ecco: An error correcting comparator theory.

Area of Science:

• Cognitive Psychology
• Behavioral Neuroscience
• Computational Neuroscience

Background:

• The comparator hypothesis explains associative learning by comparing conditioned stimulus (CS) representations.
• Existing formalizations face challenges in explaining certain learning phenomena and stimulus salience effects.

Purpose of the Study:

• To propose a novel formalization of the comparator hypothesis, termed Error-Correcting COmparisons (ecco).
• To address shortcomings of current comparator models and integrate principles of error correction in associative learning.

Main Methods:

• The ecco model retains the comparator process and contingency-based CS-CS association learning.
• It incorporates total error correction for CS-US association learning, drawing from Rescorla-Wagner principles.
• Model behavior is evaluated across standard associative learning paradigms: acquisition, blocking, and retrospective revaluation.

Main Results:

• ecco successfully accounts for phenomena like acquisition, blocking, backward blocking, and unovershadowing.
• The model overcomes limitations of existing comparator models, including difficulties with discriminations and paradoxical salience effects.
• ecco demonstrates characteristic comparator theory phenomena, such as retrospective revaluation.

Conclusions:

• The ecco model offers a robust framework for understanding associative learning within the comparator hypothesis.
• By integrating error correction, ecco provides a more comprehensive explanation for various learning phenomena.
• This formalization advances our understanding of how organisms learn associations and adapt their behavior based on experience.