Since I wrote about the Dream Maker, various new ideas and theories about dreams have appeared.
July 2, 1990: In , I explained my theory of a "Dream Creator" and a "Dream Reactor", and stated that the Dream Creator has motivation to create dreams which have the effect that they have on the Dream Reactor.
December 21, 1996: I wrote a more detailed exposition of the theory in a web page, as shown at .
Antti Revonsuo has developed the theory of Threat Simulation, as published in ,  (with co-author Katja Valli) and . Revonsuo's basic observation is that threatening situations are represented in dreams more often than they occur in real life, yet at the same time they occur more in the dreams of those who actually experience more threatening situations. This is consistent with the hypothesis that dreams are a form of practice at dealing with these threatening situations.
In , Revonsuo describes dreams as likely being "the consequence of an active and organized process", and states that the neural mechanisms of dreaming "function in a selective, orderly and organized manner rather than randomly". He also mentions the non-lucidity of dreams (but not using the term "non-lucidity"), as caused by "perceptual realism" and "delusional lack of insight", and the consequence that dreams create realistic simulations which are to be taken seriously by the dreamer.
This "active and organized process" seems similar in some ways to my concept of the Dream Maker. However, the Dream Maker is a specific agent motivated to purposefully construct dreams which satisfy some measurable criterion of the extent to which they satisfy the purpose for which they are being constructed. Whereas Revonsuo states that dream production is "an automatic, hard-wired, regularly activated feature of human brain function". He recognises that the perceptual realism of dreams contributes to their effectiveness as simulations, but does not suggest that the dream-creating agent may manipulate the content of dreams in specific ways in order to maintain (or restore) such perceptual realism.
One of the major discoveries about dreaming is the discovery of the REM (Rapid Eye Movement) state that occurs during sleep, and its correlation with the occurrence of dreaming. This discovery has played a role in the development of the activation-synthesis dream theory of Hobson, which supposes that the neural mechanisms underlying REM and dreams are closely intertwined (and which seems to have been accepted as the standard "scientific" theory of dreaming, replacing Freud's somewhat less scientific theories which were the previous orthodoxy). But in , Mark Solms shows that REM and dreaming (and in particular, "typical" REM-style dreaming) can be doubly dissociated. Solms shows that REM is neither necessary nor sufficient for dreaming, and although REM and dreaming are normally correlated, the neural mechanisms underlying these two phenomena are mostly distinct.
To determine the independence of REM and dreaming, Solms considers the effects of various types of brain lesion on each. One relevant area in the brain is the ventromesial quadrant of the frontal lobes. Lesions to this area often cause a complete loss of dreaming activity. They also cause adynamia, which is a loss of initiative and motivation. Solms concludes from this that dreams have something to do with motivation. He goes so far as to conclude that "these motivational mechanisms are essential for the generation of dreams", but seems not to take this line of reasoning a bit further and consider that therefore someone or something must be motivated to do the generation.
In my original Usenet posting I stated:
I propose a split personality theory of dreaming. One personality is the Dream Reactor, who experiences and reacts to the dream, and is identified with the waking personality in its motivations. The second is the Dream Creator, which has a different set of motivations. Sometimes its motive is to frighten the Reactor, but not always, because not all dreams are frightening. Some more general motive can be supposed.
The interpretation of the effect of adynamia on the "Dream Creator" (or as I now call it, the "Dream Maker"), is that if a brain lesion causes a loss of motivation, then the Dream Creator will lose its motivation to do what it does, which is to create dreams.
Another recent paper is , where Michael Franklin and Michael Zyphur consider the theories of Solms and Revonsuo (and others). They essentially accept Revonsuo's claim that dreams are a form of rehearsal, and they attempt to extend the theory to account for non-threatening dream content as well. They point out the importance of non-lucidity (which they call "situated cognition") if dreams are to be a useful simulation of the brain's response to the situations that occur in those dreams.
 by Alok Jha of the Guardian is an article that gives a good overview of recent experimental and theoretical work in the dream field.
Given that the brain is basically an information processing system, and given that dreaming is something that the brain does, we are forced to conclude that dreaming must be a form of information processing. The only questions to be answered are: what is being processed, and what is the result?
The Hobson-Allan activation-synthesis theory more or less denies that dreams are meaningful information, and treats them as meaningless consequences of some neural activity, where such neural activity itself may or may not have a specific information processing function.
For those inclined not to believe that dreams are the result of random meaningless neural activity, there are two major choices as to the role of dreams in an information processing system:
Freudian theories and other popular traditional theories of dreaming have tended to regard dreams as the significant output of information processing. Freud regarded dreams as the result of symbolic transformations of hidden subconscious desires. Other popular theories regard dreams as portents of the future, or a means of providing or suggesting solutions to current problems of the dreamer.
Rehearsal-based theories take the opposite view, and regard dreams as primarily the input of a major information processing step. Dreams are constructed scenarios, the purpose of which is to generate the reactions that the brain would have to those scenarios. To facilitate this reaction, the brain's general powers of awareness are reduced so as to eliminate the possibility of lucidity, and the outputs of motor control regions in the brain are suppressed so that reactions can occur to the hypothetical dream scenarios without the dreamer injuring themselves or exposing themselves to risk.
