Deconstructing Dreams: The Spandrels of Sleep. O. Flanagan.: The Journal of Philosophy. XCII, 1995. Pp. 5-28

Flanagan's main goal is the understanding of consciousness, which he thinks requires a method that not only grants equal respect to phenomenology (first-person reports of experience), psychology, cognitive science, and neuroscience, but that also takes evolutionary biology and anthropology into account. Drawing on all these considerations, Flanagan presents what he calls a double aspect model of dreaming. He maintains that dreams, or the subjective experiences that we have while asleep, have no interesting biological or evolutionary function, and that dreams can nevertheless be useful in the project of self-understanding. The author makes the following argument. Mentation occurs during both REM sleep and non-REM sleep. The latter is more or less continuous with waking thought, while REM mentation, both phenomenologically and neurologically, is radically different, in fact, closer to psychosis. The cortex attempts to do with REM mentation what it always does, namely, make sense of stimuli; so it more or less successfully tries to fit these stimuli into the narrative structures that are already in place, structures that have to do with the dreamer's self-representations and her or his ongoing life concerns. It follows that even in REM dreams the cortex is expressing what is uniquely on a person's mind.

Flanagan points out that this view leaves room for dream symbolism and even for something similar to the distinction between manifest and latent content. His conclusion, then, is that while dreaming does not serve some evolutionary purpose, we, as creatures to whom self-understanding is important, have ingeniously devised methods of putting dreams and dream interpretation to use. “Spandrels serve functions even though they are sequelae of the design the architect is focused on putting in place. Being a spandrel does not make something nonfunctional.” So also with dreams.