PSYCH 120A Chapter Notes - Chapter 4.2: Bigram, Transfer Function
Document Summary
When all feature detectors are activated as a group, this activates the letter detector. Feature nets: network of detectors organized in layers. Subsequent layers are concerned with larger scale objects. Activation level: status of the detector at a moment; how energized the detector is. Response threshold: detector will fire and send its signals to other connected detectors. Detectors likely involve complex assemblies of neural tissue (not necessarily neurons) Some detectors are easier to activate (weaker input) depending on the initial activation level. Detectors that have fired frequently in the past will have a higher activation level. Layer filled with detectors for letter combinations. Triggered by lower level detectors and send output to higher level detectors. Each starts with a certain activation level influenced by frequency and recency. Unfamiliar letter combos need strong inputs to bring relevant detectors to threshold. Letter detectors are wired so they receive input from other feature detectors. If too many letter detectors are firing uncertainty.
