Memory Notes PDF PDF

Preview

Summary

Lecture notes under Dr. Daniel Berstien, fall 2018...

Description

MEMORY PS3251! TUES 1-4 *MC lecture Written - readings *for each class, be prepared with 3 MC questions from previous lecture to put on the exam *be prepared with the readings for each class LECTURE 1: (sept 4) Reasons to care about memory: - thank about our past & sense of self what causes forgetting? what makes things memorable? *match your learning environment to your retrieval environment (state dependent memory) - encoding specificity & transfer appropriate processing Principle of Memory: - Predict and Control (future) History: - Mnemosyne (Greek Goddess of Memory) >500BC credited as the first philosopher possessing the gift of reason and logic. Also provided people the power to store experiences. Gave birth to the 9 muses. - Norse God Odin (God of wisdom war battle etc etc) and the RAVENS Huginn (thought) and Muninn (memory) - memory is the form of mental life. to the GREEK - to the norse, aspect of human life but not the core of humanity Historical Models of Memory: - Aristotelian view: ! recall operates according to three principles: 1. contiguity 2. similarity 3. causality - rationalist view: Operations performed on memory content are innate and abstract in nature - empirical view: Contents of memory are acquired through sensory experience - Constructivist view:! contents of memory acquired through sensory exp. and subj. to idv interpretation; forgotten info can be inferred

Modern views of memory: - computer view: memory involves the manipulation of concepts through a higher order control process - connectionist view: memory is manipulated through interconnected processing of varying strengths - NO higher order of control process Research approach: verbal learning approach info processing approach methodological approaches to mmm research - experimenting - observation - computer simulations - mathematical models - neuroimaging techniques Philosophical Origins: John Locke: blank slate, tubula rasa; all knowledge is acquired through experiences Wilhelm Wundt: uses INTROSPECTION to have people report the ATOMS or elementary components of sensory experiences. The classic approach is MEMORY IS LIKE A MOLECULE Memory is like understanding matter. Introspection: can we learn about memory about how we ourselves use memory? a problem with this is that it’s hard to verbal some things. psychological process are often inaccessible to our consciousness. Conscious exp. is not simply receiving info its an active organization of input. we are more like problem SOLVERS than problem RECEIVERS bc people actively organize sensory info, the info will vary from one person to the next. WHICH MEANS FAILURE TO REPLICATE better ways to study memory… - experimentation - take objective measurements Hermann Ebbinghaus: used CVCs (i.e. WOP), used paired-associate learning; forgetting curve demonstrates how we forget anything we learn and how we forget over time. The savings method is

a duck, a carpet, a donkey, a carpet, LECTURE 2 the Associationist Approach to Mem Research association is strengthen by the # of times we experience Thorndike’s Learning Curve: box/cat trial, you notice that after repetition and LEARNING, they can perform action immediately. Stronger association. The LAW OF USE (LOSE IT OR USE IT) Thorndike’s Forgetting Curve: the decline in the # unrewarded experience, you can weaken an associations. —> DECAY? Maybe unused associations get weaker as the function of time. —> some people say though that decay isn’t proven, once it’s encoded, you cant really forget it, you just don't use it unless cued. Jenkins and Dallenbach: the group required to stay awake for 8 hours forgot SIGNIFICANTLY MORE than the group that slept 8 hours. So we could conclude that is it NOT just DECAY, we can conduce that sleep effects memory. It’s consolidating memory. Or perhaps the LAW OF USE explains this - the group awake more needs their memory more so it more likely to LET GO of the useless info. or INTERFERENCE could explain this as well, with similar info in memory causes an association to be forgotten —

readings notes L1 - L4 Ebbinghaus on Memory: every kind of mental state, once used then no longer needed, does not cease to exist after, instead stored in memory. (1)Voluntarily reproducing old experiences (recall) (2) Involuntary reproducing experience — laws of association (3) Unconscious reproduction; continued existence of previous mental states resulting from frequent reoccurrence of any condition/process but not aware to individual. The conditions in which people remember are so unique and vary from individuals! The skills our conditions have required us to remember are different. Jenkin and Dallenbach reading(1924): all groups required to learn the words to perfection, sleep group and awake group (sleep group did better at recall) —however when individual was “hypnotized” they improved. FORGETTING AND THE LAW OF DISUSE (McGEOCH):

