Stretched Dream Science: The Essential Contribution of Long-Term Naturalistic Studies

by

Dennis Schmidt

Naturalistic observation has a lesser status than experiment in most sciences. In the field of dream study, practitioners of naturalistic and experimental disciplines coexist, with limited mutual respect. Long-term naturalistic observation, though, has unique sensitivities that make it the most effective or the only possible method for many important studies. While there are challenges to integrating naturalistic and experimental disciplines, there are possibilities, and there is a scientific imperative.

KEY WORDS: naturalistic dream study; longitudinal dream study; philosophy of science
 

In July 1994, Comet Shoemaker-Levy 9, the "String of Pearls" comet, crashed into Jupiter.

Almost every large telescope on and off Earth, and a great many small ones, were trained on Jupiter for a few days. Although the impacts occurred on the far side of Jupiter, not directly visible, analysis of what could be observed soon required fundamental revision of astrophysicists' concepts of comets, and taught much about Jupiter (Yeomans, 1995). The cometfall was a scientific windfall.

Wait a minute. You call that science?

Well, everyone does. The scientific community, the funding agencies, and the general public all agree: Astrophysics is science.

This seems anomalous, in light of one common idea about what makes a science: There is no controlled experimentation in astrophysics. Astrophysicists carefully plan how to observe, and they draw on experimental sciences in both planning and interpreting their observations, but no direct manipulation of the subjects of study is possible.

My day job is in the field of astrophysics. I often think of dream study as akin to astronomy and astrophysics. Dream study too is primarily observational, based on solitary nighttime looking into a world that seems alien to our daytime world.

In astronomy and astrophysics, there has been unending development and refinement of the primary instruments of observation and recording, the telescope and the camera. From Lippershey and Galileo's simple tubes and crude lenses (King, 1979) have come the Hubble Space Telescope; a fleet of other space-based telescopes, spanning the electromagnetic spectrum from infrared through X and gamma rays; the Very Long Baseline Array that has made a radio telescope of a whole side of Earth; adaptive optic telescopes that enable looking through Earth's atmosphere as if it's not there; and so on (Kuhn, 1994, chapter 5). For recording images, photography long ago supplanted Galileo's hand drawings, and modern telescopes have devices that record not only spatial appearances but the arrivals and energies of individual photons, showing the chemical compositions and physical processes of extremely remote objects (see, e.g., papers on solid state detectors, charge-coupled devices (CCDs), and photon counters, in Wall and Boksenberg, 1990). The work of developing the instruments is a large part of the effort in the fields of astronomy and astrophysics.

Astronomers and astrophysicists also recognize the value of using their most sensitive instruments to do surveys, exploratory studies, descriptive studies, etc. (see, e.g., several papers in Wamsteker et al., 1994). And they routinely archive, keep accessible, and extensively use the records of their programs of naturalistic observation (Heck and Murtagh, 1992, 1993).

In dream study, though, development of the arts, methods, and skills of making and recording the primary observations, and of using those records in extended study, is largely ignored by the part of the field that regards its studies as scientific. The work of individuals who cultivate these instruments and use them in extended naturalistic observational studies, while respected to a degree by other workers, has in general a secondary status even within the Association for the Study of Dreams, and is all the more marginalized among researchers elsewhere.

(By naturalistic observation I mean careful observation of some aspect of naturally occurring activity, causing as little disturbance as possible to the activity or its customary setting. This definition is less restrictive than that by which "...naturalism seeks rich descriptions of [phenomena] as they exist and unfold in their native habitats" (Gubrium and Holstein, 1997, 6), although extended naturalistic dream studies do involve seeking rich descriptions of the phenomena.)

Many workers in the dream field look upon journal-based and related experiential studies as of only personal import, with at most a potential for suggesting scientific development that might then establish what individuals glimpse in their personal study. Alan Moffitt, who supported all kinds of dream work, expressed this view (1990):

But dream study is not analogous, in an important way. In astronomy, the professional can see anything that the amateur can see, and see it better, once its location is known. But in dream study, it is amateurs and other experiential workers who use the instruments -- years-long viewing and intimate association -- that are most sensitive to certain classes of phenomena. Some phenomena well-known to workers who study their dreams through long records of daily or nearly daily recall are unknown in the laboratory.

Let us consider this scenario:

A person visiting an unfamiliar place, a college, recognizes both the setting and an incident there as an almost photographic replay from a dream within the past few months. Searching the record locates the dream, allowing comparison of the new experience with the description of the dream.

A couple of years later, evaluating the first year in an academic program, the person thinks of the same dream again, realizing its key incident as a metaphor for the experience of that first year: In the dream, being beckoned off track was literal, while in the academic program that description would be metaphoric.

Through later years, that metaphor seems to have extended application in a sequence of episodes that seem to comprise a story of the person's development in dealing with the dynamics of insecurity and challenges to personal integrity in the professional quest.

Many years after the dream, another story, similarly discovered through surprising clear correspondences between waking events and earlier dreams, is found to have an important element that was also a striking element of the "college" dream. That element now serves as a link bringing together the person's concerns in the first story -- diversions from the professional quest -- and the dominant motif of the other story, the dynamics of passion and commitment in love and work.

This person may have been more or less aware of the intrapsychic forces at play in either of these stories that seem to be running as undercurrents in the person's life. But each story itself, the specific drama spanning years, came to awareness only through noticing dramatic elements of a few dreams and dramatic waking correspondences with them, sparsely distributed over several years. The person feels that the stories are useful in contending with the dynamics.

Let us consider further that these stories also involve other people as principals, some of whose waking and dreaming experiences include elements suggesting intersecting compatible stories. That is, these other people have roles in some of the events whose associations suggest hidden stories to the original person, and those events also suggest complementary hidden stories, in their lives, to those people.

It seems to the person studying these correspondences that they reveal stories that are not arranged by the individual's everyday conscious mind, nor even just the individual's everyday unconscious mind. The stories have the appearance of being arranged across what seem to be temporal and interpersonal barriers.

While the example I have outlined is hypothetical, it is realistic: People who are doing long-term naturalistic study of their dreaming are recording cases like this one. My purpose here, though, is not to present cases for evaluation. It would not be possible to give an adequate presentation of even one case in the space of this paper. My purpose here is to argue that even the possibility of a case like the one I have outlined establishes the necessity of recognizing, as an essential component of dream study methodology, the observational disciplines that alone can reveal such cases.

To be likely to detect a correspondence such as that between the incident in the dream and the incident at the college, a study would first have to be open to noticing and recording such a correspondence. The study would also have to select dreams, waking events, and associations in such a way as to be likely to record:

• both events involved in some such correspondence;
• any such association that the dreamer may have; and
• any such association that the researcher may have in subsequent review of the record.

