1 Background

This Briefing Note (BN) introduces the basic processes and limitations of human memory. Along with discussing the various types of memory, it demonstrates how memory is related to task achievement. Understanding these processes and the limitations of memory along with learning how to improve your memory will lead to better performance not only in flying, but also in a variety of other domains.

2 Introduction

All intelligent systems that process information require memory to direct actions and to store the results of those actions. Speaking, reading, listening, decision-making and taking action are only efficient when the brain has the ability to properly process, store, and recall information.

Human memory has a variety of functions. It is able to maintain a very detailed "picture" of the past that helps to identify and classify sensory stimulations (sight, hearing, touch, smell and taste). Memory also stores experiences and knowledge to help us in similar situations in the future.

In spite of its power, human memory is limited and can fail when storing or retrieving information. Some things are easy to recall and others are not. For instance, sometimes it is easy to recall what you ate yesterday evening but you forget the frequency for the last air traffic control (ATC) clearance. It is possible to have information “on the tip of your tongue,” but only to be able to recall it after you needed it.

Overall, memory is a complex process that is integrally involved in all information processing steps (perception, comprehension, decision and action). Memory is not simply a single process; rather, it is generally considered to be composed of multiple stages in which information is processed and stored differently. These stages are classified according to the amount of time information is stored at each stage. The three basic types of memory are:

• Sensory memory
• Short-term memory (STM)
• Long-term memory (LTM).

This BN addresses each of these three types of memory and discusses the relationship between memory and task achievement.

3 Data

Memory is a topic that has been researched in great detail, and vast amounts of data exist. As seen in Table 1, memory develops over a person’s lifespan, and the amount of information we can remember at any particular moment increases as we approach adulthood. In Table 1, Digit Span refers to the number of digits that an average person can recall when read a list of numbers. Digit Span can be increased with practice, but often returns to baseline if practice is stopped.

Table 1: Digit Span as a Function of Age, Adapted from Woodworth and Schlosberg (1954), Experimental Psychology Revised.

The amount of information that we are holding in memory can affect our reaction times. As seen in Figure 1, as the number of memorized digits increases, the amount of time it takes to identify whether or not a “target” number was in the memorized set increases. This demonstrates that as we hold more information in memory, our response reaction

Figure 1: Reaction Time as a Function of Number of Memorized Digits, Adapted from Sternberg (1966) in Engineering Data Compendium: Human Perception and Performance Volume II.

times will become longer since we have to search more information to find the target information being sought.

4 Sensory Memory

Information enters sensory memory after the information has been sensed and attention been given to the sensed information (See Figure 2). Sensory memory allows us to retain impressions of sensory information after the original stimulus no longer exists. Sensory memory exists for both visual (iconic memory) and auditory (echoic memory) senses. Sensory memory retains a quite accurate and whole “picture” of the information that was sensed and attended to.

Figure 2: Sequential Process of Information in Memories

Sensory memory is characterized by being outside of conscious control and occurs automatically. The role of sensory memory is to initiate the first steps of perception. These steps include extracting and identifying the features of a stimulus after the initial reception of information regarding the stimulus. Not only can sensory memory identify features, it can also recognize meaningful patterns of cues in the information.

In general, the duration of sensory memory is very short—less than 0.5 seconds. Visual sensory memory, for example, lasts about 250 milliseconds. This short duration explains why still pictures displayed consecutively, with a frequency higher than 4 Hz, are seen to be continuous. This is how we can perceive motion with our eyes.