LECTURE 2 CONTINUED (SEPT 11) video: memory needs important connection (encoding) there is no such thing as a true photographic memory but good memory techniques, it’s all about MEANING. Scientists saw that the frontal lobe and temporal lobe lit up when looking for meaning for words, the more meaning the more you remember it. Strong memories are not formed, they’re made! Jenkins and Dallenbach continued: first study to put decay against interferences, they concluded that forgetting is not a matter of decay but a matter of interference of the old by the new. The awake group had more verbal experience causing more interference with the nonsense word pairs they studied. SO because of the new associations, it interfered with their recall. *J&D did NOT include that sleep enhanced learning, and many modern studies on sleep/memory consolidation fail to control interference. *a study in 1995 (Siegal) found that rats who slept had the SAME brain activity as when they were LEARNING the maze. These rats had better learning and performance. this was DIRECT evidence that sleep had an ACTIVE role in memory consolidation. 2 types of interference: 1. retroactive interference (RI) —> later learning weakens prior learning for example our upper level psychology programs make it harder to recall what we learned in our 1100 courses. 2. proactive interference (PI) —> prior learning hurts new learning (functional fixedness) like having learn stick when you drive automatic Associationists Methodology Retroactive Interference: Control Group: List 1 — A-B List 2 —A-D Experimental Group: List 1 — DAX-ZIL List 2 — POF-WEP *to measure PI you take the percent correct for the control group subject the correct foe th experimental group then divide by correct by control group

PI =

[% Correct(Control Group) – % Correct(Experimental Group)] % Correct(Control Group) McGeoch (1932) interference effects - forgetting due to new information entering memory

- operate on weak content only - LTM uses semantic code McGeoch says that memory neither strengthens nor weaken over time so the memory once encode is PERMANENT trace. Memory strength established during conscious processing (the encoding phrase) Retroactive Interference List 1 – DAX-ZIL List 2 – DAX-WEP Test – DAX-? Correct Response is ZIL

1 RI p(List 2 Responses)

0.75 0.5 0.25 0 5X

10X

20X

40X

Failure to correctly respond ZIL MUST BE DUE TO incorrectly responding WEP… THE REPONSE COMPETITION HYPOTHESIS Prediction 1: Retroactive Interference (RI) should increase as List 2 Associations get stronger Prediction 2: Forgetting due to RI should EQUAL the tendency to make List 2 responses OKAY given the assumption that the probability of saying the SECOND list should INCREASE as RI INCREASES too, the GRAPH should look like two lines INCREASING. ^HOWEVER LOOK AT THIS IT DOESN’T MAKE SENSE More practice with list two actually REDUCE LIST TWO RESPONSES SO THE RESPONSE COMPETITION HYPOTHESIS IS WRONG!!!! video: HM (Henry Molison) removed a lot of the hippocampus to remove seizures but he could NO LONGER make new memories…

it means the that hippocampus must be responsible for making new long term memories. wwwwHM could remember motor skills but not events so it means that memories cannot be just one thing.. there are different kinds! and the different kinds depend on DIFFERENT PARTS OF THE BRAIN. How does a long term memory get written in the first place? involves an anatomical change in the brain when new memories are formed. The growth of new connections is what allows a memory to persist. The Unlearning Hypothesis If learning list 2 weakens list 1 association, then what should we see?

RED LINE List 1 Responses Likelihood Of Report

BLUE LINE List 2 Responses

1 5 10 20 Conclusions: List 2 responses increase, and list 1 responses decrease. Therefore, as we learn learn 2 associations it weakens list 1 associations???? *this is INCONSISTENT with McGeoch Response Competition Hyp. *missing evidence slide FOR Evidence AGAINST unlearning hyp; postman and stark (1968) Postman and Stark (1968) List 1 – KIP-TEK (to Perfection) List 2 – KIP-POJ (1, 5, 10, or 20 Times) Test

Was KIP presented with TEK, HIF, or LOC? and Was KIP presented with POJ, RIN, or WOX?

Likelihood Of Correct Response

RED LINE List 1 Responses BLUE LINE List 2 Responses

1 5 1020 this experiment was different because it tests RECOGNITION while the previous were testing CUED RECALL! They are arguing that the memory is still there, because recognition can get at a memory trace in a way recall cannot. SO….

