Perceptual Selection: Attention

PERCEPTUAL  SELECTION :ATTENTION

Our discussion throughout has been shaped by the challenges we face in getting an accurate view of the world around us. The challenges come from many sides—the enormous number of objects we can recognize; the huge diversity in views we can get of each object as we move around in the world; the possibility of interpret-ing the sensory information we receive in more than one way. Besides all that, here’s another challenge we need to confront: When we look out at the world, we don’t just see one object in front of our eyes; instead, we look at complex scenes containing many objects and activities. These many visual inputs are joined by inputs in other modalities: At every moment, we’re likely to be hearing various noises and perhaps smelling certain smells. We’re also receiving information from our skin senses—per-haps signaling the heat in the room, or the pressure of our clothing against our bod-ies. How do we cope with this sensory feast? Without question, we can choose to pay attention to any of these inputs—and so if the sounds are important to us, we’l focus on those; if the smells are alluring, we may focus on those instead. The one thing we can’t do is pay attention to all of these inputs. Indeed, if we become absorbed in front of our eyes, we may lose track of the sounds in the room. If we lis-ten intently to the sounds, we lose track of other aspects of the environment.

How should we think about all of this? How do we manage to pay attention to some inputs and not others?

Selection

When a stimulus interests us, we turn our head and eyes to inspect it or position our ears for better hearing. Other animals do the same, exploring the world with paws or lips or whiskers or even a prehensile tail. These various forms of orienting serve to adjust the sensory machinery and are one of the most direct means of selecting the input we wish to learn more about (Posner & Rothbart, 2007).

For humans, eye movement is the major means of orienting. Peripheral vision informs us that something’s going on, say, in the upper-left section of our field of vision. But our peripheral acuity isn’t good enough to tell us precisely what it is. To find out, we move our eyes so that the area where the activity is taking place falls into the visual field of the fovea (Rayner, Smith, Malcolm, & Henderson, 2009). In fact, motion in the visual periphery tends to trigger a reflex eye movement, making it difficult not to look toward a moving object (Figure 5.37).

However, eye movements aren’t our only means of selecting what we pay attention to and what we ignore. The selective control of perception also draws on processes involving mental adjustments rather than physical ones. To study these mental adjustments, many experiments rely on the visual search task we’ve already discussed. In this task, someone is shown a set of forms and must indicate as rapidly as possi-ble whether a particular target is present. We noted earlier that this task is effortless if the target can be identified on the basis of just one salient feature—if, for example, you’re searching for a red circle among items that are blue, or for the vertical in a field of horizontals. We also mentioned earlier that in these situations, you can search through four items as fast as you can search through two, or eight as fast as you can search through four. This result indicates that you have no need to look at the figures on the screen one by one; if you did, then you’d need more time as the number of fig-ures grew. Instead, you seem to be looking at all of the figures at the same time.

It’s an entirely different situation when someone is doing a conjunction search—a search in which the target is defined by a combination of features.

Thus, for example, we might ask you to search for a red vertical among distracters that include green verticals and red horizontals (Figure 5.13C), or to search for a blue circle hidden among a bunch of blue squares and red circles. Now it’s not enough to search for (say) red-ness or for the vertical’s orientation; instead, you must search for a tar-get with both of these features. This requirement has an enormous impact on performance. Under these conditions, the search times are longer and depend on the number of items in the display; the more items there are, the longer it takes you to search through them.

What’s going on here? Apparently, we don’t need to focus our attention when looking for the features themselves. For that task, it seems, we can look at all the items in front of us at once; and so it doesn’t matter if there are two items to examine or four or ten. But when we’re looking for a conjunction of features, we need to take the extra step of figuring out how the features are assembled (and thus whether, for example, the redness and the vertical are part of the same stimulus or not). And this step of assembling the features is where attention plays its role: Attention allows us, in essence, to focus a mental spotlight on just a single item. Thanks to this focus, we can analyze the input one stimulus at a time. This process is slower, but it gives us the information we need; if at any moment we’re analyzing only one item, then we can be sure the features we’re detecting all come from that stimulus. This tells us directly that the features are linked to each other—and, of course, for a conjunction search (and for many other purposes as well), this is the key.

Notice, then, that attention is crucial for another issue—the binding problem. This is, you’ll recall, the problem of figuring out which elements in the stimulus information belong with which and, several lines of evidence confirm a role for attention in achieving this “binding.” In some studies, for example, people have been shown brief displays while they’re thinking about some-thing other than the display. Thus, the research participants might be quickly shown a red F and a green X while they’re also trying to remember a short list of numbers they’d heard just a moment before. In this situation, the participants are likely to have no difficulty perceiving the features in the visual display—so they’ll know that they saw something red and something green, and they may also know they saw an F and an X . In a fair number of trials, though, the participants will be confused about how these various aspects of the display were bundled together, so they may end up reporting illusory conjunctions—such as having seen a green F and a red X . Notice, therefore, that simply detecting features doesn’t require the participants’ attention and so goes forward smoothly despite the distracter task. This finding is consistent with the visual search results as well as our earlier comments about the key role of feature detection in object recognition. But, in clear contrast, the combining of fea-tures into organized wholes does require attention, and this process suffers when the participant is somehow distracted.