But even if we accept that dreams are the input to an information processing step, we are still left wondering what the output is. The direct output – movement made in response to the dream scenario – is completely suppressed. Any conscious memory of the dream is also suppressed, in that usually we forget the content of most dreams, unless we are unnaturally awoken from our sleep just when the dream is occurring.
Given that external outputs are suppressed, any information processing that occurs during dreaming must result in some change to brain state, if it is to be useful at all. If this change is some kind of "learning", it must be a type of learning which is completely internal, where one part of the brain "teaches" another part of the brain what it knows.
This suggests a possible means of "motivating" the Dream Maker: if there is some means by which it can be determined that one part of the brain, part A, has taught some useful information to another part of the brain, part B, causing part B to change state by a measurable amount, then the degree of this change in state can be considered a measure of the "success" of the dream. This implies that the Dream Maker is motivated to generate dream content which generates the greatest possible amount of "teaching" (as measured by the amount of change that occurs).
This explanation is given at a fairly abstract level, and I have not made any assumption about what is being "taught". The evidence for Revonsuo's threat simulation theory suggests very strongly that whatever it is that is being taught, it is something for which major teaching opportunities are to be discovered by considering various threatening situations. At the same time, the consistent occurrence of various types of dream which are not specifically threatening suggests that the teachable information aspect is not restricted to the handling of threat situations.
If dreams are a form of "rehearsal", how analogous is this rehearsal to the type of rehearsal that occurs when we think about hypothetical situations when we are awake? The main differences between sleep dreaming and awake "dreaming" are the following:
Revonsuo deals with the difficulty that most dreams are forgotten by assuming that the benefits of rehearsal in dreams correspond to those aspects of skill development that do not necessarily depend on conscious memory. In , Franklin and Zyphur review a number of ideas on this subject, and seem to favour the conclusion that rehearsals in dreams are very similar to waking rehearsals in the benefits they provide, except that dream rehearsals are optimised for rehearsing realistic responses to threatening situations. They follow up their discussion of threat rehearsal with discussion of other types of rehearsal, but do not really seem to ask the question as to whether the same optimisation (with its attendant tradeoffs) is so relevant to rehearsal of non-threatening situations, and whether dream-based rehearsal has any real advantage over waking mental rehearsal (or just plain practical rehearsal) of those situations.
An alternative is to assume that dream rehearsal causes the occurrence of some type of information processing that quite specifically cannot occur in the waking mode. One possibility arises from my non-routineness theory of consciousness. This theory suggests that consciousness consists of a multi-step system for processing "non-routine" situations, and, in particular, whatever it is about the current situation that is most non-routine. The main steps consist of:
Dreams certainly contain a "stream of consciousness", and at any particular point in a dream we can readily identify which aspect of the situation in the dream most occupies our conscious attention. It is therefore plausible that the purpose of dreaming may have some relation to the operation of consciousness.
But we have already determined that it is not the purpose of dreams to actually calculate the solutions to the problems that consciousness considers. Which leaves us to consider that dreaming may play a role in tuning or adjusting the process of consciousness.
A critical feature of the hypothesised conscious information processing system is the detection of and response to "non-routineness". A current situation which is less routine is by definition not very much like other particular situations that have occurred previously, and therefore requires a more sophisticated combination of the person's experience of past situations, in as much as those situations bear any relation at all to the current situation.
Threatening situations tend to be non-routine, because if they aren't, then you are living in such dire circumstances that your chances of surviving them are rather low anyway. In the modern civilised world, in times of peace, threatening situations might be so infrequent that we hardly need to worry about them, and we would be better investing our mental efforts in preparing to make the most of the positive opportunities that the world presents us with. But for many people, in many historical (and prehistorical) circumstances, it has been the case that threatening situations have been sufficiently uncommon to be considered "non-routine", but sufficiently common that it is worthwhile making an effort to prepare for such situations when they occur.
How would dreams facilitate the processing of non-routineness? If the effect of non-routineness is to initiate indiscriminate broadcast of modulatory neurotransmitters, then the measured level of non-routineness can be considered as a simple quanitity. The determination of "correct" non-routineness becomes the determination of the correct level of non-routineness for any given situation. This suggests the following model of dreaming:
The only long-term result of this information processing is the non-routineness contingent on the occurrence of particular types of situation. The details of the specific situations, and the dreamer's hypothetical responses to them, are all irrelevant, which is why the machinery of dreaming causes them to be forgotten once the dreaming is finished.
The science of dreaming has gone from treating dreams as information processing outputs, to treating them as not being proper information at all, to finally treating them as meaningful information again, but this time as inputs.
Researchers and theorists have been led to consider what the significance of these dreams as inputs is, and what is the function of the brain's internal processing of the hypothetical situations that occur in dreams, and how this fits into an evolutionary theory of human consciousness. The author of this paper wishes to add his own idea of a motivated and purposeful "Dream Maker" to the theoretical mix, and hopes that other dream scientists can consider what this idea implies about the possible biological and information processing functions of our dreams.