“It’s like waking from a dream; I just don’t remember,” said Henry Moliason, also known as patient H.M., one of the most celebrated patients among scientists who study memory. In 1953, Henry lost the ability to form new long-term memories after an experimental brain surgery intended to tame his epilepsy went awry. He could no longer remember where he went, who he talked to, what he saw in a given day or week. He was a blank slate each morning.
One important result of this memory loss was that Henry could no longer participate in that magical communion that happens when humans come together and share details of their past, present, or imagined future: conversation. He could not share stories about the world or himself with others because he had no memory bank to pull from.
Conversation is something most humans take for granted. And perhaps for that reason, we aren’t able to see how critical it is to the way we construct identity, build culture, and even form nations, argues Irish neuroscientist and author Shane O’Mara in his book Talking Heads: The New Science of How Conversation Shapes Our Worlds. These conversations depend on what O’Mara calls “collaborative memory,” a narrative act of co-creation on the fly. To imagine a shared future we must come together to talk about it, he suggests.
Read more: “Faulty Memory Is a Feature, Not a Bug”
I spoke with O’Mara about the power of assembly, what makes someone good at having difficult conversations, and the new technologies that are allowing scientists to simultaneously track four brains, all engaged in the world-building act of conversation.
I wanted to start with Henry Moliason, whose story opens your book. Why did you choose to lead the book with Henry?
Henry is the most dramatic case of amnesia that we have on record. He has dominated the literature in a way that no other patient has because the outcome of his operation was so unexpected, and his amnesia didn’t resolve over the next 50-odd years of his life.
What to me was very striking about his case was that he illustrates really well this idea that we update each other’s memories all the time through conversation. Henry did not have a conversationally updatable memory. You could trade gossip with him, you could trade information about your family, and he would have a very nice conversation with you, but then you leave the room and it’s gone.
Amnesia arises from damage to quite a variety of brain systems, but damage to the hippocampal formation, like the kind Henry had from the operation, has come to stand as the poster case for amnesia. It just happens that Henry’s case, for all the sociological reasons, was especially interesting, because you can point to a brain region easily, you can see it in a brain scanner. That’s why it has really dominated our thinking.
Related to the subject of how we update our memories in conversation with others, you describe a concept in the book called collaborative memory. What is this and how does it work?
There are really two approaches to thinking about memory. One is this austere quantitative approach, where you train somebody on material, you score it very carefully, the material is standardized from person to person. You know what the age-related norm for X, Y, or Z material will be. And that’s been a very productive and very important approach to memory. It allows the standardization of cognitive tests. And we owe that to the German experimental tradition in psychology.
The alternative way of thinking about memory is much like what we’re doing here now, which is you’re throwing questions at me, and I have to scramble around and reconstruct a story for you. This is a “narrative approach to memory,” which serves a social function and is enhanced by social interaction. So collaborative memory really revolves around the idea that two people working together to recall an event or an idea will bring to mind a much richer set of details. This happens all the time. My wife and I talk about holidays we’ve been on, and she’ll remember one thing, I’ll remember another. It’s something that happens as the result of a live interaction between two people.
When we think about memory or cognition on the fly or out in the wild, these two different things go on all the time. You have to remember very specific pieces of information in order to do things correctly in the world. How do you get your money out of an ATM? You must remember the sequence of numbers. You must do it in the right order or your card will be swallowed by the machine.
But equally, when we sit down over dinner and we’re talking to a spouse or kids, the flow of memory is narrative. You’re telling stories about what it is that you might want to do in the future, and you’re flicking backwards and forwards along this mental timeline, which is actually enabled for you by the hippocampal formation. One of the key things that allows you to do is not just recall the past or interpret the present. What it allows you to do is imagine a future. This is what humans have done since time immemorial. We made the long march out of Africa. We didn’t know what we would find because we hadn’t been there. Nobody had returned to tell us. Humans have done this kind of thing all the time. We do it in big ways. We do it in small ways.
You argue that this kind of conversational imagining of the future is essential to human culture and even nation building. Why?
The counterfactual is, look at all the nations built by chimpanzee relatives. The other apes don’t sit around in circles having a debate about why the Alpha is not distributing the spoils correctly. They have really brutal dominance hierarchies. Life is harsh and difficult. Humans are really quite different where that’s concerned. If you look at non-mechanized societies or tribal societies, a central mode of interaction is people sitting together and talking. The elder or the leader typically disparages themselves because their position is contingent. If the tribe doesn’t want them, they don’t come back from the hunt. It’s slightly less brutal in our modern world. You make a mess as a politician, you don’t come back from the election.
