 
| |
|
| |
|
| |
|
| |
The hippocampus, along with several closely interconnected adjacent brain structures, can be found just inside the left and right sides of the brain (called the temporal lobes) and toward the middle (or medial) area of the brain, a region neuroanatomists call the medial temporal lobe. Along with several other brain structures within this region, the hippocampus has traditionally been associated with memories for explicit facts (e.g., on Mondays, I have physics class) or specific events (e.g., I ate Cornflakes for breakfast this morning), which collectively are referred to as declarative memories. The anatomical borders of this essential learning circuit are not yet well defined, but most researchers now point to a broader hippocampal region, which includes not only the hippocampus, but also its overlaying cortical structures. When the hippocampal region is damaged by a stroke or other trauma, the most readily apparent symptom is anterograde amnesia, the loss of ability to store new declarative memories, despite the fact that all but the most recent pre-trauma memories remain intact. Because these older memories are not affected by hippocampal damage, it is generally believed that the hippocampus is not the final storage site for memory, but plays a crucial yet still not well understood role in the formation of new memories.
The inability of amnesics to store new information, while older memories remain intact, is familiar to those who have watched someone go through the early stage of Alzheimers Dementia, a disease whose initial pathology is characterized primarily by neuronal damage to the hippocampus and related structures.
Research in the neuropsychology of memory has traditionally focused on trying to characterize what amnesic patients can or cant do. Most of these studies have sought to identify learning tasks that are either profoundly impaired or else completely spared following hippocampal damage. This approach presumes that a declarative memory system is localized within the hippocampal region, so that any impairment or damage to the hippocampus would present itself as a deficit confined solely to declarative memory behaviors. A key challenge for this approach has been to precisely define what is, or is not, a declarative memory so as to allow accurate prediction of which abilities will be altered by hippocampal damage.
Neuropsychological studies of human memory impairments are generally limited to opportunistic encounters with patients who have sustained limited, but rarely precise or complete, damage to specific brain structures, such as the hippocampus. Only through research on animals can scientists impose the precise experimental controls required to understand brain function in detail. A major impediment to a more complete understanding of the brain mechanisms of memory has been the difficulty of assessing declarative memory in animals: One cant ask a rat if it remembers what it ate for breakfast. For this reason, our own research has emphasized learning tasks and theories of hippocampal function that facilitate effective comparison between animal and human learning.
In contrast to previous research, which focused on memory abilities that are either completely eliminated or completely spared by hippocampal damage, we have paid attention to borderline tasks that are just barely solvable by animals and humans with hippocampal damage. Among these tasks are certain complex forms of classical Pavlovian conditioning in which an animal learns that previously insignificant cues, like bells or lights, are predictive of significant future events, like the arrival of food or shock.
Originally, many researchers did not believe that the hippocampus played any role in classical conditioning or other simple forms of associative learning. Indeed, removal of the hippocampal region does not generally cause an impairment in the acquisition of single cue-to-outcome associations. However, when more than one cue is involved, or when there are multiple phases of learning, the hippocampus does appear to play an important role. Tasks that involve several successive stages of learning have been especially useful because they allow us to how experience with an initial problem can alter or bias a learners strategy for solving subsequent problems. Hippocampal-damaged animals and people often behave normally on an initial phase of training. Subtle differences in subsequent phases of training, however, suggest that hippocampal-damaged subjects use learning strategies which are different from those used by normal control subjects.
These alterations in performance based on prior experience are examples of transfer generalization, because general principles acquired from prior experience are transferred to a new task. The similarities or differences between initial and subsequent stages of training can cause either a facilitation (positive transfer) or impairment (negative transfer) to later learning.
Although our focus on transfer generalization in amnesic humans and hippocampal-lesioned animals has been motivated by seeking clues to the functional significance of the hippocampus, impaired transfer performance on real-world skills is a hallmark of many forms of brain disorder, especially amnesia and Alzheimers Dementia. While memory-impaired patients often show initial promise in rehabilitation training, their performance drops precipitously when small and seemingly insignificant aspects of the training situation are changed. That is, these memory-impaired patients fail to show appropriate positive transfer from a training task to a similar, but subtly different, real-world task. As a result, most attempts at rehabilitation for seriously memory-impaired patients have met with minimal success.