• Nodes: that represent information
• Associations: connections between nodes
• Spreading Activation
• Activation travels from one node to another via associations between the nodes
• Each node has a response threshold
• Activation level of a node depends on
• level of activating input
• recency of activating input

• If the activity reaches response threshold
• activation spreads simultaneously to all other nodes that are linked to the one that reached threshold
• Support for the idea of spreading activation?
• Hints
• State-dependent learning
• Priming
• Mnemonics

Example of a spreading activation model in which the length of each line (link) represents the degree of association between two concepts

Time to travel through the network (semantic hierarchy example)

Another Example Task: Answer TRUE or FALSE (Collins & Quillian, 1969)

• A model to account for storage of complex information (Anderson & Bower, 1973)
• Propositions carry the information
• Boots is a dog
• Boots has a dog
• Children like candy
• Cats do not like dogs
• I bought my Mom a stuffed bunny
• Einstein’s theory of relativity is the subject of tomorrow’s lecture
• Joey would not be pleased if he were to interpret this gesture as condescending

• Spreading activation along all connected links; parallel, automatic process
• Why do we experience retrieval blocks and tip-of-the-tongue phenomenon?
• How is it possible that we find information that is only weakly and distantly connected?
• How do we avoid activation of an overwhelming number of nodes?
• Guided search
• Need a way to guide the search through the network, avoiding activation of irrelevant links and focussing on potentially useful links
• Plethora of associations: Need to limit activation

• Over 100 billion neurons, constantly active
• Ramon y Cajal (1852-1934)
• first to show the neuron as the smallest unit of the nervous system
• Estimated 60 trillion synapses in human cortex (Shepard & Koch, 1990)
• Neuron (figure)

• Multiple, overlapping neural networks, each capable of learning
• Networks of populations of neurons
• Size and number of synapses increase as a result of learning
• Strength of connections between neurons changes with learning

• Critical building block of learning
• neurons become more strongly linked when they are active at the same time
• Use of a function strengthens the neural structure that supports it; disuse weakens it
• Use it or lose it!
• Connectionist or neural net models
• model the simplicity of the single neuron and the complexity of the distributed action of networks of populations of neurons

• Can we explain the tip-of-the tongue phenomenon? What is a retrieval block?
• Memories are seldom unitary; they have many features, some of which may not be accessible
• The distributed pattern involves the same neurons that are activated in other patterns. Activate too many of them and all you have is noise. The critical pattern can not emerge until the system has settled.
• Long-term storage of vast amounts of data yet ability to accurately recall
• rich networks of connections; many retrieval paths
• Acquisition of knowledge
• the network continually learns
• Effects of prior knowledge
• assimilating new information into existing patterns
• alterations in existing memories
• connections strengthened or weakened
• Pattern completion
• errors look like human errors
• filling in gaps, relying on existing patterns, following well-established routes
• Information is distributed
• redundant storage
• graceful degradation
• models are robust to minor damage
• human-type deficits
• No need for a central controller (homunculus)