First, we would like to emphasize that we do agree with the author on several points of her theory: During dreams, fragments of both recent and remote experiences reappear, in particular emotionally relevant ones; the seemingly bizarre links between these dream elements may actually be explained by idiosyncratic semantic networks (which can be revealed, for example, through psychoanalysis or other techniques); and there is good evidence that rapid eye movement (REM) sleep supports subsequent memory retrieval of emotional events (e.g., Nishida et al. Reference Nishida, Pearsall, Buckner and Walker2009). Furthermore, on a phenomenological level, dreams do share several characteristics with ancient art of memory (AAOM) mnemotechniques: Both are hyperassociative – that is, they connect individual contents through rich networks of links (which for mnemotechniques increases the number of cues which make these contents accessible) – and these associations may appear random and bizarre on the first sight but actually serve a hidden logic.
Despite this apparent similarity between dreams and mnemotechniques, we argue that dreams are unlikely to function as effective mnemotechniques. The ultimate goal of mnemotechniques is to improve memory, but dream memory is deficient compared with memory for waking experiences (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000). Not only are dreams remembered rarely, but dream memories also typically contain much less detail than does memory for experiences acquired during waking state. This may be due to various psychological factors, including the apparent strangeness of dreaming experience (which impedes subsequent recall; see Bartlett Reference Bartlett1932), as well as due to physiological factors such as the low level of functional EEG connectivity in brain regions that are crucial for declarative memory (Fell et al. Reference Fell, Staedtgen, Burr, Kockelmann, Helmstaedter, Schaller, Elger and Fernández2003). A mnemotechnique that sets individual memory contents into a framework that is inherently difficult to remember is paradoxical. (Imagine a device for motor control that renders your movements random and uncontrollable.)
It could be argued that while dreams themselves are difficult to remember, they support the consolidation of their constituents; in principle, an efficient mnemotechnique does not require that the entire connection of various items be remembered, but at least that the individual items within this connection can be easily accessed and therefore better remembered. Indeed, this seems to be what the author has in mind when she hypothesizes that “[i]f hyperassociative dream scenes do encode episodic memories and these scenes are retained as hippocampal indices, these are non-conscious phenomena used, in normal wake, to ‘search’ for memories in cortical networks” (sect. 5.1, para. 1). Thus, the idea is that dreams trigger subsequent retrieval of parts of the episodes that are retained in them through nonconscious processes. Even if this is the case, the parallel to mnemotechniques is at best remote – mnemotechniques are used to support conscious recall of entire fictitious or real memory episodes, not to trigger memory unconsciously for individual constituents of these episodes.
But leaving this aside for now, what is the empirical evidence for a role of dreams in episodic memory consolidation? It is well established that non-REM (NREM) sleep (Plihal & Born Reference Plihal and Born1997) and NREM mentations (Wamsley & Stickgold Reference Wamsley and Stickgold2010) support later retrieval of declarative memories. REM sleep appears to play a more prominent role for subsequent memory of emotional memories (Wagner et al. Reference Wagner, Gais and Born2001) and procedural memories (Plihal & Born Reference Plihal and Born1997), although evidence is mixed in this respect because a recent study suggests that pharmacological reductions of REM sleep facilitate the consolidation of motor sequences (Rasch et al. Reference Rasch, Pommer, Diekelmann and Born2009). Evidence for a role of REM dreams in the support of later retrieval of episodic memories is extremely scarce (Cipolli et al. Reference Cipolli, Fagioli, Mazzetti and Tuozzi2005; Fiss et al. Reference Fiss, Kremer and Lichtman1977).
Finally, the physiological order of REM sleep and NREM sleep is inconsistent with the proposed model. The author suggests that “elaborative encoding of episodic memories during REM dreaming may be followed by the instantiation of episodic junctions during NREM” (sect. 8, para. 1). However, REM sleep usually follows NREM sleep – the first REM episode usually occurs at the end of the first sleep cycle, and generally there is more REM sleep in the second half of night and more NREM sleep in the first half of night. Thus, the physiological order of REM sleep and NREM sleep would rather suggest another sequence of processes: Consolidation and integration of episodic memories via the formation of semantic network connections probably already occurs during NREM sleep, and these connections may later be activated (and, possibly, more thoroughly integrated) during REM sleep (Diekelmann & Born Reference Diekelmann and Born2010).
To summarize, the analogy between REM dreams and AAOM mnemotechniques is convincing at only a superficial phenomenological level but is odd, at a closer look, for several reasons: Unlike mneomotechnically encoded material, REM dreams (1) are inherently difficult to remember, (2) do usually not allow conscious subsequent retrieval of all interconnected elements, and (3) have been found to support subsequent episodic memory in only rare cases. Furthermore, the physiological order of REM sleep and NREM sleep suggests that REM sleep supports, if anything, a secondary consolidation or integration step.
