Functions of Dreaming: The Emperor'S New Clothes?

FUNCTIONS OF DREAMING: THE EMPEROR'S NEW CLOTHES? PAUL WILLNER* It has been known for many years that periods of dreaming occur in a regular cyclical pattern, throughout the night, in association with rapid eye movements (REM) and a pattern of low-voltage fast activity (desynchronization ) of the electroencephalogram (EEG) similar to that seen during alert wakefulness [I]. It has been assumed without question that a phenomenon so pervasive, regular, and dramatic must have developed during the course of evolution to serve important psychological or physiological functions, and considerable effort has been devoted to investigating what these functions may be. However, if we do stop to question this assumption, the supporting evidence is seen to be at best weak. The evidence most frequently adduced to argue for the adaptive significance of REM sleep is its obligatory character: during deprivation of REM sleep, attempted REM episodes occur more and more frequently ("REM pressure"), and following deprivation, the duration of REM sleep is somewhat higher than normal ("REM rebound") [2-4]. However , while it is true that pressure and rebound are typically observed if a vital function, such as eating, is prevented, these deprivation phenomena do not in themselves prove the presence of a physiological need. Exactly comparable "motivational" characteristics are apparent, for example, in the drug-seeking behaviour of dependent animals, whose response to drug deprivation exactly mirrors the behaviours seen following food deprivation [5]. Nevertheless, we do not interpret heroin addiction as an evolutionary imperative, and the case for REM sleep, based on its motivational characteristics alone, is no stronger. This paper advances a novel argument, that dreaming sleep may be, in effect, an accident of evolution: it is proposed that the cyclical oscillation between REM and non-REM sleep may reflect the operation of *Psychology Department, City of London Polytechnic, Old Castle Street, London El 7NT, England.© 1991 by The University of Chicago. All rights reserved. 0031-5982/91/3404-0737101.00 588 Paul Willner ¦ Functions ofDreaming systems developed to subserve important waking functions. It is emphasized at the outset that this thesis is speculative in the extreme; it may, however, be useful in formulating a new direction for research in this area and is presented in that spirit. Cognition and Emotion The search for a function for REM sleep has focused increasingly on a possible role in cognitive function. Over the past 25 years, a series of theories have sought to implicate REM sleep in memory processes, often using computer analogies. These positions have developed from a broad statement that dreams may serve to remove irrelevant or unwanted material [6] to a detailed proposal that REM sleep serves to prevent overloading in a parallel distributed memory system by a process of "reverse learning" [7, 8]. A notable feature of the recent theories is the disproportion between the complexity of the theoretical superstructure and the poor quality of the empirical support. Animal studies have consistently reported an increase in paradoxical (REM) sleep following a learning experience [9, 10], and this observation has been extended to humans [H]. However, negative reports are not uncommon, particularly in the human literature [9]. And in animals, the effects vary with the strain of animal tested, the type of task, and the amount of training, leaving it unclear to which aspect of training the REM sleep changes relate. A particular problem for the hypothesis that REM increases are related to memory is that the post-training increase in REM is seen in animals that learn rapidly but not in animals that learn more slowly but do eventually reach the same level of performance [12]; this suggests that motivational or motor factors may be more relevant. A similar problem attends the human studies. For example , the experiment of Mandai et al. [11], in which subjects learned Morse code immediately before bed, included no control for the potentially stressful or activating effects of the learning experience. A similar question may be raised in relation to the observation of REM sleep increases following a period of intense study by college students [13]. More conclusive data should, in principle, come from studies assessing the effect of REM sleep deprivation on memory function. Again, however , the results are far...