1. Limits on the Quantity & Duration of Information Held in Short-Term Memory Psychology 355: Cognitive Psychology Instructor: John Miyamoto 04/20/2016: Lecture 04-3 Note: This Powerpoint presentation may contain macros that I wrote to help me create the slides. The macros aren’t needed to view the slides. You can disable or delete the macros without any change to the presentation.

2. Outline Mention Friday section & essay quiz How much information can be retained in STM? How long does information last in STM? ♦ Brown-Peterson task ♦ Retroactive and proactive interference ♦ Proactive interference and the Brown-Peterson paradigm Psych 355, Miyamoto, Spr '16 2 Lecture probably ends here Results for Memory Span for Digits

3. Memory Span for Digits: About 7  2 Digit Span Memory Task: ♦ Digits are presented one at a time. (Experimenter controls how many digits will be presented.) ♦ After subject sees the digits, subject must respond with the exact sequence of digits that were presented. ♦ Dependent variable: Percentage of trials with perfect performance when the number of digits equals 2 or 3 or... or 10 or 11 or..... Result: Retaining 3 or 4 digits is easy. Retaining 15 or more digits is impossible. Typical limit is around 7  2 digits is the memory span for digits Psych 355, Miyamoto, Spr '16 3 Luck & Vogel (1997): Memory Span for Colors & Positions

4. Memory Span for Color & Location Psych 355, Miyamoto, Spr '16 4 Luck & Vogel (1997) is described in Goldstein's Figure 5.8 and Figure 5.9. Subject sees some colored squares on one slide followed by another slide with colored squares. Subject’s task is to say whether the two slides are identical or different. Result: Performance drops off radically after presentation contains more than 3 squares. Typical Results for Memory Span with Different Contents

5. Psych 355, Miyamoto, Spr '16 5 Typical Results for Memory Span with Different Contents We can repeat this experiment with different types of items. ♦ Auditory (spoken) digits ♦ Nonsense syllables – CVC form like "KOR" ♦ Short words chosen at random ♦ Long words chosen at random George Miller's "Magic Number Seven Plus or Minus Two". Conclusion: There is a strong capacity limit on STM, i.e., there is a limit on the total number of items that can be retained in STM. Return to "Magical Number 7 +/- 2: Raise Issue of Chunking

6. Psych 355, Miyamoto, Spr '16 6 Magical Number Seven Plus or Minus Two Miller, G. A. (1956) The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81–97. There is a strong limit on the quantity of information stored in STM. ♦ Roughly the same limit applies to many different contents, e.g., digits, nonsense syllables, short words, long words, colors in different positions on a screen, tones of different pitch, etc. Limitation on STM capacity only applies if there are no meaningful relationships between separate items. When meaningful relationships exist among the items, many more items can be held in STM. Some Issues re Measures of Memory Span

7. Psych 355, Miyamoto, Spr '16 7 Some Issues Regarding Measures of STM Capacity Limits The preceding examples use low-meaning stimuli. Would the results differ if we used stimuli with more meaning? Example: Waiter or waitress taking a dinner order: ♦ I'll have the mushroom soup, the filet of sole with green beans, and rice. (14 words) ♦ I'll have a ceasar salad, the flank steak medium rare, carrots and peas, and mashed potatoes. (16 words) ♦ I'll have..... Answer: Yes, the results are different when the stimuli are meaningful. How should we interpret this? (See next slide). Chunking & Recoding

8. Psych 355, Miyamoto, Spr '16 8 STM Capacity Limits Apply When Stimuli Are Unrelated The capacity limit, 7 ± 2, applies to sequences of stimuli.... that do not have an internal organization. ♦ Example (no internal organization): EGG, PEN, FOG, CAR, BELT, FLY,.... ♦ Example (has internal organization): YESTERDAY I WENT SHOPPING FOR A NEW COAT. I WANTED A WARM COAT THAT HAS A HOOD,.... "Chunk" – a larger unit of information built out of related smaller units. Example of Chunking – Animals List

