Learning and Memory
Vocabulary:
Memory: The ability to remember events, learned experiences change the brain through plasticity, it can be used later to make predictions. They aren’t static, can be incorrectly recalled, or dynamically recalled in varying contexts
Engram: The physical traces of a memory, the set of cells active at a time of learning (experimental episode) that are also active during retrieval, undergo changes like LTP together
Active: A memory trace currently retrievable by natural cues, neurons are reactivated when an appropriate retrieval cue is presented
Silent: The engram is there but it is inaccessible by natural cues, retrieval is impaired, Synaptic connections or excitability are weakened
Dormant (Latent): Stored by not currently being retrieved, on ‘standby’, with the right conditions it can become active
Unavailable: Memory traces that have been erased or degraded to the point that even artificial stimulation cannot recover it. Happens via synaptic pruning or neurodegeneration
Contextual memory: Memories happen in physical spaces at a certain time of day under certain emotional states in certain scenarios, or under various contexts for a memory.
Episodic memory: Has contexts, stored in the hippocampus
Hippocampus: Binds spatial and temporal contexts together to form a memory
Hebbian theory: Created by Donald Hebb in 1949, “neurons that fire together wire together”, true for neurons in a certain cell assembly
Cooperativity: There is more than one input at the same time
Associativity: Weak inputs are potentiated when the co-occur with stronger inputs
Specificity: Only stimulated synapses are potentiated
Cell assembly: Neurons that are in the same firing group in Hebbian theory, may be a representation for the memory of something
Tri-synaptic circuit: A loop, first described by Cajal using golgi staining, a coronal section that is a pathway within the hippocampus
Perforant pathway: Entorhinal cortex (EC) to dentate gyrus (DG)
Mossy fibers: DG → CA3, contains unmyelinated axons
Schaffer collaterals: CA3 → CA1
Long-term Potentiation (LTP): The increase in amount and magnitude of EPSPs with subsequent use, basis for Hebbian theory, has to do with AMPA and NMDA receptors. As EPSPs activate the cell, more glutamate enters AMPA, and then the neuron depolarizes. This causes the Mg2+ blockage to leave the NMDA receptor, allowing Ca2+ in, triggering pathways that create more AMPA receptors. More receptors means that it is easier to depolarize in the future. There may also be retroactive messengers to tell the presynaptic cell to send more glutamate
Long-term Depression (LTD): Occurs when pulses that normally cause LTP (quick) are presented slowly instead, or at low intensities. Synaptic connects become weaker. There isn’t enough depolarization to cause the MG2+ blockage on NMDA receptors to leave for each receptors, though some calcium still gets let in by a few that are unblocked, causing a cellular cascade to remove AMPA receptors, ultimately weakening a synapse and making the postsynaptic cell harder to depolarize and activate. There may also be retroactive messengers that tell the presynaptic neuron to release less glutamate
NMDA receptor: Can bind glutamate to depolarize a cell, need a Mg2+ block removed to allow Ca2+ in, which is then used in either LTP or LTD, meaning the use of this can be used as an incidence detector
Mass action: The principle that memory impairment depends on the amount of cortex removed and not the location of where (not true ultimately)
Equipotentiality: The principle that any remaining healthy cortex can take over the jobs of other damaged areas
Place cells: Activate when located in a certain place, in the hippocampus (DG, CA1, CA3), involved in contextual memory
Head-direction cells: Activate when the head is moved in a certain direction, important for spatial navigation, involved in contextual memory
Boundary vector cells: Activate when on the edge of something, example a wall or a cliff, involved in contextual memory
Grid cells: Like a place cell, but mapped along a ‘grid’ that is tessellated, will fire at every “vertex”, maps a room, in the medial EC of the hippocampus, involved in contextual memory. The grid is typically equilateral triangles that form hexagon line planes. For flying animals, this is 3D as well
Time cells: Encode temporal information, allow the brain to track passage of time, in the hippocampus, create a temporal map, help with when something should happen (in the dark, how long does it usually take to get to the other side), NOT clock cells
Morris water maze: Looks into spatial memory, water is opaque and there is a platform, animal swims around to ‘accidentally’ find it and is left there for 30 seconds, they are moved and they have to find their way back
Hypermnesia (HSAM): Only about 60 in the world, remember a very high level of detail about personal experiences, can still experience false memories, contextual memory is highly encoded for all senses so everything is easier to retrieve, a form on synesthesia
Forward replay: The re-occurrence of sequenced cell activations, happens in learning but temporally compressed. Occurs in non-REM sleep and in the hippocampus and PFC
Reverse replay: Happens while awake, disrupting this disrupts learning, memory consolidation
REM sleep: Can help to link memory and facilitate problem solving
Pre-play: Sequences of activations that occur before a memory is made, predicting a sequence
Ribot’s law: Describes the grading of memory forgetting, first in is the last out, more forgetting recent than past
Multiple trace theory (MTT): You know from your other class but assumes old memories are more resistant to disruption because they have made more copies via reactivation
Competitive trace theory: MTT and standard systems consolidation model are both true but it is context dependent and both are right, “it depends”
Prediction error: When something expected goes against what was thought to happen. Positive was better than expected, negative was worse then expected
Schematization: Through multiple exposures and memory encodings, there are created ideas of things. If there are too many memory retrievals, ideas of something may stray further from reality (remembering dead dad 10 years after death, memories start to feel ‘fake’)
Non-invasive brain surgery (NIBS): Can still impact a memory
Stress: Short term (acute) can improve memory consolidation, but long term chronic can hinder it due to high concentration of glucocorticoid receptors in the CA1. Severe cases can lead to atrophy in hippocampal volume, and increased risk of Alzheimer’s disease. Can foster resilience too, complex dynamic of balance
Echoic memory: Sensory memory for sound
Iconic memory: Sensory memory for sight
