Llewellyn proposes a novel and ambitious account of the function of dreaming. Her key claim is that “REM [rapid eye movement] dreaming may be the archetypal state for the elaborative encoding of emotional declarative memories” (sect. 4, para. 7). On this view, dreaming serves a mnemonic function by virtue of its adherence to ancient art of memory (AAOM) principles. Based on the analysis of self-experience in dreams, I argue that while there is indeed a surface similarity between dreaming and AAOM in wakefulness, this does not warrant conclusions about the functions of dreaming. I then develop constructive suggestions for the comparison between AAOM in wakefulness and dreams.
In philosophy of mind, it is widely accepted that dreams involve phenomenal experience during sleep (e.g., Revonsuo Reference Revonsuo2006): it is like something to dream (and not just to remember dreams [cf. Dennett Reference Dennett1976; Malcolm Reference Malcolm1959]). While many studies have shown that dreams typically involve a phenomenal self (e.g., Occhionero et al. Reference Occhionero, Cicogna, Natale, Esposito and Bosinelli2005), the phenomenology of selfhood in dreams is impoverished compared with wakefulness. With the exception of lucid dreams, dreams are characterized by the metacognitive deficit: dreamers do not realize that they are dreaming and are unable to distance themselves critically from the unfolding dream narrative. Dreams are experienced not as simulations but as real (Metzinger Reference Metzinger2003a; Windt & Metzinger Reference Windt, Metzinger, Barrett and McNamara2007).
By contrast, AAOM is a technique, a mnemonic tool involving the deliberate and selective imaginative composition of material with the goal of making it more memorable. As Llewellyn's discussion of AAOM shows, this involves several steps, such as dividing the to-be-remembered material into smaller parts and reorganizing it in a personally meaningful, emotionally salient manner. On the level of phenomenal experience, the resulting narrative retains its imaginative character and is experienced as the product of deliberate construction. Indeed, it is hard to see how AAOM could be effectively used as a mnemonic tool if the resulting narratives were not experienced as simulations: They would be confused with reality and lose their instrumental value.
A related difference between dreams and AAOM has to do with the phenomenology of agency. Llewellyn rightly points out that dreams typically involve the phenomenology of agentively interacting with real persons and objects in a real world. This type of agency, however, is distinct from the type of agency that is crucial for AAOM – namely, the phenomenology of deliberately constructing an imaginative narrative. This latter type of agency is typically missing in dreams. Consequently, dreams cannot be described as the product of intelligent and deliberate authorship on the part of a phenomenal self (Windt, in press). Though the narrative structure and apparent meaningfulness of dreams resembles that of AAOM, dreams lack the phenomenology of agency for the process of dream construction. Llewellyn argues that REM dreaming more closely resembles imagination than perception in terms of brain activation and eye movement patterns (sect. 4.2.2, para.4); however, such similarities on the neurobiological level of description are compatible with saying that there are important differences on the phenomenological level of description, for instance with respect to the phenomenology of agency. Moreover, a recent study in which researchers were able to predict the contents of dream reports from sleep-onset awakenings suggests that activation in the visual areas during sleep-onset dreams is equivalent to visual perception (Horikawa et al. Reference Horikawa, Tamaki, Miyawaki and Kamitani2013).
Memory also distinguishes dreams from AAOM. Spontaneous dream recall is a fleeting, highly unstable phenomenon, and a vast amount of research has shown that a majority of dreams are forgotten (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000; Schredl et al. Reference Schredl, Wittmann, Ciric and Götz2003). Detailed dream recall, as in Llewellyn's “Quicksand” dream, is the exception, not the rule. By contrast, a successful application of AAOM principles requires the integration of hard-to-remember facts into a memorable narrative. Though Llewellyn argues that nonconscious hippocampal indexing might account for the mnemonic function of dreams, she owes an explanation of why dreams themselves are not more memorable, especially if, as she claims, elaborative encoding during REM dreams “transpires at the phenomenological level” (sect. 4.2.5, para. 5).
In sum, the dissimilarities between dreaming and waking self-experience prevent dreams from contributing to elaborative encoding in the ways suggested by Llewellyn. Even if we concede that memory is the stuff dreams are made on – and this is simply the empiricist view that, in the absence of external sensory inputs, dreams must be derived from memory sources – there is reason to doubt Llewellyn's more substantial claim that elaborative encoding is what dreaming is for.