9. Psych 355, Miyamoto, Spr '16 9 Example of Chunking Your Task: Remember the following 12 words: Can you remember the words in the list? ANSWER = mascots of Pac-12 schools: husky, cougar, duck, beaver, cardinal, golden bear, bruin, trojans, wildcats, sun devils, buffalo, Utes We "chunk" the individual items under the concept “mascots of Pac-12 schools.” husky cougar duck beaver cardinal golden bears bruins trojans wildcats sun devils buffalo Utes Chunking in Everyday Experience

10. Typically everyday experience is meaningful. Not like trying to remember a series of unrelated digits in the lab! In everyday experience, people constantly reorganize the current information in terms of related general knowledge. In a very unfamiliar situation, you may feel overwhelmed with information overload – you aren’t able to chunk effectively in the unfamiliar situation. Psych 355, Miyamoto, Spr '16 10 Recoding - Definition

11. Recoding "Recoding" – Changing the representational format (change in code) Recoding of information in STM can affect ability to retain information in STM. Psych 355, Miyamoto, Spr '16 11 Codes in STM & LTM

12. Psych 355, Miyamoto, Spr '16 12 Codes in Short-Term & Long-Term Memory A "code" is a format in which information is represented in the cognitive system. (Not the same as the "neural code") Phonological codes – words represented as sequences of sounds. Visual codes – mental imagery, diagrams, perceptual memories Semantic codes – meanings. ------------------------------------------------------------------------------- Short-term memory (STM) and long-term memory (LTM) use all of these codes plus possibly others that are not listed here. ♦ Particular tasks may be biased towards one kind of coding. Next: Evidence for different types of mental codes. This evidence is just a taste – later lectures will present much more evidence. Evidence for Phonological Codes

13. Psych 355, Miyamoto, Spr '16 13 Evidence for Phonological Codes Phonological codes (representing words in terms of the sound of the words) ♦ Conrad (1964): People who are asked to remember visually presented letters, make mistakes that confuse a correct letter with similar sounding letters. Example: Suppose you are asked to remember A  F  T  R. (The letters are presented visually one after the another a screen.) Common mistake: A  S  T  R or A  F  P  R Notice "S" sounds like "F" and "P" sounds like "T". The letters were presented visually, so the effect of sound similarity is due to the way the subject represents the stimulus, and not the stimulus alone. Evidence for Visual Codes – Mental Rotation Studies

14. Evidence for Visual Codes: Shepard's Mental Rotation Experiments TASK: As quickly as possible, decide whether the two figures shown to the right have the same or different shapes. Response time for "identical" figures is a linear (straight-line) function of the angle of rotation between the figures. Psych 355, Miyamoto, Spr '16 14 Response Time Angle of Rotation Mental Rotation Experiment - Interpretation

15. Psych 355, Miyamoto, Spr '16 15 Result is easy to explain if we assume that subjects are rotating a mental image. Results are hard to explain if mental representation is exclusively propositional. ♦ “Propositional representation” = Symbolic code like human language or a computer language. Response Time Angle of Rotation Evidence for Visual Codes: Shepard's Mental Rotation Experiments Semantic Codes

16. Psych 355, Miyamoto, Spr '16 16 Semantic Codes Confusions between long-term memories are usually based on similarity of meaning rather than on similarity of sound or appearance. Example: Suppose you must remember the following words. DOG, PONY, WOLF, ELEPHANT, PENCIL, WHALE ♦ Recognition memory test: Did the list contain "PONY"? Did the list contain "EAGLE"? ♦ Later you are more likely to say "Yes" to "HORSE" than to "LOG." (Intrusion Error: False recognition due to similarity of meaning) ♦ Later you are more likely to say "No" to "PENCIL" than to "WOLF." (Omission Error: False rejection due to dissimilarity of meaning) Semantic codes also play a role in STM but these examples are more complicated. Summary re Codes

