Lecture 4 - Non Declarative and Semantic Memory

what is non declarative memory?

what is non declarative memory?

any form of memory that does not require consciousness and can potentially operate without a person being aware it is in use, involved skill learning and implicit memories, anoetic level of consciousness

how is implicit memory tested?

it can be directly tested by free or cued recall and recongition

incidental learning

the idea that people learn things without even knowing

indirect tests

person is unaware of how precious unconscious memories are influencing later performance such as priming

priming task

expose participants to some type of stimulus and the one that follows is influenced by the previous one

what is the dependent variable in the priming task?

the amount of priming and improved performance at processing a stimulus based on prior exposure even in absence of any memory for initial exposure in terms of accuracy or speed of response time

what is the study phase of the modern memory paradigm?

more focused on incidental learning and usually involved a cover task like focus on a story for instance that is not focused on the stimulus that you intended to manipulate

what is the test phase of the modern memory paradigm?

it wont reference anything in the study phase

naming task/perceptual identification

see a bunch of words in the test phase, show the words and you have to name or explain it and measure the response time to that, it could be a picture or just a word you’re naming

stem completion

give the first piece of the word and ask them to complete it, measure the completion rate

fragment completion

give them some of the letters and ask them to complete the words in terms of the words or the sentence, measure the completion rate

lexical decision task

identify the word as in which one makes sense and measure the response time

implicit vs explicit memory experiment

can separate implicit from explicit memories by reading a word for example broccoli makes amnesic patients easily complete the fragment

direct memory tests

free recall and recognition

implicit memory test

fragment completion test

how did amnesiacs react to the direct memory tests?

the amnesia patients did worse than the controls

how did amnesiacs react to the implicit memory tests?

both control and patient groups did the same showing that implicit memories stay intact while explicit ones don’t showing a separation between the two

automatic retrieval

related to implicit memories, requires minimal cognitive processes so you can immediately respond to something even though you have no cognitive recollection that its there

controlled retrieval

requires conscious attention and control, its explicit

distinguishing implicit memory experiment

participants first had to study a word list then do a fragment completion test there was an inclusion condition and an exclusion condition

inclusion condition

the participants have to complete the sentence with any word they can think of and it could either come from the study list or just stuff you come up with on your own, its automatic so you implicitly remember the study list items without even knowing and then controlled retrieval happens

exclusion condition

same as the inclusion but you exclude the studied words and only say the things you didn’t see, force participants to recall the list in order to respond to it and report the other words so there is controlled retrieval that is happening but also implicit cause you still report some of the studied words by accident

how does implicit vs explicit memory persist over a delay? experiment

in the study phase, studied each word for 5 seconds and there was a delay of either a week or an hour, then do a recognition test for direct memory test and fragment completion for indirect

how does implicit vs explicit memory persist over a delay? direct recognition

accuracy was really good after an hour delay but for the 1 week delay they didn’t remember anything at all

how does explicit vs implicit memory persist over a delay? fragment completion

in the 1 week delay condition and the 1 hour delay some of the words were forgotten but they could complete them if given part of the word as a primer, also free recall was just as strong for words they did not remember seeing as for words they did remember seeing

stochastic independence

performance for one word on one test is independent from the same item on another test

norming task

before providing any type of stimulus you need to test it many many times to make sure there are no other factors getting in the way

nonconscious priming task longitudinal study experiment

people were shown a bunch of different images 3 times and they had to name them then 17 years later they were mailed the fragmented images they were shown previously and measured if they were able to identify them or not compared to a control group who had not been exposed to the images previously

nonconscious priming task longitudinal study experiment results

the longitudinal study group did way better than the control showing that even though they were primed 17 years ago the effect is still significant

stimulus modality experiment

participants had to read or hear a list of words and had to do a recognition test or a perceptual identification task, some people that were in the hearing group were tested visually while some in the reading group were tested verbally switching the modality between test and study phase

stimulus modality experiment results

switching modalities had no effect on the recognition test performance therefore it did not impact explicit memory however for the perceptual identification task it abolished any priming effect showing that implicit memory is perceptually driven while explicit is more conceptual

non verbal perceptual priming experiment

participants were shown possible and impossible objects then they were asked if the object was facing right or left in the study phase, then in the test phase all objects were flashed briefly on a computer screen then they were asked if its an impossible or possible kind of optical illusion

non verbal perceptual priming experiment results

response time was faster for the possible objects and the same pattern was observed in amnesia patients, this is due to perceptually dependent priming so if the object is impossible it doesn’t store anything because there is no overall coherent structure to store

conceptual priming task

in the study phase, the participants had to respond to ambiguous sentences then they were tested at different delays so they solve the sentences again but they are different from the ones at study, they then complete an old/new recognition test

