REM – Rapid Eye Movement : Nucleus papilion neurons play an important role


REM – Rapid Eye Movement – is not only the name of a successful American rock band, but also and not least a characteristic eye movement in paradoxical sleep, so in the stage with high dream activity.

This sleep phase has a peculiarity: Although the muscle tone of the sleeping person completely relaxed, the eyes suddenly move back and forth.

The name “paradoxical sleep” is well deserved. Characteristic of these are signs of deep sleep (muscle atony) in connection with a brain activity, which is very similar to those in the waking state, and eye movements.

This sleep phase was discovered in the 1950s by French and American researchers and consequently called rapid eye movement sleep (REM sleep), i.e. sleep with rapid eye movements.

Why can this strange phenomenon be useful?

For 70 years, scientists have been dreaming of getting to the bottom of the mystery. Thanks to the productive cooperation between the universities of Bern and Fribourg, this dream could now come true.

Butterfly wings arranged neurons

For several years, the team led by Franck Girard and Marco Celio at the University of Freiburg has studied neurons under the microscope, which occur in the brain stem and form a structure that is reminiscent of butterfly wings, which is why she was baptized Nucleus papilio.

“These neurons are associated with multiple nerve centers, especially those responsible for eye movement, and those involved in sleep control,” explains Franck Girard.

“Therefore, we asked ourselves the following question: may the nucleus papilio neurons play a role in the control of eye movements during sleep?”

The researchers from Bern gathered the loop around the nucleus papilio neurons even more closely and were able to demonstrate with the help of optogenetic methods (combined optical and genetic techniques) that their artificial activation causes rapid eye movement, especially during this sleep phase.

Stronger together

To test this hypothesis, the Freiburg researchers turned to the research group headed by Dr. C. Gutiérrez Herrera and Prof. A. Adamantidis at the Department of Neurology at the Inselspital, University Hospital Bern, and Department for BioMedical Research of the University of Bern, who are investigating sleep in mice.

“To our surprise, we found that these neurons are particularly active in the phase of paradoxical sleep,” reports Dr. Carolina Gutierrez.

The researchers from Bern gathered the loop around the nucleus papilio neurons even more closely and were able to demonstrate with the help of optogenetic methods (combined optical and genetic techniques) that their artificial activation causes rapid eye movement, especially during this sleep phase.

Conversely, the inhibition or elimination of these same neurons blocks the movement of the eyes.

After the “how” the “why”!

Now that it is clear that the nucleus papilio neurons play an important role in eye movement during REM sleep, it is important to find out what function this phenomenon has.

Is it due to the visual experience of dreams?

Does it matter in preserving memories?

“Now that we are able to specifically activate the nucleus papilio ‘on demand’ in mice by optogenetic methods, we may be able to find answers to these questions,” says Antoine Adamantidis.

The next step, however, will be to confirm the activation of nucleus papilio neurons during REM sleep in humans.

The researchers have not yet found the key to their dreams, but they’ve come a long way.

A better understanding of the neural circuits involved in paradoxical sleep is therefore a prerequisite for understanding for instance how these neurons are prone to degenerative changes in diseases such as Parkinson’s.

Dream recall rates vary considerably between individuals (Schredl et al., 2007). Multiple models have been developed in an attempt to explain this variability (Freud, 1958Schonbar, 1965Cohen and Wolfe, 1973Cohen and MacNeilage, 1974). Among these, the arousal-retrieval model (Koulack and Goodenough, 1976) is supported by reliable empirical evidence (For a review, see Schredl, 1999Schredl et al., 2003a,b).

This model proposes a mechanism for how dream content is transferred from short-term consciousness to long-term memory storage. The model assumes that traces are not encoded during the dreaming process itself. One possible explanation for the lack of encoding could be related to the substantial deactivation of the prefrontal cortex during both non-rapid eye movement (NREM) and rapid eye movement (REM) sleep (Muzur et al., 2002Nir and Tononi, 2010Mutz and Javadi, 2017).

