MEMORY RESEARCH REVEALS WHERE RECOLLECTION RESIDES
For immediate release.
NEWS RELEASE NR-24-06 (10/15/06). For more information, please contact Sara Harris at (202) 962-4000 or sharris@sfn.org.
MEMORY RESEARCH REVEALS WHERE RECOLLECTION RESIDES
ATLANTA, October 15, 2006 - The location and mechanisms of recollection may reside in a specific set of brain cells, according to new reports. Information gained from another study might provide an essential step in developing therapies to protect memory from aging or disease.
Scientists also have identified proteins responsible for strengthening synapses -- the gap where information is transferred between neurons -- which is critical for learning and memory. And a new theory uncovers how our brain stores memory by forming an internal model of our self.
Understanding how these cells and proteins interact in the brain may lead to new strategies for preserving memory in the face of aging or disease.
Individual neurons in a seahorse-shaped area deep in the brain called the hippocampus represent particular types of different information that make up recollection, such as the name of the person or object, their context, and sequence of related events, says Howard Eichenbaum, PhD, of Boston University.
Research by Eichenbaum and his colleagues shows that animals can remember all of these types of information. They also find that selective removal of the hippocampus eliminates these capacities, and that, correspondingly, individual neurons in the hippocampus represent particular types of information that support these capacities. Also, recent findings from several other laboratories have suggested a brain circuitry supports the hippocampus in these functions, he adds.
"These findings suggest the existence of two streams of information processing between the cerebral cortex and the hippocampus," Eichenbaum says. "One stream may support our sense of familiarity for particular people or objects, whereas the other stream adds information about where the objects were experienced. During recollection, the hippocampus orchestrates the reactivation of the cortical representations for where the event happened, and other information, giving rise to the subjective experience of recollection."
Recollection involves reaching beyond mere familiarity with a person or item to retrieve qualitative information about where and when it was experienced, what happened before and after, and related experiences. "This capacity is supported by widespread cortical areas interacting with the medial temporal lobe, and in particular, with the hippocampus," says Eichenbaum. His work supports the argument that the sea horse-shaped brain area called the hippocampus serves recollection by performing memory-binding operations that are critical to each of the characteristic features of recollection.
Eichenbaum is concerned with the questions: What is conscious recollection? Do animals have it? How is it accomplished by the brain? One way to appreciate what is involved in recollection, he says, is to consider the distinction between a vivid reminiscence and something less, a sense of familiarity with a particular person or object without recollection.
"We have all been in the situation where we have met someone who seems highly familiar but we cannot recall who they are or why we know them," he says. "Sometimes, we just give up and say, 'Don't I know you?' " Alternatively, he adds, we avoid embarrassment when some clue or mental searching helps us suddenly retrieve the name, where we met before, and the circumstances of that meeting. Our most vivid recollections involve replaying the entire episode in which we met the person, and that memory might lead to remembering additional encounters or other related information.
"These considerations tell us that, whereas familiarity is defined as the ability to identify a current stimulus, such as a person or object, as previously experienced, recollection involves the additional recovery of information about associated items, like a name, for example, the context, such as the place in which the person or object was experienced, the flow of events that composes the experience, and additional related experiences," he says.
Long-lasting changes in the strength of synaptic transmission underlie learning and memory. Researchers have found that long-term potentiation (LTP) -- a phenomenon in which brief, repetitive activity causes a long-lasting strengthening of synaptic transmission -- involves the rapid activity-dependent trafficking of glutamate receptors to the synapse.
Another study suggests that memories are supported by a broad, distributed network of cortical regions, which could provide an important step in developing therapies to protect our memory from aging and disease.
"Previously, a lot of attention has been paid to how memories are initially encoded and stored, says Rui Costa, DVM, PhD, at the Duke University Medical Center, who collaborated with Paul Frankland, PhD, at the Hospital for Sick Children in Toronto. "For example, studies have begun to successfully identify the molecular machinery -- the receptors, ion channels, second messengers and transcription factors -- that is engaged in the first seconds, minutes and hours following learning."
The activation of this machinery is required for the restructuring of existing synaptic connections, as well as the growth of new synaptic connections, underlying the initial stabilization of the memory trace in brain networks. "However, recent studies have highlighted how reactivation of an existing memory trace opens up a fresh window for new plasticity," Costa says. "Memory reactivation may occur in either online situations, such as during explicit recall, or offline situations, like during sleep, and may occur over and over again."
