M-I Neurosciences Dr. Thomas M. Reeves

Learning and Objectives: 1. Describe the current understanding of the areas of the involved with memory and retrieval. 2. Describe the current understanding of the areas of the brain involved in the learning of motor and perceptual skills. 3. Describe selective neuropathologies affecting learning and memory.

Introduction

In recent decades, clinical and experimental observations have enabled considerable progress in our understanding of the nature of learning and memory. A large body of knowledge now exists which documents this evolving story, in both fact and theory, of how are established and retrieved, and how these processes may fail, for example with age, disease or injury. The material in this syllabus summarizes key aspects of the subject area of learning and memory, emphasizing the growing appreciation for distinct types of memory systems implemented in different anatomical regions of the brain. After a careful reading of this material, the student should possess a conceptual framework to better understand the methods and findings in this area.

In a broad sense, learning is the process by which experience changes our nervous system and hence our behaviors. Thus, learning is one way in which the organism adapts to its environment, which itself may be in flux. These learning episodes may range from extremely simple to extraordinarily complex. Figure 1 shows that a basic distinction can be made between associative vs. nonassociative learning. As the name implies, associative learning is a process whereby new associations are established between experiences. In this way, stimuli are associated with responses, events with other events, ideas with other ideas, objects with other objects, and so on. In fact, the capacity and applications of associative learning are evidently infinite. At an intuitive level, when most people think of 'learning and memory' they are referring to the associative forms. However, the nervous system also reacts to certain environmental stimuli in nonassociative contexts. For example, through the process of habituation, we learn to ignore monotonous stimuli (e.g., we forget we are wearing the hat), and through sensitization a single stimulus can lead to a generalized augmented responsiveness (e.g., after hearing a gunshot, we are startled by a slamming door). While the study of nonassociative learning has lead to important mechanistic insights, particularly in cellular learning in simple invertebrate systems, the following material will focus on associative memory.

Learning Memory

Nonassociative Associative Short-Term Long-Term

Habituation Sensitization Declarative Nondeclarative Figure 1.

By definition, learning refers to a process of acquisition of information, and memory to the storage or retrieval of information. Figure 1 shows that memory is also divided into subtypes, namely short-term memory (STM) and long-term memory (LTM). This concept, of distinct memory systems with different time frames of operation, has proven quite useful in explaining experimental observations. It is now known that STM and LTM reflect the operation of different neural systems and processes. STM can be considered as having a limited storage capacity (seven, plus or minus two items) that "decay" and become inaccessible after a relatively brief interval (estimates range from 12 to 30 seconds). In addition to decay, loss of information from STM can occur by interference when new information displaces older information. Information can be maintained in STM for relatively long periods if ‘rehearsal’ is used, for example by mentally (subvocally) repeating the information to be maintained. In many cases, the reason one wishes to maintain information in STM is to allow time for it to be encoded into the LTM, and thus become more permanently available. LTM can store a vast quantity of information for very long periods of time. Figure 2.

A description of memory as a two-part system (STM, LTM) is an oversimplification, because research in this area has verified the operation of multiple memory subsystems along the temporal dimension. Figure 2 illustrates the distinctions in these memory systems, beginning with memory registers that maintain sensory information for very brief intervals, which may be further processed in STM, where the information may, in turn, be consolidated into LTM. Storage and retrieval at each stage is subject to information loss () due to the processes of interference (competing information) and decay (reflecting intrinsic limitations of neural systems). Mechanisms of the sensory registers are modality specific: the visual register is termed ‘iconic’ memory (sometimes called 'eidetic' memory), with visual traces persisting for ~ 250 milliseconds, and the auditory register is termed ‘echoic’ memory, and persists for about 3 seconds. Finally, modern cognitive neuroscientists are increasingly using the term ‘,’ as a concept closely allied to STM, but expanded to include the contents of consciousness at any one time, and also described as an active memory store in which information can be manipulated.

