William James (1842-1910) is one of the historical giants of the field of psychology, and he is often quoted in the modern literature. One of his most famous quotes provides a starting point for this chapter. Roughly 125 years ago, James wrote:
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. Focalization, concentration, of consciousness are of its essence. It implies withdrawal from some things in order to deal effectively with others, and is a condition which has a real opposite in the confused, dazed, scatterbrained state which in French is called distraction…. (James, 1890, pp. 403-404)
In this quote, James is describing what modern psychologists call selective attention-that is, the skill through which a person focuses on one input or one task while ignoring other stimuli that are also on the scene. But what does this skill involve? What steps do you need to take in order to achieve the focus that James described, and why is it that the focus “implies withdrawal from some things in order to deal effectively with others”?
Early studies of attention used a setup called dichotic listening: Participants wore headphones and heard one input in the left ear and a different input in the right ear. The participants were instructed to pay attention to one of these inputs-the attended channel-and to ignore the message in the other ear-the unattended channel.
To make sure participants were paying attention, investigators gave them a task called shadowing: Participants were required to repeat back what they were hearing, word for word, so that they were echoing the attended channel. Their shadowing performance
was generally close to perfect, and they were able to echo almost 100 % of what they heard. At the same time, they heard remarkably little from the unattended channel. If asked, after a minute or so of shadowing, to report what the unattended message was about, they had no idea (e.g., Cherry, 1953). They couldn’t even tell if the unattended channel contained a coherent message or random words. In fact, in one study, participants shadowed speech in the attended channel, while in the unattended channel they heard a text in Czech, read with English pronunciation. The individual sounds, therefore (the vowels, the consonants), resembled English, but the message itself was (for an English speaker) gibberish. After a minute of shadowing, only 4 of 30 participants detected the peculiar character of the unattended message (Treisman, 1964).
We can observe a similar pattern with visual inputs. Participants in one study viewed a video that has now gone viral on the Internet and is widely known as the “invisible gorilla” video. In this video, a team of players in white shirts is passing a basketball back and forth; people watching the video are player to another. Interwoven with these urged to count how many times the ball is passed from one players (and visible in the video) is another team, wearing black shirts, also passing a ball back and forth; viewers are instructed to ignore these players. difficulty with this task, but, while doing it, they usually don’t see another event.
Viewers have no that appears on the screen right in front of their eyes. Specifically, they fail to notice when someone wearing a gorilla costume walks through the middle of the game, pausing briefly to thump his chest before exiting. (See Figure 5.1; Neisser & Becklen, 1975; Simons &Chabris, 1999; also see Jenkins, Lavie, & Driver, 2005.)
Even so, people are not altogether oblivious to the unattended channel. In selective listening experiments, research participants easily and accurately report whether the unattended channel contained human speech, musical instruments, or silence. If the unattended channel did contain speech, participants can report whether the speaker had high was male or female, or low voice, speaking loudly softly. (For reviews of this early work, see Broadbent, 1958; Kahneman, 1973.) Apparently, or was or then, physical attributes of the unattended channel are heard, even though participants are generally clueless about the unattended channel’s semantic content.
In one study, however, participants asked to shadow one passage while ignoring ere a Embedded second within the passage. unattended channel was a series of names, and one third of the participants did hear roughly their own name when it was spoken-even though (just like in other studies) they heard almost nothing else from the unattended input (Moray, 1959)
And it’s not just names that can “catch” your attention. Mention of a recently seen movie, or of a favorite restaurant, will often be noticed in the unattended channel. More broadly, words with some personal importance are often noticed, even though the rest of the unattended channel is perceived only as an undifferentiated blur (Conway, Cowan, & Bunting, 2001; Wood & Cowan, 1995).
How can we put all these research results together? How can we explain both the general insensitivity to the unattended channel and also the cases in which the unattended channel “leaks through”?
