04 - 2.4 Biology of Memory

2.4 Biology of Memory

outcome it produces. It seems reasonable to expect that many pathological behaviors that come into the clinic might also be automatic and blind through repetition. Of interest, the evidence suggests that the eventual dominance of the S-R habit in behavior does not replace or destroy more cognitive mediation by learned S–S*, R–S*, and/or S– (R–S*) relations. Under some conditions, even a habitual response might be brought back under the control of the action–reinforcer association. The conversion of actions to habits and the relation of habit to cognition are active areas of research. REFERENCES Abramowitz JS, Arch JJ. Strategies for improving long-term outcomes in cognitive behavioral therapy for obsessivecompulsive disorder: insights from learning theory. Cogn Behav Pract. 2014;21:20–31. Bouton ME. Learning and Behavior: A Contemporary Synthesis. Sunderland, MA: Sinauer; 2007. Bouton ME. Learning theory. In: Sadock BJ, Sadock VA, Ruiz P, eds. Kaplan & Sadock’s Comprehensive Textbook of Psychiatry. 9th ed. Philadelphia: Lippincott Williams & Wilkins; 2009:647. Hockenbury D. Learning. In: Discovering Psychology. 5th ed. New York: Worth Publishers; 2011:183. Illeris K, ed. Contemporary Theories of Learning: Learning Theorists...In Their Own Words. New York: Routledge; 2009. Kosaki Y, Dickinson A. Choice and contingency in the development of behavioral autonomy during instrumental conditioning. J Exp Psychol Anim Behav Process. 2010;36(3):334. Maia TV. Two-factor theory, the actor-critic model, and conditioned avoidance. Learning Behav. 2010;38:50. Sigelman CK, Rider EA. Learning theories. In: Life-Span Human Development. Belmont: Wadsworth; 2012:42. Urcelay GP, Miller RR. Two roles of the context in Pavlovian fear conditioning. J Exp Psychol Anim Behav Process. 2010;36(2):268. 2.4 Biology of Memory The topic of memory is fundamental to the discipline of psychiatry. Memory is the glue that binds our mental life, the scaffolding for our personal history. Personality is in part an accumulation of habits that have been acquired, many very early in life, which create dispositions and influence how we behave. In the same sense, the neuroses are often products of learning—anxieties, phobias, and maladaptive behaviors that result from particular experiences. Psychotherapy itself is a process by which new habits and skills are acquired through the accumulation of new experiences. In this sense, memory is at the theoretical heart of psychiatry’s concern with personality, the consequences of early experience, and the possibility of growth and change. Memory is also of clinical interest because disorders of memory and complaints about memory are common in neurological and psychiatric illness. Memory impairment is also a side effect of certain treatments, such as electroconvulsive therapy. Accordingly, the effective clinician needs to understand something of the biology of memory, the varieties of memory dysfunction, and how memory can be evaluated. FROM SYNAPSES TO MEMORY

Memory is a special case of the general biological phenomenon of neural plasticity. Neurons can show history-dependent activity by responding differently as a function of prior input, and this plasticity of nerve cells and synapses is the basis of memory. In the last decade of the 19th century, researchers proposed that the persistence of memory could be accounted for by nerve cell growth. This idea has been restated many times, and current understanding of the synapse as the critical site of change is founded on extensive experimental studies in animals with simple nervous systems. Experience can lead to structural change at the synapse, including alterations in the strength of existing synapses and alterations in the number of synaptic contacts along specific pathways. Plasticity Neurobiological evidence supports two basic conclusions: First, short-lasting plasticity, which may last for seconds or minutes depending on specific synaptic events, including an increase in neurotransmitter release; and second, long-lasting memory depends on new protein synthesis, the physical growth of neural processes, and an increase in the number of synaptic connections. A major source of information about memory has come from extended study of the marine mollusk Aplysia californica. Individual neurons and connections between neurons have been identified, and the wiring diagram of some simple behaviors has been described. Aplysia is capable of associative learning (including classic conditioning and operant conditioning) and nonassociative learning (habituation and sensitization). Sensitization had been studied using the gill-withdrawal reflex, a defensive reaction in which tactile stimulation causes the gill and siphon to retract. When tactile stimulation is preceded by sensory stimulation to the head or tail, gill withdrawal is facilitated. The cellular changes underlying this sensitization begin when a sensory neuron activates a modulatory interneuron, which enhances the strength of synapses within the circuitry responsible for the reflex. This modulation depends on a second-messenger system in which intracellular molecules (including cyclic adenosine monophosphate [cAMP] and cAMP-dependent protein kinase) lead to enhanced transmitter release that lasts for minutes in the reflex pathway. Both short- and long-lasting plasticity within this circuitry are based on enhanced transmitter release. The long-lasting change uniquely requires the expression of genes and the synthesis of new proteins. Synaptic tagging mechanisms allow gene products that are delivered throughout a neuron to increase synaptic strength selectively at recently active synapses. In addition, the long-term change, but not the short-term change, is accompanied by the growth of neural processes of neurons within the reflex circuit. In vertebrates, memory cannot be studied as directly as in the simple nervous system of Aplysia. Nevertheless, it is known that behavioral manipulations can also result in measurable changes in the brain’s architecture. For example, rats reared in enriched as opposed to ordinary environments show an increase in the number of synapses ending on individual neurons in the neocortex. These changes are accompanied by small increases in cortical thickness, in the diameter of neuronal cell bodies, and in the number and length of dendritic branches. Behavioral experience thus exerts powerful

effects on the wiring of the brain. Many of these same structural changes have been found in adult rats exposed to an enriched environment, as well as in adult rats given extensive maze training. In the case of maze training, vision was restricted to one eye, and the corpus callosum was transected to prevent information received by one hemisphere from reaching the other hemisphere. The result was that structural changes in neuronal shape and connectivity were observed only in the trained hemisphere. This rules out a number of nonspecific influences, including motor activity, indirect effects of hormones, and overall level of arousal. Long-term memory in vertebrates is believed to be based on morphological growth and change, including increases in synaptic strength along particular pathways. Long-Term Potentiation The phenomenon of long-term potentiation (LTP) is a candidate mechanism for mammalian long-term memory. LTP is observed when a postsynaptic neuron is persistently depolarized after a high-frequency burst of presynaptic neural firing. LTP has a number of properties that make it suitable as a physiological substrate of memory. It is established quickly and then lasts for a long time. It is associative, in that it depends on the co-occurrence of presynaptic activity and postsynaptic depolarization. It occurs only at potentiated synapses, not all synapses terminating on the postsynaptic cell. Finally, LTP occurs prominently in the hippocampus, a structure that is important for memory. The induction of LTP is known to be mediated postsynaptically and to involve activation of the N-methyl-D-aspartate (NMDA) receptor, which permits the influx of calcium into the postsynaptic cell. LTP is maintained by an increase in the number of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA; non-NMDA) receptors in the postsynaptic cell and also possibly by increased transmitter release. A promising method for elucidating molecular mechanisms of memory relies on introducing specific mutations into the genome. By deleting a single gene, one can produce mice with specific receptors or cell signaling molecules inactivated or altered. For example, in mice with a selective deletion of NMDA receptors in the CA1 field of the hippocampus, many aspects of CA1 physiology remain intact, but the CA1 neurons do not exhibit LTP, and memory impairment is observed in behavioral tasks. Genetic manipulations introduced reversibly in the adult are particularly advantageous, in that specific molecular changes can be induced in developmentally normal animals. Associative Learning The study of classical conditioning has provided many insights into the biology of memory. Classical conditioning has been especially well studied in rabbits, using a tone as the conditioned stimulus and an air puff to the eye (which automatically elicits a blink response) as the unconditioned stimulus. Repeated pairings of the tone and the air puff lead to a conditioned response, in that the tone alone elicits an eye blink. Reversible lesions of the deep nuclei of the cerebellum eliminate the conditioned response without affecting the unconditioned response. These lesions also prevent initial learning from occurring, and, when the lesion is reversed, rabbits learn normally. Thus, the cerebellum contains essential circuitry for the learned association. The relevant plasticity appears to be distributed between the cerebellar cortex and the deep nuclei.

