Working Memory Capacity

Classic Edition

Nelson Cowan

• Publisher: Routledge
• Year: 2016
• ISBN:
  9781317232377
• Edition: 1
• Topic: Psicologia
• Subtopic: Storia e teoria della psicologia

The idea of one's memory "filling up" is a humorous misconception of how memory in general is thought to work; it actually has no capacity limit. However, the idea of a "full brain" makes more sense with reference to working memory, which is the limited amount of information a person can hold temporarily in an especially accessible form for use in the completion of almost any challenging cognitive task.

This groundbreaking book explains the evidence supporting Cowan's theoretical proposal about working memory capacity, and compares it to competing perspectives. Cognitive psychologists profoundly disagree on how working memory is limited: whether by the number of units that can be retained (and, if so, what kind of units and how many), the types of interfering material, the time that has elapsed, some combination of these mechanisms, or none of them. The book assesses these hypotheses and examines explanations of why capacity limits occur, including vivid biological, cognitive, and evolutionary accounts. The book concludes with a discussion of the practical importance of capacity limits in daily life.

This 10th anniversary Classic Edition will continue to be accessible to a wide range of readers and serve as an invaluable reference for all memory researchers.

1 THE PROBLEM OF CAPACITY LIMITS

DOI: 10.4324/9781315625560-1

Working Memory Capacity and the Full Brain

In a cartoon in The Far Side by Gary Larson, a bewildered student raises his hand to ask, “Mr. Osborne, may I be excused? My brain is full.” Now, analyzing a joke runs the risk of ruining it. Yet, in this case, it seems instructive that there is something funny about the idea of a human brain being full. It seems to be common knowledge that every experience leaves an indelible mark on our memories. Unlike a computer, the normal human brain never reaches a point at which new experiences can no longer be committed to memory; the brain cannot be full. At the same time, one can be overwhelmed by new information—to the point that it seems to be too much to comprehend, too confusing or complex to file away in memory. It is that feeling, one might presume, that made Larson’s character want to leave the classroom.

The feeling of being overwhelmed by a lot of new information can occur because of the special type of memory that is typically termed working memory. It refers to the relatively small amount of information that one can hold in mind, attend to, or, technically speaking, maintain in a rapidly accessible state, at one time. The term working is meant to indicate that mental work requires the use of such information. For example, if I asked you to retrieve a book from a locked room in the psychology building and I told you that the room is on the third floor, that it is the second office to the right of the elevator but that I don’t recall the room number, that the key is in my mailbox in the mail room at the opposite end of the hall on that floor, and that the book is by an author named Smith and was published in 1985, you would have to consider all of that information to retrieve the book. Your best option might be to form an imaginary map of the situation with some verbal details filled in (or even a written map, if the writing materials were available) and to refer to that map. However, you might be unable to do so and might have difficulty keeping all of the relevant information in mind, leading to that uncomfortable feeling of, as it were, a full brain. If some of the information is forgotten (that is, if working memory proves inadequate), it still may be possible to muddle through the situation, but often with mishaps or mistakes along the way. The purpose of this book is to address the question of how to measure the amount of information that can be held in mind successfully at one time, or working memory capacity.

This concept of working memory and its limits is a key part of the human condition. The vast wealth of wisdom and experience that one has at one’s disposal is like a well-equipped tool shop, but, when it is time to solve a problem, we struggle to find all of the best tools for the job and to carry them to where they are needed. We need working memory in language comprehension, to retain earlier parts of a spoken message until they can be integrated with the later parts; in arithmetic, to retain partial results until the rest of the answer can be calculated; in reasoning, to retain the premises while working with them; and in most other types of cognitive tasks. Moreover, we need working memory not only to hold new information that has been given to us, but also to integrate it with old information. For example, take the aforementioned case of retrieving a book from an office. If you already are familiar with the layout of the building and can retrieve that information from memory, the new information can be associated with your mental image of the third floor to make it easier to remember, in an integrated, maplike form. It may take effort to achieve that integration, but the information then becomes easier to use and later to retrieve from memory.