One option focuses on what you do with the unattended input. The proposal is that you somehow block processing of the inputs you’re not interested in, much as a sentry blocks the path of unwanted guests but stands back and does nothing when legitimate guests are in view, allowing them to pass through the gate unimpeded. This sort of proposal was central for early theories of attention, which suggested that people erect a filter that shields them from potential distractors. Desired information (the attended channel) is not filtered out and so goes on to receive further processing (Broadbent, 1958)
But what does it mean to “filter” something out? The key lies in the nervous system’s ability to inhibit certain responses, and evidence suggests that you do rely on this ability to avoid certain forms of distraction. This inhibition, however, is rather specific, operating distractor basis. In other words, you might have the ability to inhibit your response to this distractor on a distractor-by- and the same for that distractor, but these abilities are of little value if some new, unexpected distractor comes along. In that case, you need to develop a new skill aimed at blocking the new intruder. (See Cunningham & Egeth, 2016; Fenske, Raymond, Kessler, Westoby, & Tipper, 2005; Frings & Wühr, 2014; Jacoby, Lindsay, & Hessels, 2003; Tsushima, Sasaki, & Watanabe, 2006; Wyatt & Machado, 2013. For a glimpse of brain mechanisms that support this inhibition, see Payne & Sekuler, 2014.)
The ability to ignore certain distractors-to shut them out-therefore needs to be part of our theory. Other evidence, though, indicates that this isn’t the whole story. That’s because you not only inhibit the processing of distractors, you also promote the processing of desired stimuli.
We saw in Chapters 3 and 4 that perception involves a lot of activity, as you organize and interpret the incoming stimulus information. It seems plausible that this activity would require some initiative and some resources from you-and evidence suggests that it does.
In one experiment, participants were told that they would see large “+” shapes on a computer screen, presented for 200 ms (milliseconds), followed by a pattern mask. (The mask was just a meaningless jumble on the screen, designed to disrupt any further processing.) If the horizontal bar of the “+” was longer than the vertical, participants were supposed to press one button; if the vertical bar was longer, they had to press a different button. As a complication, participants weren’t allowed to look directly at the “+.” Instead, they fixated on (i.e., pointed their eyes at) a mark in the center of the computer screen-a fixation target-and the “+” shapes were shown just off to one side (see Figure 5.2).
For the first three trials of the procedure, the participants events proceeded just as expected, and the task was relatively easy. On Trial 4, though, things slightly different: While the target “+” was on the screen, the were fixation target disappeared and was replaced by one of three shapes-a triangle, a rectangle, or a cross. Then, the entire configuration (the target and this new shape) replaced by the was mask.
Immediately after the trial, participants asked: Was there anything different on this were trial? Was anything present, anything changed, that wasn’t there on previous trials? Remarkably, 89% of the participants that there was no change; they had failed to see or anything other than the (attended) “+” To probe the participants further, the researchers told them (correctly) that during the previous trial the fixation had target momentarily shape. were then asked what that disappeared and had been replaced by a The participants shape had been, and were given the choices of a triangle, a rectangle, or a cross (one of which, of course, was the right answer). The responses to this question essentially random. Even when probed in this way, participants seemed not to have seen the shape directly in front of their eyes (Mack & Rock, 1998; also see Mack 2003).
This pattern has been named inattentional blindness (Mack & Rock, 1998; also Mack, 2003) a pattern in which people fail to see a prominent stimulus, even though they’re staring straight at it. In a similar effect, called “inattentional deafness,” participants regularly fail to hear prominent stimuli if they aren’t expecting them (Dalton & Fraenkel, 2012). In other studies participants fail to feel stimuli if the inputs are unexpected; this is “inattentional numbness” (Murphy & Dalton, 2016).
What’s going on here? Are participants truly blind (or deaf or numb) in response to these various inputs? As an alternative, some researchers propose that participants in these experiments did see (or hear or feel) the targets but, a moment later, couldn’t remenmber what they’d just experienced (e.g., Wolfe, 1999; also Schnuerch, Kreiz, Gibbons, & Memmert, 2016). For purposes of theory, this distinction is crucial, but for now let’s emphasize what the two proposals have in common: By either account, your normal ability to see what’s around you, and to make use of what you see. is dramatically diminished in the absence of attention.
