- Slides: 48
• Attention is the process of concentrating on specific features of the environment, or on certain thoughts or activities. This focusing on specific features of the environment usually leads to the exclusion of other features of the environment. • These examples (sam and tree and moth) illustrate how attention involves engagement of the mind and how this engagement affects our experience. • This connection between attention and what is happening in the mind was described over 100 years ago by William James (1890) in his textbook Principles of Psychology:
• Millions of items. . . are present to my senses which never properly enter my experience. Why? Because they have no interest for me. My experience is what I agree to attend to. . Everyone knows what attention is. It is the taking possession by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or trains of thought. . It implies withdrawal from some things in order to deal effectively with others.
• Thus, according to James, we focus on some things to the exclusion of others. • As you walk down the street, the things that you pay attention to—a classmate you recognize, the “Don’t Walk” sign at a busy intersection, and the fact that just about everyone, except you, seems to be carrying an umbrella—stand out more than many other things in the environment.
• Attention is central to many aspects of cognition. • Attention has an effect on perception (paying attention to something increases the chances you will perceive it), • on memory (you are more likely to remember something later if you were paying attention to it when it first occurred), • on language (reading a sentence involves paying momentary attention to the words in the sentence, one after the other), • and on solving problems (your success in solving the problem may depend on what aspect of the problem captures your attention).
Selective Attention: When Does Selection Occur? • Much of the early research on attention used auditory stimuli and focused on the process of selective attention—the ability to focus on one message and ignore all others. • This research began with experiments indicating that if we are paying attention to one message, it is difficult or impossible to take in the information in another message presented at the same time. • You can demonstrate this to yourself as follows
Hearing Two Messages at Once (Enlist the help of another person ) • Select two books on different topics that you have not read before. • Your task is to read one of these selections to yourself while the person reads the other selection out loud. • Do this for about a minute, and note how well you are able to remember both passages. • Were you able to understand both passages? • Experiments in which people are asked to pay attention to one of two simultaneously presented messages show that it is possible to focus on one message, but that not much information is obtained from the other message. • One of the first of these experiments was done by Colin Cherry (1953), who used a procedure called dichotic listening.
Dichotic Listening • For dichotic listening, one message is presented to the left ear and another message is presented to the right ear. • Participants are instructed to pay attention to one message (the attended message) and to ignore the other one (the unattended message) and to repeat the attended message out loud, as they are hearing it. • This procedure of repeating a message out loud is called shadowing. • The shadowing procedure is used to ensure that the participants are focusing their attention on the attended message.
• Cherry’s participants shadowed the attended message while receiving the other message in the unattended ear. • However, when they were asked what they heard in the unattended ear, participants could say only that they could tell there was a message and could identify it as a male or female voice. • They could not report the content of the message. Other dichotic listening experiments have confi rmed this lack of awareness of most of the information being presented to the unattended ear. For example, • Neville Moray (1959) showed that participants were unaware of a word that was repeated 35 times in the unattended ear.
• Cherry’s experiment is often described as a demonstration of the cocktail party phenomenon—the ability to pay attention to one message and ignore all other messages—because it resembles what people routinely achieve in a noisy party when they focus on one message and ignore all the others. • Cherry showed that a listener can select one message, but how is this selection achieved? • One answer to this question, the filter model of attention, was proposed by Donald Broadbent (1958)
Early Selection: Broadbent’s Filter Model • Donald Broadbent’s (1958) filter model is a classic in psychology because • it was one of the first to describe the human as an information processor, • and it was the first to depict the course of this information processing with a flow diagram. • Broadbent’s model, which was designed to explain selective attention, states that information passes through the following stage
• Sensory memory, which holds all of the incoming information for a fraction of a second and then transfers all of it to the next stage. • The filter identifies the attended message based on its physical characteristics— things like the speaker’s tone of voice, pitch, speed of talking, and accent—and lets only this message pass through to the detector in the next stage. All other messages are filtered out. • The detector processes information to determine higher-level characteristics of the message such as its meaning. Because only the important, attended information has been let through the filter, the detector processes all of the information that enters It. • Short-term memory receives the output of the detector. Short-term memory holds information for 10– 15 seconds and also transfers information into longterm memory, which can hold information indefinitely.
