Lecture 7: Retrieval from STS and Levels of processing

Question 1

What is the sternberg scanning task?

Answer 1

• Quest: how much time does it take to make one mental comparison in STM?
• Task: memory searching (scanning). The question simply was does the probe item match one of the numbers that was shown in the memory set?
• On half the items it matched and on the other half it didn’t match.
• DV= how quickly the person could answer yes or no to the probe item (pressed a button)
• Size of memory set varied, whether probe matched varied through a series of trials. At the end, they took the average of the trials overall and the average for the varying sizes of memory set

Question 2

What is the assumption that guides how retrieval from STS is modelled?

Answer 2

• time between a probe stimulus and a response is occupied by a series of discrete stages (Probe digit –> encoding stage –> memory comparison stage –> response stage)

Question 3

What is the encoding stage?

Answer 3

probe item gets encoded into stm, which takes time.

Question 4

What is the memory comparison stage?

Answer 4

The next stage is does this item match something in our meomry set? Memory comparison stage. Comparing probe item to the numbers that were encoded during the memory set

Question 5

What is the response output stage?

Answer 5

Finally you give a response or an output based on the comparison

Question 6

What are the three hypotheses regarding search in STS?

Answer 6

1) Parallel
2) Serial exhaustive
3) Serial self terminating

Question 7

What is the parallel hypothesis?

Answer 7

• According to this item, you can compare the probe item to each of the items in the memory set all at once (encoding stage and comparison stage happen simultaneously). Thus, it would not matter how many items are in the memory set because it happens all at once.
• Efficient
• if this hypothesis is correct, the results would be a straight line

Question 8

What is the serial exhaustive hypothesis?

Answer 8

• Comparing each item to the probe, one after the other. You go through each item in the memory set to search for the probe item, even once you find the probe item in the memory set you continue to the end of the list (hence the word exhaustive - don’t terminate when you find something).
• Not very efficient
• Response time will always be dependent on the set size (not on yes vs. no)
• Shorter reaction time when there are less items.
• Upwards slope

Question 9

What is the serial self terminating hypothesis?

Answer 9

• No trials = no termination because there is no match, on the yes trials once you get a match you terminate.
• • No responses, the bigger the set the longer it takes you to get through everything because you always have to go through the entire set
• • The yes responses will be faster when the memory set is small, and slower when the set size is large. Sometimes you terminate early, but sometimes you terminate late. The average will be somewhere in the middle.
• This difference becomes even more pronounced when the probe is early in the memory set (rather than later or last).
• • When there is only one item, the yes and no trials are the same speed
• • Self terminating assumption predicts that the no trials should be slower as the items increase, and the yes trials will get slower but not as much. This is called an interaction (memory set size had a different effect on yes as compared to no trials)

Question 10

Which of the hypotheses was supported by the result

Answer 10

• Found a serial exhaustive scanning mechanism
• Could be because maybe its so fast to make a comparison that it would take more energy to stop it than to just go through an exhaustive list
• • Turns out it is pretty fast
• 397.2 ms is where the line interacts with the y axis, this is the speed it takes If you weren’t doing a comparison, everything but the comparison.
• Once you add in one mental comparison you add in time to make the comparison happen which is 37.9 ms (time it takes to make a single mental comparison. For every item that’s added in the reaction increases by a unit of 37.9 ms

Question 11

How does changing the nature of the probe effect the task?

Answer 11

• Same task (adding in the quality of the probe stimulus)
• Clear vs. degraded probe digit (i.e., quality of the probe stimulus)
• How does our system deal with a degraded stimulus?
• Encoding stage or memory comparison stage?
• First option: encoding stage, sees degraded stimulus and decides not to pass it on before it fixes it up. Fixing it takes time. Does it matter how many items are in the memory set? Does the extra time interact with the number of items in the memory set?
• No it will take longer initially (during the encoding stage, i.e., longer than 397.2) but it will still take the same amount of time to compare the items in the comparison stage (37.9 ms). The slope remains the same, it just moves higher.
• Option 2: the encoding stage just passes on the degraded stimulus without fixing it. This adds time onto the comparison stage (12.1 is just an example). For example each mental comparison is now taking 50 seconds (i.e., 37.9 + 12.1).
• This is an interaction now between the number of items and stimulus quality. They have their impact in the same stage of processing (the comparison stage)

Question 12

What does the encoding stage locus predict?

Answer 12

Encoding stage locus predicts additivity

Question 13

What does the memory stage locus predict?

Answer 13

Memory stage locus predicts interaction

Question 14

What were the actual results of the degraded/clear probe task?

Answer 14

The actual results were additive (which indicates that the fixing of the degraded stimulus happened in the encoding stage)

Question 15

What are the advantages of additive factors logic?

Answer 15

Simple logic used to isolate stages in a serial process.

Question 16

What are the disadvantages of additive factors logic?

Answer 16

Only useful when examining a simple serial process. Assume discrete stages.