Think about how these effects matter outside of the laboratory. Chabris and Simons (2010), for example, call attention to reports of traffic accidents in which a driver says, “I never saw bicyclist! He came out of nowhere! But then-suddenly-there he was, right in front of me.” Drew, Võ and Wolfe (2013) showed that experienced radiologists often miss obvious anomalies in a patient’s CT scan, even when looking right blindness in eyewitnesses to crimes, see at the anomaly. (For similar concerns, related to inattentional Jaeger, Levin, & Porter, 2017.) Or, as a more mundane example, you go to the refrigerator to find the mayonnaise (or the ketchup or the juice) and don’t see. it, even though it’s right in front of you.
In these cases, we lament the neglectful driver and the careless radiologist, and your inability to find the mayo may cause you to worry that you’re losing your mind (as well as your condiments). The reality, though, is that these cases of failing-to-see are than “merely” having a stimulus in front of your eyes. Perception requires entirely normal. Perception requires more some work.
The active nature of perception is also evident in studies of change blindness- observers’ inability to detect changes in scenes they’re looking directly at. In some experiments, participants are shown pairs of pictures separated by a brief blank interval (e.g., Rensink, O’Regan, & Clark, 1997). The pictures in each pair are identical except for one aspect-an “extra” engine shown on the airplane in one picture and not in the other; a man wearing a hat in one picture but not wearing one in the other; and so on (see Figure 5.3). Participants know that their task is to detect any changes in the pictures, but even so, the task is difficult. If the change involves something central to the scene, participants may need to look back and forth between the pictures as many as a dozen times before they detect the change. If the change involves some peripheral aspect of the scene, as many as 25 alternations may be required.
A related pattern can be documented when participants watch videos. In one study, observers watched a movie of two women having a conversation. The camera first focused on one woman, then the other, just as it would in an ordinary TV show or movie. The crucial element of this experiment, though, changed. For example, from one camera angle, participants could plainly see the red plates on the table between the women. When the camera shifted to a different position, though, the plates’ color had changed was that certain aspects of the scene changed every time the camera angle to white. In another shift, one of the women gained a prominent scarf that she didn’t have on a fraction of a second earlier (see Figure 5.4). Most observers, however, noticed none of these changes (Levin & Simons, 1997; Shore & Klein, 2000; Simons & Rensink, 2005).
Incredibly, the same pattern can be documented with live (i.e., not filmed) events. In a remarkable study, an investigator (let’s call him “Leon”) approached pedestrians for directions to a certain building. During the conversation, two men carrying a door approached and deliberately walked between Leon and the research participant. As a result, Leon was momentarily hidden (by the door) from the participant’s view, and in that moment Leon traded places with one of the men carrying the door. A second later, therefore, Leon was able to walk away, unseen, while the new fellow (who had been carrying the door) stayed behind and continued the on a college campus and asked conversation with the participant.
Roughly half of the participants failed to notice this switch. They continued the conversation as though nothing had happened-even though Leon and his replacement were wearing different clothes and had easily distinguishable voices. When asked whether anything odd had happened in this event, many participants commented only that it was rude that the guys carrying the door had walked right through their conversation. (See Simons & Ambinder, 2005; Chabris & Simons, 2010 also see Most et al., 2001; Rensink, 2002; Seegmiller, Watson, & Strayer, 2011. For similar effects with auditory stimuli, see Gregg & Samuel, 2008; Vitevitch, 2003.)
Early versus Late Selection
It’s clear, then, that people are often oblivious to stimuli directly in front of their eyes-whether the stimuli are simple displays on a computer screen, photographs, videos, or real-life events. (Similarly people are sometimes oblivious to prominent sounds in the environment.) As we’ve said, though, there are two ways to think about these results. First, the studies may reveal genuine limits on perception, so that participants literally don’t see (or hear) these stimuli; or, second, the studies may reveal limits on memory, so that participants do see (or hear) the stimuli but immediately forget they’ve just experienced.
Which proposal is correct? One approach to this question hinges on when the perceiver selects the desired input and (correspondingly) when the perceiver stops processing the unattended input. According to the early selection hypothesis, the attended input is privileged from the start, so that the unattended input receives little analysis and therefore is never perceived. According to the late selection hypothesis, all inputs receive relatively complete analysis, and selection occurs after the analysis is finished. Perhaps the selection occurs just before the stimuli reach consciousness, so that become aware only of the attended input. Or perhaps the selection occurs later still-so that all inputs make it (briefly) into consciousness, but then the selection occurs so that only the attended input is remembered.
