Thinking : Memory, human memory models


Constructing a model of human memory provides a way of testing theories about memory, of identifying problems and failures and a way of building computer representations of human memory. This model is generally accepted and receives its input from the environment through sensors and has influence on the environment by way of a response. Three main blocks of memory are identified and each play a specific role. There is some supporting evidence to show that these components of human memory have at least some physical equivalent in the brain. For instance, it is now generally accepted that Short Term and Long Term memory are different structures with different properties.

Short term memory is often refereed to as working memory because it is this component that is used during conscious decision making and problem solving activities.


Although it has been shown that models such as this one have considerable value in helping to explain the functions of memory and in predicting outcomes, it is not clear just how well such a model can really represent a biological memory within a human brain. It is argued that since the operation of the brain is entirely different from that of a computer based model and also that the brain cannot have these distinctive components as shown in the model, that such models can have no value.

In general, the question seems to be, can one system be properly represented on another system even when the underlying architecture of the two systems is entirely different?

Could we ever create a human like memory on a computer?

Long Term Memory - Declarative - Semantic


The word semantics relates to the study of meaning. Semantic memory is that part of Long Term Memory which contains knowledge about the world; knowing the meaning and definitions of things. Several theories have been developed to further explain the working of semantic memory. Such theories include frames and scripts which help to explain how everyday memory and knowledge is stored.

Storage Capacity

People have access to an enormous amount of stored knowledge. Consider the following questions.

  • What is the capital of France?
  • Who is said to have discovered America?
  • Can cows fly?
  • What does a Doctor do?
  • How do you call somebody on the telephone?

These are quite easy questions yet they imply a great deal of stored knowledge. A typical adult has a vocabulary of between 20,000 and 100,000 words.

Storage and access

How is all of this knowledge stored and how is it accessed?

There must be some strategy or storage method. Consider the following task.

  • Name a fruit beginning with the letter p.
  • Name an animal beginning with the letter l.

and now try,

  • Name a fruit ending with the letter b.
  • Name an animal ending with the letter w.

Why is it easier to access information using the initial letter as a key rather than the last letter?

Further evidence that memory is organised rather than simply stored can be seen by considering the following questions.

  • Did Aristotle have feet?
  • What is Beethoven’s telephone number?


It is suggested that people have schemes by which they organise information and consequently conversation between individuals is most effective if both share common schemata. Consider the following passage from a study carried out by Bransford and Johnson.

First you arrange items into different groups. Of course one pile might be sufficient depending on how much there is to do. If you have to go somewhere else due to lack of facilities, then that is the next step; otherwise you are pretty well set. It is important not to overdo things.

That is, it is better to do too few things at once than too many. In the short run this may not seem important but complicatins can easily arise. A mistake can be expensive as well. At first the whole proceedure seems complicated.

Soon however, it will become just another facet of life. It is difficult to see any end to the necessity for this task in the immediate future but then one can never tell. After the proceedure is completed, one arranges the material into different groups again. Then they can be put into their appropriate places. Eventually they will be used once more and the whole cycle will have to be repeated. However, that is part of life.

This passage was found to be difficult to comprehend by students until they were told that the topic was washing clothes.

Linguistic Determination

Semantic memory is not simply an association between words. It is concerned with concepts and ideas. Some, notably Benjamin Lee Whorf, argue that a language may determine the way one thinks. For instance, an Eskimo would have many different words to describe different kinds of snow but these would be useless to someone from the tropics. Can someone from the tropics therefore, ever think about snow in the same way as an Eskimo. However, not all concepts can be easily expressed in words so semantic memory must use more than a simple linguistic encoding.

Research into colour perception across varied linguistic and cultural groupings has indicated that all people regardless of culture percieve colour in the same way. This suggests that thought is not determined by language. However, it is possible that language may influence thought, although exactly how it might do this, as yet, unclear.


Agnosia is impairment in the ability to recognise objects, it is a disruption of the process taking someone from perception to meaning. One patient who suffered a severe head injury could not, for instance, recognise an onion by sight or by touch but when asked 'What is the vegetable that makes you cry when you peel it' produced the correct response. She knew that an eagle was a bird of pray but when questioned, indicated that she thought that it had four legs and teeth.

Patients who have difficulty in defining words precisely may still be able to produce appropriate subcategory names. They may know that a dog is an animal or a rose is a flower without being able to describe the concepts themselves. This broadly supports a hierarchical frame/semantic net type organisation for semantic memory.

