Why do we forget our dreams?

Dream research indicates that many people may dream 4 -6 times per night (Scientific American (2011) states we dream on average, 5 times per night, working out to approximately 1, 852 dreams per year) – some people remember all their dreams, whilst others recall none at all and believe they ‘don’t dream’. Dream worker, Jeremy Taylor of the Marin Institute of Projective Dream Work suggests that because dreams are often complex, unusual and distinctly different from our real-life experiences they may be difficult to comprehend and express in thought and language. They may also be multi-layered and conceptually abstract – a state Taylor describes as being “not yet speech-ripe”. Dream recall may also be affected by disinterest in the dreaming process (it is common to find dream recall improves simply by taking an interest in dreaming or actively encouraging yourself – using various methods described in this Blog – to remember dreams upon waking); alcohol or drug consumption; stress or exhaustion; moon phases and biorhythms; traumatic events; or inability/unwillingness to wake up at the time of dreaming. 

Content-based Theories

Other ‘content-based’ theories of dreaming have attempted to explain ‘sleep amnesia’ – for example Salience Theory, which states that we may forget dreams simply because they are meaningless and have no application in waking life. L Strumpell, a theorist working in Freud’s day, posited that dreams are forgotten because dream content is not organised in a comprehendible manner. In Ellman & Antrobus, The Mind in Sleep: Psychology and Psychophysiology (1991), the authors suggest that dream kinesthesia may be responsible for forgetting dreams – when we wake, our physical movements cause us to ’lose’ the dream. Another possibility is that our memories are formed through repetition and finding connections to other parts of our lives. So when dreams are especially unique or too undefined to be relatable, it’s harder to tie them to real-life events and remember them. 

Interference Theory

As with all aspects of dreaming, there is no definitive scientific explanation as to why some dreams are forgotten. One study published in Journal of Clinical Psychology (2006), conducted by the Southern Illinois University at Carbondale tested Freud’s theory of repression against Interference Theory (dream kinesthesia falls under this theoretical approach to dreaming) and found preference for the latter explanation, although it is impossible to isolate a single factor responsible for sleep amnesia. In 2009, Scientific American published a study by Christof Koch entitled ‘A ‘Complex’ Study of Consciousness’ which explored that happens to the mind during sleep. Koch pointed out the incongruity between the brain’s high levels of activity during REM sleep and our inability to recall dreams upon waking, plus the fact that nerve operation does not differ much during this stage as from periods of ‘quiet wakefulness’. He suggests that the brain waves which occur during REM sleep may be the cause of sleep amnesia, preventing us from remembering dream content. Natalie Angler, a scientific journalist for the New York Times states that 75% of dreams recalled in dream laboratories and dream journaling, recall negative dream content – supporting theories which suggest that our dreams are a way of working though troubling thoughts or emotions. According to current science, clarity of dreams depends on neurochemical conditions in the brain.  Dreams are forgotten due to deficiencies in the hormone norepinephrine in the cerebral cortex, and more generally, due to the fact that dreams are not highly conscious activities in the brain and thus are not consolidated as a memory.

Earnest Hartmann, professor at the Psychiatry department of Tufts University School of Medicine and director of the Sleep Disorders Center at Newton-Wellesley Hospital, states that forgetting dreams is usually attributed to neurochemical conditions of the brain during REM sleep – the most compelling evidence being an absence of the hormone norepinephrine in the cerebral cortex – an area of the brain with functions in language, memory, consciousness and thought. A 2002 study published in the American Journal of Psychiatry supports the theory that a presence of noreprinephrine in humans enhances memory, although its role in learning and recall is unclear. However, a lack of norepinephrine does not fully explain why we forget dreams so easily. Recent research suggests that dreaming exists on a continuum of mental activity and functioning – focused concentration is on one end of this spectrum, whilst dreaming and mind-wandering is at the other. Dreaming – whether ‘day dreaming’ or during REM sleep – is a less consciously directed form of thought and therefore it is not easy to remember – because we are good at forgetting non-essential, non-relevant details and tend to remember things of emotional significance. Mulling over important thoughts activates our dorsolateral prefrontal cortex (DLPFC), a brain region that facilitates memory. Although most dreams vanish, certain ones tend to remain. These dreams were so beautiful/anxiety-provoking or bizarre, they captured our attention and increased activity in our DLPFC. Thus, the more impressive your dream or thought, the more likely you are to remember it.

