Cog Neuro Exam 2 (Notes)

Why do we need memory?

Memory allows us to retain information for future use.

Types of Shorter forms of Memory

• Sensory Memory

• Working Memory

• Short term memory

Types of Long-term memory

• Declarative Memory

• Non-Declarative Memory

What is Sensory memory

high capacity, very short-duration memory that happens between sensory systems

• within this short window, it can be as effective as having attention directed to the information in advance

Types of Sensory Memory

• Iconic memory (Visual) → lasts less than 500ms

• Echoic Memory (Auditory) → can last up to 10s

What is Iconic Memory

• Sensory memory associated with visual systems

• lasts less than 500ms

• As iconic memory decays → there is a slow decaying in V1 response

What is the capacity of short-term memory

• ~7 item

• longer duration of 15-30s

What is working memory

an expansion of the concept of short-term memory to include manipulation of maintained information

• recalling information that is no longer present

How does selective attention influence sensory memory

it can boost the amount of items you retain

• if a cue is given within the first 100ms of a delay (after an image has been shown and removed), a person can perform just as well as if they were given the cue in advance

  • any more (900ms) cuts a person’s visual sensory memory capacity in half → benefits of sensory memory lost; can only recall what was encoded into WM

Describe Atkinson & Shiffrin’s Modal Model of Memory

A: Sensory Register

B: Short-term storage

C: Long-Term Storage

D: Attention

E: Rehearsal

What does Atkinson & Shiffrin’s Modal Model of Memory propose?

• a serial structure (hierarchical): attention shifts info from sensory memory to STM, then rehearsal shifts from STM to LTM

• Proposes a unitary STM store: any type of information all goes in one short-term memory box

What are the two problems with Atkinson & Shiffrin’s Modal Model of Memory?

1. found patients that have impaired STM but intact LTM (vice versa: impaired LTM, intact STM)

   • patient HM had impaired STM and intact LTM

     • serial structure doesn’t make sense if you can bypass STM and encode into LTM

2. STM impaired patients had a lot of selectivity in their impairments (e.g. impaired visual STM but intact audio STM)

   • pushes back on unitary STM store; all types of information isn’t being encoded in the same place in the brain

Is the Atkinson & Shiffrin’s Modal Model of Memory widely adccepted?

no, it has been dismissed

Describe Baddeley & Hitch’s Tripartite Model of Memory

A: Central Executive

B: Phonological Loop

C: Episodic Buffer

D: Visuospatial Sketchpad

E: Articulatory Loop

F: Acoustic Store

What does Baddeley & Hitch’s Tripartite Model of Memory propose

• separate short-term memory stores for acoustic information (phonological loop) vs. visual/spatial information (visuospatial sketchpad)

• a control centrer that arbitrates between and coordinates STM subsystems (central executive)

What is the evidence for separate STM stores for different types of information?

1. when holding a word list in mind with an imagery strategy (imagining a picture of the word-list), you remember fewer words if you are simultaneously tracking visual movement (or performing any other type of visual task)

   • similarly, if you are using a rehearsal strategy (verbal repetition), you remember fewer words if you are simultaneously repeating nonsense words (or performing any other type of verbal task)

2. looking at patients with brain damage: if you give different types of task (spatial span task or verbal version of spatial span task) people with brain damage in different areas struggle with these tasks: dissociation is most important (selective impairement)

   • Damaged parieto-occipital regions (especially right hemisphere) → impairment in spatial span task; intact on other tasks

   • premotor and supramarginal gyrus → impairment in verbal version of spatial span task; (speech production and perception still intact)

What is the spatial span task?

1. Theres a bunch of different blocks on the table

2. the experimenter taps them in a specific order

3. participant waits for 7-10s

4. participant has to tap the boxes in the same order

tests visuo-spatial sketchpad

What is the verbal version of the spatial span task

1. Participant hears a bunch of letters in a row

2. DELAY: they wait 7-10s

3. Then, participant has to repeat back the letters in the same order

tests phonological loop

What is working memory?

• contains the concept of short-term memory (+ more things)

• analogy: your mind’s blackboard

• TWO COMPONENTS

  • memory component (maintenance) → representation of information that used to be there but is no longer there, that has to be maintained

  • working component (manipulation) → some type of manipulation or updating given your particular task goal; mentally manipulating information + anytime you have to re-organize or update information in working memory

What is working memory necessary for?