The key point is that humans engage in what’s best described as deliberative assembly. We sit and we deliberate together and we hash things out together. This is a pervasive feature of human life, from boardrooms, to meetings in sports clubs, to teachers’ assemblies. You have to have places where people can meet, exchange information, talk, and think together about how they want the future to be. Often it’ll be about settling the past, but again, we do it in the same way because we have jury trials, where juries get to deliberate on what has happened. This kind of deliberative assembly, where we gather together to hash out our differences, is a unique feature of human life. It’s not something that other species are capable of engaging in for obvious reasons. We have language. We have symbolic methods of communication. And we have these exceptionally powerful cognitive skills that allow us to flit between past, present, and future.
You argue that scientists have underestimated the power of conversation. Why do you think this has happened?
It’s one of those things that’s sitting at the end of your nose. Things that are directly in front of you are really, really hard to see. I’m often struck when you go to a scientific conference, you’re presenting your data on a poster. Somebody will walk up to you and they’ll say, “Tell me the story.”
Also, we’ve only recently been able to track individual contributions to a conversation and locate those in individual brains. Over the last five years, new tools have emerged that allow us to measure activity in three or four brains that are interacting at the same time while a dynamic conversation is ongoing. This is called hyperscanning, which has two principal methodologies.
One is to use EEG on a number of people who are interacting. That’s problematic because it tends to be artifact prone. You get this electrical activity in the EEG and you have to control for the noise. The other is becoming a very popular technique, and it’s one I’m using in my lab: functional near infrared spectroscopy, or fNIRS.
Basically, what you’re doing is pulsing near infrared light into the brain. It’s back scattered, back reflected, and you’ve got a detector that picks it up, and it’s differentially absorbed, depending on whether the local blood flow is oxygenated or deoxygenated. That’s basically the same method as you use in fMRI, but it’s much cheaper. fMRI often costs $2 to $3 million dollars, whereas the setup that I’m just putting into place now—which will allow me to scan four people at once—costs closer to 400,000 euros. Now, the problem with it is, it only works for the cortex. You’ll only get the outside of the brain, 2 centimeters deep. But that’s pretty good, and it’s noise-free.
You billed the book as the new science of conversation, and I wondered, has anything changed then in the years since it was first published in 2023?
The key thing is that we now have this range of tools that we didn’t have even just a couple of years ago, that will allow us to do things in conversation that we otherwise wouldn’t have been able to do. Aside from fNIRS, the obvious big one is LLMs, which allow us to parse vast amounts of speech. In the experiments that we will be doing soon, we will be having people wear fNIRS caps, and we will be recording speech channels for each person, during a decision-making interaction. And we’ll be getting them to make repeated decisions about things, and we’ll be able to capture what they’re saying. I tried to do this with hippocampal patients [with damage to their hippocampus] quite a few years ago, and it was immensely difficult coding what they were saying. The advent of LLMs coupled to other kinds of tracking of activity in the brain during cognitive tasks will open up new ways of thinking about how we can analyze conversation that we couldn’t do just a couple of years ago.
What are some of the characteristics of good conversation?
It depends on the type of conversation. But the core of a good conversation is one where there’s a high degree of rapport established between both parties. People enjoy connecting in conversation. People enjoy revealing information about themselves. People enjoy swapping gossip. It offers really good social glue. But a good conversation is one where both people walk away feeling that they’ve profited in some way.
Of course, some conversations are difficult. For example, investigative interviews, which are not easy, either for the person who’s doing the investigation or the person who’s been questioned. But the data show that people who are good at information gathering in these kinds of interviews are people who are effective at building rapport. They’re open, respectful, don’t engage in needless dominance displays, they’re not ego invested. In other words, they’re good psychotherapists. These kinds of people are also good at making the person feel good about the fact that they’ve given up some information.
Is there a conversation that you’ve had in your lifetime that really stands out as among the most memorable or most meaningful?
In the notes at the end of the book, I thank a friend of mine. When I originally set out to write this book, about halfway through, I found it was not going in the direction that I wanted it to go. My friend Adrian Lynch and I were out for a walk one night and I was chatting about the book and he said, “Have you read this book, Imagined Communities by Benedict Anderson?” And I said, “No, tell me more.” This book is the most cited book in political science. It’s really a stunning book, so I went and got it and I thought, “God, this is what I need.” Immediately, as a result of that conversation, the direction of the book became clear to me. When we start to imagine the future, we do it together in a conversation.
Enjoying Nautilus? Subscribe to our free newsletter.
Lead Image: Eugène François Fines / Wikipedia Commons