Declarative vs. Nondeclarative Memory

Of vital importance, in contemporary concepts of learning and memory, is the distinction between declarative and nondeclarative forms of memory. Declarative memory is the type of memory usually referred to when the term ‘memory’ or ‘remembering’ is used. With experience, each person accumulates a vast store of facts, words, names (of people and things), memories of places, events, concepts, rules, laws and customs; the list Figure 3. is endless. These are all examples of declarative memory. A useful is that we can declare the contents of declarative memory. In contrast, it is difficult, if not impossible, to specify the contents of nondeclarative memory, which is also referred to as "procedural" memory. The stored information that allows one to execute skilled motor acts, to play a musical instrument, or to perform the articulatory movements of speech, are all examples of . In addition to motor activities, procedural memory also encompasses perceptual skills (e.g., learning to spot a motionless animal in the woods) and some cognitive skills (e.g., learning how to learn). Also subsumed under nondeclarative memory are forms of learning involving or learning with a strong emotional component, which is known to involve of the in the . Returning to our mnemonic, one cannot declare the contents of procedural memory. For example, it would be inadvisable to try to learn how to swim by reading a book about swimming. Only by engaging in the motor act (practicing the procedure) can this form of learning effectively occur.

Learning theorists have further subdivided declarative memory into semantic and (Figure 3). Semantic is memory for facts or meaning, while episodic refers to for events or episodes that occur in a given place and time. There is clinical evidence that the semantic/episodic distinction is based on different processes implemented in different brain regions. For example, patients with frontal lobe lesions often exhibit "source ," with selective inability to remember when and where events occurred. Also, patients with Alzheimer's Disease often show specific disabilities in episodic memory during the initial phases of the disease process. It is also notable that hippocampal damage, early in childhood, leads to more severe deficits affecting episodic memory than (Vargha-Khadem et al, 2001).

Neuroanatomy of Memory

A major theme, of modern learning and memory theory, is that different memory modalities, declarative and nondeclarative, are more than merely descriptions of behaviors: they represent the operation of different anatomical brain regions (neurological substrates). Accordingly, they are differentially affected by ageing, disease, surgeries, or (in animal models) experimental manipulations. In fact, our knowledge of the anatomical segregation of memory function within the brain, has been provided in large measure by observing how the system fails when a portion of the brain is removed surgically (for example, to minimize intractable epilepsy), or is suppressed by natural causes (e.g., stroke). In addition, technical advances in research methods have permitted a greater understanding of CNS involvement in learning and memory. These include: 1. the development of primate models of human amnesia 2. the development of advanced brain imaging techniques (e.g., MRI, PET) -MRI (magnetic resonance imaging) has allowed the localization of anatomical lesions in humans which can be associated with memory deficits -PET (positron emission tomography) has identified areas of the brain activated during storage and retrieval of memories

A failure of memory is called amnesia. In cases where the cause of amnesia is known (injury, surgery, disease, etc.), it is helpful to distinguish between and . Anterograde amnesia refers to an inability to form new memories after the trauma (event precipitating the amnesia). Retrograde amnesia refers to an inability to memories established prior to the trauma. As explained in the following sections, a systematic examination of the pattern of anterograde and retrograde resulting from various neurotraumatic events (planned experimental or naturally occurring in the clinic) has contributed to our current understanding of the brain structures involved in the formation and storage of memories (refer to Table 1, below). Anatomical Correlates: Declarative Memory