• holding a goal in mind

• linking relevant pieces of information together to make inferences

• directing attention and action towards the goal

• connecting them over time (e.g. strategize, plan)

Classic working Memory Tasks (Maintenance)

No tasks ask to manipulate information, just to maintain it

• Delayed Match to Sample

  • presented with stimulus

  • image goes away

  • delay

  • participant presented with a choice of stimuli, asked to select the correct one

• Visual Array

  • presented with an array of visual items

  • array goes away

  • delay

  • asked question about the stimulus

    • asked which stimulus is correct (i.e. is this color correct?)

    • asked to reconstruct stimulus or array

• Digit/Letter Span

  • Presented with a list of digits or letters sequentially

  • participant must correctly reconstruct at the end

Classic Working Memory Tasks (manipulation)

• Backwards span

  • same as digit/letter span but participant is asked to recall/reconstruct the span backwards

• N-back task

  • see a stream of stimuli (what the stimuli is does not matter)

  • asked to press a button anytime the same stimulus appears after N times

    • N = 1 → one back: anytime the stimuli appears back to back

    • N = 2 → two back: responding when there are 2 letters between the same stimulus

      • A is shown, two other letters are shown in between, A is shown again (press button)

  • typically N = a high number (3 or 5)

  • higher N backs are harder because participant has to store and recall a string of information at once and manipulate said information rather than just one (for 1-back)

    • on every trial you have to update whats in your working memory

Explain the involvement of the PFC in working memory

• PFC Activity persists during working memory delay even after the extenral stimulus is removed

  • scientists found (in monkeys) that there is a lot of persistent firing in PFC neurons during working memory delay even after external stimulus is removed

    • not related to perception, not related to action explicitly

    • even when delay period is longer, the neurons firing still persists

    • Lled people to believe PFC is involved in working memory maintenance

• PFC activity increases with load (when there is more information to maintain) and stops when the item is no longer needed

• Persistent activity seems to predict/be associated with WM performance in humans

  • higher spikes for when something needs to be remembered correctly

• PFC Lesioned Monkeys

  • PFC-lesioned monkeys become impaired at recalling location of food (stimulus that needs to be remembered) that is out of sight unless there is a learned cue

    • learning cues involves a different type of memory system (one not affected by the lesion)

• Performance on WM tasks improves dramatically if room is darkened during delay period → this is because there is a reduction of visual stimuli (less competition of visual stimuli when trying to remember)

What does PFC persistent activity represent for memor?

• It was initially thought to be the storage of the information (maintenance of location or specific stimuli)

• however current evidence suggests its more about control, selection, and task goals and not about the sensory content itself

• leaves question: if the content that is being maintained is not in the PFC then where is it? → the sensory cortex

What brain regions is the sensory content being maintained in?

• the sensory cortex

What is the sensory recruitment hypothesis?

the idea that the same regions that primarily process information maintain it during working memory

• if this is true then we should be able to see evidence in sensory systems (like the visual system) of maintenance of information over working delay period

Gating Mechanism of Working Memory

• brain faces dilemma:

  • stability (gating out) → keeping current information stable + protected from distractions + noise

  • flexibility (gating in) → allow new information to enter and replace old information

• If the brain updated constantly → working memory would be too unstable

• if brain never updated → it would be too rigid and unable to respond to new information

• so brain needs a control system (a gate) that decides when to let information in and when to keep it out → basal ganglia as the gate (basal ganglia also involved in motor skills)

  • basal ganglia decides when that memory should be updated

What is Automatic Behavior

Behavior that we do without thinking:

• Habitual, obligatory, rapid, and stimulus-driven

• Often engaged without consideration of future consequences 

What is controlled behavior

behavior that is:

• Internally guided, slow, deliberative, consistent with goals and broader context

• Harder to do under load/fatigue

• Typically feels effortful

What are the areas of the prefrontal cortex

• Lateral prefrontal cortex

• Frontal pole → very front, underneath your forehead

• Orbitofrontal cortex → right behind your eyes

• Medial prefrontal cortex (including anterior cingulate cortex)

What does the PFC do

• most widely interconnect part of cortex

• Heteromodal association cortex

  • PFC must receive inputs from all secondary sensory areas in order to perform its tasks related to cognitive control

• Sends widespread feedback projections that can influence processing in other brain systems

• Regions of brain that continue expanding latest in development (adolescence)

• PFC show mixed selectivity

What is mixed selectivity

• means individual neurons respond to multiple tasks/stimulus features or combinations of features rather than single features

  • this is useful because it allows for multiple stimuli to be processed by single neuron increasing efficiency of PFC

Phineas Gage

• (1849) Foreman of railroad construction crew. Triggered dynamite charge which sent the tamping rod he was holding through his head.