• Broadbent’s model is called an early-selection model, because the filtering step occurs before the incoming information is analyzed to determine its meaning. • One way to think about the filter is to consider a sieve at use at the beach, trapping the coarse grains of sand letting through the small grains (Figure 4. 4 a). • The filter in Broadbent’s model filters messages in a similar way, but instead of filtering based on the size of particles, it filters based on physical characteristics of the message, such as the speaker’s pitch or rate of speaking (Figure 4. 4 b).
Criticism • Broadbent’s theory stimulated a great deal of research on attention, but the results of some of this research posed problems for his theory. • For example, Neville Moray (1959) used the dichotic listening procedure and had his participants shadow the message from one ear. But when Moray presented the listener’s name to the other, unattended ear, about a third of the participants detected it.
• Moray’s participants had recognized their names even though, according to Broadbent’s theory, the name should have been filtered out before reaching the detector. • (Remember that the filter is supposed to let through only one message, based on its physical characteristics. ) • Clearly, the person’s name had not been filtered out and, most important, it had been analyzed enough to determine its meaning. • You may have had an experience similar to Moray’s laboratory demonstration if, as you were talking to someone in a noisy room, you have suddenly heard someone else saying your name (Following Moray’s lead, other experimenters showed that information presented to the unattended ear is processed enough to provide the listener with some awareness of its meaning)
Dear Aunt Jane (experiment) • For example, J. A. Gray and A. I. Wedderburn (1960) did the following experiment, which is sometimes called the “Dear Aunt Jane” experiment, as an undergraduate research project at the University of Oxford. • As in Cherry’s dichotic listening experiment, the participants were told to shadow the message presented to one ear. • As you can see from Figure 4. 6, the attended (shadowed) ear received the message “Dear 7 Jane, ” and the unattended ear received the message “ 9 Aunt 6. ” However, rather than reporting the “Dear 7 Jane” message that was presented to the attended ear, participants reported hearing “Dear Aunt Jane.
• Switching to the unattended channel to say “Aunt” means that the participant’s attention had jumped from one ear to the other and then back again. • This occurred because they were taking the meaning of the words into account. • Although Broadbent had shown it is difficult to switch between channels, the meaning of the words presented in the “Dear Aunt Jane” experiment caused participants to switch channels anyway. • Because results such as these could not be explained by Broadbent’s filter theory, Anne Treisman (1964 a) proposed another theory, which she called the attenuation theory of attention.
Intermediate Selection: Treisman’s Attenuation Theory • Treisman proposed that selection occurs in two stages, and she replaced Broadbent’s filter with an attenuator (Figure 4. 7). • The attenuator analyzes the incoming message in terms of • (1) its physical characteristics—whether it is high-pitched or lowpitched, fast or slow, • (2) its language—how the message groups into syllables or words, and • (3) its meaning—how sequences of words create meaningful phrases.
• Note that this is similar to what Broadbent proposed, but in Treisman’s model, language and meaning can also be used to separate the messages. • Treisman proposed, however, that the analysis of the message proceeds only as far as is necessary to identify the attended message. • For example, if there are two messages, one in a male voice and one in a female voice, then analysis at the physical level is adequate to separate the low pitched male voice from the higher-pitched female voice. • If, however, the voices are similar, then it might be necessary to use the message’s meaning to separate the two messages.
• Once the attended and unattended messages are identified, both messages are let though the attenuator, but the attended message emerges at full strength, and the unattended messages are attenuated— they are still present, but are weaker than the attended message. • Because at least some of the unattended message gets through the attenuator, Treisman’s model has been called a “leaky filter” model. • The final output of the system is determined in the second stage when the message is analyzed by the dictionary unit. • The dictionary unit contains stored words, each of which have thresholds for being activated (Figure 4. 8). • A threshold is the smallest signal strength that can barely be detected. Thus, a word with a low threshold might be detected even when it is presented softly or is obscured by other words.
• According to Treisman, words that are common or especially important, such as the listener’s name, have low thresholds, so even a weak signal in the unattended channel can activate that word, and we hear our name from across the room. • Words or words that are unimportant to the listener have higher thresholds, so it takes the strong signal of the attended message to activate these words. • Thus, according to Treisman, the attended message gets through, plus some parts of the weaker unattended message. Treisman’s model, like Broadbent’s, is often called an early-selection model because the attended message can be separated from the unattended message early in the information-processing system. • However, because further selection can also occur later, we have called Treisman’s model an intermediate-selection model.
