The Nightshirt Sightings, Portents, Forebodings, Suspicions

Dreams and the Art of Memory: A New Hypothesis About Dream Bizarreness

Dreams wouldn’t be dreams without being bizarre. Their imagery is absurd from the standpoint of waking life, yet they seem sensible during the experience itself—at least, as we remember the experience. With the exception of lucid dreams, we are unaware during dreaming that we are dreaming. Yet even on recollection, there often seems a method to dreams’ madness—they are bizarre, yet they do seem to contain elements of reality, reflections of our waking life that are occasionally witty, even brilliant. Dreaming has supplied countless artists and scientists with inspiration that has paid off in their work.

Dreams’ ‘sensible bizarreness’ has led people from time immemorial to assume that dreams have an origin. They have been thought at various times to come from gods or demons, to be portents of the future, and to symbolically represent our wishes. All these theories of dreaming agree on one fundamental: that dreams are meaningful and thus can be interpreted.

It is only in our century, with the extreme reductionism of scientific psychology and, more recently, the ability to study neural processes directly, that dreams have lost much of their mystique. The scientific revolt against Freud’s inherently untestable theory that dreams are wish fulfillments, as well as against other seemingly mystical theories like that of Carl Jung, has led some of the foremost dream researchers to the view that dreams are inherently meaningless, random firing of neurons that are constructed as meaningful only after-the-fact.

The latter view was first put forward by Nobel Prize-winning biologist Francis Crick, the discoverer of DNA. Crick argued that dreams are just the discharging of mental static, random and meaningless associations, essentially the brain farting. His view was developed and nuanced by neurologist J. Allen Hobson, probably the foremost contemporary scientific dream researcher. Hobson has devoted his career to debunking any notion that dream content might be meaningful or subject to symbolic interpretation. Most recently, he has argued that dreaming represents a kind of ‘warmup’ for waking cognitive processing.

Hobson has admitted to harboring an extreme antipathy toward psychoanalytic interpretation of dreams—an antipathy rooted in a distaste for the admittedly infuriating habit of old-school psychoanalysts to reduce all human behavior to wish fulfillment. Like the sophists of old, or the Medieval Inquisition, they can turn everything you say around and assert that it “really” reflects some secret repressed desire, and there is no way to argue against such an interpretation without one’s argument being turned around to support it. Hobson’s desire to empty dreams of meaning is thus understandable. Yet the hatred of the Freudian legacy arguably biases his work.

There is however another way of seeing the dreaming mind: that dreaming directly reflects the process of memory construction–indeed, that dreams are the associative links of memory as they are being forged.

It has been well established that during sleep memories are consolidated and that recently learned material and recently acquired skills are somehow solidified. In laboratory experiments, people who have learned new material remember it better after “sleeping on it” than if they don’t. The hippocampus, an organ that plays a huge role in making memories, is extremely active during sleep. And during sleep, complex material is simplified, reduced to the “gist.” Some sleep researchers have concluded that dreams are involved in memory formation. However, those who have tested this by studying actual dream content have limited themselves to overt references in dreams to daytime events–e.g., thinking or reading about something and seeing if it appears in a dream, or suppressing the thought of a friend and seeing if that friend appears in the dream (i.e., compared to not suppressing such a thought). The bizarre details are assumed to be detritus or garbage, simple hyperactivation of the brain’s associative networks.

I am unaware of any research in psychological science that has actually taken seriously the possibility that all dream elements represent meaningful associative links. Yet I suggest there is a lot of circumstantial evidence that dreams are, in their entirety, and precisely through their bizarreness, mnemonic images. Most of this evidence is unknown to psychologists, however, because it comes from the unlikely domains of history, anthropology, and, oddly enough, esoteric philosophy.

To build such a hypothesis about dream bizarreness, I suggest looking at the method commonly used by people in nonliterate societies, and in pre-Gutenberg Europe, to remember things they have learned: Ars Memoria, or the Art of Memory—also called mnemotechnics. Dreaming, I suggest, is simply the Art of Memory operating automatically while we sleep.

