Thinking : Memory, memory capability


We may occasionally and subjectively estimate the effectiveness or the capability of our own memory. Some of us may even reflect as to how good it is compared with other peoples memory or even the memory of computers. Some more objective work has been carried out in terms of the accuracy of human memory and also in terms of speed and capacity. Much of this work highlights the limitations of memory whilst still accepting the fact that human memory is an impressive system.

Human memory is not limitless but no one has ever documented a case where a human is unable to store more information because his/her memory is full. The speed of human memory can be seen as a delay in our ability to access facts. Accuracy of human memory has considerable implications in many areas of life but none of these has been investigated more thoroughly that the area of eye witness testimony.


It has been stated that human memory is not, nor can it be limitless. What exactly does this mean to people in their every day lives. There are many possibilities. Memory could be so vast that to all intents and purpose, given the life expectancy of humans, there is no real limit to what we can remember. Memory could be very leaky so that there is always space for new information as old information leaks away. Memories could be compressed as new information comes in and merges with old. The consequence of this is that details may be lost whilst the generalisations remain.

So which is true, or is there some other explanation for our continuing ability to store new information in spite of a lifetime of accumulated knowledge?

Memory Capacity


The capacity of the human brain is unlikely to be limitless. It is a difficult job however to estimate this capacity in terms of storage space as can easily be none with a digital computer. It is likely that there would not be a point reached where a human memory is ‘full up’ and no more can be stored. It is more likely that new memories would displace or interfere with old memories.

Von Neumann

Von Neumann

Von Nuemann

Von Neumann made an early estimate of 1020 bits based on the number of impulses conducted in the brain during a lifetime. This estimate is believed to be an over estimation. An estimate of bit capacity based on the number of synapses in the brain would yield a figure of 1013 to 1015 bits.



Thomas Landauer

Thomas Lander used other techniques to estimate the size of human memory based on actual functional capacity. Analysis of experiments based on the acquisition of memories in various forms such as text, pictures, music etc. lead to a conclusion that humans remember approximately 2 bits per second under all experimental conditions. A lifetime of learning would therefore lead to a memory capacity of 109 bits, or a few hundred megabytes.

Landers experimental work was aimed at finding out what sort of computer system, in terms of capacity, would be needed to mimic the capabilities of a human. However, it is not simply capacity that is important.

Storage method, retrieval and speed are also important characteristics. Although this experiment yielded a fairly low estimate, it did not account for much procedural knowledge, mainly concentrating on declarative forms.

Memory Speed


The speed at which we can process information is limited by the physical construction of our brains and the organisation of the mechanisms within. Neurons (brain cells) have a characteristic operating time which sets certain limitations on speed. The organisation of memories within the brain may also set other limitations. Working memory may also limit processing speed. Effects such as the mental rotation tasks suggest that working memory is organised to process spatial information in a specific way.

Neuronic Constraints

The way neurons work must impose physical constraints on the performance of the brain. If very approximate figures are used, then each neuron can be said to have a characteristic operating time of 1ms and a simple neural circuit of 10ms. This means that a simple neural circuit can perform about 100 computations per second compared with a modern microprocessor performing 10,000,000 instructions per second.

Cognitive Functions

Cognitive functions will take at least 10 times longer than the minimal 10ms. Thus the fastest deliberate thought would be in the order of 100ms or more. Typically, the fastest cognitive thoughts take 1s or more. It is possible for experts to play a game of 10s chess where the time allowed for each move is 10s. Humans could not play 1s chess. Various studies have been carried out to show that humans take several seconds to perform a single problem solving operation. It is also suggested that humans can read information out of Long Term memory in well under 100ms but will take several seconds to write new information into LTM.

Many everyday experiences rely on the fact that our brains are slow at certain things. Because our vision system takes several milliseconds to process an image, it is possible to broadcast television pictures at 25 frames per second without us being able to detect the breaks. We see a continuous moving picture.


Studies of the brain during aging have shown that slowing does occur. Experiments have shown that where speed is a factor in performance, that older people perform less well than younger people. However, if speed is within the control of the individual, then older people can take advantage from their greater experience. In other words, of older people slow things down a little their memory and computational abilities remain good.

Memory Accuracy


Graph : x=Age, Y=Unis of Impairment

How accurate is a person's memory?

Although we may often decide that our memories are at fault because we fail to remember something, when we do remember things we are often convinced that our memories of those incidents are faultless.

Most of us have witnessed or been involved in arguments that are basically disagreements about what happened on a particular occasion.

Two people who witness a situation, may to have different memories about the same event. In some cases, these differences can concern some key item.

If this happens between two senior managers in business say, the results can be costly.

Eyewitness Testimonies

There has been considerable interest in the area of eyewitness testimony. In court cases, a great deal of reliance is placed on eyewitness testimony so it is important to work out how reliable it really is. The psychologist, J.M.Cattell (in 1895) showed that subjects were not very good at recalling, with any great accuracy, everyday events such as what the weather had been like the previous week. In fact he concluded that people cannot state much better what the weather was like a week ago than what the weather will be like a week hence. It has also been found that people have poor memories for everyday objects such as coins. Instead they seem to have generalised recollections for categories rather than specific memories for individual items.

Even though it may be expected that people remember unusual events such as crime, more clearly, there are many factors that can interfere with memory. Such factors include extremes of emotion, here memory for the event may be stronger but it may also be less for the associated detail. Memory for faces and facial reconstructions (artist impressions) have been shown to be unreliable. Leading questions can also influence memory.

Loftus’ Car Crash

In one experiment (Elizabeth Loftus), subjects were asked to watch a film of a car crash. All subjects were than asked the same questions about the crash (About how fast were the cars going when they hit each other?) except that the word 'hit' was replaced by one of the words, 'smashed', 'collided', 'bumped', or 'contacted'. Estimates were higher when the word 'smashed' was used (40.8mph) and lowest when the word 'contacted' was used (31.8mph) with the expected range in between. Furthermore, when subjects were questioned a week later about if there was any broken glass, those subjects who were given the word 'smashed' were consistently more likely to report 'yes' than the others. In fact there was no broken glass.

An example of a leading question that could influence memory, or at least the report from memory, is:

Instead of asking, "Did you see a broken headlight", you ask: "Did you see the broken headlight."


"Your Memory: A Users Guide" : Alan Baddeley: Penguin. 1993 provides information about each of the topics, speed, capacity, accuracy and limitations. The world wide web is also an interesting source of information.

Unified Theories of cognition: Allen Newell: Harvard. 1990 also provides some information relating to the capability and speed of human memory and thought processes.