This book presents important new findings and new ideas about remembering: what people remember, why they remember it, and how we might best describe the processes involved. These issues have been considered before, most notably by Sir Frederick Bartlett in the 1932 book from which our title is derived. Arguing that standard laboratory methods did not do justice to what he called the subject's “effort after meaning,” Bartlett introduced both new procedures (recall of meaningful stories) and new theoretical concepts (e.g., “schema”) into the study of memory. Although the influence of his work was slow to develop at first, its importance is now widely acknowledged: Story recall has become a standard laboratory technique, and “schema” a standard theoretical idea. No longer revolutionary, Bartlett's insights have been effectively assimilated into the traditional psychology of memory.

In the mid-1970s I undertook a general critique of contemporary mental-model-oriented cognitive psychology, including the study of memory (see Cognition and Reality [Neisser, 1976]). In the hope of stimulating a serious reconsideration of the prevailing assumptions in memory research, I presented a paper sharply critical of traditional approaches at the Cardiff conference on “Practical Aspects of Memory” in 1978. After nearly a century of research, psychologists still knew almost nothing about recall of early childhood experiences, about memorizing songs and poems, about memory for places, faces, and names, about the fate of knowledge acquired in school, about oral history, about testimony, about prospective remembering, or even about individual differences.

In his book Memory Observed (1982), Neisser claimed that almost all interesting questions about memory have been ignored by experimental psychologists. The conference that is documented by this book was meant to help remedy this situation by launching an ecological approach to memory research. Contributors were invited to attend the conference because their research was considered ecologically valid. When I received my invitation, I viewed this with some misgivings, because I identify with the “establishment” memory researchers who apparently have done uninteresting work. Also, my research addresses different issues and, perhaps, is less clearly ecologically valid than is the research other contributors are reporting here. However, I gained some comfort by noting at least a superficial similarity between the ecological approach to memory research and the approach advocated by functionalists such as Dewey (1910). The approaches agree in their claim that the functions of memory must be viewed as allowing adaptation of an organism to its environment and in their preference for talking in terms of processes or mental operations rather than in terms of mental structures. Most of the research of the early functionalists would seem to meet Neisser's criteria for ecological validity, although they conducted their research in the laboratory as well as in natural contexts. The advantage that I gain by noting these similarities is that the functionalists provide an excellent example of the coordination of laboratory research and ecological concerns.

This is the second volume in the Emory Symposia in Cognition series. It was originally conceived in a seminar on memory that the two of us conducted together in the fall of 1984–a seminar characterized, like all Cognition Project seminars, by a vigorous airing of many different points of view. Although we devoted a good deal of time and thought to the ecological approach (as is also typical of Cognition Project seminars), more traditional lines of research were not neglected. Indeed, we could not neglect them if we were to do justice to our topic. The study of memory has undergone a remarkable renaissance in the past few years; good ideas and important research have begun to appear from many different directions. There are new findings everywhere: in the naturalistic study of memory, in the laboratory study of basic processes, in development, in neuropsychology, in the understanding of clinical memory disorders. Although we could not do justice to all of those areas in the seminar (nor, alas, in this volume), we did our best to focus on good ideas wherever they might come from. A lot of those good ideas are in this book.

Although the primary goal of the seminar was to understand as much as we could about memory, it was clear that we were working toward a conference as well. The first Emory Cognition Project Conference–whose results have since been published by Cambridge University Press as Concepts and Conceptual Development: Ecological and Intellectual Factors in Categorization (1987)–was held that same fall, and we were pleased with its success.

The focus of this chapter is on how memory for real-life events–both specific episodes and general schemas–develops in the early years of childhood. To begin the discussion, it is useful to consider an example of such memory in a very young child. This example is taken from a very rich corpus of material that was obtained by periodically recording the presleep monologues of a child, Emily, between the ages of 21 months and 3 years. The first example in the Appendix is taken from the very earliest recordings to give a flavor of this material. At a later point I shall consider how the monologues changed and developed over time.