It may be that dreams are not encoded into our memory, because they are unrehearsed, unpredictable – and simply because we fail to pay attention to them. When we are awake and want to remember something, we make a conscious effort to remember it – to subscribe it to memory so it can be effectively recalled and acted upon at an appropriate time. It is well known – and documented on this Blog – that the best way to recall your dreams is to make a conscious decision to remember them and give yourself a ‘reason’ to recall – i.e. for the purposes of recording or interpreting them. Studies of dream recall – using subjects who rarely recall a dream (less than one per month) and those who routinely recall the majority of their dreams (several per night) found that the biggest difference between the groups was the motivation to pay attention to, and recall, dreams. The people who best recalled their dreams were people who had decided that their dreams were worth remembering. 

Memory Consolidation Theory

Memory Consolidation Theory states that blocking REM sleep impairs the ability to perform procedural tasks - tasks that involve the learning of a skill through a sequence of steps that involve making predictions. For example, rats trying to find their way around a complex maze to get a reward remember much better how to do the task if they have had REM sleep. Without REM sleep, the knowledge becomes damaged. So REM sleep would seem to facilitate this type of learning. It has also been shown that, in complex learning, memory is improved if we have both slow-wave sleep and REM sleep, otherwise the knowledge doesn't seem to survive quite so well. So, clearly, there is some connection between memory and REM sleep. But is there evidence that the REM state and dreaming exist fundamentally to carry out memory consolidation? The first piece of evidence that goes against this possibility is that taking MAOIs, the antidepressants that suppress REM sleep, does not lead to memory impairment. So, although there is some evidence that certain types of learning do seem to be improved by REM sleep, most dreaming cannot be about retaining new learning experiences, according to Griffin and Tyrell (2006).

The second fact that goes strongly against the memory consolidation hypothesis is that many people fail to remember their dreams! It is very rare to vividly remember all dreams, unless you've trained yourself to do so, despite the fact that we dream for about two hours a night. Dream researcher Allan Hobson, an expert trained to recall his dreams, has himself pointed out the fact that, when he looked at the number of dreams he had recorded against the number he had forgotten, it was something like 0.0002%. So it seems hard to see how we could be dreaming to make our memories permanent if we forget the material we are processing as soon as we open our eyes (yet, as certain types of memory do seem to be facilitated by REM sleep, any state-of-the-art theory of dreaming must be able to account for it).