Each hypothesis captures part of the truth. On the one side, there are cases in which people seem unaware of distractors but are influenced by them anyway-so that the (apparently unnoticed) distractors guide the interpretation of the attended stimuli (e.g., Moore & Egeth, 1997; see Figure 5.5). This seems to be a case of late selection: The distractors are perceived (so that they do have an influence) but are selected out before they make it to consciousness. On the other side, though, we can also find evidence for early selection, with attended inputs being privileged from the start and distractor stimuli falling out of the stream of processing at a very early stage. Relevant evidence comes, for example, from studies that record the brain’s electrical activity in th stimulus has arrived. These studies confirm that the brain activity for attended inputs is milliseconds after a distinguishable from that for unattended inputs just 80 ms or so after the stimulus presentation-a time interval in which early sensory processing is still under way (Hillyard, Vogel, & Luck, 1998; see Figure 5.6).
Other evidence suggests that attention can influence activity levels in the lateral geniculate nucleus, or LGN (Kastner, Schneider, & Wunderlich, 2006; McAlonan, Cavanaugh & Wurtz, 2008 Moore & Zirnsak, 2017; Vanduffel, Tootell, & Orban, 2000). In this case, attention is changing the flow of signals within the nervous system even before the signals reach the brain. (For more on how attention influences processing in the visual cortex, see Carrasco, Ling, & Read, 2004; Carrasco, Penpeci-Talgar, & Eckstein, 2000; McAdams & Reid, 2005; Reynolds, Pasternak, & Desimone, 2000; also see O’Connor, Fukui, Pinsk, & Kastner, 2002; Yantis, 2008.)
e. Demonstration 5.1: Shadowing
Many classic studies attention involved a task called “shadowing” The instructions for this task go like this:
You are about to hear a voice reading some English text. Your job is to repeat what you hear, word for word, as you hear it. In other words, you’ll turn yourself into an “echo box” following along with the message as it arrives and repeating it back, echoing as many of the words as you can.
As a first step, you should try this task. You can do this with a friend and have him or her read to you out of a book while you shadow what your friend is saying. If you don’t have a cooperative friend or any other recording of a voice speaking in try shadowing a voice on a news broadcast, a podcast English.
Most people find this task relatively easy, but they also figure out rather quickly that there are steps they can take to make the task even easier. One obvious adjustment is to shadow in a quiet voice, because otherwise your own shadowing will drown out the voice you’re trying to hear. Another adjustment is in the rhythm of the shadowing: People often settle into a pattern of listening to a phrase, rapidly spewing out that phrase, listening to the next phrase, rapidly spewing it out, and so on. This pattern of phrase-by-phrase shadowing has several advantages. Among them, your thinking about the input as a series of phrases (rather than as individual words) allows you to rely to some extent on inferences about what is being said-so that you can literally get away with listening less, and that makes the overall task of shadowing appreciably easier.
Of course, these inferences as well as the whole strategy of phrase-by-phrase shadowing depend on your being able to detect the structure within the incoming message-providing another example in which your perception doesn’t just “receive” the input; it also organizes the input. How much does this matter? If you have a cooperative friend, you can try this variation on shadowing: Have your friend read to you from a book, but ask your friend to read the material backward. (So, for the previous sentence, your friend would literally say, “backward material the read to friend your ask.. “). Your friend will probably need a bit of practice to do this peculiar task, but once he or she has mastered it, you can try shadowing this backward English.
With backward English, you’ll find it difficult (if not impossible) to keep track of the structure of the material-and therefore much harder to locate the boundaries between phrases, and much harder to make inferences about what you’re hearing. Is shadowing of this backward material harder than shadowing of “normal” material?
Now that you’re practiced at shadowing, you’re ready for the third step. Have one friend read normally (not backward!) to you, and have a second friend read something else to you-both at the same time. (Ask the second friend to choose the reading, so that you don’t know in advance what it is.) Again, if you don’t have cooperative friends nearby, you can do the same with any broadcast or recorded voices. (You can play one voice on your computer and another from a podcast. Or you can have your friend do one voice while the TV news provides another voice. Do whichever of these is most convenient for you.) Your job is to shadow the first friend for a minute or so. When you’re done, here are some questions:
· What was the second voice saying? Odds are ood that you don’t know.
