1 Journal of Personality and Social Psychology 1998, Vol. 75, No. 1, Copyright 1998 by the American Psychological Association, Inc /98/S3.00 The Effects of Current-Concern- and Nonconcern-Related Waking Suggestions on Nocturnal Dream Content Charles D. Nikles II, David L. Brecht, Eric Klinger, and Amy L. Bursell University of Minnesota, Morris In previous research, presleep suggestions influenced nocturnal dream content. It was hypothesized that suggesting topics associated with participants' current concerns would influence dream content more than suggesting other topics. Ten students spent 4 nights in a sleep laboratory: an adaptation night, a baseline night, and 2 nights under suggestions to dream about a concern-related or other topic. Concern-related suggestions influenced dream content largely its central imagery more than did other suggestions, which did not differ from nonsuggestion. Number of transformations within dreams was uncorrelated with dream vividness, contrary to extended activation-synthesis theory. Thus, the concern-related status of suggestions moderates their effectiveness and, inconsistent with extended activation-synthesis theory but consistent with current-concerns and distributed-activation theories, motivational and volitional processes actively influence dream content. The study of motivational influences on dream content has traditionally contributed to advances in the understanding of personality (e.g., Freud, 1900/1961), especially when personality is viewed as ' 'the dynamic organization within the individual of those psychophysical systems that determine his unique adjustments to his environment" (Allport, 1937, p. 48). The present investigation provides additional evidence on the influence of motivational factors, among the systems dynamically organized within personality, on the shaping of subjective experience in this instance, on the flow of dream content. Previous research has shown that verbal stimuli introduced during sleep influence dream content insofar as they are related to participants' current concerns (Hoelscher, Klinger, & Barta, 1981). Other research has indicated that suggesting to people who are about to fall asleep that they dream about particular topics sometimes influences subsequent dreams (e.g., Walker & Johnson, 1974). However, the successful suggestions were often Charles D. Nikles II, David L. Brecht, Eric Klinger, and Amy L. Bursell, Division of Social Sciences, University of Minnesota, Morris. Charles D. Nikles II and Amy L. Bursell are now at the Department of Psychology, University of Minnesota, Twin Cities Campus. David L. Brecht is now at Hopkins High School, Hopkins, Minnesota. Charles D. Nikles II and David L. Brecht contributed equally to the design, execution, analysis, and reporting of this study. This research was funded in part by grants from the Undergraduate Research Opportunities Program at the University of Minnesota and the Morris Academic Partnership program at the University of Minnesota, Morris. We gratefully acknowledge the assistance of Timothy J. Ray in essential technical support, Jeffrey Ratliff-Crain in the analysis and reporting of the data, and Nicole A. Herdina and Joline C. Ness in the ratings of dream properties. We also greatly appreciate the assistance of Bryon Peasly and Amanda Schanus with electrode application and dream report transcription, respectively. Correspondence concerning this article should be addressed to Eric Klinger, Division of Social Sciences, University of Minnesota, Morris, Minnesota Electronic mail may be sent to of emotionally arousing topics. The present investigation probes the further question of whether such suggestions are effective primarily when the suggested topic is associated with one of the dreamer's current concerns. This possibility is inferable from current-concerns theory, which originated as a comprehensive psychological theory of conscious mental flow (Klinger, 1971), including the flow of dream content. The construct of current concern refers to a goal-specific latent state of an organism between the time it becomes committed to a goal and the consummation or abandonment of the goal (Klinger, 1971,1975,1977). There is a different current concern corresponding to each of an individual's goals. Although current concerns are associated with conscious thoughts, other cognitive processing, and emotional responses related to the goal pursuit, the construct itself refers to a nonconscious, time-binding brain process that underlies the goal pursuit and its conscious manifestations. In its present form, current-concerns theory predicts that becoming committed to any goal potentiates emotional responses to and cognitive processing of cues associated with that goal pursuit (Klinger, 1971, 1977, 1996). The emotional responding (called protoemotional in its initial phases) begins within at most about 300 ms of the stimulus and possibly within the first 100 ms (Klinger, Goetzman, Hughes, & Seppelt, 1996). Depending on the ability of the stimulus to satisfy certain criteria, cognitive processing is terminated or carried forward, potentially to the point of conscious thought or dreaming. This is a motivational and volitional effect in the sense that pursuing a goal (and hence having a current concern about that goal) disposes toward processing of its cues, which leads to incorporating elements of the goal pursuit into dream content (Klinger, 1971, 1990). Because concern-related cues are processed in preference to others, cues unrelated to current concerns will have, at best, much weaker effects. This differential effect has been demonstrated for stimuli experimentally introduced during sleep (Hoelscher et al., 1981) as well as in a wide range of waking cognitive investigations (Klinger, 1996). It also pre- 242
