In the previous section, you learned about two types of explicit memory: semantic and episodic. The hippocampus seems to be especially important for the encoding, storing, and retrieving of episodic memories, but less so for semantic memories.
During the 1930s and 1940s, a neurosurgeon by the name of Wilder Penfield and his colleagues performed research that, at the time, seemed to indicate that engrams for episodic memories were located in a specific part of the cerebral cortex. Penfield's goal was to link particular mental functions with activity in particular cortical areas. In order to achieve this goal, he and his colleagues activated areas throughout the cerebral cortex in patients undergoing neurosurgery. They found that, when areas within the temporal lobes of the cortex were stimulated, a small proportion of their patients (less than 4%) reported experiences that, to the patients, seemed like the retrieval of episodic memories from long ago. These experiences were full of vivid perceptual details and the patients felt as if they were actually reliving the events:
One of Penfield’s patients was a young woman. As the stimulating electrode touched a spot on her temporal lobe, she cried out: “I think I heard a mother calling her little boy somewhere. It seemed to be something that happened years ago ... in the neighborhood where I live.” Then the electrode was moved a little and she said: “I hear voices. It is late at night, around the carnival somewhere, some sort of travelling circus. I just saw lots of big wagons that they use to haul animals in.“ (Blakemore, 1977; quoted in Baddeley, 2004, p. 114)
Penfield and his colleagues believed that they were accessing episodic memories of real life events — memories containing all the perceptual details of the original events. Based on these results, they concluded the following about episodic memories:
- Episodic memories are stored in the temporal lobes of the cerebral cortex.
- Episodic memories are encoded as detailed perceptual reproductions of the original events.
This research provided what seemed to be conclusive evidence for what we will call the reproduction theory of explicit memories. This theory states that, for each episode in our lives, we encode and store an exact reproduction — one that is accurate to the smallest perceptual details.
The reproduction theory of explicit memories, however, is not supported by any other scientific evidence. In fact, the best evidence available supports the claim that we forget almost everything that has ever happened to us. Two theories of forgetting help to explain why this happens:
- Decay theory: memory traces in both the sensory and short-term stores decay (disappear) rapidly because of biological changes.
- Displacement theory: memory traces in the short-term store are displaced (pushed out) rapidly because of its small capacity.
What ends up in the long-term store from any single episode in our lives is a small fraction of the information initially processed in sensory memory, as shown in Figure 1 (from Section 6-9).
Figure 1. The amount of information transferred from one memory store to another during a single life episode
If Penfield and his colleagues weren't retrieving episodic memories, then what were their patients experiencing? In order to answer this question, there are two things we need to keep in mind. First, only a small proportion of patients reported these experiences. If episodic memory codes are located in the temporal lobes, then we would have expected that most patients would have retrieved vivid episodic memories when their temporal lobes were stimulated. Second, activation of sensory areas of the brain often causes hallucinations — vivid perceptual experiences of objects or events that are not actually there. In the case of Penfield's patients, it is likely that activation of their temporal lobes caused them to experience complex hallucinations coupled with the illusory feeling that the hallucinated events had actually happened to them. In fact, it now is generally agreed that these patients were not experiencing memories of life events. Most research suggests that there is no specific area of the cortex that stores episodic memory codes. Rather, episodic memory codes probably are distributed throughout the cerebral cortex; and they also depend on activation of structures in the limbic system (such as the hippocampus), at least for memories that formed within about the past 10 years (Smith & Squire, 2009).
If episodic memory codes are not reproductions of past events — that is, if most of the perceptual details decay or are otherwise lost soon after the events occur — then why do we often remember many perceptual details when retrieving these memories? In fact, most of us seem able to visualize in detail the locations in which many events took place as well as the people who were there. How is this possible if the memory codes don't actually include all these details?
Reconstruction of Episodic Memories
When we encode in working memory an event that is to become an episodic memory, we attend to and elaborate only parts of the event. For example, let’s say that you are watching a frightening movie about a serial killer. During the movie, you are much more likely to attend to and encode details relevant to the main theme of the movie — such as the large knife that the killer has used to mutilate his victims — than you are to attend to and encode details that were irrelevant to this theme — such as the fact that the room in which he has imprisoned his victims contains a blanket. Later on, if asked whether a blanket was present in the room, you will not remember this detail if you did not attend to it. When asked to describe a particular scene from the movie, you will piece together the bits of information you actually did attend to and encode, and construct something that contains many gaps (because most of the details were not encoded). In order to retrieve a complete memory — one that can be consciously recalled and described — you must fill in these gaps with what you think probably happened. This process of "filling in the gaps," referred to as reconstruction, is performed unconsciously (that is, you are not aware that you are filling in the gaps). In general, the memory codes that form the basis of episodic memories consist of only a small number of “memory fragments” — bits of encoded information that, together, contain only a few details of the original event. The memory that is consciously retrieved, however, consists of much more than these memory fragments.
