Lecture 6 - Future Directions in Memory Research

what is episodic memory?

what is episodic memory?

Form of memory that allows an individual to recollect happenings from his or her past (very subjective), according to tulving memory for personally experienced events that occurred in a particular place at a specific time

chimp vs human working memory experiment

test where 1 to 9 numbers are randomly scattered across the touch screen and the chimp had to memorize it, when they touch a number the rest of them are hidden and they have to remember the layout

chimp vs human working memory experiment results

humans only got the pattern right once after 30 tries but chimps remembered the numbers 90% of the time which makes them a good model for human memory

morris water maze experiment

pool filled with water and rodents are put in it then they go quickly to the platform, first its random search but then after some time they learn where the platform is located

what are the 2 views of how animals remember?

the first is that they only have noetic memory so they dont experience recollection while the other view is that they remember where the platform was in the water previously so they use that to get to the platform the next time more accurately

episodic like memory in monkeys study

4 chimps are trained to use a joystick and computer stations in their cages where they can practice, then there is a delayed matching to sample task, in the first they are shown pictures then need to identify which one they saw

delayed matching to sample

look at what, where and when, get feedback and reward every time they react

spatial delayed matching to sample

object on the right or left repeated several times then they have to say which side they say the object in the retrieval phase

temporal delayed matching to sample

chimps trained to know which item corresponds to 1 second vs 10 seconds 

episodic like memory in monkeys study results

50% is the chance level cause we always have 2 options, all the chimps are doing very well above chance in all the experiments, in the second part all 3 experiments are combined in order and where and what are doing way better than time still but with the when component they are doing worse compared to when the experiments were done separately

another way to test when experiment

look at different pictures everytime and timing is up to them, the chimps are trained to know that if they tough it they get a reward, all chimps are shown 2 pictures in the test phase and have to decide which one came earlier (temporal relationship, symbolic distance

symbolic distance

how far apart the images are spatially, 0: no images between, 1: 1 picture in between, 3: only one option you can do image 5 or 1 so there are many images in between

another way to test when experiment results

If the tested images are further apart it is better and if the earliest image they are responding to is the most recent they do better, shows that maybe interference rather than time past might be the reason for forgetting 

another way to test which/context component experiment

the what component is cube vs cylinder, where component is left or right and which component is black vs mesh room, all rodents are exposed to both rooms and get to explore for 2-5mins then they are tested

another way to test which/context component experiment results

show novelty preference so they will spend more time exploring the new thing but there was no difference in novel and old item in exploration phase, in controls they do lots of exploration of novel box while in lesioned animals they explore the novel item the same amount as the studied item showing that if hippocampus is lesioned the what, where and which memory is impaired

scrub jay experiment for episodic like memory in what, where and when components

scrub jays are very good at storing and caching meal worms in different places, worms are good but go bad quick while peanuts not as good but last long, in the 1st trial they got peanuts then 120h go buy then they are caching the worms then 4hrs go buy so they should pick worm cause fresh, in 2nd trial they first cache worms then 120hrs go buy then peanuts then 4hrs go buy and predict they chose peanuts cause worm has already gone bad

scrub jay experiment for episodic like memory in what, where and when components (degrade condition)

learn that worms naturally degrade and after delay they inspect more for nuts

scrub jay experiment for episodic like memory in what, where and when components (replenish condition)

acts as control, learn that worms are replenished so they don’t decay so they control for possibility of forgetting of worms because they have seen them a long time ago

future planning in animals experiment

3 different compartment and most of the time in B, food is not cacheable, In compartment A they would have breakfast but in compartment C no breakfast, after training unexpectedly one evening they are giving food, in compartment C they should store the food that they are given unexpectedly cause they know there won't be any food the next morning 

future planning in animals experiment follow up

in compartment A they always get dog kibble but peanuts on mornings  they were confined to compartment C, In evening they are given both foods to cache, If they jays are able to forward plan they should cache different food rather than the same food because they prefer variety

future planning in animals experiment results

episodic like memory that is related to the past and forward planning because they do seem to cache different rather than the same compartment so the predictions made were observed

what is a schema?

