Lecture 1 ENGRAM Episodic Memory Formation 2021-2022 PDF

23-9-2021 Performance Mechanisms of Long-Term Memory Consolidation Retrieval Loss of information c 2001 Sliwuer AMOchm. Iik . 1 2 Different Kinds of Memory * What about semantic learning? 3 4 Patient HM: Summary Episodic Memory: Patient HM Background -epilepsy patient (documented by Bre...

23-9-2021 Performance Mechanisms of Long-Term Memory Consolidation Retrieval Loss of information c 2001 Sliwuer AMOchm. Iik . 1 2 Different Kinds of Memory * What about semantic learning? 3 4 Patient HM: Summary Episodic Memory: Patient HM Background -epilepsy patient (documented by Brenda Milner, Sue Corkin, and others) Brenda Milner -bilateral removal of medial temporal lobe (1953), including amygdala and most of hippocampus, by William Scoville -symptoms after surgery epilepsy was alleviated but: memory impairment 5 6 1 23-9-2021 Removal hippocampus leads to deficits in EVENT memory -No recall of events more than a few minutes ago. Examples: What is the date? No recognition of doctor. Forgetting the question… -Impossibility to memorize new events, and deficit to recall events that had occurred up to three years before surgery. -Thus: - complete anterograde episodic memory deficit - graded retrograde deficit 7 8 Some long-term memories are spared: Skill learning Was this memory deficit for events a total surprise? (a) Ilie mirror-tracing task (bi Performance of HM. on mirror-tracing task Epilepsy attack is often preceded by -olfactory sensation -mood change -motoric phase -grand mal or petit mal -gross motor movements, or complicated motor behaviors Execution of complicated behavioral patterns indicates retrieval from memory No new memories (of events during epileptic episode) are formed 9 Skill learning does not depend on hippocampous, instead, skill representations include changes in topography in sensory areas 10 Lots of types of memory formation are spared -FACTS memory (semantic) & distant EVENT memory (autobiographical/epsiodic) -ability to learn new facts (facts regarding a person, new vocabularly) -ability to learn new skills (e.g. a perceptual discrimination) -normal intelligence -normal reasoning abilities -short-term memory Long-term memory formation is not a monolithic ability! 11 Nudo et al., 1996 12 2 23-9-2021 Who was Patient HM ? Some anatomy Episodic memory requires binding of disparate foci of cortical activity. Because of the extensive connectivity of amygdala and hippocampus with all cortex, both structures could play a central role in the encoding of episodic memories Henry Molaison Henry Gustav Molaison, known widely as H.M., was an American memory disorder patient who had a bilateral medial temporal lobectomy to surgically remove the antenor two thirds of his hippocampi. ... Born: February 26, 1926, Hartford, Connecticut, United States Died: December 2, 2008. Windsor Locks, Connecticut, United States Full name: Henry Gustav Molaison Education: East Hartford High School 13 14 Hippocampal-cortical interactions through circuit of Papez Hippocampal-cortical interactions gyrus 15 16 Hippocampus in humans Locsno" o’ cut Hippocampal-cortical interactions through Entorhinal cortex Mossy Fiber Pathway Schaffer Collateral Pathway 17 18 3 23-9-2021 Hippocampus is an evolutionary old structure, which is comparable across different species Hippocampus, sea horse 19 Roughly the same anatomical divisions in rat, monkey, and man 20 Information Flow Mossy Fiber Pathway Schaffer In Collateral Pathway In Out 21 22 Hippocampus and amygdala are both connected to whole brain and thus welll suited to play a central role in memory I. Amygdala: -a mass of grey matter next to the hippocampus. -afferents and efferents: -cortex -hippocampus -hypothalamus -olfactory bulb *efferents to ventral striatum II. Hippocampus: -primitive cortex at the fringe of medial temporal lobe -afferents:-cortex (predominantly entorhinal) -amygdala, hypothalamus Amygdala-cortical interactions -unidrectional flow through the hippocampus: -dentate gyrus  CA3  CA1  Subiculum 23 -efferents: -subiculum to entorhinal cortex -CA3 to cingular cortex (through fornix & anter thal nucl) -amygdala 24 -hypothalamus 4 23-9-2021 Standard model of episodic memory consolidation (Larry Squire) The ‘standard’ theory of episodic memory consolidation 1) Initially, hippocampal linking of neocortical storage sites 2) Through a slow consolidation process (years), neo-cortical storage becomes ‘stand alone’. 3) The more remote a memory, the less hippocampus is needed ( graded retrograde amnesia). 