Although sensory memory is very short in duration, it is almost limitless in the amount of information it can hold. Our ability to attend to the information, however, is limited. In practice, sensory memory retains all information received by our senses for a very short period of time, but our ability to pay attention to the information limits what actually makes it through to the next stages. The specific information that makes it through is determined by active and passive selection processes within the memory system:

• Active processes are oriented-by-concepts in which there is an intention to extract useful and expected information. Useful and expected information is determined by the goals of the information-seeking behaviors. These goals represent the concepts. Active processes are guided by attention mechanisms that focus on specific portions of the total information since there is not enough time to analyze all information held in sensory memory. These attention processes allow us to analyze a specific aspect of the stimulus with great accuracy. Most of the balance of the sensory information that we do not attend to is lost.
• Passive processes are oriented-by-data in which our brains automatically process and analyze information. No selective attention is needed to obtain and process the information. Most of this information is subliminal, and we are never aware that we are actually processing the information, although we may be able to respond if necessary. One example of passive processes at work involves not-so-subliminal alarms and warnings. Alarms and warnings are designed to elicit automatic processing. Alarms and warnings must be perceived and processed without the need to selectively focus attention to notice the warning. As such, the intensity of alarms and warnings must be much greater than the minimum thresholds of the human senses and must be of sufficient intensity to override all other active processes that we may be engaging in at the same moment. After this initial automatic response to the alarm or warning, we will then engage in active processes as we turn our attention to the alarm and attempt to determine what caused it.

5 Short-Term Memory (STM)

After passing through sensory memory, and if more attention is devoted, information then enters short-term memory (STM), which stores a limited amount of information for a limited amount of time. The duration of information retention in STM ranges from seconds to minutes (on average 15-30 seconds).

STM content is different from that of sensory memory. Stored information is not in the form of a whole picture of the events as with sensory memory. STM stores information as an immediate interpretation of the events rather than as a picture of the actual events. At this point, there can be a difference between the interpretation of the events stored in STM and the actual observed events originally in sensory memory.

The information held in STM may be:

• Recently processed by sensory input via sensory memory
• Items recently retrieved from long-term memory
• The result of recent mental processing.

5.1 Key concepts of STM

The key concepts associated with STM are duration of retention and finite capacity. In order to overcome the short duration and limited capacity of STM, information must be consciously repeated or rehearsed. Rehearsal can include articulating information out loud or by mental repetition. During rehearsal, information re-enters the STM store and is retained for another short period of time. A good example of how we use rehearsal is when we meet an old friend who gives us his phone number. If we do not have a storage device to immediately store the phone number, we need to repeat it out loud or mentally until we can write it down or store it in some other device. Repeating the phone number for a long time could even lead to the number being stored in long-term memory.

Several experiments have been conducted to determine the size of STM. Results indicate that STM holds 7 + 2 pieces of information. In other words, STM can store 5-9 items at any given point. However, it is important to understand that these items of information can be either very small or very large. The important point is that the items are some form of meaningful information. As such, a piece of information could be a single digit, a letter, a whole word or something more complex such as a whole phrase or multiple-digit number. The complexity of information that can be stored within STM is something that can be increased with practice.

5.2 Limitations of STM

The relatively limited capacity of STM and the short time information is held there accounts for why it is so easy to forget information at this early stage. Forgetting in STM occurs by two main mechanisms, interference and mnemonic trace decrease.

• Interference. The effects of interference are largely related to the limited capacity of STM. Interference occurs when information cannot be processed because STM is already full. One of two things typically happen at this point. New information can replace old information, in which case the old information is lost; or old information can block new information from entering STM, in which case the new information will not be remembered. Unfortunately, we cannot always consciously choose which information should remain in STM. Generally, “stronger” information will remain in STM. Stronger information is whichever information is easiest to recall. The strength or the weakness of information is relative between the competing information sources. The strength of information depends on its frequency of occurrence, importance to the current task and our past experiences. STM issues are particularly relevant to jobs where there is a great deal of task sharing. Maintaining information in STM requires using a large amount of attentional resources. These are not always available when we are trying to manage multiple tasks, each of which has its own important information competing for our limited STM.
• Mnemonic Trace Decrease. A mnemonic trace is a fragment of the original memory and is the functional unit of a memory. Without rehearsal, the mnemonic trace of a piece of information stored in STM will disappear. When the mnemonic trace is gone, the information is lost from STM.