Question 17

What is the framework for levels of processing?

Answer 17

• emphasis on what do you do with material
• type of encoding important
• storage is a by-product of encoding
• orienting tasks: - physical, phonemic, semantic

Question 18

What did the levels of processing class demo illustrate?

Answer 18

• Dr. Herdman read a list of words

- People who were in the semantic condition did much better on average than the phonemic group ( generating rhymes)

Question 19

What are the two assumptions made by Craik and Lockhart about the levels of processing (and against rehearsal/the assumptions of a three store model)?

Answer 19

1) Memory persistence NOT affected by rehearsal (called type I)
2) Memory persistence depends on encoding level (called type II)
- did not believe that rehearsal in and of itself was responsible for getting information into LTM, what matters processing information at a deeper level

Question 20

What is the evidence against rehearsal?

Answer 20

• Varied amt. of rehearsal (trying to demonstrate that rehearsal is not that important)
• • “cabbage, bag, rat, cupboard, blonde, table, bath etc.”
• • told to rehearse the word, until they see the next
• Results: no relationship between probability of recall and # of rehearsal cycles
• • Put into question that rehearsal is the mechanism or acquisition of getting things into memory

Question 21

How did Hyde and Jenkins show that encoding level matters?

Answer 21

• Varied type of orienting task
• • Count # letters (in each word they saw)
• • Check for E (cross out letter E’s in the words that they saw)
• • Pleasantness ratings (rate how pleasant or unpleasant words were)
• • Incidental condition vs. Intentional condition
• • The deeper the level of processing, the better they did on the items (i.e., the pleasantness ratings task). When doing the pleasantness ratings, both incidental and intentional learning groups did the same on average. Because they were semantically encoding those items (associating them with meaning)
• • For the count and letter E conditions, they didn’t do very well overall, but there was a slight increase on average if they knew it was a memory task (i.e., intentional)
• • Intention helps a bit if you’re doing shallow level processing

Question 22

What was the recognition task by craik & tulking? And what did it show about semantic encoding?

Answer 22

• Orienting
• • Physical (was the word that was presented in capital letters?)
• • Phonemic (did it rhyme?)
• • Semantic (does the word fit into the sentence?)
• Then had people do a recognition task (rather than free recall like class demo)
• recognition task (present some items that they had seen and present some that they had not seen)
• recognition performance increases as a function of how deep you process that information. The semantic group did better than the phonemic group and the phonemic group did better than the physical.

Question 23

What did Tresselt and Mazner find out about copying words and processing words at a deeper level?

Answer 23

Orienting:
- cross out vowels
- copy the words
- judge degree to which words fits a category (economics)
-Incidental recall
- Results: recall for judge category twice as high
as just coping the words

Question 24

What are incidental vs. intentional memory tasks?

Answer 24

• Incidental (not knowing something is a memory task, just asked to recall items from memory afterwards) vs. Intentional learning (told it is a memory task, make sure you learn those items etc.)

Question 25

What did Bowler and Karlin find in their study of nonverbal material and how did it show that encoding level matters?

Answer 25

• Orienting
• showed the participants pictures of people
• • Note sex of person
• • Judge face likableness
• • Rate honesty
• Recognition task (bc its nonverbal information. Cannot free recall pictures - would have to draw them out)
• Groups doing deeper level processing (likable or honesty) did better at recognition than the sex (shallow level processing). The percentage of recognition for gender was just above chance (60% for incidental, and 56% for intentional).

Question 26

Are there differences in-memory performance at the within semantic processing level (i.e., varied complexity of sentence frame)?

Answer 26

• Craik and Tulving Pointed out that it goes beyond just level of processing and depth
• varied complexity of sentence frames (Simple, Medium or Complex)
• Judge if word fit into sentence, incidental recall
• Within the semantic level, peoples memory performance level got better and better the more complex the semantic material (i.e., the deeper they had to think)

Question 27

What did Morris, Bransford and Franks find when they manipulated the types of recognition test?

Answer 27

• Two types of memory tasks
• Rhyming condition vs. Semantic condition
• • Standard recognition: Semantic > phonemic (if you test people with a semantic recognition task, the semantic condition does really well and the phonemic doesn’t)
• • Rhyme recognition: Phonemic > semantic (if you test with rhymes the rhyme group does well, which means that the information may be getting in there with rhyming condition but cannot be accessed readily (i.e., without a certain type of cue, in this case a phonemic cue). Information need to be recalled in a certain way)
• Suggests that the type of test is important

Question 28

What does memory depend on?

Answer 28

Memory depends on what you do with material.

• copying words not useful!
• simple rehearsal not useful!

Question 29

What is the idea behind additive factors logic?

Answer 29

According to additive factors logic, if a task contains distinct pro­cesses, there should be variables that selectively influ­ence the speed of each process. Thus, if two variables influence different processes, their effects should be sta­tistically additive. However, if two variables influence the same process, their effects should statistically inter­act.