An analogous pattern of cerebellar plasticity is believed to underlie motor learning in the vestibuloocular reflex and, perhaps, associative learning of motor responses in general. Based on the idea that learned motor responses depend on the coordinated control of changes in timing and strength of response, it has been suggested that synaptic changes in the cerebellar cortex are critical for learned timing, whereas synaptic changes in the deep nuclei are critical for forming an association between a conditioned stimulus and an unconditioned stimulus. Fear conditioning and fear-potentiated startle are types of learning that serve as useful models for anxiety disorders and related psychiatric conditions. For example, mice exhibit freezing behavior when returned to the same context in which aversive shock was delivered on an earlier occasion. This type of learning depends on the encoding of contextual features of the learning environment. Acquiring and expressing this type of learning requires neural circuits that include both the amygdala and the hippocampus. The amygdala may be important for associating negative affect with new stimuli. The hippocampus may be important for representing the context. With extinction training, when the context is no longer associated with an aversive stimulus, the conditioned fear response fades. Frontal cortex is believed to play a key role in extinction. CORTICAL ORGANIZATION OF MEMORY One fundamental question concerns the locus of memory storage in the brain. In the 1920s, Karl Lashley searched for the site of memory storage by studying the behavior of rats after removing different amounts of their cerebral cortex. He recorded the number of trials needed to relearn a maze problem that rats learned prior to surgery, and he found that the deficit was proportional to the amount of cortex removed. The deficit did not seem to depend on the particular location of cortical damage. Lashley concluded that the memory resulting from maze learning was not localized in any one part of the brain but instead was distributed equivalently over the entire cortex. Subsequent research has led to reinterpretations of these results. Maze learning in rats depends on different types of information, including visual, tactile, spatial, and olfactory information. Neurons that process these various types of information are segregated into different areas of the rat cerebral cortex, and memory storage is segregated in a parallel manner. Thus, the correlation between maze-learning abilities and lesion size that Lashley observed is a result of progressive encroachment of larger lesions on specialized cortical areas serving the many components of information processing relevant for maze learning. The functional organization of the mammalian cerebral cortex has been revealed through neuropsychological analyses of deficits following brain damage and through physiological studies of intact brains. The cortical areas responsible for processing and storing visual information have been studied most extensively in nonhuman primates. Nearly one half of the primate neocortex is specialized for visual functions. The cortical pathways for visual information processing begin in primary visual cortex (V1) and proceed from there along parallel pathways or streams. One stream projects ventrally to inferotemporal cortex and is specialized for processing information concerning the identification of visual objects. Another stream projects dorsally to parietal cortex and is specialized for processing information about spatial location. Specific visual processing areas in the dorsal and ventral streams, together with areas in prefrontal cortex, register the

immediate experience of perceptual processing. The results of perceptual processing are first available as immediate memory. Immediate memory refers to the amount of information that can be held in mind (like a telephone number) so that it is available for immediate use. Immediate memory can be extended in time by rehearsing or otherwise manipulating the information, in which case what is stored is said to be in working memory. Regions of visual cortex in forward portions of the dorsal and ventral streams serve as the ultimate repositories of visual memories. Inferotemporal cortex, for example, lies at the end of the ventral stream, and inferotemporal lesions lead to selective impairments in both visual object perception and visual memory. Nonetheless, such lesions do not disrupt elementary visual functions, such as acuity. Electrophysiological studies in the monkey show that neurons in area TE, which is one part of inferotemporal cortex, register specific and complex features of visual stimuli, such as shape, and respond selectively to patterns and objects. Inferotemporal cortex can thus be thought of as both a higher-order visual processing system and a storehouse of the visual memories that result from that processing. In sum, memory is distributed and localized in the cerebral cortex. Memory is distributed in the sense that, as Lashley concluded, there is no cortical center dedicated solely to the storage of memories. Nonetheless, memory is localized in the sense that different aspects or dimensions of events are stored at specific cortical sites—namely, in the same regions that are specialized to analyze and process what is to be stored. MEMORY AND AMNESIA The principle that the functional specialization of cortical regions determines both the locus of information processing and the locus of information storage does not provide a complete account of the organization of memory in the brain. If it did, then brain injury would always include a difficulty in memory for a restricted type of information along with a loss of ability to process information of that same type. This kind of impairment occurs, for example, in the aphasias and the agnosias. However, there is another kind of impairment that can occur as well, called amnesia. The hallmark of amnesia is a loss of new learning ability that extends across all sensory modalities and stimulus domains. This anterograde amnesia can be explained by understanding the role of brain structures critical for acquiring information about facts and events. Typically, anterograde amnesia occurs together with retrograde amnesia, a loss of knowledge that was acquired before the onset of amnesia. Retrograde deficits often have a temporal gradient, following a principle known as Ribot’s law; deficits are most severe for information that was most recently learned. A patient with a presentation of amnesia exhibits severe memory deficits in the context of preservation of other cognitive functions, including language comprehension and production, reasoning, attention, immediate memory, personality, and social skills. The selectivity of the memory deficit in these cases implies that intellectual and perceptual functions of the brain are separated from the capacity to lay down in memory the records that ordinarily result from engaging in intellectual and perceptual work. Specialized Memory Function

Amnesia can result from damage to the medial portion of the temporal lobe or from damage to regions of the midline diencephalon. Studies of a severely amnesic patient known as HM stimulated intensive investigation of the role of the medial temporal lobe in memory. HM became amnesic in 1953, at 27 years of age, when he sustained a bilateral resection of the medial temporal lobe to relieve severe epilepsy. The removal included approximately one half of the hippocampus, the amygdala, and most of the neighboring entorhinal and perirhinal cortices (Fig. 2.4-1). After the surgery, HM’s seizure condition was much improved, but he experienced profound forgetfulness. His intellectual functions were generally preserved. For example, HM exhibited normal immediate memory, and he could maintain his attention during conversations. After an interruption, however, HM could not remember what had recently occurred. HM’s dense amnesia was permanent and debilitating. In HM’s words, he felt as if he were just waking from a dream, because he had no recollection of what had just taken place. FIGURE 2.4-1 Structural magnetic resonance images of the brains of patients HM and EP through the level of the temporal lobe. Damaged tissue is indicated by bright signal in these T2weighted axial images. Both patients sustained extensive damage to medial temporal structures, as the result of surgery for epilepsy in HM, and as the result of viral encephalitis in EP. Scale bar: 2 cm. L, left side of the brain. (Reprinted from Corkin S, Amaral EG, González RG, Johnson KA, Hyman BT. H.M.’s medial temporal lobe lesion: Findings from magnetic resonance imaging. J Neurosci. 1997;17:3964; and Stefanacci L, Buffalo EA, Schmolck H, Squire LR. Profound amnesia after damage to the medial temporal lobe: A neuroanatomical and neuropsychological profile of patient E.P. J Neurosci. 2000;20:7024, with permission.)