4) Hippocampus always needed for new episodic memories ( anterograde amnesia). 25 26 Testing predictions from the standard theory of episodic memory formation Brain regions activated during event Which structures in temporal lobe produce the deficit? An animal model of amnesia Time… Time… Hippocampus 27 28 Concurrent Object Discrimination Testing retrograde, graded amnesia (WMkS) Fig. Z Retention of 100 object discrimination problems learned approximately 2, 4, 8, 12, and 16 weeks before hippocampal surgery (20 pairs per tune penod). Retention was assessed 2 weeks after surgerv in monkeys with lesions (H ) (O) (n 1 1) or after an equivalent interval in unoperBnctao stow (•’ uca ammaU 0» standard error of the mean. - 29 Stuart Zola-Morgan, Science, 1989 30 5 23-9-2021 Delayed Non-Match to Sample Testing anterograde amnesia 31 Amygdala and Hippocampus contribute equally to memory formation 32 Findings: 1) Retrograde, graded amnesia 2) Anterograde amnesia 3) Amygdala and Hippocampus both contribute Conclusion:  Standard theory has been confirmed Do you believe this conclusion?? Fig- 5 Comparison of groups in present study with those in an earlier study [8] trained and tested in an identical manner. Curves show average postoperative performance on recognition task with extended delays (left) and list lengths (right). Numerals to the left of each curve indicate the average -nber of trials (T) each group required to relearn the basic task Abbreviations: N, normal control bilateral amygdalectomy; H, bilateral hippocampectomy; A +h. bilateral amygdalectomy plus unilateral hippocampectomy; H+a, bilateral hippocampectomy plus unilateral amygdalectomy; A+H. combined bilateral amygdalectomy plus hippocampectomy. Saunders et al., 1984 33 34 Could effects be due to damage to surrounding parahippocampal and rhinal cortex?? Anything suspicious here? More effective Less effective And that’s the reason 35 36 6 23-9-2021 A rhinal lesion alone Meunier et al., 1993 37 Meunier et al. (1993) found that Rhinal lesions produce the deficit which was thought to be caused by A+H lesions. They also suggest that perirhinal cortex explains most of the deficit caused by rhinal lesions. 38 Thus, hippocampus and amygdala do not produce the anterograde memory deficit on DNMS task. To confirm, selectively lesion the hippocampus and the amygdala. Ixaiupk* image* u»«d lo plan surgeries. Sagittal in tot onal sccimni III. In sagittal section. the Mere© carbar 12), the mcdioialcral Icsel al which the hippocampus was targeted (3), the angk of approach <M. the deepest point of injection (5). the point at which the needle exiled the hippocampus (6), and the point at which the needle exiled the skull (7). See text for details. 39 HGl Rt I. sinus M shown taxii R. Hampton, Hippocampus, 2004 40 Murray & Mishkin, J Neurosci., 1999 Figure 9. Group mean scores on the DNMS performance test. The curves on the left show the effects of imposition of increasingly longer delays between simple presentation and choice, whereas the curves on the right show the effects of list-length testing. For each list item the minimal delay is 20 sec X the length of the list. SELECTIVE A 11. Monkeys with bilateral exdtotoxic lesions of the amygdala and hippocampus; NORMAL CONTROL, unoperated control monkeys. 41 Murray & Mishkin, 1998 42 7 23-9-2021 What is episodic memory and did the tests used tap this ability? Do the results from this animal model indicate that the hippocampus is not important for episodic memory formation ? Typical of episodic memory is that the learning is incidental and fast Learn discrimination AB in context X Learn discrimination CD in context Y Retention of AB is better in context X than in context Y Retention of CD is better in context Y than in context X Retention in familiar context is better (even though the learning was incidental) Interference by unfamiliar context In a rat lesion study (Honey et al.), this context effect disappears after hippocampus lesions In the monkey studies, memory formation required training. Tests tapped skill learning / object recognition 43 And what is context? It certainly includes spatial information! Ventral Stream Incuts Area TE Area TEO 44 Let’s have a look at studies of spatial processing in rats Dorsal Stream Inputs Retrosplenial Ctx Posterior Parietal Ctx Dorsolateral Prefrontal Ctx STSd Spatial information is the glue for event-memory… 45 Place cells 46 Wilson & McNaughton (1993): A hippocampal ensemble code for space. Nadel & O’Keefe, 1978 Experiment: ensemble recording in CA1 during explorative behavior Main findings: 1. Pyramidal cells in CA1 show place fields 2. Sparse coding: -In any location, few cells are active Some of those have overlapping place fields -Exploration of the total environment is covered by a small subset of neurons in the recorded population 3. Population states (within recruited population) are correlated with the animal's spatial location in a particular environment 4. Accurate prediction of the animal's location is possible purely on the basis of population state. This confirms the role of the hippocampus for spatial encoding. "... opens the possibility of the interpretation of neuronal activity in the absence of explicit behavior, such as during periods of sleep..." 