Llewellyn's is an attractive hypothesis, engagingly presented. In structure – not, of course, in substance – it is like Freud's dream theory. Freud's dream machine exercises a set of operations – the components of the dream work – that, when applied unconsciously in waking, produce a neurotic symptom. Llewellyn's dream machine exercises a set of operations – the components of the ancient art of memory (AAOM) – that, when applied deliberately in waking, elaboratively encode an episodic memory.

Both dream generators account ingeniously for the peculiarities of dream experience, and both achieve an explicit aim. By condensation, displacement, “consideration of representability,” and secondary revision, a Freudian dream expresses a forbidden Oedipal wish (Freud Reference Freud and Strachey1900/1981). A Llewellynian dream – by association, organization, narration, embodiment, movement, and location – encodes the to-be-remembered material. In both cases the latent dream is obscured – in Freud's model by cunning, in order to evade the “psychic censorship,” and in Llewellyn's (she has no use for a calculating homunculus) by guileless neural processes, in order to realize what I shall call the AAOM-in-REM. These formal likenesses entail another. Against the school of “bottom-up” dream construction, Llewellyn's theory, like Freud's, is a “high” theory: an explicitly “top-down” machine.

What if – both on the neural axis and in regard to imaginative complexity – dreaming is neither a “high” nor a “low” process, but an egalitarian one? A case in point is what might be called Llewellyn's telencephalocentric treatment of levels of acetylcholine (ACh) in waking and in sleep. Table 1 lists and describes the main cholinergic neural groups in the central nervous system. It is fair to ask, can the AAOM-in-REM accommodate the realities of ACh-in-REM, from hindbrain to forebrain? (In Table 1, cell groups are designated by accepted nomenclature [Wainer & Mesulam Reference Wainer, Mesulam, Steriade and Biesold1990]; citation is provided only for recent claims concerning function.)

Table 1.

Cholinergic cell groups activated in rapid eye movement (REM) sleep

EEG: Electroencephalogram

Indisputably in REM sleep – if you will, in Table 1 -in-REM – neural excitation is high in each of the cell groups Ch1–Ch8. Also indisputably, REM sleep and the neural distribution Ch1–Ch8 are highly (and probably inseparably) conserved across species. A sufficient theory of human dreaming must incorporate those facts. Table 1-in-REM can do so.

But can Table 1-in-REM account, as Freud's and Llewellyn's models may seem to do, for the properties of dreams? Pending the outcome of that question's empirical investigation, I propose a tentative answer, or at least a prologue to an answer: By analogy to the cholinoceptive retinal wave of early neural development, in REM sleep or its prototype in at least all vertebrate species, “waves” of excitation sweep over the entire cholinoceptive central axis. These waves, like retinal ones, propagate stochastically but acquire coherent structure. Thus a “dream wave” might generate a dream scene. In its favor in this respect, Table 1-in-REM easily generates obligatory dream locomotion rather than tendentious (Freud) or highly recommended (Llewellyn) dream locomotion. The table readily accounts for the typicality of typical dreams: fearful dreams, for example, or dreams of fleeing or of flying. Against Llewellyn's reading of Sprenger et al. (Reference Sprenger, Lappe-Osthege, Talamo, Gais, Kimmig and Helmchen2010), the table produces actual REM sleep saccade trajectories, not creative ones. As for Ch3 (diagonal band of Broca, horizontal limb), if one takes into account relative species primacy of olfaction, Table 1-in-REM would generate few olfactory dreams in humans but – this is written only partly in jest – many olfactory dreams in dogs. Most important, where cholinergic drive clamps “idea” onto “activation,” the table generates meaning: probabilistic, to be sure, and often oddly realized in REM's mixed landscape of cortical activation and inactivation, but meaning nonetheless. For Table 1-in REM, even an “Icarus” dream is a cinch.

To entertain the idea that the experienced, intact brain constructs dreams out of directionally unbiased swaths of central cholinergic excitation is to jettison a host of dream theories, both “high” and “low.” The AAOM-in-REM is well worth salvaging, however – even if it must relinquish pride of place to a more distributed anatomy, and pride of encoding to a less singular function.

Finally and incidentally – this is the stuff of a different commentary – I am mystified, as if awakening from a dream, by the AAOM's situation within the temporal state architecture of sleep. Exactly what happens where, and when?