Despite these objections, the comparison between dreaming and AAOM might be interesting in other ways. Lucid dreams involve insight into the fact that one is now dreaming and the ability to control the unfolding dream narrative (Voss et al. Reference Voss, Schermelleh-Engel, Windt, Frenzel and Hobson2013). They are also more easily remembered than nonlucid dreams, and short- and long-term memory, both of which are deficient in nonlucid dreams, can reach waking levels (LaBerge Reference LaBerge, Barrett and McNamara2007). Because lucid dreams thereby fulfill the requirements for a successful application of AAOM principles, trained lucid dreamers could actively and deliberately apply them while dreaming. While limiting its scope to lucid dreams arising from REM sleep (Dresler et al. Reference Dresler, Wehrle, Spoormaker, Koch, Holsboer, Steiger, Obrig, Sämann and Czisch2012; Voss et al. Reference Voss, Holzmann, Tuin and Hobson2009), this would be a new way of testing Llewellyn's claim that “REM dreaming may provide the most conducive state for the elaborative encoding of personal emotionally salient memories” (sect. 6, para. 4).
The comparison between REM dreaming and AAOM is not limited, however, to lucid dreams. Whereas Llewellyn argues that elaborative encoding is attenuated in wakefulness as compared to dreaming (sect. 4, para. 7), I would like to suggest the contrary: AAOM might be efficient precisely because it involves a selective but incomplete simulation of the phenomenology of dreaming. On this view, the surface similarities between AAOM and dreaming – such as hyperassociativity, emotional salience, narrative structure, the involvement of self-related imagery, and the importance of visual and spatial imagery (cf. Windt Reference Windt2010) – account for the efficacy of AAOM in wakefulness even though REM dreams themselves do not fulfill an elaborative encoding function. In particular, AAOM is effective only because it fails to replicate other characteristics of (nonlucid) dreams, such as the metacognitive deficit, the lack of agentive control for the unfolding narrative, and their incomplete integration with autobiographic memory. This switch in perspective from the function of dreaming to the efficacy of AAOM in wakefulness might provide a fruitful extension of Llewellyn's comparison between AAOM and dreaming for future research.
Llewellyn proposes a novel and ambitious account of the function of dreaming. Her key claim is that “REM [rapid eye movement] dreaming may be the archetypal state for the elaborative encoding of emotional declarative memories” (sect. 4, para. 7). On this view, dreaming serves a mnemonic function by virtue of its adherence to ancient art of memory (AAOM) principles. Based on the analysis of self-experience in dreams, I argue that while there is indeed a surface similarity between dreaming and AAOM in wakefulness, this does not warrant conclusions about the functions of dreaming. I then develop constructive suggestions for the comparison between AAOM in wakefulness and dreams.
In philosophy of mind, it is widely accepted that dreams involve phenomenal experience during sleep (e.g., Revonsuo Reference Revonsuo2006): it is like something to dream (and not just to remember dreams [cf. Dennett Reference Dennett1976; Malcolm Reference Malcolm1959]). While many studies have shown that dreams typically involve a phenomenal self (e.g., Occhionero et al. Reference Occhionero, Cicogna, Natale, Esposito and Bosinelli2005), the phenomenology of selfhood in dreams is impoverished compared with wakefulness. With the exception of lucid dreams, dreams are characterized by the metacognitive deficit: dreamers do not realize that they are dreaming and are unable to distance themselves critically from the unfolding dream narrative. Dreams are experienced not as simulations but as real (Metzinger Reference Metzinger2003a; Windt & Metzinger Reference Windt, Metzinger, Barrett and McNamara2007).
By contrast, AAOM is a technique, a mnemonic tool involving the deliberate and selective imaginative composition of material with the goal of making it more memorable. As Llewellyn's discussion of AAOM shows, this involves several steps, such as dividing the to-be-remembered material into smaller parts and reorganizing it in a personally meaningful, emotionally salient manner. On the level of phenomenal experience, the resulting narrative retains its imaginative character and is experienced as the product of deliberate construction. Indeed, it is hard to see how AAOM could be effectively used as a mnemonic tool if the resulting narratives were not experienced as simulations: They would be confused with reality and lose their instrumental value.