The most compelling evidence for more than one memory system comes from studies of amnesics, and the current interest in the memory of amnesics can be traced to a single clinical patient known as H.M. The story of H.M. starts in 1953, when at 27 years of age he underwent surgery by Dr. William Scoville to treat severe temporal lobe epilepsy. The surgery resulted in bilateral damage to the temporal lobes, involving the and some surrounding cortical structures (see Figure 4). On recovery, Scoville was surprised to discover that the surgery had produced a severe anterograde (forward) amnesia. H.M. showed an almost total inability to remember anything that occurred after the operation. He was unable to learn the location of his room; he was unable to find his way to the bathroom; he was unable to learn the names of visitors, even those who had been to see him frequently over the years. He repeatedly re-read the same papers and repeated tasks over and over without giving any evidence of having done them before. Dr. Scoville's associate, the psychologist Dr. , did not find any obvious personality changes. On standard intelligence tests, H.M. still tested normal (approx. 108). What was possibly more surprising, however, was that H.M. was able to learn some tasks. One example was a test of mirror drawing which tests motor control (a procedural Figure 4. memory task). On this, H.M. showed improvement over repeated testing, in spite of the fact that he never remembered having done the test before. H.M. was also able to repeat a list of words (providing that he was not distracted), and he had a normal recall of numbers in sequence (known as a digit span test). In other words, H.M. appeared to show relatively normal short term memory of information, but was totally lacking in the ability to retain new information beyond a short length of time. The case of H.M. demonstrates different aspects of memory. First, the test of mirror drawing showed that , one form of procedural memory, is apparently separate from the declarative (episodic) Table 1.: Some Clinical Causes of Amnesia Causes Examples Site of damage Vascular (occlusion of both Patient R.B. Bilateral medial temporal lobe, the hippocampus in posterior cerebral arteries) particular Medial bilaterally (hippocampus and other Tumors --- related structures if tumor is large enough) Trauma Patient N.A. Bilateral medial temporal lobe Surgery Patient H.M. Bilateral medial temporal lobe Infections Herpes simplex enceph. Bilateral medial temporal lobe Vitamin B1 deficiency Korsakoff’s syndrome Medial thalamus and mamillary bodies Electroconvulsive therapy (ECT --- Uncertain for depression memory of the test experience. Second, declarative memory as shown by the word list and digit span tests was retained in the short term but not at length, suggesting separate mechanisms for short term and long term memory.

The Role of the Medial Temporal Lobe Medial temporal lesions

in primates (Squire et Subsequently, lesion studies began al). Ablations: H, to be conducted in monkeys to identify the hippocampus; EC, anatomical substrates of the amnesia entorhinal ctx; PHC, created by this type of lesion, and to better perihippocampal ctx; characterize the nature of deficits PRC, perirhinal ctx. correlated with larger and smaller ablations of medial temporal lobe areas. Initial studies focused on duplicating the lesions experienced by H.M. and continuing studies have looked at the effects of removal of different components of the original larger lesion. A series of experiments by and colleagues (see Figure 5) indicated that lesions of hippocampus alone still were accompanied by memory Figure 5. deficits, but these deficits were not as severe as for the more complete lesions. Further studies indicated that the additional deficits accompanying larger lesions were due to cortical (rather than amygdala) damage, although amygdala damage has been shown to produce a significant deficit in value or reward based learning (stimulus- reward training), including conditioned fear and other forms of affective memory. Of these cortical areas, those adjacent to the entorhinal cortex are responsible for nearly 2/3 of the input to the entorhinal cortex, which provides, in turn, the major source of projections to the hippocampus and , supporting the idea that damage to these areas could impair memory function.

