Chapter 10
Knowing about Knowledge
In This Chapter
Conceptualising knowledge
Grouping knowledge into structures
Categorising things as things
We’d like you to think about One Direction for a moment (a delight for some, a chore for others!). Whether you like the group or not, you’ve probably heard its songs, you may know some of the members’ names and a few facts about them, and perhaps you can think of what they look like. All these aspects are called semantic knowledge.
As you can see, your brain can represent knowledge about the same item in various forms:
- Fact: You know that One Direction is a boy band created on The X Factor.
- Visual: You can easily picture the handsome features of your favourite band member.
- Sound: You can imagine the group’s music.
- Smell: You know that the aroma of One Direction’s new fragrance ‘Between Us’ is alluring.
- Touch: You can remember that the texture of sandpaper is rough.
- Taste: You can recall the taste of your grandmother’s roast dinners!
Note that you probably don’t have direct knowledge of how the members of One Direction feel or taste – unless you know them very well!
Cognitive psychologists are interested in how you store and organise all the information you acquire, whether it’s about One Direction or anything else. This chapter looks at two of the most important questions when thinking about knowledge: how does the brain represent this vast array of information and how does it know to link (or bind; see Chapters 8 and 9 on short- and long-term memory, respectively) all this information together?
Psychologists have developed many theories about how knowledge is stored. Here we discuss the idea of concepts and how they may be organised into hierarchies, hubs and spokes, schemas and scripts. We also look at a couple of theories of knowledge representation and at how cognitive psychologists think the brain represents knowledge.
Thinking of Knowledge as Concepts
Most cognitive psychologists think of knowledge as being stored in the form of concepts: abstract representations of categories. In order for this system to work, an individual needs to use the concepts consistently across time and be able to share them with different people. This ability is vitally important for communication: if people have different concepts for the same word, communication becomes impossible!
Concepts can be of anything for which you have a word and be of any level: for example, you can have a concept of people, a concept of boys, a concept of husbands in general and a concept of your particular husband.
Introducing the idea of concepts
You may be thinking of concepts as being like entries in a dictionary or an encyclopaedia. This interpretation of concepts is tempting, but overly simplistic. It works when you’re asked questions in a pub quiz, but people rarely use concepts in exactly the same way every time in life.
Usually, you need to understand different aspects of a concept when you have different goals. In fact, concepts of objects must link to how they’ll be used, which means that the brain has connections between the brain areas for concepts and the brain areas for movement and action.
Cognitive psychologists are extremely interested in the format by which the brain stores concepts and knowledge. In general, they identify four main formats for representations:
- Images: The brain can store some knowledge only in the form of images of the world (therefore in one modality or one format or sense). These images represent a particular moment in time and a particular area of the visual scene, like a photograph.
- Feature records: The brain stores some knowledge in terms of how useful combinations of features are. These features are all in the same modality. The idea is that creatures have representations of how useful a combination is. For example, a frog may detect motion of a small round object and combine these two meaningful features into the representation of an insect (a delicious snack).
- Symbols: Technically called amodal symbols. Representations of these types aren’t restricted to one format. Instead they contain information about how other items may interact, including a list of all the properties that belong to a category. These properties are highly abstract. For example, to the frog, a fly has several features including its taste, sound, movement pattern and look.
- Statistics: Technically, statistical patterns in neural nets. This approach is very computational and is based on the idea that when a concept is active, a series of connected and related set of features are active and this pattern is the concept. Concepts are abstract entities that use information obtained across the different senses and which aren’t tied to any particular modality.
Some debate exists about whether you can have concepts for things you don’t have a word for (see Chapter 16). Most models of knowledge don’t allow for representations of such things: they assume that you have to create a word for something first in order to represent it.
Ordering concepts: Hierarchies
After you realise that knowledge is stored as abstract concepts (see the preceding section), you need to establish how this information links together: such as, how does a concept of a person link to the concept of your husband? This issue contains two related problems: how do concepts at different levels link together and how do concepts at the same level link? Therefore, you need to consider the different levels of representation.
We look at two ways in which the brain may represent that knowledge: hierarchies in this section and the hub-and-spoke model in the next.
Look at Figure 10-1 and name the object before reading on.
When you’re asked to identify something (label it), you tend to do so using different levels, which psychologists call a hierarchy:
- General: The top or superordinate level is general, vague and abstract and doesn’t provide a great deal of specific information. For example, you may say that Figure 10-1 shows an ‘animal’.