A related difference between dreams and AAOM has to do with the phenomenology of agency. Llewellyn rightly points out that dreams typically involve the phenomenology of agentively interacting with real persons and objects in a real world. This type of agency, however, is distinct from the type of agency that is crucial for AAOM – namely, the phenomenology of deliberately constructing an imaginative narrative. This latter type of agency is typically missing in dreams. Consequently, dreams cannot be described as the product of intelligent and deliberate authorship on the part of a phenomenal self (Windt, in press). Though the narrative structure and apparent meaningfulness of dreams resembles that of AAOM, dreams lack the phenomenology of agency for the process of dream construction. Llewellyn argues that REM dreaming more closely resembles imagination than perception in terms of brain activation and eye movement patterns (sect. 4.2.2, para.4); however, such similarities on the neurobiological level of description are compatible with saying that there are important differences on the phenomenological level of description, for instance with respect to the phenomenology of agency. Moreover, a recent study in which researchers were able to predict the contents of dream reports from sleep-onset awakenings suggests that activation in the visual areas during sleep-onset dreams is equivalent to visual perception (Horikawa et al. Reference Horikawa, Tamaki, Miyawaki and Kamitani2013).
Memory also distinguishes dreams from AAOM. Spontaneous dream recall is a fleeting, highly unstable phenomenon, and a vast amount of research has shown that a majority of dreams are forgotten (Hobson et al. Reference Hobson, Pace-Schott and Stickgold2000; Schredl et al. Reference Schredl, Wittmann, Ciric and Götz2003). Detailed dream recall, as in Llewellyn's “Quicksand” dream, is the exception, not the rule. By contrast, a successful application of AAOM principles requires the integration of hard-to-remember facts into a memorable narrative. Though Llewellyn argues that nonconscious hippocampal indexing might account for the mnemonic function of dreams, she owes an explanation of why dreams themselves are not more memorable, especially if, as she claims, elaborative encoding during REM dreams “transpires at the phenomenological level” (sect. 4.2.5, para. 5).
In sum, the dissimilarities between dreaming and waking self-experience prevent dreams from contributing to elaborative encoding in the ways suggested by Llewellyn. Even if we concede that memory is the stuff dreams are made on – and this is simply the empiricist view that, in the absence of external sensory inputs, dreams must be derived from memory sources – there is reason to doubt Llewellyn's more substantial claim that elaborative encoding is what dreaming is for.
Despite these objections, the comparison between dreaming and AAOM might be interesting in other ways. Lucid dreams involve insight into the fact that one is now dreaming and the ability to control the unfolding dream narrative (Voss et al. Reference Voss, Schermelleh-Engel, Windt, Frenzel and Hobson2013). They are also more easily remembered than nonlucid dreams, and short- and long-term memory, both of which are deficient in nonlucid dreams, can reach waking levels (LaBerge Reference LaBerge, Barrett and McNamara2007). Because lucid dreams thereby fulfill the requirements for a successful application of AAOM principles, trained lucid dreamers could actively and deliberately apply them while dreaming. While limiting its scope to lucid dreams arising from REM sleep (Dresler et al. Reference Dresler, Wehrle, Spoormaker, Koch, Holsboer, Steiger, Obrig, Sämann and Czisch2012; Voss et al. Reference Voss, Holzmann, Tuin and Hobson2009), this would be a new way of testing Llewellyn's claim that “REM dreaming may provide the most conducive state for the elaborative encoding of personal emotionally salient memories” (sect. 6, para. 4).
The comparison between REM dreaming and AAOM is not limited, however, to lucid dreams. Whereas Llewellyn argues that elaborative encoding is attenuated in wakefulness as compared to dreaming (sect. 4, para. 7), I would like to suggest the contrary: AAOM might be efficient precisely because it involves a selective but incomplete simulation of the phenomenology of dreaming. On this view, the surface similarities between AAOM and dreaming – such as hyperassociativity, emotional salience, narrative structure, the involvement of self-related imagery, and the importance of visual and spatial imagery (cf. Windt Reference Windt2010) – account for the efficacy of AAOM in wakefulness even though REM dreams themselves do not fulfill an elaborative encoding function. In particular, AAOM is effective only because it fails to replicate other characteristics of (nonlucid) dreams, such as the metacognitive deficit, the lack of agentive control for the unfolding narrative, and their incomplete integration with autobiographic memory. This switch in perspective from the function of dreaming to the efficacy of AAOM in wakefulness might provide a fruitful extension of Llewellyn's comparison between AAOM and dreaming for future research.