Learn something new every day
More Info... by email
Spreading activation is a model of working memory, also known as short-term memory, that seeks to explain how the mind processes related ideas, especially semantic or verbal concepts. The spreading activation model is one way cognitive psychologists explain the priming effect, which is the observable phenomenon that a person is able to more quickly recall information about a subject once a related concept has been introduced. According to this model, semantic long-term memory consists of a vast, interrelated network of concepts. When a person is presented with any concept, the concepts most closely connected to it are activated in that person's mind, preparing or "priming" him or her to recall information related to any of them.
According to the theory of spreading activation, each semantic concept has a node in the neural network that is activated at the same time as the nodes for related concepts. If a person is presented with the concept "dog," nodes for concepts like "bark," "beagle" and "pet" might be activated, priming him or her to think about these related words. Depending on which concept relating to "dog" is presented next, the person is able to recall any information that might be relevant to the task at hand. One such task might be to evaluate the accuracy of semantic statements. The person could, for instance, more quickly verify the statement "A beagle is a dog" if he or she already knows that the topic at hand is "dog."
The stronger the connection between the ideas, the more quickly the person is able to recall relevant information. A person can probably verify very quickly the statement "A bird is an animal," because birds are very common, typical examples of the category "animal." On the other hand, the same person would likely take significantly longer to process and verify the statement "A chinchilla is an animal," because a chinchilla is an atypical member of the category. The model of spreading activation would account for this difficulty by saying that the node for "chinchilla" would not necessarily be activated by the category "animal."
Of course, the associations between semantic concepts vary greatly from person to person. Someone who has a pet chinchilla, for example, will have far greater connections between "animal" and "chinchilla" than the general population. In this way, the semantic categories described by spreading activation are a product both the actual content and of the individual experience. For this reason, the spreading activation model is very useful for describing how the mind has responded to a semantic task, but not necessarily useful for predicting how a person will respond to any given task.
Funnily enough, Telsyst, there's a seminal article that remarks about exactly the same metaphor: From Computers to Anthills (by Gardenfors).
This is an interesting model of how the mind works. I can't help but be reminded of anthill optimization when reading this piece.
If an ant finds a trail left chemically by an ant, that ant is more likely to follow that trail. This ant will be leaving its own chemicals which makes the chemical trail stronger and so more ants will be attracted to this trail.
This optimization causes great numbers of ants to move with precision. A large number of ants could just wander off in many directions, but this keeps them in line.
The pheromones do dissipate over time, so if a trail doesn't lead anywhere, it isn't picked up by other ants, and so that trail
is swept away.
A forager ant looking for food, when finding a meal, can then go back to the main group following its own chemical trail, thereby strengthening that trail. When arriving back to the main group, it will lead other ants down the same trail, leaving no chance for error.