conceptual priming task results

the participants do pretty well overtime but there is still a drop but accuracy is still 75% but for amnesia patients, after a long delay they do worse than the controls but still above baseline so pretty good accuracy which shows that patients with amnesia have intact implicit memory in both perceptual and conceptual priming tasks

neural correlates of priming

pet scans show that there s a blood flow during word stem completion so priming could be associated with reduced activity in posterior perceptual areas cause there is a reduction in neural response with priming then after exposure to stimulus subsequent processing is faster requiring less neural activity that comes with it

repetition suppression

decrease in neural activity for the repeated presentation of the stimulus in part of the brain that is processing the stimulus so in the brain region of interest, it lights up first then at the second presentation not as much, ie habituation

what are the 3 models explaining repetition suppression?

the fatigue model, the facilitation model and the sharpening model

fatigue model

neurons are exhausted so they will not get engaged in the second presentation so less brain activity

facilitation model

repetition causes faster processing stimuli such that it leads to short duration of neural firing

sharpening model

neurons become more optimally tuned so distributed representation becomes sharper resulting in fewer responsive neurons in total

mirror drawing task

participants could only see their hand through a mirror and had to draw an image, those with amnesia managed to improve the more they did it

word trigram task

participants had to read triads of words in the mirror and no words repeated over time, then they measured the mean reading time per triad over a period of days, overtime mean reading time decreased so they were gaining skills, the two groups were doing about the same

korsakoff patients

patients with anterograde amnesia but due to alcohol

memory for motor sequences/sequence learning experiment

participants saw boxes and every time they light up they had to press a button, in the first group the sequence of lit up boxes was repeated while the other group was at random, they evaluated motor skills and looked at response time of these boxes and the experimental ones

memory for motor sequences/sequence learning experiment results

response time in the random one does not decrease cause they cant learn the pattern but for the repeated sequence group response time decreases over time

what is semantic memory?

general world knowledge that is fairly stable and shared with the community, it does not require a self component or mental time travel, conscious and declarative form of memory

what are the functions of knowledge?

facilitates learning and remembering, provides a framework for interpreting and organizing events, it allows us to predict the outcome of things

functions of knowledge experiment

goal is to try to see if having conceptual knowledge is beneficial, baseball experts and non experts read a bunch of fictional baseball games and then had to recall them

functions of knowledge experiment results

experts had better memory for the information overall cause they remembered more relevant information and were better for prediction of outcomes

importance of networks experiment

lexical decision task and the prime word was nurse then the target word doctor for instance and have to say if its a meaningful word or not where the category is shared vs non shared so for instance nurse and potato, response time was measured

importance of networks experiment results

results show that there was a priming effect between response time for shared category level but not for different category priming levels

semantic priming

the facilitation of related ideas, so activity spreads through semantic networks so certain concepts are activated based on what is currently being thought about, if one node is activated everything in association will be activated too

mediated priming/layover activation

activation of a concept or word indirectly influences the activation of a related but not directly linked concept through an intermediate association, for instance lion activates the intermediate tiger which then activates stripes

semantic interconnectivity

the more associations the faster the retrieval because there are more retrieval pathways available

fan effect

more associations causes more interference

automatic priming

priming is more or less an automatic process, not a lot of cognitive effort needed, dominated by implicit memory

controlled priming

where there are controlled pairs and there is activation with respect to a goal and it starts after an initial automatic component 

semantic organization (automatic vs controlled priming) experiment

participants are given category names then followed by lexical decision task and delay between the two was varied, for controlled processing condition they took inherently unrelated concepts and give the participants the expectation that the prime would be followed by a word for another category, in the lexical task they had to say if its an english word or not, looked at priming for unexpected and expected related pairs,

stimulus onset synchrony (SOA)

show the participants something then there is some kind of delay then show them some other thing, the longer the SOA the more time there is for controlled processing

SOAs in automatic priming

looked for it in the short SOAs cause there wouldn’t be enough time to make cognitive effort and control processes would be in longer SOA

semantic organization (automatic vs controlled priming) experiment results

priming (facilitation) was observed at short SOAs cause BIRD automatically activated ROBIN and this was also observed for unexpected related pairs so BODY and HEART, but for the controlled long SOA there was a priming effect only for long delays cause more cognitive effort needed

what are the 2 components involved in semantic priming?