In monkeys, many parallels to human amnesia have been demonstrated after surgical damage to anatomical components of the medial temporal lobe. Cumulative study of the resulting memory impairment eventually identified the medial temporal structures and connections that are crucial for memory. These include the hippocampus—which includes the dentate gyrus; hippocampal fields CA1, CA2, and CA3; and the subiculum— and the adjacent cortical regions, including the entorhinal, perirhinal, and parahippocampal cortices. Another important medial temporal lobe structure is the amygdala. The amygdala is involved in the regulation of much of emotional behavior. In particular, the storage of emotional events engages the amygdala. Modulatory effects of projections from the amygdala to the neocortex are responsible for producing enhanced memory for emotional or arousing events compared to neutral events. Detailed study of amnesic patients offers unique insights into the nature of memory and its organization in the brain. An extensive series of informative studies, for example, described the memory impairment of patient EP. EP was diagnosed with herpes simplex encephalitis at 72 years of age. Damage to the medial temporal lobe region (see Fig. 2.4-1) produced a persistent and profound amnesia. During testing sessions, EP was cordial and talked freely about his life experiences, but he relied exclusively on stories from his childhood and early adulthood. He would repeat the same story many times. Strikingly, his performance on tests of recognition memory was no better than would result from guessing (Fig. 2.4-2A). Tests involving facts about his life and autobiographical experiences revealed poor memory for the time leading up to his illness but normal memory for his childhood (Fig. 2.4-2B). EP also has good spatial knowledge about the town in which he lived as a child, but he was unable to learn the layout of the neighborhood where he lived only after he became amnesic (Fig. 2.4-2C).

FIGURE 2.4-2 Formal test results for patient EP, showing severe anterograde and retrograde deficits, with intact remote memory. A. Scores were combined from 42 different tests of recognition memory for words given to patient EP and a group of five healthy control subjects. The testing format was either two-alternative forced choice or yes–no recognition. Brackets for EP indicate the standard error of the mean. Data points for the control group indicate each participant’s mean score across all 42 recognition memory tests. EP’s average performance (49.3 percent correct) was not different from chance and was approximately five standard deviations (SDs) below the average performance of control subjects (81.1 percent correct; SD, 6.3). B. Autobiographical remembering was quantified by using a structured interview known as the Autobiographical Memory Interview. Items assessed personal semantic knowledge (maximum score 21 for each time period). Performance for the recent time period reflects poor memory for information that could have been acquired only subsequent to the onset of his amnesia. For EP, performance for the early adult period reflects retrograde memory deficits. Performance for the childhood period reflects good remote memory. Similar results for semantic and episodic remembering were obtained from these time periods. (Data from Kopelman MD, Wilson BA, Baddeley AD. The autobiographical memory interview: A new assessment of autobiographical and personal semantic memory in amnesic

patients. J Clin Exp Neuropsychol. 1989;5:724; and Reed JM, Squire LR. Retrograde amnesia for facts and events: Findings from four new cases. J Neurosci. 1998;18:3943). C. Assessments of spatial memory demonstrated EP’s good memory for spatial knowledge from his childhood, along with extremely poor new learning of spatial information. Performance was compared to that of five individuals (open circles) who attended EP’s high school at the same time as he did, lived in the region over approximately the same time period, and, like EP (filled circles), moved away as young adults. Normal performance was found for navigating from home to different locations in the area (familiar navigation), between different locations in the area (novel navigation), and between these same locations when a main street was blocked (alternative routes). Subjects were also asked to point to particular locations while imagining themselves in a particular location (pointing to landmarks), or they were asked about locations in the neighborhoods in which they currently lived (new topographical learning). EP showed difficulty only in this last test, because he moved to his current residence after he became amnesic. (Data from Teng E, Squire LR. Memory for places learned long ago is intact after hippocampal damage. Nature. 1999;400:675.) (Adapted from Stefanacci L, Buffalo EA, Schmolck H, Squire LR. Profound amnesia after damage to the medial temporal lobe: A neuroanatomical and neuropsychological profile of patient E.P. J Neurosci. 2000;20:7024. Printed with permission.) Given the severity of the memory problems experienced by EP and other amnesic patients, it is noteworthy that these patients nonetheless perform normally on certain kinds of memory tests. The impairment selectively concerns memory for factual knowledge and autobiographical events, collectively termed declarative memory. Amnesia presents as a global deficit, in that it involves memory for information presented in any sensory modality, but the deficit is limited, in that it covers only memory for facts and events. Hippocampal pathology in patients with amnesia can also be revealed using highresolution magnetic resonance imaging (MRI). Such studies indicate that damage limited to the hippocampus results in clinically significant memory impairment. In addition to the hippocampus, other medial temporal lobe regions also make critical contributions to memory. Thus, a moderately severe memory impairment results from CA1 damage, whereas a more profound and disabling amnesia results from medial temporal lobe damage that includes the hippocampus and adjacent cortex. Memory impairment due to medial temporal lobe damage is also typical in patients with early Alzheimer’s disease or amnestic mild cognitive impairment. As Alzheimer’s disease progresses, the pathology affects many cortical regions and produces substantial cognitive deficits in addition to memory dysfunction. Amnesia can also result from damage to structures of the medial diencephalon. The critical regions damaged in diencephalic amnesia include the mammillary nuclei in the hypothalamus, the dorsomedial nucleus of the thalamus, the anterior nucleus, the internal medullary lamina, and the mammillothalamic tract. However, uncertainty remains regarding which specific lesions are required to produce diencephalic amnesia.

Alcoholic Korsakoff’s syndrome is the most prevalent and best-studied example of diencephalic amnesia, and in these cases damage is found in brain regions that may be especially sensitive to prolonged bouts of thiamine deficiency and alcohol abuse. Patients with alcoholic Korsakoff’s syndrome typically exhibit memory impairment due to a combination of diencephalic damage and frontal lobe pathology. Frontal damage alone produces characteristic cognitive deficits along with certain memory problems (e.g., in effortful retrieval and evaluation); in Korsakoff’s syndrome the pattern of deficits thus extends beyond what is commonly found in other cases of amnesia (see Table 2.4-1). Table 2.4-1 Memory and Cognitive Deficits Associated with Frontal Damage The ability to remember factual and autobiographical events depends on the integrity of both the cortical regions responsible for representing the information in question and several brain regions that are responsible for memory formation. Thus, medial temporal and diencephalic brain areas work in concert with widespread areas of neocortex to form and to store declarative memories (Fig. 2.4-3).