· Could you at least hear when the second voice started and stopped? If the second voice or hesitated in the middle of reading, did you hear that? Odds are good coughed, or giggled, that you did.
In general, people tend to be oblivious to the content of the unattended message, but they hear the physical attributes of this message perfectly well-and it’s that pattern of results that we need to explain. The textbook chapter explores what the explanation might be.
e. Demonstration 5.2: Color-Changing Card Trick
The phenomenon of “change blindness” is easily demonstrated in the laboratory, but it also has many parallels outside of the lab. For example, stage magicians often rely change blindness in their performances-with the audience being amazed by objects seeming to on (some version of materialize or dematerialize, or with an assistant being mysteriously transformed into an entirely different person. In most of these cases, though, the “magic” involves little beyond the audience’s failing to notice straightforward swaps that had, in truth, taken place right before their eyes.
A similar effect is demonstrated in a popular video on YouTube. You can find the video by using your search engine to find “color changing card trick” or you can try this address:www.youtube.com/watch?v=voAntzB7EwE. This demonstration was created by a wonderfully playful British psychologist named Richard Wiseman. Several of Wiseman’s videos are available on YouTube, but you can find others on a website that Wiseman maintains: www.quirkology.com.
Watch the video carefully. Were you fooled? Did you show the standard change-blindness pattern: failing to notice large-scale changes in the visual input?
Notice also that you-like the audience at a magic performance-knew in advance that someone was trying to fool you with a switch that you wouldn’t notice. You were, therefore, presumably on your guard-extra vigilant, trying not to be fooled. Even so, the odds are good that you were still fooled, and this is, in itself, an important fact. It tells us that being extra careful is no protection against change blindness, and neither is an effort toward being especially observant. These points are crucial for the stage magician, who is able to fool the people in the audience despite their best efforts toward detecting the tricks and penetrating the illusions.
The same points tell us something about the nature of attention: It’s not seems, just to “try hard to pay attention.” Likewise, instructions to “pay close attention” may, in many especially useful, it circumstances, have no effect at all. In order to promote attention, people usually need some information about what exactly they should pay attention to. Indeed, if someone told you, before the video, to pay attention to the key (changing) elements, do you think you’d be fooled?
(For more videos showing related effects, search the Internet using the key words “invisible gorilla” and follow the links to the various demonstrations.)
Selection via Priming
Whether selection is early or late, it’s clear that people often fail to see stimuli that are directly in front of them, in plain view. But what is the obstacle here? Why don’t people perceive these stimuli?
In Chapter 4, we proposed that recognition requires a network of detectors, and we argued that these detectors fire most readily if they’re suitably primed. In some cases, the priming is produced by your visual experience-specifically, whether each detector has been used recently or frequently in the past. But we suggested that priming can also come from another source: your expectations about what the stimulus will be.
The proposal, then, is that you can literally prepare yourself for perceiving by priming the relevant detectors. In other words, you somehow reach into the network and deliberately activate just those detectors that, you believe, will soon be needed. Then, once primed in this way, those detectors will be on “high alert” and ready to fire.
Let’s also suppose that this priming isn’t “free. Instead, you need to spend some effort or allocate some resources in order to do the priming, and these resources are in limited supply. As a result, there’s a limit on just how much priming you can do.
We’ll need to flesh out this proposal in several ways, but even so, we can already use it to explain some of the findings we’ve already met. Why don’t participants notice the shapes in the inattentional blindness studies? The answer lies in the fact that they don’t expect any stimulus to appear, so they have no reason to prepare for any stimulus. As a result, when the stimulus is presented, it falls on unprepared (unprimed, unresponsive) detectors. The detectors therefore don’t respond to the stimulus, so the participants end up not perceiving it.
What about selective listening? In this case, you’ve been instructed to ignore the unattended input, so you have no reason to devote any resources to this input. Hence, the detectors needed for the distractor message are unprimed, and this makes it difficult to hear the distractor. But why does attention sometimes “leak” so that you do hear some aspects of the unattended input? Think about what will happen if your name is spoken on the unattended channel. The detectors for this stimulus are already primed, but this isn’t because at that moment you’re expecting to hear your name. Instead, the detectors for your name are primed simply because this is a stimulus you’ve often encountered in the past. Thanks to this prior exposure, the activation level of these detectors is already high; you don’t need to prime them further. So they will fire even if your attention is elsewhere.
