RECOGNITION AND SELECTION

The perceptual system has to do much more than just organize and integrate sensory information, and depth perception. The system also has to recognize what an object is. This requires that it relates the perceptual information to what you know about the object. Your perceptual system provides you with much more information than you can process, and to do one specific task, you must ignore most of the objects in the environment, such as in a basketball game. You must ignore the fans, and other distractions in the gym. In the following section, we consider the processes that help you recognize and attend to objects in your environment and how these processes have been shaped by learning.

You are practicing pattern perception when you recognize and identify shapes, faces, melodies, words, pictures, objets, and so forth. Pattern perception illustrates the recognition of a set of stimuli that are arranged in such a way that we would say they constitute a form or pattern. Thus, we define a pattern as a collection of stimuli arranged in a reasonably specific way.

The neural equipment in the visual system that is required for pattern perception seems to exist in humans at birth. Scientists have demonstrated that newborn babies can reliably discriminate among different patterns and that they show clear pattern preferences.

Pattern perception is important for every sense. You recognize the words a friend has spoken, a familiar face, and the smell of freshly brewed coffee. In order to recognize a word or an object, the visual system has to compare the sensory input to information stored in long-term memory. This raises a number of basic issues. How is the comparison performed? What is stored in memory?

When you look at this page and see the letter B, how is it that you recognize this stimulus as the second letter of the alphabet? To do so, the visual system must somehow take the sensory input and compare it to information stored in long-term memory. How is this accomplished? A simple theory that you have stored little copies, or templates, to represent each letter of the alphabet. When a letter is detected, the recognition system then quickly compares the sensory information to each of the stored templates. When a match is detected, recognition occurs. However, some problems arise from this hypothesis. What happens when there are many different variations of the same letter? It would seem that you would need a template for every possible variation. To allow for variation, you need to assume a nearly infinite number of templates have been stored in memory. Because of this, psychologists have come to reject this theory.

Many psychologists now believe that instead of storing an exact copy of an object, you store the object's average characteristics. For example, birds tend to be small in size, to be feathered, to fly, and to sing. These and other characteristics could be stored in memory and used to recognize whether the sensory input corresponded to a bird. From this perspective, being a bird would not require having all of the essential characteristics. Thus, a perfect match would not be required. In this theory, there should be quite a bit of variability in how long it takes you to recognize something as a bird. For example, a bird such as a robin has many characteristics common to all birds, and can be thought of as a prototype. In contrast, an ostrich is a poorer example. The prototype-matching theory predicts that it should take longer to judge that an ostrich is a bird. As predicted, the closer the pattern is to the prototype, the faster and more accurately the stimulus is recognized.

A feature-analysis approach to pattern perception would propose that the resultant perception of the pattern depends on detecting and recognizing the relevant features. According to a popular version of this theory, pattern perception by feature analysis proceeds through four ordered stages: (a) detection, (b) pattern dissection, (c) feature comparison in memory, and (d) recognition.

The previous theories work for simple objects such as letters, but what about complex ones such as horses. Are thousands of feature detectors somehow linked together? According to Irving Biederman's theory of recognition by components, complex objects can be represented by a relatively small set of simple shapes he calls geons. He estimates that the average individual may be familiar with as many as 30,000 objects, and that recognizing those objects easily could be handled by no more than 36 geons. According to Biederman, object perception begins by the perceiver segmenting the object, usually at regions where edges join or where curves markedly change. These components are then compared against the geons to determine a best match.

Up to now, we have discussed pattern perception as a process that begins with detecting information in the environment, continues with interpreting that information, and results in recognizing the pattern. This kind of information processing is called bottom-up processing because it begins at the level of the receptors and works its way up to the higher brain centers. But researchers in pattern perception recognize that this process can go in reverse order, when the higher perceptual centers give directions to the lower centers about information to be extracted. This top-down processing is illustrated by the concept of perceptual set. By set, we mean that the perceiver has some kind of perceptual readiness or expectancy; that is, the observer is "set" to perceive something.

Attention is defined as a cluster of integrated events and processes that determine which stimuli receive further processing. In vision, focusing one's attention on a parcticular object or event is easy. However, this is not so for every sense. Consider the cocktail party effect. Imagine that you are talking in small groups, so that the room contains a half-dozen conversations, all of which may be about equally audible. Yet you have no trouble focusing attention on the conversation of your group. Moreover, you can easily listen in on a conversation of another group. In this case, what you are attending to is not determined by a simple peripheral mechanism. Instead, central mechanisms are being used to determine which sensory information receives further processing.