FIGURE 2.4-3 Brain regions believed to be critical for the formation and storage of declarative memory. Medial diencephalon and medial temporal regions are critical for declarative memory storage. The entorhinal cortex is the major source of projections for the neocortex to the hippocampus, and nearly two thirds of the cortical input to the entorhinal cortex originates in the perirhinal and parahippocampal cortex. The entorhinal cortex also receives direct connections from the cingulate, insula, orbitofrontal, and superior temporal cortices. (Adapted from Paller KA,: Neurocognitive foundations of human memory. In: Medin DL, ed.: The Psychology of Learning and Motivation. Vol. 40. San Diego, CA: Academic Press; 2008:121; and Gluck MA, Mercado E, Myers CE.: Learning and Memory: From Brain to Behavior. New York: Worth; 2008:109, Fig. 3.16.) Retrograde Amnesia Memory loss in amnesia typically affects recent memories more than remote memories (Fig. 2.4-4). Temporally graded amnesia has been demonstrated retrospectively in studies of amnesic patients and prospectively in studies of monkeys, rats, mice, and rabbits. These findings have important implications for understanding the nature of the memory storage process. Memories are dynamic, not static. As time passes after learning, some memories are forgotten, whereas others become stronger due to a process of consolidation that depends on cortical, medial temporal, and diencephalic structures.

FIGURE 2.4-4 A. Temporally limited retrograde amnesia for free recall of 251 news events. Scores were aligned relative to the onset of amnesia in patients (N = 6) and to a corresponding time point in age-matched and education-matched healthy individuals (N = 12). The time period after the onset of amnesia is labeled AA (anterograde amnesia) to designate that this time point assessed memory for events that occurred after the onset of amnesia. Standard errors ranged from 2 to 10 percent. Brain damage in the patient group was limited primarily to the hippocampal region. B. Temporally limited retrograde amnesia in rats with lesions of the hippocampus and subiculum. Rats learned to prefer an odorous food as the result of an encounter with another rat with that odor on its breath. Percent preference for the familiar food was observed for three trainingsurgery intervals. At 1 day after learning, the control group performed significantly better than the rats with lesions (P < .05). At 30 days, the two groups performed similarly, and both groups performed well above chance. Error bars show standard errors of the mean. (Adapted from Manns JR, Hopkins RO, Squire LR. Semantic memory and the human hippocampus. Neuron. 2003;38:127; and Clark RE, Broadbent NJ, Zola SM,

Squire LR. Anterograde amnesia and temporally graded retrograde amnesia for a nonspatial memory task after lesions of hippocampus and subiculum. J Neurosci. 2002;22:4663, with permission.) The study of retrograde amnesia has been important for understanding how memory changes over time. The dynamic nature of memory storage can be conceptualized as follows. An event is experienced and encoded by virtue of a collection of cortical regions that are involved in representing a combination of different event features. At the same time, the hippocampus and adjacent cortex receive pertinent high-level information from all sensory modalities. Later, when the original event is recalled, the same set of cortical regions is activated. If a subset of the cortical regions is activated, the hippocampus and related structures can facilitate recall by facilitating the activation of the remaining cortical regions (i.e., pattern completion). When the original event is retrieved and newly associated with other information, hippocampal–cortical networks can be modified. In this way, a gradual consolidation process occurs that changes the nature of memory storage (see Fig. 2.4-3). The neocortical components representing some events can become so effectively linked together that ultimately a memory can be retrieved without any contribution from the medial temporal lobe. As a result, amnesic patients can exhibit normal retrieval of remote facts and events, as well as autobiographical memories. Distributed neocortical regions are the permanent repositories of these enduring memories. In contrast to what is observed after damage restricted to the hippocampus, extensive retrograde impairments for facts and events from the distant past can also occur. Damage to the frontal lobes, for example, can lead to difficulty in organizing memory retrieval. Accurate retrieval often begins with an activation of lifetime periods and proceeds to an identification of general classes of events and then more specific events, but this process becomes difficult following frontal damage. Damage to other cortical regions can also impair memory storage. Networks in anterolateral temporal cortex, for example, are critical for retrieving stored information because these areas are important for long-term storage itself. Patients with focal retrograde amnesia exhibit substantial retrograde memory impairments together with only moderately impaired new learning ability. Some capacity for new learning remains, presumably because medial temporal lobe structures are able to communicate with other areas of cortex that remain undamaged. MULTIPLE TYPES OF MEMORY Memory is not a single faculty of the mind but consists of various subtypes. Amnesia affects only one kind of memory, declarative memory. Declarative memory is what is ordinarily meant by the term memory in everyday language. Declarative memory supports the conscious recollection of facts and events. The classic impairment in amnesia thus concerns memory for routes, lists, faces, melodies, objects, and other verbal and nonverbal material, regardless of the sensory modality in which the material is presented. Amnesic patients can display a broad impairment in these components of declarative memory while a number of other memory abilities are preserved. The heterogeneous set

of preserved abilities is collectively termed nondeclarative memory. Nondeclarative memory includes skill learning, habit learning, simple forms of conditioning, and a phenomenon called priming. For these kinds of learning and memory, amnesic patients can perform normally. In controlled laboratory settings, the acquisition of a variety of perceptual, perceptual–motor, and cognitive skills can be tested in isolation, and amnesic patients are found to acquire these skills at rates equivalent to the rates at which healthy individuals acquire the skills. For example, amnesic patients can learn to read mirror-reversed text normally, they exhibit the normal facilitation in reading speed with successive readings of normal prose, and they improve as rapidly as healthy individuals at speeded reading of repeating nonwords. In addition, amnesic patients can, after seeing strings of letters generated by a finite-state rule system, classify novel strings of letters as rule based or not rule based. Classification performance is normal despite the fact that amnesic patients are impaired at remembering the events of training or the specific items they have studied. Priming Priming refers to a facilitation of the ability to detect or to identify a particular stimulus based on a specific recent experience. Many tests have been used to measure priming in amnesia and show that it is intact. For example, words might be presented in a study phase and then again, after a delay, in a test phase when a priming measure such as reading speed is obtained. Patients are instructed to read words as quickly as possible in such a test, and they are not informed that memory is being assessed. In one priming test, patients named pictures of previously presented objects reliably faster than they named pictures of new objects, even after a delay of 1 week. This facilitation occurred at normal levels, despite the fact that the patients were markedly impaired at recognizing which pictures had been presented previously. Particularly striking examples of preserved priming come from studies of patient EP (Fig. 2.4-5), who exhibited intact priming for words but performed at chance levels when asked to recognize which words had been presented for study. This form of memory, termed perceptual priming, is thus a distinct class of memory that is independent of the medial temporal lobe regions typically damaged in amnesia.