Activation Synthesis Theory

Hobson and McCarley proposed the Activation Synthesis Theory.  Laboratory studies of brain waves show that, just before we go into REM sleep and during it, powerful electrical signals pass through the brain like a wave. On electroencephalogram recordings (EEGs), they appear as sudden spikes. The signals arise from the pons (P) in the brainstem, from the neurons that move the eyes, and then travel up via a part of the midbrain called the geniculate body (G) to the occipital (O) cortex in the higher brain - so are known as PGO spikes (they also constitute what is termed our orientation response, which, when we are awake, is what directs our attention to any sudden change in the environment, such as a sound or movement). Hobson and McCarley's theory was that these PGO spikes were sending a random ‘barrage’ of stimulation through the brain every so often, activating the whole cortex as a result; the higher brain had to try and make some sense of this random barrage, and dreams were the result. Dreams, therefore, were an epiphenomenon: they had no intrinsic meaning. They were just the brain's efforts to synthesise some sense from random signals. Evidence has accumulated over the last 30 years to disprove this theory. The first piece of evidence that disproved it emerged once PET scanning of the brain was developed. According to Hobson and McCarley's original theory, a barrage of random stimulation coming up periodically from the brainstem was synthesised by the prefrontal cortex into dreams, but scans of the brain in the REM state showed that the cortex was very selectively activated. The emotional brain (the limbic system) and the visual brain were highly activated but the prefrontal cortex was excluded from this stimulation (the very part supposed to be doing the synthesising). Indeed, Hobson himself, over the last few years, has been so drastically redrafting the theory that it is just a pale shadow of its original presentation.  Even he now agrees with the evidence that (instead of global forebrain activation being responsible for dream synthesis) it is the emotional ‘brain’ which is responsible for dream plot formation. This evidence on its own disproves the theory. However, there is more. Research accumulated over the last 40 years (and universally accepted by dream researchers) shows that dreams are coherent and that they typically relate to previous waking experiences and memories. There also tends to be continuity in the type of dream content over time and this could not be so if there were a random stimulus. Hobson and McCarley also theorised that REM sleep serves to 'rest' the cells in the brainstem which produce serotonin and noradrenalin, because in REM sleep these particular neurotransmitters are not used by the brain. Their idea was that these neuronal pathways were being rested so that we would wake up the next day, refreshed by REM sleep. Consequently, then, the more REM sleep people had, the more refreshed they should be. But researchers looking at the sleep patterns of depressed patients found that they had massive amounts of REM sleep in proportion to slow-wave sleep and yet, far from waking up refreshed, they were waking up exhausted! How did Hobson account for this? He just said: "It is a paradox". 

Yet another problem with this theory, which Hobson now admits, is that it can't explain why certain dreams have positive emotions and some have negative emotions. But the final nail in the coffin for the Activation Synthesis Theory's is the finding that deep brainstem lesions do not generally stop dreaming, whereas certain lesions in the cortex do, despite the existence of brainstem-initiated REM sleep.

Freudian Theory

Freud’s theory asserts that we forget our dreams due to repression or self-preservation. As, under Freudian theory, dreams are a form of subconscious wish-fulfillment, we forget certain dreams as a result of our minds ‘blocking’ material – wishes, desires and fears – that we are unable or emotionally ill-equipped to handle. The dream content may be too traumatic to remember. There has recently been strong evidence to show that REM involves the ‘expectation dopamine pathway’. Professor Mark Solms, Chair in Neuropsychology at the University of Cape Town in South Africa, has been pre-eminent in synthesising this research, showing that when people go into the REM state, the motivation circuit in the brain - the expectation pathway - is activated. As Freud talked about motivation and emotion, this lends some support to Freudian theories of dream analysis. But what it leaves out of the picture is that, when you activate the expectation pathway, you are activating consciousness, not necessarily some subconscious conflict. So it is also possible to argue that is no real evidence there in support of Freud. Secondly, Freud's theory has real difficulties explaining why people so often have anxiety or aggression dreams. Freud said dreams were for fulfilling wishes - but who would want nightmares? Who would want to get beaten up or sexually assaulted in their dreams? Freud claimed that we dream to protect sleep, to prevent us being awakened by threatening, subconscious wishes. However, the REM state, in which most dreams occur, is a regularly occurring biological programme in humans and other mammals, and not something which arises to protect sleep. To recap, expectation pathways activate conscious, not subconscious, experience. There is no evidence at all that dreams are sexually motivated and Freud can't plausibly explain why we would wish for anxiety dreams. The REM state occurs in all mammals, so it is not just a human activity, protecting sleep, as Freud suggested. According to Griffin and Tyrell (2006) of the Human Givens Institute, a cat is unlikely to be dreaming about its Oedipus complex, so the attempt to revive Freud's theory seems to be based more on wishful thinking than on realistic considerations of its defects. So, according to the research of Griffin and Tyrell, Freud's theory just doesn’t explain in any coherent fashion the fact that dreams involve far more than wishes and that only a minority of them can actually be characterised as wishes; further his claim that all dreams are sexually motivated is no longer given any credence. I would argue that Freud’s theory is based on the notion that dream content undergoes symbolic transformation  and consists of both manifest and latent content – hence the need for ‘dream work’ to unlock the symbolic meaning of the dream material which may not appear comprehensible or relevant to real-life on the surface. Aggression, sex or anxiety-provoking dream symbolism may be interpreted in a way which unveils subconscious desires and drives. Griffin and Tyrell neglect to mention that subconscious wishes and desires may well cause anxiety or involve forms of violence or taboo sexuality – indeed, under Freudian theory, this is all the more likely – the very function of dreaming, according to The Interpretation of Dreams (1900) was to work through unacceptable or harmful thoughts, wishes and desires which the waking mind was emotionally unable to process or indulge. They also fail to note that humans experience far more complex emotions and thought processes than other mammals – something which may be reflected in dream content and function. 