47 48 8 23-9-2021 Recording from 80 hippocampal cells at once… So, hippocampus codes for space Relevance for episodic memory? Hippocampal units get converging, multimodal neocortical input and thus can be a link between distributed cortical ensembles that are co-active. During hippocampal bursting activity (while sleeping), neo-cortical neurons linked to hippocampal units form cortical links (Hebbian learning, consolidation). As a result of consolidation, a hippocampal-cortical network is formed: Hippocampal units represent a location or a spatial context, to which a rich set of experiences/events/episodes/landmarks is linked. Relevant for episodic memory formation, and contradicts classical theory of episodic memory consolidation! Wilson & McNaughton, Science, 1993 49 50 Again: Classical Consolidation Theory of Episodic Memory Neurons become linked sleep (consolidation) Neurons activated during event Odor info Odor info Visual info Visual info Stand alone memory trace Odor info Visual info Alternative: The multiple trace theory of episodic memory consolidation (Nadel) Time… Time… Iterative process of feedback and cortical input Hippocampus Neurons Spatial info 51 52 Multiple TraceTheory of Episodic Memory General Principles Neurons become linked sleep (consolidation) Neurons activated during event Final memory trace 1) Hippocampal codes (which represent spatial information) provide necessary, independent indexes for different memories 2) Hippocampus is necessary for retrieval of even the oldest memories Odor info Odor info Visual info Time… Visual info Odor info Visual info Time… 4) Retrieved memories are also re-stored Iterative process of feedback and cortical input Spatial info 3) Retrograde amnesia should not be ‘graded’ after complete hippocampal damage 5) If hippocampus is part of a hippocampal-cortical ensemble that stores episodic memories, then hippocampus activity might in part reflect local cues, landmarks, object properties, in addition to spatial information Hippocampus Neurons 53 54 9 23-9-2021 Testing predictions from the multiple trace theory of memory formation Molecular evidence for independent codes, making use of time course of IEGs Arc 55 o 2001 SMauer Amoclate*. lot 56 Molecular evidence for independent codes 57 58 Guzowski et al., Nature Neuroscience, 1999 Hippocampus remains necessary for retrieval of even the oldest episodic (autobiographical) memories Loft Hippocampus Right Hippocampus Ryan et al., Hippocampus, 2000 59 Mean (SEMI activations in left and rivht hippocamFIGURE J. pus for recollection of recent event*, remote event*. and rest, from three labjectt who were unaware of the event cue* that were to be pretented in the tea oner 60 10 23-9-2021 Reactivation and reconsolidation: Multiple Traces for an episodic memory? After complete hippocampal lesions, retrograde (autobiographical) memory deficit is ‘flat’, not graded TRAINING / EVENT Controls Training in a task Amnesics - o amnesia news details A Ok performance Reminder (the type of reinforcement, a conditional stimulus, or a cue or context, on a single trial) Reminder (the type of reinforcement, a conditional stimulus, or a cue or context, on a single trial) Better performance Amnesia ECS, hypothermia, protein synthesis blockers Training in a task Ok performance ECS, hypothermia, protein synthesis blockers contols news details controls - news delate At Tout number of druil. frody recalled .born per Sonal episodes and news esmu rhal o.tor red before sod alter unncsis Itnaual. Pre-5 refers to childhood, Pre- 2 to adoleueuce, and Pre-I to carle adulthood- Hr Total number of details recalled with prompl. FIGURE S. Training in a task -•-amnesic personal detail -^-controls personal deists -o amnesic news details smncKCT - peraona- delate -a- controls - personal details RECALL Training in a task ins about personal episodes and newt events that occurred before and after unrirau (trauma). Pre- 5 reden Ur duldluaod, Pre- 2 to adoles¬ and Pte-1 to early adulthood. 