SO THERE IS EVIDENCE AGAINST THE UNLEARNING HYPOTHESIS. this graph shows that the previous study did NOT have a strong enough cue to recall. SO THE LEARNING HYP IS WRONG associations are NOT UNLEARNED, because people did not forget the associations when given the proper cues. MAYBE the associations are REPRESSED??? come in Postman, Stark and Fraser (1968) with the RESPONSE SET SUPPRESSION HYPOTHESIS. The difference is that there are two groups in which they are both required to learn the lists to perfection.

Postman, Stark, and Fraser (1968) List 1 –A-B (e.g., JEP-WOP) to Perfection List 2 – A-D (e.g., JEP-SOV) to Perfection

Likelihood Of Correct Response

Group 1 Test = List 1 Recall A? (JEP?) Correct response is B (or WOP) Group 2 Test = List 2 Recall A? (JEP?) Correct response is D (or SOV)

Red LINE List 2 Recall Group Blue LINE List 1 Recall Group

So Group 2 should cued recall.

do better than group 1 due to repression, it is still

20 min

48 hrs

CONCLUSIONS: list 1 responses were impaired when list 2 was learned relatively recently and list 1 responses were not impaired when List 2 was such less recent. *recency of learning list 2 suppresses list 1 response? (video: Clive Wearing — “waking up for the first time” over and over) here comes the DISTINCTIVENESS HYPOTHESIS (and alternative to the response set suppression hypothesis)

A Distinctiveness Hypothesis*** (Underwood & Freund, 1968) Interference occurs because, due to their similarity, people cannot separately access A-B and A-D pairings The experiment: Prediction: Interference should be less if learning of A-B pairings is made DISTINCT from learning of A-D pairings Group 1: List 1 –A-B (e.g., JEP-WOP) – 32 Times and then on SAME DAY List 2 – A-D (e.g., JEP-SOV) to Perfection Group 2: List 1 –A-B (e.g., JEP-WOP) – 32 Times and then THREE DAYS LATER List 2 – A-D (e.g., JEP-SOV) to Perfection *test for both groups *list 2 response required will spearing the list in time lead to less proactive interference? ! —> so we would predict that GROUP 2 should do better because there was distinctiveness between learning the two lists. results: evidence FOR the distinctiveness hypothesis, the more distinctive the lists were in memory, the more they were able to remember, THIS interference occurs because experiences are hard to distinguish from one another. THIS ONE HYPOTHESIS WORKS!!!!!! IF YOU WANT TO LEARN SOMETHING, MAKE IT DISTINCT

Another Study by Postman and Gray in support for Distinctiveness Hyp: Group 1 List 1 –A-B (e.g., JEP-WOP) – To Perfection List 2 – A-D (e.g., JEP-SOV) - To Perfection

Group 2 List 1 –A-B (e.g., JEP-WOP) – To Perfection List 2 – A-D (e.g., JEP-SOV) - To Perfection AND While learning List 2 report List 1 responses Test for both groups (1 week later) *list 2 response required Results: group 2 does better. WHY? because learning list 2 while reporting list 1 response creates a distinctiveness! D increases as the time between two similar associations increases AND D increases if people think about two similar associations simultaneously The underlying Principle: ENCODING VARIABILITY thinking about the sam stimulus differently will create distinctive representations in memory. How to avoid interference? create two associations separate from learning list 1 and list 2. i.e. Bank — violin bank — lamp creating different meanings for each list

conclusion: failure of recall or forgetting does NOT mean that the info has been lost.

- interferences has mimcl effect on being able to recognize responses that people have difficulty recalling.

- forgetting is cause by the failure to access the learned info - accessing info in memory depends distinctiveness - Similarity to other information in memory blocks access to “target” information LECTURE FOUR: THE MULTI-STORE MODEL —> this is powerful and current model of memory

The demise of the associationism…due to ENCODING VARIABILITY (flashback to Ebbinghaus, though he created seemingly meanings pairs CVC…there is till meaning between nonsense syllables) —> this is essentially encoding variability because

JEP-WOP may mean something different then JEP-TIC SO memory CANNOT be studied in its purest form because stimulus has MEANING through the use of prior experience — always a factor. the methodology of Associationism is fundamentally flawed… because it assumes that JEPWOP is always the SAME Another problem… how many memories are there?