Each reactivation provides an opportunity for modifying a memory trace, or incorporating new information. Repetitive reactivations may eventually lead to dramatic changes in the way memories are organized at the systems level.
Because our memories of everyday life -- of people, places and events -- define who we are, the loss of these memories in old age or in disorders, such as Alzheimer's disease, is devastating. In humans, these types of memory initially depend on the medial temporal lobe system, including the hippocampus. However, as these memories mature, they are thought to become increasingly dependent on other brain regions, such as the cortex.
The recent work used brain mapping, mouse-genetic and pharmacological approaches to understand how new memories in the hippocampus are transformed into lifelong, or remote, memories in cortical networks. "In particular our studies show that recall of remote memories activates multiple cortical brain regions," says Costa, "suggesting that these memories are supported by a broad, distributed network of cortical regions.
"Understanding how the topology of this memory network affects its function is an essential step in developing of therapeutic strategies designed to protect our memories from the ravages of disease or normal aging."
Research has identified proteins that play a critical role in long-term potentiation (LTP), which strengthens synapses and enhances learning and memory.
Roger Nicoll, MD, University of California, San Francisco, and colleagues study the cellular and molecular mechanisms underlying learning and memory in the mammalian brain. They use a combination of electrophysiological and molecular techniques to reveal how connections become stronger with use over time, the mechanism that allows the brain to process, store and recall information.
One key player is glutamate, the neurotransmitter released from synapses, which acts primarily on two types of receptors: AMPA receptors and NMDA receptors. LTP is induced by NMDA receptor activation, but the receptor site where LTP is expressed has been more problematic, says Nicoll.
LTP involves the rapid activity-dependent trafficking of glutamate receptors to the synapse. "Much of our current work is focused on how activity controls the trafficking of this receptor complex and how the increase in synaptic strength during LTP is stabilized and maintained," he says.
Recent work suggests that LTP is due to a modification involving the rapid recruitment of AMPA receptors to the synapse. "We have found that stargazin, the mutated protein in the ataxic and epileptic mouse stargazer, is necessary both for the expression of surface AMPA receptors in certain cerebellar cells and for their targeting to the synapse," he says.
Stargazin is a small membrane protein, and is a member of a family of proteins referred to as transmembrane AMPAR regulatory proteins (TARPs). Varieties of these proteins are found throughout the brain, says Nicoll. TARPs bind to AMPA receptors as well as to the synaptic scaffolding protein. It is via this interaction that the synaptic scaffolding protein anchors AMPA receptors to the synapse, says Nicoll.
In other research, Japanese scientists have uncovered how our brain stores memory by forming an internal model of our self.
The "neuronal machine" of the cerebellum, an area of the brain in the back of the head that controls motion and motor learning, are able to learn on the basis of the unique memory mechanism incorporated in its neuronal circuitry, says Masao Ito, MD, PhD, at the RIKEN Brain Science Institute in Japan.
With this learning capability, the cerebellum forms an internal model that copies the dynamic properties of a physical part. "The machinery is supposed to form an internal model that even copies a mental representation in the cerebral cortex," he says. The general concept is that the cerebellum provides models of physical parts and mental representations, which are essential in the control of physical and mental activities.
The major learning process in the neuronal machine of the cerebellum is long-term depression (LTD), the weakening of a neuronal synapse that lasts from hours to days. It occurs in a unique structure of the cerebellum, says Ito. About one hundred thousand "parallel fibers" and one "climbing fiber" converge onto an output neuron of the cerebellum, the Purkinje cell, a class of GABAergic neuron located in the cerebellar cortex. A conjunctive activation of these two types of input leads to a persistent decrease in the efficacy of transmission from activated parallel fibers to a Purkinje cell, he says.
In the past decade, extensive analyses using pharmacological and genetic methods revealed more than 30 different molecules involved in the induction of LTD. Complex chains of chemical reactions occur in the microspace of dendritic spines of Purkinje cells. These reactions lead to the removal of glutamate receptors from the synaptic membrane.
"A spatial distribution pattern of LTD-expressing synapses along Purkinje cell dendrites may represent a memory trace," says Ito. "In a practical sense, these findings provide us useful pharmacological and genetic means to manipulate LTD."