61 cence, Nadel et al., 2000 62 SJ Sara et al., 2000 Some evidence from molecular biology for re-consolidation Some evidence from molecular biology for re-consolidation A role for LGs Using the anti-sense technique BDNF, ZIF-268 Anti-sense RNA o 2001 SMauer AmorWea. Ine 63 64 3h 24h Consolidation You will read about this… Bottom line: different genes for consolidation and reconsolidation Paradigm: Less time freezing = forgetting 65 JLC Lee et al., Science, 2004 fore conditioning with BDNF antisense ODN. Impaired LTM (24 hours after conditioning) freezing, with intact STM at 3 hours, in rats infused before conditioning with BDNF anti¬ sense ODN (1 nmol/p.1) (black, n = 7) relative to BDNF missense ODN (white, n = 7). Repeat¬ ed-measures analysis of variance (ANOVA) re¬ vealed a group X test interaction (Fp.u) = 5.820, P < 0.05). BDNF antisense ODN (black, n = 15) administration reduced freezing levels at the LTM test (ANOVA, *P < 0.01) relative to the BDNF missense ODN (white, n = 14) group. 66 11 23-9-2021 Hippocampus codes not only spatial information… but also landmark/object information Re-consolidation is affected by pairing with Zif268 antisense, but not by pairing with BDNF antisense B Cond 48h 24h > LTM -> PR-LTM D Cond 4 24h > LTM — 48h PR-LTM 4 BDNF Zlf268 Landmark information: Recording in different recording boxes placed in same room Separate LGs for consolidation and re-consolidation: Strong evidence for multiple trace theory 67 68 Leutgeb et al., Science, 2005 Conclusions All Predictions from multiple-trace theory are supported 1) Hippocampal codes (which represent spatial information) provide necessary, independent indexes for different memories 2) Hippocampus is necessary for retrieval of even the oldest memories 3) Retrograde amnesia should not be ‘graded’ after complete hippocampal damage 4) Retrieved memories are also re-stored 5) If hippocampus is part of a hippocampal-cortical ensemble that stores episodic memories, then hippocampus activity might in part reflect local cues, landmarks, object properties, in addition to spatial information 1) Hippocampus is part of a cortico-hippocampal ensemble that represents episodic memories 2) The hippocampus shows fast plasticity (thus, one-trial learning), which allows it to guide slower cortical consolidation 3) The hippocampus orchestrates cortical binding and indexing, hippocampal-cortical networks representing memories, according to principles of multiple-trace theory 4) Even similar activity patterns in cortex representing episodic memories can be distinguished by independent sparse codes in hippocampus 69 Note: Amygdala 70 And some extra thoughts The amygdala is recruited into the memory trace when there is an emotional component (e.g., fear) (so still plays an important role in memory) 71 1) In case of procedural memory storage (skills, semantic material) other subcortical structures interact with cortex during consolidation. 2) These ‘guiding’ structures (e.g., basal ganglia), as well as cortex, lack fast plasticity (simplifying…) 3) Therefore, practice is needed. 4) Cortical skill memory traces that use same population of neurons, instead of forming indexed, differentiated memory traces, compete and interfere with each other. 5) The distinction between neocortex and hippocampus may be related to the differences in ease with which LTP can be induced (easier in hippocampus) 72 12 23-9-2021 Additional issues and topics The end 73 74 Structural Plasticity in the Hippocampus Long-Term Potentiation: A model for hippocampal and cortical learning? (a) Sizes of home range Timothy IIH> 75 77 76 EA Maguire Papers 78 13 CIPL (cm) 79 A shrinking anterior hippocampus, and a growing posterior hippocampus in taxi drivers? 83 81 Children born without hippocampus, show normal language development, i.e. Normal semantic memory (Vargha-Kadem et al., 1997) Performance of old vs young rats in Morris Water Test The hippocampus and episodic memory in Alzheimer’s disease Clinical relevance of hippocampus research Hippocampal cross-sectional area (mm2) Maze Learning The effect of aging on spatial / episodic memory First visit Second visit Instability of hippocampal representations in old rats Gallagher et al., 1995 Percentage of animals reaching criterion 23-9-2021 80 82 84 14 23-9-2021 Young Conclusions 1) Hippocampus is part of cortico-hippocampal ensemble that represents episodic memories 2) The hippocampus shows fast plasticity (thus, one-trial learning), which allows it to guide slower cortical consolidation 3) The hippocampus orchestrates cortical binding and indexing, according to principles of multipletrace theory 4) In case of procedural memory storage (skills, semantic material) other structures are involved to guide binding and consolidation. 