You have
Access
First, we would like to emphasize that we do agree with the author on several points of her theory: During dreams, fragments of both recent and remote experiences reappear, in particular emotionally relevant ones; the seemingly bizarre links between these dream elements may actually be explained by idiosyncratic semantic networks (which can be revealed, for example, through psychoanalysis or other techniques); and there is good evidence that rapid eye movement (REM) sleep supports subsequent memory retrieval of emotional events (e.g., Nishida et al. Reference Nishida, Pearsall, Buckner and Walker2009). Furthermore, on a phenomenological level, dreams do share several characteristics with ancient art of memory (AAOM) mnemotechniques: Both are hyperassociative – that is, they connect individual contents through rich networks of links (which for mnemotechniques increases the number of cues which make these contents accessible) – and these associations may appear random and bizarre on the first sight but actually serve a hidden logic.
Despite this apparent similarity between dreams and mnemotechniques, we argue that dreams are unlikely to function as effective mnemotechniques. The ultimate goal of mnemotechniques is to improve memory, but dream memory is deficient compared with memory for waking experiences (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000). Not only are dreams remembered rarely, but dream memories also typically contain much less detail than does memory for experiences acquired during waking state. This may be due to various psychological factors, including the apparent strangeness of dreaming experience (which impedes subsequent recall; see Bartlett Reference Bartlett1932), as well as due to physiological factors such as the low level of functional EEG connectivity in brain regions that are crucial for declarative memory (Fell et al. Reference Fell, Staedtgen, Burr, Kockelmann, Helmstaedter, Schaller, Elger and Fernández2003). A mnemotechnique that sets individual memory contents into a framework that is inherently difficult to remember is paradoxical. (Imagine a device for motor control that renders your movements random and uncontrollable.)
It could be argued that while dreams themselves are difficult to remember, they support the consolidation of their constituents; in principle, an efficient mnemotechnique does not require that the entire connection of various items be remembered, but at least that the individual items within this connection can be easily accessed and therefore better remembered. Indeed, this seems to be what the author has in mind when she hypothesizes that “[i]f hyperassociative dream scenes do encode episodic memories and these scenes are retained as hippocampal indices, these are non-conscious phenomena used, in normal wake, to ‘search’ for memories in cortical networks” (sect. 5.1, para. 1). Thus, the idea is that dreams trigger subsequent retrieval of parts of the episodes that are retained in them through nonconscious processes. Even if this is the case, the parallel to mnemotechniques is at best remote – mnemotechniques are used to support conscious recall of entire fictitious or real memory episodes, not to trigger memory unconsciously for individual constituents of these episodes.
But leaving this aside for now, what is the empirical evidence for a role of dreams in episodic memory consolidation? It is well established that non-REM (NREM) sleep (Plihal & Born Reference Plihal and Born1997) and NREM mentations (Wamsley & Stickgold Reference Wamsley and Stickgold2010) support later retrieval of declarative memories. REM sleep appears to play a more prominent role for subsequent memory of emotional memories (Wagner et al. Reference Wagner, Gais and Born2001) and procedural memories (Plihal & Born Reference Plihal and Born1997), although evidence is mixed in this respect because a recent study suggests that pharmacological reductions of REM sleep facilitate the consolidation of motor sequences (Rasch et al. Reference Rasch, Pommer, Diekelmann and Born2009). Evidence for a role of REM dreams in the support of later retrieval of episodic memories is extremely scarce (Cipolli et al. Reference Cipolli, Fagioli, Mazzetti and Tuozzi2005; Fiss et al. Reference Fiss, Kremer and Lichtman1977).
Finally, the physiological order of REM sleep and NREM sleep is inconsistent with the proposed model. The author suggests that “elaborative encoding of episodic memories during REM dreaming may be followed by the instantiation of episodic junctions during NREM” (sect. 8, para. 1). However, REM sleep usually follows NREM sleep – the first REM episode usually occurs at the end of the first sleep cycle, and generally there is more REM sleep in the second half of night and more NREM sleep in the first half of night. Thus, the physiological order of REM sleep and NREM sleep would rather suggest another sequence of processes: Consolidation and integration of episodic memories via the formation of semantic network connections probably already occurs during NREM sleep, and these connections may later be activated (and, possibly, more thoroughly integrated) during REM sleep (Diekelmann & Born Reference Diekelmann and Born2010).
To summarize, the analogy between REM dreams and AAOM mnemotechniques is convincing at only a superficial phenomenological level but is odd, at a closer look, for several reasons: Unlike mneomotechnically encoded material, REM dreams (1) are inherently difficult to remember, (2) do usually not allow conscious subsequent retrieval of all interconnected elements, and (3) have been found to support subsequent episodic memory in only rare cases. Furthermore, the physiological order of REM sleep and NREM sleep suggests that REM sleep supports, if anything, a secondary consolidation or integration step.