Two Types of Priming
The idea before us, in short, has three elements. First, perception is vastly facilitated by the priming of relevant detectors. Second, the priming is sometimes stimulus – driven-that is, produced by the stimuli you’ve encountered (recently or frequently) in the past. This is repetition priming-priming produced by a prior encounter with the stimulus. This type of priming takes no effort on your part and requires no resources, and it’s this sort of priming that enables you to hear your name on the unattended channel. But third, a different sort of priming is also possible. This priming is expectation-driven and under your control. In this form of priming, you deliberately prime detectors for inputs you think are upcoming, so that you’re ready for those inputs when they arrive. You don’t do this priming for inputs you have no interest in, and you can’t do this priming for inputs you can’t anticipate.
Can we test these claims? In a classic series of studies, Posner and Snyder (1975) gave participants a straightforward task: A pair of letters was shown on a computer screen, and participants had to decide, as swiftly as they could, whether the letters were the same or different. So someone might see “AA” and answer “same” or might see “AB” and answer “different.”
Before each pair, participants saw a warning signal. In the neutral condition, the warning signal was a plus sign (“+”). This signal notified participants that the stimuli were about to arrive but provided no other information. In a different condition, the warning signal was a letter that actually matched the stimuli to come. So someone might see the warning signal “G” followed by the pair “GG.” In this case, the warning signal served to prime the participants for the stimuli. In a third condition, though, the warning signal was misleading. It was again a letter, but a different letter from the stimuli to come. Participants might see “H” followed by the pair “GG.” Let’s consider these three conditions neutral, primed, and misled.
In this simple task, accuracy rates are very high, but Posner and Snyder also recorded how quickly people responded. By comparing these response times (RTs) in the primed and neutral conditions, we can ask what benefit there is from the prime. Likewise, by comparing RTs in the misled and neutral conditions, we can ask what cost there is, if any, from being misled.
Before we turn to the results, there’s a complication: Posner and Snyder ran this procedure in two different versions. In one version, the warning signal was an excellent predictor of the upcoming stimuli. For example, if the warning signal was an A, there was an 80% chance that the upcoming stimulus pair would contain A’s. In Posner and Snyder’s terms, the warning signal provided a “high validity” prime. In a different version of the procedure, the warning signal was a poor predictor of the upcoming stimuli. For example, if the warning signal was an A, there was only a 20% chance that the upcoming pair would contain A’s. This was the “low validity” condition (see Table 5.1).
Let’s consider the low-validity condition first, and let’s focus on those few occasions in which the prime did match the subsequent stimuli. That is, we’re focusing on 20% of the trials and ignoring the other 80% for the moment. In this condition, the participant can’t use the prime as a basis for predicting the stimuli because the prime is a poor indicator of things to come. Therefore, the prime should not lead to any specific expectations. Nonetheless, we do expect faster RTs in the primed condition than in the neutral condition. Why? Thanks to the prime, the relevant detectors have just fired, so the detectors should still be warmed up. When the target stimuli arrive, therefore, the detectors should fire more readily, allowing a faster response.
The results bear this out. RTs were reliably faster (by roughly 30 ms) in the primed condition than in the neutral condition (see Figure 5.7, left side; the figure shows the differences between conditions). Apparently, detectors can be primed by mere exposure to a stimulus, even in the absence of expectations, and so this priming is truly stimulus-based.
What about the misled condition? With a low-validity prime, misleading the participants had no effect: Performance in the misled condition was the same as performance in the neutral condition. Priming the “wrong” detector, it seems, takes nothing away from the other detectors-including the detectors actually needed for that trial. This fits with our discussion in Chapter 4: Each of the various detectors works independently of the others, and so priming one detector obviously influences the functioning of that specific detector but neither helps nor hinders the other detectors.