To be likely to detect the metaphoric fit of a dream to a category of important past, present, or future life experiences entails additional requirements. There must be openness to such metaphoric correspondence between dreaming and waking life experience. And there must be intimate familiarity with important concerns and dynamics in the life of the individual.

To be likely to detect the linkage, via a striking common element, of two important hidden stories, each suggested by a cluster of literal and metaphoric correspondences like those given in illustration, compounds the requirements for detecting a set of correspondences suggesting one hidden story.

And to be likely to detect the intersection of such hidden stories in the lives of two or more people entails extending the study to include some unknown quota of people two of whom would discover or reveal such hidden stories, and the intersection of those stories, in the course of the study. It also entails being open to noticing and recording such an intersection.

The detection and recording of the events that are the elements of the associations that suggest the personally important hidden stories that give the appearance of being arranged across temporal and interpersonal barriers -- this detection is prerequisite to considering whether the correspondences, hidden stories, and intersections of hidden stories have any inferable causes, or whether they are products of chance and creative imagination, or whether they suggest principles outside a conventional causal framework. My concern here is with the satisfying of that prerequisite of detecting and recording key events, and with noticing and recording key associations. Discriminating among possible significances of the correspondences and the compositions based upon them is beyond my concern in this discussion.

To satisfy the requirements for detecting and recording the key events, whose occurrences are unpredictable, rare, and often widely separated in time, a study must collect from each dreamer reports of dreaming and waking life events and conditions, daily or nearly daily for a period of years. Since it is unpredictable what elements will be involved in correspondences that will cluster and suggest hidden stories, there is little to guide one in deciding what to record, except to try to make as faithful and complete a record of each recalled dream and waking event as possible, in the hope that what would prove relevant will make it into the record and will be found later. To satisfy the requirements for detecting and recording literal and metaphoric correspondences between dreaming and later waking life events, a study must involve regular review of the record, with an openness to such correspondences. To satisfy the requirements for detecting and recording intersections of one's hidden stories with others' hidden stories, a study must include regularly juxtaposing the experiences and associations of multiple dreamers over a long period of time.

The requirement of nearly daily observation for years means that the primary setting in which dream memories are collected for these studies will be, ipso facto, the everyday bedroom of the dreamer. Those who are doing these studies are already doing them in the only possible laboratory for them; there is no possibility of moving them into a laboratory that will not be the everyday bedroom.

The researchers doing these studies are the dreamers themselves. In all studies of dreaming, the seeming privacy of both dreaming and remembering means that the person doing the initial gathering of data is the dreamer. In studies intending to be sensitive to the kinds of correspondence that are my focus here, and to their association with important personal concerns of the dreamer, the necessary involvement of the dreamer is extended, usually to the point that the dreamer is the only candidate researcher.

The view that findings from in vivo observations of dreaming have no part in the scientific dialogue until they are quantified in experiment means that theorists do not take account of kinds of phenomena such as those in my hypothetical example. Such neglect would be an error even if there were not dream workers observing such phenomena. Since no experimental protocol in use comes close to satisfying the requirements for sensitivity to these classes of phenomena, theorizing based on experiment should be explicit about this insensitivity, as a limitation on the scope of the theory. The neglect to consider these classes of phenomena is an important error, both because there is evidence that there are such phenomena, and because these are phenomena that would have fundamental import in our conceptualizing of dreaming. I will discuss these issues further below.

Whether later understanding will show possibilities for fruitful experimentation concerning the kinds of phenomena under discussion is a question for the future. One might imagine, for example, attempting to discriminate among alternative hypotheses concerning the causation of the correspondences found. Experimentation with that aim, if possible at all, would still depend on detecting the elements and their correpondences, and so would still depend on the "instrumentation" -- personal disciplines and methods of recording and reviewing -- developed and used by experiential workers. Indeed, this experimental study would necessarily be experiential as well as experimental.

It is worth noting that records collected within a regimen that satisfies the requirements for sensitivity to types of correspondence with which I am concerned may also serve the needs of studies with different purposes, needs, and methods. For example, if the records for a long-term naturalistic study are collected in such a way as to be scorable on dimensions of concern for a different study, say a quantitative study, then the same data may provide samples for that study -- perhaps even statistically large samples, depending on the definition of the population being studied. While such possibilities enhance the potential value of records collected in long-term naturalistic studies, and may even demonstrate further unique sensitivities derived from such a regimen of data collection, I will restrict attention in this discussion to the sensitivities illustrated earlier.

The example presented above shows phenomena that raise important general questions. Do what we dream about, and what we dream about it, show consistency over years? Do our dreams reflect, or possibly guide, long-term growth and changes in our selves? Do they reveal otherwise hidden stories active deep in our lives? Do they include previews of unsuspected later experiences? Do they show connections with the dreaming of others in our lives?

And if there are hidden stories, previews, connections, etc., what is their nature? How do we see them develop? How do they seem related to the rest of our lives?

Questions like these are of great interest, but they elude elucidation in the laboratory. Laboratory studies can not begin to look deeply enough, or steadily enough, or for long enough, into a person's dreaming and waking to see and recognize phenomena that might answer such questions.

The context in which people in modern Western culture have most frequently attended to some of these questions -- those concerning hidden stories, growth, and change -- has been individual psychotherapy. In some psychotherapeutic practices, prolonged attention to a person's dreaming is used to discover and understand covert dynamics, such as "complexes" formed in early childhood; and to monitor ongoing change and growth. There is a high degree of agreement among practitioners who attend to dreams that they do reveal hidden stories active deep in our lives, and that it takes patient study to comprehend them. An important reason for the difficulty in seeing these stories is that they are emotionally charged, having implications that feel threatening to the person (Freud, 1953). The person (covertly) tries hard to avoid them, and so would be unlikely to see them without assistance, in dreaming or waking life.

Psychotherapy is also a context in which people, especially therapists, see surprising connections between one's dreaming (usually that of the client) and events in the dreaming or waking experience of another (usually the therapist) (Devereux, 1953; Eisenbud, 1970; Van de Castle, 1994, 409-411). Therapists also see examples of dreams giving previews of unsuspected future events (ibid.). They observe that both types of anomalous phenomena serve natural functions in the hidden stories of the client and the therapist.

Psychotherapeutic studies of the subtleties of the structure of experiences that reveal stories previously hidden to client and therapist alike are constrained by the therapeutic imperative. When the indicated therapeutic direction diverges from the direction of inquiry that would be most generally instructive about the phenomena, as it may, the therapist is bound to take the therapeutic course. That is, once the psychodynamics in a discovered story are apparent, warrant for further study of the phenomena involved in the discovery is limited or absent. Also limiting the yield, from psychotherapy, of studies of phenomena involved in the discovery of hidden stories is the apparent tendency for dramatic "paranormal" dreaming in therapy to be associated with a felt failing of the therapist, an association that may make the therapist doubly reluctant to report the phenomena (ibid.).

