The Making of the Mind: The Neuroscience of Human Nature (18 page)

The retrieval of an episodic memory is a reconstructive process. By understanding this fundamental fact about our recollections of the past, it is readily apparent how constructing the future makes use of the same cognitive machinery. Reconstructive retrieval is not at all like our everyday experiences of retrieving printed or digital information in our twenty-first-century culture. To retrieve a fact from a book, you find the book, open it, search for the fact, and read it verbatim; or you do the same thing within an electronic document. The same notion of search, find, and perceive applies also to video and audio
information stored on compact disks or on the Internet. But with retrieval from episodic memory we draw on an ancestral form of memory for events. It is a form of memory that served us well in the oral culture of our prehistoric past, yet continues to do so today in a world of writing and audiovisual literacy, all supported by massive external memory via the Internet. The key difference is that in a purely oral culture, all information available to people
had
to be remembered using only the mind without the benefits of writing and other kinds of external systems of storage.

Reconstructive retrieval uses the concepts of semantic memory to create a plausible recollection rather than a verbatim recollection. Activated concepts create an expectation of what should have occurred in the past rather than what necessarily did occur. The concepts and facts of semantic memory are organized in mental structures called schemas. Schemas play a critical role in our perception of the world by providing expectations. For example, your ability to perceive the objects in the room in which you are now sitting is aided by anticipating what objects normally would appear in such a place. If you were sitting in a coffee shop a schema that organizes your past experiences with coffee shops would become active. The tables and chairs that you expect to see would differ from those you might expect in your kitchen or dining room of your home.

Related concepts are organized into a schema, which can be thought of as a cluster of closely connected concepts. The plenitude of schemas that make up all that we know about the world is, in turn, organized in semantic memory, again according to their relationships. The end result is a massively interconnected network of concepts and schemas. Activate one concept and there will be a path that the activation can travel to light up, eventually, any other imaginable concept. For example, think of a cat. This concept is closely linked to the concept of mouse. It is likely that you would then think of a cat prowling for mice. The concept of barn might then be activated, because it is a place where a cat might prowl for mice, but other concepts could also come to mind, such as house or backyard. But, if barn did come to mind, then the train of thought might immediately move on to the schema for a farm, with a cluster of related concepts popping into awareness (e.g., horses, pigs, chickens, cows, and crops). The network organization of semantic memory produces the
associative structure of thought with which we are all intimately familiar. By following the associations through the network of semantic memory it is possible for virtually any thought, through some possibly lengthy chain of associations, to lead to any other thought.

Still, for any given schema and its associated concepts, some schemas are more closely linked than others. The farm schema is closely linked in semantic memory to the schema for the pristine countryside, for instance. By contrast, it is only very tangentially linked to a factory schema. By contrast, a factory is closely connected to the concept of city. It is difficult for people to think of farms as factories, because these ideas are so far apart in semantic memory. Similarly, the association of city and pollution is tightly wound in semantic memory, whereas farm and pollution are distant from each other. It tends to surprise us that laws are needed to prohibit pollution from large industrial farms that raise thousands of chickens or hogs. Whereas urban factories and pollution are immediately associated, farms are thought of as part of the wholesome countryside, not sources of toxic waste.

Our retrieval of past experiences from episodic memory passes through the organizational network of semantic memory. That is to say, these schemas provide expectations that help us to reconstruct past events. As part of their reconstruction, they can distort memories in order to conform to the expectations of the schemas active at the moment. Thus, schemas help us to reconstruct past events, but they do so by enabling us to fabricate how these events most likely unfolded. Reconstructive retrieval, then, refers to schema-guided construction of episodic memories that interpret, integrate, embellish, and otherwise alter the initial experience stored in episodic memory.
¹¹

A classic illustration of reconstructive retrieval comes from the downfall of America's thirty-seventh president, Richard M. Nixon. The testimony to Congress of John Dean, an assistant to the president, concerning the cover-up of a burglary at the Watergate Hotel, led to Nixon's resignation. Dean's testimony captivated Congress, the press, and the nation not only because of its factual content but also because of Dean's remarkably detailed recollections. Dean became known as the human tape recorder, given his confident recollection of conversations that had taken place weeks and months before. As it happened, Nixon had secret tapes of these same Oval Office conversations.
When the tapes became public, it was possible to compare Dean's actual conversations with his recollections, with the exception of a mysterious eighteen-minute gap in the tapes. Comparisons of the tapes with transcripts of Dean's testimony provided an extraordinary window on how episodic memory actually functions, not in a lab but in the Oval Office and Congress. Ulrich Neisser observed the following in his study of Dean's testimony:

Comparison with the transcript shows that hardly a word of Dean's account is true. Nixon did not say any of the things attributed to him here: He didn't ask Dean to sit down, he didn't say Haldeman had kept him posted,…he didn't say anything about Liddy or the indictments. Nor had Dean himself said the things he later describes himself as saying.
¹²

It was discovered that Dean's confidence in his recollections for details was unwarranted. Distortions showed that Dean hardly ever recalled the verbatim content of a given conversation, even though he was extremely confident that his recollection was veridical. In cases where his conversations were correctly recalled, it was more about the gist of the conversation rather than the details. Although Dean was truthful and accurate in conveying the gist of what happened, it turned out that “what seems to be specific in his memory actually depends on repeated episodes, rehearsed presentations, or overall impressions.”
¹³

Thus, Dean's recollections illustrate two important points about how schemas reconstruct episodic memories. The first is that the schema reconstructs the general idea or gist that is based on repetitions of essential elements or overall impressions. It is important to emphasize that Dean's testimony was factually accurate when considered at the level of gist. His recollections of the important facts were indeed correct, despite the small mistakes in the details. The second is that in recollecting an event, the schema richly embellishes it with details, as if it were a current event available for others to see and hear, without awareness that the details are fabricated. This is not an isolated example. Studies of eyewitness testimony have consistently found that the confidence of a witness is not predictive of the accuracy of recollection. One can be highly confident that a detail is correct, but still be dead wrong.