Working memory: STM that lasts seconds to minutes, happens in the PFC and used for execution of cognitive tasks, obtained from sensory or LTM, used in cognitive tasks (STM isn’t), in the PFC, cingulate, and parietal cortex. Dopaminergic. Controlled through selective attention (choosing the relevant information and focusing on it while ignoring irrelevant data) and active manipulation (processing not just storing). Done in lateral frontal and inferior parietal lobes
Oscillatory synchrony: WM is bound via the same kind of waves occurring at the same time
Wisconsin Card Sorting Task (WCST): Test for response inhibition as tasks are switched, controversial because it more heavily relies on visual processing, led to the hypothesis that there is a PFC deficit in schizophrenia as WM is in the PFC
Acoustic code: The phonological loop, recall of words is better when they aren’t similar as when they are they interfere, showing WM isn’t semantic
Episodic buffer: A temporary storage system that integrates info from phonological loop and visuospatial sketchpad to bind visual spatial and verbal information to each other, giving a sense of time and ling
Dopamine: The gating signal for filtering out other stimuli from focus and attention direction, a learning signal, deficit in ADHD
Implicit memory: Procedural, priming, associative, habituation, cannot be expressed verbally, no conscious access to it
Procedural memory: For motor skills, cognitive skills (like reading), and the ability to form habits (basal ganglia loops with dopamine), grows over extensive and repeated experience, relies on corticobasal ganglia dopamine loops
Priming: Phenomenon where exposure to one stimulus subconsciously influences a response to a subsequent stimulus, can las 2 hours of words, more effective within the same domain
Associative learning: Classical conditioning, learned association from simultaneous presentation yada yada blah blah blah, in the cerebellum mostly
Habituation: The diminishing of a physiological response to a repeated stimulus (babies seeing an infant and being less interested over time). Examples/properties:
If the stimulus is withheld response recovers over time
The weaker the stimulus, the more rapid the habituation, strong stimuli might not show habituation (like really loud noises suddenly)
Habituation stimulus shows generalization, like a rat being calmed by a person and then being given to another person and not being scared of them either
Another strong stimulus is presented and dishabituation occurs
Gated habituation, lessened with reward/punishment
Can be short or long term, needs de novo protein synthesis for long term
Dishabituation: Presentation of a new stimulus reignites interest, happens after habituation
Extinction: Being re-exposed to a feared stimulus without the bad thing, rewriting memory. This could be the same mechanism as habituation, new memory reimpressed on the original, involves inhibitory neurons
Sensitization: Enhancement of physiological responses in response to a repeated stimulus (opposite of habituation), for example yelling at a kid when they leave their socks on the ground and the next time you yell at them for that they cry even more
People:
Cajal: Used golgi staining to describe the tri-synaptic circuit that builds the hippocampus
Richard Semon: Coined engram in 1921
Donald Hebb: Hypothesized Hebbian learning, neurons that fire together wire together
Timothy Bliss + Terje Lomo: Used electrophysiology in rabbits, stimulated the perforant path to show long term magnitude of EPSPs and greater synaptic strength, larger granule cell responses
Karl Lashley: Failed to find the engram by selectively lesioning rat brains and seeing if they could still remember a maze. They could keep finding their way, leading to the conclusion that memories are stored throughout the cortex, defined mass action and equipotentiality
Endel Tulving: Described memory as mental time travel, mechanisms for remembering are the same as imagining, described episodic and semantic as separate
George Miller: Contibuted to chunking and STM storage (7±2)
Patricia Goldman-Rakic: Studied WM and associated it with the PFC, showed it was also distributed across the cortex
Atkinson and Shiffrin: Modelled memory as only going sensory → STM → LTM (Modal Model), though wrong as demonstrated by people with ability to do some but not others (H.M.), memories were showed to be located in different places. LTM but not STM people K.F. and E. E. have damage to language areas (showing language areas are critical to STM)
Baddeley and Hitch: Memory model with central executive and visuospatial sketchpad and phonological loop (you know this from your other class you literally have this exam tomorrow GET IT TOGETHER), phonological auditory and visuospatial for mental images, separate storage, controlled and manipulated by central executive. Phonological left and visuospatial right. Has some neuropsychological evidence and provides evidence for multi-tasking though has vague definitions and doesn’t common on how LTM impacts STM, more for storage and not dynamic processes
Eric Kandel: Did work with sensitization, showed it in aplysia, uses serotonin
Ways to manipulate an engram:
Observational: Look at overlaps between activated neuronal populations at different times, like at learning vs. retrieval
Loss-of-function: Asking is disrupted engram cells impair memory recall, if they do then the cells are a necessity for memory recall
Gain-of-function: Asking if activating engram cells induces recall, if they do then the cells are sufficient for the memory. This technique can do mimicry of a memory, giving an engram a false context and creating a false memory
Stages of memory (you know these based on PSYC 251 be so fr)
Acquisition/encoding for new memory
Consolidation for stabilizing memories, has protein synthesis dependency in the hippocampus
Storage is complicated
Smell goes straight to the hippocampus and skips the thalamus, meaning it can greatly trigger memories contextually
Memories are dynamic, sometimes being used semantically and sometimes episodically. They have to be framed differently based on different scenarios
For a memory to be updated/modified, it has to be activated (retrieved) and new information has to be added, like with prediction error. Old memories are resistant to change (like with PTSD). They need to be flexible and dynamic as our lives are constantly changing.
To change a memory, there must be protein degradation and then protein synthesis
If memories are too flexible they aren’t reliable, but if they aren’t flexible enough we cannot adapt to an environment
Executive functions are planning/organizing, task inhibition, self-monitoring, emotional regulation, goal-directed persistence, time management, and decision making (all things lower in ADHD)