Because working memory is limited, there sometimes is important mental work that fails to get done; multiple ideas or facts fail to become integrated as they should. Consequently, people occasionally retain inconsistent memories. For example, you may have planned for a week or so to pick up a certain rare type of vegetable for a dish you are planning to cook for some guests. You may then learn that the guests cannot come and that the dinner is going to be postponed for a month. Still, learning that is not necessarily sufficient to make you revise your shopping plans. It may be only when the shopping plans and the postponement of the dinner are present in working memory at the same time that you suddenly realize that there is no longer a good purpose to buy the rare vegetable.

Psychologists have been studying and thinking about concepts similar to working memory for a long time—at least as long as the modern field of experimental psychology has existed, for slightly more than 100 years. Yet, there still is little agreement about just how much information can be held in working memory at one time. In experiential terms, what is the limit on how much can be experienced at once or on how much we can be conscious of at once? These are the types of fundamental questions that will be considered in this book. They help to define what it means to be a human being and they get at processes that profoundly affect the capabilities and limitations of our thought processes.

Broad and Narrow Definitions of Capacity

We have asked, fundamentally, what is the human limit in working memory capacity. Now it is important to refine the question. According to a very broad definition, working memory capacity is simply the ability to remember things in an immediate-memory task (a task with no delay between the end of the presentation of items to be recalled and the period of recall itself). For example, if given a list of words to recall, if one recalls six of the words, one’s capacity would be six words; if one recalls three words, one’s capacity would be three words; and so on. It has to be said that that is not a very satisfactory definition of capacity because there is no reason to believe that it could lead to any information beyond what we already know from the data. When we ask about capacity in a narrower sense, we are asking not about the ability of the entire processing system in the task, but rather about the ability of a specific component of the system. The narrower component with which this book is most concerned is the amount that an individual can hold in mind at one time, which I think to be synonymous with the individual’s focus of attention. Given that this component has a lot to do with the experience of being human, its potential value is easy to understand. The trouble is, when it comes to the human mind, the components are not visible and easily separable from one another, making this a very difficult type of issue to address. If it were easy, this book would not be necessary.

Occasionally (starting now), I will use analogies to communicate ideas that may offer some insights. Suppose someone asks about the capacity of a ferry boat. Imagine that the boat will take people only in vehicles and that the boat is limited only in how many vehicles it can take on board. The practical capacity of the boat might be defined as how many people it will hold. However, the answer depends on how many people are packed into each vehicle. If the boat remains at the dock until it is full (after which no more vehicles are admitted) and then departs on its journey, the number of people transported each time will vary within a certain range. However, a person will understand that variation better if he or she knows about the boat’s fundamental limit in terms of vehicles. So, the limit in terms of vehicles is a narrow sense of capacity and understanding that limit lends insight to our understanding, even if we are ultimately more interested in the practical question of how many people can be transported. Of course, for practical purposes, someone also might want to know what other means of transportation are available for crossing the body of water or, in terms of working memory, what mental processes are available for holding information for a task at hand.

To appreciate the benefit of analyzing performance into fundamental components, consider a seminal study related to working memory that was conducted by Sperling (1960). In that study, arrays of letters and other characters were briefly flashed on a computer screen. The task was to record the character array or some designated portion of it. When held responsible for the entire array, subjects were able to recall only about four items, no matter how many items the array held. That result is illustrated in Figure 1.1.

FIGURE 1.1 An illustration of the method and results of Sperling (1960)

When subjects were held responsible for the entire array, they could recall only about 4 of the characters on average, no matter how many the array contained (between 4 and 12). However, in some conditions, a tone was presented very soon after the array ended, indicating which of three rows was to be recalled. This procedure served the purpose of sampling from the information available to the subject, rather than requiring a full reporting of it. With such a tone cue, subjects were able to recall up to three or four items from the designated row. At this point, it was necessary to postulate two different capacities. The information from the entire array, or most of it at least, was said to be held in a vivid, but transient, sensory memory. This accounted for the fact that subjects’ memory for a row of characters designated by a partial-report tone cue was excellent. Apparently, the sensory memory of the entire stimulus array managed to persist from the time of the array to the time of a closely following tone cue (although the more delayed the tone cue was, up to a quarter second or so, the more diminished the sensory memory became). However, there was some additional limit on working memory that made it impossible for subjects to report more than about four items on average from whatever set was relevant. Later, we will see that this capacity limit appears to be typical of a very fundamental component of the processing system_ it turns out to be a capacity limit narrowly defined, which may result from a limit in the capacity of the focus of attention (Cowan, 2001).

