Brief Lecture Notes for Unit 7
In the Atkinson-Shiffrin model of memory, there are three types of memory: sensory store, short-term memory or STM, and long-term memory or LTM. (If you don't see a flowchart-like diagram below, it's because my scanner and I have had a spat.) Think of sensory store as a "pre-awareness" or "pre-processing" memory (a very brief "memory buffer" into which incoming information is stored before it is consciously processed), STM as your "working memory", and LTM as your "permanent memory" or "memory storehouse".
In this model of memory, these three types of memory can be compared and contrasted along three different dimensions.
The first dimension, decay time, has to do with how long information can be held within that memory "space" before it is spontaneously erased or forgotten (before the so-called memory trace is said to decay).
The second dimension, storage capacity, has to do with the "size of the memory buffer", or how much information can be held in this memory space at any given time.
The third dimension, encoding mode, describes how information is stored or organized within each memory space.
There are four processes by which information can be transferred from one memory space to another (again, the diagram above, if any, will prove helpful):
Much of this unit will have to do with a detailed look at experimental confirmation of the above assertions. Five critical experiments are outlined below.
Sperling's span of apprehension experiments
Purpose: To demonstrate the existence of sensory store and to determine its properties, many of which can be explained by the existence and nature of the span of apprehension, a limit in the amount of information that can be transferred from sensory store to STM in a given amount of time. It is a function of the rate of transfer of information (which is processed in a serial fashion: see Unit 5 for a discussion of serial vs. parallel processing). Because of the short decay time of sensory store, information not transferred from sensory store to STM during the decay time limit is lost.
Method and results: In the first of two experiments, Sperling presented arrays of 12 letters (3 rows of 4 letters each) tachistoscopically (see Unit 3). Subjects typically stated that they saw all 12 letters, but could only report 3 or 4 of them before the memory trace faded. In the second experiment, subjects were told that, after seeing the array, they would hear a musical tone (pitched high, medium, or low) telling them which row of the array to report, and the time lapse between the presentation of the array and the onset of the tone was varied as the independent variable. With immediate onset (0 sec. delay), subjects can typically report all 4 letters of the indicated row, but with a delayed onset of 1 sec., recall worsens to about 1-2 letters (can you figure out why this makes sense given the theoretical explanation above?)
Dichotic listening experiments
Purpose: To illustrate the fact that (with a few exceptions as noted below) only attended information is transferred from sensory store to STM (subject to the span of apprehension limitation above); unattended information, in general, is lost and does not reach conscious awareness.
Method and results: A subject wears headphones, and two different auditory messages are presented, one to each ear. The subject is forewarned that at some random moment both messages will stop, and s/he will be asked to recall as much of the unattended message as possible. To ensure that attention is kept focused only on one message, the subject is asked to shadow (repeat out loud) the right-ear message. In general, only about 1 sec. worth of information from the unattended, left-ear message can be recalled (why 1 second?) Memory for the unattended message is almost nil otherwise: even changes in the speaker or in the language used by the speaker generally go unnoticed. However, interestingly, if the subject's name is imbedded into the left-ear message, subjects usually hear it, showing that unattended information is attenuated but not entirely incapable of diverting attention. (Can you see why this is a valuable property for a memory system to have?)
Free recall experiments
Purpose: To elucidate differences between STM and LTM, and to demonstrate that to an appreciable extent at least (some psychologists disagree), the two memory systems are separate and relatively independent of one another.
Method and results: Subjects are presented with a list of words, read out loud to them at a constant rate, and asked to try to remember as many of the words as they can. Recall of the words need not be in order, but is unconstrained (hence "free" recall). In general, words at the beginning and at the end of the list are more likely to be recalled than words in the middle; the first of these phenomena is called the primacy effect and the latter the recency effect. These recall phenomena are ubiquitous; for instance, in a semester-length course like this one, on the final exam information from the first few and the last few lectures is more likely to be recalled than material in between, and when people are asked to recall life events at random from their entire life, early childhood memories and very recent memories are more likely to be generated than a chance model of recall would suggest (along with, in this case, a third phenomenon, a reminiscence bump in which events from ages 15-25 are also disproportionately likely to be recalled; see Unit 9 for a possible explanation).
Modification of this basic procedure suggests that the primacy effect is an LTM effect (due to the fact that early words get more rehearsal time and thus have a higher probability of reaching LTM in the first place), while the recency effect is a STM effect (due to the fact that the last few words on the list are still in STM at the time of recall). To prove this, we can modify the procedure to eliminate one (but not the other) of these effects. The primacy effect can be minimized (if not reduced altogether) by increasing the rate at which the words on the list are read to the subject, while the recency effect can be eliminated by introducing a time delay (with an accompanying distractor task to prevent ongoing maintenance rehearsal) between the end of the list and the onset of the recall process.