- Basic: This middle level is more informative than the general level. People find thinking of basic-level traits easier. For example, when shown Figure 10-1, many people go to the basic level and name it as a ‘monkey’ or more accurately an ‘ape’.
- Specific: The bottom or subordinate level is specific to the object and highly informative. For example, you may label the creature in Figure 10-1 an ‘orangutan’ (or even ‘Tuan’ if you know its name!).
This hierarchy is similar to the way scientists classify living things in the natural world, starting with Kingdom: Animalia and going right down to the species level (Pongo pygmaeus in the case of the pictured orangutan).
People use concepts at every level depending on the context, but typically they use basic-level names. American psychologist Eleanor Rosch and colleagues conducted a study in which they showed a series of pictures, like the one in Figure 10-1, to participants and asked them to name the item in each picture. Rosch found that people used the basic-level names 1,595 times compared to 14 times for specific-level names and only once for general-category names. This shows a preference for the basic level.
With one type of object, everyone uses the specific level: faces. When you see a face of someone you know, you name the person (that is, use the specific level name) rather than say ‘man’ or ‘person’.
Expertise also seems to affect the level you use. A primatologist (who studies monkeys) probably looks at Figure 10-1 and uses the specific level (Bornean Orangutan). Mind you, instead of expertise, perhaps a better word is familiarity. If you present people with familiar buildings, they’re faster to use specific-level names than basic-level names.
Naming tends to occur at the basic level, but it’s not the fastest level for categorising objects. For speed, people tend to use the general category level, suggesting that it’s the first level that comes to mind.
Wheeling away at the hub-and-spokes model
When you consider a concept, such as ‘cheese’, you may picture it, or think of its smell, taste or rubbery texture; that is, you can think of a concept in every sensory modality, not typically considered in the hierarchical approach. With this in mind, British neuroscientist Karalyn Patterson and colleagues proposed the hub-and-spoke model as an alternative to hierarchies.
The hub-and-spoke model is based on the idea of a bicycle wheel, with a central hub and spokes radiating out (see Figure 10-2). The hub represents the core aspects of the concept without any consideration of sensations. It links to each of the spokes, which are sense-specific representations. These spokes are linked to how you perceive and may use the concept.
Organising Knowledge in Your Brain
As well as describing knowledge in terms of concepts (see the preceding section), you can also think about how knowledge is represented in another way. You can group or chunk concepts together (similar to the chunking we describe in Chapter 8).
When they’re grouped together, cognitive psychologists call this way of representing knowledge a schema. If that concept refers to a way of behaving or acting, they refer to it as a script.
We describe these two forms of knowledge in this section, as well as discuss how people may represent spatial knowledge.
For example, some people may have a script for using a fast-food drive-through. The driver’s schema involves knowing that he has to drive to the window, make the order, pay for it and drive somewhere else to collect the order. (We assume this is correct – we don’t drive or haven’t used a drive-through; we’re writing this using a schema derived from watching American movies!) Having a schema or script is useful, because it enables the driver to know how to behave to achieve his goal (buying a burger) and what to expect. This has benefits in terms of processing, because the driver can use the same schema in many different contexts (any drive-through) without having to work out what to do from scratch each time.
Schemas are intricately involved in the way people store information. Without schemas, the world is a hugely confusing and bewildering place. For example, some brain-damaged patients have no problem with scripts but serious problems with concepts (they suffer from semantic dementia). Other patients, with damage to the brain’s frontal lobes, have no problem with concepts but great difficulty planning and organising behaviour, suggesting that they don’t have stored scripts.
Scheming your way to knowing
Schemas are large structures of knowledge, linking a group of concepts together to form knowledge about events or things. When a schema concerns events, cognitive psychologists call it a script (see the next section). When a schema is about a thing, they call it a frame.
Schemas integrate your existing knowledge and influence how future information is stored. Because schemas are based on existing knowledge, people have more difficulty recalling information that’s inconsistent with their schemas. Remembering schematic information requires little effort, but information not in your schema is harder to remember and requires more effort to process.
For example, if you’re presented with a picture of a room that has a slightly unexpected object in it and then are given a memory test, you tend not to remember that slightly unusual object (in fact you often ‘remember’ items that aren’t in the room but are consistent with your schema for what should be in the room). If the object is highly unusual, however, it attracts your attention and is easier to remember.
Frederic Bartlett, a British psychologist, conducted what has become a classic study of the influence on schemas: the War of the Ghosts study. He gave his Cambridge students a Native-Canadian ghost story to read. He found that the students failed to remember information that was inconsistent with their existing schema (inconsistent with their culture).