Late-Selection Models • Some researchers proposed late-selection models, which state that selection of stimuli for final processing doesn’t occur until after the information has been analyzed for its meaning. • An experiment by Donald Mac. Kay (1973) supports this idea of late selection. • He had participants listen to ambiguous sentences, such as “They were throwing stones at the bank, ” that could be taken more than one way. (In this example, “bank” can refer to a riverbank or to a financial institution. ) • These ambiguous sentences were presented to the attended ear, while biasing words were presented to the other, unattended ear. • For example, as the participants were shadowing “They were throwing stones at the bank, ” either the word “river” or “money” was being presented to the unattended ear. • After hearing a number of the ambiguous sentences, participants were presented with pairs of sentences such as the following: They threw stones toward the side of the river yesterday. They threw stones at the savings and loan association yesterday.
• When they indicated which of these two sentences was closest in meaning to one of the sentences they had heard previously, Mac. Kay found that the meaning of the biasing word had affected the participants’ choice. • For example, if the biasing word was “money, ” participants were more likely to pick the second sentence. • This occurred even though participants reported that they were unaware of the biasing words that were presented to the unattended ear. • Because the meaning of the unattended word (“money”) was affecting the participant’s judgment, this word must have been processed to the level of meaning.
• Figure 4. 9 symbolizes the differences between the early- and late-selection approaches to selective attention in terms of what characteristics of messages are processed. • According to the early-selection view, only the physical characteristics of the message are processed before selection occurs. According to the lateselection view, both the physical characteristics and the meaning are processed before selection occurs. • Because there is evidence to support both views, how can we choose between them? One idea that has been proposed is that the information that gets processed during a selective-attention task is determined by the nature of the task Lavie, 1995). • Nilli Lavie (1995) proposed that the crucial variable is task load—how much of a person’s cognitive resources are used to accomplish a task
How Does Task Load Affect Selective Attention • Lavie describes a high-load task as one that uses most or all of a person’s resources and so leaves no capacity to handle other tasks. • A low-load task uses few resources, leaving some to handle other tasks. • These two situations are illustrated by the following examples: • High load: Samantha is practicing for her piano recital. She is working on a particularly difficult part, repeating it over and over, and is so intent that nothing can distract her. • Low load: Ragu is quickly scanning a celebrity gossip magazine that he has little interest in. He is easily distracted, especially when someone he knows enters the room.
• Samantha’s task is high load because she had to focus all of her resources on mastering the piano passage. • No resources remained to deal with other stimuli, so nothing distracted her. • In contrast, Ragu’s task was low load and left him with enough resources to notice and perhaps even interact with people entering the room. • The effect of task load on the processing of other stimuli has been studied in the laboratory using a technique called the flankercompatibility task.
Method Flanker-Compatibility Task • In the ﬂanker-compatibility task, a participant is presented with a central display, that may contain a target, and a “ﬂanker” distractor stimulus off to the side (Figure 4. 10 a). • The participant’s task is to detect the target in the central display as rapidly as possible, while ignoring the distractor stimulus. • For example, if participants are told that the target is the square, they would push a key as quickly as possible as soon as they detect the square.
• The question posed by this task is “Can the participant so totally focus their attention on detecting the target in the central display that the identity of the distractor will not affect their performance? ” You might think it would be easy to do this. • After all, the distractor is off to the side, so it is just a matter of focusing attention on the central display and ignoring the distractor. Nonetheless, the distractor can affect detection of the target. • This is demonstrated by using two types of distractors: (1) A compatible distractor, which is the same as the target (Figure 4. 10 a) • and (2) an incompatible distractor, which is different from the target (Figure 4. 10 b
• The task in Figure 4. 10 is called the low-load condition, because there is only one potential target. • The participants’ task is therefore easy—they just have to determine whether or not the shape that is present is the target. • When Shawn Green and Daphne Bavelier (2003) measured reaction times for detecting targets in displays like the ones in Figure 4. 10, they found that the reaction time was longer for the incompatible distractor (Figure 4. 10 c). • Thus, even though the participant was instructed to focus attention on the circular display that contained the target, some information from the distractor was being processed.
• What happens when task load is increased? Figure 4. 11 shows the high-load condition, in which the target is still the square, but there are five additional shapes. • The results for this condition are shown in Figure 4. 11 c. In this condition, the longer incompatible reaction time does not occur. (The difference between the compatible and incompatible reaction times is not statistically signifi cant. ) • When the load is high, the type of distractor does not affect reaction time. This occurs because the participant needs to use all of his or her resources to deal with the more complex display, so there are no resources left to process the distractor.
