Based on the analogy of the ancient art of memory (AAOM) techniques with dream content, Llewellyn hypothesizes in the target article that an off-line memory processing – termed elaborative encoding– is taking place during rapid eye movement (REM) sleep and that this is instigated during non-REM (NREM) sleep and facilitates the recall of episodic memories. That there is more to off-line memory processing than simple consolidation (repeating the newly acquired information exactly) is in line with the current theories (see Stickgold & Walker Reference Stickgold and Walker2013). The major challenge posed by this hypothesis is the idea that dreaming, especially REM dreaming, reflects or participates in this function of REM sleep since one can argue that a lot of processes occur during this sleep stage, not all of which are reflected in dreaming – if recalled.
Looking at the AAOM examples, the parallel between AAOM and dream content is less than clear for me: According to Llewellyn, AAOM techniques were used to learn episodic memories that were not related directly to the person by combining those memories with personal material. Dreaming, however, often reflects personal concerns (Domhoff Reference Domhoff2003) and emotional salient daytime experiences (Schredl Reference Schredl2006).
The parallel between dreams and AAOM also does not explain why dreams are quite easily forgotten (Schredl Reference Schredl, Barrett and McNamara2007). It is important to keep in mind that dreams are only memories of the dreaming process (Schredl Reference Schredl2008). One study (Schredl Reference Schredl2000), for example, did not find a correlation between dream salience, dream bizarreness,, and dream recall frequency. The theory that best explains the problems the waking mind has with recalling dreams was proposed by Koukkou and Lehmann (Reference Koukkou and Lehmann1983). The functional state-shift theory is based on the empirical evidence on state-dependent learning: the waking state differs from the REM-sleep state regarding cholinergic modulation (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000), and thus there are difficulties in remembering the dreamed experiences. In addition, the fact that the sleep profile of the night shows brief REM periods at the beginning with short, mundane dreams and longer REM periods with the more bizarre and elaborate dreams at the end of the night (Antrobus Reference Antrobus1991; Dement & Kleitman Reference Dement and Kleitman1957a) does not completely support the idea of instigating the REM sleep processes in subsequent NREM sleep. Furthermore, sleep-onset dreams, NREM dreams, and even daydreams often cannot be differentiated from REM dreams on a phenomenological level (Foulkes & Fleisher Reference Foulkes and Fleisher1975; Nielsen Reference Nielsen2000). The Kekulé example cited by Llewellyn (sect. 4.2.1, para. 6), for instance, is a sleep-onset dream (Strunz Reference Strunz1993). These issues are, however, not as important as is the question about how to test whether dreaming – defined as subjective experiencing during sleep – has a function additional to the function of sleep or REM sleep.
To illustrate the problems within this line of research, I complement the overview of Llewellyn, who cited only the nap study by Wamsley et al. (Reference Wamsley, Tucker, Payne, Benavides and Stickgold2010). The first pilot study linking dream content to overnight memory consolidation was conducted by De Koninck et al. (Reference De Koninck, Prevost and Lortie-Lussier1996), who found that task-related dream content is related to increased performance in the procedural task (e.g., performing a task while wearing goggles that invert the visual field) the next day. Unfortunately, this result was not confirmed by a subsequent study using another procedural task: mirror tracing (Schredl & Erlacher Reference Schredl and Erlacher2010). A completely different approach was applied by Erlacher and Schredl (Reference Erlacher and Schredl2010): Lucid dreamers were instructed to train for a coin-tossing task within their dreams, and those who successfully did this exhibited improved performance the next morning.
The major problem with these studies is that you have to interrupt sleep (several REM awakenings) to get a clear picture of the person's dreams that night, and this might interfere with the off-line memory processes. If only spontaneously recalled dreams were elicited (Wamsley et al. Reference Wamsley, Tucker, Payne, Benavides and Stickgold2010), there is a problem that the trait-variable dream recall might be associated with memory performance and not the recall of a particular dream about the task. The main methodological issue, however, is that you have to elicit dream content (verbal or written reports) in order to know whether the person had dreamed about the memory task (or the episodic memories that the researcher is interested in), and thus it cannot be ruled out that the memory-enhancing effect is not due to the dreaming during the night or can be attributed only to the remembered dream. In the Erlacher and Schredl study, for example, the successful lucid dreamers might have been more confident because of their lucid dreaming and thus performed better in the task. Having these methodological issues in mind, the suggestions by Llewellyn as to how her hypothesis might be tested have to be evaluated very carefully. Studying patients with brain lesions who report loss of dreaming (Solms Reference Solms1997) has several drawbacks because one does not know what other cognitive processes were affected by the lesions. But the main question is how one can be sure that they are not dreaming and are only unable to recall any dreams.
If dreaming is defined as subjective experiencing during sleep (similar to the subjective experiencing during wakefulness), a complete cessation of dreaming might not be possible – at least in patients able to perform memory tasks or speak about their subjective processes while awake. The most promising approach suggested by Llewellyn is to study AAOM processes in the waking state by using functional magnetic resonance imaging (fMRI) techniques and look to see whether similar brain activation patterns can be found during REM sleep. This, however, can still not prove that dreaming is involved in this off-line memory processing. As a dream researcher, I would like to study the dreams of persons who use AAOM techniques in their waking lives regularly or very gifted persons. If their dreams are completely different with regard to bizarreness or content related to episodic memories, this would not support the AAOM analogy to “normal” dreaming.
To summarize, the hypothesis proposed by Llewellyn is stimulating for the field, but many important methodological issues need to be addressed before making progress in this line of research.