The Art of Memory

The Art of Memory works by free association. Here’s a deliberately oversimplified example: A college freshman hears in her History 101 lecture that the Normans invaded England in 1066. How is she going to remember that fact and that date? On their own, the bare facts, Normans, England, 1066, may mean little or nothing to her—she may have never been to England though she has a lot of vague impressions from TV and movies, but she doesn’t know what a Norman looks like and, like most numbers, the date 1066 is just a string of digits representing a random time almost a millennium in the past.

Mnemonists since antiquity have all given a single, simple solution to the problem of how to memorize a fact like “the Normans invaded England in 1066,” and if she is lucky, our hypothetical undergraduate may have been taught it by a teacher or adviser at some point: If you want to remember a new fact, something you didn’t already know, the key is to associate it—that is, link it—to something you do know. That is, link the new to the old. In doing this, you need to distort it or substitute the to-be-learned material with something else through the use of metaphors, shorthand symbols, and especially puns. The result is always a bizarre, absurd mental image. Such an image will almost always be highly idiosyncratic, based on the individual’s unique knowledge and life experience.

So, say our freshman is a maven of 60s rock trivia and knows the years of every Bob Dylan album like the back of her hand. The year 1066 thus might immediately call to her mind 1966, the year Blonde on Blonde was released. And while she may not know what an actual historical Norman looked like, she may have an uncle named Norman. Instead of repressing such “absurd” connections as they occur to her during the lecture, a clever, mnemonically-trained student will actually allow her naturally playful mind to make and even embellish those associations on the fly—allowing, perhaps, a mental image of a weirdly blond-haired version of her uncle taking a tour of the Tower of London, perhaps remembered from a postcard or TV program.

In short, the mnemonist merely lets flourish whatever crazy association occurs to her. Experience with this method shows that it doesn’t take mental effort or concentration—only willingness not to censor an association that spontaneously comes up.

Here’s the thing: If our hypothetical student doesn’t know this Art of Memory trick, her mind may do it for her when she’s sleeping, and she will dream an absurd image not that different from the one I described. If placed on an analyst’s couch, she might readily produce an image like blond uncle Norman touring the Tower of London. Induced to free associate on the image, she may recall the lecture. The therapist would realize that this is what Freud called a “day residue”—a memory of the previous day. (Actually it is common to dream of events of the previous two days). No examination of surface content–the level at which dream researchers investigate–would uncover these associations.

This is a hugely oversimplified example, involving a simple learned fact. Dreams are really much more complex because they encode not just things we have learned, i.e. new knowledge, but our most pertinent day-to-day experiences in all their sensory and emotional richness. The byzantine networks of association in dreams reflect the richness of daily life and the complexity and paradoxes of our priorities and longer-term memories, such as memories of childhood. The puns in dreams are generally far more multifarious than our blond Norman example, linking multiple associations and utilizing multiple sensory modalities, not just the sounds of words. Think brilliant sight gags; imagine touch, taste, smell, and emotion puns, not just verbal ones. The bizarre objects and places in dreams, if you unpack them, connect to multiple things we did or thought about in recent waking life as well as to long-term preoccupations and things in our past.

The Mnemotechnic Hypothesis

So I suggest that dreams are probably mnemonic devices, but taken to a higher degree of complexity than the conscious mind of even the most highly trained mnemonist could achieve. This is why dreams take a lot of digging–indeed, precisely free association, the method advocated by Freud but discredited by today’s psychological scientists, to arrive at even a fraction of the experiences they encode, and they will always largely evade our efforts and parts of them will just seem senseless. But it is still, I suggest, the basic principle. And it is why, even if the function of dreaming is not simply to represent our wishes, Free-association really works to unpack the brilliant and uncanny webs of association in a dream. Try it–even a simple dream, when unpacked, can fill pages and pages in a notebook with all the associations and day residues contained in it.

