My alarm woke me early. I was in a hotel room in London, near the headquarters of the BBC. I hadn't slept well. When I looked in the bathroom mirror, I saw someone pale and slightly terrified. I had reason to feel nervous. In just over an hour I would be speaking live to an audience of millions on the BBC's flagship radio discussion program, Start the Week. As I gazed into the mirror, I was aware that I was talking, silently, in my head. My words were a reassurance. They were aimed at me. “Relax,” I said. “You've been on Start the Week before.” I had the impression that I was speaking to myself but was also hearing something internally, the familiar shadow of a voice.

This is a story about everyday experience: the thoughts, images and sensations that go through your head as you are soaking in the tub, chopping onions in the kitchen or waiting for the door to open on an important meeting. When asked, people often say that their inner lives contain a lot of words. Psychologists use the term “inner speech” for this phenomenon, in which people talk to themselves silently in their head. It has a cousin, “private speech,” in which people talk to themselves audibly. If you say words to yourself, such as “Remember to get some coffee” or “Stick to the plan,” without making a sound, then you are using inner speech. If you say something similar to yourself out loud, it is private speech.

Both forms of language seem to have varied purposes, including planning and monitoring our behavior, regulating our emotions and fostering creativity. Among adults, inner speech seems to be more common than the private variety and, of particular interest to psychologists, is thus the form that probably plays the biggest part in our thinking. It is also quite a bit more difficult to study. When I was starting out in research in the 1990s, there was hardly any scientific literature on the topic. That situation has changed dramatically over the past couple of decades, partly because researchers have developed new experimental techniques for studying inner speech and partly because we now have a richer notion of how it functions, what forms it takes, and how it can benefit and hinder a thinker. In fact, we are starting to realize that inner speech elucidates some big questions about the mind and brain.

A Chat with Oneself

Henry is lying on a play mat with a toy train in each hand, rhapsodizing about the make-believe city he is about to create. “First the cars. Then a big train,” he says to himself. Henry is three years old. Walk into any nursery or preschool, anywhere in the world, and you will see (and hear) something similar. It can get noisy, with a classroom of kids thinking to themselves out loud. But this natural phenomenon of children's private speech provides some important clues about where the words in our head come from.

Scholars have long pondered the private speech of young children. In the 1920s Swiss developmental psychologist Jean Piaget proposed that this type of self-talk reflected the inability of youngsters to take other people's perspectives and adapt their speech to their listeners. In this view, private speech was the result of a failure to communicate with others. That was why it was thought to drop away as children got older, and they became more skilled at taking the perspectives of their listeners into account.

In the 1930s a Russian psychologist named Lev Semyonovich Vygotsky proposed an alternative explanation for private speech: children deliberately repurpose words that they have previously used successfully in social interactions with other individuals. Instead of regulating the behavior of others, they were getting the hang of using language to control themselves. Research in the intervening decades has bolstered Vygotsky's theory of how inner speech develops and how it comes to have the functions it has.

Discovering Vygotsky's writings as a student in developmental psychology, I remember being struck by the simplicity of his idea. It felt as though the theory of how verbal thought develops should be more complicated. But although his notion was itself straightforward, its ramifications were quite complex. Vygotsky was suggesting that the silent self-talk people engage in as adults is an internalized version of the conversations we have with others when we are developing as children. Nearly a century after Vygotsky wrote down his insights, I and other inner speech researchers are only beginning to unpack what they mean for understanding how words function in our thinking.

One of the most important implications of Vygotsky's theory is that inner speech should have the same structure as out-loud conversation: namely the quality of a dialogue between different points of view. This concept of thinking as mental dialogue is not new—it traces back at least as far as the philosopher Plato—but I latched onto its potential to reframe some deep mysteries of human cognition. One such mystery is about control: How it is that an intelligent system can come up with, and implement, new ideas about how to act? A robot can get very smart at responding to what happens in the environment, but what makes it come up with the idea of doing anything for itself? If the system has to be told what to do, then it is lacking one of the essences of intelligence.

What excited me about dialogue is that it is, by its very nature, self-regulating. When you are in conversation with another person, there is no third party standing there waving a conductor's baton to show you where the conversation should go next. You and your conversational partner regulate each other through the normal processes of questioning, challenging, responding, agreeing, and so on. Understanding self-talk in these terms seemed to hold out the prospect of explaining how human thought can be open-ended—not always directed toward a particular goal—and inherently flexible.

To do dialogue, though, you need to be able to represent something of the point of view of the person with whom you are in conversation. (It was the failure of just this kind of perspective taking that Piaget thought explained young children's private speech.) You often do not know in advance what the other person is thinking, but once you figure it out, you need to be able to keep it in mind and update that representation of his or her point of view as the conversation unfolds. Scientists now know a fair bit about the neural basis for such perspective taking, thanks in part to studies carried out using functional MRI and other medical imaging techniques that can reveal which brain regions carry out a given task.