Based on the analogy of the ancient art of memory (AAOM) techniques with dream content, Llewellyn hypothesizes in the target article that an off-line memory processing – termed elaborative encoding– is taking place during rapid eye movement (REM) sleep and that this is instigated during non-REM (NREM) sleep and facilitates the recall of episodic memories. That there is more to off-line memory processing than simple consolidation (repeating the newly acquired information exactly) is in line with the current theories (see Stickgold & Walker Reference Stickgold and Walker2013). The major challenge posed by this hypothesis is the idea that dreaming, especially REM dreaming, reflects or participates in this function of REM sleep since one can argue that a lot of processes occur during this sleep stage, not all of which are reflected in dreaming – if recalled.
Looking at the AAOM examples, the parallel between AAOM and dream content is less than clear for me: According to Llewellyn, AAOM techniques were used to learn episodic memories that were not related directly to the person by combining those memories with personal material. Dreaming, however, often reflects personal concerns (Domhoff Reference Domhoff2003) and emotional salient daytime experiences (Schredl Reference Schredl2006).
The parallel between dreams and AAOM also does not explain why dreams are quite easily forgotten (Schredl Reference Schredl, Barrett and McNamara2007). It is important to keep in mind that dreams are only memories of the dreaming process (Schredl Reference Schredl2008). One study (Schredl Reference Schredl2000), for example, did not find a correlation between dream salience, dream bizarreness,, and dream recall frequency. The theory that best explains the problems the waking mind has with recalling dreams was proposed by Koukkou and Lehmann (Reference Koukkou and Lehmann1983). The functional state-shift theory is based on the empirical evidence on state-dependent learning: the waking state differs from the REM-sleep state regarding cholinergic modulation (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000), and thus there are difficulties in remembering the dreamed experiences. In addition, the fact that the sleep profile of the night shows brief REM periods at the beginning with short, mundane dreams and longer REM periods with the more bizarre and elaborate dreams at the end of the night (Antrobus Reference Antrobus1991; Dement & Kleitman Reference Dement and Kleitman1957a) does not completely support the idea of instigating the REM sleep processes in subsequent NREM sleep. Furthermore, sleep-onset dreams, NREM dreams, and even daydreams often cannot be differentiated from REM dreams on a phenomenological level (Foulkes & Fleisher Reference Foulkes and Fleisher1975; Nielsen Reference Nielsen2000). The Kekulé example cited by Llewellyn (sect. 4.2.1, para. 6), for instance, is a sleep-onset dream (Strunz Reference Strunz1993). These issues are, however, not as important as is the question about how to test whether dreaming – defined as subjective experiencing during sleep – has a function additional to the function of sleep or REM sleep.
To illustrate the problems within this line of research, I complement the overview of Llewellyn, who cited only the nap study by Wamsley et al. (Reference Wamsley, Tucker, Payne, Benavides and Stickgold2010). The first pilot study linking dream content to overnight memory consolidation was conducted by De Koninck et al. (Reference De Koninck, Prevost and Lortie-Lussier1996), who found that task-related dream content is related to increased performance in the procedural task (e.g., performing a task while wearing goggles that invert the visual field) the next day. Unfortunately, this result was not confirmed by a subsequent study using another procedural task: mirror tracing (Schredl & Erlacher Reference Schredl and Erlacher2010). A completely different approach was applied by Erlacher and Schredl (Reference Erlacher and Schredl2010): Lucid dreamers were instructed to train for a coin-tossing task within their dreams, and those who successfully did this exhibited improved performance the next morning.
The major problem with these studies is that you have to interrupt sleep (several REM awakenings) to get a clear picture of the person's dreams that night, and this might interfere with the off-line memory processes. If only spontaneously recalled dreams were elicited (Wamsley et al. Reference Wamsley, Tucker, Payne, Benavides and Stickgold2010), there is a problem that the trait-variable dream recall might be associated with memory performance and not the recall of a particular dream about the task. The main methodological issue, however, is that you have to elicit dream content (verbal or written reports) in order to know whether the person had dreamed about the memory task (or the episodic memories that the researcher is interested in), and thus it cannot be ruled out that the memory-enhancing effect is not due to the dreaming during the night or can be attributed only to the remembered dream. In the Erlacher and Schredl study, for example, the successful lucid dreamers might have been more confident because of their lucid dreaming and thus performed better in the task. Having these methodological issues in mind, the suggestions by Llewellyn as to how her hypothesis might be tested have to be evaluated very carefully. Studying patients with brain lesions who report loss of dreaming (Solms Reference Solms1997) has several drawbacks because one does not know what other cognitive processes were affected by the lesions. But the main question is how one can be sure that they are not dreaming and are only unable to recall any dreams.
If dreaming is defined as subjective experiencing during sleep (similar to the subjective experiencing during wakefulness), a complete cessation of dreaming might not be possible – at least in patients able to perform memory tasks or speak about their subjective processes while awake. The most promising approach suggested by Llewellyn is to study AAOM processes in the waking state by using functional magnetic resonance imaging (fMRI) techniques and look to see whether similar brain activation patterns can be found during REM sleep. This, however, can still not prove that dreaming is involved in this off-line memory processing. As a dream researcher, I would like to study the dreams of persons who use AAOM techniques in their waking lives regularly or very gifted persons. If their dreams are completely different with regard to bizarreness or content related to episodic memories, this would not support the AAOM analogy to “normal” dreaming.
To summarize, the hypothesis proposed by Llewellyn is stimulating for the field, but many important methodological issues need to be addressed before making progress in this line of research.