Now that we have seen how memory doesn't work—namely, like an infallible recording of everything that happens to you—we turn to studies of how it does work.

Conscious, intentional recollection of an event or an item of information is called explicit memory. It is usually measured using one of two methods. The first method tests for recall, the ability to retrieve and reproduce information encountered earlier. Essay and fill-in-the-blank exams require recall. The second method tests for recognition, the ability to identify information you have previously observed, read, or heard about. The information is given to you, and all you have to do is say whether it is old or new, or perhaps correct or incorrect, or pick it out of a set of alternatives. The task, in other words, is to compare the information you are given with the information stored in your memory. True–false and multiple-choice tests call for recognition.

Recognition tests can be tricky, especially when false items closely resemble correct ones. Under most circumstances, however, recognition is easier than recall. Recognition for visual images is particularly impressive. If you show people 2,500 slides of faces and places, and later you ask them to identify which ones they saw out of a larger set, they will be able to identify more than 90 percent of the original slides accurately (Haber, 1970).

The superiority of recognition over recall was demonstrated in a study of people's memories of their high school classmates (Bahrick, Bahrick, & Wittlinger, 1975). The participants, ages 17 to 74, first wrote down the names of as many classmates as they could remember. Recall was poor; even when prompted with yearbook pictures, the youngest people failed to name almost a third of their classmates, and the oldest failed to name most of them. Recognition, however, was far better. When asked to look at a series of cards, each of which contained a set of five photographs, and to say which picture in each set showed a former classmate, recent graduates were right 90 percent of the time—and so were people who had graduated 35 years earlier. The ability to recognize names was nearly as impressive.

Sometimes, information encountered in the past affects our thoughts and actions even though we do not consciously or intentionally remember it, a phenomenon known as implicit memory (Schacter, Chiu, & Ochsner, 1993). To get at this subtle sort of memory, researchers must rely on indirect methods instead of the direct ones used to measure explicit memory. One common method, priming, asks you to read or listen to some information and then tests you later to see whether the information affects your performance on another type of task.

Suppose that you read a list of words, some of which began with the letters def (such as define, defend, or deform). Later, if you were asked to complete word fragments (such as def-) with the first word that came to mind, you would be more likely to complete the fragments so they turned into words from the list than if you had never seen the list—even if you could not remember the original words very well (Richardson-Klavehn & Bjork, 1988; Roediger, 1990). That is, the words on the list have “primed” (made more available) your responses on the word-completion task. Priming isn't limited to words; priming people with unusual sentence constructions causes them to adopt those constructions a week later (Kaschak et al., 2011). Fragments of pictures can also act as primes. In one study, people briefly saw fragments of drawings depicting objects and animals. Then, 17 years later, they were mailed the same fragments and also fragments of new drawings, with a request to name what the fragments depicted. Even when people couldn't remember having been in the original experiment, they identified the primed objects much better than the new objects (Mitchell, 2006). These studies show that people know more than they think they know—and that they can know it for a very long time.

Another way to measure implicit memory, the relearning method, or savings method, was devised by Hermann Ebbinghaus (1885/1913) in the 19th century. The relearning method requires you to relearn information or a task that you learned earlier. If you master it more quickly the second time around, you must be remembering something from the first experience.

Although people usually refer to memory as a single faculty, as in “I must be losing my memory” or “He has a memory like an elephant's,” the term memory actually covers a complex collection of abilities and processes. If a video camera is not an accurate metaphor for capturing these diverse components of memory, what metaphor would be better?

Many cognitive psychologists liken the mind to an information processor, along the lines of a computer, though more complex. They have constructed information-processing models of cognitive processes, liberally borrowing computer-programming terms such as input, output, accessing, and information retrieval. When you type something on your computer's keyboard, a software program encodes the information into an electronic language, stores it on a hard drive, and retrieves it when you need to use it. Similarly, in information-processing models of memory, we encode information (convert it to a form that the brain can process and use), store the information (retain it over time), and retrieve the information (recover it for use). In storage, the information may be represented as concepts, propositions, images, or cognitive schemas, mental networks of knowledge, beliefs, and expectations concerning particular topics or aspects of the world.

In most information-processing models, storage takes place in three interacting memory systems. A sensory register retains incoming sensory information for a second or two, until it can be processed further. Short-term memory (STM) holds a limited amount of information for a brief period of time, perhaps up to 30 seconds or so, unless a conscious effort is made to keep it there longer. Long-term memory (LTM) accounts for longer storage, from a few minutes to decades (Atkinson & Shiffrin, 1968, 1971). Information can pass from the sensory register to short-term memory and in either direction between short-term and long-term memory, as illustrated in Figure10.3.

Figure10.3

Three Memory Systems

This model, which is known informally as the three-box model, has dominated research on memory since the late 1960s. The problem is that the human brain does not operate like your average computer. Most computers process instructions and data sequentially, one item after another, and so the three-box model has emphasized sequential operations. In contrast, the brain performs many operations simultaneously, in parallel. It recognizes patterns all at once rather than as a sequence of information bits, and it perceives new information, produces speech, and searches memory all at the same time. It can do these things because millions of neurons are active at once, and each neuron communicates with thousands of others, which in turn communicate with millions more.

Because of these differences between human beings and machines, some cognitive scientists prefer a parallel distributed processing (PDP) model or connectionist model. Instead of representing information as flowing from one system to another, a PDP model represents the contents of memory as connections among a huge number of interacting processing units, distributed in a vast network and all operating in parallel—just like the neurons of the brain (McClelland, 1994, 2011; Rogers & McClelland, 2014; Rumelhart, McClelland, & the PDP Research Group, 1986). As information enters the system, the ability of these units to excite or inhibit each other is constantly adjusted to reflect new knowledge.

In this chapter, we emphasize the three-box model, but keep in mind that the computer metaphor that inspired it could one day be as outdated as the metaphor of memory as a camera.