In trying to understand the reconstruction theory of episodic memories, it may help to think of the retrieval of an episodic memory as being similar to the reconstruction of a complete dinosaur skeleton from a small and incomplete set of fossilized bone fragments:
For the paleontologist, the bone chips that are recovered on an archeological dig and the dinosaur that is ultimately reconstructed from them are not the same thing: the full blown dinosaur is constructed by combining the bone chips with other available fragments, in accordance with general knowledge of how the complete dinosaur should appear. Similarly, for the rememberer, the engram (the stored fragment of an episode) and the memory (the subjective experience of recollecting a past event) are not the same thing. The stored fragments contribute to the conscious experience of remembering, but they are only part of it. (Schacter, 1996, pp. 69-70)
When a set of fossilized dinosaur bones is found, it typically consists of only a small portion of the original skeleton. In order to reconstruct a complete skeleton from the small number of bones, paleontologists use their general knowledge of what dinosaurs (probably) looked like — knowledge that they have accumulated since 1841, when the first dinosaur fossils ( which were found in 1822) were recognized as being from species that had existed ages before. Because we now have so much knowledge about dinosaurs, our reconstructed versions probably are very accurate. Nevertheless, in earlier years, the builders of reconstructed skeletons made many mistakes because their knowledge of the original creatures was limited by the fact that the last dinosaurs died about 66 million years ago. In a similar way, we use our general knowledge of what usually happens, or what we think must have happened, when we reconstruct an episodic memory from the small number of encoded and stored memory fragments.
When an episodic-memory code is activated by a retrieval cue, we use the knowledge that we have accumulated over our lifetimes to fill in the large gaps in the memory code (see Figure 2).
Figure 2. Reconstruction of an explicit memory
This set of reconstructive processes leads to a remembrance that is very much like a reconstructed dinosaur: it probably is accurate in broad terms, but is wrong in at least some details. Furthermore, reconstruction occurs unconsciously, so we are unaware that we have added information to the retrieved memory — information that is not included in the memory code. In fact, Loftus and Ketcham (1991) concluded that the process of memory reconstruction introduces inaccuracies into each and every episodic memory that we retrieve:
Memories don’t just fade, as the old saying would have us believe; they also grow. What fades is the initial perception, the actual experience of the events. But every time we recall an event, we must reconstruct the memory, and with each recollection the memory may be changed — colored by succeeding events, other people’s recollections or suggestions, increased understanding, or a new context…. Truth and reality, when seen through the filter of our memories, are not objective facts but subjective, interpretive realities. We interpret the past, correcting ourselves, adding bits and pieces, deleting uncomplimentary or disturbing recollections, sweeping, dusting, tidying things up. Thus our representation of the past takes on a living, shifting reality; it is not fixed and immutable [unchangeable], … but a living thing that changes shape, expands, shrinks, and expands again, an amoebalike creature with powers to make us laugh, cry, and clench our fists. Enormous powers — powers even to make us believe in something that never happened. (p. 20)
Let’s take a real-life example of the retrieval of an episodic memory that shows the types of inaccuracies introduced when reconstructing a complete memory from a small amount of encoded information. In 1968, Jack Hamilton, a pitcher for the California Angels, threw a fast ball that hit Tony Conigliaro, an outfielder for the Boston Red Sox, on the left side of his face. Years later, Hamilton recalled the event:
“I never hit a guy that hard.... He [Conigliaro] went right down, he just collapsed.... I’ve had to live with it, I think about it a lot,” Hamilton, now fifty-one years old and the owner of a Midwestern restaurant chain, told The New York Times after hearing about Conigliaro’s death [in 1990]. “Watching baseball on TV, anytime a guy gets hit, I think about it. It was like the sixth inning when it happened. I think the score was 2-1, and he was the eighth hitter in the batting order. With the pitcher up next, I had no reason to throw at him.” It was a day game, Hamilton recalled, because he remembered visiting Conigliaro in the hospital later that afternoon. After the accident he remembered wondering whether or not to return to Fenway Park later that year for another series of games; eventually he decided to make the trip. (Loftus & Ketcham, 1994, p. 76)
Although this reconstructed memory had one element of truth to it (Conigliaro was hit by a baseball pitched by Hamilton during a game played in Boston), the rest of it was almost completely inaccurate. Conigliaro was hit by Hamilton's pitch in the fourth inning, not the sixth; it happened during the final road trip to Boston, not earlier in the season; the game was at night, not during the day; the score was 0-0, not 2-1; Conigliaro was the sixth batter in the batting order, not the eighth and, therefore, the pitcher would not have been up next. Hamilton had recalled this memory hundreds, perhaps even thousands, of times over the 22 years since it had first occurred. During each reconstruction, he filled in the gaps in his memory code with more and more inaccurate information until most of the details of the story had changed. Nevertheless, the general theme of the story (that Tony Conigliaro had been hit hard by one of Hamilton’s pitches at Fenway Park) stayed the same.
Thus, the reconstruction of episodic memories (something similar probably happens with semantic memories, too) is one reason why we forget them, especially those episodic memories that were initially encoded and stored many years ago. In the rest of Section 6, we will look more closely at why we forget episodic and semantic memories.
Study Questions for Section 6-13
- What did Penfield and his colleagues discover when activating parts of the cerebral cortex in the temporal lobes?
- What did Penfield and his colleagues conclude about the storing of episodic-memory codes?
- What are the decay and displacements theories of forgetting, and how do they explain forgetting from sensory memory and short-term memory?
- What do modern memory researchers believe regarding the location in the brain of episodic memory codes?
- What do modern neurologists and memory researchers think happened when Penfield stimulated the temporal lobes of his patients?
- What is the reconstruction theory of explicit memories?
- Ar what level of awareness does the reconstruction of episodic memories take place?
- How does the reconstruction theory explain the forgetting of the details that make up our episodic memories?
- What does the reconstruction theory imply about the accuracy of our episodic memories?
- What is likely to happen to the accuracy of a episodic memory that is retrieved frequently over a number of years? Why?