• Survived, but accident destroyed parts of prefrontal cortex.’

• Changed personality. Couldn’t hold down job, became vulgar, impulsive, and temperamental.

What are the effects of Frontal lobe damage

• Environmental dependency

  • Imitation

  • Utilization → if they see an item that typically triggers a specific type of behavior they will act on that despite inappropriate environmental contexts

• Perseveration → trouble stopping oneself from repeating an action

  • Wisconsin card sorting task

    • Will keep using same sorting rule even if they know the rule is wrong

• Poor planning abilities

• Disconnect between knowledge and action

  • Know what they are supposed to do but have a hard time carrying out these actions in correspondence with what they’re supposed to do

• Loss of goal-directedness

• Exhibit socially inappropriate behavior

  • Often commit crimes or do behaviors that would go against social norms

    • Rage

• Similar behavior to children

• Most metrics of cognitive ability in tact

  • Vocabulary, language function

  • Semantic and episodic memory

  • Perceptual processing

  • Motor function

  • IQ score in typical range

What are the two concepts under inhibitory control?

• response inhibition → the ability to stop, interrupt, or abort responses

• inference control → resolves competition between incompatible responses or representations

  • sources of attentional control → direct attention to relevant information (dorsal/ventral attention networks)

  • sites of attentional control → targets for modulating relevant information processing (eg., V1, V4, MT)

What are two tasks that test response inhibition?

• Go/no-go task → push a button when certain visual stimuli appear (go trials) and withholds response to other stimuli (no-go trials)

  • No-go stimuli typically has x in middle

  • Most of responses are go responses and it is hard to inhibit pressing button the few times the no-go stimuli is shown

• Stop-signal task → respond as quickly as possible to a stimulus, unless there’s another signal (e.g., auditory tone) soon after, in which case stop yourself

What areas are involved in successful stopping?

interactions between inferior frontal gyrus(IFG), pre-SMA, the basal ganglia, and the primary motor cortex

• Baseline activity in motor cortex determines whether stopping is ultimately successful

damage to the inferior frontal gyrus (IFG) impairs stopping

• Circuit mediated through pre-SMA and basal ganglia

How is the inferior frontal gyrus involved in stopping or response inhibition

• engaged for all stop trials (successful and failed) ⇒ recruitment of inhibitory control

• damage to IFG impairs stopping

Thought Suppression task

Think/no-think task:

• method:

  • participants are cued to either retrieve a previously learned association (“think”) or to actively prevent it from coming to mind (“no-think”)

• Conclusion:

  • Suppressing word associations during encoding leads to decrease in hippocampal activity, and poorer recall

  • this activity is mediated by lateral PFC (IMPORTANT)

What does damage to lateral PFC impair?

filtering of sensory signals

What is the Flanker task?

• Press a left or right button depending on whether the middle arrow is pointing left or right

• Supposed to ignore the flanker arrows

• No conflict when middle arrows and flanker arrows are pointing the same direction

  • Congruent, no control needed

• Conflict when middle arrows and flanker arrows are pointing in opposite directions

  • Incongruent, interference control needed

How is control and automatic behaviors involved in the stroop task?

• More automatic pathway to say the color the word spells out rather than the color of the font

  • Stimulus features that support conflicting responses (incongruent) can result in poorer performance (slower, less accurate responses) than response-congruent stimuli

    • When the ink color and the word itself conflict, then the response time is slower and responses are less accurate

  • Control can support processing of the task-relevant feature

    • Inhibit automatic response to do what the task is asking

• inferior prefrontal cortex (IPFC) → biases you in favor of reading the word rather than saying the color of the ink

What is the Anterior Cingulate cortex?