A stereotype is a special kind of schema that combines all the concepts relating to a particular group of people. Stereotypes are often negative and very rarely contain accurate statements. Stereotyping is usually tested by social psychologists, but the nearby sidebar ‘Illusory correlations’ gives an example of a cognitive psychological explanation about the formation of stereotypes – why they’re inaccurate and the result of a lazy brain.
Scripting knowledge
Scripts are special kinds of schemas that are about how events are sequenced in time and how people should behave in a certain way if the right conditions apply. For example, going into a classroom, you sit behind a desk, get your pen and paper out, and wait (in silence!) for the teacher to arrive. (Are we hoping for too much here?)
A study conducted by American psychologists John Bransford and Marcia Johnson demonstrated how scripts help people organise knowledge. They gave participants a passage of text similar to the following one: it’s a script of an activity, but which one?
It’s a really frustrating task, but luckily you don’t have to do it that often, though more often is better of course. First you have to get it out. You also need all the extra bits, because different bits are used for different things. When it’s all ready you can begin, but not before you’ve made sure that there’s nothing in your way: if there is you have to move it and it’s better to do that at the start than partway through. If you don’t you have to stop and move it and start it again. When you start you have to do it following the appropriate guidelines: this should be really simple – even a child can do it. You should do it thoroughly, making sure to cover every area. If you don’t then it can be quite bad later on. You may need to change parts at different times, but that depends on you. When you’ve finished, you may have to remove the contents and then put it away.
At first, participants have difficulty understanding or remembering elements of the passage unless they’re given the title. But when they know the title they can remember almost all its elements and understanding becomes easy. This experiment shows that knowledge is stored as a group of concepts rather than each one being stored in isolation.
The script refers to vacuum cleaning. Test this passage out on your friends – see how much they can remember without and then with this label.
Finding your way around with routes
When considering knowledge, you often think of knowing facts or how to do things. Routes are important sets of knowledge that combine these two aspects. In order to walk from home to school (or somewhere more exciting, like the library!), you must know the route and the landmarks.
People need two types of knowledge of their environment:
- Routes: Specific paths from one place to another
- Surveys: Map-like knowledge of the environment (associated with more experience within an environment)
Knowledge of routes and surveys depends very much on the individual: it’s egocentric. If you ask people to draw a map of their country, they over-exaggerate the distances between places closer to them.
When people think of their own cities, they focus on five main elements that develop based on their experience within the environment:
- Paths: Channels that people walk (you know, paths!)
- Edges: City boundaries, including walls
- Districts: Self-contained sections of cities
- Nodes: Important parts of an environment – focal points that are used, such as parks
- Landmarks: Important focal points of an environment that aren’t useable, such as historic monuments
Representing Items in Your Head
People often consider knowledge as a fixed construct, such as books in a library, and knowing how people incorporate new knowledge into their heads is certainly important. But knowledge is often used in dynamic settings: by which we mean that you often compare new information that you have to knowledge already stored.
In this section, we look at a few cognitive psychological theories that explore these ideas in a bit more detail. This approach to thinking about knowledge is primarily concerned with concrete objects (real things you can touch, such as, er, concrete).
Defining attributes
Simple models of knowledge use rules or definitions. Definition-based models work by finding attributes (features) that are common to particular objects. The attributes that are common then form a rule.
If you try to list the attributes of a fish, like most people you’re likely to list a number of similar features.
People have a list of attributes in their brains for every kind of object. When you see something new, you list all the attributes and compare them to the ones stored in your knowledge.
Say that you think a fish is made of these attributes: scales, a fin, it swims, it lives in water. Now you see a newt. It looks like it has scales (but doesn’t), swims (sometimes), lives in water (sometimes). If you haven’t experienced it before, you may think it’s a fish – until you obtain enough knowledge to separate the category of fish into fish and newts.
Similar to the object-perception models we describe in Chapter 6, the suggestion is that people break down objects into constituent parts to recognise them. Indeed, perception and knowledge researchers have come up with similar models (no wonder, because you can’t have knowledge without an awareness for how things look, sound, feel and so on!).
The list of common attributes that link all fish are the features people use to define something as a fish. These common attributes must fit all examples in order for the category to work.
The common attributes that fit basic categories are more specific than those that fit general-level categories. Specific-level categories have detailed attributes that define them (see the earlier section ‘Ordering concepts: Hierarchies’ for more on the categories). These traits make them unique and different from the other objects in the basic category. For example, the goldfish has all the common attributes of a fish, but has the added distinctive trait that it’s gold (well, orange).