Let’s look next at the high-validity primes. In this condition, people might see, for example, a “J” as the warning signal and then the stimulus pair “JJ.” Presentation of the prime itself will fire the J- detectors, and this should, once again, “warm up” these detectors, just as the low-validity primes did. As a result, we expect a stimulus-driven benefit from the prime. However, the high-validity primes may also have another influence: High-validity primes are excellent predictors of the stimulus to come. Participants are told this at the outset, and they have lots of opportunity to see that it’s true. High-validity primes will therefore produce a warm-up effect and also an expectation effect, whereas low-validity primes produce only the warm-up. On this basis, we should expect the high-validity primes to help participants more than low-validity primes-and that’s exactly what the data show (Figure 5.7, right side). The combination of warm-up and expectations leads to faster responses than warm-up alone. From the participants’ point of view, it pays to know what the upcoming stimulus might be.
Explaining the Costs and Benefits
The data make it clear, then, that we need to distinguish two types of primes. One type is stimulus- based-produced merely by presentation of the priming stimulus, with no role for expectations. The other type is expectation-based and is created only when the participant believes the prime allows a prediction of what’s to come.
These types of primes can be distinguished in various ways, including the biological mechanisms that support them (see Figure 5.8; Corbetta & Shulman, 2002; Hahn, Ross, & Stein, 2006; but also Moore & Zirnsak, 2017) and also a difference in what they “cost.” Stimulus-based priming appears to be “free”-we can prime one detector without taking anything away from other detectors. (We saw this in the low-validity condition, in the fact that the misled trials led to responses just as fast as those in the eutral trials.) Expectation-based priming, in contrast, does have a cost, and we see this in an aspect of Figure 5.7 that we’ve not yet mentioned: With high-validity primes, responses in the misled condition were slower than responses in the neutral condition. That is, misleading the participants actually hurt their performance. As a concrete example, F-detection was slower if G was primed, compared to F-detection when the prime was simply the neutral warning signal (“+”). In broader terms, it seems that priming the “wrong” detector takes something away from the other detectors, and so participants are worse off when they’re misled than when they receive no prime at all.
What produces this cost? As an analogy, let’s say that you have just $50 to spend on groceries. You can spend more on ice cream if you wish, but if you do, you’ll have less to spend on other foods. Any increase in the ice cream allotment, in other words, must be covered by a decrease somewhere else. This trade-off arises, though, only because of the limited budget. If you had unlimited funds, you could spend more on ice cream and still have enough money for everything else.
Expectation-based priming shows the same pattern. If the Q-detector is primed, this takes something away from the other detectors. Getting prepared for one target seems to make people less prepared for other targets. But we just said that this sort of pattern implies a limited “budget.” If an unlimited supply of activation were available, you could prime the Q-detector and leave the other detectors just as they were. And that is the point: Expectation-based priming, by virtue of revealing costs when misled, reveals the presence of a limited-capacity system.
We can now put the pieces together. Ultimately, we need to explain the facts of selective attention, including the fact that while listening to one message you hear little content from other messages. To explain this, we’ve proposed that perceiving involves some work, and this work requires some limited mental resources-some process or capacity needed for performance, but in limited supply. That’s why you can’t listen to two messages at the same time; doing so would require more resources than you have. And now, finally, we’re seeing evidence for those limited resources: The Posner and Snyder research (and many other results) reveals the workings of a limited-capacity system, just as our hypothesis demands.
The Posner and Snyder study shows that expectations about an upcoming stimulus can influence the processing of that stimulus. But what exactly is the nature of these expectations? How precise or vague are they?
As one way of framing this issue, imagine that participants in a study are told, “The next stimulus will be a T” In this case, they know exactly what to get ready for. But now imagine that participants are told, “The next stimulus will be a letter” or “The next stimulus will be on the left side of the screen” Will these cues allow participants to prepare themselves?
These issues have been examined in studies of spatial attention-that is, the mechanism through which someone focuses on a particular position in space. In one early study, Posner, Snyder, and Davidson (1980) required their participants simply to detect letter presentations; the task was just to press a button as soon as a letter appeared. Participants kept their eyes pointed at a central fixation mark, and letters could appear either to the left or to the right of this mark.