The prefrontal cortex is essential for executive functions involved in the encoding of complex content. Therefore, according to the model, a period of wakefulness is necessary to enable long-term storage of short-term dream content. If this occurs, subsequent retrieval from long-term storage is enabled.

Several lines of research provide some support for the arousal-retrieval model of dream recall. Studies of individuals who experience frequent arousals during sleep, due for example to insomnia or sleep apnea, show increased dream recall frequency (DRF; Schredl, 199920092010Schredl et al., 1999). However, DRF in individuals with abnormal sleep may be confounded by their sleep pathologies, while there might be factors other than arousals contributing to increased DRF in this population group.

Two recent studies managed to circumvent at least one major confound evident in earlier research by recruiting healthy participants. De Gennaro et al. (2010), utilizing polysomnography, recruited 40 individuals to investigate the effect of a single night of total sleep deprivation on DRF the morning following recovery sleep. The authors found a near-complete abolition of morning dream recall.

They propose that one explanation relates to the significant decrease in the number of awakenings on the recovery night, a finding they propose to be consistent with the arousal-retrieval model. Another study utilized a design where high frequency recallers (HFRs; n = 18) are compared directly to low frequency recallers (LFRs; n = 18; Eichenlaub et al., 2014a). They investigated various sleep parameters, including arousals and awakenings. A significant difference with regard to “intra-sleep wakefulness” was found, i.e., individuals with high rates of dream recall spent significantly more time awake following sleep onset.

This study provided critical evidence for the arousal-retrieval model; however, it should be interpreted with caution as questions remain as to whether mechanisms other than (or in addition to) arousal-retrieval cause higher DRF in HFRs. For example, it may be precisely because HFRs are alerted to their dreams via awakening that they report a higher frequency of dreams in the first place. Alternatively, HFRs may actually produce more dreams, which, in combination with increased wakefulness, results in higher dream recall.

To investigate the latter possibility, a study by the same group recruited healthy HFRs (n = 21) and LFRs (n = 20) and measured regional cerebral blood flow (rCBF) during both sleep and wakefulness (Eichenlaub et al., 2014b). The study found that (a) compared to LFRs, HFRs showed significantly increased rates of rCBF in the temporo-parietal junction (TPJ) during NREM stage 3 sleep (NREM3), REM sleep, and wakefulness, and (b) significantly increased rCBF in the medial prefrontal cortex (mPFC) during REM sleep and wakefulness. Based on this, the authors propose that the TPJ and mPFC play an important role not only in relation to dream recall during wakefulness, but also in the dreaming process itself. The study by Marzano et al. (2011) also gives credence to the importance of frontal and temporo-parietal areas in the dreaming process.

They investigated possible neurophysiological correlates associated with successful recall upon awakening during REM and NREM 2 sleep. The authors found that an increase in frontal theta activity during REM sleep, and lower alpha activity in the right temporo-parietal areas during NREM2 sleep, were associated with subsequent successful dream recall.

Authors from both studies note that lesion studies provide support for the important, yet not exclusive, role of the TPJ and mPFC in dream production. These studies demonstrate that complete or near-complete cessation of dreaming frequently occurs with damage to the mPFC and TPJ (Murri et al., 1985Doricchi and Violani, 1992Solms, 1997). Overall, these results suggest that HFRs not only have increased intra-sleep wakefulness, which promotes dream recall, but may also have increased dream production.

However, measuring dream production directly remains a methodological challenge. This is because subjective dream recall does not necessarily produce reliable estimates of actual dream frequency (Schredl et al., 2003bParke and Horton, 2009Kahan and LaBerge, 2011). An alternative index of dream production is REM density (the frequency of rapid eye movements during REM sleep). Studying dream production via REM sleep parameters serves as a reasonable starting point as REM sleep typically yields the highest rates of dream recall (up to 90%) compared to NREM sleep (10–54%; for reviews, see Stickgold et al., 1994Nielsen, 20002004Schredl et al., 2007).