Findings from psychotherapy also have other limitations. The therapeutic relationship is temporary, so the duration of observations, while long compared to that in laboratory studies, will be shorter than intervals often recorded between two key corresponding events in journal study. And the therapist's observations are subject to the therapist's own emotional reactions, and to the therapist's theoretical orientation. Freud's opinion, for example, that "the creation of a dream after the event...alone makes prophetic dreams possible" (Freud, 1950), limited the kinds of hidden stories that he would consider in relation to reports of prophetic dreams.

But case studies of hidden stories discovered in psychotherapy are nevertheless valuable. Although under-reported, inevitably biased, and limited in scope in their view of principles and parameters of hidden stories, they sensitively reveal the existence of covert phenomena involving what is emotionally important in the dreamer's life. Each case illuminates the general class of phenomena by providing one piece of the puzzle.

People sharing dreams in peer groups also discover hidden stories active in their lives, and in the shared life of the group (Reed, 1977, 1985; Watkins, 1989; Ullman, 1996). Various workers conclude from extended group work that dreaming always functions to maintain or restore health, using "normal" and "paranormal" channels (Taylor, 1992; Ullman, 1996). Group members get to know their own and each other's consistency and growth gradually, as their dreams reveal their stories. They see themselves and each other dreaming not only of unknown currents in their lives, but also of unsuspected future events. They often find their dreams surprisingly linked.

Group dream work shares with psychotherapy, and extends, the potential for discovering interpersonally linked dreaming through sharing stories. Absence of the therapeutic imperative, while possibly limiting the degree of focus achieved, removes restrictions on studying connections discovered; and lower tension in the generally egalitarian role relations may remove inhibitions on reporting findings.

What groups share, though, within themselves or with the outside world, is subject to their overt and covert agreements and disagreements about interpersonal interactions. Their emotional dynamics and theoretical orientations and interests bias their studies, as they bias psychotherapeutic investigations (and indeed all investigations of dreaming). The frequency of sampling each person's dreaming will in general be lower than the frequency of the person's dream recording. Group membership also tends to be temporary, even if long-lasting, so the duration of observations is limited, as it is with psychotherapy.

We have only a few reports (e.g., from workers cited in this section and the following section) demonstrating dramatic group dream phenomena occurring naturally in people's lives. Few groups document their explorations and discoveries methodically. Reports that we have, though, show how extended dream-sharing reveals both the existence and the natural functioning of otherwise unnoticed shared stories.

Extended dream study in psychotherapy or in a peer group depends on the individual's recalling and making note, either mentally or physically, of the dreams and other events that become subjects of the study. Either of these methods of study has the further requirement of a specific long-sustained social arrangement. Individual journal-keeping entails the same initial requirement of recalling and noting, but is independent of any coordination with others in subsequent study of the record. It is in that sense a more basic form of long-term naturalistic study.

Individuals studying their own dreams for a long time also experience novel events that they nevertheless recognize from past dreams; discover dreamed connections with others' lives; and find themselves participating in stories that seem to come from within themselves but that don't seem to fit in their lives. They discover unsuspected themes and variations when they review dream records that span years. Most surprisingly, multiple alien-seeming hidden stories, discovered one by one through numerous but relatively rare "paranormal" correspondences, may gradually interconnect in an unsuspected order of organization that eventually does connect to what is emotionally important in the dreamer's life. (Schmidt, 1995; W. Phillips, 1996, Chapter 11; Schmidt et al., 1996)

It is possible to discover such an order of organization probably only through patient naturalistic self study. Only dedicated nearly daily systematic recording of dreaming and waking life experiences for several years has a chance of catching enough of the unpredictable interrelated events that the dreamer may recognize an organization in them. And to realize the personal meaning in the big story requires the intimacy and direct access to the dreamer's memory and associations that only the dreamer has. Journal-based self study can surpass other methods in sensitivity to these stories and their personal meanings, because the sampling frequency and study duration are constrained only by the dreamer's requirements; because only the dreamer has the view from within the dreamer's life; and because the dreamer's interests in the study may freely transcend the therapeutic.

In self study, though, as in other methods, the observer's emotional dynamics, theoretical orientation, and interests affect what one notices and how one attends to it. In self study, as in psychotherapy, there may be no perspective balancing that of the person doing the interpreting. And while the person's interest in the life under study will not waver, interest in the study may. The life being studied will always act in its own interest, which will sometimes override the interest of the study.

The prolonged discipline and personal openness that are required to see and reveal anomalous long-term coherence and its personal meaning make these studies difficult; reports of them are rare. There has also been no forum for reports of these studies to enter the scholarly dialogue. But their implications are profound. The documentation in personal journals of long-interval, often "paranormal" correspondences, and their roles in discovering hidden stories that become personally meaningful, is of great value to the field.

Extended naturalistic dream study, in psychotherapy, peer groups, or alone with one's journal, consistently shows that dreaming reveals hidden stories that are active deep in people's lives, stories that may be anomalously intertwined with other people's stories and that may refer to the unsuspected future as well as past and present. These findings depend on prolonged, regular, frequent recall and recording of dreaming and associations to dreaming. This requirement is due partly to emotional defenses obscuring the view, partly to the complexity and subtlety of the active stories, partly to their long time scale, and partly to the relative rarity of key illuminating events. While these phenomena are reliably observed, their invisibility except through the required discipline of study keeps them out of the view of the laboratory.

Any method of study that would be sensitive to the phenomena of concern to me here, both the elementary phenomena and their involvement in personally significant hidden stories, must involve at base a practice of recording dreams and other life events nearly daily over a period of years of ordinary living. For this reason, there is no loss of generality, in the discussion that follows, in limiting consideration to long-term natural-life journal-keeping as the fundamental form of extended naturalistic dream study.

In the face of the claim that extended naturalistic studies reveal important phenomena invisible experimentally, many scientists nevertheless resist the suggestion to elevate the status of observations and records from systematically-kept personal journals to that of scientific data. They have several reasons.

The first major criticism is that journal records inevitably fall short of standards for scientific work. Because naturalistic observers usually do not define a priori what to look for and record, or how to evaluate the observations, the critic argues, other investigators can not reliably evaluate them or attempt to replicate them.

