Primary Memory

1. Primary Memory

2. Questions for this section • Is there more than one kind of primary memory? • What is the capacity of primary memory? • What do serial position and activation effects tell us about the nature of primary memory? • What does rehearsal accomplish? • How is information lost from primary memory? • How do we retrieve information from primary memory?

3. Sensory Memory • Sensory memory is responsible for initial (fast) processing of sensory stimulation. • Characteristics • Preattentive and automatic. • Info. loss occurs through decay. • Iconic Memory • Vision • Echoic Memory • Auditory

4. Sperling 1960 A classic experiment on iconic memory. Subjects are presented brief displays (~50 msec) of letters such as this: G L K V Q H P B X N Y D Whole Report: Ss are told to recall as many letters as they can. Partial Report: Ss are cued (tone) AFTER the display is gone on which row to recall

5. Logic and Results from Sperling (1960) On average, Ss remember about 4 letters for the WR conditions, and they remember about 3.5 letters for the PR conditions. If we keep in mind that for WR, the 4 letters that are recalled are out of 12 possible, they only recall about 33%. However, for PR, the 3.5 letters that are recalled are out of 4 possible… thus recall in this condition is close to 90%. The increased recall percentage in the PR conditions is called the Partial Report Superiority Effect. Of course, the questions now becomes WHY? Sperling argued that after the offset of the stimuli, a visual “icon” was present, and subjects recalled information from that icon.

6. % Correct 0 ~300 msec WR Offset to Tone Delay Another twist to the Sperling (1960) experiments… Sperling also varied the amount of time between stimulus offset and the recall cue (tone) for the PR conditions. These data are shown below: What do these data tell us? That the icon only lasts for a very brief (~300 msec) amount of time.

7. Echoic Memory • Darwin, Turvey & Crowder’s (1972) “Three-eared man” study. • Also found a PRSE, but the duration lasted as many as 2 seconds. • Why would echoic memory last longer than iconic memory? • Suffix effect. • If recall cue is physically similar to words on a list, recency effect disappears.

8. What is it? IP psychologists described it as a “place” separate from LTM. Contemporary cognitive psychologists argue that it is simply the activated portion of LTM. How much info.can it hold? Depends on the content of the information (sometimes we can “chunk” info.). Also depends on our level of arousal (Yerkes- Dodson Law). Short-term Memory

9. Recall Accuracy Serial Position Serial Position Effects When we present info. (e.g. words) for subjects to remember, and we then plot their accuracy as a function serial position, we will normally see primacy and recency effects (see below). Primacy Effect Recency Effect

10. Recall Accuracy Serial Position Explaining Primacy Glanzer and Cunitz (1966) argued that the primacy effect occurs because Ss are able to rehearse the items more at the beginning of the list. How would you test this hypothesis? Glanzer andCunitz’s data… No Primacy

11. Recall Accuracy Serial Position Explaining Recency Glanzer and Cunitz (1966) argued that recency effects occur because those items at the end of the list are the most activated items. How would you test this hypothesis? Glanzer and Cunitz’s data No Recency

12. Serves two purposes in STM. Keeps info. active in STM Gets info. from STM to LTM for later retrieval. Craik and Lockheart (1972) Maintenance rehearsal Keeps info. active in STM. Elaborative rehearsal Helps create LTM trace for later retrieval. How could you test these assumptions? Rehearsal

13. Information Loss from STM • Peterson and Peterson (1959) • Presented subjects with consonant-vowel-consonant (CVC) trigrams to remember for a short period of time. • While trying to remember, they also had to count backwards by threes from a random number. • Example…

14. F O V

15. 893

16. RECALL

17. Recall Accuracy Delay (seconds) Peterson and Peterson’s (1959) data P & P demonstrated that the DURATION of STM was not very long. That is, if information is not rehearsed, it does not stay in STM for more than a few seconds After 3 secs., Accuracy is ~50% After 15 secs., Acc is ~10% 0 3 15

18. Explaining Loss from STM • Peterson and Peterson argued for a decay explanation of their data. • Keppel and Underwood (1962) argued that the P & P data were due to proactive interference. • When items at the beginning of the experiment interfere with memory for the items at the end of the experiment.

19. Retrieval from STM • Another important question we may ask is how we get the information out of STM… that is, how do we retrieve it? • Sternberg (1966) performed a classic study to answer this question.

20. Sternberg (1966) • In this study, Ss were presented w/ “memory sets” that varied from 1 – 6 letters. After presentation of the memory set a “target” letter was presented. Ss had to respond as quickly and as accurately as they could whether or not the target was a member of the previous memory set. • Two questions regarding retrieval emerge: • Is STM retrieval a parallel or serial process? • Is STM retrieval a self-terminating or exhaustive process?

21. Reaction Time 1 6 Sternberg’s (1966) data… “No” “Yes” Do these data indicate that search is parallel or serial? Self-terminating or exhaustive? WHY?

22. An alternative account of Sternbergs’ data… These very data can be explained by a parallel search… how? If we assume that activation is limited and that each item in the memory set needs some activation to remain in STM, then items in the larger memory sets receive proportionately less activation that items in the smaller set. A parallel search that depends on the activation level of each item can explain these data very well.