Importantly, there is empirical support for using REMs to index the occurrence of dreaming: REMs function as a physiological correlate of ponto-geniculo-occipital (PGO) wave activity, while PGO waves sub-serve the occurrence of dream imagery (Pace-Schott, 2005Miyauchi et al., 2009Leclair-Visonneau et al., 2010Desseilles et al., 2011). Therefore, based on PGO activity serving as a common underlying mechanism, the incidence of REMs can be associated with the occurrence of dream imagery during REM sleep.

To our knowledge, there is only one study comparing REM density in healthy HFRs and LFRs (Vallat et al., 2017a). This study compared 18 HFRs with 18 LFRs and showed that there was no significant difference in REM density between the two groups. Therefore, the question as to whether HFRs not only report more dreams but also produce more dreams warrants additional consideration.

The current study has two aims. The first aim is to investigate whether HFRs and LFRs differ in their profile of nocturnal awakenings (including both time spent awake and the number of awakenings, based on measures from the whole night and from REM sleep in particular). The second aim is to investigate whether HFRs produce more dreams across the night when compared to LFRs.

The following hypotheses were tested:

Hypothesis 1:

(a) High frequency recallers will spend significantly more time awake after sleep onset compared to LFRs.

(b) High frequency recallers will experience a significantly increased number of awakenings across the night, and from REM sleep in particular, compared to LFRs.

Hypothesis 2:

High frequency recallers will exhibit significantly higher rates of REM density across the night compared to LFRs.


The present study had two aims. Firstly, we aimed to show that HFRs have a higher awakening profile (characterized by both time spent awake after sleep onset and the number of awakenings) compared to LFRs.

We found that, consistent with findings related to polysomnographic data from Eichenlaub et al. (2014a), HFRs spent significantly more time awake after sleep onset. Our findings also showed that HFRs had more frequent awakenings in comparison with LFRs. However, contrary to our prediction, HFRs experienced the majority of their awakenings from NREM 2 rather than REM sleep, while LFRs experienced similar amounts of awakenings during NREM 2 and REM sleep.

Secondly, via examining REM density as a proxy for dream production, we tested the prediction that HFRs would have higher REM density than LFRs. We did not find any between-group differences in this marker of dream production.

Regarding our first aim, the pattern of increased wakefulness after sleep onset and increased total number of awakenings experienced by HFRs fits well with the encoding and retrieval mechanisms postulated by the arousal-retrieval model. For example, our data supports the hypothesis that an increased number of awakenings leads to an increased number of opportunities for dream traces to be encoded. Furthermore, the longer periods of wakefulness following awakenings that HFRs experienced may also enhance the encoding of dream content from short-to long-term memory, according to the model. These results build upon existing evidence to provide support for the validity of the arousal-retrieval model of dream recall.

With regard to a mechanism underlying the arousal-retrieval model, Eichenlaub et al. (2014a) propose that increased wakefulness after sleep onset, as well as awakenings from sleep in general in HFRs, show heightened brain reactivity. This assertion is based on results from their event-related potential (ERP) study which revealed that HFRs respond more strongly to novel auditory stimuli during both wakefulness and sleep. The neurophysiological mechanism underlying this heightened brain reactivity to stimuli in HFRs is thought to be a P3a-like component, or P3a-like wave detected on electroencephalogram (EEG). The P3a-like component (a sub-component of the P300 wave), is strongly associated with orientation of attention to external stimuli (Friedman et al., 2001). It is accepted that the larger the P3a-like wave is, the stronger is the attention orientation response (Dominguez-Borras et al., 2008Lv et al., 2010). HFRs exhibited larger P3a-like waves in response to novel auditory stimuli across vigilance states compared to LFRs. Eichenlaub et al. (2014a) postulate that a stronger attention orientation response to external stimuli in HFRs is one of the neurophysiological mechanisms underlying awakenings and longer periods of wakefulness after sleep onset in these individuals.