On the basis of these, and related, observations, some conclusions may be drawn regarding the role of the medial temporal structures in the process of memory formation. As evidenced by Figure 6. patient HM, vast amounts of (pre- surgery) memory remained intact, even in the absence of the hippocampi bilaterally. These memories were manifestly not stored in the hippocampus. Rather, most evidence now points to a complex set of operations that involves the hippocampus in the process of consolidating new memories into more permanent long term storage in neocortical association areas. This is a deliberate oversimplification of an extraordinarily complex process, but as an initial conceptual model does prove useful in explaining a number of observations, including clinical symptoms and experimental results. Moreover, the pattern of neuronal pathways which connect the association neocortex with the medial temporal lobe structures is consistent with a special role of the hippocampus in the process of memory formation. Figure 6 summarizes key aspects of this connectivity, showing the location of the entorhinal cortex (EC) in midsagittal and coronal perspective (top). The EC is of pivotal importance, anatomically and functionally, as it the primary gateway for cortical information to enter and exit the hippocampus. Note that the EC communicates with widespread neocortical association areas via bidirectional pathways through the parahippocampal and perirhinal cortices. The EC, in turn, generates a substantial projection to the dentate gyrus of the hippocampus (the 'perforant path'), and this neuronal activity is relayed through hippocampal CA fields and finally feeding back upon the EC itself. In addition to this neuronal "loop" (EC-->dentate-->CA1-->CA3-->subiculum-->EC), comparatively smaller numbers of EC fibers connect directly with the hippocampal CA fields and the subiculum (dotted arrows in Figure 6).

The nature of the neuronal processing which takes place, in this elaborately connected Figure 7 hippocampal/cortical system, is only just beginning to be understood. One theory which accounts well for anterograde amnesia, in medial lobe-damaged patients, is referred to as memory indexing. First proposed by Teyler and DiScenna (1986), this theory assumes that actual memories consist of changes within a distributed array of neocortical modules. The hippocampus stores a map or index of the location of the these modules. The hippocampus is initially necessary for Time from training to surgery reactivation of the memory but over time, through reactivation of hippocampal cortical circuits, the memory index itself is transferred to the neocortex. Thus, memory indexing theory predicts that the structures initially involved in memory storage are different from those involved in permanent storage. The consolidation phase, according to this theory, persists until the transfer is complete, the hippocampus is no longer necessary (remember patient HM could access detailed memories of his life before the hippocampectomy surgery). One experimental test of the memory indexing theory, is to produce hippocampal lesions at varying times after acquisition of a particular association. If the hippocampus is important for a transient period of time, we would predict that the lesions would have amnesic effects for newly acquired information, but not for more remote learning. This study was done by Zola-Morgan and Squire (1990). They trained monkeys on a series of 20 object discriminations on 5 separate occasions: 16, 12, 8, 4 and 2 weeks prior to surgical removal of hippocampus. Retention was then tested following recovery from surgery. They found that the lesions disrupted discriminations acquired 2 or 4 weeks

Figure 8. Figure 9.

before surgery but not earlier ones (see Figure 7). Thus, hippocampal lesions caused a retrograde amnesia for information acquired up to one month before surgery. These findings support the contention that the hippocampus plays a necessary role in storage for a transient time period following learning. Again, the reader should bear in mind that this is merely one theory of hippocampal functioning, and does not rule out additional roles and mechanisms for this structure.

Considerable evidence implicates the multimodal 'association' areas (see Figure 8) of the neocortex in long term memory storage. As indicated above, a rich set of interconnections establish links between the medial temporal lobe structures, and widespread neocortical areas. A structural component of this connectivity are long arcuate fasciculi that interconnect the cortical association areas (Figure 9).

The Role of the Diencephalic Areas

Damage to the midline diencephalon also causes amnesia. Specific areas identified as important include the mamillary nuclei (MN) and the mediodorsal thalamus (MD) (see Figure10). The idea that damage to the MN was related to memory was supported by the finding that in alcoholics with Korsakoff’s syndrome and its associated memory impairment, the MN and MD are consistently damaged. In other patients studied with CT scans after midline thalamic infarctions that caused memory impairment, the areas most frequently found with damage included MN and MD, as well as the internal medullary lamina, which would effectively disconnect several thalamic midline nuclei, including anterior nuclei and the MD.