Passive rehearsal experiments
Purpose: To illustrate that mere mindless rehearsal of information does not facilitate transfer of information from STM to LTM, and thus to highlight the important difference betweeen maintenance rehearsal and active rehearsal.
Method and results: In an illustrative study of this type, subjects were read lists of words. They were told that some of the words would begin with the letter "G", and that their task was to remember the most recently encountered G word (the last G word on each list). The lists were constructed so that the number of intervening words (not starting with G) varied, e.g., "garden, gun, friend, chair, gate, apple, gold, umbrella, cat, table, gift, cloud"; this comprised the independent variable. Obviously, G words followed by a large number of non G words would receive more (maintenance) rehearsal than those followed by a small number of non G words. Yet there was no correlation between the amount of rehearsal time and the probability of later recall when subjects were surprised, at the end of the experiment, by a request to list all the G words encountered. I hope the implications for your study habits are obvious! Only active rehearsal transfers information into LTM; to learn something, your mind has to be "in gear". Click here for more details on that from a practical standpoint.
False recognition experiments
Purpose: To illustrate that STM and LTM are organized along different lines (different encoding modes).
Method and results: Subjects were presented with lists of words to be remembered, and recall tasks that were either immediate or delayed. Enough information was presented to produce less than perfect recall, and recognition errors were examined and classified into three types: visual errors (words that look alike), auditory errors (words that sound alike), and semantic errors (words with similar meanings). For instance, if a word on the list was "dough", a visual error would be "cough", an auditory error "row", and a semantic error "batter". In general, auditory errors predominated during immediate recall conditions, but semantic errors predominated during delayed recall conditions.
Other facts about memory
LTM can be subdivided into two kinds of sub-memory systems known as episodic and semantic. Episodic LTM is your personal memory (for episodes from your life, which are unique to you); semantic LTM is your impersonal memory (for facts and general knowledge, which are not inherently unique to you). Thus, "What was your kindergarten teacher like?" is a question that taps into episodic LTM (everyone's answer will be different), while "What color are oak leaves in the fall?" taps into semantic LTM (everyone's answer, if informed and correct, will be the same). Episodic, but not semantic, memories tend to be emotionally tinged; see Unit 11 for some practical implications of this fact. Episodic LTM is organized chronologically, while semantic LTM is organized topically or semantically.
Memory is not a passive process but is generally constructive. This is why memories can be distorted by elements of the recall process itself. For instance, if a witness to a car accident is asked, "How fast were the cars going when they smashed into each other?", s/he is likely to give a higher estimate than if asked, "How fast were the cars going when they bumped into each other?" Once consolidated in this way, the memory will seem factual and objective and resistant to further change, a fact well known to defense attorneys and opinion pollsters.
Theories of forgetting
Forgetting is not a mere function of time or trace decay: for a neurologically intact 80-year-old, 75-year-old memories are likely as sharp and clear, if not more so, than very recent memories. Forgetting is likely due in most cases to a combination of two processes: interference, in which similar information impedes recall, and cue failure, in which the lack of a retrieval cue (think of a "hint" as an example) makes it hard to remember where you "put" a given memory. In emotionally charged situations, the more controversial notion of repression or "motivated forgetting" of unpleasant realities may also come into play; see Unit 11 for more. The fact that, with age, people's memories of their childhoods tend to take on a warm, nostalgic glow is probably due to the selective repression of unpleasant memories.
Study Guide
1. How do sensory store, STM, and LTM differ in terms of decay time, storage capacity, and encoding mode? Use these facts to discuss and explain the Atkinson-Shiffrin model of memory. What are four information transfer processes within this model, and how do they differ?
2. Discuss five critical experiments that shed light on the Atkinson-Shiffrin model. What is meant by the span of apprehension, and what causes it? What is dichotic listening, and what does it illustrate about attention and recall? What is free recall, what are the primacy and recency effects, and what causes each? How do passive rehearsal experiments illustrate the difference between maintenance rehearsal and active rehearsal? What are three kinds of recall errors, and how to these illustrate the differences between STM and LTM encoding modes?
3. How do episodic and semantic memory differ? Give some examples of each. What does it mean to say that memory is constructive? Give some examples.
4. Discuss and critique four theories of forgetting.