• “ACC” refers to the Anterior Midcingulate cortex (aMCC) or dorsal regions (dACC)

  • linked to sensory, motor, and affection control

• ACC is connected to lateral PFC

What does the ACC do

• ACC is involved in monitoring for conflict

  • many situations require us to engage control involve conflicting responses

  • just monitor conflict levels to determine when to increase control → Conflict needs to occur for ACC modulation to happen

• One of the most debated regions of the brain, but a common view is that it links monitoring & evaluation with control

• monitors conflict

• when it detects conflict it communicates to the LPFC to increase control

How do we test whether the ACC is involved in detecting conflict

• measure brain activity in ACC when doing tasks involving high conflict

• observed that activity scales with conflict for those tasks

Explain adapting to conflict ***(needs to be edited)

• People often perform better on an incongruent trial if the previous trial was also incongruent vs. if it was congruent

• These conflict adaptations have been tied to interactions between ACC and LPFC across trials. Resulting in increased control and decreased conflict for trials where there was a previous conflict

What are the ERPs related to error detecting

• ERPs from moments where an error is made during a case of conflict are averaged over time

  • Error-related negativity (ERN) occurs ~100ms after the response is made (prior to/without feedback).

  •  Its source has been localized around rostral ACC (also error-related fMRI signals)

explain post-error slowing

when response time slows in new trials after trials where errors were made; occurs without correction

How do we adapt to errors

•  Errors often lead to increased caution on subsequent trials → slower and/or more accurate responses

• Post error slowing → slower, more cautious response time in new trials after trials  where errors were made; occurs without correction
  

•  ACC does error monitoring

  • Evidence: stimulating or inhibiting ACC with tDCS increases or decreases ERN and changes post-error behavior.

Define and explain task switching

Task switching → the process of reconfiguring control in the PFC to represent a different goal than the one you previously had

• Don’t know how its done but know the effects

• Task switching does not come for free

• Reaction times (RT) are slower and error rates are higher on switch trials compared to repeat trials (“switch costs”).

•  Patients with PFC damage may be completely unable to switch

How do the ACC monitor the LPFC in reference to Hierarchical representations of control?

• ACC may parallel hierarchy observed in LPFC

  • more rostral → abstract goals

  • more caudal → specific action based goals

• Need to have a parallel gradient of ACC that maps the abstractness or specificity of the goals: ACC monitoring signals may parallel to control hierarchy in LPFC

  • Anterior (abstract) → posterior (specific/action based)

  • anterior cingulate cortex (ACC) → monitoring system that tells the LPFC when it needs to adjust control

    • monitors PFC to tell when to give signal to adjust control → this process is paralleled to the gradient of abstractness of the goals in the different regions of the PFC

How are goals represented in the PFC

• PFC has a hierarchy of organization to represent all goals simultaneously

  • This maps onto an organized hierarchy

    • Anterior regions of PFC → abstract goals and task

    • As you go more posterior in PFC→ goals that are more concrete tasks (more directly related to actions)

How does control work in reference to goals?

• Control doesn’t create (transmit), it modifies (modulates) existing pathways/representations!

• To modify appropriately, control requires information about the current context and goals

  • Have to have a representation of the current context in order to be able to choose what things to do

What goals are represented in PFC

• The problem: what goes do we even represent

  • The most detailed form of the goal: e.g. typing password to get to notes

  • Or more general: e.g. doing well on the exam

Answer: we should represent all the goals

What is the working definition of goals

• a persistent neural state in the PFC that can bias downstream regions

Adjusting control proactively vs. relatively

• Rather than relying on signals during a task (e.g., conflict, errors) to reactively adjust control, it often makes more sense to allocate control proactively

• If you know you are going to do a hard task, why wait to update control after instead of doing it before

• Evidence shows that we can adjust control proactively

  • If more trials are going to be hard → increase control

  • If more trials are going to be easy → decrease control

• Better conflict adaptation when expecting more conflict

• When there are more incongruent trials → people perform better over time than when there are less

• Better performance with higher incentives → If you pay people for performing the task better they perform better 

  • Evidence for the idea that control can be adjusted ahead of time

Define decision making

a commitment to an action or beleif on the basis of evidence; how we extract evidence until we land on a decision

What are the two highly studied forms of decision making in neuroscience

• perceptual decision making → which response best fits the sensory information that is coming in through my eyes

• value based decisions → which response fits my preferences

What task is used for perceptual decision making?