2 CONCERNS, SUGGESTIONS, AND DREAMS 243 sumably operates under natural conditions, in which the cues would be largely contained in the preceding mentational stream itself. Although the effect may well be weaker for presleep cues, current-concerns theory would still predict that people will more easily carry out an intention to dream about particular topics when these represent current concerns that is, when the volitional force of that intention (a current concern in its own right) coalesces with the already existing disposition to process in dreams the naturally occurring cues of that topic. The idea that motivational variables influence dream content is ancient. The role of wish and wish fulfillment was central to Freud's (1990/1961) theory. In more recent times, the validity of current concerns and related motivational processes as influences on dreams has received various kinds of empirical support and has been acknowledged, though not systematically developed, in major contemporary dream theories, such as in a psychological extension (Seligman & Yellen, 1987) of the neuroscience-focused activation-synthesis theory (Hobson, 1988; Hobson & McCarley, 1977), in the cognitive-modeling approach of distributed-activation theory (Antrobus, 1991), and in a critical synthesis of research on dream content (Domhoff, 1996). However, the nature of the motivational influence varies in these latter theories. In activation-synthesis theory, the sense of plot and meaning in dreams is supplied by the dreamer's cognitive integration (synthesis) of the images produced by the activations. Hobson seems not to have systematically delineated a role for motivation here in fact, he has appeared to deny such a role in his more formal statements about the theory although in his examples of dream interpretation (Hobson, 1988) he implied that being concerned about something increases the likelihood of associated ideas becoming part of the cognitive synthesis by which dreamers make sense of and construct plots around the essentially chaotic, random images triggered by the neural bursts. Thus, Hobson (1988) asserted that his dream about Mozart "meant" "I would love to see Mozart," clearly a motivational influence on content selection, but one that he appears to rule out in emphasizing the independence of pontine bursts from psychological processes (p. 220). This issue is explored further in a later section. Antrobus (1991), on the other hand, proposed a model in which motivational factors interact reciprocally and continuously with imaginal, conceptual, and motor factors in ways that determine dream content: "[a] better model for dreaming may be one which integrates the information both within and between modality-specific, motivational, conceptual, and motor modules in parallel. That is, the information in each module constrains the processes of the others" (p. 108). Furthermore, the linear influences of goal states and motivation can be represented in DREAMIT:S [a model of imagery and thought] by assigning... bias values to particular local units in a network.... [The process will therefore come] to respond differently to the external input as a function of different plan states, (p. 117) Depending on the precise definitions, this feature of Antrobus's model corresponds to a long-held tenet of current-concerns theory (Klinger, 1971, 1975, 1996) and is consistent with its hypotheses regarding effects on dreaming. This investigation, which assessed the effects of suggesting concern-related and concern-unrelated topics to dreamers, thus served a number of ends: (a) it sought to refine information regarding the conditions for successful presleep suggestions of dream topics, (b) it tested a derivation from current-concerns theory, (c) it assessed consistency with two other major contemporary formulations of dreaming, and (d) it provided a test of a further derivation from one of them. Previous Evidence for Effects of Suggestions on Dream Content Research has demonstrated that nocturnal dreams and thoughts can be influenced by presleep suggestions. Stoyva (1965) found that a posthypnotic suggestion, administered just before sleep onset, to dream about a certain topic influenced the content of the REM dreams reported by highly hypnotizable participants. Half of these participants, after a posthypnotic suggestion to dream about a narrative that had been played during the trance, reported relevant dream content (Tart, 1967). In a similar study, approximately two thirds of all dream reports incorporated elements of the suggested narrative (Tart & Dick, 1970). Barber, Walker, and Hahn (1973) showed further that suggestions affected nocturnal thoughts (reported mostly from NREM periods) as well as dreams and that for suggestions to have an effect, participants needed to be neither highly hypnotizable nor hypnotized. The greatest effects on dreams were obtained when authoritatively worded suggestions (""fou will dream about..." ) were used with hypnotized participants and permissively worded suggestions ("Try to dream