automatic priming which is independent of expectations/intentions, it comes for the semantic network so nothin is really surprising, then there is controlled processing which is more about the expected relationship, it takes time, reflects expectations and can only come at longer delay s

semantic congruity effect

people are faster at picking the smaller of the small items rather than the larger of the big items when comparing items from a series so responses are faster when the direction of comparison is congruent with the end of the stimulus dimension

semantic distance effect

have something that is continuous, faster to respond ot things that are further way from each other so easier to say for instance that mouse is smaller than horse compared to saying goat is smaller to sheep

serial position effect

things at the extremities are faster and easier to compare, so whatever is at the beginning and end of the series, thins at the end tend to be easier than those at the beginning

superordinate categorization

more general (ex; animal)

basic categorization

easiest to think about (ex; dog)

subordinate categorization

specific type of something (ex; golden retriever)

levels of categorization experiment

show a bunch of people images and see if they are able to name them fast or slow (naming latency), results showed that basic categorization had the fastest response time cause it is used the most often, then subordinate is slightly faster than superordinate

classical view of categorization

categories are defined by necessary and sufficient features, it does not explain central tendency, graded membership or family resemblance, its more likely to use probabilistic information and features vary in importance and diagnosticity

necessary

having the feature is required

sufficient

having the feature is enough to be part of the category

prototype theory

the prototype is an average of all category members and central tendency is the prototype itself

prototype theory experiment

there were blue and red dots then participants would have to study the association of one type of dot with a particular category and the other with another one, tested by seeing if they could correctly categorize the prototype at test by taking the shape of the dots and averaging them so once you see the prototype you made see if it fits with the particular category

what is the issue with the prototype theory?

its an average of everything so don’t have the specifics so prototype only conveys  central tendency but not category size, variability or correlation with other categories

exemplar theory

everything you see in the world is an example and you compare them with everything you stored, this is a better method cause you can capture central tendency, family resemblance, information about category size, variability, correlation with other categories and new information about a category

schemas

Knowledge can influence memory by organizing information, really highly organized sort of network, which could be in terms of semantic information or events

what are the 5 ways schemas influence memory?

encoding, selection, abstraction, interpretation, integration and at retrieval

selection

schemas identify which information is more relevant

abstraction

selected info is converted into simpler sort of way so you can remember it a bit better

interpretation

allows you to fill in the gaps really easily

integration

get the information and integrate into your schemas someway and that influences your memory 

reconstruction at retrieval

trying to come up with a whole sentence based on the schema you have available

abstraction study

Showed people images in the middle and they were shown one of 2 labels in the study phase, In the test phase, with the labels given create that vision, extracting things in different ways

schemas and memory bias experiment

• participants told a native american folk tale then later asked to recall the parts from that passage, the participants insert information filling in gaps with english folk tales

• had to wait in graduate student office before experiment began than taken to another room and tested on memory of what was in that room, a lot of the people filled in the gaps and reported office supplies related to that gap

schemas and memory bias fMRI study

looks at how different modern movements were activated looking at activity in brain region, instead of making any movements said to think about the movements and activation similar to when the participants were actually doing the movements, Objects represented in distributed network so like sound, motion, words

semantic dementia

degradation of semantic networks, all modalities of input and output are damaged so you cant understand stimulus or produce a response to it, visual perception and episodic memory are unaffected,

what areas of the brain are affected in semantic dementia?

areas surrounding the hippocampus, so there is atrophy in the temporal lobes and that spreads posteriorly and medially with disease progression

prunin of semantic tree in semantic dementia

hierarchical structure of the network in our brain, we start losing information for the bottom and it moves up as disease progresses so loss of specific and basic level of categorization but superordinate still intact 

semantic dementia in patient JL

longitudinal study showed deterioration of semantic knowledge, had to name a bunch of images back to back and got it right ½ the time when initially diagnosed, a year later far less accurate and only good for basic level then the following year confused between basic level and could only do well with general categories, typical items took longer to deteriorate in memory

target usual vs unusual information in semantic dementia patients

so a usual item would be a green vegetable while an unusual one would be an orange one in severe dementia patient the typical items last longer in memory so longer for them to deteriorate than unusual items

how is semantic dementia assessed?

word to picture matching, drawing object to name so which object matches with another, which tool to use, sound to picture matching or knowledge of correct colors

semantic hub

the anterior temporal lobe, category specific regions connect to a hub where associations across modalities are integrated

episodic memory in semantic dementia experiment

in the study phase there was an object picture naming task and in the test phase there was an episodic retrieval task where out of all these objects they had to pick the ones they saw in the study phase, in the mixed retrieval task they were asked if they recognized a slightly different example of what they saw

episodic memory in semantic dementia experiment results

controls did well for both types of retrieval tasks, in semantic dementia they are good with episodic retrieval task but for mixed retrieval they do significantly worse, for alzheimer’s there is a drop in performance for both retrieval tasks

double dissociation task

Trying to see if semantic dementia patient can be separated from alzheimer’s, Showed 2 different trees and see which one is more likely to show up in each environment, then say which one they see

• results shows that in alzheimer’s they could correctly which one of those is actually fitting into the picture but for episodic memory they dont do well, with semantic dementia the opposite is true showing that declarative memory is not supported by a single system