The clinical diagnosis of Korsakoff's Figure 10. syndrome is characterized by persistent episodic memory loss (both anterograde and retrograde) and preserved semantic memory, intelligence and learned behavior (Graham & Lantos, 2002; Kopelman, 2002). Some of the neuropathological changes are illustrated in Figure 11, showing the characteristic atrophy of the mamillary bodies (white arrow) and hemorrhagic areas in the walls of the third ventricle (black arrow). There is a loss of and myelinated nerve fibers in several brain structures including the mamillary bodies, as well as the dorsomedial and laterodorsal thalamic nuclei. Memory distortions and are a frequent behavior feature in Korsakoff’s patients. The episodic memory impairment, with relative sparing of semantic memory, is one piece of evidence supporting separate anatomical substrates of these two memory modalities. Figure 11.

Role of the Basal Forebrain

Many patients with aneurysms of the anterior communicating artery exhibit persisting memory impairment. Damage associated with these insults occurs in the basal forebrain, the principal source of cholinergic innervation of the cortex. This area includes: 1. the Medial Septal Nuclei - providing cholinergic innervation to the hippocampus via the fornix 2. the Nucleus Basalis - providing cholinergic innervation widely to frontal, parietal and temporal cortices

Additional support for the idea that cholinergic innervation of the cortex originating from the basal forebrain is important in learning and memory was provided in reports of patients with Alzheimer’s disease, exhibiting memory impairment as a prominent early symptom. These patients show decreased choline acetyltransfersase activity (an enzyme critical for the synthesis of acetylcholine) in cortex and hippocampus, and markedly reduced cell numbers in the basal forebrain. The exact role of these structures and acetylcholine in learning and memory remains to be determined.

Alzheimer's Neuropathology

The frequency and severity of Alzheimer's Disease (AD) in modern society warrants a special consideration of this condition. AD remains the predominant cause of in the elderly. The American Psychiatric Assoc. defines dementia as "...the development of multiple cognitive deficits that include memory impairment and at least one of the following cognitive disturbances: , , agnosia or a disturbance in executive functioning.” Examination of the of AD patients at autopsy frequently reveals severe atrophy exemplified by narrowing of the gyri and widening of the sulci (see Figure 12). Accumulation of Beta-amyloid in the brain is a hallmark of Alzheimer’s disease. However, no strategies for curing the disease have been developed because the precise cause of the disease has not been discovered. Therapies in AD are limited to symptomatic treatments, for example using cognitive enhancer drugs (“nootropics”). Some limited success has been obtained using anticholinergic nootropic drugs (e.g., tacrine) which boost acetylcholine levels in the cortex.

Figure 12.

A major histopathological feature of AD is the neurofibrillary tangle, an intracellular structure often occupying large volumes of the cell body and apical dendrite of the . Figure 13 shows examples of these tangles at the light microscopic level (Bodian stain), in this case afflicting CA1 cells of the hippocampus. Notice that the tangles have filled the pyramidal-shaped cell bodies. Tangles are formed by hyperphosphorylation of a microtubule- associated protein known as tau, causing it to aggregate in an insoluble form.

It is notable that the entorhinal cortex is particularly susceptible to neurofibrillary tangle Figure 13. neuropathology. Braak and Braak (1997) quantified the incidence of tangles for various brain regions (Figure 14). Clearly the EC is one of the earliest and most severely affected areas, and its involvement increases with the severity of the disease.

Anatomical Correlates: Procedural Memory

Procedural memory is very specific to the nature of association or skill learned. It is very Figure 14. specific to the unconditioned stimulus/conditioned stimulus association developed in classical conditioning (e.g., the food (UCS) and the bell (CS) in the original conditioning of Pavlov’s dogs) or to the nature of the skill learned through repetition for motor learning.

Cerebellum and Most forms of motor learning probably first involve the and deep cerebellar nuclei, then subsequently, after the developed skill becomes ‘automatic’, some transfer occurs to the basal ganglia and associated ‘promotor’ motor cortical connections, leading to behaviors which are well coordinated.

