It has been proposed that mental experience in sleep, such as dreaming activity, both reflects and contributes to the off-line consolidation of recent episodic memories (e.g. Cipolli et al. Reference Cipolli, Fagioli, Mazzetti and Tuozzi2004; De Koninck et al. Reference De Koninck, Christ, Hebert and Rinfret1990; Fiss et al. Reference Fiss, Kremer and Lichtman1977; Wamsley & Stickgold Reference Wamsley and Stickgold2011). However, the mechanisms underlying a beneficial effect of sleep on memory consolidation through dreaming remain largely unknown. Llewellyn proposes that rapid eye movement (REM) dreaming supports the “elaborate encoding” of episodic memories, eventually leading to the emergence of novel commonalities or associations between recent and remote memories that are then instantiated as “junctions” and consolidated during non-REM (NREM) sleep. We found Llewellyn's hypothesis sensible in considering synergetic roles for REM and NREM sleep in memory consolidation processes. Indeed, animal studies (Giuditta et al. Reference Giuditta, Ambrosini, Montagnese, Mandile, Cotugno, Grassi Zucconi and Vescia1995) and human studies (Ficca et al. Reference Ficca, Lombardo, Rossi and Salzarulo2000) suggest that information processing through consecutive iterations of NREM-REM sleep cycles plays a pivotal role in the off-line consolidation of recent memories.
In this respect, and as acknowledged by Llewellyn, several authors already ascribed complementary roles to the main sleep states. For instance, it was proposed that recent memories are protected against retroactive interference during NREM sleep and then consolidated during REM sleep (Scrima Reference Scrima1982) or at the molecular level that NREM slow-wave sleep (SWS) promotes post-acquisition processes of neuronal reverberation, whereas transcriptional events necessary for long-term memory consolidation are sparked during the subsequent REM sleep. Hence, converging data support the hypothesis of sequential processes of memory consolidation during sleep (Giuditta et al. Reference Giuditta, Ambrosini, Montagnese, Mandile, Cotugno, Grassi Zucconi and Vescia1995). More generally, the selection of adaptive memory traces and the weakening of nonadaptive ones may predominate during NREM sleep, the selected memory traces being then consolidated during the immediately ensuing REM sleep episode.
Besides differences in conceptualizations, most proposals converge to suggest that the succession of cerebral states corresponding to a typical sleep cycle (i.e., NREM followed by REM sleep) influences the consolidation of recently learned information in long-term memory. In contrast with this prevailing view, Llewellyn argues that episodic memories are first processed during REM sleep and then consolidated during NREM sleep. This position raises several questions that would need to be addressed.
First, if it is true that episodic memories are elaborately encoded during REM dreaming and then consolidated as “junctions” during NREM sleep, it logically entails that REM sleep–dependent memory processes would take place before those occurring during NREM sleep – that is, in the opposite order of sleep states within a cycle, which also assumes that the first NREM episode within a night of sleep might not play any role in the processing of memories. However, such a hypothesis is invalidated by the abundant literature ascribing a major role to the first NREM sleep episode in memory consolidation (e.g., Gais et al. Reference Gais, Plihal, Wagner and Born2000; Plihal & Born Reference Plihal and Born1997; Reference Plihal and Born1999; Yaroush et al. Reference Yaroush, Sullivan and Ekstrand1971). Second, and within the continuity of the previous argument, the first sleep cycles in a night are proportionally the richer in NREM (SWS) sleep, whereas the last sleep cycles are the richer in REM sleep. Therefore, consolidation according to Llewellyn would predominantly occur during the first part of the night, whereas elaborative encoding would take place in the second half, again leading to the same logical contradiction.
Notwithstanding sleep physiology, we also found Llewellyn's proposal incongruent with the chronobiological characteristics of dreams production. Indeed, dream contents change as the night progresses: REM sleep mentation is more vivid, more emotional, and bizarre later in the night (Agargun & Cartwright Reference Agargun and Cartwright2003; Fosse et al. Reference Fosse, Stickgold and Hobson2004; Verdone Reference Verdone1965; Wamsley et al. Reference Wamsley, Hirota, Tucker, Smith and Antrobus2007). Because Llewellyn envisions these features of dream contents as reflections of elaborative encoding processes, elaborative encoding should occur during late REM sleep, thus leaving little time for NREM sleep processes to instantiate the so-called junctions.
Taken together, these elements question the roles attributed by Llewellyn to NREM and REM sleep states. While the succession of sleep stages may not be mandatory for memory consolidation processes (e.g., Scrima Reference Scrima1982), this interplay should be conceptualized and grounded in physiology whenever the differential actions of NREM and REM sleep follow an orderly sequence, as it is the case in Llewellyn's proposal.
Another key concept in Llewellyn's proposal is that novel associations between recent and remote memories are generated through dreaming activity during REM sleep. The evidence presented to us in support of this hypothesis is questionable. Indeed, Llewellyn mostly relies on studies by Wagner et al. (Reference Wagner, Gais, Haider, Verleger and Born2004) and Walker et al. (Reference Walker, Liston, Hobson and Stickgold2002) to infer a REM dreaming-related ability to create novel associative patterns. However, REM sleep was not specifically tested in the study by Wagner et al. (Reference Wagner, Gais, Haider, Verleger and Born2004), only the effect of a complete night of sleep deprivation. Furthermore, although Walker et al. (Reference Walker, Liston, Hobson and Stickgold2002) found an association between REM sleep and improved ability in solving anagrams, the implicit rules extracted after a night of sleep were actually already present at encoding (also in Wagner et al. Reference Wagner, Gais, Haider, Verleger and Born2004). Therefore, it is disputable whether these studies support the creation of de novo associations during sleep rather than the simple extraction of hidden patterns. Although activation of dopaminergic circuits during REM sleep may indeed favor dreaming and unusual associations (Perogamvros & Schwartz Reference Perogamvros and Schwartz2012), it is probably more appropriate considering Wagner's and Walker's studies to suggest that (possibly REM) sleep facilitates insight into hidden rules, a process that leads to the construction of novel, higher-level schemas (Walker & Stickgold Reference Walker and Stickgold2010).
To conclude, Llewellyn's proposal has merits in proposing a synergetic role of sleep stages in memory processes. However, the proposal is also counterintuitive, considering the chronobiology of dreams and the architecture of sleep both within and between sleep cycles over the course of the night. It also remains disputable whether and how elaborative encoding processes would generate novel associations during sleep.
You have
Access
It has been proposed that mental experience in sleep, such as dreaming activity, both reflects and contributes to the off-line consolidation of recent episodic memories (e.g. Cipolli et al. Reference Cipolli, Fagioli, Mazzetti and Tuozzi2004; De Koninck et al. Reference De Koninck, Christ, Hebert and Rinfret1990; Fiss et al. Reference Fiss, Kremer and Lichtman1977; Wamsley & Stickgold Reference Wamsley and Stickgold2011). However, the mechanisms underlying a beneficial effect of sleep on memory consolidation through dreaming remain largely unknown. Llewellyn proposes that rapid eye movement (REM) dreaming supports the “elaborate encoding” of episodic memories, eventually leading to the emergence of novel commonalities or associations between recent and remote memories that are then instantiated as “junctions” and consolidated during non-REM (NREM) sleep. We found Llewellyn's hypothesis sensible in considering synergetic roles for REM and NREM sleep in memory consolidation processes. Indeed, animal studies (Giuditta et al. Reference Giuditta, Ambrosini, Montagnese, Mandile, Cotugno, Grassi Zucconi and Vescia1995) and human studies (Ficca et al. Reference Ficca, Lombardo, Rossi and Salzarulo2000) suggest that information processing through consecutive iterations of NREM-REM sleep cycles plays a pivotal role in the off-line consolidation of recent memories.
In this respect, and as acknowledged by Llewellyn, several authors already ascribed complementary roles to the main sleep states. For instance, it was proposed that recent memories are protected against retroactive interference during NREM sleep and then consolidated during REM sleep (Scrima Reference Scrima1982) or at the molecular level that NREM slow-wave sleep (SWS) promotes post-acquisition processes of neuronal reverberation, whereas transcriptional events necessary for long-term memory consolidation are sparked during the subsequent REM sleep. Hence, converging data support the hypothesis of sequential processes of memory consolidation during sleep (Giuditta et al. Reference Giuditta, Ambrosini, Montagnese, Mandile, Cotugno, Grassi Zucconi and Vescia1995). More generally, the selection of adaptive memory traces and the weakening of nonadaptive ones may predominate during NREM sleep, the selected memory traces being then consolidated during the immediately ensuing REM sleep episode.
Besides differences in conceptualizations, most proposals converge to suggest that the succession of cerebral states corresponding to a typical sleep cycle (i.e., NREM followed by REM sleep) influences the consolidation of recently learned information in long-term memory. In contrast with this prevailing view, Llewellyn argues that episodic memories are first processed during REM sleep and then consolidated during NREM sleep. This position raises several questions that would need to be addressed.
First, if it is true that episodic memories are elaborately encoded during REM dreaming and then consolidated as “junctions” during NREM sleep, it logically entails that REM sleep–dependent memory processes would take place before those occurring during NREM sleep – that is, in the opposite order of sleep states within a cycle, which also assumes that the first NREM episode within a night of sleep might not play any role in the processing of memories. However, such a hypothesis is invalidated by the abundant literature ascribing a major role to the first NREM sleep episode in memory consolidation (e.g., Gais et al. Reference Gais, Plihal, Wagner and Born2000; Plihal & Born Reference Plihal and Born1997; Reference Plihal and Born1999; Yaroush et al. Reference Yaroush, Sullivan and Ekstrand1971). Second, and within the continuity of the previous argument, the first sleep cycles in a night are proportionally the richer in NREM (SWS) sleep, whereas the last sleep cycles are the richer in REM sleep. Therefore, consolidation according to Llewellyn would predominantly occur during the first part of the night, whereas elaborative encoding would take place in the second half, again leading to the same logical contradiction.
Notwithstanding sleep physiology, we also found Llewellyn's proposal incongruent with the chronobiological characteristics of dreams production. Indeed, dream contents change as the night progresses: REM sleep mentation is more vivid, more emotional, and bizarre later in the night (Agargun & Cartwright Reference Agargun and Cartwright2003; Fosse et al. Reference Fosse, Stickgold and Hobson2004; Verdone Reference Verdone1965; Wamsley et al. Reference Wamsley, Hirota, Tucker, Smith and Antrobus2007). Because Llewellyn envisions these features of dream contents as reflections of elaborative encoding processes, elaborative encoding should occur during late REM sleep, thus leaving little time for NREM sleep processes to instantiate the so-called junctions.
Taken together, these elements question the roles attributed by Llewellyn to NREM and REM sleep states. While the succession of sleep stages may not be mandatory for memory consolidation processes (e.g., Scrima Reference Scrima1982), this interplay should be conceptualized and grounded in physiology whenever the differential actions of NREM and REM sleep follow an orderly sequence, as it is the case in Llewellyn's proposal.
Another key concept in Llewellyn's proposal is that novel associations between recent and remote memories are generated through dreaming activity during REM sleep. The evidence presented to us in support of this hypothesis is questionable. Indeed, Llewellyn mostly relies on studies by Wagner et al. (Reference Wagner, Gais, Haider, Verleger and Born2004) and Walker et al. (Reference Walker, Liston, Hobson and Stickgold2002) to infer a REM dreaming-related ability to create novel associative patterns. However, REM sleep was not specifically tested in the study by Wagner et al. (Reference Wagner, Gais, Haider, Verleger and Born2004), only the effect of a complete night of sleep deprivation. Furthermore, although Walker et al. (Reference Walker, Liston, Hobson and Stickgold2002) found an association between REM sleep and improved ability in solving anagrams, the implicit rules extracted after a night of sleep were actually already present at encoding (also in Wagner et al. Reference Wagner, Gais, Haider, Verleger and Born2004). Therefore, it is disputable whether these studies support the creation of de novo associations during sleep rather than the simple extraction of hidden patterns. Although activation of dopaminergic circuits during REM sleep may indeed favor dreaming and unusual associations (Perogamvros & Schwartz Reference Perogamvros and Schwartz2012), it is probably more appropriate considering Wagner's and Walker's studies to suggest that (possibly REM) sleep facilitates insight into hidden rules, a process that leads to the construction of novel, higher-level schemas (Walker & Stickgold Reference Walker and Stickgold2010).
To conclude, Llewellyn's proposal has merits in proposing a synergetic role of sleep stages in memory processes. However, the proposal is also counterintuitive, considering the chronobiology of dreams and the architecture of sleep both within and between sleep cycles over the course of the night. It also remains disputable whether and how elaborative encoding processes would generate novel associations during sleep.