FIGURE 2.4-5 Preserved priming in patient EP relative to seven control subjects. A. Stem-completion priming on six separate tests. Priming reflected a tendency for subjects to complete three-letter stems with previously encountered words when they were instructed to produce the first word to come to mind (e.g., MOT___ completed to form MOTEL). Priming scores were calculated as percentage correct for studied words minus percentage correct for baseline words (guessing). B. Perceptual-identification priming on 12 separate tests. Subjects attempted to read 48 words that were visually degraded. Priming scores were calculated as percentage correct identification of previously studied words minus percentage correct identification of nonstudied words. Brackets indicate standard error of the mean. (Data from Hamann SB, Squire LR. Intact perceptual memory in the absence of conscious memory. Behav Neurosci. 1997;111:850.) (Reprinted from Stefanacci L, Buffalo EA, Schmolck H, Squire LR. Profound amnesia after damage

to the medial temporal lobe: A neuroanatomical and neuropsychological profile of patient E.P. J Neurosci. 2000;20:7024, with permission.) Another form of priming reflects improved access to meaning rather than percepts. For example, subjects study a list of words, including tent and belt, and then are asked to free associate to other words. Thus, they are given words such as canvas and strap and asked to produce the first word that comes to mind. The result is that subjects are more likely to produce tent in response to canvas and to produce belt in response to strap than if the words tent and belt had not been presented recently. This effect, called conceptual priming, is also preserved in amnesic patients, even though they fail to recognize the same words on a conventional memory test (Fig. 2.4-6). FIGURE 2.4-6 Preserved conceptual priming in amnesia. In the free association test, subjects studied a set of words (e.g., lemon) and 5 minutes later viewed cue words that included associates of the studied words (e.g., orange). Subjects were asked to produce the first word that came to mind in response to each cue word. Results are shown separately for the control group (CON; n = 12), amnesic patients with large medial temporal lobe lesions (MTL; n = 2), and amnesic patients with lesions believed to be limited to the hippocampal region (H; n = 3). The left panel shows conceptual priming scores computed as the percentage of studied words produced in the free association test minus a baseline measure of the likelihood of producing those words by chance. All of the groups performed similarly in the conceptual priming test. The right panel shows results from a yes–no recognition memory test using comparable words. Both patient groups were impaired relative to the control group. The dashed line indicates chance performance. The data points for the MTL and H groups show the scores of individual patients averaged across four separate tests. Brackets show standard errors of the mean for the control group. (Reprinted from Levy DA, Stark CEL, Squire LR. Intact conceptual priming in the absence of declarative memory. Psychol Sci. 2004;15:680, with permission.)

Not all types of priming are preserved in amnesia. Some priming tests have been designed to examine the formation of new associations. When priming tests are based not on preexisting knowledge but on the acquisition of new associative knowledge, priming tends to be impaired. In other words, priming in certain complex situations can require the same type of linkage among multiple cortical regions that is critical for declarative memory. Memory Systems Table 2.4-2 depicts one scheme for conceptualizing multiple types of memory. Declarative memory depends on medial temporal and midline diencephalic structures along with large portions of the neocortex. This system provides for the rapid learning of facts (semantic memory) and events (episodic memory). Nondeclarative memory depends on several different brain systems. Habits depend on the neocortex and the neostriatum, and the cerebellum is important for the conditioning of skeletal musculature, the amygdala for emotional learning, and the neocortex for priming. Table 2.4-2 Types of Memory Declarative memory and nondeclarative memory differ in important ways. Declarative memory is phylogenetically more recent than nondeclarative memory. In addition, declarative memories are available to conscious recollection. The flexibility of declarative memory permits the retrieved information to be available to multiple response systems. Nondeclarative memory is inaccessible to awareness and is expressed only by engaging specific processing systems. Nondeclarative memories are stored as changes within these processing systems—changes that are encapsulated such that the stored information has limited accessibility to other processing systems. Semantic memory, which concerns general knowledge of the world, has often been categorized as a separate form of memory. Facts that are committed to memory typically become independent of the original episodes in which the facts were learned. Amnesic patients can sometimes acquire information that would ordinarily be learned as facts, but the patients learn it by relying on a different brain system than the system that supports declarative memory. Consider a test requiring the concurrent learning of eight object pairs. Healthy individuals can rapidly learn which the correct object in each pair is, whereas severely amnesic patients like EP learn only gradually over many weeks, and at the

start of each session they cannot describe the task, the instructions, or the objects. In patients who do not have severe amnesia, factual information is typically acquired as consciously accessible declarative knowledge. In these cases, the brain structures that remain intact within the medial temporal lobe presumably support learning. In contrast, when factual information is acquired as nondeclarative knowledge, as in the case of EP’s learning of pairs of objects, learning likely occurs directly as a habit, perhaps supported by the neostriatum. Humans thus appear to have a robust capacity for habit learning that operates outside of awareness and independent of the medial temporal lobe structures that are damaged in amnesia. Frontal Contributions to Memory Although amnesia does not occur after limited frontal damage, the frontal lobes are fundamentally important for declarative memory. Patients with frontal lesions have poor memory for the context in which information was acquired, they have difficulty in unaided recall, and they may even have some mild difficulty on tests of item recognition. More generally, patients with frontal lesions have difficulty implementing memory retrieval strategies and evaluating and monitoring their memory performance. NEUROIMAGING AND MEMORY The understanding of memory derived from studies of amnesia has been extended through studies using various methods for monitoring brain activity in healthy individuals. For example, activation of posterior prefrontal regions with positron emission tomography (PET) and functional MRI have shown that these regions are involved in strategic processing during retrieval, as well as in working memory. Anterior frontal regions near the frontal poles have been linked with functions such as evaluating the products of retrieval. Frontal connections with posterior neocortical regions support the organization of retrieval and the manipulation of information in working memory. Consistent with the evidence from patients with frontal lesions, frontal–posterior networks can be viewed as instrumental in the retrieval of declarative memories and in the online processing of new information. Neuroimaging has also identified contributions to memory made by parietal cortex. Multiple parietal regions (including inferior and superior parietal lobules, precuneus, posterior cingulate, and retrosplenial cortex) are activated in conjunction with remembering recent experiences. Although many functions have been hypothesized to explain this parietal activity, a single consensus position has not yet been reached, and it may be the case that several different functions are relevant. Neuroimaging studies have also illuminated priming phenomena and how they differ from declarative memory. Perceptual priming appears to reflect changes in early stages of the cortical pathways that are engaged during perceptual processing. For example, in the case of stem-completion priming, in which subjects study a list of words (e.g., MOTEL) and then are tested with a list of stems (e.g., MOT___) and with instructions to complete each stem with the first word to come to mind, neuroimaging and divided visual-field studies have implicated visual processing systems in extrastriate cortex, especially in the right hemisphere. In contrast, conscious recollection of remembered

words engages brain areas at later stages of processing. Neural mechanisms supporting priming and declarative memory retrieval have also been distinguished in brain electrical activity recorded from the scalp (Fig. 2.4-7). In sum, priming differs from declarative memory in that it is signaled by brain activity that occurs earlier and that originates in different brain regions. FIGURE 2.4-7 Brain potentials associated with perceptual priming versus declarative memory retrieval. Paller et al. (2003) studied 16 volunteers, who took a memory test involving three sorts of faces: new faces, faces they had seen recently and remembered well, and faces they had seen but did not remember because the faces had been presented too briefly to process effectively. In a companion experiment with a priming test, speeded responses were found, indicative of priming. Frontal recordings of brain waves elicited by the primed faces included negative potentials from 200 to 400 ms after face presentation that differed from the brain waves elicited by new faces. These differences were particularly robust for trials with the fastest responses (data shown were from trials with responses faster than the median reaction time). Remembered faces uniquely elicited positive brain waves that began about 400 ms after face presentation. Brain potential correlates of face recollection occurred later than those for perceptual priming and were larger over posterior brain regions. (Adapted from Paller KA, Hutson CA, Miller BB, Boehm SG. Neural manifestations of memory with and without awareness. Neuron. 2003;38:507, with permission.) Hippocampal activity associated with the formation and retrieval of declarative memories has also been investigated with neuroimaging. In keeping with the neuropsychological evidence, the hippocampus appears to be involved in the recollection of recent events (Fig. 2.4-8). Retrieval-related hippocampal activity has

