According to Theme 2, our cognitive processes are remarkably efficient and accurate. However, when we occasionally do make a mistake, that mistake can often be traced to a “smart mistake,” such as overusing the strategy of top-down processing. Because we
overuse top-down processing, we sometimes demonstrate change blindness; we fail to detect a change in an object or a scene. Overusing top-down processing can also lead us to demonstrate inattentional blindness; when we are paying attention to some events in a scene, we may fail to notice when an unexpected but completely visible object suddenly appears (Most et al., 2005). Let’s consider these two kinds of visual- processing errors.
Change Blindness. Imagine that you are walking along a sidewalk near your col- lege campus. Then, as in Part A of Figure 2.7, a stranger asks you for directions to a particular building. Right in the middle of this interaction, two workers—who are carrying a wooden door—walk between you and the stranger. (See Part B of Figure 2.7.) When they have passed by, the original stranger has been replaced by a different stranger. Would you notice that you are no longer talking with the same individual? You may be tempted to reply, “Of course!”
When Daniel Simons and Daniel Levin (1997a; 1997b; 1998) tried this stranger- and-the-door study, only half of the bystanders reported that one stranger had been FIGURE 2.7
A Study on Change Blindness. These photos are four frames from a video of a study on change blindness. Frames A through C show the sequence of the shift, and Frame D shows the original stranger and the “substitute stranger,” standing side by side.
A B
C D
Top-Down Processing and Visual Object Recognition 49
replaced by a different stranger. Many were still “clueless” when they were explicitly asked, “Did you notice that I’m not the same person who approached you to ask for directions?” (Simons & Levin, 1998, p. 646). Take a moment to try Demonstration 2.4. How quickly can you detect the difference between these two similar scenes?
This chapter examines how we see objects. When perceiving an entire scene, our top-down processing encourages us to assume that the basic meaning of the scene will remain stable. This assumption is rational, and the mistaken perception makes sense (Saylor & Baldwin, 2004). In the real world, one person does not suddenly morph into a different person!
Laboratory research provides other examples of change blindness. For instance, Rensink and his colleagues (1997) asked participants to look at a photo, which was briefly presented twice. Then a slightly different version of the photo was briefly pre- sented twice. This sequence of alternations was repeated until the participant detected the change.
This result showed that people quickly identified the change when the change was important. For example, when a view of a pilot flying a plane showed a helicop- ter either nearby or far away, participants required only 4.0 alternations to report the change. In contrast, they required 16.2 alternations to report a change that was unim- portant, such as the height of a railing behind two people seated at a table. Again, these results make sense (Saylor & Baldwin, 2004). The basic meaning of the scene with the pilot is drastically different if the helicopter is nearby, rather than distant. In contrast, the height of the railing doesn’t change the meaning of the scene at the table. Furthermore, people are somewhat likely to notice an improbable change, such as a window in one picture that is entirely missing in the next picture. In contrast, they seldom notice a probable change, such as a window in one picture that is covered with a curtain in the next picture (Beck et al., 2004).
Additional studies confirm that people are surprisingly blind to fairly obvious changes in the objects that they are perceiving (e.g., Rensink, 2002; Saylor & Baldwin, 2004; Scholl et al., 2004; Simons et al., 2002). In general, then, when we look at a scene with many objects, most people do not store a detailed representation of that scene.
However, Heather Pringle and her coauthors (2004) have discovered individual differences in this skill. People who are accurate in detecting changes are typically more skilled in paying attention to a large visual scene, and they are also more skilled in storing an accurate image of a geometric shape.
Demonstration 2.4
Detecting the Difference Between Two Pictures
Turn to the front of this book, on the right-hand side, where you’ll see two color photos of children in a park. Look back and forth between these two scenes until you have detected which feature is different. The answer is at the end of the chapter, on p. 63.
Inattentional Blindness. As we noted on page 48, inattentional blindness occurs when you are paying attention to something, and you fail to notice an unexpected but completely visible object. In general, psychologists use the term change blindness when people fail to notice a change in some part of the stimulus. In contrast, they use the term inattentional blindness when people fail to notice that a new object has appeared. In both cases, however, people are using top-down processing as they concentrate on some objects in a scene. As a result, when an object appears that is not consistent with their concepts, expectations, and memory, people often fail to recognize this changed object (change blindness) or this new object (inattentional blindness).
Let’s now consider a dramatic study about inattentional blindness. Simons and Chabris (1999) asked participants to watch a videotape of people playing basketball. They were instructed to mentally tally the number of times that members of a spec- ified group made either a bounce pass or an aerial pass. Shortly after the video began, a person dressed in a gorilla suit wandered into the scene and remained there for 5 sec- onds. Amazingly, 46% of the participants failed to notice the gorilla! Other research confirms that people often fail to notice a new object, if they are paying close atten- tion to something else (Intraub, 1999; Most et al., 2001; Most et al., 2005). Inciden- tally, Daniel Simons’s Web site contains some interesting demonstrations of his research: http://viscog.beckman.uiuc.edu/djs_lab/demos.html
Theme 2 of this textbook states that our cognitive processes are remarkably effi- cient and accurate. How can we reconcile the data on change blindness with this theme? One important point is that many of the visual stimuli that people fail to see are not high in ecological validity (Rachlinski, 2004). Studies are high in ecological validity if the conditions in which the research is conducted are similar to the natu- ral setting where the results will be applied. Frankly, I doubt if anyone reading this book has seen someone in a gorilla suit strolling through a basketball game!
Simons and Levin (1997a) emphasize that we actually function very well in our normal visual environment. If you are walking along a busy city street, a variety of per- ceptual representations will rapidly change from one glance to the next. People move their legs, shift a bag to another arm, and move behind traffic signs. If you precisely tracked each detail, your visual system would rapidly be overwhelmed by the trivial changes. Instead, your visual system is accurate in integrating the gist or general inter- pretation of a scene. You focus only on the information that appears to be important, such as the distance of an approaching bus as you cross the street, and you ignore unimportant details. Change blindness and inattentional blindness illustrate a point we made in connection with Theme 2: Our cognitive errors can often be traced to the use of a rational strategy.
We have been discussing research that illustrates how we make errors in object recognition if we are not paying close attention to the object. In Chapter 3 (Per- ceptual Processes II: Attention and Consciousness), we will examine attention in more detail.
So far, we have discussed the visual system, perceptual organization, and theories of object recognition. We have also emphasized the importance of top-down process- ing in perception. Now let’s consider another topic in some detail. One of the most active areas of research on object recognition is the challenging topic of face perception.