Random motion task → cloud of dots shown and participant is supposed to make a decision about what direction the dots are moving in

• This task is also done with monkeys → track their eye movements to see what the decide on

What is the drift diffusion model

overview:

• popular form of evidence accumulation model

• used to predict peoples’ choices and how long they take to make those choices

graph

• x-axis → time

• each line (blue and red lines) represents one trial: a decision that you might be making

  • lines also indicate the amount of evidence you have towards a motion decision

• each of the two decision you could make are the top and bottom line

• middle line: decision threshold

reading the graph

• start in the middle

• stimulus comes on → you’ll see bits of motion come into your visual system

• gonna start accumulating evidence towards a left-ward or right-ward motion decision

• once you cross horizontal line → mean you have passed decision threshold (have enough evidence to make decision)

analysis

• the steepness of the line tells you how fast or slow the decision making process was

• NOT ALL DECISIONS ARE THE SAME

What is evidence for the sensory recruitment hypothesis

• FMRI study proved this

  • paradigm:

    • presented with stripes that are oriented diagonal

    • given different set of stripes

    • told which stripes need to be maintained

      • diagonal → 1st

      • OR other → 2nd

    • presented with another stimulus and asked to say whether it matches with the first/second one or not

  • results:

    • on a plot, we see persistent activity in visual cortex (goes down but not all the way back down) during this time

      • people have devised tests to see what information is being encoded during that activity

        • computer trained to use visual cortex activity to distinguish when subjects are looking at one orientation vs another (while participant is looking at stimulus)

        • test during WM delay while theres nothing on the screen

        • if that model works it suggests that the delay period activity is organized in the same way → computer model had 70% accuracy

  • conclusion:

    • suggests that the content of stimuli is in that brain activity observed in the visual cortex

    • responses in the visual system seem to track the information that is trying to be maintained

Analyze this Graph

• graph is recording of neurons in area MT of monkeys as they random dot motion task

  • y-axis: firing rate

  • x-axis: time since the dots come on the screen

• color gradient shows how MT neurons respond to the motion stimulus depending on how coherently in one direction the dots are moving

  • dark → bright red = increasing motion coherence

    • 0% coherence → random motion

    • 50% coherence → dots are clearly moving in same direction

      • 50% of dots are moving in one direction

• takes 50 milliseconds(ms) for activity to ramp up → thats just the time that it takes for information to travel from visual processing to MT

  • once it ramps up → activity is stable

• conclusion: the higher the motion coherence the higher the firing rate

  • when more dots move in the same direction, MT neurons fire more vigorously

  • Firing in MT scales with the amount of motion you are perceiving

  • not really decision making → this is just perception but this information is gonna feed forward to other brain areas that are important for decision making

random dot motion task (with monkeys)

• monkeys look at cloud of dots

• dots will go away and monkeys eyes will either go to left target or right target depending on which way they believe the dots are moving

explain the face/house perceptual decision making task

perceptual decision making task

• paradigm: pictures of houses and faces are overlayed in a confusing way

  • intermediate stimuli that look a little bit like a face and a little bit like a house

• Graph:

  • if you look at face-responsive visual areas like FFA and house-responsive visual areas like PPA → see smooth gradients of activity:

    • the more face like it is → more FFA activity

    • more house-like image is → more PPA activity

PPA and FFA are tracking the amount of perceptual information in the stimulus

• not really decision making → this is just the input for decision-making

what are the important concepts of decision-making

• evidence/evidence accumulation → accumulation of information supporting each of the possible choices

• decision threshold → the threshold that needs to be reached, once enough evidence has been accumulated, for a decision to be made;

  • once you’ve accumulated enough information to pass this threshold, the decision has already been made, you just need to make the response

• noise → evidence we accumulate is affected by random factors

all of these concepts are part of an evidence accumulation model, specifically, a drift diffusion model

What are the ways that decisions vary

drift rate → the speed at which we move toward a decision threshold (how fast we made the decision)

• the slope of the line:

  • steep slope → high drift rate; getting evidence at a fast rate

    • you hit the decision threshold faster

    • higher drift rate → leads to faster and more accurate responses;

      • if you accumulate information faster → reach the right information faster → make a response faster

  • shallow slope → slow drift rate; getting evidence at a slow rate

decision threshold → how much information/evidence is needed in order to make a decision (distance between bounds) can vary

• can differ between people and context

• even if drift rate is the same → amount of time to make decision changes if the bounds are changed

• higher thresholds → lead to slower and more accurate responses

response bias

• bias towards one decision:

• offset at beginning that puts you closer to one of the bounds than the other → more likely to respond + will be faster to respond in that way

Neural basis of perceptual decision-making

MT

• signals strength of motion in a given instance → momentary evidence towards a decision

  • strong motion → higher MT firing

  • weak motion → lower MT firing

• NOT evidence accumulation (does not show accumulation of evidence over time)

LIP (lateral inter-parietal area)

• shows accumulation dynamics

• DONT see a stable response over time

  • steep slope → most coherent motion

  • shallow slope → lower amounts of motion

• LIP neurons are accumulating evidence of the information they get from MT until they reach some type of threshold that signals a decision

  • usually studied in monkeys

• LIP part of intraparietal sulcus on the lateral parietal sulcus

  • Intraparietal sulcus different on monkeys than on humans

Evidence for MT/LIP circuit for perceptual decision making

• Lateral Intraparietal Area (LIP)— it's a region of the parietal cortex involved in decision-making, spatial attention, and the accumulation of sensory evidence over time (as shown in that drift-diffusion style graph).