been observed in memory tests with many different types of stimuli. The hippocampus is also active during the initial storage of information. Whereas left inferior prefrontal cortex is engaged as a result of attempts to encode a word, hippocampal activity at encoding is more closely associated with whether encoding leads to a stable memory that can later be retrieved (Fig. 2.4-9). These findings confirm and extend the idea that medial temporal and frontal regions are important for memory storage and that they make different contributions. FIGURE 2.4-8 Activity in the left and right hippocampal regions measured with functional magnetic resonance imaging (fMRI) during declarative memory retrieval. Data were collected from 11 participants who saw words at study and test and from 11 different participants who saw pictures of namable objects at study and test. Recognition memory accuracy was 80.2 percent correct for words and 89.9 percent correct for objects. Areas of significant fMRI signal change (targets vs. foils) are shown in sagittal sections as color overlays on averaged structural images. The green box indicates the area in which reliable data were available for all subjects. With words, retrieval-related activity was observed in the hippocampus on the left side (A) but not on the right side (B). With namable objects, retrieval-related activity was observed in the hippocampus on both the left side (C) and the right side (D). (Reprinted from Stark CE, Squire LR. Functional magnetic resonance imaging (fMRI) activity in the hippocampal region during recognition memory. J Neurosci. 2000;20:7776, with permission.)

FIGURE 2.4-9 Functional activations of prefrontal and medial temporal regions that were predictive of later memory performance. Single words were presented visually, each followed by an instruction to remember (R cue) or to forget (F cue). Trials were sorted based on the remember or forget instruction and on subsequent recognition performance. Activity in the left inferior prefrontal cortex and left hippocampus was predictive of subsequent recognition but for different reasons. Left inferior prefrontal activation (A) was associated with the encoding attempt, in that responses were largest for trials with a cue to remember, whether or not the word was actually recognized later. The time course of activity in this region (B) was computed based on responses that were time locked to word onset (time 0). Left inferior prefrontal activity increased for words that were later remembered, but there was a stronger association with encoding attempt, because responses were larger for words followed by an R cue that were later forgotten than for words followed by an F cue that were later remembered. In contrast, left parahippocampal and posterior hippocampal activation (C) was associated with encoding success. As shown by the time course of activity in this region (D), responses were largest for words that were subsequently remembered, whether the cue was to remember or to forget. (Reprinted from Reber PJ, Siwiec RM, Gitelman DR, Parrish TB, Mesulam MM, Paller KA. Neural correlates of successful encoding identified using functional magnetic resonance imaging. J Neurosci. 2002;22:9541, with permission.) SLEEP AND MEMORY

The speculation that memories are processed during sleep has a long history. Freud noted that dreams can reveal fragments of recent experiences in the form of day residues. Although many questions about how and why memories may be processed during sleep remain unresolved, recent experiments have provided new empirical support to bolster the idea that memory processing during sleep serves an adaptive function. It is now clear that memory performance can be facilitated when sleep occurs after initial learning, and that sleep-related facilitation can be observed for many different types of memory. Memory storage appears to be specifically aided by processing during deep sleep within a few hours after learning, especially in stages 3 and 4 (slow-wave sleep). Some results indicate that slow-wave sleep facilitates the storage of declarative memories but not nondeclarative memories. Direct evidence for this proposal has been obtained using stimulation with olfactory stimuli (Fig. 2.4-10), stimulation with direct current electrical input at the approximate frequency of electroencephalographic slow waves, and other methods. Furthermore, neuronal recordings in animals have revealed a phenomenon of hippocampal replay, in which activity patterns expressed during the day are later observed during sleep. In summary, declarative memories acquired during waking can be processed again during sleep, and this processing can influence the likelihood of subsequent memory retrieval when the individual is awake. The facilitation of declarative memory is typically manifest as a reduction in the amount of forgetting that occurs, not as an improvement in memory.

FIGURE 2.4-10 Evidence for memory processing during sleep. Subjects first learned object–location associations when a rose odor was present. Following learning, subjects slept wearing a device for delivering odors to the nose, and the rose odor was administered during the first two slow-wave sleep periods of the night (in 30-second periods to prevent habituation). Memory facilitation was observed when object–location associations were tested the following morning in the absence of odor stimulation. Facilitated memory was not found when stimulation occurred during slow-wave sleep but not during learning, when stimulation occurred during learning and then during rapid-eyemovement (REM) sleep, or when subjects were kept awake. Moreover, odor stimulation during slow-wave sleep was found to produce anterior and posterior hippocampal activation (lower panels). (Reprinted from Rasch B, Büchel C, Gais S, Born J. Odor cues during slow-wave sleep prompt declarative memory consolidation. Science. 2007;315:1426, with permission.) ASSESSMENT OF MEMORY FUNCTIONS A variety of quantitative methods are available to assess memory functions in neurological and psychiatric patients. Quantitative methods are useful for evaluating and following patients longitudinally, as well as for carrying out a one-time

examination to determine the status of memory function. It is desirable to obtain information about the severity of memory dysfunction, as well as to determine whether memory is selectively affected or whether memory problems are occurring, as they often do, against a background of additional intellectual deficits. Although some widely available tests, such as the Wechsler Memory Scale, provide useful measures of memory, most single tests assess memory rather narrowly. Even general-purpose neuropsychological batteries provide for only limited testing of memory functions. A complete assessment of memory usually involves a number of specialized tests that sample intellectual functions, new learning capacity, remote memory, and memory selfreport. The assessment of general intellectual functions is central to any neuropsychological examination. In the case of memory testing, information about intellectual functions provides information about a patient’s general test-taking ability and a way to assess the selectivity of memory impairment. Useful tests include the Wechsler Adult Intelligence Scale; a test of object naming, such as the Boston Naming Test; a rating scale to assess the possibility of global dementia; a test of word fluency; and specialized tests of frontal lobe function. New Learning Capacity Memory tests are sensitive to impaired new learning ability when they adhere to either of two important principles. First, tests are sensitive to memory impairment when more information is presented than can be held in immediate memory. For example, one might ask patients to memorize a list of ten faces, words, sentences, or digits, given that ten items is more than can be held in mind. The paired-associate learning task is an especially sensitive test of this kind. In the paired-associate task, the examiner asks the patient to learn a list of unrelated pairs of words (for example, queen–garden, office– river) and then to respond to the first word in each pair by recalling the second word. Second, tests are sensitive to memory impairment when a delay, filled with distraction, is interposed between the learning phase and the test phase. In that case, examiners typically ask patients to learn a small amount of information and then distract them for several minutes by conversation to prevent rehearsal. Recollection is then assessed for the previously presented material. Memory can be tested by unaided recall of previously studied material (free recall), by presenting a cue for the material to be remembered (cued recall), or by testing recognition memory. In multiple-choice tests of recognition memory, the patient tries to select previously studied items from a group of studied and unstudied items. In yes–no recognition tests, patients see studied and unstudied items one at a time and are asked to say “yes” if the item was presented previously and “no” if it was not. These methods for assessing recently learned material vary in terms of their sensitivity for detecting memory impairment, with free recall being most sensitive, cued recall intermediate, and recognition least sensitive. The specialization of function of the two cerebral hemispheres in humans means that left and right unilateral damage is associated with different kinds of memory problems.