In this first example, Emily talks about a real event that, according to her mother's report, had occurred 2 months earlier when one of their cars was being repaired and they had to use the other one, an unusual event in Emily's experience. Her mother was sure that no similar event had taken place in the interim period. Evident in this example are some of the characteristics of presleep monologues of 2-year-olds that have been noted by other researchers, for example, repetition of utterances with substitution and rephrasing. Of some interest here are occasional apparent intrusions into the main topic (e.g., the reference to her friend Carl). However, more impressive is the sustained topic over roughly 30 separate (although not different) propositions.

This chapter consists of three relatively independent sections. The first provides an overview of recent developments in various fields of memory, the second suggests an ecological way of thinking about memory for personally experienced results, and the third offers a conjecture about the relation between personal memory and spatial orientation. My own confidence in the several sections is mirrored by their order of presentation: I think of section I as almost incontrovertible, of section II as an essentially reasonable proposal, and of section III as frankly speculative.

The new memoria

In the past 10 or 15 years, the psychology of memory has undergone a fundamental change. This shift, which began well before the current interest in the possibilities of an ecological approach, has not been the work of any one individual or research group. What has happened is not merely the emergence of a new method or a new theory, but a change in our notions of what kinds of things people remember in the first place. In effect, we have a new definition of what memory is of.

Twenty years ago, the principal function of memory was assumed to be the storage of individual experiences and actions. The strings of numbers, words, or pictures that subjects were asked to remember in the laboratory were surrogates for the sequences of specific stimuli (or percepts or responses or ideas or whatever) that they presumably experienced and remembered in the outside world.

There is a class of real events that we know about and remember, though we have not experienced them. Learning about such events does not require clairvoyance or divine revelations, but only listening to a friend, reading a newspaper, or watching television. The fact is that we may know about real events in the world in two different ways, either by personal experience (directly, firsthand) or by the report of someone else (indirectly, secondhand). I shall call events that are known in these two ways experienced events and reported events, respectively. This chapter attempts, first, to demonstrate the pertinence of this distinction to research on real-event memory; second, to advance a theoretical analysis of memory of the two classes of events in terms of the information available to specify them; and, third, to examine some empirical results relevant to this analysis.

A little personal background is in order to set the topic and the approach in perspective. My interest in these issues stems from several years of research into memory for texts and particularly news reports (Larsen, 1983). News reports are different from the texts of most memory experiments because they are about real events and because they are important in everyday life. Therefore, news memory seemed to deserve a place in the emerging field of ecological memory research (Neisser, 1982a). But in this research the emphasis was heavily on autobiographical memory.

This conference on ecological and traditional approaches to memory was held in 1985, 100 years after the publication of Ebbinghaus's monograph (1885/1964) marking the beginnings of a science of memory. Although this happens to be a centennial coincidence rather than an observance, I want to emphasize some continuities between the tradition deriving from Ebbinghaus and more ecologically oriented research. There is a tendency to view the ecological approach to memory as contrasting sharply in both method and theory with the Ebbinghaus tradition. Indeed, some psychologists might see the research represented in this volume as exciting in direct proportion to the liberation it shows from the presumed dreariness and artificiality of the verbal learning tradition descending from Ebbinghaus. I argue here that this would be a mistake and, instead, suggest that there is much in common between laboratory studies of memory and the study of memory for events that did not occur in a psychology laboratory. Further, I believe that Ebbinghaus himself would have approved of most of what he would have heard at the conference.

Continuities across approaches

Verifiability

Ebbinghaus's most important methodological contribution to the scientific study of memory, in my opinion, is that he carefully controlled the conditions at encoding. In the Ebbinghaus tradition, the experimenter is present at the encoding and can therefore score for the accuracy of remembering later on. The problem of verifiability, as Brewer puts it, presents difficulties for the majority of studies of autobiographical memory.