• if we mess with MT → should impact the evidence accumulation process in LIP (because it should be getting evidence from MT and accumulating it overtime)

example

• if we use electrodes and stimulate MT neurons that care about rightward motion → we should see that information accumulates in LIP for rightward motion decision more quickly

  • has downstream effects on what the monkey chooses and choice RTs

example graph

• red line → LIP with fake rightward motion activity (due to MT stimulation)

  • looks equivalent to if there was more visual stimulus showing rightward motion even though there is none

  • even when there’s noting there, because we’re adding stimulation, you are more likely to make rightward motion decision

• blue line → LIP control (with not interference)

  • control → if there is no distinct leftward or rightward motion, you should always pick 50/50

what are the brain regions that code for value in a common currency (across stimulus attributes)

these regions track different kinds of rewards and code them into some universal value signal

ventral-medial prefrontal cortex (ventral mPFC)

• overlapping definition of orbitofrontal cortex

ventral striatum

• striatum includes the caudate and the putamen

• nucleus accumbens (NAcc)→ part of basal ganglia

  • more related to reward

Describe this value tracking task

• start with a value case:

  • start with two stimuli and you just have to pick one

• learn through trial and error that one stimuli gives you a lot of money and one gives you not very much money

• can track how much value people assign to the stimuli + track brain activity

• can do the same for social reward

  • one stimulus gives you a picture of someone smiling at you, other gives you a picture of someone frowning at you

• In both cases, if you look at activity in ventral medial PFC regions → see more activity for both social value and for monetary value

  • evidence for these regions caring about value in some type of common currency

Describe the ‘willingness to pay’ paradigm

paradigm

• ask someone how much they are willing to pay for an item

• people give you a number

• later you can make people make choices

  • do you want ‘X’ item or ‘___’ value of money

  • do you want ‘X’ item or ‘Y’ item

• very easy to show that peoples willingness to pay scales nicely with whether they choose that item later

• works for all type of stimuli

  • money, food, random objects, etc.

brain activity

• in PFC → more activity for more willingness to pay for these items (regardless of what the items are)

making same point that there is this common currency for value

what is the homunculus problem?

• the PFC controls behavior by directing attention, selecting actions, and overriding impulses but what controls the PFC?

  • if you say “a higher executive system,” then who is controlling that system → leads to infinite regress of controllers

What is the dilemma for tasks like the N-back task? How may it be solved?

problem:

dilemma between maintaining info (gating out) vs. updating (gating in) information in working memory: stability vs. flexibility

• can be broadened to other WM tasks (both in lab and in real life situations)

solution:

this gating problem may be solved in art by similar gating mechanisms as motor responses (i.e., basal ganglia)

• basal ganglia is also important for opening gates in working memory; deciding when new information is going to come in or when its not going to come in

• gating in the motor system can be extended to cognitive concepts like WM

what is long term memory

systems that can encode, store, and retrieve information over long periods of time from minutes to a lifetime

• long can mean anything here → anything greater than working memory

• unlike working memory → information doe not have to be “active” to be remembered for LTM

• LTM is high capacity (much higher capacity than WM)

• usefulness of LTM → provides a way to exploit the relationship between the past and future

  • things that are remembered from the past can be useful in the future/present

• subdivided into declarative vs. nondeclarative

what are the two subdivisions of LTM

• Declarative (explicit) → you can talk about what you know

  • e.g. memories of your life, facts (you can tell someone these are things you remember)

• Nondeclarative (implicit) → things you don’t have access to through your language or verbal system

  • e.g. knowing how to ride a bike (things you just do without real thinking or capacity to explain to someone else)

What are the two major subbranches of declarative memory

• episodic memory → reflects experiences in your life for specific events, objects, places, specific things that happened

  • contextualized with yourself at the center (memories, self-related experiential knowledge)

• semantic memory → reflects generalized knowledge or facts

  • just knowing facts, decontextualized, detached from where they were learned, not a specific episode

what is encoding

storing things in memory

what is retrieval

bringing memories back to mind

what is consolidation

changing memories over time

Give an overview of Karl Lashley’s work

Project:

• wanted to find out where episodic memories were stored

• used mice going through mazes, lesioned their brains to try to see where episodic memories (that helped them navigate the mazes) were stored

Conclusion:

• the more he destroyed of the mice’s brains, the worse their memories were

• concluded that memory traces were distributed across the brain not localized

  • this conclusion was unsatisfying because it seems incorrect

Give and overview of Wilder Penfield’s contribution to knowledge about episodic memory

• Penfield was a prominent neurosurgeon in the 1950s

  • invented a lot of modern neurosurgical techniques

project:

• would inject electrical activity into the brains of patients while they were awake and under anesthesia → would try to understand what different parts of the brain did

conclusion:

• found that sometimes when he stimulated it seemed to evoke things like memory

• seemed to be happening when he stimulated the temporal lobe: possible memory traces in the temporal lobe

Give and overview of patient HM’s injury, cause of injury, and associated deficits (pattern of memory loss)

overview:

• H.M experienced seizures in childhood that became severe as a teenager

• underwent experimental surgery in his 20s that resected his bilateral medial temporal lobes because thats where a surgeon thought HM’s epilepsy was originating from

result:

• surgery was successful in reducing the severity of his epilepsy but his memory was profoundly impaired

HM’s pattern of memory loss:

• had anterograde amnesia + retrograde amnesia (specifically temporally graded retrograded amnesia)

• his early memories seem to be intact (same as healthy individuals)

• HM’s deficits were remarkably specific to episodic memory → could not form new episodic memories in any modality (spatial, visual, verbal, etc.)

• however all other cognitive abilities were intact ( normal working memory, semantic memory, procedural memory ( speech understanding.. etc.)

• due to impairment in hippocampus → could recall semantic aspects of memories but not episodic memories

Describe HM’s performance on Mirror Tracing Task

Task

• participants have to trace a shape without looking at hands directly → can only at hands through looking in a mirror

• Had HM perform this task over repeated sessions

outcome

• HM shows the same pattern of improvement as a normal person but just doesn’t remember any of it

conclusion

• episodic memory is distinct from other forms of memory

What composes the medial temporal lobe

• perirhinal cortex (PRC)

• entorhinal cortex (ERC)

• parahippocampal cortex (PHC)

all of these are direct inputs/outputs to the hippocampus

• hippocampus is evolutionarily old and conserved across species + looks like a seahorse

Identify the structures highlighted in blue and green

green → hippocampus

blue → MTL cortex

• made of the entorhinal cortext, the perirhinal cortex, and the parahippocampal cortex

What are the three phases of episodic memory

1. encoding → the process of converting sensory information into a memory trace

2. retrieval → the process of retrieving information from memory storage

3. consolidation → stabilization of encoding information into long-term memory stores; some type of change of the stored information in some way—something about the storage is changing over time

What is a subsequent memory paradigm

seeks to answer question

• if the hippocampus is important for memory (which is what we think from HM), can we find evidence that activity in the hippocampus predicts whether you are going to remember something or not?

Task

• individual is sitting in fMRI scanner during the task

• three trials where each trial a pariticpant is presented with a different word

• measure brain activity while participant sees each of these words

• later, memory task is given where participant is asked whether a word from the trials was seen

  • if person says yes → they remembered

  • if person says no → they forgot

• responses of the memory tests are taken and trial brain activity is labeled based on whether participant later remembered or forgot the word

results

• we find a few different brain regions where

  • get more brain activity for subsequently remembered items than for subsequently forgotten items

• brain activity located in the MTL and PFC

conclusion

• MTL (and PFC) involved in successful encoding

how does MTL support successful encoding

plays a role in binding (hippocampus)

• memories have a lot of different features (people, place, time, internal context like emotion)

• combination of features of what make the memories unique

• idea is that hippocampus may play a particular role in binding all of those features together into one episode

• Binding of items in context (BIC) model

pattern separation → keeping similar memories separate

explain the Binding of item in context (BIC) model

• the neocortical areas (the rest of the brain) have representations of what an wear (object and spatial representations)

• these representations get fed forward into the perirhinal and parahippocampal cortex