Created definition rules must be specific enough to differentiate between similar types of objects, but not too specific. For example, imagine that you define a cat as a four-legged, hairy animal that purrs, and then you spot a Sphynx cat. It doesn’t have hair, and so using that rule you may not think that it’s a cat; but it really is (and cute in an alien-looking way!).
Rules tend to be processed in the frontal areas of the brain. Neuroscientists have shown that when categorising objects using rules, the frontal motor areas are more active than other areas of the brain.
Comparing to averages
Prototype theories are a different approach to think about how people store knowledge in the brain. (We consider a version of prototype theory when looking at how objects are recognised in Chapter 6.) The idea is that the brain stores an average example of something.
When you think of the average fish, you probably picture something like our description in the earlier ‘Defining attributes’ section (scales, fins and so on). You categorise quickly everything that’s close to that image as a fish. When you’re presented with something that looks dissimilar to that image, you don’t think of it as a fish. Therefore, when you see an unusual fish type, such as the lionfish, classifying it at the general level as a fish is harder (though classifying it at the specific level and providing its name is easier, if you’re familiar with that name, of course!).
This theory works nicely for objects that have prototypes, but it’s problematic for more difficult concepts that may not have an average. For example, what’s an average ‘game’? The answer probably depends on whether you’re sporty, like board games or are a gambler. Can you really find a prototype that accurately combines football, Monopoly and poker? Defining a set of attributes that links these three things is very difficult.
Examining exemplar theory
Possibly the simplest way of representing knowledge is to store memories of every individual category member: called exemplars. The logic is that every time you see an object of a particular category, you store a new representation of it.
For example, you see a black cat and add it to your exemplar store of cats. You then come across a white cat and also add it to your knowledge of cats. Now you know that cats can be black or white. With more and more exemplars stored, you have an accurate representation of a category.
Sufficient evidence exists that the brain stores exemplars of objects. Unfortunately, memories of exemplars can prevent people from correctly categorising objects. For example, many people have exemplars from seeing cheetahs; subsequently, categorising a leopard becomes more difficult, because of the similarities between the two species.
When storing exemplars, the sensory parts of the brain are used more than the memory sites of the brain.
Putting Aside Knowledge in Your Brain
A simple way of investigating how and where the brain stores concepts (in the sense of the earlier section ‘Thinking of Knowledge as Concepts’) is to measure the brain activity when people think about an object. We look at two such studies now.
Storing in modules
One set of theories suggests that modules in the brain store and process particular things. As we state in Chapter 1, the idea of modules is one of the core assumptions of cognitive neuropsychology. If it’s true, different parts of the brain are responsible for storing different types of knowledge. In which case, you should be able to observe category-specific deficits, where someone has a brain injury that means that he loses knowledge of one type of concept but not others.
Some patients do indeed have difficulty identifying pictures of living things but display relatively preserved abilities to identify non-living things. The thought is that they have a category-specific knowledge deficit for living things but their knowledge of things such as tools is unaffected. Similar cases exist of people unable to name foods or kitchen utensils, but these are rarer than deficits in naming living things.
Another explanation for category-specific deficits suggests that the deficit is in terms of specific information use (that is, category-specific knowledge is really a loss of the most specific detailed level of knowledge). Living things may have more specific details than non-living things, which leads to the observed patterns.
Distributing knowledge
Most of the research on brain imaging when people are thinking about concepts shows quite a wide-ranging activation. Some simple results show that when people think about a physical object, the part of the brain associated with seeing is active. But when they think about an abstract concept (such as ‘freedom’) that ‘seeing’ part isn’t active.
Similar results are obtained when people think about concepts involving action. If you think about a bicycle, the chances are that your brain activates the areas of the brain associated with movement. Whereas if you think about a stationary object (say a chair), the movement areas of the brain aren’t active. Indeed, if the parts of the brain that process movement are temporarily switched off (using Transcranial Magnetic Stimulation), people have more difficulty thinking about moving.
Based on these findings, the hub-and-spoke model is quite plausible (see the earlier section ‘Wheeling away at the hub-and-spokes model’).
Karalyn Patterson suggests that the core of the concepts (the hubs) are stored in the anterior temporal lobes, which don’t contain any sense or motor information. The sensory information (the spokes in the model) that relate to a concept are stored in the various sensory parts of the brain. The hubs integrate all the different types of knowledge. Other researchers, however, suggest the existence of a convergence zone, where part of the brain integrates all the sensory, conceptual and motor information about concepts. This may be located in the superior temporal sulcus.