Long Term Memory - Declarative - Episodic


Episodic memory involves the memory of particular incidents from ones personal experiences or involvement's such as sitting an examination last week or shopping at duty free in the airport before going away on holiday or remembering a significant news item.

Episodic memory relates to personal experience of a situation or episode. Two people perceiving the same event may have different episodic memories of that event. This may depend on their own interpretation steered by their individual experiences.

Episodic memory may be treated in one of several ways.

  1. It may be forgotten quite rapidly.
  2. A particular episode may be absorbed into semantic memory and contribute to a schema or generalisation concerning similar events. In such a case, specific details are lost and only the generalisations retained.
  3. A particular event may be retained in episodic memory and the details may be remembered specifically.


Experiments have shown that even two and three month old babies show some signs of acquiring episodic memories. When they are allowed to invoke movement in a mobile from a string attached to one of the babies' feet, a kicking response is evoked. Babies can remember this response over many days and show some signs of association with environment. When trained on several mobiles, babies also show some signs of performing generalisation.

Long Term Memory - Declarative - Autobiographical


A particular kind of episodic memory concern episodes, which only relate to personal experiences. These memories are known as autobiographical memories. Autobiographical episodic memory usually includes details about a particular time, place and details about objects and events that were experienced. The distinction between semantic and autobiographical episodic memory should not be emphasised too strongly.

Semantic memory concerning weddings for instance would be built up from several autobiographical episodes. This semantic memory may then in turn be used to help understand and interpret a new experience. Studies have shown autobiographical memory to be quite accurate. Studies have shown autobiographical memory to be quite accurate. Studies with eyewitness testimony have conversely shown that this sort of memory is highly suspect.

Linton's study

Several experiments have been carried out to test the accuracy and extent of autobiographical memory. Marigold Linton, in 1982, undertook a six year personal study of her own memory. Each day she placed at least two events in a diary and used these later to test her recall.

She noted a tendency to forget the more unpleasant events. She also noted that items picked out at random for testing that had been picked out before were better remembered. This shows that rehearsal was playing an important role.

Brewer's study

In 1988, Brewer carried out a similar study with ten subjects but the events they recorded were timed random events. Recall was tested after 0, 23 or 46 days. During these tests, there were many omissions. On many occasions the wrong event was produced and there were fewer although significant number of correct recollections.

Watergate witness

During the Watergate trials in the USA, John Dean gave such detailed accounts of events that the press called him the man with the tape recorder memory. Since many of the events had actually been taped, it became possible to verify the testimony. It turned out that the broad outline of Dean's testimony was indeed accurate but the detail was highly inaccurate. For instance, Dean's own role was presented as more important and central than it actually proved to have been.

It seems that autobiographical memories are reasonably free from error when considering the broad outline of events. However, errors begin to occur once the details of specific events are required.

Brain Damage

Brain damage caused by a stroke or a blow to the head for instance can have a variety of effects on autobiographical memory. Some patients can remember distant events but fail to remember events closer to the accident. Some patients can remember broad outline but have great difficulty recalling any detail at all.

Long Term Memory - Non-Declarative - Procedural


Procedural memory is known to be extremely robust. It is the memory used for skills; for doing things. The ability to perform tasks such as riding a bicycle, knitting or soldering is within the domain of procedural memory. It is known that people with severe memory loss, even for their own name, still have intact procedural memory. Patients suffering from learning problems are often sill able to learn new skills, even though they may not know that they have learned them. This may also explain why expert skills are difficult to describe to others.

Amnesia Experiments

There have been many experiments carried out in the area of procedural learning with amnesiac patients. These experiments are interesting because the patients, due to their amnesia, are usually incapable of declarative learning.

As long ago as 1911, the Swiss neuropsychiatrist, Claparade, secreted a pin in his hand when meeting a lady suffering amnesia. The next day, the lady refused to shake his hand although she did not know why.

In 1968, Corkin reported the case of a densely amnesic patient, due to lesions in the hippocampus, was still able to learn simple motor skills even though he could not learn even simple new facts. This patient had no conscious knowledge of his new skills.

That skills or procedural knowledge can be learned without and conscious awareness may account for the fact that experts often find it very difficult to verbalise skills that have been learned over many years.

The Hippocampus

It has been shown that the hippocampus plays a vital role in the laying down of new memories and that damage to this area of the brain can prevent the acquisition of new knowledge. However, it also seems that even with this deficit, new procedural skills can still be learned.

Graphic : Location of the Hippocampus

Long Term Memory - Non-Declarative - Conditioning


Conditioning leads to a specific type of memory. It results from pairing actions with reward or from pairing actions with parallel stimuli. There are essentially two kinds of conditioning:

Classical Conditioning is where a neutral event such as a bell ringing is regularly paired with a response-evoking event. Eventually the response can be evoked from the neutral stimulus.