5) These ‘guiding’ structures (e.g., basal ganglia), as well as cortex, lack fast plasticity (simplifying…) 6) Therefore, practice is needed. r (within episodes) r (within episodes) 85 Wilson and McNaughton (1994): Reactivation of hippocampal ensemble memories during sleep. Experiment: Ensemble recording in CA1: - During pre-exploration sleep (PRE) - During explorative behavior period (RUN) - During post-exploration sleep (POST) 86 Cross-correlation 1.2 POST Cross-correlation 1.3 10| ii 300 -400 10' 0 400 3CO Time lag (ms) Cell 1 Cell 2 Overlapping fields Cell 1 Cell 3 Non-overlapping SCIENCE • VOL. 265 • 29 JULY 1994 87 - Wilson & McNaughton, Science, 1994 88 Wilson and McNaughton (1994). Reactivation of hippocampal ensemble memories during sleep. Experiment: Ensemble recording in CA1: - During pre-exploration sleep (PRE) - During explorative behavior period (RUN) - During post-exploration sleep (POST) Main findings: PRE: not many high correlations RUN: increased number of high correlation in the firing of neurons with overlapping place fields. POST: re-activation of the correlation pattern observed during RUN-phase. --> Expression of some type of memory 89 90 15 23-9-2021 Hoffman &McNaughton, Science, 2002: Correlation between cortical ensembles during sleep (i.e. during hippocampal ensemble reactivation)? Relevance for episodic memory CA1 units get converging, multimodal neocortical input and thus can be a link between distributed cortical ensembles that are co-active. A test in the monkey During hippocampal bursting activity (while sleeping), neo-cortical neurons linked to CA1 units form cortical links (Hebbian learning, consolidation). As a result of consolidation, a hippocampal-cortical network is formed: CA1 units represent a location or a spatial context, to which a rich set of experiences/events/episodes/landmarks is linked. Crucial support for this hypothesis: during consolidation CA1 activity of neurons with overlapping place fields becomes correlated - Correlation is not due to within-hippocampus processing of spatial information. - Why? CA1 units show no lateral connections. - Therefore, correlation between two CA1 units with overlapping place fields must reflect correlation among cortical inputs. 91 92 A Rest 1 Task Rest 2 WK 10 min 10 min F" 30 min 1 5-40 min 4000 2000 0 2000 4000 Time (ms) B 30 min -4000 2000 0 2000 4000 Time (ms) 93 94 95 96 16 Visual info Hippocampus Neurons Time… Odor info 97 101 99 Iterative process of feedback, cortical input, feedback Visual info Neurons become linked sleep (consolidation) Relevance for Episodic Memory Odor info Neurons activated during event Iterative process of feedback and cortical input Spatial info Independent codes for different memories (O’Reilly and Rudy, 1999): Requirements: - Hippocampal activation pattern should not be a direct reflection of the stimulus input. - Hippocampal code must be capable of re-activating stored representations in a precise way. CA3: generates random codes because of diffuse cortical projections  this allows random codes to be associated with a cortical input pattern  (it does not allow re-activation of that input pattern, because of the diffuse back projections) CA1: Receives random code from CA3, and gets very precise (topographical) cortical input  CA1 code becomes linked to cortical input pattern  it is CA1 that is instrumental for memory retrieval General Background The hippocampus contributes to the formation of fast incidental conjunctions between very different stimulus representations (episodic memory). Cortex contributes to the slow, incremental formation of conjunctions between more similar stimuli in learning tasks (semantic and skill learning). 23-9-2021 98 100 102 O’Reilly 17 23-9-2021 Odor info Odor info The major hippocampal pathways Neurons become linked sleep (consolidation) Neurons activated during event Visual info Time… Iterative process of feedback and cortical input Hippocampus Neurons in CA1 CA1 population activity In a location is a random code generated in CA3 Visual info Time… Iterative process of feedback, cortical input, feedback Correlations not due to within-hippocampus ‘lateral’ connections! 103 104 The setup: Stimulation of commisure and Mossy fiber pathway The rest is up to you  105 106 18