4 Simulation theories of dreaming

According to the simulation view, dreaming is a special case of phenomenal consciousness, or the phenomenal level of organization being activated in the brain. Waking consciousness and dreaming are manifestations of the same natural biological phenomenon in the brain, but they occur in different contexts and under different conditions. The simulation theory of dreaming is anchored to a more general theory of consciousness, which in turn is anchored philosophically to weak emergent materialism and multi-level explanation (Bechtel 2008, 2011; Craver 2007; Revonsuo 2006, 2010). In a multi-level explanation of a mental phenomenon, several different explanatory dimensions surround the target phenomenon: the downward-looking explanation specifies its neural correlates and mechanisms; the backward-looking mechanisms specify what has causally brought about or modulated the phenomenon (e.g., day residues or traumatic experiences that directly influenced specific contents of dreaming; the ontogeny of dreaming—how dreaming came about during individual development; phylogeny—how dreaming emerged and might have been selected for during evolutionary history[2]); and the upward-looking (functional) explanation—how does dreaming guide or change consequent mental states or external behaviours? Only after all these explanatory dimensions can be accounted for may we be said to have a comprehensive theory of dreaming, including its function(s) (see also Revonsuo 2006, 2010; Valli 2011).

So far, one general and three separate, more specific simulation theories have been proposed. From a more general perspective, some versions of the Continuity Hypothesis (CH) can be regarded as a simulation theory, as some proponents of it consider the world-simulation itself to be a functional form of dreaming (e.g., Foulkes 1985, pp. 201–202). Three other, more specific simulation theories have been proposed: the protoconsciousness theory (Hobson 2009), which covers the role of dreaming in ontogeny; the Threat-Simulation Theory (TST), which covers the negative contents of dreaming and provides an evolutionary account for them; and the Social Simulation Theory (SST), which covers the social contents of dreaming, including the positively charged ones. Taken together, these theories are at the same time both covering and economical: the simple principle of "internally activated world-simulation" underlies all of them (see figure 1). The proto-consciousness theory accounts for how and why the basic form of the virtual-reality generator comes about in the developing brain, and how during early brain maturation both dreaming and waking consciousness emerge together in interaction. It is, however, the most speculative of the three simulation theories, as we cannot hope to test it with data about subjective experiences describing the postulated fetal dream experience: what is it like to be a proto-conscious dreaming fetus? Thus, its weakness is that dream reports or any other direct evidence of the existence of subjective dream-like states cannot conceivably be empirically collected to test the validity of the theory.

Figure 1: Simulation theories of dreaming. All simulation theories assume that dreaming can be defined as a world-simulation, the form of which is functional. The protoconsciousness-theory is more focused on explaining the form of dreams instead of their specific contents. Threat simulation and Social simulation theories try to explain the content of dreams as having a specific function, while the Continuity Hypothesis assumes the content of the simulation to be evolutionarily non-functional.

The TST and SST, as we have explicated in our target article (Revonsuo et al. this collection) and in earlier publications elsewhere (Revonsuo 2000, 2006; Valli & Revonsuo 2009) are testable as they issue specific predictions concerning the frequency and quality of dream contents under different circumstances. They are also covering, TST potentially accounts for normal dreaming as well as several special types of dreams, where negative dream contents are particularly abundant and dominate (bad dreams, recurrent dreams, nightmares, post-traumatic dreams, children's earliest dreams, dreams in parasomnias such as RBD, night terrors, and so on). Together TST and SST potentially cover a very large proportion of the statistically most frequent dream contents, and the predictions derived from these theories have specific empirically testable consequences as to the quantity and quality of these types of dream contents. As Dresler (this collection) points out, simulation theories also have the advantage of being highly consistent with the peculiar behavioural, neurophysiological, and phenomenal features of dreaming such as isolation from sensory input, motor activity, cognitive reflection, and reality testing. These features are necessary preconditions for running powerful, phenomenologically realistic but behaviorally isolated virtual reality simulations in the sleeping brain. The simulation theories thus have a lot of explanatory power. The concept of world-simulation unifies numerous separate phenomena related to dreaming and makes sense of them under a single concept. In this the simulation theories of dreaming fulfil the requirements of simplicity, coverage, and economy as well as having predictive and explanatory power. Compared to some of the other ideas Dresler presents in his commentary, it appears that currently the simulation theories are amongst the strongest frameworks for the form and function of dreaming.