Models of Memory

Memory: the central topic of cognitive psychology

Memory is paradoxical: amazingly good and yet amazingly bad (sometimes)

We are memories. We get our sense of self through our memories

Basic Aspects of Memory

1. Encoding

getting information into the system. This has been the focus so far. Attention is critical for later memory. If something is never attended to, it can’t be encoded.

2. Storage (retention)

preserving information over time. Can be as long as a lifetime. It seems to be functionally infinite.

3. Retrieval

recovering information from the memory system. The main purpose for memory. We are (usually) very fast and accurate retrieving information. Thus, our memory system has to be efficiently stored. We can even quickly conclude that we do NOT know something. (What color of Obama’s toothbrush?)

How is memory measured? (The Three R’s)

1. Recall

information must be reproduced.  (what is capital of Kenya?) Stringent test of memory. There are no cues to aid you. Examples: Essay and fill in the blank test. If they are able to recall info, they usually know information relatively well.

2. Recognition

information must be recognized. (Is capital of Kenya Tokyo, London, or Nairobi?) Not always but usually recognition is easier than recall. Teachers like these because they can change difficulty. It depends on the similarity between the correct answer and the false items. False items are referred to as foils or distractors. Recognition can be made just as hard as recall: Is the capital of Kenya Cairo, Khartoum, or Nairobi?

3. Relearning

information can be relearned faster if it was learned previously. This is a very sensitive measure of memory. The person to discover this is the father of memory research. He was Hermann vonEbbinghaus (1850-1909). His goal was to quantify memory and forgetting. He is the father of thenonsense syllables. (BAF, ZUP, TEJ) .

Used these made up words because no one will have ever seen or heard these words before. Took away a confound. He viewed his research as a true measure of memory. He created many lists of nonsense syllables. He had subjects study these words and memorize them. Then he would have them come back in and relearn the words. He conducted the study over two years (163 trials). He used himself as a subject (bias).  He studied the list till he had two perfect recitations. If he met that criterion, he would stop studying. He keeps track of the length of time to meet criterion. This was the original learning time. Then he we would delay 20min to 31 days. He would then get this list again and time how long it took him to recite the list perfectly two times. He would track this time to meet the criterion as well. This was the relearning time.

Method of Savings: original learning – relearning time = savings (how much better you do to relearning something from having learned it the first time). Expressed as a percentage of learning time = savings/original learning time. Data can be plotted to discover trends (Dependent variable: percent savings, Independent variable: Elapsed Time). Curve goes down over time but not linearly. Very rapid forgetting early on. Relatively low rates of forgetting over time. After 20 over 50% remembered. After 1 hour, under 50%. After 9 hours, at 40%. Over time though, memory never got below about 30%.

Ebbinghaus proved memory could be quantified and measured. Most forgetting takes place immediately.

Does overlearning help? Additional study after criterion achieved leads to increased saving. Nonsense syllables are ecologically invalid.

Are nonsense syllables devoid of memory? Not really because people try to find meanings

Is memory static and passive? Attic analogy. Putting stuff in attic to use around christmas.

Later researchers: remembering is active, reconstructive process (Bartlett).

Models of Memory:

Two basic types

Structural models

emphasize the organization of memory. (books in a library)

Process models

emphasize the operations carried out in memory (like a set of directions)

Historically, people started with structural models.

Data to be explained:



Serial position effect – subjects see a list of words, the words are presented one at a time. Afterwards, they’re asked to recall them (in any order). It turns out that what words you recall depends on at what time you saw the word: Serial Position Curve. People tend to remember the beginning of the list and the end of the list. Early good performance is called primacy effect. Late good performance is called recency effect.


Why are primacy effects higher than recency effects?

Atkinson and Shiffrin (1968)

A serial model of memory is used again. It is an information processing model as well. Info moves through sequence of stages and does its own thing in cog system as it moves along. Designed to explain the serial position effect

Three components:

1. Sensory memory

relatively complete record of environment. Decays quickly (few hundred msec)

2. Short term memory (STM)

working memory of limited capacity (~7 items). Without rehearsal, decays quickly (~20 sec).

3. Long term memory (LTM)

more permanent record of information. Can last a lifetime. Functionally infinite. Coded in a variety of ways. Information lost in various ways?

Neurological Support

Damage to the hippocampus seems to disrupt transfer from STM to LTM. Seen in one type of amnesia. Fencing accident with lance or something going up through nose and damaging hippocampus. He remembers life before accident. Can’t create new long term memories though. He was known as patient NA.

Primacy Effect: The first few words are repeatedly rehearsed in the cog system. Therefore it is more likely to enter the LTM. End of study: easily retrieved from LTM

Recency Effect: Last few words are in STM at the end of the experiment. Simply read off before they disappear.

You can mess with subjects by preventing them from recalling immediately after experiment. This will wipe STM. Having them do an additional task immediately after seeing the words. One thing would be count backwards by 3s from 471. STM will fill up with these numbers instead of words after about 30secs. Then recency effect is not seen. It looks like the middle of the pack. Primacy effects still hold up even after the trick.


Stuff in middle of list not always forgotten.

Von Restorff Effect (Isolation Effect)

Unusual, distinctive item in the middle of the list: very high level of recall.

Hard to explain via Atkinson and Shiffrin.

Related articles
  • Models of Memory (
  • Long Term Memory (
  • Concept and Knowledge Structures (
  • Process Model of Memory (
  • Pattern Recognition (
  • Short Term Memory (