Thus, episodic memory is nothing like an audio-video disc that stores a verbatim record of a conversation or event. The record of human experiences can be altered at the time when information is initially encoded and stored in long-term memory. And it can also be changed in the process of reconstructing the event at the time of retrieval. How does the process of reconstructing an autobiographical experience unfold? For example, try to recollect your first day of school, your first kiss, your first day of your first job, your graduation from high school, the day of your wedding, or the birth of your first child. The process typically begins by retrieving the period of your life during which the event took place. People say to themselves, “when I lived in X” or “when I worked for Y.” This is simply a fact available in the semantic store of long-term memory rather than a detailed episodic memory. Once a general period is identified, the retrieval process turns to themes and moods evoked by the event even though the specific event is not yet concretized. The theme and mood of falling in love, for example, sets the stage for remembering general rather than specific events. Finally, concrete images come to mind as sensory reconstructions of a specific event. Even though the phenomenal experience is of a recollection of the who, what, when, and where of a specific event, these episodic memories are always supported by and integrated with the themes and general periods of life that depend on factual or general information from semantic memory.
¹⁴

A common trigger for recollecting our personal past is old photographs. Seeing oneself, family members, and friends in the places one used to live provides a retrieval cue for events from that period of life. In the process of reflecting on the picture and attempting to reconstruct long-forgotten events, the details can readily be mistakenly remembered, distorted, or totally fabricated without awareness of the error. For example, in one study, young adults were asked to recollect three events that took place at school during their childhood.
¹⁵
Based on a questionnaire completed earlier about their autobiographical experiences, two of these events were real, one that took place in grade five or six and another that went back a bit further to grade three or four. The third event, supposedly that took place in grade one or two, was actually false. This was an event that never actually happened, namely, the time a friend named Jane put slime in the teacher's desk. This pseudo-narrative was
made up by the researchers and read to the participants along with accounts of the two real events. For each event, there was a group photograph of the student's class from the appropriate grades. Starting with the account from grades five and six, along with the group photo, the participants tried to recall as much as possible about the event and judged the extent to which they felt they were reliving the event. Finally, they judged their confidence that the event had really occurred all those years ago. Strikingly, seventy percent of those who looked at the photo and heard the description actually recollected the slime incident, with as much vividness as the real events. Without the photo, just under half remembered this false recollection.

Reconstructive retrieval recruits the interpreter in the left hemisphere. Like a detective attempting to reconstruct a crime scene, the interpreter does the same with autobiographical experiences from the past. Details are inferred, a storyline is concocted, and a coherent episodic memory can be constructed from whole cloth. The role of the interpreter in reconstructive retrieval has been highlighted with studies of split-brain patients.
¹⁶
In one study, a series of pictures was presented that portrayed a story. The pictures were shown either to the silent right hemisphere or the interpreter of the left hemisphere. In doing so, it was discovered that the right hemisphere remembers the literal story as portrayed only by the pictures. However, the left hemisphere made inferences that went beyond the literal pictures. These inferences allowed the left hemisphere to remember the gist of the story with various details embellished using inference and logic rather than any real recollection. Because the interpreter seeks to make inferences even when they may not be warranted, the left hemisphere plays a large role in false memories. From neuroimages collected as a person recollects past events, Michael Gazzaniga concluded that “although both hemispheres are activated when recalling true items, the left hemisphere becomes more activated when it is experiencing false reports.”
¹⁷

Before considering how the brain appears to accomplish mental time travel, it is instructive to experience reconstructive retrieval first hand. Try to remember what you were doing at this time of day one year ago. Take a few moments to think about the events that transpired one year ago at this time. When we try to remember an event from long-term memory, usually a schema from long-term memory is activated, one that is relevant to the
circumstance in which you now find yourself and are trying to reconstruct. For example, suppose it is now 10 a.m. and you remember that one year ago you most likely would have been at work at that time. From this inference of where you would have been at such and such a time, the appropriate schema would become active, in this case a schema that summarizes the experiences from your working life, particularly those from the job you held a year ago. Or, if you instead inferred that a year ago at 10 a.m. you would have been in class, then a school- or college-related schema would jumpstart the recollection with expectations of what you were most likely doing at that time. With those expectations, reconstructive retrieval can begin to piece together and simulate an episode from the past. Notice that such a reconstructive process is much like solving a problem.

As another example, try to recall your tenth birthday party. Is your recollection accurate or is it, like John Dean's testimony, more of an amalgam of past experiences. Does it include experiences from birthday parties of friends rather than your own? Does it include the elements of past parties that occurred most often over the years of childhood rather than the specific elements of the tenth birthday party? Do your recollections include highly specific details that you were certain happened? Can you be certain, however, that they are not embellishments of the reconstructive process brought about by simulating the episode as if it were here and now?