PS261 Lecture Notes - Lecture 8: Feature Learning, Prototype Theory, Railways Act 1921
Document Summary
Pigeons are trained on a simple decimation: red light on response is rewarded (s+, red light off response is not rewarded (s-) When tested with lights of different wavelengths, responding varies with distance from original stimulus: pigeons generalize (somewhat, transposition test, varying the stimulus along a dimension, peak response is to trained wavelength (s+) When the discrimination is between two stimuli that vary along that same dimension: red light (650nm) response is rewarded (s+) Abdul asif: green light (500nm) response is not rewarded (s-) Peak response wavelength is shifted away from s- (towards even redder stimuli) Animals can learn to produce or inhibit a response to a stimulus. Amount of responding depends on the difference between the inhibitory and the excitatory responses hatch on the right. Generalization of both responses causes the peak shift: always away from the s- Grouping of stimuli that share features of functionality (or some other organizing principle)