5.3 Managing STM

It is possible to employ various strategies to help manage STM and to increase the effectiveness of our memory storage processes. Strategies for managing STM include:

• Knowing the limitations of STM, being able to anticipate the point at which you will start to forget information, will allow you to better manage your tasks and lessen the amount of important information that is forgotten.
• Using training and experience to know what information to select and elaborate on. It is important to practice elaborating on those items that are most important to the task at hand. Practice and training will increase your ability to know which information to store in STM for each particular task.
• Using support devices. These devices can help when you know the information you need to remember is going to exceed STM capacity. These devices could be as simple as a pen and paper or as complex as a computer system. Using support devices is critical in aviation. When the amount of information appears to be greater than your STM capacity you should automatically use some type of support device. The support device can also be another team member if there is not sufficient time to write down the information. If you have some doubt about the accuracy of the information, do not hesitate to confirm it with another team member or with its source.

6 Long-Term Memory (LTM)

Long-term memory (LTM) is where experiences and knowledge are stored permanently. The length of time information can be stored and the capacity of LTM are unlimited. The challenge is to retrieve the correct information when it is needed. Failures in retrieval account for most of the forgetting related to LTM. Information stored in LTM is never truly lost, but it can be very difficult, or almost impossible, to retrieve. Therefore, it is important to focus not only on how to integrate and maintain information in LTM, but also on how to retrieve and interpret it correctly.

6.1 Encoding phase

Encoding is the process by which information stored in STM is transferred as cognitive data to be integrated and stored in LTM. This process assigns meaning and weight to the new information and integrates it with already stored information. The more deeply information is encoded — the more meaning and weight that are assigned — the easier recall will be later. Encoding can also be deepened by the establishment of links or associations between new information and stored information. Links and associations allow the organization of information into a network of hierarchies. Stronger links and associations lead to easier retrieval. In addition, each new piece of information is spatially and temporally indexed based on the location in the brain and time the information was processed. Finally, encoding may be conscious or unconscious, and attention and learning intent can also play important roles.

Each of these mechanisms provides a context for the encoded information (associated spatial/temporal data, links with stored information, knowledge organization, processing depth, etc.) that goes far past simply getting the information into LTM. This context is also extremely important when it comes time to retrieve the information.

Information may be encoded in a variety of ways. Encoding is generally described as being either verbal or visual in nature. Although we can encode information by both methods, people usually prefer or favor one type of encoding over the other. Because people may have different encoding preferences, good training courses must present information in a way that is conducive to each type of encoding. Presenting information that promotes only one encoding style would not be good for those people who favor the other encoding style.

6.2 Storage Phase

Encoded information can be stored in LTM indefinitely. However, natural “fading” may occur in the various links and associations. Active processes are required to prevent this fading. The first of these processes is consolidation. Consolidation strengthens the mnemonic trace of information and decreases the likelihood of interference. Information is transformed and integrated into stored knowledge. Consolidation explains differences in memorization and recall of recent events (episodic memory: recent facts within their contexts) versus older events (semantic memory: concepts and context-independent facts). Episodic memory is fragile; semantic memory is much more stable and reliable. Consolidation is time consuming and lasts well after the information is initially encoded. Much of consolidation occurs during sleep. Also, rehearsal is vitally important to the process of consolidation. Both sleep and rehearsal can help prevent loss of information during consolidation. Any information that is lost during the consolidation phase can affect the storage of other information that was associated/linked to the lost information.

After consolidation, memory may need to be rebuilt based on the new information. Old memory is compared with the new information, reprocessed and updated accordingly. Old memories are not necessarily lost; however, the new, updated memory will likely be more powerful and easier to remember.

The consolidation and rebuilding processes demonstrate that LTM is not a static entity. Rather, it is active and requires constant reorganization. Our individual knowledge is changing and enriched with experiences as we live. Each person has his own knowledge according to his experiences and rebuilds and reorganizes his LTM accordingly.