Accordingly, different kinds of memory tests must be used when unilateral damage is a possibility. In general, damage to medial temporal or diencephalic structures in the left cerebral hemisphere causes difficulty in remembering verbal material, such as word lists and stories. Damage to medial temporal or diencephalic structures in the right cerebral hemisphere impairs memory for faces, spatial layouts, and other nonverbal material that is typically encoded without verbal labels. Left medial temporal damage can lead to impaired memory for spoken and written text. Right medial temporal damage can lead to impaired learning of spatial arrays, whether the layouts are examined by vision or by touch. A useful way to test for nonverbal memory is to ask a patient to copy a complex geometric figure and then, after a delay of several minutes, without forewarning, ask the patient to reproduce it. Remote Memory Evaluations of retrograde memory loss should attempt to determine the severity of any memory loss and the time period that it covers. Most quantitative tests of remote memory are composed of material in the public domain that can be corroborated. For example, tests have been used that concern news events, photographs of famous persons, or former one-season television programs. An advantage of these methods is that one can sample large numbers of events and can often target particular time periods. A disadvantage is that these tests are not so useful for detecting memory loss for information learned during the weeks or months immediately before the onset of amnesia. Most remote memory tests sample time periods rather coarsely and cannot detect a retrograde memory impairment that covers only a few months. In contrast, autobiographical memory tests can potentially provide fine-grained information about a patient’s retrograde memory. In the word-probe task, first used by Francis Galton in 1879, patients are asked to recollect specific episodes from their past in response to single word cues (e.g., bird and ticket) and to date the episodes. The number of episodes recalled tends to be systematically related to the time period from which the episode is taken. Most memories normally come from recent time periods (the last 1 to 2 months), whereas patients with amnesia often exhibit temporally graded retrograde amnesia, drawing few episodic memories from the recent past but producing as many remote autobiographical memories as do normal subjects (see Fig. 2.4-4). Memory Self-Reports Patients can often supply descriptions of their memory problems that are extremely useful for understanding the nature of their impairment. Tests of the ability to judge one’s memory abilities are called tests of metamemory. Self-rating scales are available that yield quantitative and qualitative information about memory impairment. As a result, it is possible to distinguish memory complaints associated with depression from memory complaints associated with amnesia. Depressed patients tend to rate their memory as poor in a rather undifferentiated way, endorsing equally all of the items on a self-rating form. By contrast, amnesic patients tend to endorse some items more than

others; that is, there is a pattern to their memory complaints. Amnesic patients do not report difficulty in remembering very remote events or in following what is being said to them, but they do report having difficulty remembering an event a few minutes after it happens. Indeed, self-reports can match rather closely the description of memory dysfunction that emerges from objective tests. Specifically, new-learning capacity is affected, immediate memory is intact, and very remote memory is intact. Some amnesic patients, however, tend to markedly underestimate their memory impairment. In patients with Korsakoff’s syndrome, for example, their poor metamemory stems from frontal lobe dysfunction. In any case, querying patients in some detail about their sense of impairment and administering self-rating scales are valuable and informative adjuncts to more formal memory testing. Psychogenic Amnesia Patients sometimes exhibit memory impairment that differs markedly from the typical patterns of memory loss that follow brain damage. For example, some cases of amnesia present with a sudden onset of retrograde amnesia, a loss of personal identity, and minimal anterograde amnesia. These patients may even be unable to recall their name. Given the psychological forces that prompt the onset of amnesia in these cases, they are commonly termed psychogenic amnesia, or sometimes hysterical amnesia, functional amnesia, or dissociative amnesia. Differentiating psychogenic amnesia from a memory disorder that results from frank neurological injury or disease is often straightforward. Psychogenic amnesias typically do not affect new-learning capacity. Patients enter the hospital able to record a continuing procession of daily events. By contrast, new-learning problems tend to be at the core of neurological amnesia. The main positive symptom in psychogenic amnesia is extensive and severe retrograde amnesia. Patients may be unable to recollect pertinent information from childhood or from some part of their past. Formal neuropsychological testing has shown that the pattern of memory deficits varies widely from patient to patient. This variability may reflect a patient’s commonsense concepts of memory, even when symptoms are not the result of conscious attempts to simulate amnesia. Some patients may perform poorly only when asked to remember past autobiographical events. Other patients also may fail at remembering past news events. Some patients perform well when memory tests appear to assess general knowledge, such as remembering the names of celebrities or cities. Learning new material is usually intact, perhaps because such tests appear to concern the present moment, not traveling into the past. Occasionally, patients with psychogenic amnesia exhibit broad memory deficits such that they cannot perform previously familiar skills or identify common objects or common vocabulary words. By contrast, patients with neurological amnesia never forget their name, and remote memory for the events of childhood and adolescence is typically normal, unless there is damage to the lateral temporal or frontal lobes. Patients with psychogenic amnesia sometimes have evidence of head trauma or brain injury, but nevertheless the pattern of deficits cannot be taken as a straightforward result of neurological insult. The clinician’s challenge is not to distinguish psychogenic amnesia from neurological amnesia, but to distinguish psychogenic amnesia from malingering. Indeed, the diagnosis of psychogenic amnesia can be difficult to substantiate and may be met with skepticism by hospital staff. Some features that argue in favor of a genuine psychogenic disorder include: (1) memory test scores that are not as low as possible, and never worse than chance levels; (2) memory access that is improved by hypnosis or amobarbital (Amytal)

interview; and (3) a significant premorbid psychiatric history. In some cases, psychogenic amnesia has been observed to clear after a period of days, but in many cases it has persisted as a potentially permanent feature of the personality. IMPLICATIONS Memory Distortion Current understanding of the biology of memory has significant implications for several fundamental issues in psychiatry. Given the selective and constructive nature of autobiographical remembering and the imperfect nature of memory retrieval more generally, it is surprising that memory is so often accurate. How much can we trust our memories? Subjective feelings of confidence are apparently not perfect indicators of retrieval accuracy. Moreover, memory distortion can clearly lead to unfortunate consequences, such as when mistaken eyewitness testimony harms an innocent individual. In fact it is possible to remember with confidence events that never happened. For example, it is possible to confuse an event that was only imagined or dreamed about with an event that actually occurred. One factor that contributes to memory distortion is that similar brain regions are important both for visual imagery and for the long-term storage of visual memories (Fig. 2.4-11). Another factor that contributes to memory distortion is that memory functions best in remembering the gist of an event, not the particulars from which the gist is derived. In a noted demonstration, people listen to a list of words: Candy, sour, sugar, tooth, heart, taste, dessert, salt, snack, syrup, eat, and flavor. Subsequently, when asked to write down the words they heard, 40 percent of them wrote down the word sweet, even though this word did not appear in the list. Thus, many people in this demonstration failed to discriminate between the words that had been presented and a word that was strongly associated to all the words but had not itself been presented. The word sweet can be thought of as a gist word, a word that represents the other words and that captures the meaning of the whole list. Presumably, the words in the study list evoked a thought of the word sweet, either at the time of learning or during the memory test, and people then tended to confuse merely thinking of the word with actually hearing it.