Few records of naturalistic dream study contain descriptions of conditions and procedures sufficient for anyone reliably to reproduce them. The critic argues that deficient descriptions reflect technique that is inadequate for science. The often unsystematic methods are seen as those of explorers who are not scientists. When there are systematic methods, they are idiosyncratic and they usually change over time, as the explorer notices patterns or frequencies not registered before, or as personal interests change. And people using free-form logbooks to guide exploratory study will regularly step outside of any data collecting discipline that they usually keep. A waking life event jogs the memory or suggests that an unrecorded detail is significant. The journal keeper, seeking to understand the dreaming, will consider the newly recalled detail, and its waking life correspondence, along with what was recorded earlier (albeit keeping in mind how and when the new significance came to attention). Scientific discipline, the critic may argue, would require leaving it out of the study. Indefiniteness, instability, and violation of discipline in data gathering procedures are seen to preclude scientific evaluation of observations.

There is also unevaluable distortion in reports of journal-based studies, due to unavoidable editing and censorship. Lapses of memory, alterations of memory, and selectivity imposed by limited time for recording all affect the original record. The dreamer's emotional dynamics also affect the record; sometimes the dreamer is aware of these influences and sometimes not. Even with good original records, a full public report of what the dreamer has recorded privately may not be possible, for personal or interpersonal reasons. The critic argues that the reader can not be confident that the report is complete and undistorted.

In all these ways, the critic asserts, dream journals and reports of observations recorded in them fail to satisfy the same scientific criteria as, say, archives of astronomical observations.

A second major reason for scientists' reluctance to accept the claim that extended naturalistic observations are necessary is that they judge as unconvincing many of the inter-event correspondences suggesting hidden stories that support that claim.

Attributions of psychodynamic or anomalous correspondence between dreaming and waking life often depend on considering dream elements (and maybe waking life events as well) as metaphoric. This was the case, in the example presented earlier, in thinking that being beckoned off track in the dream might refer to being pulled off track, figuratively, in a later professional endeavor. Such attributions are more intuitive than rational; the critic objects that they do not meet any standard of scientific operationalism. Science uses operationalism (or operationism) as a discipline to exclude from discourse concepts not anchored concretely in observation (Bridgman, 1927; Russell, 1928). Metaphoric interpretation, whether guided by theory or ad hoc, whether intersubjectively convincing or not, is not objectively or operationally compelling. Without metaphoric interpretation, the critic points out, many apparent correspondences vanish, and with them the appearance of an extended story or an anomalous phenomenon.

When correspondence of form does seem real, the idea that it is meaningful often seems to the critic more gratuitous than rooted in theory. The critic suspects that chance may be the correct explanation.

The argument that long naturalistic observation makes a unique contribution to our comprehension of dreaming rests on claimed observations of naturally occurring correspondences between dreaming and waking life experiences. Many scientists doubt the reality or the significance of such correspondences.

A third major reason not to depend on extended qualitative naturalistic studies, the scientist may argue, is that, to the extent that there is something to learn from looking at dreaming over time, there are better ways to do it.

The critic points out that quantitative content analysis can answer questions about consistency and change (Hall and Van de Castle, 1966; Domhoff, 1996). Experience validates results from properly designed small samples (even very small samples, with the "Most Recent Dream" technique), minimizing reliance on extended recording of dreams by individuals. Content analysis can extend to people's associations to their dreaming (Cipolli et al., 1992). Questionnaires may reliably answer questions, the critic says, about individuals' dreaming over time, and people's stories, at a fraction of the effort of journal studies (Bernstein and Roberts, 1995; Bernstein et al., 1995). Longitudinal experiments can answer questions about long-term phenomena in the dreams of individuals (Cartwright, 1996; Foulkes, 1993). And, the argument continues, if there are anomalous phenomena to study, the laboratory is an appropriate place to study them (Ullman et al., 1989; Child, 1985; Van de Castle, 1994, 414-433).

This is an impressive set of objections.

The third major objection is the easiest to answer. The general answer is that while the methods cited do answer questions about long-term or anomalous phenomena of dreaming, they don't answer the most important questions that extended naturalistic studies ask; they don't reveal the same phenomena.

Quantitative content analysis can answer certain questions about consistency and change in individual dreamers, with the same reliance on extended naturalistic observation as has any other journal-based study. For the statistical analysis, the researcher scores the journal records on the selected content dimensions. While it illustrates a way to tap some of the value of extended naturalistic observation, this kind of analysis does not pick up any of the information that would reveal hidden stories, previews, or connections with others' dreaming. The key characteristic of Hall-Van de Castle scoring is that it concerns qualities or aspects or types, not identities, of the specific contents of dreams. One records, for example, "known unrelated adult male individual," not "Gary," and thereby loses access to whatever correspondences between other events and this dream depend on the identity and not just the type of this dream element. The granularity of scoring for statistical content analysis is too coarse to support identification of the kinds of associations that suggest hidden stories. But the Hall-Van de Castle scoring method does not concern associations of dream contents to thoughts, other dreaming, or waking life anyway. It includes no way to record such associations, so any that exist, of whatever granularity, are invisible to it. The significance of these characteristics of Hall-Van de Castle scoring in the present consideration is this: Naturalistic observers find that previews, connections with other dreaming, and some kinds of hidden stories become apparent not through frequencies of qualities or types but through associating specific dream content with elements of other dreams or waking life. So quantitative content analysis will not be useful for discovering those previews, connections, and hidden stories.

The usual use of quantitative content analysis is sociological, to study group norms of various qualities, aspects, and types of dream contents. The "Most Recent Dream" method and other methods using only few dreams from any individual are for studies of this type (Domhoff, 1996).

The associations of interest to Cipolli et al. in their content analytic studies are the dreamer's "source" associations to dream elements; that is, the dreamer tries to identify what might be the source of an element. As is the case with dream content in these studies, interest is in the types, not the identities, of the associations. Each "source" association is categorized as past episode, abstract self-referred, or semantic memory. The focus of the studies is the contingencies of distribution of these categories of source associations -- that is, what affects the mix of these types of source association. Even within the restricted set of associations considered (which excludes associations to future events and associations not felt as "sources"), the studies would not identify potential bindings of pairs of experiences in hidden stories.

Bernstein et al.'s Dream Content Questionnaire asks the same kinds of questions as Hall-Van de Castle content analysis, about qualities, aspects, and types of dream contents, but asks the dreamer to assess their "typical" occurrences and frequencies retrospectively, without recourse to records. The object of Bernstein et al.'s study is to evaluate this technique's usefulness in assaying dimensions of personality. Results in that endeavor are equivocal.

With this method of inquiry, one would not expect to learn the actual occurrences and frequencies of classes of dream content. Bernstein et al. acknowledge substantial differences between recorded and reported content. Even if the questionnaire did measure frequencies of qualities, aspects, and types of dream contents, it would remain able to answer only the same kinds of questions as the more detailed Hall-Van de Castle content analysis method.