For some trials, a neutral warning signal was presented, so that participants knew a trial was about to start but had no information about stimulus location. For other trials, an arrow was used as the warning signal. Sometimes the arrow pointed left, sometimes right; and the arrow was generally an accurate predictor of the location of the stimulus-to-come. If the arrow pointed right, the stimulus would be on the right side of the computer screen. (In the terms we used earlier, this was a high-validity cue.) On 20 % of the trials, however, the arrow misled participants about location.
The results show a familiar pattern (Posner et al., 1980). With high-validity priming, the data show a benefit from cues that correctly signal where the upcoming target will appear. The differences between conditions aren’t large, but keep the task in mind: All participants had to do was detect the input. Even with the simplest of tasks, it pays to be prepared (see Figure 5.9).
What about the trials in which participants were misled? RTs in this condition were about 12% slower than those in the neutral condition. Once again, therefore, we’re seeing evidence of a limited- capacity system. In order to devote more attention to (say) the left position, you have to devote less attention to the right. If the stimulus then shows up on the right, you’re less prepared for it-which is the cost of being misled.
Attention as a Spotlight
Studies of spatial attention suggest that visual attention can be compared to a spotlight beam that can “shine” anywhere in the visual field. The “beam” marks the region of space for which you are prepared, so inputs within the beam are processed more efficiently. The beam can be wide or narrowly focused (see Figure 5.10) and can be moved about at will as you explore (i.e., attend to) various aspects of the visual field.
Let’s emphasize, though, that the spotlight idea refers to movements of attention, not movements of the eyes. Of course, eye movements do play an important role in your selection of information from the world: If you want to learn more about something, you generally look at it. (For more on how you move your eyes to explore a scene, see Henderson, 2013; Moore & Zirnsak, 2017.) Even so, movements of the eyes can be separated from movements of attention, and it’s attention, not the eyes, that’s moving around in the Posner et al. (1980) study. We know this because of the timing of surprisingly slow, requiring 180 to 200 ms. But the benefits of primes can be detected within the first 150 ms after the priming stimulus is presented. Therefore,, the the effects. Eye movements are benefits of attention occur prior to any eye movement, so they cannot be a consequence of eye movements.
But what does it mean to “move attention”? The spotlight beam is just a metaphor, so we need to ask what’s really going on in the brain to produce these effects. The answer involves a network of sites in the frontal cortex and the parietal cortex. According to one proposal (Posner & Rothbart, 2007; see Figure 5.11), one cluster of sites (the orienting system) is needed to disengage attention from one target, shift attention to new target, and then engage attention on the new target. A second set of sites (the alerting system) is responsible for maintaining an alert state in the brain. A third set of sites (the executive system) controls voluntary actions.
These points echo a theme we first met in Chapter 2. There, we argued that cognitive capacities on the coordinated activity of multiple brain regions, with each region providing depend a specialized process necessary for the overall achievement. As a result, a regions can disrupt the overall capacity, and if there are problems in several regions, the disruption problem in any of these can be substantial.
As an illustration of this interplay between brain sites and symptoms, consider disorder we mentioned earlier-ADHD. (We’ll have more to say about ADHD later in the chapter.) Table 5.2 summarizes one proposal about this disorder. Symptoms of ADHD are listed in the left column; the right column identifies brain areas that may be the main source of each symptom. This proposal is not the only way to think about ADHD, but it illustrates the complex, many-part relationship between overall function (in this case, the ability to pay attention) and brain anatomy. (For more on ADHD, see Barkley, Murphy, & Fischer, 2008; Brown, 2005; Seli, Smallwood, Cheyne, & Smilek, 2015; Zillmer, Spiers, & Culbertson, 2008.)
In addition, the sites listed in Table 5.2 can be understood roughly as forming the “control system” for attention. Entirely different sites (including the visual areas in the occipital cortex) do the actual analysis of the incoming information (see Figure 5.12). In other words, neural connections from the areas listed in the table carry signals to the brain regions that do the work of analyzing the input. These control signals can amplify (or, in some cases, inhibit) the activity in these other areas and, in this way, they can promote the processing of inputs you’re interested in, and undermine the processing of distractors. (See Corbetta & Shulman, 2002; Hampshire, Duncan, & Owen, 2007: Hon, Epstein, Owen. & Duncan, 2006; Hung, Driver, & Walsh, 2005; Miller & Cohen, 2001.)