• both the PHC and PRC converge into the hippocampus

• hippocampus is in a good spot to

  • get all in of these different type of information

  • bind all the information together

How do we actually know whether the hippocampus is important for binding

task

• have people remember words across two different contexts

  • task contexts tell you what you should do with the word

    • place task-context→ you should try to imagine whatever the word is

    • read task-context → read the word backwards in your head

  • purpose of the different contexts is to create multiple components

    • now you have to recognize whether something happened or not + remember which task it happened in

two stage memory test

• put people in a scanner while they are doing this + do the same type of memory test except there is one extra component

  • item recognition: did you see this word or not

  • context memory: which task did you see this word on

• requires you to have bound that information together, not just recognize that its old

• based on what they respond you can call the word:

  • forgotten → they forgot the word + the context

  • item only → remember the word but not the context

  • item + context → remember the word and the task they did when they saw the word

• we are interested in the distinction → if hippocampus is important for binding, then it should respond more for item + context conditions

  • perirhinal cortex does not seem to distinguish item + context from other conditions, just cares if you remember anything at all

conclusion

• hippocampus supports successful binding of multiple aspects of an experience during encoding

how does the hippocampus keep memories separate?

• two similar memories activate the same/similar neurons in the visual cortex because they have lots of similar features

pattern separation:

• but the hippocampus has a very separate way of coding things → even though the memories are very visually similar they are going to activate different neurons in the hippocampus

How does the MTL support succesful encoding?

hippocampus is the main reason; hippocampus is involved in both

• binding together all of the co-occuring features of an event

• pattern separation (keeping similar memories separate)

Which of the following best describes the logic of the subsequent memory paradigm used in fMRI studies of memory encoding

Brain activity during encoding is sorted based on whether each item is later remembered or forgotten

Which of the following describes the contribution of different MTL structures to encoding?:

A. Perirhinal cortex activity during encoding predicts later item memory but not recollection of contextual detail

B. Perirhinal cortex predicts later recollection of contextual detail

C. Hippocampal activity during encoding predicts recollection of contextual details

D. A and C

D. A and C

when is hippocampal activity the highest during encoding

when you can remember both the item (the word) and the contextual items (the task you were performing when you were encoding the word)

when was perirhinal activity the highest during encoding?

showed a boost just if you remembered the item (the word); activity was also high if you remembered the item and the contextual details (the task you were performing when you were encoding the word)

Describe the process of encoding and retrieving things from memory

Encoding

1. features of an experience are encoded as a pattern of activity in the sensory cortices (visual, auditory, etc.)

2. the hippocampus binds all of this information together

   • hippocampus serves as a pointer to the other systems of the brain, indexing the features of the memory

Retrieval (if the hippocampus has done this binding successfully)

1. Later when the hippocampus gets a cue (a reminder of that memory) it pattern completes (hippocampal pattern completion) the rest of the memory:

   • the whole pattern of the memory is reactivated in the hippocampus

   • hippocampus is recreating that pattern of cortical activity

2. that is going to reactivate the sensory features of the memory in cortex

   • at least some parts of the cortex that were active during encoding of a memory are reactivated (reinstated) when that memory is retrieved

what is hippocampal pattern completion

when the hippocampus receives a cue (partial reminder of the memory)

What are the two processes that occur when you reconstruct an entire memory experience from a cue or a reminder

Hippocampal Pattern Completeion

• a partial cue that reactivates part of the memory is going to engage this auto-complete process where the entire memory pattern is recreated in the hippocampus

• once the memory is recreated in the hippocampus it can be recreated in cortex

Cortical Reinstatement aka Cortical Reactivation

• The same cortical neurons that were activated when you experienced the event get activated when you recall it

What do we see in the sensory systems during cortical reinstatement? What is the evidence in support of our conclusion?

We see activity in the sensory systems while people are remembering things (i.e. faces, sounds, etc.)

Example 1: Auditory + visual regions

• had people look at images + listen to sounds → saw activity in primary sensory areas during this condition

• saw weaker activity in the same sensory areas when participants were asked to recall the images/sounds they’d been shown previously

Example 2: FFA + PPA

• if we show people blocks of faces and houses → FFA responds to faces, PPA responds to houses

• when we have people recall faces → FFA becomes weakly activated again (but only when we have people recall faces)

• when we have people recall houses → PPA becomes weakly activated again

conclusion

• retrieval of visual vs. auditory memories reactivates visual vs. auditory regions.

• retrieval of face vs. location content reactivates FFA vs. PPA

• when we bring things back to mind we are engaging the same cortical regions as when we first experience them

• cortical reinstatement pattern of reactivating sensory regions is widespread across many cortical regions