Cartwright's and Foulkes' longitudinal experiments also would not discover hidden stories or the kinds of correspondences that suggest them. The studies would probably not be long enough, being one to five years in duration. But more importantly the frequency of sampling is far too low to give a reasonable expectation of recording even one pair of events that would have a kind of correspondence that, in context, might suggest a hidden story. It is not surprising that these experiments are not sensitive to the phenomena that are my concern, because their interests are different. These studies are concerned with the dream as a possible index of a developmental process in a specific dynamic life stage -- adult adjustment to divorce in Cartwright's study, child mental development in Foulkes'.

The studies of dream telepathy and precognition conducted in the 1960s and 1970s at Maimonides Hospital demonstrated elementary connections resembling those that occur and interrelate naturally. But those studies concerned only correspondences between the specifics of dream content and other events close or juxtaposed in time, in a context of brief focused intention to produce correspondence. They could not have shown the spontaneous long-term interrelations, and their natural functioning in the lives of the dreamers, that naturalistic observers discover and record. The principal investigator in the Maimonides studies, Montague Ullman, now favors naturalistic work in groups (Ullman, 1996).

The problem with all these proposals as alternatives to extended naturalistic dream study is that they are not sensitive to the most provocative phenomena reported by naturalistic observers. By their design, they would not detect any spontaneous previews or hidden stories implicit in connections or associations with specific dream content, even if they were present.

Astronomers are keenly aware of the need always to temper conclusions based on the absence of something in an observation. If the observing method would not detect an object or phenomenon if it were present, no conclusion follows from not seeing it; no theory should rely on its not being there. To investigate it, or questions of its existence, one must look with an instrument that could detect it (Freudling et al., 1995; McDowell, 1996). Dream study must consider biases due to insensitivities of observing methods as carefully as astronomers and astrophysicists do. To investigate some important questions, there is no alternative to extended naturalistic study.

The objections that records of naturalistic observations are unreliable and difficult to evaluate, and that it is unconvincing that they show otherwise unseen important phenomena, are more difficult to answer. Responses to these objections must recognize the validity of the concerns behind them, while arguing that the difficulties cited do not negate the need or the possibility of using naturalistic observations. The general form of the answers is that the difficulties are real, although not as bad as the objections suggest; and that since extended naturalistic observations may uniquely reveal important phenomena, we must use them, despite difficulties.

It is true that naturalistic observers do not work to agreed conventions and standards, measuring data on fixed dimensions by means of conventional operations. There are several reasons for this, some allowing remedy and some bound to remain.

Community-wide dialogue among long-term journalers about aims, techniques, and experiences is just now developing (Schmidt et al., 1996; Pearson et al., 1998). Informal communications have taught us that our findings are not idiosyncratic, that we see similar phenomena despite differences in how we observe. Discovering previews and active hidden stories through noticing how dream elements correspond with other events seems reproducible even if the methods are not founded in operational agreements.

The current dialogue may move journalers toward agreements about how to record data and what data to record. If we could agree on conventions for how to record various kinds of data, and if the agreed methods were practical, not upsetting the balance of dream naturalists' personal disciplines, we could improve our prospects for comparing, evaluating, and pooling observations. Agreement on recording conventions might also serve as a step toward agreement on a standard minimal set of data to record. Again, if the agreed operations were practical, and if the data were on dimensions of real interest, then we might further enhance the shared value of our records.

But such agreements are not easy to reach, because they could jeopardize other desirable conditions. Each naturalistic observer has developed a personal discipline involving a delicate balance of accurately witnessing and recording each dream in its own terms, regularly recording standard data of interest, finishing a record in an acceptable amount of time, making the records accessible and usable for review and chosen kinds of study, and maintaining interest and comfort to keep the discipline steadily for years. Disturbing that discipline could jeopardize the naturalist's primary mission of exploration. Data dimensions that we define will not supplant the intuitive exploratory work required to discover our unsuspected stories. Explorers seeking such stories will not use those dimensions and their recording conventions if they impede exploring. And whatever their discipline, explorers will follow their stories where they lead. Those stories are the unique contribution of experiential explorers; preserving and enhancing the viability and effectiveness of the personal disciplines developed for discovering the stories is more important than reaching complete operational agreement.

Another part of the complaint that reports of experiential studies are unreliable is that they are inevitably susceptible to editing and censorship. Serious investigators develop skills and practices, though, to reduce the risks. They train themselves to recall dreaming regularly, and to record the memory accurately and thoroughly, without censoring and without bias, but with as much awareness as possible of their own urges to bias or censor the record. They train themselves to practice careful discrimination among various categories of experience (to distinguish, for example, whether two dream elements are related through continuity in the dream, or by a waking logical bridge not from the dream, or through a waking association). Dream naturalists also practice disciplines to report their findings as fully and honestly as possible. When personal or interpersonal needs dictate an alteration in the story reported, the researcher changes as little as possible (and never in a direction to enhance the appearance of interesting phenomena), and informs the reader of the scope of the alteration.

These disciplines of the naturalist are never-ending struggles, but they result in records of high quality and reliability. The intersubjective agreement that we discover about what our hidden stories are like and what phenomena we see in them, even without operational agreements about how and what to observe, testifies to the robustness of the phenomena and the quality of our idiosyncratic disciplines.

Beyond questions about the quality of reports of naturalistic studies are doubts of the claim that such studies show, in physically anomalous and psychodynamic associations, important phenomena that are otherwise invisible. These doubts involve reluctance to recognize intuitive and metaphoric associations. There are two errors in rejecting the controversial findings of naturalistic observers on the basis of their reliance on intuition and metaphor.

First, they don't all rely on intuition and metaphor. The primary indication of extended hidden coherence is often, as in much of my work (Schmidt, 1995), a set of obvious and striking correspondences of form and quality that suggests, through additional obvious and striking correspondences of form and quality, that the individual correspondences are related. Lack of theory makes these phenomena little amenable to statistical analysis, but they appear even to statistically sophisticated observers to require explanation.

Using intuition and metaphoric interpretation does add something, though, and it is a second error to reject that. Intuition and metaphoric insight may bring out otherwise unseen associations that may contribute to realizing stories that make sense, sometimes compelling sense, of sets of otherwise puzzlingly related correspondences and associations. The argument that while such sense founded on non-operational association may be appealing we don't know that it is real, and that we should therefore reject it from consideration, has several weaknesses.

The first weakness is in the suggestion that science rejects ideas because they arise intuitively and resist empirical verification. If it did, cosmological theory, for one, would have no place in scientific discourse. Many, perhaps most, major mathematical and scientific insights arise intuitively in imagination (Hadamard, 1945). Researchers do not discard such insights; they know that it may take a long time to accommodate operational theory to them. While we may expect eventual accommodation to or disqualification of each intuition, there are always many being held in uncertainty.