Parasitical Connections Theory
Crick and Mitchison's Parasitical Connections Theory suggests that we dream to forget. Their idea came from studying work done on computer programmes that simulated neural intelligence. An overload of incoming information could trigger ‘parasitical connections’ between unrelated bits of information that interfered with memory, and an unlearning system had to be developed to knock these out of the computer systems. Crick and Mitchison postulated that a complex associational network, such as the cortex, might become overloaded in the same way, and that the PGO spikes were an unlearning mechanism, in the form of random 'bangs' coming up from the brainstem every so often, to knock out these fairly weak parasitical neural links. As, at that time, most dreams were thought to be bizarre in content, this was taken as evidence for the existence of these parasitical connections. Crick and Mitchison theorised that, if we didn't have dreaming, we would go on making more and more bizarre connections, which would imply that, if we block REM sleep, our memories should become more addled. If this theory is correct, then depressed people on antidepressants that block REM sleep (monoamine oxidase inhibitors - MAOIs) should suffer memory impairment they don't. If anything, depressed people on MAOIs report memory improvement rather than increased confusion in memory recall.Yet another problem with the theory is that, over the last decade or so, there have been significant technical advances in the recording of what actually happens during dreaming. The overwhelming majority of dreams are, in fact, quite routine, everyday experiences; it is the tiny percentage of dreams that we recall that seem bizarre: dreams recorded in the sleep laboratory, when sleepers are woken as soon as they go into REM sleep, are mostly not bizarre at all. As a result of this discovery, Crick revised his theory to suggest that it might still, at least, explain those few dreams that do have a bizarre component to them. In other words, his theory has been so drastically modified that very little of it remains at all. Finally, since Crick and Mitchison formulated this theory, not a shred of evidence has arisen to show that the human brain makes parasitical connections - that is something known only to occur with computer networks.

