Corlett et al., 2014, Dreams, reality and memory: confabulations in lucid dreamers implicate reality-monitoring dysfunction in dream consciousness, Cognitive Neuropsychiatry.
This study investigated the relationship between lucid dreaming and reality monitoring, which is currently a really big interest of mine. The article is a bit hard to follow at times, because it seems very clear that their results are in the opposite direction of sensible predictions (they hint that this is the case). As a result, the Discussion section has a lot of HARKing, but interesting thoughts none-the-less.
They recruited healthy participants (about 30), and had them fill out the “Baseline Lucidity Assessment” (BLA) survey originally used in Neider et al. (2011). They split participants into high/low lucidity groups based on their BLA scores. While this is likely a reasonable approach, this my first major concern – that it is not certain the BLA measures lucidity (or equivalently, “dream awareness”) alone. It is only 5 questions, and some of them might are certainly not lucidity-specific. The questions, which are all treated equally in total scoring (summed up), are as follows:
- I can usually remember my dreams.
- My dreams have a lot of detail that I can often remember.
- Sometimes I can tell that I am dreaming during a dream.
- Sometimes I can control what happens in my dreams.
- Sometimes I can dream about things that I have decided I want to dream about.
They split into low (<=12) and high (>=18) lucidity groups, where scores range from 5-25.
The task participants complete isn’t diagramed, but is one often referenced by Schnider, who seems to be a leading researcher in confabulation (Nat Neuro Review here). The task is that participants view runs of images. The images are the same within each run, but their task is to identify whether each image is new/old with respect to the current run (as opposed to the previous runs).
Using signal detection theory analyses on their behavioral data, though I think they are using binary (yes/no) decisions rather than confidence ratings I am accustomed to seeing when SDT is applied to memory data. They find no differences between high/low dream recallers in d-prime, which should represent the ability to distinguish “truly old” and “truly new” items on each run. However they find a difference decision criterion, which is indicative of “the threshold subjects use to decide whether to respond that an item is a repeat.”
The catch is that while they setup an expectation that dream awareness should be indicative of better performance on this task, they claim to find that the high lucidity group have a more liberal response bias (i.e., “more likely to indicate that a picture was familiar to them, even if it was novel”). Now I’m not sure how bad this is, and how much it explicitly relates to reality monitoring in the sense that it is discussed by Marcia Johnson. As the authors state here, there are both neuroscientific and psychological reasons to expect lucid dreamers might be better at a reality monitoring task. The Neider study referenced above showed that a high lucidity group (also using BLA) performed better on the Wisconsin card sorting task, and both that task and this task are supposed to recruit orbitofrontal cortex. Psychologically, reality checks have large overlap with the act of reality monitoring, a practice that increases lucid dreaming frequency.