about...") were used with unhypnotized participants. These previous researchers did not systematically manipulate the content of the suggestions given. They have therefore left an important question unanswered: Which properties of suggestions determine how much they affect dreams? Stoyva (1965) instructed participants to dream about simple action sequences, such as "climbing a tree" and "rowing a boat." Tart (1967) played a recorded narrative that placed the participant in a threatening situation. Tart and Dick (1970) played narratives selected to be "detailed, interesting, pleasant, and emotionally involving'' (p. 305). Barber et al. (1973) instructed participants to dream about the assassination of President Kennedy, which had occurred less than a year before the study was conducted. Thus, all participants in each of these studies received either identical suggestions or suggestions of the same general type. In each instance, the suggested dream topics were emotionally evocative, but whether this or other properties of the suggestions contributed to their effectiveness cannot be determined from these data. An investigation that did manipulate suggested dream topics (Foulkes & Griffin, 1976) found no effects of any kind of suggestion on dream content. The chief manipulation was a distinction between the lists from which the suggested dream topic was randomly selected: a list of dream topics suggested by participants or a list proposed by the investigators. The dream topics listed by participants would most likely have been associated with participants' current concerns and would be predicted to have yielded greater incorporation. However, this investigation used nonlaboratory, morning-after dream recall. The judges'
3 244 NIKLES, BRECHT, KLINGER, AND BURSELL task was to infer from each participant's dreams which of the originally listed dream topics had actually been selected. Given the low level and complex selectivity of longer term dream recall, and given the all-or-none judgment requested of the judges rather than a rating of incorporation, this method may simply have been too blunt an instrument for testing these hypotheses. Indications From Nonsuggestion Presleep Manipulations There is a substantial literature regarding the effects of presleep stimuli, without suggestions to dream about them, on subsequent dream content (Arkin & Antrobus, 1991; Cartwright, 1974; Cohen, 1972,1979; Cohen & Cox, 1975; Kuiken, Rindlisbacher, & Nielsen, 1991; Saredi, Baylor, Meier, & Strauch, 1995). This is not the place to repeat such a review. The results of more recent investigations, which show quite varied effects of presleep nonsuggestion stimuli, are consistent with results in earlier literature reviewed by Arkin and Antrobus (1991): There was very little direct incorporation of such stimulation into dreams, and there was an array of indirect effects that appeared to vary from one experiment to the next. This appeared to be true whether participants were passive recipients of stimulation or active participants in creating stimulation. The possible exception was presumably concern-related activity as in Cartwright's (1974) study, in which there was some evidence of significant direct incorporation. There has been too little exact replication of methods to establish which of the indirect effects might be stable and predictable. The relevant conclusion for the hypotheses of this investigation, which focused on direct incorporation, is that there is very little such incorporation of presleep stimuli that are unrelated to participants' current concerns, and the evidence for incorporation of unsuggested presleep stimuli that are concern-related (in comparison with concerns not cued before sleep) is at best uneven. (This is a sharply different conclusion from that relating to personally relevant stimulation during sleep; Hoelscher et al., 1981). Evidence for Differential Effects of Current- Concern-Related Stimuli on Dream Content and for Emotional Mediation Investigations pursued within the framework of current-concerns theory (Hoelscher et al., 1981; Klinger, 1975, 1977, 1987, 1996) have indicated that suggestions of dream topics that are emotionally evocative and thematically relevant to participants' goal pursuits are likely to influence dreams more than would emotionally and motivationally neutral topics. This is because current concerns potentiate cognitive responses to concern-related cues and because this effect is probably at least partly mediated by emotional processes. Concern Effects on Cognition Research has shown that current concerns influence cognitive processes in both waking and sleeping participants. For instance, people are especially responsive to environmental cues that pertain to their current concerns (Klinger, 1978), and they continue to screen for concern-related stimuli while asleep (Hoelscher et al., 1981). Exposure to current-concern-related stimuli influences thought and dream content much more strongly than exposure to concern-unrelated stimuli, regardless of whether participants are awake or asleep in REM states. Parallel effects occur on attention and recall (Klinger, 1978). The fact that the effect occurs during sleep indicates that it is involuntary and automatic, a conclusion confirmed by evidence on interference effects of concern-related distractors in lexical decision (Young, 1987) and emotional Stroop tasks (Cox & Blount, 