Armed with these insights, my collaborators and I have been testing a new idea about how mental dialogues happen, based on the suspicion that they recruit the same parts of the brain used in perspective taking. In an fMRI experiment led by my colleague Ben Alderson-Day of Durham University in England, participants produced two forms of inner speech while lying in a brain scanner. We asked our volunteers to generate some inner speech that had a monologic structure; in other words, it did not involve a conversational exchange between different points of view. We also asked participants to conduct an inner dialogue. In each case, we presented a particular scenario as the theme for the inner speech, such as a visit to one's old school. For the monologue condition, participants might be giving a speech to some students; in the dialogue, chatting to their former principal.

We predicted that both kinds of inner speech would recruit the standard language systems that activate when people are asked to produce any kind of speech: specifically, areas at the boundary between the brain's left frontal and temporal lobes and an area farther back in the part of the brain known as the superior temporal gyrus. We thought that inner dialogue would be special, however, in additionally activating parts of the brain known to be involved in thinking about other minds. These brain regions underpin our so-called social cognition system, which functions to help us represent other people's thoughts, beliefs and desires.

Credit: Tami Tolpa; Sources: “The Brain's Conversation with Itself: Neural Substrates of Dialogic Inner Speech,” by Ben Alderson-Day et al., in Social Cognitive and Affective Neuroscience, Vol. 11, No. 1; January 2016 (monologue vs. dialogue); “Exploring the Ecological Validity of Thinking on Demand: Neural Correlates of Elicited vs. Spontaneously Occurring Inner Speech,” by Russell T. Hurlburt et al., in PLOS One, Vol. 11, No. 2, Article No. E0147932; February 4, 2016 (rote vs. spontaneous)

The results supported our predictions. When people were doing inner dialogue, their language system seemed to be working in conjunction with a part of their social cognition system, located in the right hemisphere close to the junction between the temporal and parietal lobes. Subjects did not exhibit this pattern of brain activity when they generated silent monologues. Although these findings need to be replicated, they provide preliminary evidence for a collaboration, spanning the two brain hemispheres, between two systems that are usually understood to have different functions. This neural linkage of language and social cognition seems to support Vygotsky's intuitions that when people are talking to themselves, they are having an actual conversation.

Caught in the Act

There are lots of reasons to be cautious in interpreting neuroimaging findings and, in this case, for overturning what had previously been understood about the neuroscience of inner speech. Most previous studies had simply asked participants to repeat sentences to themselves silently in their head in a monologic, nonconversational manner—the kind of inner speech you might do as you wander around a supermarket trying to remember the last few items on your list. It is supremely useful when the moment demands it but a long way from the creative and flexible inner dialogues that stem from treating ourselves as participants in a social exchange. Our research team put the conversational properties of inner speech front and center, but we were still asking our volunteers to do something quite unnatural: to talk to themselves on demand rather than waiting for inner speech to bubble up naturally. The problem is that cognitive neuroscientists need to be able to control things to understand what an experiment's findings really mean. Hanging around for inner speech to happen naturally seems to run against the idea of a rigorous experimental method.

What we need are ways of capturing inner speech as it occurs. Recently our team has taken a step in that direction by using a sophisticated method for garnering descriptions of people's inner experience known as descriptive experience sampling (DES). In this method, participants are trained to report on moments of inner experience when cued by a beeper. The process prompts the subjects to focus on whatever they are thinking, feeling, hearing, and so forth at the moment just before the beep went off and to take brief notes on those experiences as they occur. The following day the volunteers are interviewed in great depth about each moment of experience captured by the beep so that researchers can describe whether it was characterized by inner speech, sensory awareness or any of several other common phenomena.

My colleagues and I conducted the first ever study to couple this powerful method with fMRI. In it, we ran the conventional inner speech experiment, asking people to say particular words to themselves silently as they lay in the scanner. We also used DES to capture moments of experience as they happened naturally. We picked out those beeps in which we were fairly sure inner speech had occurred, based on the DES interviews, and compared the brain activations with those we had obtained in the standard task.

The differences were striking. Whereas the standard “rote repetition” method activated Broca's area (a part of the brain that is often implicated in the production of internal and external speech), spontaneous inner speech gave more pronounced activations farther back in the temporal lobe, in Heschl's gyrus. In terms of patterns of brain activation, naturally occurring inner speech contrasted dramatically with the kind that is produced on demand.

These findings have broad implications for how we go about investigating inner experience in cognitive neuroscience. They raise hard questions about how investigators approach the study of inner speech and what we can assume about any kind of mental experience we might think can be generated on demand. They underscore the need for what I like to call slow neuroscience: harnessing the power of neuroscientific techniques to very careful descriptions of human experience.

There are other reasons for taking care to describe inner speech in all its varieties. In Vygotsky's theory, dialogue and monologue are not the only variables of internal self-talk. A big feature of his scenario is the idea that, as language is internalized to form private and then inner speech, its form changes. Vygotsky saw several ways in which this might happen, including different kinds of abbreviation or condensation. In my anxious thoughts in my London hotel room, I caught myself saying a full sentence to myself: “You've been on Start the Week before.” At other times, the language I direct at myself is much more stripped down. If I hear a shrill beeping sound from the kitchen when I'm cooking, I might say to myself something such as “The oven timer is going off.” It is more likely, though, that I will simply say, “The timer.” Vygotsky noted that inner and private speech are often abbreviated, relative to utterances directed at another person. In self-talk, we usually do not have to put things into full sentences, in part because the utterance is for the self, and we therefore do not have to spell out all the details. The great Russian-American novelist Vladimir Nabokov recognized how our thoughts can have a compressed form relative to what we might say out loud. “We think not in words but in shadows of words,” he wrote in his notes for Pale Fire, according to a 1964 interview.

Oddly enough, no one had examined this feature of internal language until recently. Simon McCarthy-Jones, now at Trinity College Dublin in Ireland, and I put together an online questionnaire asking people about different qualities of inner speech. Our team has also used a smartphone app to gather such data as people go about their daily lives. The results of our initial study, published in 2011, reveal four main qualities of inner speech: its dialogical nature, its tendency to be condensed, the extent to which it can incorporate other people's voices, and its role in evaluating or motivating our behavior. Only a minority of people indicated that their inner speech tends to be condensed, but this quality is common enough to warrant further investigation.

Above all, this questionnaire-based research confirms the view that inner speech is not just one thing. It appears to come in different forms that may be adapted to different functions and that will quite possibly have different neural underpinnings. A challenge for the future is to try to understand whether the brain handles condensed inner speech differently from its expanded form. That will require either a way of eliciting condensed inner speech experimentally in the brain scanner or further developments in capturing it as it occurs naturally. Inner speech remains an elusive target of study.

A Key to Creativity

The study of inner speech has taken great strides since I started pondering it as a graduate student in the 1990s. A facet of mental life that was generally considered impenetrable to science has yielded to new experimental methods and neuroscientific techniques. And as it happens, this intimate aspect of consciousness can illuminate some important questions about the human mind.

For a start, inner speech can provide some clues about the origin of human creativity. Once people have the architecture for internal conversations, we can use it in all sorts of ways, from arguing with ourselves to conversing with an entity that is not there. Because we have internalized dialogues with others, we retain an “open slot” for the perspectives of other beings: whether or not they are present, are still alive or ever even existed. My dialogues with God, a deceased parent or an imaginary friend can be as richly creative as those I have with myself. Asking ourselves questions and then answering them may be a crucial bit of apparatus for taking our thoughts into new territories.

Another routine experience with links to self-talk is one of the most familiar and private of them all. The moment you open a book, your inner speech is hijacked in all kinds of interesting ways. Neuroscientists have shown that reading a fictional character's speech activates the same parts of the brain we use to process other people's voices. Using an online survey, our team recently asked a large sample of keen readers about the “voices” they heard when they were reading fiction. Around one in seven of our respondents said that the voices of fictional characters spoke as vividly in their mind as if there had been another person in the room uttering the words.

Some of our participants gave us more detail on their experience of fictional voices. Using the tools employed in literary scholarship to analyze narrative, we examined their open-ended descriptions for more clues to literature's power to colonize our thoughts. For at least some of our respondents, the voices of fictional characters continued to resonate even after the book had been put down. A few even adopted the personas of fictional characters as they went about their daily lives—looking out through the eyes of Mrs. Dalloway (the character from Virginia Woolf's novel of the same name) during a routine visit to Starbucks, for example. Marco Bernini of Durham University calls this phenomenon “experiential crossing.” These findings give us important clues about how our mind might represent the voices and characters of the social beings with whom we share our world.

The new science of inner speech tells us that it is anything but a solitary process. Much of the power of self-talk comes from the way it orchestrates a dialogue between different points of view. Like the collaboration my colleagues and I saw between the language system of the left hemisphere and the social cognition networks of the right, the inner speech network must be able to “plug in” to other neural systems as the situation demands—when we have verbal thoughts about the past and future, when we use words to talk ourselves through demanding tasks or when our mind simply wanders, with no particular objective in mind. If researchers get the science right, verbal thought stands to elucidate all these features of our cognition.

It may be because it is such an ordinary thing that inner speech has received so little scientific attention. But the next time you find yourself psyching yourself up for a challenge, talking yourself through a dilemma, ticking yourself off after a mistake or just planning your evening in a cozy out-loud mumble, you might want to think about the private, intimate wonder of your self-directed use of words. In everyday life, as in the lab, the voices of inner speech have much to tell us.