Dreams and Memory

A new theory asserts that dreams are crucial for episodic encoding of memories.

Posted Dec 15, 2013

Why do we dream? In the scientific quest for understanding potential functions of dreams many investigators have proposed that dreams may be important for memory but few have presented any convincing empirical evidence for such a link. It seems reasonable to me to assume a link exists given that both of the major sleep states participate in memory processing. It is entirely possible that brain mechanisms that mediate memory processing may also produce cognitive contents that reflect or participate in that memory processing. But again the data that directly speak to the issue are not yet there so far as I can tell.

Some investigative teams have shown that one can identify various types of memory fragments in both REM and NREM dreams, but demonstrating that dream contents contain memory fragments is a far cry from demonstrating a functional role for dreams in memory processing. We have no way of telling whether the demonstration that memory fragments occur in dreams is a trivial finding or not. After all memory fragments occur in waking consciousness as well, so finding memory fragments in dreams may not tell us anything special about dreams per se.

In addition, it is difficult to tell whether dreams could use any other kind of mental content besides memories to “compose” a dream narrative. It would therefore be surprising if we found no memory fragments in dreams at all. Their presence in dreams are a mere consequence of the fact that memory fragments must be ubiquitous in all forms of cognitive processing. Where can the cognitive system go to get mental content if not memory stores? Even if dreams were, as Freud argued, about future (not past) wishes/desires, dreams would still need to tap memory stores to construct mental content. So the fact that memory fragments occur in dreams may just be a trivial prerequisite for constructing any mental content at all.

In the effort to clarify the role, if any, of dreams in memory processing it would help if the field of dream studies had a theory about the logically possible roles dreams might play in memory processing. Such a theory would enable us to better evaluate what role memory fragments are playing in both the dream system and in the memory systems. With such a theory we could evaluate the available data on the issue and identify what kind of data we need to move the whole field forward.

 In the December issue of the journal Behavioral and Brain Sciences (Behav Brain Sci. 2013 Dec;36(6):589-607. doi: 10.1017/S0140525X12003135), Professor Sue Llewellyn from the Faculty of Humanities, University of Manchester, United Kingdom, proposed just such a theory. She is to be commended for the effort and it is a shame that no scientist (including myself) who specializes in the study of dreams has proposed such a theory themselves. It often takes “outsiders’ to prod a whole discipline forward and hopefully Professor Llewellyn efforts will do just that. 

Llewellyn proposed a key role for dreams in elaborative encoding of episodic memories during rapid eye movement (REM) dreaming. She proposed that the cognitive system in REM can profitably be understood as utilizing the ancient art of memory (AAOM) principles to enhance memory. The ancients used visualization, bizarre association, method of loci /organization, narration, embodiment and other associational techniques to improve their memories. These techniques certainly worked as scholars are agreed that the ancients exhibited prodigious feats of memory. Do they operate in REM dreaming? Llewellyn thinks so.

It is plausible that episodic memory networks interconnect profusely within the cortex thus setting up semantic networks that operate on associational principles. Llewellyn proposes that these create omnidirectional "landmark" junctions. A REM dream scene is retained by the hippocampus as an index and instantiated as a junction in cortical networks in NREM stage 2 sleep thus setting up cognitive platforms for memory encoding during REM-NREM sleep cycles. There is much more to the theory than this snippet suggests but space precludes detailed dicussion.

What I like about the theory is that it implicitly argues that episodic memory encoding must in some sense depend upon dreams. But as several commentators on Llewellyn’s paper pointed out, there is no clear evidence that loss of dreaming results in memory deficits. The problem with these critics’ positions however is that it is really difficult to find people who have completely lost the ability to dream.

On the other hand there is now pretty good evidence that some people who claim to have never experienced dreams-yet their memories are intact. Similarly when dreaming is suppressed via chronic use of certain antidepressants, there is no discernible effect on memory encoding. So the jury is still out on the links between memory and dreams but Llewellyn’s theory at least puts discussion of these complex issues firmly on the table of dream scientists everywhere.