Sleeping Brains Understand Words

By Neuroskeptic | October 3, 2014 3:01 pm

Have you ever heard someone describe a task as being so easy that they ‘could do it in their sleep’? A fascinating new study from a team of French neuroscientists shows that this statement may be literally true, far more often than you’d think: Inducing Task-Relevant Responses to Speech in the Sleeping Brain

Sid Kouider and colleagues’ elegant experiment went as follows. Volunteers were asked to perform a word categorization task: spoken words were played to them and they had to press a button with their left hand (say) if the word was a kind of animal, or press a button with their right hand if it was an object.

So far, so simple – but the kicker was that participants were allowed to fall asleep during the task. The experiment took place in a quiet, dark room to help them nod off. Once a volunteer was soundly asleep, the task continued – more animal and object words were played to them while they slept.

The key question was: did the volunteers’ brains continue to perform the task while they were asleep? This might seem like a hard hypothesis to test – how can a brain ‘perform’ a button pressing task, without pressing any buttons, and how would we know even if it? Well, the participants were wired up to an EEG system to record brain electrical activity, before the experiment began. Based on the EEG data from the awake phase of the experiment, Kouider et al were able to record the different neural activations that accompanied pressing a button with either the left or the right hand. (These activations happen on opposite sides of the brain, fittingly.)


The authors then examined whether these same ‘button pressing’ patterns occurred in response to the stimuli presented during sleep – and amazingly, they did, in most cases. The truly remarkable result was that the sleeping brains ‘produced’ the correct responses to the stimuli. If an animal word was played, the brain’s activity was usually consistent with it making a (say) left hand button press.

So this is pretty amazing and suggests that the brain can perform a high-level language task, involving understanding the meaning of words, while asleep. There are some questions, of course. As Kouider et al say:

First, one might question whether participants in our study were truly asleep… in order to be fully confident that the trials that we included in our analysis genuinely reflect a state of sleep, microarousals and arousals (associated with button presses or not) were detected and trials in the direct vicinity of these events were discarded.

Finally, this paper made me think of the Chinese Room – a philosophical thought-experiment in which a man with an elaborate instruction book is able to respond, in Chinese, to questions posed in Chinese, even though he doesn’t know the language and has no (conscious) understanding of what he’s saying. Is a sleeping brain rather like that man? A sleeping brain has no conscious experience of the outside world, so far as we know. Yet somehow it knows how to respond to words…!

ResearchBlogging.orgKouider S, Andrillon T, Barbosa LS, Goupil L, & Bekinschtein TA (2014). Inducing task-relevant responses to speech in the sleeping brain. Current Biology, 24 (18), 2208-14 PMID: 25220055

CATEGORIZED UNDER: EEG, papers, philosophy, select, Top Posts
  • feloniousgrammar

    My mate sometimes talks about “the monkey” or “monkeys” in his sleep. When I respond with, “the monkeys are making too much noise?”, for instance, he affirms it. I’m tempted to try this experiment at home.

  • Pingback: Sleeping Brains Understand Words - Improve Brain Memory IQ Mind Brain News and Informative Articles | Brain Training()

  • Jespersen

    It’s a fascinating experiment indeed, but isn’t all of this entirely expected?

    I think it’s fairly well-established that the brain maintains learned connections during sleep (e.g. the rats and the mazes of Lee & Wilson 2002) or during the hypnagogic phase (the “Tetris effect”), that it can incorporate and respond to external stimuli (the basis of a number of lucid-dreaming experiments), and that this includes language comprehension (ever had a dream in which you, you know, talked with people?). So the fact that subjects can classify stimuli according to learned patterns even when asleep doesn’t surprise me.

    >Is a sleeping brain rather like that man? A sleeping brain has no
    conscious experience of the outside world, so far as we know. Yet
    somehow it knows how to respond to words…!

    I’m going with Shimon Edelman’s report of lucid dreaming experiments here (his lecture slides), but one of the conclusions was that sleeping people’s brain processes are essentially identical to ‘awake’ responses, except heavily constrained. Or, as Stephen LaBerge put it, “When we think
    in dreams, we really do think (whether clearly or not is another matter)”.

    Of course, this only concerns *dreams*. This might as well be the first case I’ve heard of which investigates (presumably dreamless) sleep.

    • Neuroskeptic

      Yes – the experiment (well, actually the second of two experiments that I merged together in my blog post) ensured that subjects were in stage 2 NREM sleep before starting the testing phase; and any trial in which the subject moved or seemed to not be in stage 2 sleep, was discarded.

      • Jespersen

        Very well then, thanks for clarifying!

        Now I almost can’t wait until the press misreports the experiment as ‘you can learn things during your sleep’, thus spawning a new wave of bullcrap businesses purporting to teach you a second language while you take a nap…

      • P. A. nichols

        Maybe this explains ‘post-hypnotic suggestion’?

  • Shalryn

    Somebody actually received money to study this? Why? It’s one of the most obvious abilities possessed by humans and animals. Anyone who has ever dreamed a dream that contained speech gets this. I have also dreamed dreams in which a speaker did not speak a language that I knew, and surprisingly enough, I didn’t understand that language in my sleep, either. My dog could also duplicate the experiment about which has been written here. All I had to do was wait until she was sleeping in REM, then talk about a certain treat. The dog would not wake up, but the drool factor proved beyond a doubt that she heard, understood and responded to the stimulus. She could also be made to run by someone saying, “Get the bad crow,” while she slept. All sorts of creatures have the functionality-while-asleep trait. All you have to do is pay attention to see that.

    • feloniousgrammar

      In defense of the experiment, it demonstrated this in a formal experiment; which, for further study, is a foundation that isn’t anecdotal.