A second weakness in the argument against intuitive association is in the selectivity with which the criterion is applied. No one rejects all inexact correspondences. We all have examples, in our own experience, of recognizing nonliteral waking life antecedents of dream elements. We are convinced of the relationship of antecedence, even though we could not establish it operationally. I dream of writing two pieces of music, one with steeply ascending chord progressions, the other with a gradually descending melody, and I associate that with my wakefully recognizing, the night before, a feature in many West African songs, steep ascent followed by gradual descent. I doubt that anyone will question my associating the dream with the earlier waking experience. This correspondence is near the beginning of a spectrum of degrees of abstractness. Once one recognizes a correspondence that is not quite literal but is nevertheless compelling, then it is hard to find a point that is not arbitrary for saying, "No, that association is too far-fetched." One is left without legitimate appeal to strictures attributed to operationalism for rejecting intuitive and metaphoric identifications of correlation.

But the argument that the operational philosophy of science prescribes such strictures, if only we could apply them consistently, is also weak. Even in stronger forms than are usually applied in social sciences, operationalism does not proscribe discussion, in terms not fully formally operational, of possible relations among empirical data. Operationalism is a philosophy and a set of conventions, closely related to logical positivism, for making abstractions from empirical observation (defining dimensions for abstraction, and abstracting on those dimensions) and for using those abstractions in discourse. It is a restriction of the kinds of abstractions to use. In his thorough revision of logical positivism, Rudolf Carnap (1956) "clearly exhibits a desire not to prescribe what should be regarded as meaningful from some metascientific or philosophical point of view but rather to describe what is commonly and usefully regarded as empirically meaningful" (Schlesinger, 1967). If we find that metaphoric interpretation makes empirical observations comprehensible, operationalism does not stand in the way of using it, regardless of the complications that it entails. And if we suspect that we see metaphor as an active principle structuring objective as well as subjective experience, operationalism does not prohibit our investigating the idea.

If operationalism does not exclude intuitive and metaphoric associations from scientific discourse, the critic would nevertheless be more comfortable if there were some empirical way of evaluating them. There have been a few experimental studies that one might expect to give some guidance in judging whether a dream element is a representative, literal or metaphoric, of a certain experience (Ullman et al., 1989; Cipolli et al., 1992; Strauch and Meier, 1996, Chapter 8). These studies have involved either unconstrained associations identified by the dreamer as believed (past) sources of dream elements, or associations to known presleep stimuli or other material by independent readers. The aim of these studies, though, is not to identify ways to judge whether elements associated are causally or genetically related. In fact, just as in naturalistic studies, there is no evidence of genetic relatedness of the associated elements in these studies, beyond a subjective impression of plausibility (possibly an intersubjective impression, when there are independent readers). (The studies themselves are of coarse categories or characteristics of associations. Except for intending specifically to find a past source, the unconstrained association by the dreamer is similar to the associating typically done in extended naturalistic study; but the details that might suggest stories of interest to the naturalist do not receive attention in these studies.)

Naturalist and critic alike are left with no operationally convincing way to discern the origins of correspondences identified in extended naturalistic studies. We do not know how, or even whether, nature has arranged each correspondence identified. The naturalist, regarding a correspondence as possibly indicating a commonality in the genesis of the associated elements, tries associations in combination with each other, giving the most continuing attention to those that are the most dramatic and those that are the most productive in combination. There may be no better test of the value of an association than subjective judgment of how well it contributes to plausible interpretation of observations, and how plausible, how comprehensive, and how elegant that interpretation is.

But elegant plausible interpretation of otherwise baffling phenomena is a very valuable result. Researchers in other fields, such as sociology and anthropology, have articulated methods similar to those used by dream naturalists, because of their value for generating an "intuitive grasp" or a "substantive theory". Severyn Bruyn emphasizes the importance to the phenomenologist or participant observer of immersion in the subject of study in its natural setting, free of preconceptions and hypotheses, "in order to obtain an accurate intuitive grasp of it," to "intuit essences and essential relations existing in the symbolic data" (Bruyn, 1970, 284, 285). Glaser and Strauss (1970, 288-289) have argued both the particular potential of qualitative research to reveal substantive theory, and the inappropriateness of holding it to a standard that ignores its unique usefulness:

These arguments show that the hazards inherent in naturalistic studies do not justify excluding the studies from scientific discourse, nor according them a reduced status.

But a dilemma remains. These studies are not science in a usual sense, and they leave us not knowing how to evaluate the sometimes incredible answers that they give to the important questions that they ask. The field of dream study, unlike astronomy, has no major observatories capable of taking up these questions with as much sensitivity and power as experiential workers can bring to them. Science, far from meeting an expectation to be "the final arbiter of truth" (Van de Castle, 1994, 208), to limit our thought to what we are sure of, seems in this situation to have no means to come to a decision.

Relief from the dilemma of having to and being unable to judge the truth can come in two ways. The first way is to recognize that the presumed obligation is illusory. The second way is to learn new ways to reduce uncertainties in the observations and ideas, while accommodating whatever uncertainties remain.

Certainty of knowledge, certainty of the reality and significance of observed correlations, is not a requirement of science; it is rather a goal that science serves. Scientists, impatient with having no "scientific" method to resolve the provocative questions raised by careful but unavoidably imperfect naturalistic observations and uncertain or imprecise imaginative thought about them, may wish to dismiss the troublesome reports; but science requires the opposite response. An uncertain observation or conjecture may be inaccurate, but it may be accurate; so it is an error to reject it, just as it is an error to accept it. We must retain in thought, not reject, what we are not sure of.

When the methods and observations are conventional and non-controversial, practicing scientists know this. In any active field, there are competing hypotheses, and inconsistent observations are common and accepted as part of the process. Astrophysicists have evidence that the physical universe can not be as old as some of the stars must be (Wilford, 1994; anonymous, 1996). They live with this inconsistency, not rejecting either set of evidence. They avoid both risks, of rejecting a true idea and of accepting a false idea, by holding the incompatible ideas in uncertainty. They have faith that later observations and imaginative insights will bring some resolution.

The complementarity of those two risks, of rejecting true ideas and accepting false ones, is well known in many fields. In statistical hypothesis testing, they are called type I errors (accepting false hypotheses) and type II errors (rejecting true hypotheses) (Bethea et al., 1984, 180-181); in information retrieval theory, they are called failures of precision (retrieving irrelevant information) and failures of recall (failing to retrieve relevant information) (Parsaye et al., 1989, 315-318); in radio theory, they are called low selectivity (accepting noise as part of the signal) and low sensitivity (rejecting part of the signal as if it is noise) (Orr, 1972, Section 10-2).