LTD induction driven by error signals provides a mechanism for forming an internal model that copies dynamic characteristics of physical parts, such as fingers, hands, arms or legs. Deviations between the performance of an internal model and that of an actual physical part lead to the generation of error signals, which act to modify the internal model until deviations are minimized. "Referring to an internal model in the cerebellum, we are able to perform precise movements even without checking sensory feedback," says Ito. "For example, we can point at a target accurately even with the eyes closed."
The learning capability of the cerebellum is demonstrated in various forms of motor learning such as adaptations occurring in eye movements, eye blink, and various hand and arm movements. Motor systems in the spinal cord, brainstem or cerebral cortex controlling these movements are connected to certain regions of the cerebellum, where internal models are formed. Ito collected evidence for internal models or motor learning using three methods.
Recording from Purkinje cells during eye movement reveals that their electrical activities represent an internal model for an eyeball, he says. A computer simulation of an internal model reproduces the learning capability of the cerebellum. LTD deprivation in a cerebellar region by pharmacological or genetic means leads to the loss of relevant learning capability.
In acting out a thought, an area of the cerebral cortex associated with conscious activities interplays with a cerebellar internal model formed in the unconscious domain. This hypothesis has been and is being tested by imaging studies of human brain activities and by a precise tracing of neuronal connections between the cerebellum and the cerebrum.
"On the basis of the internal model control, we foresee that implicit learning is a process by which an internal model of the self is formed and reformed in the unconscious domain of the brain, the cerebellum," says Ito. "Questions of how we move and even think unconsciously and how abnormal mental activities emerge in schizophrenia and autism can find answers in this hypothesis."
Other research looking into the role of the prefrontal cortex (PFC) in working memory has discovered a clearer understanding of the way memory is stored.
Working memory is the active retention of information, its manipulation and transformation, and its use in guiding behavior. It is central to systems and cognitive neuroscience studies of high-level cognition because it underlies many complex behaviors, such as planning, problem solving, and, in humans, many aspects of language function.
Individual differences in working memory capacity predict such abilities and outcomes as general fluid intelligence, reading proficiency, standardized test performance, and even lifelong income. Despite its centrality to cognition, does one or more working memory systems exist in the brain? asks Bradley Postle, PhD, of the University of Wisconsin-Madison. The answers may one day lead to treatments for protecting or restoring memory.
"For years, the received wisdom has been that the prefrontal cortex (PFC) is critical for all aspects of working memory, including one of its core functions, the short-term storage of information," says Postle. "Recent developments, however, require important modifications to our understanding of working memory storage."
One, he says, is that successful storage depends on delay-period activity in nonPFC areas traditionally classified as sensory, associational, and motoric. A second is that, for even the simplest tasks that would seem to require little more than the retention of information across a brief delay period, the functions of PFC are much more complex than just the storage of sensory information. Current research in this area draws from monkey electrophysiology, human neuroimaging and transcranial magnetic stimulation, and computational modeling.
Memory for visual motion in the monkey serves as a case study of working memory-related function of sensory systems. Psychophysical studies of the effects of delay-period masking reveal the remarkable fidelity of the visual memory code and the temporal dynamics of its retention and transformation, says Postle. Electrical microstimulation and recordings characterize the distributed nature of the processes underlying these functions.
Neuroimaging and repetitive transcranial magnetic stimulation studies in humans suggest that the principles underlying the storage of visual motion information in the monkey generalize to human working memory for many types of information, including linguistic. "Neuroimaging studies have also begun to address the question, 'If not storage, what are alternative explanations of delay-period activity in PFC?,' and highlight a role for PFC is the control of inference," he says.
Another important role for PFC is the integration of cognitive with motivational information to guide behavior, Postle says. "This is seen in studies that manipulate the value of reward delivered to a monkey for successful performance of a spatial delayed-response task. Trials featuring a preferred reward are associated with better behavioral performance and enhanced working memory-related activity in the PFC.
"Although the work suggests that PFC supports multiple functions, most computational spiking network models of PFC have focused on single cognitive functions," he says. "Is the PFC organized in a modular fashion, with different functions each implemented by a different network?"
Recent modeling work suggests an alternative: the PFC may operate via the rapid modulation of single networks from one set of computations to another, as circumstances require. This new understanding of working memory in the PFC could lead to improved therapies to combat disorders that degrade memory, such as Alzheimer's.