      • Shalryn

        Okay, you got me there. Valid point. I suppose one gets so used to seeing proof of something that one doesn’t consider the scientific setting. Thanks for the reminder.

        • Neuroskeptic

          Also, it’s not obvious to me that brains would be able to process language coming from the outside world.

          In dreams, we often dream about talking and hearing speech, but that’s entirely internally generated.

          • Overburdened_Planet

            I’m new to Discover forums (I’ve been a hard copy subscriber for many years) and I very much enjoyed your article.

            It reminded me of another article, but that post disappeared after I refreshed the page.

            I was able to post a couple of other comments that remained, but I wanted you to read my post that was relevant to your article and response to Shalryn above.

            Possibly it’s awaiting moderation, but I’m not sure why.

            Are links permitted? It’s a link from Discover for an article entitled: “MRI Brain Scans Show Signs of Consciousness in Some “Vegetative” Patients”

            Are you able to find and release that post, or should I try again?

            Here’s some of that post, not the same wording or complete: Using an fMRI, some patients were able to answer basic yes or no questions by activating different parts of their brains when researchers asked patients to think of playing tennis, where areas of their motor cortex activated, or if asked to think of being in their house, spatial areas in their brains became active.

            They were told to associate thoughts about tennis with “yes” and thoughts about being in his house with “no,” and to ensure that patients were making conscious choices, researchers switched the rules and asked patients to associate tennis with “no” and his house with “yes,” all with consistent success.

  • John Owen

    Great, now wives and girlfriends everywhere will be able to point to this study and argue that things their men say, and agree to, are valid even if we were asleep when we said it.

    Thanks, science.

    • Bobareeno

      Don’t worry— wives and girlfriends are too busy imagining things that we say to even think about what we actually say.

    • Don’t Even Try It!

      Could this “Sleep talk” be used in a court of law?

      • Overburdened_Planet


        Last year, I got ‘sleep married.’

        My then girlfriend at the time said I had proposed to her in my sleep.

        I denied it, but she had a tape recording and said she would sue me for breach of contract.

        Now I sleep in the bathtub with the door locked and the water running.


    • Monica Mallow

      Perhaps responses are correct but the dreaming mind frequently produces circumstances that are different than the waking world. For example I dreamed that I was flying. Without devices. But I can’t do that in the waking state. I’ve also had dreams where I was single or captive. Both untrue. So it depends on what your sleeping brain is actually responding to, if the listener will be honest.
      But it also begs the question…do you have something to hide?

      • John Owen

        How exactly does you flying in a dream with no devices have the slightest thing to do with whether I’m hiding anything from my girlfriend? Seems like you’ve dated one too many cheaters.

    • Overburdened_Planet

      Last night, my wife woke me up and said she had a dream about me.

      She dreamed we were driving in the middle of the desert and I stopped the car, let her out, and drove away.

      I said “Honey, don’t worry. Go back to sleep and I’ll pick you up.”

  • Pingback: Pe Net 06.10.2014 | RAVEDA()

  • constitutionalive

    I often solve problems, write theses in my mind when sleeping/dreaming…wake up write immediately B4 forget, feel great, usually creative/innovative ideas ! luvv working/thinking this way :))

    • David Priest

      Stage one (N1) has also been shown to involve problem solving mentation. The”Tetris” experiments.

  • Dave Gray


  • Overburdened_Planet

    I’ve been a hard copy subscriber to Discover for years but new to emailed articles.

    This article is a nice treat, fascinating, and it reminds me of another article.

    I might be describing some of the details incorrectly, but there’s a study with comatose patients who were able to communicate using Yes/No responses by visualizing two physical tasks that corresponded to, (or represented) a Yes or No response.

    I remember one task was to visualize playing tennis to indicate a Yes (maybe it was No, but you get the idea, so tennis = Yes), and let’s say visualizing walking up stairs represented No.

    Some patients were able to hear and visualize these tasks-as-responses as a way to communicate without having to speak or move.

    I believe the same type of EEG monitoring occurred, where staff could see the part of the brain that activates when playing tennis or walking up stairs, requiring different and distinct regions of the brain to register in response to spoken questions.

    Do you like music?

    (Visualizing tennis = Yes).

    Do you like the smell of cut grass?

    (Visualizing stair walking = No).

    Thanks for the article, and the memories.

  • StarryTelling

    Once I was typing at my computer at work late at night and I fell asleep sitting up. I kept typing and what I typed was what I was dreaming. Kind of the opposite of this but….

  • Thom Baguley

    It seems over-strong to say this is a “high-level language task, involving understanding the meaning of words”. This is in essence an association based task. It is is consistent for example with with reports of sleep-talkers having “conversations”. Apparently these are driven by surface associations rather than meaning … Still interesting and a cute experiment, but not that surprising.

  • Pingback: I’ve Got Your Missing Links Right Here (11 October 2014) – Phenomena: Not Exactly Rocket Science()

  • Pingback: I’ve Got Your Missing Links Right Here (11 October 2014) | Gaia Gazette()

  • connielogg

    I have listened to books in my sleep and remember them very well

  • Pingback: Weekend Link Love – Edition 317()

  • Pingback: Link love: language (60) | Sentence first()

  • Pingback: El cerebro entiende palabras cuando estás dormido; podemos usar el lenguaje para programarnos | Lado B()

  • Pingback: ¿El cerebro entiende palabras mientras dormimos? | maestroviejo()

  • Pingback: El cerebro entiende palabras aunque estemos dormidos()

  • Pingback: El cerebro entiende palabras cuando estás dormido; podemos usar el lenguaje para programarnos – Vive Sana()



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar