If you need to learn French, it will not help you – unlike the cartoon character Dexter – to put on a CD while you are asleep with a voice saying cheese omelet – omelette du fromage over and over again. Nor will it help you (or Chandler Bing) to quit smoking by playing a tape during the night with the text You are falling fast asleep. Deeper. Deeper. Deeper. You are now completely asleep. You don’t need to smoke. Cigarettes don’t control you. You are a strong, confident woman who does not need to smoke. A strong, confident woman. (Friends – season 3 – episode 18: the one with the hypnosis tape). Sleep-learning like this does not work!
If you do the little experiment from the sidebar (which I don’t expect you do), you will be surprised by the very small amount of information that you can recall or even recognize. Maybe, you remember some fragments. Have you learned during your sleep?
Probably not! What is the problem? As the term sleep-learning indicates, the learning has to occur during sleep. On a typical night, a subject cycles between different depths of sleep and dreaming. You can see this cycling in figure 1, which is a simplified outline based on Moser et al. (2009). Normally, you start your trip to Morpheus awake and then enter sleep stage N1 after approximately 10 to 15 minutes. This is a very light sleep which is easily disturbed, for example by calling your name. After this phase, you gradually fall deeper asleep: N2 and N3. Behaviourally, it becomes more difficult to arouse you, e.g. increasing louder sounds must be used to awaken you. Each of these stages is also characterized by typical brain waves.
Figure 1. Sleep stages in a typical night (healthy adults 21 – 60 yrs) – AASM protocol; based on Moser et al. (2009), p. 145 and table 3, p. 142. Brain wave examples from Fraiwan et al. (2012), p. 12.
The first cycle (N1 – N2 – N3) takes about 90 minutes. Then follows a period of Rapid Eye Movements (REM). In this example, the first REM-period starts just before 01:00 hours. Subjects awakened during these REM-periods typically report dream content. The first REM-periods are rather short but increase in length as the night progresses. Intermittently, a subject awakens during the night; about 9% of the total time in bed. For most parts, this is forgotten the next morning.
The right hand side of figure 1 shows the proportion of each stage during a typical night of a healthy <60 years old population. As you can see, information presented during the time in bed, does not mean that you are in deep sleep while listening to this information. In fact, about 9% of the time in bed, you are awake. Already in 1955, Simon and Emmons have convincingly demonstrated that, under controlled conditions, no learning occurred. They presented the information only when the subjects were deep asleep, as demonstrated by their EEG recordings.
The conditions under which the results were found tend more to support the contention that some learning takes place in a special kind of waking state wherein Ss apparently do not remember later on if they had been awake. This may be of great practical importance from the standpoint of economy in study time, but it cannot be construed as sleep-learning.(Simon and Emmons, 1955, p. 341)
Of course, this great practical importance should not be exaggerated. If you did the little experiment at the beginning of this post, maybe you awoke during the night and listened carefully to the list of phone numbers. Maybe, at that time, you tried hard not to forget them. Only, the next morning all was gone, usually even the memory of being awake. How many times were you roughly awoken in the midst of a beautiful (or terrifying) dream, persuading yourself not to forget it. Alas, the next morning, nothing is left. Being wide awake during the night is no guarantee for remembering things in the morning.
During sleep, your working memory and perception is more or less intact. Your eardrum for example doesn’t stop vibrating! You hear things, you can smell odours, you feel pain and you respond to them. And while dreaming your cognition works at full speed. Only, you don’t remember all those things the next morning. Why? Scientists really don’t know at this moment. A far fetched hypothesis is that the situation resembles a little bit to the phenomenon of state-dependent learning. You can learn something in one (particular) state and can only reproduce it in a more or less similar state. But, this means that the effect of sleep-learning should only be observable while … asleep. Not very practical.
Can you really not learn anything during your sleep? In 2012, Arzi and his colleagues surprised the world with their publication in Nature NeuroScience Humans can learn new information during sleep. The title pitched high expectations but reading the article however soon revealed that it dealt with a completely different type of learning: partial reinforcement trace conditioning with odours. This is a big word for the following experiment. The researchers paired one tone (e.g. 400 Hz, 40 dB) with a pleasant odour and another tone (1200 Hz, 40 dB) with an unpleasant odour during sleep. It is a well-known effect that pleasant odours (e.g. shampoo) cause a more intense ‘sniff-reaction’ than unpleasant odours (e.g. rotten fish). This also happens while you are asleep. A bit like Pavlov’s dog who started salivating after hearing the bell; see some historical information about classical conditioning for background information. After presenting these tone-odour combinations numerous times during sleep, do their subjects respond to the tone alone? Due to their particular experimental design, this could be tested during sleep and after awakening. In their first experiment, the authors clearly showed that subjects produce a more intense sniffing respons to the tone that was previously associated with the pleasant odour than to the tone that was combined with the unpleasant odour. Both during sleep and after awakening!
In a second experiment, they tried to differentiate between REM and nREM-sleep. See figure 2, for a visual explanation of the experimental design and the results. Remember, some subjects received these trial blocks only in REM-sleep and some only in nREM sleep.
Figure 2. Design and results of Arzi et al. (2012)
The experiment used a very strict inclusion criterion and only included trials if there was no arousal within a window of 30 s after tone onset. The critique of Simon & Emmons (1955) does not apply here. The authors made the following conclusion.
Despite these limitation [only one measure for wake retention, small number of subjects due to awakenings], our results reveal learning of novel information during natural human sleep and implementation of this new learning in sleep and ensuing wake. Moreover, this learning occurred without later awareness of the learning process. This implies that, beyond the general health advantages associated with good sleep, humans may be able to utilize toward learning new information a state in which they spend about a third of their lives.Arzi et al. (2012), p. 1464.
Although the researchers did show that conditioning is possible during sleep with transfer to the waking state, their conclusion sounds a little bit exaggerated to me, as is the title of their article Humans can learn new information during sleep. First, remember that only in their first experiment there was a main effect of conditioning. In the second experiment, there only seems to be transfer in the nREM training. There are also a few ‘strange’ phenomena. During sleep, learning seems better in REM-sleep than in nRem sleep. The transfer to the waking state reverses the effect. Why? No reasonable argument is given. This is a rather complex experiment and also not very well described (see full-text). At numerous places, you can ask questions about the measurement of the sniffing respons, the exclusion of subjects, the seemingly arbitrary timing for stimulus duration, interstimulus and intertrial interval, inclusion interval. There is also a confounding of the period in the night with the REM-nREM manipulation. The REM-group received their training mainly in the second part of the night. Although the researchers tried and succeeded for the most part in controlling these different variables, there are still many blind spots and replication of these results are badly needed.
Last but not least, it seems to me that there is still a long way to go from learning a sniff-respons during sleep to utilizing the third part of our lives [sic] with learning new, relevant, verbal information. And it is this kind of learning that is usually implied and understood by most of the audience with the term sleep-learning.
The post Sleep-learning? Forget it! appeared first on Cogpsy.info.