Selective attention refers to the different processing of coexisting perceptual events, meaning that the perceiver has the ability to focus on one stimulus while ignoring other stimuli that are present. Such a process has obvious advantages, because it allows the perceiver to maximize the information that is picked up from one particular source while reducing the potential sensory interference from another source.

In studying selective attention, psychologists have made use of a technique known as dichotic listening. In this technique, a subject wearing earphones simultaneously receives two different auditory messages, one in each ear. The subject is told to pay attention to the message in one ear (the attended message) and to ignore the message in the other ear (unattended message). To ensure that the subjects are attending, researchers typically ask them to repeat aloud the words they hear in the attended message. This technique is called shadowing, because the subjects' verbalizations are, in effect, a shadowed version of the attended message. At the end of the session, subjects are asked various questions about both the attended and the unattended messages. The results from these types of studies have shown that when the unattended message is switched to a foreign language, such as German, or when the words are spoken backward, the subjects are able to recall little, if any, of the unattended message. When the content of the messages is very similar, the subjects' attention wanders between the two so that neither message is accurately perceived.

Although dichotic listening studies support the existence of selective attention, they also provide evidence that attention can be divided. In one experiment, the unattended message informed the subjects that "you may stop now." Yet subjects ignored that message and continued to shadow the attended message. But when the unattended message included the person's name, the subjects were much more likely to hear that message and stop their shadowing. Obviously, some monitoring of the unattended message was taking place. The cocktail party effect provides support for the existence of divided attention. The fact that you heard your name in another conversation suggests that at some level you were listening to more than one conversation, but you would probably fail in any attempt to recall the content of the different auditory messages. Divided attention is generally difficult to achieve, but there are some situations that are conducive to its occurrence.

One way to improve the occurrence of divided attention is through practice. When there is consistency in a task, that is, when the stimuli occur in highly predictable ways, with practice the appropriate responses can be made automatically, through automatic processing, thus allowing us to attend to several tasks at once. This processing does not require conscious attention. Controlled processing refers to those situations in which the connections between stimuli and responses vary enough from time to time to require focused attention. For example, you can drive your car from school to home with little attention to driving or to navigating the familiar route (automatic processing), while at the same time conversing with a fellow student about school (controlled processing).

Many stimuli exists in our perceptual world. It is through the process of attention that we select certain stimuli from others in the perceptual array. We might search for a friend we have temporarily lost in a crowded department store or scan our cluttered desk for the pencil that we just lost. But sometimes stimuli seem to seek us out; there is something about a particular stimulus that draws our attention to it. Context is important, meaning that stimuli that are different or unusual are likely to get our attention. Color is another important variable, as are movement and size. Advertisers clearly recognize what stimulus qualities attract attention and capitalize on that information in selling their products. Repetition is another way to maximize attention. When the instructor says the same thing several times in a lecture, the students are more likely to think it is important and will attend more closely to it. Intensity is also an important variable. The loudest speaker in a group is likely to attract your attention. In short, perception directs attention and attention directs perception. This focusing of perception to a limited portion of our sensory world means that our perceptual capability is enhanced, but not maximized.

The term perceptual learning refers to an increase in a person's ability to extract information from the environment. In essence, it means that the person learns to be a better perceiver. Perceptual abilities differ among individuals for a variety of reasons, but most often because of differences in the functioning of their sensory systems. Some people have greatly enhanced sensory skills in such areas as visual acuity, color vision, hearing range, tactile sensitivity, and so forth, because of the physiological nature of their sensory systems. But perceptual learning does not refer to differences produced by physiological advantage. Instead, perceptual learning is a label reserved for those cases where perceptual ability has been increased due to learning.

Eleanor Gibson and her colleagues identified a number of changes that occur when perceptual abilities are significantly improved. Obviously, much of perceptual learning is simply learning to extract information from the environment that was not being extracted before. However, perceptual learning is not just the extraction of larger quantities of information - you can probably see why that strategy would have diminishing returns after awhile - but also has to do with how that information is extracted.

Gibson says that there is an increase in the economy of information extraction: the time required to make decisions is reduced, and one learns to make finer discriminations. Further, there is an equal decrease in stimulus generalizations. Attention is maximized, so that irrelevant stimuli are ignored and one focuses on the most relevant stimuli. If the task is a visual one, new eye movement strategies may be learned. One learns distinctive features that are usually overlooked by perceivers who see them only as part of a whole. But the perceptual learner picks out those distinctive features and uses them as a basis for more accurate perceptual judgments.