6.3 Declarative and Procedural Knowledge

Knowledge is often broken down into two categories, declarative and procedural. Declarative knowledge is conscious, verbal and static, and it includes episodic and semantic memories (See Figure 3). Declarative knowledge can most easily be described as knowledge about facts, methods and other static pieces of information. In contrast, procedural knowledge refers to physical and mental skills that are dynamic in nature and involve many steps. Unlike declarative knowledge, which can be easily described and shared with others, procedural knowledge is more tacit, meaning it cannot be articulated easily to others. However, despite an inability to articulate the steps, the procedure itself is carried out with ease, especially as expertise increases. It is important to differentiate between procedural knowledge and knowing the steps involved in a procedure. Knowledge of the steps of a procedure and being able to clearly define the steps to others is actually declarative knowledge. Performing the procedure itself without having to think about the steps is procedural knowledge. For example, in aeronautics, being able to write down a procedure described in a manual (like Standard Operational Procedures) is declarative knowledge. Performing the procedure by putting together all of the steps without having to look at the manual or think about the steps is procedural knowledge. A person who can perform the procedure with ease might have difficulty actually writing down all of the steps.

Figure 3: Procedural and Declarative Knowledge in LTM

6.4 Knowledge Organization in LTM

Declarative and procedural knowledge stored in LTM can be organized in one of two forms:

• Knowledge network

or

• Schema.

Knowledge Network. The amount of information stored in LTM is so great that it must be organized efficiently to be functional. This functionality is created by numerous interconnections between different pieces of knowledge. The more similar two pieces of knowledge are to each other, the shorter and stronger the interconnection. Each time a piece of information must be recalled, different pieces of similar knowledge are activated, and the interconnections lead to the correct place in memory where the desired information is stored. In a network representation such as this, knowledge is described as objects with specific properties. The links between the objects generate a network and reduce the amount of information that needs to be retained in any one place. To clarify, in a hierarchical structure of objects, the lower-level objects share many of the properties of upper-level ones. Consequently, the full description of each piece of information does not need to be saved. Only the unique properties of each object need to be saved since the shared properties are already saved and can be applied when needed. Objects are classified by categories based on their shared qualities, and proximity between objects in the network is used to aid in storage and subsequent recall. The network is dynamic, as all new knowledge has to be integrated and may change the interconnections among objects.

Schema (pl. Schemata). Schemata are another way to organize knowledge in LTM. Schemata use sequences of blocks of knowledge in order to achieve goals. A schema is defined by its goal and is comprised of a list of expected actions — mental or physical — or events that may occur in a given situation. Schemata are developed based on specific past experiences or experiences that are similar to the given situation. Once a schema is activated, it allows you to immediately have all relevant information for achieving the task or sub-task. In aviation, you have probably developed schemata for a variety of tasks such as takeoffs and landings. You likely have a sequence of events that you follow each time you take off such as extending flaps, checking all control movements and verifying task sharing. You may have a generic schema that you apply to all take-offs, or you may have a very specific schema for a specific airport because it has some special characteristics. In either case, if the current sequence of events matches your schema, then the task is performed and the schema is unchanged. In the event that some aspect of the current situation does not match your schema, you may either apply another schema, alter your existing schema to fit this new situation or create a completely new schema. Altering an existing schema or creating a new one takes attentional resources as you must use other types of knowledge to solve the problem at hand. The altered or new schema will integrate the new information and will be available the next time you face the same or a similar situation. As you gain experience you will increase the number and complexity of your schemata. As such, your changing schemata become more specific, complete and useful.

6.5 Recall

After information has been stored, it is useful only if we can later retrieve it and apply it to the current situation. This retrieval process is termed recall, which is a complex process that requires activation of knowledge in LTM. This activation may be voluntary or involuntary. In either case, we use retrieval mechanisms that recognize the indicators created in the initial encoding process.