a whole network of things you can call on, influences memory encoding, retention and retrieval, they are influences by new experiences and takes into account commonalities across experiences that are emphasized while unique aspects are weighed down

what are the pros and cons of schemas?

they are associative networks but they encode in a single episode so you just see it in one shot then you are done with it, single shot may not be important

evidence of shared schemas

similar networks are engaged across people during movie watching, looking at the same thing and thinking about different things but activation across people is the same showing that people share schemas

statistical structures triplets

3 things that follow each other, create different ones to show people and ask them which one comes first or after

adaptation

concept of creating the entire schema, learning or the process through which we learn how to accommodate and assimilate

assimilation

new information is added to the existing schema

accommodation

modification or updating of schema to account for new information

what brain regions are involved in schema formation?

the medial prefrontal cortex and the hippocampus, both activated when the brain is exposed to new information

how does congruency work between the MPFC and the MTL?

the medial frontal cortex is inhibiting the medial temporal lobe because information is congruent because MPFC has the right mechanism for it so it doesn’t need help from MTL

how does incongruency work between the MPFC and the MTL?

if information is less congruent the hippocampus will help the MPFC make the schema cause it loves to bind things but if its incongruent MPFC might not even be active so there is no schema and the hippocampus will takeover and make connections between the two things

compete

MPFC activates relevant schemas, computes congruency and if its congruent it will inhibit the hippocampus

cooperate

MPFC and hippocampus will complement each other and hippocampus will store relationships among events

phase 1: schema acquisition

If the controls are the rodents with intact hippocampus they will do well but when you have the lesion they won't be able to dig for the correct location cause they are not familiar with the layout of the event arena anymore

phase 2: schema updating

digging and the are getting better as time goes on because they are getting better at detecting the location of the flavored food pellets, if they are in the non cued condition they aren’t prompted to remember a specific location, after training they are introduced to 2 new associations with 2 new rewards, then they are tested one day later

phase 2: schema updating (results)

new locations can be learned rapidly because they can be incorporated into a familiar layout, schemas speed up consolidation, then they lesion hippocampus and in the controls as well as hippocampus lesioned rodents they do the same with the new associations because the 1 day delay was enough to consolidate the information into the schemas

schemas on processing information

for congruent information there would be more brain signals and only takes 300ms after stimulus presentation to see neural ERP difference between incongruent and congruent picture

schemas supported by hippocampus (MPFC) experiment

A and B trial and see both at the same time with hippocampus binding them in someway, then BC trial same thing, then AC transfer where you infer the relationship between the 2 items with the common one that is B so it requires associations across event and look at neural responses

schemas supported by hippocampus (MPFC) experiment results

C should not be activated during the AB trial, at 2nd repetition there would be a change in neural signal referring to you activating your schemas, in the 3rd repetition even more of the scenes would be activated, hippocampus and MPFC regions show greater communication across repetitions, fMRI evidence for reinstatement of associated information during encoding

MPFC necessary for interference study

look at lesion map and warm colour indicates more lesions, AB and BC structures encoded and ask about AC connection, then look at proportion correct so saying that A goes with C’ then look at ventral medial PFC lesion to see where they do worse

VMPC

place that is necessary for building, storing and/or using schemas to interpret new information

MPFC generalizes knowledge experiment

in the incongruent trial that would be ice cream with cactus then in the test phase show them a slightly different cactus and ask if item is old or new and the idea is that similar items should create false memories so not focused on item related information

MPFC generalizes knowledge experiment results

normal controls are making more errors if the items were congruent than incongruent, those with MPFC lesion are saying old to everything showing they don’t get the false memories and we know this region makes schemas so they wont be focused on activating schema but rather looking at details of the dessert to figure out if its old or new, VMPC patients are not impacted by congruency including showing no congruency related increase in false memories like controls cause they aren’t actually activating schemas which results in more accurate memories

features of schemas

they support encoding of new related information, speeds up consolidation, influence of schemas on information processing is rapid and potentially automatic, depends on hippocampus and medial prefrontal cortex and allows behavioural generalization