1995; Johnsen, Laberg, Cox, Vaksdal, & Hugdahl, 1994; Riemann, Amir, & Louro, 1995; Riemann & McNally, 1995). Emotional Mediation Indirect evidence has suggested that these effects are mediated by emotional processes (Klinger, 1996). That is, the effect of committing oneself to a goal is to potentiate protoemotional responses to cues associated with the goal pursuit, and these responses in turn instigate cognitive processing of the cues. To at least some extent, therefore, protoemotional responses control subsequent cognitive processing. This sequencing is possible because the emotional features of at least verbal stimuli are processed soon after stimulus presentation, probably before full semantic processing has been completed (Bock & Klinger, 1995; Klinger, 1996). The likely role of emotional influences specifically on dreams was suggested in an earlier investigation (Piccione, Jacobs, Kramer, & Roth, 1977). Participants' daily events that were most often incorporated into their sleep-laboratory dreams were associated with more intense emotions than were other daily events. This suggests that the mediational evidence found in waking cognition extends to dreams as well. Implications for Suggested Dreams The experimental part of the above evidence examined effects of concern-related immediate stimuli on ongoing cognitive and dream activity. However, one might expect that these effects would generalize to presleep suggestions of dream topics. That is, because current-concern-related topics are relevant to people's goals and have high emotional value, they may be more effective as suggestion topics than concern-unrelated topics. There are at least two possible mechanisms for producing concern-related topics' greater effectiveness. First, suggesting concern-related material as dream topics associates the suggestion with existing motivational structures, the current concerns, thereby increasing the likelihood of retaining the suggestion. Second, it primes the already existing tendency toward emotional and cognitive response to stimuli related to the suggested concern, including those stimuli existing within the dreamer's own stream of sleep mentation. These consequences of suggesting concern-related dream topics, then, provide predictions within the current-concerns framework that such suggestions will have greater impact than suggestions to dream about concern-unrelated topics. Additionally, current-concerns theory predicts that people will dream spontaneously in the absence of any presleep suggestion primarily about current-concern-related topics rather than concern-unrelated topics. In the absence of concern-related
4 CONCERNS, SUGGESTIONS, AND DREAMS 245 stimuli during sleep, the probability of dreaming about any one concern at a given moment will be low (Hoelscher et al., 1981), but most dream images should contain material associated with one or another of the dreamer's many current concerns. The focus of dreaming on concern-related topics occurs, according to current-concerns theory and presumably Antrobus's (1991) distributed-activation theory, because current concerns continuously potentiate responses to goal-related stimuli, including to the goal-related internal cues contained in one's ongoing mentation. In this view, then, dreaming consists of concern-related imaginal responses to goal-associated cues. This view is supported by results (Saredi et al., 1995) indicating that 98% of 8 participants' dream reports contained material identifiable with at least one of their individual current concerns, as assessed with the Motivational Structure Questionnaire (MSQ; Cox & Klinger, 1988; Klinger, 1987). Two investigations, using rather different methods, failed to confirm this finding (Rados & Cartwright, 1982; Roussy et al., 1995). Rados and Cartwright were unable to find a significant association between participants' REM dream content and their postsleep reports of current concerns, although they found a significant association of dream content with that of presleep thought samples. Roussy et al. found dreams were not associated at a significant level with either presleep thought samples or presleep descriptions of significant concerns, but the trends were well in the predicted direction and their design contained a number of features that would be expected to reduce the power of the tests. A weak association between immediate presleep thought content and dream content is, furthermore, not surprising. Typical college-student participants list approximately 30 current concerns on present MSQ answer sheets and have averaged approximately 50 concerns on more probing previous versions. One feature of waking thought is that it cycles through an individual's numerous concerns, and there is some evidence that having dealt with a topic at one time inhibits for a while its further appearance in conscious thought (Klinger, 1973), except perhaps in the case of emotionally charged, ruminative instances. Similarly, it is not surprising that Rados and Cartwright (1982) found little relation between their dream and postsleep concern reports. Their participants produced ' 'three to five REM reports... and one report from Condition 3" (p. 434) in which participants, having just awakened, were "asked, while they were still in bed, to report what was going on in their lives'' (p. 434). It is unclear how many different concerns might have been reported in one report by a drowsy