Emergence of Emotion Selectivity in Deep Neural Networks Trained to Recognize Visual Objects

Datasets and analysis code of the following publication: Peng Liu, Ke Bo, Mingzhou Ding and Ruogu Fang (2024). Emergence of Emotion Selectivity in Deep Neural Networks Trained to Recognize Visual Objects. PLOS Computational Biology. DOI: 10.1371/journal.pcbi.1011943 For any questions please contact the first author at mail pliu1 [at] ufl [dot] edu Contents: Code_DataAnalysis - Extracted Selectivity -- IAPS and NAPS datasets -- Neurons In Alexnet and VGG networks --Networks are pre-trained on ImageNet and randomly initialized - Extracted Overlapped Selectivity across IAPS and NAPS. - Extracted tuning performance changes from two datasets and the VGG network -Code to replicate the key results including --Tuning quality -- Number of overlapped neurons -- Enhance neuron activity -- Lesion neurons TrainedNetworks --Pre-trained VGG network on ImageNet --Pre-trained Alexnet network on ImageNet After pre-training these networks on ImageNet, we fixed their weights and trained them to classify pleasant, neutral, and unpleasant images into three emotion categories using both IAPS and NAPS datasets. Image datasets Access image datasets by request from https://csea.phhp.ufl.edu/media/iapsmessage.html for IAPS and https://lobi.nencki.edu.pl/research/8/ for NAPS.