The objections to accepting the dream journal as a scientific instrument are arguments that selectivity -- rejecting false ideas -- is important enough in science that the goal of high sensitivity -- recognizing and accepting true ideas -- may be partially sacrificed to it.

In some contexts, selectivity is more important than sensitivity. Filtering out noise can enable detection and use of a signal that would otherwise be lost. Statistical filtering in evoked potential studies, for example, enables neurological diagnosis on the basis of signals that appear swamped by noise.

But in other contexts, sensitivity is more important than selectivity. An example is in detecting wind shear from aboard an aircraft. The National Transportation Safety Board named a Honeywell wind shear alert system as a factor in a 1994 USAir crash, for failing to warn the crew (E. H. Phillips, 1994). The system inhibited alerts when the flaps were in transit -- common during takeoff and landing, when wind shear is especially dangerous -- because the flaps would cause some false alarms. In the wake of the accident, Honeywell modified the system, to let all the alerts come through. Now flight crews have to live with uncertainty about whether an alert is real; but in this case it is more important not to miss a real one than to eliminate all the false ones.

Of course we want high sensitivity and high selectivity. In science, there is no reason to endorse either type of error. Science, as a quest for knowledge, must neither accept nor reject what it does not know. Science strives for theory, but on the way it holds many observations and ideas in uncertainty. We advance our science by finding answers to our questions, but also by increasing our capacity to hold in mind whatever appears relevant to our inquiry and to keep clear on the epistemic status of each thing -- how well or how tentatively we know or suspect it. Science is not obligated to, and it must not, judge the truth of what it has not investigated with sufficient sensitivity and sufficient selectivity.

The same conclusion follows, perhaps surprisingly, from Ockham's razor. While Ockham's razor is usually invoked to justify excluding some specific element deemed extraneous in a theory, its principle logically applies as well to extrapolation of a theory into areas not investigated. The long form of Ockham's principle states, "Nothing is to be assumed as evident, unless it is known per se, or is evident by experience, or is proved by the authority of Scripture" (Ockham, 1967). To assume that intuitive insight is inaccurate, for example, or that metaphoric correspondence is accidental, is to assume what is not known per se and is not evident to experience.

The second way to relief from a felt dilemma between obligation and inability to discriminate the truth is to get better at the discrimination. What can we do to improve our constructive use of the skills and sensitivities developed by dream naturalists, and to improve our confidence in what we may learn from them?

A first step is to increase awareness of what naturalistic observation over long periods discovers. Those who do such observing need to report on their work. They deserve to be heard, to have their work critiqued according to its reason, its carefulness, and its elegance in interpreting what is seen. Those who are not familiar with this work deserve an opportunity to see it, to get a feel for what it shows and how, and to think about it.

Laying this groundwork will require active efforts, by both speakers and listeners, to bridge the gulfs of vocabulary and assumption that separate them. Even with no antagonism in the listening or the speaking, those gulfs are real barriers to communication.

With more of a common view of what experiential workers observe, scientists and journalers alike may turn careful attention to how to gain more clarity. Present and new journal keepers, including people who are already scientists, can look upon their journals in a new light, as sensitive instruments able to serve not only the journalers themselves but also our common quest to comprehend dreaming. We may critically consider the further development of methods that naturalistic observers use to look beyond the personal immediate of their dreaming.

Effective content-based recall of past dreams and waking events is fundamental to investigating hidden stories based on long-interval correspondences. Without good recall, one will not find the original records that provide the details and the contexts that are essential in both developing and evaluating hidden stories. Beyond finding records of the events that form the backbone of a story, we who do these studies are often aware of not finding perhaps even more records that we know exist and are relevant. We are also regularly surprised to find records that we could not have found by intention. Improving recall is important for enhancing our ability to evaluate whether dramatic events stand alone, or are strengthened by correspondence with other events, or are weakened by being commonplace (or, perhaps, are strengthened by being commonplace).

Good recall is difficult to achieve, first because both indexing and data entry for automated searching are labor intensive, and second because identifying and implementing effective search strategies for this work are challenging. Continuing advances in computer technology, and pioneering efforts in indexing, database, and hypertext development for journal study, show promise for significant advances in retrieval for review and discovery. Henry Reed, Sarah Richards, and Bjo Ashwill, among others, have worked on fully computer-based systems, while Juhani Kääriäinen, Cynthia Pearson, and I, among others, have worked on hybrid paper-and-computer-based systems.

Careful study of the structures of hidden stories and of the correspondences involved in their discovery characterizes my work (Schmidt, 1995) and that of Gloria Sturzenacker (1997) and Cynthia Pearson (1998). Part of this work involves exploring methods for discovering and representing -- in diagrams, "composite narratives" and other forms -- the correspondences and the hidden stories that emerge from them. (My "composite narratives" resemble forms that Don Kuiken (1991) has developed for interpreting correspondences within single dreams. We developed our methods independently.) Developments in the structural representation and visualization of the forms that we see -- correspondences, "synchs", "constellations", "Arabesques", "hidden stories", "natural allegories", and "life maps" (terms used by investigators Pearson, Schmidt, and Sturzenacker) -- will be instrumental in extending our comprehension of the structures.

Discovering shared or intersecting hidden stories requires the development of networks of people sharing dreams, with an openness to collective dream phenomena. Local dream groups often discover intersecting stories among their members. Henry Reed's work on the Sundance Community Dream Journal (1977, 1985) involved dream reports collected over time from a distributed community of dreamers engaged in the common Sundance endeavor. The reports showed surprising correspondences in detail, consistent enough to suggest underlying hidden common stories. Susan Watkins (1989) made similar discoveries in a community of subscribers to a newspaper. Linda Lane Magallón (1997, 1998) and others have begun to use the World Wide Web as a medium for dream sharing with attention to collective phenomena.

In addition to developments in the content-based retrieval of records, the structural representation of correspondences, and the detection of intersections between different dreamers' experiences, new applications of quantitative methods may also be useful in the study of phenomena uniquely detected by long-term naturalistic observation.

While Hall-Van de Castle content analysis is not sensitive to the correspondences that reveal hidden stories, previews, or connections with others' dreaming, this method may have other applications in these studies. One may wonder whether dream elements identified by other means as elements of such correspondences tend to be of types that occur rarely. Unusualness is often a factor in subjectively or tentatively judging significance. Content analytic studies can tell us relevant conditional frequencies of occurrence of the types. If the elements of the correspondences do tend to be of rare types, then findings from content analysis may become useful in prospecting for candidate members of key correspondences. If on the other hand the elements are of common types, we will know that commonness is not an indicator of insignificance. Either way, the new knowledge will be a factor in theorizing about the correspondences and the stories they reveal.