Expectation Fulfillment Theory

Evolutionary theory argues that animal brains – including humans – have evolved in ways which best help predict effective survival instincts and behaviours. The evolution of the cortex in mammals led to an increase in the processing capacity of the brain – instead of acting solely on instinctual impulses, the animal is able to weigh up the benefits and risks of a particular activity. In more technical terms, it enabled the ancient dopamine prediction circuits of the limbic system to be subjected to a higher order risk analysis, based on the additional computing power provided by the evolved cortex. However, that left another problem to be solved - the limbic system communicates with the cortex via behavioural impulses (emotions). If these are not acted upon (for instance, because the strategy is deemed too risky or because the cortex has set other priorities or motives — such as deciding, in certain circumstances, that it is more important to protect young than to chase a possible food source) they don't go away. In the case of humans, this state of unfulfilled expectation can also occur when we think about something in the future or the past that causes emotional arousal in the present, but which can't - by its very nature - be acted upon. These uncompleted emotional impulses or ‘expectations’ stay switched on, taking up processing capacity in the expectation system. So far, two distinct strategies have evolved for dealing with this biological conundrum. Firstly – found in the spiny anteater - is the development of a much bigger cortex to store all these expectations whilst retaining sufficient spare computing power for making new, ongoing risk assessments. This may also be the strategy evolved by dolphins, which have an exceptionally large cortex. The muscle paralysis that accompanies REM sleep places dolphins at risk of drowning, so they can have hardly any REM sleep. The second and much more efficient method is dreaming. Whilst dreaming, we act out the unrealised expectations from waking by pattern matching them to analogous sensory patterns — images and events stored in memory — as it is through pattern matching that the REM system works. An expectation is an imagined scenario, using images from memory. This accords with Freud’s theory of dreams as wish-fulfillment. In dreaming, we are asking memory to provide a scenario that matches a scenario that is already a part of memory — the event that aroused the expectation. So the matching scenario has to be the best fit that memory can provide. Think of it this way — if you left hand and ask your brain for a best-fit pattern match, it can't use your left hand because that is the one you want a match for — so it must use your right hand, as the best-fit pattern match for your left. This does not happen in waking because we pattern match our expectations to whatever stimulates them in the environment, not to a memory. If you want an ice cream, the expectation is fulfilled when you are actually eating it. The dream, then, by fulfilling the expectation, completes the circuit and switches off the arousal. But that is not the end of the matter, for we have now converted an unrealised expectation into a factual memory of completing it. Ordinarily, the hippocampus (the conscious memory store) archives our memories of recent events and quickly deconstructs those memories, sending them to various parts of the cortex — the parts concerned with vision, hearing, touch etc — for storage, in order to facilitate efficient pattern matching. But, if the dream is allowed to be stored as a real memory, it will corrupt the memory store and greatly diminish our ability reliably to predict the outcome of similar experiences in the future. This is avoided by preventing the hippocampus from sending the dream information to the cortex for long-term storage (see Busáki (1995), ‘The hippocampo-neurocortical dialogue’, Cerebral Cortex 6, pp. 81—92). As explained earlier, PET scans and other types of research have shown that, during the dreaming process, the prefrontal cortex is closed down. So it is no accident that the prefrontal cortex is switched off during dreaming; or that the hippocampus doesn't de-construct information and send it all around the brain - because what the hippocampus is doing in dreaming is getting rid of expectations that didn't pan out while we were awake. It is getting them out of the way and making them inaccessible, in effect, so as to allow us to build up a proper, intelligence prediction and expectation system and an accurate storage of knowledge (this also explains the evidence for memory consolidation — if you take away all the false expectations, the memories that are consolidated are more accurate). The Expectation Fulfillment Theory can therefore explain why dreams are about stimulating or provocative events, particularly about emotionally arousing expectations and why dreams may be consistent over time, in addition to the developmental aspects of dreaming. It can explain the other tests conducted and published by Bill Domhoff, but more than that, it explains the cutting-edge evidence that the brain is ever malleable, by explaining how it can be so malleable. It was Abraham Maslow who introduced the idea that, until basic needs are met, people can't engage with questions of meaning and spirituality - what he calls self-actualisation. Another contributor was William Glasser, who put forward the idea that fulfilment of people's needs for control, power, achievement and intimacy depends on their ability to behave responsibly and conscientiously - he argued vehemently that mental illness springs from these needs not being met.

The purpose of the brain is to predict, so that we can get our needs met. We need to have a system that can continually adapt itself, and the Expectation Fulfilment Theory shows how the brain does that by cancelling out the expectations that didn't work. It enables us to have a contemporary register of what really does get needs met in our lives, so that we can more accurately predict what we need to do in the future (but we can only work with the experiences we have had - if, as a child, a woman experienced both abuse and love from her father, she may continually seek a relationship with abusive men, until eventually she can learn that love exists separately from abuse, for example). It may well be questioned then, whether by recalling our dreams, as sometimes happens, we are simply ‘undoing’ the dream work. The answer is no - because the arousal is switched off once the expectation is acted out. When we are awake, the cortex is switched on, enabling us to compare dream content to what is really happening around us and, thus, to distinguish between dream and reality. Occasionally the cortex is alerted by some incongruity in the dream experience - such as flying - and we become aware that we are dreaming (lucid dreaming) But, according to some dream theorists, this risks undoing the dream work of deactivating the experiences — because we now know we are experiencing a fantasy (in more primitive mammals, if an altering of the cortex were to happen, it is less likely their brains could make the distinction between dream and reality; the fantasy would be treated as real and would therefore corrupt the memory stores). 

