9.4 Concepts and Categories: How We Think

In October 2000, a 69-year old man known by the initials JB went for a neurological assessment because he was having difficulty understanding the meaning of words, even though he still performed well on many other perceptual and cognitive tasks. In 2002, as his problems worsened, he began participating in a research project concerned with the role of language in naming, recognizing, and classifying colors (Haslam et al., 2007). As the researchers observed JB over the next 15 months, they documented that his color language deteriorated dramatically; he had great difficulty naming colors and could not even match objects with their typical colors (e.g., strawberry and red, banana and yellow). Yet even as his language deteriorated, JB could still classify colors normally, sorting color patches into groups of green, yellow, red, and blue in the exact same manner that healthy participants did. JB retained an intact concept of colors despite the decline of his language ability–a finding that suggests that we need to look at factors in addition to language in order to understand concepts (Haslam et al, 2007).

JEFF KAUFMAN/THE NEW YORKER COLLECTION/CARTOONBANK.COM

Concept refers to a mental representation that groups or categorizes shared features of related objects, events, or other stimuli. A concept is an abstract representation, description, or definition that serves to designate a class or category of things. The brain organizes our concepts about the world, classifying them into categories based on shared similarities. Our category for dog may be something like “small, four-footed animal with fur that wags its tail and barks.” Our category for bird may be something like “small, winged, beaked creature that flies.” We form these categories in large part by noticing similarities among objects and events that we experience in everyday life. For example, your concept of a chair might include such features as sturdiness, relative flatness, and an object that you can sit on. That set of attributes defines a category of objects in the world–desk chairs, recliner chairs, flat rocks, bar stools, and so on–that can all be described in that way.

Concepts are fundamental to our ability to think and make sense of the world. We’ll first compare various theories that explain the formation of concepts and then consider studies that link the formation and organization of concepts to the brain. As with other aspects of cognition, we can gain insight into how concepts are organized by looking at some instances in which they are rather disorganized. We’ll encounter some unusual disorders that help us understand how concepts are organized in the brain.

Psychological Theories of Concepts and Categories

There is family resemblance between family members despite the fact that there is no defining feature that they all have in common. Instead, there are shared common features. Someone who also shares some of those features may be categorized as belonging to the family.

BLEND IMAGES/SUPERSTOCK

Early psychological theories described concepts as rules that specify the necessary and sufficient conditions for membership in a particular category. A necessary condition is something that must be true of the object in order for it to belong to the category. For example, suppose you were trying to determine whether an unfamiliar animal was a dog. It is necessary that the creature be a mammal; otherwise it doesn’t belong to the category dog because all dogs are mammals. A sufficient condition is something that, if it is true of the object, proves that it belongs to the category. Suppose someone told you that the creature was a German shepherd and you know that a German shepherd is a type of dog. German shepherd is a sufficient condition for membership in the category dog.

Most natural categories, however, cannot be so easily defined in terms of this classical approach of necessary and sufficient conditions. For example, what is your definition of dog? Can you come up with a rule of “dogship” that includes all dogs and excludes all non-dogs? Most people can’t, but they still use the term dog intelligently, easily classifying objects as dogs or non-dogs. Three theories seek to explain how people perform these acts of categorization.

Family Resemblance Theory

Eleanor Rosch put aside necessity and sufficiency to develop a theory of concepts based on family resemblance, that is, features that appear to he characteristic of category members but may not be possessed by every member (Rosch, 1973, 1975; Rosch & Mervis, 1975; Wittgenstein, 1953/1999). For example, you and your brother may have your mother’s eyes, although you and your sister may have your father’s high cheekbones. There is a strong family resemblance between you, your parents, and your siblings despite the fact that there is no necessarily defining feature that you all have in common. Similarly, many members of the bird category have feathers and wings, so these are the characteristic features. Anything that has these features is likely to be classified as a bird because of this “family resemblance” to other members of the bird category. FIGURE 9.6 illustrates family resemblance theory.

Figure 9.6: Family Resemblance Theory The family resemblance here is unmistakable, even though no two Smith brothers share all the family features. The prototype is brother 9. He has it all: brown hair, large ears, large nose, mustache, and glasses.

Prototype Theory

Building on the idea of family resemblance, Rosch also proposed that psychological categories (those that we form naturally) are best described as organized around a prototype, the “best” or “most typical” member of a category. A prototype possesses most (or all) of the most characteristic features of the category. For North Americans, the prototype of the bird category would be something like a wren: a small animal with feathers and wings that flies through the air, lays eggs, and migrates (see FIGURE 9.7). If you lived in Antarctica, your prototype of a bird might be a penguin: a small animal that has flippers, swims, and lays eggs. According to prototype theory, if your prototypical bird is a robin, then a canary would be considered a better example of a bird than would an ostrich because a canary has more features in common with a robin than an ostrich does. People make category judgments by comparing new instances to the category’s prototype. This contrasts with the classical approach to concepts in which something either is or is not an example of a concept (i.e., it either does or does not belong in the category dog or bird).

Figure 9.7: Critical Features of a Category We tend to think of a generic bird as possessing a number of critical features, but not every bird possesses all of those features. In North America, a wren is a better example of a bird than a penguin or an ostrich.

Exemplar Theory

In contrast to prototype theory, exemplar theory holds that we make category judgments by comparing a new instance with stored memories for other instances of the category (Medin & Schaffer, 1978). Imagine that you’re out walking in the woods, and from the corner of your eye you spot a four-legged animal that might be a wolf or a coyote but that reminds you of your cousin’s German shepherd. You figure it must be a dog and continue to enjoy your walk rather than fleeing in a panic. You probably categorized this new animal as a dog because it bore a striking resemblance to other dogs you’ve encountered; in other words, it was a good example (or an exemplar) of the category dog. Exemplar theory does a better job than prototype theory in accounting for certain aspects of categorization, especially in that we recall not only what a prototypical dog looks like but also what specific dogs look like. FIGURE 9.8 illustrates the difference between prototype theory and exemplar theory.