Related evidence comes from individuals who suffer from severe attention deficits because of damage in the parietal cortex. These individuals can do visual search tasks if the target is defined by a single feature, but they’re deeply impaired if the task requires them to judge how features are conjoined to form complex objects (Cohen & Rafal, 1991; Eglin, L . Robertson, & Knight, 1989; L .Robertson, Treisman, Friedman-Hill, & Grabowecky, 1997).

But what exactly does it mean to “focus attention” on a stimulus or to “shine a mental spotlight” on a particular input? How do we achieve this selection? The answer involves priming—a warming up of certain detectors so they’re better prepared to respond thanthey otherwise would be. Priming can be produced in two ways: First, exposure to a stimulus can cause “data-driven” priming—and so, if you’ve recently seen a red H, or a pic-ture of Moses, this experience has primed the relevant detectors; the result is that, the next time you encounter these stimuli, you’ll be more efficient in perceiving them. But there’s another way priming can occur, and it’s based on expectations rather than recently viewed stimuli. For example, if the circumstances lead you to expect that the word CAT is about to be presented, you can prime the appropriate detectors. This top-down priming will then help if your expectations turn out to be correct. When the input arrives, you’ll process it more efficiently because the relevant detectors are already warmed up.

Evidence suggests that the top-down priming we just described (dependent on expectations rather than recent exposure) draws on some sort of mental resources, and these resources are in limited supply. So if you expect to see the word CAT, you’ll prime the relevant detectors, but this will force you to take resources away from other detectors. As a result, the priming is selective. If you expect to see CAT, you’ll be well pre-pared for this stimulus but less prepared for anything else. Imagine, therefore, that youexpect to see CAT but are shown TREE instead. In this case, you’ll process the input less efficiently than you would if you had no expectations at all. Indeed, if the unexpected stimulus is weak (perhaps flashed briefly or only on a dimly lit screen), then it may not trigger any response. In this way, priming can help spare us from distraction—by selec-tively helping us perceive expected stimuli but simultaneously hindering our perception of anything else.

In the example just considered, priming prepared the perceiver for a particular stimulus—the word CAT. But priming can also prepare you for a broad class of stimuli— for example, preparing you for any stimulus that appears in a particular location. Thus, you can pay attention to the top-left corner of a computer screen by priming just those detec-tors that respond to that spatial region. This step will make you more responsive to any stimulus arriving in that corner of the screen, but it will also take resources away from other detectors, making you less responsive to stimuli that appear elsewhere.

In fact, this process of pointing attention at a specific location can be demonstrated directly. In a typical experiment, participants are asked to point their eyes at a dot on a computer screen (e.g., Wright & Ward, 2008). A moment later, an arrow appears for an instant in place of the dot and points either left or right. Then, a fraction of a second later, the stimulus is presented. If it’s presented at the place where the arrow pointed, the participants respond more quickly than they do without the prime. If the stimulus appears in a different location—so that the arrow prime was actually misleading— participants respond more slowly than they do with no prime at all. Clearly, the prime influences how the participants allocate their processing resources.

It’s important to realize that this spatial priming is not simply a matter of cuing eye movements. In most studies the interval between the appearance of the prime and the arrival of the target is too short to permit a voluntary eye movement. But even so, when the arrow isn’t misleading, it makes the task easier. Evidently, priming affects an inter-nal selection process—it’s as if your mind’s eye moves even though the eyes in your head are stationary.

Perception in the Absence of Attention

As we’ve just seen, attention seems to do several things for us. It orients us toward the stimulus so that we can gain more information. It helps bind the input’s features together so that we can perceive a coherent object. And it primes us so that we can perceive more efficiently and so that, to some extent, we’re sheltered from unwanted distraction.

If attention is this important and has so many effects, then we might expect that the ability to perceive would be seriously compromised in the absence of attention. Recent studies indicate that this expectation is correct, and they document some remarkable sit-uations in which people fail to perceive prominent stimuli directly in front of their eyes.

In one study, participants watched a video showing one group of players, dressed in white shirts, tossing a ball back and forth. Interspersed with these white-shirted players—and visible in the same video—a different group of players, in black shirts, also were tossing a ball. But, when participants were focusing on the white-shirted players, that was all they noticed. They were oblivious to what the black-shirted players were doing, even though they were looking right at them. Indeed, in one experiment, the par-ticipants failed to notice when someone wearing a gorilla suit strolled right through the scene and even paused briefly in the middle of the scene to thump on his chest (Figure 5.38; Neisser & Becklen, 1975; Simons & Chabris, 1999).

In a related study, participants were asked to stare at a dot in the middle of a computer screen while trying to make judgments about stimuli presented just a bit off of their line of view. During the moments when the to-be-judged stimulus was on the screen, the dot

at which the participants were staring changed momentarily to a triangle and then back to a dot. When asked about this event a few seconds later, though, the participants insisted that they’d seen no change in the dot. When given a choice about whether the dot had changed into a triangle, a plus sign, a circle, or a square, they chose randomly. Apparently, with their attention directed elsewhere, the participants were essentially “blind” to a stimulus that had appeared right in front of their eyes (Mack, 2003; Mack & Rock, 1998; also see Rensink, 2002; Simons, 2000; Vitevitch, 2003).