The “dream” of “blond uncle Norman” visiting the Tower of London is silly, obviously. The genius of mnemonic devices and dreams is that every person’s individual experience gives them a completely unique and vast set of associations readily usable for fastening new experiences and newly learned facts to older, more firmly fixed images in our memory. Many or even most of these associations are highly personal, and not the sort of thing you could or would ever explain to someone else. This personal or idiosyncratic nature of mental associations is probably a better explanation than Freudian “resistance” for people’s reluctance to free-associate about their dreams to anybody but their therapist. The distortions and apparent symbolic “disguises” in dreams are really just the hooks whereby events of the previous few days are connected to each other as well as to more longstanding themes, and those themes inevitably include lots of embarrassingly personal memories and feelings. Their idiosyncratic nature also causes dreams to be boring when told to other people.

The “hooking” characteristic of the Art of Memory, and the way it inevitably distorts to-be-learned material, cannot be stressed enough as the likely basis for dream bizarreness as a natural mnemotechnic: A new item on its own can find no purchase in memory unless it is bent, distorted somehow, to forge a connection to something old.

The mnemonic hypothesis explains, in a way that Freudian theory can’t, why dreams are so full of “day residues,” yet why these events or objects encountered during the day are almost never rendered literally, and consequently why we seldom recognize most of their daytime referents on awakening.

Psychodynamic Theories Reassessed

So, in light of this mnemonic hypothesis, let’s look again at the out-of-fashion psychodynamic theories of Freud and Jung.

The mnemotechnic hypothesis runs counter to those theories when it comes to the function of dreaming. Dreams are not fundamentally wish-fulfilments; nor are they mechanisms of compensation and individuation, as Jung argued. Yet the mnemotechnic hypothesis does not counter the basic idea of those theories that dreams can be used to uncover and examine our innermost desires or reveal our true (or whole) selves.

Here’s why: It only makes sense that our long-term memories should be organized in terms of our deep and lasting priorities, hopes, fears, and insecurities—the things that matter to us. The DNA of our long-term memories ought to be the unresolved aims, the ongoing pains, and the sources of pride and self-identity we carry through life. Daily experiences that don’t matter to our priorities would not get strongly “hooked” and thus would not be retained, at least for very long.

Literalists unwilling to accept a meaning-centered theory of dreaming, like Crick and Hobson, have suggested that dreams help us forget or discard mental waste, and the mnemonic theory agrees with them on this point. Memory and forgetting are two sides of the same coin: Unused neural connections don’t get reinforced night after night in the ongoing melting and solidification of long-term memory that dreaming represents. Dreams do help us forget, or let go, of what doesn’t matter. The difference is that dreaming also, at the same time and for the same basic reasons, represents a process of clarification, consolidation, and fixation of the things that do matter.

That memory and forgetting are interlinked is well-established in psychology. Researchers who study memory know that long-term memory is constantly reshaping itself. Every time we remember an event, we distort it—it is changed in the recollection. Over time, these metamorphoses make our memories more generic; specific recollected experiences become more and more unlike what may “really” have occurred (as videotaped by a hypothetical God’s-eye camcorder). In short, memory is pliant. If I’m right, dreaming provides a direct window not only onto the fastening of new experiences in memory but, simultaneously, onto this reworking and reshaping of older mental stuff.

It is important to be clear on the ontology of dreaming. Dreams, in the mnemotechnic view, do not cause memories to be created. Rather, they are the experience of new memories being created. It’s a fundamental difference. There is an axiom in neuroscience of learning: neurons that fire together wire together. That is, new neural connections are reinforced by firing. I suggest that dreams are the experience of this firing, the experience of the forming of new mental associations, the rewiring of the brain. Dreaming is the experience of neural rewiring, the experience of memory formation. One possibility is that it may be precisely the firing that converts the highly punny associative links of relatively short-term memory into the more logical associations of long-term memory, a process believed to be governed by the hippocampus.