Figure 9.8: Prototype Theory and Exemplar Theory According to prototype theory, we classify new objects by comparing them to the “prototype” (or most typical) member of a category. According to exemplar theory, we classify new objects by comparing them to all category members.

Concepts, Categories, and the Brain

Studies that have attempted to link concepts and categories to the brain have helped to make sense of the theories we just considered. For example, in one set of studies (Marsolek, 1995), participants classified prototypes faster when the stimuli were presented to the right visual field, meaning that the left hemisphere received the input first (see the Neuroscience chapter for a discussion of how the two hemispheres of the brain receive input from the outside world). In contrast, participants classified previously seen exemplars faster when images were presented to the left visual field (meaning that the right hemisphere received the input first). These results suggest a role for both exemplars and prototypes: The left hemisphere is primarily involved in forming prototypes and the right hemisphere is mainly active in recognizing exemplars.

ROBERT MANKO FF/THE NEW YORKER COLLECTION/CARTOONBANK.COM

More recently, researchers using neuroimaging techniques have also concluded that we use both prototypes and exemplars when forming concepts and categories. The visual cortex is involved in forming prototypes, whereas the prefrontal cortex and basal ganglia are involved in learning exemplars (Ashby & Ell, 2001; Ashby & O’Brien, 2005). This evidence suggests that exemplar-based learning involves analysis and decision making (prefrontal cortex), whereas prototype formation is a more holistic process involving image processing (visual cortex).

Some of the most striking evidence linking concepts and categories with the brain originated in a pioneering study conducted over 30 years ago. Two neuropsychologists (Warrington & McCarthy, 1983) described a man with brain damage who could not recognize a variety of human-made objects or retrieve any information about them, but his knowledge of living things and foods was perfectly normal. In the following year, neuropsychologists (Warrington & Shallice, 1984) reported on four individuals with brain damage who exhibited the reverse pattern: They could recognize information about human-made objects, but their ability to recognize information about living things and foods was severely impaired. Over 100 similar cases have since been reported (Martin & Caramazza, 2003). These unusual cases became the basis for a syndrome called category-specific deficit, an inability to recognize objects that belong to a particular category, although the ability to recognize objects outside the category is undisturbed.

Category-specific deficits like these have been observed even when the brain trauma that produces them occurs shortly after birth. Two researchers reported the case of Adam, a 16-year-old boy who suffered a stroke a day after he was born (Farah & Rabinowitz, 2003). Adam has severe difficulty recognizing faces and other biological objects. When shown a picture of a cherry, he identified it as “a Chinese yo-yo.” When shown a picture of a mouse, he identified it as an owl. He made errors like these on 79% of the animal pictures and 54% of the plant pictures he was shown. In contrast, he made errors only 15% of the time when identifying pictures of nonliving things, such as spatulas, brooms, and cigars. What’s so important about this case? The fact that 16-year-old Adam exhibited category-specific deficits despite suffering a stroke when he was only 1 day old strongly suggests that the brain is “prewired” to organize perceptual and sensory inputs into broad-based categories, such as living and nonliving things.

The type of category-specific deficit suffered depends on where the brain is damaged. Deficits usually result when an individual suffers a stroke or other trauma to areas in the left hemisphere of the cerebral cortex (Mahon & Caramazza, 2009). Damage to the front part of the left temporal lobe results in difficulty identifying humans; damage to the lower left temporal lobe results in difficulty identifying animals; and damage to the region where the temporal lobe meets the occipital and parietal lobes impairs the ability to retrieve names of tools (Damasio et al., 1996). Similarly, when healthy people undertake the same task, imaging studies have demonstrated that the same regions of the brain are more active during naming of tools than animals and vice versa, as shown in FIGURE 9.9 (Martin, 2007; Martin & Chao, 2001).

Figure 9.9: Brain Areas Involved in Category-Specific Processing Participants were asked to silently name pictures of animals and tools while they were scanned with fMRI. Areas in white, including parts of visual cortex and temporal lobe, labeled (1, 4), showed greater activity when participants named animals. Areas in black, including other parts of visual cortex and temporal lobe (2, 3) and a region of motor cortex (5), showed greater activity when participants named tools. Note that the images are left/right reversed.

How do particular brain regions develop category preferences for objects such as tools or animals? One possibility is that these preferences develop from the specific visual experiences that individuals have during the course of their lives. An alternative possibility is suggested by the study of Adam that we just considered: The brain may be prewired such that particular regions respond more strongly to some categories than others. A recent study tested these ideas by examining the activity of category-preferential regions in adults who have been blind since birth (Mahon et al., 2009). While in the fMRI scanner, blind and sighted individuals each heard a series of words, including some words that referred to animals and others that referred to tools. For each word, participants made a judgment about the size of the corresponding object. The critical finding was that category-preferential regions showed highly similar patterns of activity in the blind and sighted individuals. In both groups, for example, regions in the visual cortex and temporal lobe responded to animals and tools in much the same manner as shown in Figure 9.9.

These results provide compelling evidence that category-specific organization of visual regions does not depend on an individual’s visual experience. The category-specific organization could conceivably have arisen from interactions with objects that blind individuals have had involving senses other than vision, such as touch (Peelen & Kastner, 2009). However, when combined with the observations of Adam, the simplest explanation may be that category-specific brain organization is innately determined (Bedny & Saxe, 2012; Mahon et al., 2009).