One of the fundamental issues guiding work on autobiographical memory in recent years concerns the role of higher-order cognitive structures in these recollections. Presumably, the influence of such structures is a matter of degree, depending on such variables as the delay between the remembered event and the memory, the cues available at the time of recollection, and the character of the materials remembered. Boldly or timidly advanced, views that attribute a role to such structures require that at least some features of some autobiographical memories involve not the retrieval of mental copies of the original experiences but rather reconstructions based largely on our general knowledge of related everyday affairs. The teeth in this reconstructionist position show when its partisans employ it to explain even those memories that seem to subjects as if they did involve nothing more than the direct retrieval of mental copies.

Until quite recently, Craig Barclay and William Brewer both have defended relatively strong versions of the reconstructionist view of autobiographical memory. Now, apparently, only Barclay does. Both Brewer's chapter and Barclay and DeCooke's chapter respond to challenges to the sort of schema-based approach to mnemonic reconstruction that they have championed in the past. Barclay and DeCooke aim to improve on the previous research in order to offer additional evidence in support of this position. Brewer, by contrast, continues a recent trend toward a more moderate-sounding position (Brewer, 1986).

This chapter describes two empirical studies of autobiographical memory that focused on the personal memory component of autobiographical memory. These experiments investigated the characteristics of randomly sampled events and of events that subjects selected to be memorable. The autobiographical memory data in these experiments were obtained from naive subjects, not from the experimenter. Both experiments were designed so that it would be possible to study memory for thoughts independently from memory for actions. The second experiment was designed to study cued recall of verifiable autobiographical memories, and in this experiment systematic reports were obtained of the subjects' phenomenal experiences during the recall process.

The data from these two experiments were used to address a wide range of issues in the study of autobiographical memory. Memory for random events was compared with memory for subject-selected events, and memory for thoughts was compared with memory for actions. A number of analyses were carried out to uncover the relationships between the characteristics of the original events and memory for these events after various time intervals. The cued-recall data were used to study the accuracy of autobiographical recall and to work out the relations between the contents of the events and memory for the events. An analysis was carried out to study the relative effectiveness of various forms of cues for the retrieval of information from autobiographical memory. The data from the phenomenal memory scales were used to explore the subjects' experiences during the recall process.

This book represents an emerging theme in the field of memory research: a shift in focus from examining what memory is, a structural approach, to what memory is for, a functional approach (Bruce, 1985; Oakley, 1983). As Nelson points out, one of the primary and developmentally early functions of event memory is to organize our knowledge about the world, and as Barsalou's data demonstrate, this remains an important function of event memory throughout development. But there is another function of event memory. Event memory not only organizes our knowledge about the world, but also helps organize our knowledge about ourselves. Our sense of self and event memories are interwoven systems. We learn about ourselves by interacting with the world. Who we are is largely defined by what we do–the kinds of activities and events that we engage in.

Moreover, it is the sense of self that is crucial for autobiographical memory. Autobiographical memory is not simply memories of previously experienced events; it is memory of the self engaging in these activities. It is the sense of self that makes the memories cohere as a life history that expresses the essence of who we are. Whereas event memory serves the function of organizing our knowledge of the world, a predictive function, autobiographical memory serves the function of organizing our knowledge about ourselves, a self-defining function. I am not arguing for two separate memory systems, only for multiple functions.

There are a few people who think and write so clearly that their work is almost always influential. They can shape a whole field. What they say has to be taken very seriously. Failure to do so can waste a lot of time either because valuable insights are lost or because less-than-valuable insights are followed. Professor Neisser has shown himself to be such a person.

There are some problems that have “garden-path” solutions that are so seductive that psychologists repeatedly take them. Professor Neisser has warned us about many of these, including the reappearance hypothesis and the fact that the physical onset and offset of stimuli are not the psychological onset and offset of stimuli (Neisser, 1967). I think nesting is such a garden path, partly because it contains a kernel of truth and partly because we have been misled by similar concepts in the past. In Gibsonian ecological terms, nesting is a garden path because it makes a structure out of a process.