Examples: Vestibulo-occular reflex conditioning (VOR) - vestibulocerebellum (flocculonodular lobe) Eyeblink conditioning - cerebellar cortex and deep nuclei Several types of procedural (nondeclarative or "implicit") memories are shown in Figure 15, along with their major associated neurological substrates. Note the association of the cerebellum and basal ganglia in classical conditioning and motor skills. Direct fiber connections from these subcortical structures to the neocortex are minimal, which is one of the factors underlying the unconscious nature of procedural memory. Key structures implicated in procedural memory are shown in Figure 16. MEMORY

DECLARATIVE (EXPLICIT) NONDECLARATIVE (IMPLICIT)

FACTS EVENTS SKILLS SIMPLE NONASSOCIATIVE AND CLASSICAL LEARNING HABITS CONDITIONING

EMOTIONAL SKELETAL RESPONSES MUSCULATURE

MEDIAL TEMPORAL LOBE DIENCEPHALON REFLEX NEOCORTEX AMYGDALA CEREBELLUM PATHWAYS

Figure 15.

Neocortex Other types of procedural learning, such as the development of specific perceptual and cognitive (problem solving) skills probably involve the neocortex, but not the structures important in declarative memory (medial temporal lobe), since these types of procedural memory formation can proceed relatively normally in animals for which declarative memory ability is reduced or absent after lesions to the medial temporal lobe.

Figure 16.

References, and suggestions for further reading

Cohen, N.J. and Eichenbaum, H. Anatomical data regarding the procedural-declarative distinction. Chapter 4, pp 98, In: Memory, Amnesia, and the Hippocampal System, MIT Press, 1993.

Graham DI, and Lantos PL. (Eds.) Greenfield's Neuropathology. Arnold, London, 2002.

Jones, R.S.G. Entorhinal-hippocampal connections: a speculative view of their function. TINS 16: 58-64, 1993.

Kandel ER, Schwartz JH, Jessell TM. (2000) Principles of Neuroscience, 4th Ed., McGraw Hill, New York.

Kopelman MD. (2002) Disorders of memory. Brain, 125, 2152-2190.

Levitan, I.B. and Kaczmarek, L.K. Learning and memory. Chapter 17, pp396, 416,420, In: The Neuron: Cell and Molecular Biology, Oxford Univ. Press, N.Y., 1991.

Madison, D.V., Malenka, R.C. and Nicoll, R.A. Mechanisms underlying long-term potentiation of synaptic transmission. Ann. Rev. Neurosci. 14: 379-397, 1991.

McNaughton, B.L. The mechanism of expression of long-term enhancement of hippocampal synapses: Current issues and theoretical implications. Ann.Rev. Neurosci. 55: 375-396, 1993.

Ridley, R.M. and Baker, H.F. A critical evaluation of monkey models of amnesia and dementia. Brain Res. Rev. 16: 15-37, 1991.

Squire, L.R. and Knowlton, B.J. Memory, hippocampus and brain systems. Chapter 53, pp 834, In: The Cognitive Neurosciences, M.Gazzaniga, Ed., MIT Press, 1995.

Teyler, T.J., and DiScenna, P. The hippocampal memory indexing theory. Behavioral Neuroscience, 100, 147-154, 1986.

Vargha-Khadem F, Gadian DG, Mishkin M. Dissociations in cognitive memory: the syndrome of developmental amnesia. Philos Trans R Soc Lond B Biol Sci.; 356(1413):1435-40, 2001.

Zola-Morgan, S.M., and Squire, L.R. The primate hippocampal formation: evidence for a time-limited role in memory storage. Science, 250, 288-290, 1990.

Zola-Morgan, S. and Squire, L.R. Neuroanatomy of memory. Ann Rev. Neurosci. 16: 547-563, 1993.

Zola-Morgan, S., Squire, L.R., Clower, R.P. and Rempel, N.L. Damage to the perirhinal cortex exacerbates memory impairment following lesions to the hippocampal formation. J. Neurosci. 13 (1): 251-265, 1993.