Linnea Brush (1993) has proposed a method of content analysis with features that may open up new possibilities in the study of hidden stories. The method includes definition of specific procedures for studying correlations, trends, and other timing effects. The special strength of Brush's method is its sensitivity to what Brush calls "companion symbols", symbols co-occurring in a period of life even if never co-occurring in a dream. It is possible that such correlations may serve, like long-interval correspondences, as clues to hidden stories. Brush leaves it to the dreamer to define content categories, which may therefore be as specific and idiosyncratic as they typically must be to suggest such stories.

Brush's method also suggests the possibility of further extensions of content analysis, for looking at other kinds of correlations of specific dream content. For example, symbols that never occur in the same dream, nor even in the same period of life, but that do occur in similar contexts in different periods, may also be clues to hidden stories. There are indications of this in my work (Schmidt, 1995).

Statistical hypothesis testing is another quantitative technique, one that may be of particular interest to scientists who wonder whether literal and metaphoric correspondences might be due to chance. In this view, events that occur independently will sometimes coincide or correspond by chance. If one can estimate the independent frequencies of the kinds of events involved in correspondences, then one can estimate the chance frequency of those types of correspondence. Years-long nearly daily dream recording would seem to offer opportunities for powerful hypothesis testing. For example, having noticed over a period of years that one nearly always receives a letter from a certain irregular correspondent within a few days after that person appears in a dream, and almost never at any other time, one might investigate how likely it is that the correlation is due to chance. With a dense sampling of dream recall throughout the period, and a complete record of correspondence (i.e., mail) received, one might perform a powerful significance test.

The validity of statistical significance tests, though, depends on conditions that are virtually impossible to satisfy in extended journal-based studies. The universe from which one is sampling is difficult or impossible to define. Most scoring dimensions that would prove to be interesting are not defined a priori. And the needed independent theoretical probabilities are essentially impossible to estimate. As a result of these difficulties, it is not clear how great a potential there is for useful hypothesis testing using extended journal records.

There are some new statistical tools -- the Vapnik-Chervonyenkis Dimension and Rissanen Minimum Description Length Modeling (Rissanen, 1989), for example -- whose conditions for validity may be much easier to satisfy in naturalistic investigation. With these methods, one does not hypothesize a universe of a certain complexity and then evaluate whether random sample data plausibly come from that universe. Rather one starts with sample data and then estimates how complex the universe must be to produce, by chance, data of the observed complexity. Then one may judge whether the universe from which one is sampling is plausibly that complex. There would be significant difficulties, though, in judging the meaning of the application of even these methods in naturalistic studies. "Minimum description length" in natural language, for example, is difficult to estimate and arguably rather arbitrary. It remains to be discovered whether these methods will be useful.

We should also be cautious in using statistical tools to judge significance, remembering that chance statistics need not imply insignificance. Natural language, for example, full of meaning, is characterized by many very regular statistics.

There are further prospects for improving how we learn from long-term journal records, beyond the developments for discovery, structural analysis, and quantitative analysis already discussed. These additional prospects are the ones that involve the most challenging aspects of integrating naturalistic and experimental science: developing our capacity to hold and work with inevitable uncertainties. Respecting natural conditions that may limit one's probing but that may be essential to the phenomena and the observing; extrapolating from one's observations only with generous reserve; working with correspondences whose possible significance resists statistical analysis; honoring intuitive, literary, or religious conjecture -- these exercises of restraint and openness, which may be difficult for the scientist seeking clarity and discrimination, will leave open channels for learning that would otherwise be closed off. In time those open channels may lead to substantive theory that is grounded in reliable observation; that is of a scope that is grand beyond what is operationally testable; and that, while perhaps (or perhaps not) destined always to remain operationally tentative, is "testable" intuitively, literarily, mythically.

Harry Hunt (1991) and Kelly Bulkeley (1994) contribute to this aspect of our effort to gain clarity in our studies, with their clear views of the essential tensions in dream study and the need for many vantage points, including scientific, literary, religious, and artistic, to begin to comprehend dreaming.

Obviously it is not going to be easy to join our efforts from different perspectives. We may get some inspiration from other fields. We see in astronomy how a discipline may depend exclusively on naturalistic studies and may develop effective methods for learning from very large quantities of observational data. A field that, like dream study, concerns aspects of human subjective experience, the study of memory presents an example of how experimental studies may complement naturalistic studies (Neisser, 1982). A model that may be particularly apt for dream study is provided by the field of social primatology (Waal, 1989).

The relationship, including the tension, between naturalistic home-based dream study and experimental dream study has a close analog in primatology:

While we may get ideas from other disciplines, the integration of methods of studying dreaming will be unique. The endeavor to join our efforts will involve listening to each other's disparate assumptions, methods, observations, and thoughts. We will need to treat each other decently -- an obligation that sadly has not always been met, particularly in responses to reports and studies of anomalous phenomena (as McClenon, 1984, and Child, 1985 document).

The responsibilities that we have to each other (and to ourselves) include:

• That we be honest, conscientious, careful, thorough in all of our work.
• That we respect the limitations of our own work, being clear about what is not implicit in it. That we clearly distinguish among what we know, what we conjecture, what we take on faith, and what we infer from each of these.
• That we not ignore or defy prima facie legitimate observations by others. That we not suspect wrongdoing simply because of what an observer claims.
• That we temper our inferences from our own work, with respect for prima facie legitimate observations by others.
• That we participate, in good faith, in fair discourse about our work and others', being careful to honor the above requirements in discussing controversies.

• That we be strongly interested in any given direction of study.
• That we agree on the basic nature of existence, when such agreement would depend on accepting the same conjectures, items taken on faith, or inferences dependent on these.
• That we agree on how we may know. (It is reasonable, though, to assume common belief in sharable and communicable knowledge derived from personal observation of objects of experience and thought. Without this, dialogue is meaningless.)

The Association for the Study of Dreams goes further than any other group that I know in restoring the reciprocal functioning of complementary approaches to learning about nature. While we have a way to go, we have the potential to integrate experience, experiment, and theory in dream study. The integration is required, because experiential workers see robust phenomena that experimenters can not see and that theorists ignore. While it calls for dealing with problematic methods, the integration would bring the needed increase in sensitivity. It would also open up discussion of other questions that benefit from or require long naturalistic observing, such as language use in dreams and allegory in dream sets. Support in an integrated endeavor would also lead to improving the precision achievable with the difficult methods.

An integrated field of dream study would be a new fusion of disciplines, in which scientists appreciate open-eyed straining to see everything that there is to see, and explorers appreciate scientific striving for clarity and sureness. We would manifest in a new way this fundamental attitude of science: Look to nature, not to received wisdom or other preconception, to learn how nature works. Look through eyes both creative and discriminating, and use imagination in your seeing. In fact, use every way of looking that you can.