• The results in Figures 4. 10 and 4. 11 demonstrate differences between performance on low-load and high-load tasks, but you might be wondering whether a task that is high load for one person might be low load for another person. • For example, although mastering the piano passage is a high-load task for Samantha, it may be a relatively easy, low-load task for a master concert pianist. In fact, Green and Bavelier showed that the results in Figure 4. 11—in which increasing load eliminates the longer reaction time for the incompatible distractor condition—do not occur for people who, through many hours of practice, have become experts at playing video games.
Divided Attention: Paying Attention to More Than One Thing • Our emphasis so far has been on attention as a mechanism for focusing on one task. We have seen that sometimes we take in information from an “unattended” task, even when we are trying to focus on one task, as in the low-load condition in the flanker compatibility experiments. • But what if you want to consciously distribute your attention to a few tasks? Can we focus on more than one thing at a time? Although you might be tempted to answer “no” based on the fact that it is difficult to listen to two conversations at once, there are many situations in which divided attention—the distribution of attention to two or more tasks—can occur. • For example, people can simultaneously drive, have conversations, listen to music, and think about what they’re going to be doing later that day. The ability to divide attention depends on a number of factors, including practice and the difficulty of the task.
Practice Can Lead to Automatic Processing • Recently, as I was standing on the curb waiting for the “walk” signal, I observed a woman driving a car marked “AAA Driving School. ” I was impressed both by how slowly she was driving and by the intense look of concentration on her face. • She was paying very close attention to her task, and it clearly was not easy for her. Many people have had this experience when learning to drive, but later, with practice, they find that driving becomes much easier.
• Research on divided attention has shown that with practice, people can learn to simultaneously do two things that at first may have seemed fairly difficult. For example, when two college students first arrived at Elizabeth Spelke’s laboratory to be participants in an attention experiment, they could easily read short stories and take dictation (writing words that were spoken to them), but couldn’t do both at the same time (Spelke et al. , 1976). • However, after 85 hours of practice spread over 17 weeks, they were able to read a story rapidly and with good comprehension while simultaneously categorizing dictated words (for example, classifying “chair” as “furniture).
• Divided attention was demonstrated in another way by Walter Schneider and Richard Shiffrin (1977) using the procedure illustrated in Figure 4. 13, in which participants had to divide their attention between remembering target characters (letters or numbers) and monitoring a series of rapidly presented stimuli. • At the beginning of a trial, participants saw a memory set that consisted of from 1 to 4 target characters (Figure 4. 13 a). • They then saw a quick succession of 20 test frames (Figure 4. 13 b). • Each frame had four positions, and at each position, there was either a random dot pattern, a target (one of the characters from the memory set) or a distractor (a character from the distractor set). • The distractors were always from a different category than the characters in the memory set, so if the targets were numbers, the distractors were always letters. • Schneider and Shiffrin called this way of presenting stimuli the consistent mapping condition because the participants always knew that the target would be numbers and the distractors would be letters.
• As the 20 frames were being presented for a particular trial, the participants’ task was to detect and identify a target, if one was presented in one of the frames (Figure 4. 13 c). • In half of the trials, a single target was presented in one of the 20 frames. • In the other half, no target was presented. (We can see from the example that the target 3 did appear in one of the frames on this particular trial. ) • At the beginning of the experiment, the participants’ performance was only 55 percent correct, and it took 900 trials for performance to reach 90 percent (Figure 4. 14).
• Participants reported that for the first 600 trials, they had to keep repeating the target items in the memory set in order to remember them. • (Although targets were always numbers and distractors letters, the actual targets and distractors changed from trial to trial). • However, participants reported that after about 600 trials, the task had become automatic: The frames appeared and participants responded without consciously thinking about it. They would do this even when as many as four targets had been presented.
• What this means, according to Schneider and Shiffrin, is that practice made it possible for participants to divide their attention to deal with all of the target and test items simultaneously. • Furthermore, the many trials of practice resulted in automatic processing, a type of processing that occurs • (1) without intention (it automatically happens without the person intending to do it), and • (2) at a cost of only some of a person’s cognitive resources.
• One of my students described a situation in which she used automatic processing during her summer job working as a mail sorter at the post office. She found the job to be boring but became so good at it that she could listen to audiobooks as she sorted the mail. She said she was able to do so “unconsciously, without thinking about it, ” which is one of the properties of automatic processing. Another demonstration of automatic processing is the Stroop effect, which is illustrated in the following demonstration.