Semantic networks and spreading activation (video) | Khan Academy

Video transcript

– [Instructor] In order to solve problems and make decisions, you often
have to access information already stored in your brain. But how is that information stored? There are a lot of possibilities, but the one we’re going
to focus on in this video is the semantic network approach, which argues that concepts
are organized in your mind in terms of connected ideas. You can kind of think of this as parallel to how information might
be stored in a computer. You have different nodes,
which represent your concepts, and those nodes are connected by links. Depending on how connected the nodes are, the links might be shorter,
for closely related ideas, or longer, for less related ideas. Let me show you an example to make this a little more concrete. The first semantic network
model was hierarchical, meaning that they thought
concepts were organized from higher order categories
down to lower order categories and their exemplars. So let’s start with a
general category: animal. So that’s this node here. Animal might be linked to other
nodes, such as bird or fish. And bird might be further
linked to canary, bluebird, or more distantly, ostrich. Ostrich is probably not the
first thing you think of when you think of a bird, so
that’s why its link is longer. But simple labels aren’t the only type of knowledge we store. We can store characteristics
and properties of concepts at each node. According to the principle
of cognitive economy, which just means that
our brain is efficient, we store these properties at
the highest possible node. For example then, instead
of storing can breathe at each animal’s node,
we store that property just at the animal node. More specific characteristics
such as sings, or long legs, would be stored at lower level nodes. One pretty interesting
piece of supporting evidence for this type of hierarchical
organization comes from how long it takes people
to verify certain statements. In this kind of test, you say a statement, and ask people to tell
you if it’s true or not. As you can imagine, people
verify a canary is a canary pretty quickly. It takes them a little longer
to verify a canary is a bird, and even longer to verify
that a canary is an animal. So this is some support
that we store things in a hierarchical manner,
because the longer it takes us to verify a connection between two nodes, then the longer those links are, or the more nodes we have to go through to make that link. However, this isn’t true
for all types of animals, or even all types of categories. For example, people tend to
verify that a pig is an animal faster than a pig is a mammal. Because of this issue
and some other problems, Collins and Loftus
proposed a modified version of the semantic network. Rather than a hierarchical organization, this model says that every
individual semantic network develops based on their
experience and knowledge. So, some links might be longer or shorter for different individuals,
and there may be direct links from higher order categories
to their exemplars. One pretty cool thing
about semantic networks is that it means all the ideas in your head are connected together. So when you activate one concept, you’re pulling up related
concepts along with it. This general elevation
and availability is called spreading activation. For example, if I say “fire
engine,” not only do you think of a fire engine, but related
concepts such as trucks, fire, even the color red become activated, making it easier for you to retrieve or identify those items.