Scientists have long known that memories are formed in the brain's hippocampus, but are stored elsewhere - most likely in the neocortex, the outer layer of the brain. Transferring memories from one part of the brain to the other requires changing the strength of the connections between neurons and is thought to depend on the precise timing of the firing of brain cells.

"We know that if neuron A in the hippocampus fires consistently right before neuron B in the neocortex, and if there is a connection from A to B, then that connection will be strengthened… And so we wanted to understand the timing relationships between neurons in the hippocampus and the prefrontal cortex, which is the front portion of the neocortex” explains Casimir Wierzynski, a Caltech graduate student in computation and neural systems and co-author of the study “State-dependent spike timing relationships between hippocampal and prefrontal circuits during sleep” (2009) Neuron. The research team - led by Athanassios Siapas, a Bren Scholar in the Caltech Division of Biology and an associate professor of computation and neural systems - used high-tech recording and computational techniques to listen in on the firing of neurons in the brains of rats. These techniques helped them pinpoint a number of neuron pairs that had precisely the kind of synchronous relationship they were looking for - one in which a hippocampal neuron's firing was followed within milliseconds by the firing of a neuron in the prefrontal cortex. "This is exactly the kind of relationship that would be needed for the hippocampus to effect changes in the neocortex - such as the consolidation, or laying down, of memories" states Wierzynski.

Once these spike-timing relationships between the hippocampal and prefrontal cortex neurons had been established, the team used their high-tech eavesdropping techniques to hear what goes on in the brains of sleeping rats - since sleep, as Siapas points out, has long been thought to be the optimal time for the memory consolidation. As it turns out, those thoughts were right - but only part of the time.The team did indeed hear ‘bursts’ of neuronal chatter during sleep - but only during a phase of sleep known as slow-wave sleep (SWS), the deep, dreamless periods of sleep. "It turns out that during slow-wave sleep there are these episodes where a lot of the cells in the hippocampus will all fire very close to the same time" says Wierzynski. In response, some cells in the prefrontal cortex will fire in near unison as well, just milliseconds later. "What's interesting is that the bulk of the precise spike timing happens during these bursts, and not outside of these bursts," he adds.

On the other hand, during rapid-eye-movement (REM) sleep, the previously chatty neuron pairs seemed to talk right past each other, firing at the same rates as before but no longer in concert. “It was surprising" says Wierzynski, "to find that the timing relationship almost completely went away during REM sleep."

Since REM sleep is the phase during which dreaming occurs, the scientists speculate that this absence of memory-consolidating chatter may eventually help to explain why dreams can be so difficult to remember. As intriguing as that idea may be, the researchers caution that these findings only raise possibilities, providing avenues for further research in the field. Siapas asserts "now that we've shown this link, we have a framework we can use to study these questions further. This is just a step toward our goal of some day fully understanding the relationship between memory and sleep."

Writing in Behavioural and Brain Sciences, in a special issue devoted to the most widely promoted dream theories, Professor Domhoff of the University of California, recognised as one of the leading researchers in this field, commented on the evidence presented, concluding "if the methodologically most sound descriptive empirical findings (i.e. the findings that are most solidly established to explain dreaming) were to be used as a starting point for future dream theorising, the picture would look like this: 

1. Dreaming is a cognitive achievement that develops throughout childhood 
2. There is a forebrain network for dream generation that is most often triggered by brainstem activation (the PGO spikes)
3. Much of dream content is coherent, consistent over time and continuous with past or present emotional concerns"

So any theory of dreaming would have to account for those three most solidly established findings (also perhaps including the need to explain memory consolidation as well, if a theory were to explain the full picture).

Domhoff's final conclusion is that "none of the papers reviewed in this commentary puts forward a theory that encompasses all three of these well-grounded conclusions. This suggests the need for a new neurocognitive theory of dreaming”. In other words, according to Domhoff, theories that have dominated the field over the last 30 years do not explain why we dream, and there is need for a completely new one.

