In her commentary, Kühle claims that the way we use the term “insight” leaves many—mostly philosophical—questions unanswered. While I certainly agree in principle that solving one question often generates many others, I also believe that there is some need for clarification regarding terminology. It seems that the discussion of what insight is and what it isn’t reveals one of the key methodological differences between our disciplines. Whereas philosophy of mind is mainly involved in meta-theory and the conceptualization of psychological theories, the focus of experimental psychology lies on the testing of hypotheses, albeit neither foci apply exclusively. By definition, however, experimental psychology aims at identifying cause-and-effect relationships between observable phenomena by applying experimental methods to induce controlled manipulations of so-called “independent variables”, leading to reproducible changes in “dependent variables”. Although experiments are hypothesis-based, testing specific (confirmatory) or unspecific predictions (exploratory) derived from theory, progress is often made when such an experiment leads to an unpredicted result. Such was the case in the construction of our LuCiD scale.
In the set of lucid and non-lucid dreams investigated and reported on by our group (Voss et al. 2013), we identified a factorial structure in which eight item clusters (which differed from the theoretically predicted ones) showed sufficient common variability to consider the items within each cluster related. These eight factors accounted for a large portion of variance in dream consciousness as defined a priori, and based on theoretical considerations. The items in the item pool statistically identified as the single factor we referred to as “insight” pertained to the verbal communication that one knew one was dreaming while the dream continued. As such, insight would have to be regarded (in an epistemological sense) as understanding that at a particular moment within the dream, the dreamer acquired knowledge about his or her state of consciousness, which would be the hybrid state of lucid dreaming.
As Kühle correctly points out, this may or may not be true, however. It is just as possible that a dreamer who states upon his or her awakening: “I knew it was a dream while the dream continued” only thought that he or she knew, while in truth, he or she may have sensed, felt, or experienced that the ongoing dream action was not real. This would then pertain to a phenomenological experience similar to what Duncker (1947) refers to as “conscious participation” (p. 505), describing the sensorial experience that one is, at a particular moment, consciously aware of (pp. 508–509). On the other hand, even if we really experienced insight in a phenomenological sense, how can we be sure that this experience was not the result of the epistemological recognition of some sort of incongruence within the dream at some particular point in time? To me, this line of thought resembles that revolving around the question of whether we can be certain that a dream is really a dream and not something else. Philosophically, this is of course fascinating. But to experimental psychologists, such a discussion is unsettling because it is so difficult to translate into testable, i.e., operationalizable, hypotheses. Our admittedly very pragmatic approach is to define underlying assumptions such as “we assume that dream reports generated from REM sleep awakenings are mentations generated during REM sleep and (fractionally) remembered (at least) until questioning” or “we assume that verbal accounts are reliable and valid”. These assumptions can then again be challenged by separate experimental studies. In the case of doubting the existence of REM sleep dreams, an experiment could be set up, for example, interrupting different states of arousal such as meditation, daydreaming, NREM sleep, or REM sleep and questioning the subject with respect to immediate recollections of mental activity. A comparison would lead to the conclusion that reports from REM sleep awakenings differ fundamentally from reports gathered from other states of arousal. This has, of course, been successfully achieved and repeated many times. However, the question is still not solved. It is doubtful, for example, whether an arousal from REM sleep enables as accurate a report as an arousal from the meditative state. Similarly, we cannot exclude the possibility that REM sleep alters mnemonic processes in a different way to NREM sleep, so that obvious discrepancies in NREM and REM reports are due to state-dependent retrieval and filtering processes and not at all related to different fantasies generated during the particular state.
In the same way, it certainly is appropriate to wonder about the true nature of what we refer to as “insight”. To psychologists, the explanation that a factor name is really only an attempt to describe a commonality between different but related observations is probably satisfactory. To philosophers, this will of course not be the case. However, with psychological pragmatism in mind, I would like to point to some empirical findings (and their immanent difficulties) regarding the question on how to further explore the nature of insight in lucid dreams: when we constructed the LuCiD scale (Voss et al. 2013), we started out with a set of 50 items that were selected on the basis of theoretical consideration. In a first step, these items were tested on a large sample of dreamers, leading to 158 dream narratives considered valid. These were then analyzed for factorial structure as well as for item reliability. Several items that might have been potentially informative regarding the question of epistemology vs. phenomenology proved either indistinct in differentiating between lucid and non-lucid dreams or they yielded too high statistical item difficulties so that they had to be eliminated from further evaluation. Some examples are:
While dreaming my sensations were the same as when I imagine something or daydream during wakefulness
While dreaming I was convinced that I was awake.
I wasn’t in the dream, I had no self.
While dreaming I felt that I knew where I was sleeping.
While dreaming I was more than one person.
This finding of no-difference is of course by no means sufficiently informative to consider the question of insight in dreaming solved or even solvable. The finding of high item difficulty in particular poses some problems: items are considered difficult if they do not yield a reasonable number of affirmative answers (Moosbrugger 2008; Schermelleh-Engel & Werner 2008). Thus, an item that is not often selected as true will be eliminated from analysis although it might contain valuable information, e.g., that the statement is considered false by the majority of participating subjects. Further, in the case of subjects awakened from sleep, they may not affirm an item although it is true, simply because they are not yet able to comprehend its content (sleep inertia). For example, the item “I wasn’t in the dream, I had no self” was not often selected as true. Was this because in most cases, dreamers felt they did have a self or was it because they didn’t understand what was asked of them? I hope that this example highlights some of the problems that arise when we try to subject philosophical theory to experimental testing. Perhaps a different design, opting for a specific comparison of questions addressing epistemology vs. phenomenology during a steady state of wakefulness (such as mindwandering or meditation) might generate more concrete answers, avoiding sleep inertia effects should they exist. We look forward to such results.