod along and forget about it, for the same reason you don’t notice the things that don’t violate your expectations. And if I tell you something that violates your strongly held beliefs, you’re just going to ignore me. Possibly because it’s wrong, but also because there’s too much cognitive and emotional infrastructure invested in our strongly held beliefs to let a piece of evidence overturn them.
I think the reason I keep talking about the sociology of the interesting is that it’s notable just how much of the internet is this kind of content. It’s what I try to do in my podcasts, and in my articles at times, to generate interest. It’s a pattern that underlies a lot of the media we find entertaining. Spooky coincidences are basically our preferred form of content. And I think that’s a problem, because it doesn’t need to bear any relationship to the truth.
So when I first wrote this article on eerie coincidences, the idea was to get you to question whether an eerie coincidence was really eerie. And I guess the rule translates to content too. If something violates your weakly held beliefs and you find it interesting, maybe that’s a signal to stop and think. Maybe this isn’t that interesting. Maybe my nervous system is just tricking me into thinking that it is.
And that, I think, is where I’m going to leave it today. Until next time.
Show notes
Stuff I mentioned
· Eerie coincidences aren’t that eerie—the article this episode grew out of
· 404 Media (2024). Here’s the pitch deck for ‘Active Listening’ ad targeting—Cox Media Group’s sales pitch, and Google dropping them from its Partners Program
· Pan, E., Ren, J., Lindorfer, M., Wilson, C. & Choffnes, D. (2018). Panoptispy: Characterizing audio and video exfiltration from Android applications. Proceedings on Privacy Enhancing Technologies, 2018(4), 33–50—the Northeastern study of 17,000 apps; there’s a plain-English summary from Northeastern
· GrapheneOS—the privacy-hardened Android I use
· Google. Delete, switch, or add users—Android’s user profiles, which you can use to sandbox apps
· How to never get mad again (kind of)—the interruption theory of emotion
· Emotion and the mind—on predictive brains and cognitive misers (the cheapskates)
· Barrett, L. F. (2017). The theory of constructed emotion: an active inference account of interoception and categorization. Social Cognitive and Affective Neuroscience, 12(1), 1–23.
· Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138.
· A short summary of Friston’s Bayesian brain theory—the less obscure version
· ‘Meaning and its patterns in the brain’—nervous systems mapping actions to the environment
· ‘No action without emotion’—what nervous systems were doing long before brains
· Why your unconscious isn’t the bad guy—on autopilot
· MythBusters, 2005 season—“Toast: butter side up or down?”
· Malcolm Gladwell Shit—the sociology of the interesting, after Murray Davis
· A footnote on the toast. The folk belief turns out to be right, for physical reasons. Robert Matthews showed that toast sliding off a table doesn’t have the height to finish turning over, so it tends to land butter side down—the study won the 1996 Ig Nobel Prize for physics. Tossed in the air, it’s about 50/50. Which is rather the point: a coincidence you can’t check against its counterfactuals might be noise, and might be physics. Matthews, R. A. J. (1995). Tumbling toast, Murphy’s Law and the fundamental constants. European Journal of Physics, 16(4), 172–176.
Related stuff
· “Purple Doesn’t Exist”—the brain inventing what it can’t account for