Operant or Instrumental conditioning is where a reward is given for progressive improvement at a task that may start with random movement and due to the progressive reward, will end up in the completion of the task.

Classical Example

                : PavlovAn experimental rat in a wire-bottomed cage is given a mild electric shock to the feet at the same time as a bell is sounded. The shock causes the rat to jump slightly. Continued application of this treatment leads the rat to jump when the bell sounds, even though there is no shock presented. The famous research into classical conditioning was that of Pavlov

Practical Applications

Many phobias have been acquired as the result of conditioning. These may include a fear of spiders or a fear of open spaces. It is possible to apply various methods to eliminate these phobias by techniques such as desensitisation, modelling, flooding or operant shaping.

Panic attacks can be suffered by otherwise healthy patients. These are conditioned initially due to random fluctuations such as palpitations of the heart or breathlessness for example. A mild panic reaction to these can increase their severity. Eventually, otherwise unimportant fluctuations can lead to the onset of a panic attack. It is possible to treat such attacks by a process of education, and then control of self induced attacks.

It has also been suggested that certain types of advertising, whether deliberately or not, promote products by a type of conditioning. In many such cases, the product is associated with a pleasant neutral stimulus such as an attractive scene or an attractive person. It is then desirable for the advertiser to strengthen the association between the product and the pleasant stimulus over a period of time.

Short Term Memory - Phonological


The Phonological Loop is itself composed of two sub systems. These are an auticulatory control system which hold information by rehearsing it sub-vocally and a phonological store which is a store for auditory information, like an inner ear, and can hold information for about 2 seconds.

Experiments have shown that the phonological loop is crucial for the learning of language. Children with a deficit in capacity of the phonological store have greater difficulty in language acquisition. Adults who acquire this deficit seem to suffer no ill effects unless they try to learn a second language.

Articulatary System

It is believed that the auticulatory control system has a limited working capacity. In experiments to remember lists of letters, through sub vocal rehearsal, subjects perform better on lists such as: K, W, Y, L, R, Q than on lists like D, B, C, T, P, G. Phonologically similar sounds are harder to remember than dissimilar sounds.

The limited capacity can also be demonstrated by attempting to remember lists of words.

  • Some
  • Harm
  • Bond
  • Yield
  • Hate
  • Immediate
  • Considerable
  • Individual
  • Association
  • Opportunity

The list of words in the first column are shorter that those in the second. They are consequently much easier to remember. Baddely has shown that words containing more syllables are harder to remember because they take longer to articulate and place a greater demand on the auticulatory control system.

Using auticulatory suppression, to prevent subjects from rehearsing, lead to a finding that the average person could remember about 1.5 second's worth of sound (words).

Comparisons between languages such as Welsh and English showed that children who spoke welsh as their first language had a smaller digit span. This was accounted for by the fact that welsh digits take longer to say than English equivalents.

Phonological Store

The phonological store or auditory imagery system allows us to create images of actual sounds from our own memories. We may be able to imagine a key part of a musical recording or hear waves crashing on the beach from the memory of our holiday. Some people are able to improve their auditory performance by imagining themselves saying sounds and thus supplementing the auticulatory control system.

One way to show that we do not normally use the image of sounds when we read is to see how laborious it is to read the following.

Iff yue sowned owt thiss sentans tew yoreselph, yoo wil komprehenned it.

An observation which helps to demonstrate that the sound of words is treated differently from their meaning comes from brain damaged patients who have difficulty in reading pronounceable non-words such as fleep or spart. One patient could read bee but nor be and could read words like castle and trombone much easier than hope or justice.

Short Term Memory - Central Executive


The Central Executive is the controlling system of working memory and is more complex than either of the two slave systems. This system controls attention and it is necessary for cognitively demanding tasks such as problem solving and comprehension.

The Central Executive also has a limited processing capacity, which is related to the capacity of working memory and also to Millers magic number seven.

Memory Test

Working memory span can have a significant effect on a person's ability to comprehend. Consider the following test for working memory span. Read each of the following sentences and cover each up after it has been read. On completion, recall the last word from each sentence.

  • The greengrocer sold many apples and oranges.
  • The sailor had been around the world several times.
  • The house had large windows and a massive mahogany door.
  • The bookseller crossed the room, scowled and threw the manuscript on the chair.