Far from reflecting the sad inaccuracy of the dreaming brain (the literalist-scientist view), dream distortion in the mnemotechnic view instead represents the genius of a system honed through billions of years of evolution to fasten life experiences onto the core of our deepest priorities, thereby retaining and contextualizing them nightly as we sleep. In dreams, old and new, present and past, come together, merge, and forge something subtly different from either, a third thing, and this “third thing” is really what memory is.

Consequently the main difference from the Freudian view of dream distortion is simply that it reverses the direction of information flow: Instead of being the disguising of material so it can get out, past the guard posts and barbed wire that keep unconscious material imprisoned, it is instead simply the alteration of material so that it can get into the usually dormant (you could say “unconscious”) storehouse of long-term memory and become fixed there. You could say that Freud got it exactly backwards. Yet mirrors get things backwards too—inverting right and left—but we can still use mirrors to gain important knowledge of ourselves. We can still use them to shave with.


As I indicated, the major weakness of the mnemonic hypothesis is that there is no direct experimental evidence, and it is hard to imagine an easy way to acquire such evidence given currently available methods.

This is due in great part to the embarrassment factor. There’s no aspect of our thought that doesn’t veer into realms we’d rather not disclose. Most of our thought is private because it is inappropriate – consisting of embarrassing memories, things we have done or thought that we are ashamed of, and aspects of ourselves that are just plain immature or undignified. You may be unaware of how much of your thought is interwoven with this embarrassing stuff, but free associating on dream elements arrives at these things almost immediately. In my experience, no dream interpretation can proceed very far without hitting this domain of things that I’d never share with anyone else, except perhaps a therapist. As a subset of this domain, there are the things that are utterly “forbidden” – the realm of what Freud considered “repressed” thoughts like private and shameful fantasies and fetishes, secret desires for family members or other off-limits targets (e.g., the Freudian classic, the wish to have sex with your mother). In other words, the amazing and constant underground ocean of desires and hatreds that we all secretly harbor but conceal, even from ourselves.

But there’s a much less sinister reason much of our inner associative sea cannot be navigated by anyone but us: Most of our associations are simply things that are so idiosyncratic, so dependent on our personal life experience, that they would never make sense to anybody else, at least without too much explanation to make it worth it – like explaining a private in-joke to someone not “in.” For most of our associations, the only one “in” is oneself; they wouldn’t make sense to another person because, well, “you had to be there.” The common, archetypal symbols that so fascinated Carl Jung do occur in dreams – there is no doubt of this – but they make up only a small subset of dream imagery. This is the reason that Jung’s writings on dreams only seem to apply to very infrequent, remarkable dreams, not one’s everyday (everynight) dream life.

In other words, dreams can never be verified. The dreamer is always an n of 1.

These factors make Freudian theory untestable as well. Yet the mnemotechnic hypothesis does not share one further factor that makes Freudian theory untestable: the presumption that dreams always are wish-fulfillments. The mnemonic theory does not require that we trace dream associations all they way to supposed wishes that the dreamer is not consciously aware of. It was this inherently slippery presumption that inspired Hobson to seek evidence for his reductionistic viewpoint, and he really can’t be blamed.

But even if it is not testable in a way that might satisfy an experimental psychologist or neurologist, there is quite a bit of circumstantial evidence in favor of the mnemotechnic theory, and this evidence makes it, I believe, more parsimonious than either the Freudian or Hobsonian views. These are the main pieces of that evidence:

• We know from a vast number of experimental studies that it is during sleep that memories are consolidated and new neural connections are being made. Knowledge and skills acquired during the day are much better reproduced after a period of slumber. It would make sense that dreaming is either a reflection or a byproduct of this memory-consolidation process.

• To reiterate the main theme of this paper, dreams closely resemble the mnemonic images used by preliterate and early-Modern (i.e., pre-Gutenberg) mnemonists. Like memory images, they are composed of distorted imagery and utilize punny associations, resulting in absurd imagery and composite formations that condense multiple trains of association. The associations readily discovered in dream elements extend in multiple directions at once. (Although it may be spurious, it is hard not to compare this characteristic of what Freud labeled “condensation and displacement” to the form of the neuron itself: a nucleus with tree-like dendrites extending in multiple directions, connecting to the axons of multiple other neurons.)