An activation mechanism defines what is useful and needed knowledge based on the current cognitive activities being conducted. The more deeply encoded, organized and structured the knowledge, the easier it is to recall. Recalling one piece of information that is linked to many other pieces can make it easier to recall all of the connected information as well. Activation of one node in the knowledge structure can propagate to nearby nodes through the various interconnections. Some information may be recalled more quickly than other information because it is closer to the active network and may have stronger links.

It is important to note the significance of context for recall. The success of retrieval depends on the nearness of the recall context to the context when the initial encoding occurred. The more similar the context, the greater retrieval. This is why it is often easier to recall information when you are in the same place you learned it compared to trying to recall the information in a completely different location or context.

6.6 Forgetting in LTM

Forgetting in LTM is different from forgetting in STM. Forgetting in LTM likely stems from either encoding or retrieval problems. These problems could be due to:

• Poor consolidation that may lead to weak links or the creation of no links at all if some portion of the information was incomplete or lost. Information that has poor links is virtually impossible to recall.
• Interference because of similarity between information. If two items are very similar, one could be mistakenly recalled when the other is needed.
• Incompatibility between encoding and retrieval contexts. The situation in which you are trying to recall information may be so dissimilar from the context in which the information was originally encoded that you are not stimulated to recall the item.

6.7 Improving LTM and Recall

Techniques for improving LTM primarily focus on the creation of useful links between pieces of knowledge in order to increase the likelihood that the information will be retrieved correctly. These techniques are called mnemonics. Most mnemonics rely on the creation of a meaningful association between the information to be remembered and some other easily recognized or recalled information. The easily recognized information serves as stimulus for the recall of the other information.

A variety of mnemonic methods exist. A few examples are:

• The loci method. This method involves relating information to existing objects within a geographical space. One piece of information is associated with each object located within the space thereby providing a context to recall the information. The information can then be retrieved by returning to or visualizing the site and imagining the information related to each object in the space. Essentially, you can “walk through” the geographical space in your mind or in real life. The objects will serve as reminders of the information you associated with them.
• The key-word method. This method involves the creation of an association between a new word to be remembered and an image or statement that relates to an already known word. The known word acts as the “key” that stimulates the image or statement, which leads to the recall of the new word that was just entered into memory.
• The acronym method. This method involves the creation of a meaningful word or phrase where each letter of the word or phrase is the first letter of an item in a list. For example, to remember all of the colors in the visible light spectrum, we often use the acronym ROY G BIV (Red Orange Yellow Green Blue Indigo Violet). This method is used quite extensively in aviation for checklists or extended phrases.

Practicing mnemonics can improve your memorization skills. Mnemonic techniques are effective because they force you to give attention and meaning to the information being memorized. The meaning provides structure and organization to the information.

7 Working Memory

The term “working memory” is used in cognitive psychology to refer to a short-term store that is needed for certain mental tasks that involve dynamic or complex information and processes. Not the same as STM, working memory is defined not in terms of the amount of time information can be held, but in terms of the goal of the task being performed. Working memory gathers information necessary for attaining the goals from a variety of places. As seen in Figure 4, working memory can draw information from sensory memory, STM or LTM.

Figure 4: Relationships Among Memory Stores.

7.1 Features of Working Memory

The concept of working memory is a fairly recent addition to models of human memory. Working memory represents a collection of the structures and processes used for temporarily storing and manipulating information. In many newer models of memory, working memory replaces or encompasses STM, and, in some models, LTM. These models provide a stronger emphasis on the notion of manipulation of information instead of the passive maintenance of information assigned to STM in early models. Working memory emphasizes an activation of information used in a task. In working memory, information is available quickly and directly. Working memory combines the processes for maintaining information, directing attention to relevant information, suppressing irrelevant information and inappropriate actions, and coordinating cognitive processes when more than one task must be performed at the same time

Since working memory is goal-oriented, the amount of time information is stored is related to the duration of the task. After task completion, the content of working memory is cancelled and no longer has meaning. As such, the content of working memory evolves with the task and is organized in relation to the task goals.