FIGURE 2.4-11 Neural substrates of false memories. A. Functional magnetic resonance imaging data were acquired in a learning phase, when subjects read names of objects and visualized the referents. One half of the names were followed 2 seconds later by a picture of the object. B. In a surprise memory test given outside of the scanner, subjects listened to object names and decided whether they had seen a picture of the corresponding object. On some trials, subjects claimed to have seen a picture of an object that they had only imagined. C. Results showed that the left inferior prefrontal cortex and the left anterior hippocampus were more active during learning in response to pictures later remembered compared to pictures later forgotten. D. Several different brain areas showed a greater response to words in the learning phase that were later falsely remembered as pictures compared to words not misremembered. Activations that predicted false remembering were found in a brain network important for the generation of visual imagery in response to object names (precuneus, inferior parietal cortex, and anterior cingulate, shown in the left, middle, and right images, respectively). (Reprinted from Gonsalves B, Reber PJ, Gitelman DR, Parrish TB, Mesulam MM, Paller KA. Neural evidence that vivid imagining can lead to false remembering. Psychol Sci. 2004;15:655, with permission.) The reconstructive nature of recollection means that the interpretation of eyewitness testimony is not straightforward. Whole episodes are not available in the neocortex, but rather must be pieced together based on fragmentary components and in the context of potentially misleading influences present at the time of retrieval. Studies in adults and in children have documented that illusory memories can be created. Children are particularly susceptible to these effects, especially when subjected to leading questions and false suggestions. In view of these features of memory, when a memory for childhood abuse is remembered after many years, it is prudent to ask whether the memory is accurate. Genuine examples of memory recovery have been documented, whereby an individual produces a veridical memory for a past traumatic event after not recalling the event for

extended time periods. Numerous examples of apparent memory recovery have also been subsequently discovered to be instances of false memory. Unfortunately, there is no perfect method, in the absence of independent corroboration, for determining whether a recollective experience is based on a real event. Infantile Amnesia The biology of memory has also provided insights relevant to the phenomenon of infantile amnesia—the apparent absence of conscious memory for experiences from approximately the first 3 years of life. Traditional views of infantile amnesia have emphasized repression (psychoanalytic theory) and retrieval failure (developmental psychology). A common assumption has been that adults retain memories of early events but cannot bring them into consciousness. However, it now appears that the capacity for declarative memory does not become fully available until approximately the third year of life, whereas nondeclarative memory emerges early in infancy (e.g., classical conditioning and skill learning). Thus, infantile amnesia results not from the adult’s failure to retrieve early memories, but from the child’s failure to store them adequately in the first place. Nevertheless, studies in young infants show that a rudimentary capacity for declarative memory is present even at a few months of age. As a child develops, memories can be retained across increasingly long intervals, and what is represented becomes correspondingly richer and more full of detail. Medial temporal and diencephalic regions seem to be sufficiently developed during these early months and years. What limits the capacity for declarative memory appears to be the gradual development and differentiation of the neocortex. As the neocortex develops, the memories represented there become more complex, language abilities allow for more elaborate verbal descriptions of events, and a developing sense of self supports autobiographical knowledge. As new strategies emerge for organizing incoming information, declarative memories become more persistent, more richly encoded, and better interconnected with other information. It is not the case that fully formed childhood memories are stored but cannot be retrieved. The perspective consistent with current understanding of the biology of memory is that declarative memories formed very early in life are fragmentary, simple, and tied to the specific context of an infant’s understanding of the world. They are unlike typical declarative memories in adults, which are imbued with meaning and a complex understanding of events. Memories and the Unconscious The existence of multiple memory systems also has implications for issues central to psychoanalytic theory, including the construct of the unconscious. How one believes that past experience influences current behavior depends on what view one takes of the nature of memory. By the traditional view, memory is a unitary faculty, and representations in memory vary mainly in strength and accessibility. Material that is unconscious is below some threshold of accessibility but could potentially be made available to consciousness.

The modern, biological view begins with the distinction between a kind of memory that can be brought to mind—declarative memory—and other kinds of memory that are, by their nature, unconscious. Stored nondeclarative memories are expressed through performance without affording any conscious memory content. Our personalities are shaped by nondeclarative memories in the form of numerous habits and conditioned responses. In this view, one’s behavior is indeed affected by events from early life, but the effects of early experience persist in a nondeclarative form without necessarily including an explicit, conscious record of the events. Learned behavior can be expressed through altered dispositions, preferences, conditioned responses, habits, and skills, but exhibiting such behavior need not be accompanied by awareness that behavior is being influenced by past experience, nor is there a necessity that any particular past experience has been recorded as a complete episode. That is, an influence from early experience does not require a memory of any specific episode. One can be afraid of dogs without remembering being knocked down by a dog as a child. In this case, the fear of dogs is not experienced as a memory. It is experienced as a part of personality. Furthermore, a strong fear of dogs carries with it no implication that the brain retains a specific record of any early experience that subsequently resulted in a fear of dogs. Behavioral change can occur by one’s acquiring new habits that supersede old ones or by becoming sufficiently aware of a habit that one can to some extent isolate it, countermand it, or limit the stimuli that elicit it. However, one need not become aware of any early formative event in the same sense that one knows the content of a declarative memory. The unconscious does not become conscious. Various forms of nondeclarative memory simply influence behavior without having the additional capacity for these influences to become accessible to conscious awareness. REFERENCES Akre KL, Ryan MJ. Complexity increases working memory for mating signals. Curr Biol. 2010;20(6):502. Byrne JH, ed. Learning and Memory—A Comprehensive Reference. New York: Elsevier; 2008. Crystal JD. Comparative cognition: Comparing human and monkey memory. Curr Biol. 2011;21(11):R432. Gerstner JR, Lyons LC, Wright KP Jr, Loh DH, Rawashdeh O, Eckel-Mahan KL, Roman GW. Cycling behavior and memory formation. J Neurosci. 2009;29(41):12824. Kandel ER. The biology of memory: A forty-year perspective. J Neurosci. 2009;29(41):12748. Kandel ER, Dudai Y, Mayford MR. The molecular and systems biology of memory. Cell. 2014;157:163–186. Lee SH, Dan Y: Neuromodulation of brain states. Neuron. 2012;76(1):209. Lubin FD. Epigenetic gene regulation in the adult mammalian brain: Multiple roles in memory formation. Neurobiol Learn Mem. 2011;96:68. Paller KA, Squire LR. Biology of memory. In: Sadock BJ, Sadock VA, Ruiz P, eds. Kaplan & Sadock’s Comprehensive Textbook of Psychiatry. 9th ed. Vol. 1. Philadelphia: Lippincott Williams & Wilkins; 2009:658. Rösler R, Ranganath C, Röder B, Kluwe RH, eds. Neuroimaging of Human Memory. New York: Oxford University Press; 2008. Solntseva SV, Nikitin BP. Protein synthesis is required for induction of amnesia elicited by disruption of the reconsolidation of long-term memory. Neurosci Behavioral Physiol. 2011;41(6):654.