Generalization following symmetrical intradimensional discrimination training.

Area of Science:

• Cognitive Psychology
• Behavioral Science
• Computational Neuroscience

Background:

• Generalization models face challenges in explaining how positive (CS+) and negative (CS-) evidence interact to shape generalized responding.
• The monotonicity principle suggests additional negative evidence should decrease generalization, but recent findings contradict this.
• This study investigates generalization phenomena under conditions with multiple sources of negative evidence.

Purpose of the Study:

• To examine if additional negative evidence can increase generalization in a symmetrical intradimensional discrimination.
• To compare generalization gradients between groups receiving single versus double negative evidence.
• To test the applicability of extended associative models to explain observed generalization patterns.

Main Methods:

• Three experiments compared generalization along a color dimension (green-blue) in human participants.
• Participants learned discriminations involving a CS+ (aqua shape) and either one or two CS-s (slightly greener/bluer shapes).
• Generalization was assessed by varying stimulus color and, in one experiment, changing stimulus shape at test.

Main Results:

• Experiments 1A and 1B found no evidence that additional negative evidence increased generalization.
• Experiment 2 revealed a higher generalization gradient peak in the double-negative group when stimulus shape changed at test.
• This outcome, violating the monotonicity principle, was successfully modeled by an extended associative model incorporating cue competition across dimensions.

Conclusions:

• Associative mechanisms, particularly cue competition across multiple stimulus dimensions, can explain instances where negative evidence increases generalization.
• The findings challenge strict adherence to the monotonicity principle in certain discrimination learning contexts.
• This research contributes to understanding the complex interplay of excitation and inhibition in generalization behavior.