Griffin and Tyrell argue that the Expectation Fulfillment Theory is an adequate explanation for this phenomenon of dreaming, as it is rooted firmly in biological science, yet also recognises the rich, subjective experience of dreaming in the individual. It states that all arousals of the autonomic nervous system - the generation of an emotion, however slight - form half of a process. The second half of the process involves the brain ‘fulfilling’ the expectation (an emotion is the same as an expectation) through an action of some kind. If that doesn't happen in reality during the daytime, it happens metaphorically in a dream at night, thus completing the arousal - de-arousal process. Recurrent dream themes tend to reflect unfulfilled expectations and daily concerns, which may not be those acknowledged by the wakeful individual as being the most significant ‘concerns’. The theory also explains Domhoff’s developmental theory – dreams are consistent and coherent (representing metaphorical symbols of expectation fulfillment) and become more complex over time (as we mature we develop more complex modes of introspection and abstract thought). It also apparently explains the relationship of REM function in foetuses with that adulthood – the pattern-matching templates are programmed whilst in the womb and throughout early life – the same pattern-matching process which is used in dreams to deactivate emotional arousal. The theory is also consistent with Domhoff’s notion of a ‘forebrain network for dream generation’ which is triggered by brainstem activation and the idea that dream is consistent and relates to waking experiences and concerns. For example, persons suffering from depression may have too much REM sleep due to an increase in introspection and waking anxieties and concerns. This causes so much arousal which needs to be de-aroused/discharged during sleep that individuals often awake exhausted, rather than refreshed, due to lack of deep, dreamless delta-wave sleep. A second piece of evidence arises from an analysis of Freud and Jung's specimen dreams, which they had offered as the best convincing evidence for their respective theories. The analysis revealed that the dreams were perfect metaphorical manifestations of what was worrying them the day before - according to detailed written data they had themselves provided, and this was structurally extremely tight-fitting data. Freud's dream of Irma's injection (which I have written about elsewhere in this Blog) and the expectations/desires he had the previous day; the sequencing of the dream; the characters involved; and what they actually did in the dream provided an exact mirror image of the biggest event he had on his mind before he went to bed that night – the medical condition of a female patient and fear of accusations of negligence. There is no ambiguity there and it was likewise for Jung's dream, which I shall write about in the near future.

REM sleep has three functions: to switch off expectation and reduce the stress of managing increasing numbers of expectations which may have no relevant to the current situation; it creates space storage capacity in the cortex; and it preserves the integrity of our emotional templates.

The brain has ‘adaptive intelligence’ – starting off with basic instincts which can be modified as a result of experience, enabling the brain to continually refine its learning processes. In the Expectation Fulfillment Theory, patterns are ‘programmed’ into the brain when in REM states – during gestation or early childhood, which humans seek to complete in the environment after birth.  All learning is thought to be a form of pattern-refinement. The first evidence for this came from research findings that the honeybee has a neural transmitter called octopamine, which is similar to dopamine (the human motivation neuro-chemical). One single cell, using this neuro-chemical, motivates the bee to go out every morning to search for nectar (instinctive behaviour) and then that cell keeps a record of where the nectar is found. The next time the bee goes out, it predicts (on the basis of that record) where it will find nectar today. So, if the bee got nectar from a blue flower yesterday, it will pattern-match and go to a blue flower today, predicting and expecting that it will find nectar there again. If it doesn't get nectar from the blue flower today, it immediately revises its memory store. So the memory store will now show that blue is not such a good predictor of nectar after all. Clearly, then, the bee has an instinct plus a capability for learning; it takes an instinctive pattern, builds on current information and modifies it. That is also what was postulated by computer scientists trying to model how the brain works. Their computer program succeeded in modelling complex bee foraging behaviour, and other kinds of more complex behaviours, using this simple idea that you start with an instinctive core that can be modified through feedback from previous efforts. It is a far more efficient system for acquiring knowledge than the one suggested in the module theory - if we had masses of modules in the brain, all occupying their own areas, the brain should be pretty well fixed, and that is exactly what the evolutionary psychologists thought was so. Their idea was that those multiple modules had evolved over the two million years that we were Stone Age hunter-gatherers and that (as all of our knowledge would go back to those times) we are maladapted for the world we live in today — a very fatalistic view of human nature. But the new information allows us to be more optimistic about human capacity. Behaviour has been shown to be so very much more malleable than anyone had ever suspected.