17. Psych 355, Miyamoto, Spr '16 17 Summary re Memory Codes Recoding: Changing the code in which information is represented, e.g., changing from a phonological code to a visual code. The phonological code is the primary code for STM. The semantic code is the primary code for LTM. ♦ Other codes are also used in STM and LTM, e.g., visual code. Some tasks are easier to perform by using one type of code or another. Recoding can sometimes help to overcome capacity limits of STM. Overview of STM & WM Models

18. Psych 355, Miyamoto, Spr '16 18 Example of Recoding "It is a cold but sunny winter day. You are looking across Drumheller Fountain. Mt. Rainier is visible in the distance. Two children are playing with a ball next to the fountain. They accidentally throw the ball into the water." If you formed a mental image of the situation, then this is an example of recoding (changing from the verbal format to an image format). Recoding involves changing the cognitive “code." Recoding can increase the amount of information that one can retain in STM by putting it into a form that is more easily retained, e.g., changing verbal information into a mental image. Example of Chunking and/or Recoding – Chess Master Example

19. Psych 355, Miyamoto, Spr '16 19 Famous Example of Chunking and/or Recoding Chess Masters versus Chess Novices (Beginners) Chase and Simon (1973a, 1973b): Chess masters can remember the locations of a large number of pieces on a chess board if it is flashed in front of them. Chase and Simon's studies are discussed in Chapter 12 of Goldstein (2014, 4th edition). They also illustrate differences in how chess experts and chess novices use their memory resources. Results for Realistic Boards

20. Chess Masters versus Chess Beginners (Novices) Chess masters are much better than chess novices at remembering chess positions. Psych 355, Miyamoto, Spr '16 20 Results for Random Boards Master Novice (a) Actual game positions Master Novice (b) Random placement Master does better because he can chunk based on strategic relationships. Master does no better than novice because he can’t chunk.

21. Chess Masters versus Chess Beginners (Novices) Masters and novices are equally bad at remembering a random chess position (impossible position) on a chess board. Psych 355, Miyamoto, Spr '16 21 Chunking Summary Master Novice (a) Actual game positions Master Novice (b) Random placement Master does better because he can chunk based on strategic relationships Master does no better than novice because he can’t chunk.

22. Psych 355, Miyamoto, Spr '16 22 Summary: Chunking, Recoding, STM Capacity Limits As we take in information,..... we reorganize it into larger chunks, and..... we recode the information into other mental formats. ♦ Knowledge, learning, and past experience help us use our limited-capacity short-term memory more effectively. Are experiments that study STM capacity limits relevant to everyday life? ♦ In everyday life, we almost never try to remember sequences of unrelated information. ♦ Yes – the capacity limits are always present, even if they are hidden by our chunking and recoding strategies in everyday life. What Causes Forgetting in WM/STM?

23. Are Experiments that Study STM Capacity Limits Relevant to Everyday Life? In everyday life, we almost never try to remember sequences of unrelated information. Yes – the capacity limits are always present, even if they are hidden by our chunking and recoding strategies in everyday life. STM capacity limits strongly influence reasoning strategies even if they don't directly quantify the amount of semantically- related information that can be held in STM. Psych 355, Miyamoto, Spr '16 23 What Causes Forgetting from WM/STM?

24. Why is information lost from STM? How long does information last in STM without active processing? Two hypotheses ♦ Decay – representations of information simply "fall apart." ♦ Interference – other information "bumps" information out of STM. Brown-Peterson paradigm – an attempt to measure how long information is retained in STM without active processing. General Conclusion: Without active processing, all information is lost after 15 – 20 seconds. What causes the loss of information? ♦ Interference - one very likely cause. ♦ Decay - hard to prove convincingly that decay occurs; maybe it does. Psych 355, Miyamoto, Spr '16 24 The Brown-Peterson Paradigm - Example