participant but, following the logic described earlier, the probability of a match between one or a few concerns and a small number of REM reports is low. Implications for Activation-Synthesis Theory Differential Effects of Presleep Stimuli The predictions derived from current-concerns theory are substantially at variance with those from activation-synthesis theory. The principal features of Hobson and McCarley's (1977; see also Hobson, 1988) activation-synthesis theory are that (a) dream images are directly generated by essentially random sporadic pontine discharges (the activation) that produce sudden sensations, the dream hallucinations, and (b) forebrain activity seeks to fit these images into as coherent a pattern or plot as possible (the synthesis). Activation-synthesis theory can presently accommodate findings in which previously primed topics or current-concern-related topics show up in dreams more often than they would by chance. On one hand, the theory could regard the process by which this happens as purely associative and passive, which would remain consistent with its view that the pontine bursts are entirely independent of prior psychological processes or content. On the other hand, it could regard the process as part of the active cognitive effort to make sense of the burst-induced visual imagery, an effort that depends on the individual's store of memories and long-term potentiations. However, the differential occurrence of suggested concern-related topics, in contrast to suggested concern-unrelated topics and unsuggested concern-related topics, would require that a specific volitional process influence dream content. We see nothing in activation-synthesis theory as presently formalized that would compel such a prediction. That a prior volitional process could codetermine the selection of dream content would in itself produce a problem for the theory. This problem arises from the insistence of the theory that the pontine bursts that produce initial dream images are independent of any preceding psychological process. One way out of this problem within activation-synthesis theory would be to argue that the influence was not on content selection but on content interpretation. Content interpretation (the synthesis) appears for Hobson and McCarly (1977; see also Hobson, 1988) to be constrained primarily by the burst-induced dream images and their immediate associations, but Seligman and Yellen (1987) admitted other factors, including current concerns. Nevertheless, the impact of a remote suggestion would seem an unlikely candidate for a dream influence within this theory, especially if one assumes that "limits on memory capacity both storage and retrieval may be severe during the dream state" (Seligman & Yellen, 1987, p. 16). These deductions from activation-synthesis theory are based on its present explicit statement. We have been unable to find a specification in statements of the original theory as to how the particular configurations of pontine bursts or the visual images they evoke come to be selected. They are described without further specification as arising from "a largely random or reflex process... with little or no primary ideational, volitional, or emotional content" (Hobson & McCarley, 1977, p. 1347). In the extended version (Seligman & Yellen, 1987), specifications are limited to otherwise unspecified "day residues," "remote residual," and concurrent sensory events. Nor have we found a specification in the original theory of the precise processes whereby (or of the constraints on) the cognitive synthesis of burst-induced visual images takes place. The implications have been that the images generated by pontine bursts are quasi-random and that the synthesis is constrained by the content of the burst-induced images and by associations to those images. Thus, "best fits to the relative inchoate and incomplete data provided by the primary stimuli are called up from memory... likened to a computer searching its addresses for key words" (Hobson & McCarley, 1977, p. 1347). The extended theory (Seligman & Yellen, 1987), which posits not only pontine-originated visual bursts but also emotional bursts,
5 246 NIKLES, BRECHT, KLINGER, AND BURSELL suggests that dreamers' syntheses are influenced by current concerns. This provides a degree of common ground with currentconcerns theory and distributed-activation theory. However, unlike current-concerns theory (e.g., Klinger, 1977, 1978, 1996), there is no specification of the mechanism for this influence of current cojicerns on syntheses or of the reason that a particular concern should influence a particular synthesis at a particular time. Thus, if these formulations original and extended of activation-synthesis theory are to accommodate selective effects of presleep suggestions, they will at least need to be modified significantly and specified more precisely. From its beginnings, the current-concerns view of dreaming (Klinger, 1971) has taken the position that the influences of current concerns on cognition primarily in potentiating responses to cues associated with the respective goal pursuits remain essentially unchanged around the clock but that the cognitive consequences are modulated by the physiological context, whether that be chemical alterations of consciousness, fatigue, psychosis, or sleep. Some of these biological factors determine the extent to which the phenomenal stream is