This experiment was devised by Meredith Daneman and Pat Carpenter to test working memory span. Recalling all four words is quite good, some subjects have difficulty managing more than two.

Short Term Memory - Visio-Spatial


The Visuo-Spatial sketchpad is used to hold visual or spatial information in short-term store, where it may be manipulated. It is similar to the auditory system and indeed, there are suggestions that the Visuo-Spatial sketchpad is also composed of two separate sub systems, one visual and one spatial. Information may enter the system either from the outside world via iconic store, or from Long Term Memory.

Shephard's Boxes

An interesting experiment by Roger Shepard showed how visual imigary is manipulated within the brain in a similar way to the physical manipulation of an object in the hand, with more detailed manipulations taking longer. The task is to find out if the arrows on the unfolded boxes will meet each other, point to point.

Graphic : Shepard's Box Problem

The Problem

Graphic : Shepard's Box Solution

The Solution

Visio-Spatial Capacity

Graphic : Capital
              F puzzle

The large capital letter F shown has 10 corners. Close your eyes and continue to imagine the letter F. Starting at the bottom left corner and proceeding clockwise round the letter, Label each corner as 'YES' if it involves the top or bottom line of the letter and 'NO' if is does not.

The correct response is, yes, yes, yes, no, no, no, no, no, no, yes.

Subjects who were asked to say yes or no out loud performed much better that those asked to point to yes or no located in different locations on a page. This showed that the two visual tasks were competing for the same processing resource.

Visual or Spatial

Evidence has been collected regarding the question of whether vision is image based or spatially based. The neurologist, Gordon Holmes, during world war 1, reported the case of a brain damaged patient who was able to identify objects but could not locate them spatially to pick them up, reaching in the wrong direction when attempting to do so.

Another patient was found to be incapable of performing image based tasks such as judging the relative size of a dog and a cow but was able to perform spatial rotation and manipulation tasks normally.

Sensor Input

Sensory Channels

There are five sensory channels used by human beings to percieve the world. The illustration shows which areas of the brain relate to each sensory channel.

Graphic : Sensory Channels

Autonomic Responses

Many senses are able to form Autonomic responses with motor neurons without waiting for the intervention of the brain. This sensory-motor relay process may take at the spinal cord in emergency conditions such as burning ones finger.

The Thalmus

Sensory input from the Central Nervous System, except the olfactory system, converges on the Thalmus. The Thalmus is the main sensory relay station within the brain.

Graphic : Location of Thalamus

Sensor Registers


Sensory Registers are micro memories that are directly associated with the senses. Information in these registers may persist for only a few milliseconds (possibly longer for strong stimuli). Examples of such registers are the Iconic store that holds visual information for a short time, and echoic store that holds auditory information for a short time.

The effect can be witnessed by looking at a bright object for a short time and then closing and then covering the eyes with the palms of the hand to experience total blackness. It will be noticed that the brightest objects persist in the field of vision for a little while before fading away.

Efron's Experiment

An experiment by Efron in 1970 involved the presentation of short duration tones to the ear. It was found that regardless of the duration of the tone between say 30 and 100 milliseconds, subjects tended to perceive tone duration of 130 milliseconds.

Segner's Experiment

In 1740, the Swedish scientist Segner measured the duration of visual persistence or the iconic register. He attached a glowing coal to a freely spinning cartwheel and adjusted the rate of spin until subjects just perceived one complete unbroken circle of light. By measuring the time it took for that revolution he was able to estimate the duration of iconic persistence to about 100 milliseconds.

Human Memory Response

Motor Cortex

Conscious movement is controlled by the motor cortex in the Cerebral Hemispheres. This area occupies a band around the outside portion of the cortex stretching approximately from ear to ear. It has been shown that muscles in the body map to specific regions of the motor cortex in a similar way to senses mapping to the sensory cortex.

Fine Muscle Control

The Cerebellum is an area of the brain that is believed to be responsible for fine muscle control. All actions such as riding a bicycle, driving a car or playing tennis are co-ordinated within the cerebellum. Many of the complex movements that the human body can learn are co-ordinated here but they may not be accessible to conscious decomposition. For example, an expert football player may have difficulty in explaining exactly how a specific movement is carried out; it is carried out automatically by the cerebellum after possibly years of practice.

Graphic : Location of Cerebellum


"Your Memory: A Users Guide" : Alan Baddeley: Penguin. 1993 and "Human Memory: Theory and Practice". again by Alan Baddeley are the best references for this work.

There are also many internet sources, primarily from the university sector. However, these are often transitory and so it is best to use a search engine to search for concepts such as ‘Human Memory’.