Interestingly, detection of a larger P3a-like wave in HFRs in response to novel stimuli (either a participant’s first name or an unknown first name presented randomly and rarely among pure tones), was not homogenous across sleep stages (Eichenlaub et al., 2014a). For example, at earlier latencies (which represent the attention-orientation response), larger P3a-like waves in HFR individuals in response to novel stimuli were detected most strongly during NREM 2 sleep. Put differently, HFRs exhibited the strongest attention-orientation response to novel stimuli during NREM 2 sleep.

A stronger attention-orientation response during NREM 2 sleep could serve as one potential mechanism underlying the awakening profiles of HFRs. We found significant between-group differences in relation to the number of awakenings, as well as awakenings lasting ≥2 min from NREM2 sleep. The latter finding is of particular significance since, according to the arousal-retrieval model of dream recall, increased duration of awakenings should lead to enhanced dream recall (Koulack and Goodenough, 1976). Indeed, Vallat et al. (2017a) found that the minimum time period required for awareness of/memory for dream traces is approximately 2 min. Finally, the significant between-group difference in relation to dream recall rates from NREM 2 sleep in this study further underscores the critical role of NREM 2 awakenings in enhanced dream recall in HFRs.

Regarding our second aim, we found that HFRs did not exhibit higher REM density than LFRs, and therefore were not likely to experience more dreaming during REM sleep. There are two viable explanations for this finding. One is that differences do exist, but they were undetected in this study. More specifically, it could be argued that the methods and measures employed in our study were not sensitive enough to detect between-group differences. However, this is an unlikely explanation as there were strong and significant correlations between the REM density parameters and affective variables in a yet unpublished study from our laboratory (van Wyk, unpublished). It is unlikely that the measure would be sensitive enough to detect significant results in one investigation but not in another. This favors the second explanation for the null findings, which is that between-group differences with regard to REM density do not exist in the current sample.

In light of this, we propose that between-group differences were lacking because, perhaps, REM dreaming is not of prime relevance in the current sample. For example, there were no between-group differences with regard to REM density, REM sleep%, REM sleep latency, the number of awakenings from REM sleep, the number of awakenings lasting ≥2 min from REM sleep, nor dream recall rates from REM sleep awakenings. Furthermore, at the time of writing (2019), new findings with regard to REM sleep and REM density based on the Eichenlaub et al. (2014a) data, were published (Vallat et al., 2017a). Researchers found no significant difference with regard to REM density values, nor between the number and length of awakenings from REM sleep. They concluded that higher DRF in the HFR group “could not be explained by the REM sleep hypothesis of dreaming” (Vallat et al., 2017a). It is important to note that we independently obtained results comparable to Vallat et al. (2017a), despite utilizing a different method of measuring REM density.

Another point that should be emphasized is that the studies by Eichenlaub et al. (2014ab) and Vallat et al. (2017a), and the current study, all recruited healthy participants devoid of psychiatric symptoms and/or diagnoses. This is important as several changes in REM sleep parameters have been noted in the presence of psychopathology (Cartwright et al., 1998Ellis et al., 2014Medina et al., 2014). Therefore, one possible reason for failing to detect significant differences in any of the REM sleep parameters could be because a sample free from psychopathology was recruited in the present study and similar research.

University of Bern
Media Contacts:
Antoine Adamantidis – University of Bern
Image Source:
The image is in the public domain.

Original Research: Open access
“Neurons in the Nucleus papilio contribute to the control of eye movements during REM sleep”. C. Gutierrez Herrera, F. Girard, A. Bilella, T. C. Gent, D. M. Roccaro-Waldmeyer, A. Adamantidis & M. R. Celio.
Nature Communications doi:10.1038/s41467-019-13217-y.


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.