dreamlike. Hence, they determine the onset of dreaming, but not the onset of particular dream content within a dream state. The onset of dreaming is, in this view, simply a changeover from one mode of mental flow to another, without much impact on what themes the mental content is about. The first statement of the theory (Klinger, 1971) gave extensive consideration to the response organization of daydream and dream imagery, integrating it with emerging findings in psycholinguistics, motor programs, and response integration. It did not then, however, dwell on the neuropsychological or neurochemical substrate and has not done so since. It is precisely in this neuroscientific domain, neglected by current-concerns theory, that the original activation-synthesis model and its more recently evolved extensions (e.g., the activation level, input source, and mode of processing [AIM] model, Hobson, 1992) have distinguished themselves. Accordingly, the two positions differ primarily in their views of what modulates what: whether a continuous cognitive flow that is primarily steered by emotional responses to concern-related stimuli, including internal ones, is modulated by neurophysiological processes (the current-concerns approach) or whether dream experience that is generated by neurophysiological events is modulated by cognitive history and possibly current concerns (the activation-synthesis approach). This is largely, though not entirely, a difference in emphasis arising from the respective theorists' different disciplines, interests, Zeitgeists, and views of what made their positions distinctive. The two positions can thus readily be regarded as complementary in many respects, describing factors that modulate each other reciprocally. In one important respect, however, there appears to be a difference of more than emphasis. The current-concerns approach views not just the forms of dream imagery but also changes in dream content as governed to a substantial extent by emotional and cognitive responses to concern-related stimuli, including the internal stimuli of the dream flow itself. This source of change in dream imagery is at the least not developed in detail in the activation-synthesis models, despite receiving periodic informal mention (e.g., Hobson, 1988, pp ; Seligman & Yellen, 1987, pp ). To incorporate it formally would require a significant modification of the theory. Thus, in the present investigation, current-concerns theory would predict that presleep suggestions to dream about particular topics will be more effective if the topics are related to participants' current concerns and will lead to more incorporation of the suggested concern-related topic than if it was not suggested. Activation-synthesis theory would predict no particular differential effect. Visual Vividness and Dream Discontinuities Another hypothesis derivable from the extended activationsynthesis position (Seligman & Yellen, 1987) is that dreams that manifest the greatest number of transformations changes of course, sudden changes in scenery and personnel, and so on should also on average be among the most vivid. This deduction rests on Deduction 1 from the four postulates of the theory: It follows from Postulates 2 and 4 that there should be two distinguishable systems of information in a dream. The first, generated by the bursts, should be vivid, detailed and discontinuous with the prior plot (i.e., surprising). The second, generated by the cognitive synthesis, should be less vivid, less detailed, but continuous with the prior plot. (Seligman & Yellen, 1987, p. 5) Presumably, mentation experienced as a new dream would originate with such bursts; in that light, all dreams might be expected to begin vividly. However, bizarre discontinuities within a dream sample (i.e., those that do not evolve organically out of the preceding dream events) should theoretically also generally reflect bursts of these kinds. It follows that dream segments that contain substantial dream transformations, in the sense of discontinuities, should also on average be more vivid. Although Seligman and Yellen (1987) reported data (based on morning-after dream reports) that confirm this hypothesis, there is also some inferentially nonconfirmatory evidence. If one may take self-rated bizarreness as a proxy for discontinuity (discontinuity is usually considered a part of bizarreness; e.g., Reinsel, Antrobus, & Wollman, 1992), then it is relevant that Rechtschaffen and Buchignani (1992) found no visual differences for dreams judged high versus low in bizarreness. Currentconcerns theory supports no prediction on this issue, but the data reported here permit a further test of the hypothesis. Hypotheses The specific hypotheses of this investigation were as follows: (a) dreams will reflect participants' own current concerns more than other participants' concerns (own-concern-unrelated material) on nights with no presleep suggestions, (b) dreams will reflect own current concerns more when the dream topics suggested before sleep are associated with those concerns than when no topics are suggested, (c) dreams will reflect concernunrelated topics more when these are suggested than when no topic is suggested (insofar as the suggestion instates a goal to dream about the topic and hence renders it concern related), (d) own-current-concern-related suggestions will influence dream content more than concern-unrelated suggestions, and (e) fol-