In what follows, Gibson's use of the word ecological is taken specifically to include the following principle: Although we should always start by making a very good description of the environment before wondering how the animal interacts with the environment, we should not copy that description of the environment into the mind.

Taxonomy

Shelducks are intermediate between the true ducks and the geese, although they are usually placed with the ducks in the family Anatidae, where they join the sheldgeese in the tribe Tadornini (Johnsgard, 1961). There are seven species of shelducks: the common or northern shelduck Tadorna tadorna (L.), the ruddy shelduck T. ferruginea (Pallas), the Cape or South African shelduck T. cana (Gmelin), the Australian shelduck or mountain duck T. tadornoides (Jardine & Selby), the New Zealand or paradise shelduck T. varietaga (Gmelin), the radjah shelduck or burdekin duck T. radjah (Garnot) and the crested or Korean shelduck T. cristata (Kuroda) (which may be extinct). All are large brightly coloured ducks with many goose-like features. The lack of a camouflaged female plumage, the persistence of the pair bond and the prolonged parental behaviour shown by the male all resemble the geese, whereas the general morphology, voice and the existence of an eclipse plumage are duck-like. Johnsgard (1978) considered that, like the radjah shelduck, the common shelduck represents a rather isolated offshoot from the rest of the group. Both species are adapted for dabbling and feed chiefly on molluscs and other invertebrates, whereas the rest of the group are principally vegetarian.

The common shelduck was originally called Anas tadorna by Linnaeus in 1758, but has had a variety of other names including Tadorna cornuta (Saunders, 1889) Tadorna vulpanser and Tadorna bellonii (Yarrell, 1843).

Having selected a nest site, the breeding bird must build a nest, lay a clutch of eggs and incubate them. The number of young hatched will depend on the size of clutch which can be produced and success in hatching them, in the face of many factors which might cause failure. Incubating birds must divide their time between the eggs and the need to spend some time feeding, the balance depending on the fat reserves which can be lost over the incubation period. In this chapter I will discuss these and other problems encountered by laying and incubating shelducks.

The timing of laying

The date on which the first egg of the clutch is laid can only rarely be determined by direct observation during the egg-laying period, due to the inaccessibility of the nests and the birds' intolerance of disturbance. Instead, a number of indirect methods must be used. In a few nests, observed before laying is complete, the laying date of the first egg can be back-dated since normally one egg is laid per day (Hori, 1964a). If the hatching date and clutch size are known, the laying date can be estimated using the incubation period of 29–31 days (Hori, 1964a; Young, 1964a), plus one day for each egg in the clutch. More commonly the laying date is back-dated from the first sighting of the brood of ducklings, adding a further day which the young spend in the nest after hatching (mean of four nests observed at hatching by Young, 1964a).

Parental behaviour normally involves a number of different activities: feeding the young, providing them with shelter and protecting them from predators and other dangers. By investing time and energy in this behaviour, even at some risk to their own chance of survival, the adults enhance their genetic fitness by increasing the survival of their progeny. Shelducks, like other parents of precocial young, do not feed their broods but merely accompany them while they feed themselves. The parents do, however, actively provide shelter and protection as I shall describe in this chapter.

Leaving the nest

Newly hatched young shelducks stay in the nest for some time, usually for at least 12 h and up to four days if there is a wide spread of hatching date within the clutch (Hori, 1964a). During this period, ducklings and female call frequently. Hori was able to approach very close to a nest inside a shed and described a monosyllabic ‘aarrk’ and a soft running ‘ugg ugg ugg’ given continuously for long periods by the female. I installed a microphone in a nest burrow over the hatching period and heard similar calls, along with piping responses from the ducklings. All of the calls are given very softly and can only be heard at very close range. It is likely that the ducklings become imprinted on the mother's voice while in the nest (Gottlieb, 1965). Since the ducklings' first hours are spent in the dark, it is obviously functional to use auditory rather than visual characteristics of the mother as a basis for imprinting.