Associative Emotional Learning in Convolutional Neural Networks

2026-07-21

AI Summary

This study addresses the adaptive learning mechanisms by which organisms associate predictive stimuli with emotional valence, such as pleasure or aversion. To this end, we propose a novel Pavlovian learning paradigm that, for the first time, implements valence-based associative learning within a deep convolutional neural network by integrating visual encoding and emotional valence recognition modules. Through neural representational alignment analyses, we uncover how representations of conditioned and unconditioned stimuli converge at both single-neuron and population levels. The model successfully recapitulates key phenomena observed in human associative learning—including association formation and generalization—and exhibits neural activity patterns that closely align with empirical human data, thereby validating the biological plausibility and efficacy of our approach.

Abstract

Associative emotional learning enables organisms to adaptively link pleasant or unpleasant outcomes to the presence of predictive stimuli. Whereas computational models such as the Rescorla-Wagner model have shed light on this important function, the limitations of these models are also known, especially when they are applied to neural data. The advent of deep neural networks has opened another avenue for modeling associative emotional learning. In this work we proposed a deep neural network model of visual valence processing, consisting of a visual module that encodes complex natural scenes and a module that recognizes their emotional significance in terms of valence, a key dimension of emotion, and tested a novel Pavlovian learning paradigm on the model. The results showed that with learning, the model reproduced several observations from human associative learning studies, including association formation and generalization, and that the neural representations of the conditioned and the unconditioned stimuli became increasingly aligned both at the single unit and at the neural population level. Comparison between the model and human experimental data provided further validation of our approach. This study thus suggests that deep neural network models, when combined with appropriate learning algorithms, can be used to model behavioral and neural signatures of associative emotion/valence learning.

Problem

• associative emotional learning
• valence processing
• deep neural networks
• Pavlovian learning
• neural representations

Innovation

• associative emotional learning
• deep neural networks
• valence processing
• Pavlovian conditioning
• neural representation alignment