The second source of evidence in support of the theory comes from neurological research – brain scans which show mental activity during the learning process. For example, when people are born blind, the brain cells that would have been used to generate sight learn to read Braille instead and thus, neurons are incredibly adaptive and able to take on new tasks. 

Another example, is research which shows that the hippocampal area in taxi drivers' brains grows new cells and expands when they do 'the knowledge' - the huge number of street maps they must plot out, learn and keep in their memory.

The Expectation Fulfillment Theory also explains the fundamental question of why we forget our dreams. In mammals, the ancient dopamine prediction circuits of the limbic system evolved, becoming subject to a higher order risk/benefit analysis, based on greater computing power of the developed cortex. However, the limbic system communicates with the cortex via behavioural impulses (emotions) and if these are not acted upon – because they are deemed too risky, or because there is another more pressing motive – they do not vanish, at least not in humans. This ‘unfulfilled expectation’ can also occur when we think about something (past of future) which cannot be acted upon in the present. As aforementioned, these unexhausted expectations remain, taking up capacity in the ‘expectation system’. A dream recreates a scenario which matches one which is already stored in the memory – i.e. the event which aroused the expectation – and therefore the dream content (from memory) is the best-fit the brain can provide, and completes the ‘circuit’ by switching off the arousal. We have now converted an unrealised expectation into a factual memory of fulfilling it. Ordinarily, the hippocampus (the conscious memory store) holds our memories of recent events and quickly deconstructs those memories, sending them to various parts of the cortex - the parts concerned with vision, hearing, touch, etc - for storage. It does this to facilitate efficient pattern matching - but, if the dream is allowed to be stored as a real memory, it will corrupt the memory store and greatly diminish our ability reliably to predict the outcome of similar experiences in the future. This is avoided by preventing the hippocampus from sending the dream information to the cortex for long-term storage. PET scans and other types of research have shown that in dreaming, the prefrontal cortex is purposefully switched off. 

Therefore, it is no accident that the hippocampus doesn't deconstruct information and send it all around the brain, because what the hippocampus is doing in dreaming, is getting rid of expectations that didn't pan out while we were awake. It is getting them out of the way, making them inaccessible in effect, so as to allow us to build up a proper, intelligence prediction and expectation system and an accurate storage of knowledge. This also explains the evidence for memory consolidation - if you take away all the false expectations, the memories that are consolidated are more accurate. The Expectation Fulfillment Theory can therefore explain why dreams are about emotionally arousing events, particularly about emotionally arousing expectations; and why dreams are consistent over time. It explains the developmental aspects of dreaming; and the other tests put down by Domhoff. But, more than that, it explains the cutting-edge evidence that the brain is ever malleable, by explaining how it can be so malleable. 

Griffin and Tyrell explain why it is necessary to forget dreams by furthering the model and suggesting the 3-E approach: the Elimination of Emotional Expectation Theory. Elimination of emotional expectation lowers emotional arousal and frees up spare capacity in the cortex. ‘Dream amnesia’ eliminates failed expectations, thus updating the expectation memory stores which functions as an internal survival guide. They propose that this is the function of dreaming in the REM state. With regard to remembered dreams – this does not inhibit the function of dreaming within this model – as arousal is deactivated once the expectation is fulfilled and eliminated by pattern-matching within the dream content. The expectation has been acted out. When we are awake, the cortex is switched back on and we are able to compare dream content to our real-life environment and thus distinguish memories of dreams and reality.