POGIL Lab Module: Cognition

Scanning of STM (Short-Term Memory)

Scanning of Short-Term Memory
Motivation
How do individuals scan short-term memory? Assume you are at a payphone and you call the Information Operator to get a phone number. You don't have paper and pencil so you can't write down the number but you are holding the number in your head. Where exactly "in your head" is this number? It's in that part of memory we call short-term memory (abbreviated as STM). Other names for short-term memory are short-term store (STS), primary memory and immediate memory. It's also been called a wastebasket because information comes in but is then quickly discarded to make room for new information.

Now suppose your friend, who is standing nearby, asks you "Is there a '9' in the phone number?" How would you go about scanning the 7-digit phone number in your short-term memory to determine whether it contains a 9? Think about how you do this.

Learning Objectives
(1) Students will introspect about their own memory processes;
(2) Students will learn to derive a model based on reasoning about possible data outcomes;
(3) Students will do the Sternberg experiment and plot their data on a graph using Excel;
(4) Students will think analytically about data on a graph to derive a model

New Concepts and Vocabulary
Short-term memory, short term store, primary memory, immediate memory, visual STM, auditory STM, stimulus set, test stimulus, parallel processing, serial processing, serial exhaustive search, serial self-terminating search.

Key Activity
1. Imagine you're the one getting the phone number at a payphone and then your friend asks you if a certain digit was in the phone number. Introspect (Think aloud) about what you actually do when you scan your short-term memory and write down your introspections.


2. Do you scan all 7 digits in your STM at once? This we call parallel processing. Do you scan them one by one in turn? We call this serial processing. If you are searching them serially, do you stop when you find a match (a serial self-terminating search) or do you keep going on to the rest of the digits even once you have found a match (i.e., a serial exhaustive search)? The example we gave above is an example about auditory STM because you heard the phone number from the information operator.

A similar process occurs when, instead of hearing the numbers, you see the numbers presented in succession on a TV screen or computer. We call this visual STM. The following exercise illustrates this.

3. Suppose you are presented these numbers on a TV screen one after the other:



We call these numbers the stimulus set. Then, you're given the test stimulus "1," meaning you are asked whether or not "1" appeared in the stimulus set. Introspect about how you would scan your visual STM to answer this question. Here are some possibilities you might think about:


(1)Do you scan in parallel (comparing in one shot the test item to all the memorized items you're holding in STM)? Or

(2)Do you scan serially? If serially, do you use a serial self-terminating search (this means, you stop searching once you find a match) or do you do a serial exhaustive search (comparing the test stimulus to each item in your visual STM even once a match is found)?


(STUDENT NOW DOES THE PSYCHMATE EXPERIMENT WHERE HE /SHE PARTICIPATES AS A SUBJECT IN STERNBERG'S EXPERIMENT ON VISUAL STM.)

Let us think for a moment what predictions the different models make for the way our data should look. Remember, you are given a stimulus set containing s items, then you are shown a target item that either is in the set (in which case you respond "yes") or a target item which is not in the set (in which case you respond "no").
  • In a parallel search, it wouldn't matter how many items are in the stimulus set for you to scan the stimulus set. The reason the size of the stimulus set s doesn't matter is that no matter how many items are in the stimulus set they would all be held in STM simultaneously and searched simultaneously.
  • For a serial exhaustive search, you don't stop the search when you get a match. What this means is that you would compare the test stimulus to all the items in STM for each stimulus which gets a positive response as well as for each one that gets a negative response. Hence the slope of the RT function would be the same for positive responses (i.e., item was in the stimulus set) as for negative responses (i.e., item was not in the stimulus set).
  • In a self-terminating search, you stop in the middle of the list, on the average, for "yes" responses, but continue through the entire list for "no" responses. So, the rate at which RT increases with list length, the slope of the RT-function for negative response would be steeper than that for positive response. As a matter of fact, as list length is increased, you would expect the latency of positive response to increase at half the rate for negative responses.


Exercises

1. It says above, "In a serial self-terminating search, you stop in the middle of the list, on the average, for "yes" responses, but continue through the entire list for "no" responses." Explain why.


2. Below is the graph for a parallel search. The size of the stimulus set, s, is on the abscissa (x-axis) and RT is on the ordinate (y-axis).

  1. As s increases, what prediction can we make about the reaction time for "yes" (positive) responses? Why?
  2. As s increases, what prediction can we make about the reaction time for "no" (negative) responses? Why?

3. Below is the graph for a serial exhaustive search where s is on the abscissa (x-axis) and RT is on the ordinate (y-axis).

  1. As the size of the stimulus set s increases, what prediction can we make about the reaction time for yes" (positive) responses? Why?
  2. What prediction can we make for "no" (negative) responses? Why?
  3. Why is the graph for "no" responses higher than that for "yes" responses?

4. Now suppose the search of STM is a serial self-terminating search. Below is the graph for a serial self-terminating search where s is on the abscissa (x-axis) and RT is on the ordinate (y-axis).

  1. As the size of the stimulus set, s, increases what prediction can we make about the reaction time for "yes" (positive) responses? Why?
  2. What prediction can we make for "no" (negative) responses? Why?
  3. Why does one graph have a steeper slope than the other? Why aren't the two graphs parallel as they were for a serial exhaustive search? Explain.

5. Now look at your data in the experiment you just did. Plot RT as a function of length of list. (You can do this using Excel.)


5. What do you find? Of the models we considered, which one best fits the data you got?


6. When Sternberg plotted his data using many subjects over many trials, this is what he got:

Which of the 3 models above best fit Sternberg's data?


If you said "exhaustive search", you are correct. Explain your response.



Problem

You will now use PEAK to apply Sternberg's model to a different domain. Pretend you are taking a chemistry course and you want to know how you scan STM when the stimuli are stimuli in a chemistry course. You will prepare stimuli that you would find in a chemistry course and run the experiment using PEAK.