The capacity of working memory is finite. Working memory capacity does not vary according to age or intelligence. Rather, it depends on encoding speed and on the ability to retain information following chronologically sequenced events. If a situation is common and has been experienced many times in the past, then working memory capacity will be high. Conversely, if a situation is new or more attention must be used to identify information, working memory will be limited in its capacity during the particular situation or task. Repetition and good encoding are required to maintain information in working memory. Otherwise, it is easily lost.

Since working memory has finite capacity, it is important to store information about performing tasks and to have the information readily available. However, actually knowing which information is most important is difficult because of our complex world where many things can occur at once. Anticipation is an especially important cognitive strategy that we must develop to aid in the selection of the correct information to be processed in Working Memory. This is particularly true in aviation because of the dynamic work environment. However, it is impossible to guarantee that we will always choose the correct information to be processed in working memory. The challenge is to avoid:

• Selecting inappropriate information
• Storing useless information
• Forgetting or not properly encoding information that will be useful later.

7.2 Managing Working Memory

Managing working memory is a continuous process that occurs throughout the performance of a task. In aviation, working memory is activated as soon as flight preparation begins. During the flight, crew members must update working memory with new data related to the current situation. Failing to update working memory can be very dangerous. Even when you are engaged in routine procedures, it is vitally important that you monitor all situations around you to prevent any surprises that your working memory will not be prepared to address. Each crew member should provide working memory management support for the others.

Support may include activities such as:

• Searching for useful information
• Validating information
• Removing useless information
• Managing information flow
• Becoming a memory device for the other crewmember by remembering or recording information.

8 Key points

• The different memory stores are most easily defined by the amount of time that information is held in each.
• Sensory memory is virtually unlimited, but we can only attend to a limited amount of information with active processes. The information we attend to is then passed through to STM.
• STM is limited in its capacity (7 + 2 pieces of information), and the amount of time information can be stored. Interference or loss of mnemonic trace can lead to forgetting in STM.
• Rehearsal or the use of memory devices can aid in the ability to get information into LTM.
• Building strong links and associations is important to storage of information in LTM and the ability to recall the information later. Information can be stored as schemata or as a knowledge network.
• Mnemonics can aid in the storage and recall of information from LTM.
• Working memory is driven by task goals. Information is extracted from a variety of sources and held and processed until the task goals are met or ultimately abandoned.
• Crewmembers need to work together to help each other with memory and reaching all of the task goals.

9 Additional Reading Material and Websites References

• Baddeley, A.D. (1976). The Psychology of Memory. New York: Basic Books, Inc.
• Baddeley, A.D. (1986). Working memory. Oxford: Oxford University Press.
• BOLLES, B.E. (1988). Remembering and Forgetting: Inquiries into the Nature of Memory. New York: Walker and Co., Inc..
• DEUTSCH, D. (1975). Short-term Memory. New York: Academic press, Inc.
• GROOT de, A.D., GOBET, F. (1996). Perception and memory in chess. Studies in the heuritics of the professional eye. Assen: Van Gorcum.
• MILLER, G.A. (1956). The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information. Psychological Review, 63, 81-97.
• MIYAKE, A., SHAH, P. (1999). Models of working memory: Mechanisms of active maintenance and executive control . New York, NY: Cambridge University Press, 1999
• TULVING, E. DONALDSON, W. (1972). Organization of memory. New-York: Academic Press.
• WATKINS, M.J., TULVING, E. (1976). Context effect in recognition memory for faces. Journal of Verbal Learning and Verbal Behavior, 15, 505-517.

Related SKYbrary Articles

• Situational Awareness
• Pilot Workload
• Memory in ATC