• Dreams are full of emotionally charged images. Mnemonists also recommended using emotionally charged imagery—things that made you laugh or feel aroused, for example. In fact, sexual images and symbols, as Freud noted, are common in dreams, and sexual images are also some of the strongest mnemonic images—a fact explicitly stated by some Renaissance writers on the Art of Memory (see my post “Sex and the Art of Memory”).

• Dreams tend to fade rapidly on waking. This is the best evidence that, contrary to Freudian or Jungian views, they are not “shows” put on for our benefit, something actually meant to be pondered and interpreted. Instead, like the scaffolding at a construction site, mnemonic images quickly dismantle themselves—they just vanish—once the material is fixed in long-term memory—that is, once it is learned. That remembered material is the completed building, meant for our long-term use. (And when additions or modifications need to be made to the building, or to memory, up will go the scaffolding again, only again to be dismantled afterward.)

• Dreams are organized episodically—typically between 3 and 5 linked episodes or scenes. Another fascinating feature of mnemonic systems is that they use narratives to link images together into larger wholes.

• Dreams always “take place” in a distinctive environment. This is another characteristic organizing feature of mnemonic systems. Lawyers and orators, for instance, would “collect” public spaces like buildings to use later as memory frameworks. They would memorize a legal argument, for example, or a speech, by mentally walking through such a space, “planting” their absurd, emotionally charged memory images along an imagined route. When later delivering the discourse, they would, in their head, walk along the route, retrieving the memory images placed at each station. Often as a dream fades, it is this sense of place that is retained most strongly and even remembered spontaneously months or years later, for no apparent reason. Similar “places” are also the most likely recurrent elements in dreams; it is possible that these dream places are reused again and again, the same way memory spaces were “erased” and reused, again and again, by orators.

• Dream images are highly idiosyncratic. They tend to be either hard to express without embarrassment (for instance, due to their sexual content), or else they are intrinsically boring when related to other people. They are truly meaningful only to the individual. Mnemonic images have the same characteristic of being highly personal, utilizing highly personal associations, not usefully or easily shared.


I recently spoke to a memory psychologist who had never heard of the Art of Memory. I was surprised at this, but it is just another example of the academic myopia that often prevents researchers from seeing possibilities that may be obvious to someone in another field of study.

Hard scientists like Hobson or Crick naturally feel uncomfortable with symbolism, and with things that are nonliteral or irrational. Indeed, the mnemotechnic theory is admittedly rather literary. If they would overcome their antipathy and step outside of their laboratories and stroll through the humanities stacks in their library, they might detect that there is method in dream madness. They might find ample circumstantial evidence for the real essence and function of dreaming in other fields like art and history and philosophy.

The mnemonic hypothesis, I think, is the most parsimonious hypothesis of dreaming’s nature and function, and it accords with the most strange aspects of the dreaming experience: Dreams are extended mnemonic puns whose wit and complexity exceeds our strongest conscious interpretive acumen. The unconscious, as any writer or artist knows, is much cleverer than the conscious mind.

There are far-reaching implications of a mnemotechnic hypothesis of dreaming. Unlike a file cabinet, the storehouse of memory is more like the messy desk of the creative and disorganized coworker who can somehow still find what he’s looking for amid his random piles of paper and post-it notes. He finds what he’s looking for through association, remembering where he stuck a certain spreadsheet not by its title, but because he remembers a coffee stain on it. If our memory is not the ordered, literalistic formation we tend to think it is, an inevitable and perhaps disturbing conclusion arises: Our reality is not what we think it is, either. It may bear little relation to “actual reality.”

If Dream=Memory and Memory=Life, is life then but a dream? In an important way, yes. This insight might give us tremendous power, if we are prepared to hear it.

–Eric Wargo