In Search of Capacity Measures

A historical moment in the history of experimental psychology occurred when Jevons (1871) rolled some black beans onto a table and tried to estimate the number without actually counting the beans. His motivation included this quotation from Sir William Hamilton: “Supposing that the mind is not limited to the simultaneous consideration of a single object, a question arises, How many objects can it embrace at once?…. I find this problem stated and differently answered by different philosophers, and apparently without a knowledge of each other” (p. 281). Jevons went on to state that:

This subject seemed to me worthy of more systematic investigation, and it is one of the very few points in psychology which can, as far as we yet see, be submitted to experiment. I have not found it possible to decide conclusively in the manner Hamilton suggests, whether 4 or 5 or 6 is the limit, nor do imaginative acts of experiment seem likely to advance exact knowledge. Probably the limit is not really a definite one, and it is almost sure to vary somewhat in different individuals. (p. 281)

Jevons noted that “The whole value of the experiment turns upon the rapidity of the estimation, for if we can really count five or six by a single mental act, we ought to be able to do it unerringly at the first momentary glance” (p. 281). In more than 1,000 trials, three and four items were unerringly enumerated, whereas increasingly larger proportions of error occurred for higher numbers. (The proportion of trials correct was.95 for five beans,.82 for six beans,.72 for seven beans,.56 for eight beans, and continued to decline for higher numbers of beans.) In later chapters, we will revisit this basic technique with more recent research on what has come to be known as the process of subitizing, or enumerating objects without actually counting. For now, this first study serves as an interesting historical backdrop for a broader debate about psychological theories.

In particular, there has long been a heated debate about whether it is reasonable to expect to find constants in psychology, such as a constant working memory capacity limit in the enumeration task. This debate was characterized very well by Carr (1933), with respect to enumeration and then to constants of learning, in the abstract to his article:

There are two opposed attitudes toward experimental problems in psychology, one the “constancy” attitude, the other the “relativity” attitude. The former assumes the presence of constants which are the true values even though disturbed or hidden from view because of the variable distractive conditions which cannot be controlled. The latter attitude directs attention to the variations, and attempts to determine how and why the obtained value varies with the conditions of which it is a function.

The present work takes one sort of constancy attitude toward capacity. Now, what would be a motivation for doing so? Clearly, if a constant can be found, it greatly simplifies the theory of human performance. Finding the simplest theory that is adequate to explain the data is a basic goal of science. Moreover, it appears to be the case that, even if a constant exists, it typically will be obscured by the “variable distractive conditions” that Carr mentioned.

Cowan (2001) used the example of the gravitational constant. Although it seems obvious now that we know about it, discovery of the gravitational constant actually was a magnificent achievement. Scientists could have argued that no such regularity existed. Feathers fall to the ground much slower than pebbles, and both birds and arrows can stay in the air for a very long time. The gravitational constant was not discovered until Tycho Brahe collected extensive data on planetary movement, which counts as a type of motion in the absence of wind resistance. Johannes Kepler used such evidence to establish regularities or laws of planetary motion, without understanding the principles behind the regularities. Only then could Isaac Newton discover the gravitational constant. To be sure, such a constant still had to be tied in to other forces in the universe, and the work of Albert Einstein went a long way toward achieving that integration. Nevertheless, progress could not be made until the variable distractive conditions (wind resistance, in this case) could be identified and controlled.

What might be the variable distractive conditions for working memory capacity, and what might be the nature of the constant? As the ferry boat analogy suggested, there may be multiple ways in which a...