If there’s no delay in seeing a CVC and reporting the CVC the results should be not a lot of problems. But as the delay increases, the likelihood of reporting the right CVC is less because there is MORE interferences * so forgetting in the short time may be due to decay but forgetting in the long term may be to due to interference!! * * maybe memory is made of two different stories: longterm and shortterm!

The Cognitive Revolution: INFORMATION PROCESSING APPROACH the mind is like the computer. It performs tasks in STAGES.

so… The Demise of Associationism + evidence for 2 types of memory + emergence of cognitive psych and info-process approach = MULTI-STORE MODEL *REFER to slide for specific model layout

There are 4 properties the stores distinguish from:

- duration: how long - capacity: how much - code: what form - forgetting: how info lost The sensory store: •

Capacity = Very Large

•

Duration = About 500 milliseconds

•

Code = Physical Characteristics (any of the senses)

•

“Forgetting” = Decay

- iconic memory (visual sensory store) - echoic memory (auditory sensory store) short term store: •

Capacity = 7 + or – 2 Bits of Information

•

Duration = Up to 30 seconds

•

Code = Auditory or Acoustic

•

“Forgetting” = Decay

long term store: •

Capacity = Infinite

•

Duration = Forever

•

Code = Semantic (Meaning-Based)

•

“Forgetting” = Interference

“control” processes: processing allowing info to be transferred from one store to another - attention: allows us to choose info in the sensory store to enter the short term store

- encoding: transferring info from short term to long term store - retrieval: transferring info from longterm to short term store Sperling’s research on the large capacity of the visual store: - 12 sets of letters for 50 milliseconds - asked them to report as many letters they could (whole report technique) - on average they reported 4.5 letters - capacity is a little more than a third? - however he kept going with his experiment, he also developed the partial report technique, people only had to report one row of letters, —> twelve letters flash, tone presents immediately after, depending on the tone you have to report a certain row. Now they reported 3.2/4 letters on average! SO whats the capacity?? 80%

now!!! so you can report 80% if you are cued properly! Also depending on how many letters you ask them to report, the average is typically 4 letters —> also, when you increase the delay there is no partial report advantage, so it’s less likely to report the right letters.

why does the partial report advantage exist? according to Sperling, All available visual information is stored BRIEFLY in Iconic Memory –

Reporting letters takes time —> Requires Attention and Pattern Recognition

The Form of Info in Iconic Memory: Sperling tried to vary numbers vs. letters he found no partial report advantage! Meaning only PHYSICAL INFO can be stored (so not meaning)

“forgetting” in the sensory store could be decay but could be OUTPUT INTERFERENCE?

Averbach and Coriell: •

Flashed a row of 50 letters for 50 ms

•

After the flash, a line appears where one of the letters was

•

People to report the letter that was there

•

this is iconic memory; he was able to find out that there was an after image, we have a picture of what was shown of a 250m/s —> so if you ask people to report after 500m/s delay they cannot

•

—> as soon as you present a visual mask (has the info from image, but meaningless) the afterimage is gone.

•

so DECAY IS THE CAUSE OF FORGETTING IN ICONIC MEMORY

LECTURE FIVE: ATTENTION •

Definition: A concentration of mental activity

•

Selective Attention: Processing one source of information at the expense of other sources.

•

Divided Attention: Processing multiple sources of information simultaneously. (multitaskers) —> after a mild head injury, this gets bad.

There are limits to the amount of information people can take in at one time. —> People select information that is most relevant to them. Dichotic listening experiments: involves two different messages presented at the same time through headphones, one to left ear one to the right. People told to shad one one messag, what do people know about the ignored one? Broadbent (1954) presented different series of numbers at the same time in each ear, people had to report all the numbers they heard. People first reported digits in one ear, following by digits in the other ear. So a conclusion is that people CANNOT attend to both messages at once, need to attend to one then switch to the other. In 1958 he did a split scan experiment where he gave letters to different ears. Report in any order was 65% and the report in each pair presented 20% how much info can people attend to at once? to what extent to people process the info you ignore? Cherry 1953, people report physical info about ignore message (pitch, volume, gender of speaker) and people reported no MEANIG of the ignored message, but could report gender and tone. Shadowing is more difficult when messages overlap their physical properties. People must select info based on physical characteristics. Broadbent’s Filter Model**

- people only attend to one source of info at once - info is selected based on physical properties - ignored info is filtered out before meaning is processed —> attention is necessary for pattern recognition

Problems with the filter model:
...