Neurons in the amygdala are thought to extract from external stimuli their social and emotional significance. The context in which stimuli are received is crucial to the determination of their social importance. Changes in social and behavioral context play out on a longer timescale than individual stimuli (e.g., a touch stimulus), and experimental evidence shows that context can be encoded by the enhanced or suppressed baseline firing of individual neurons that persists across multiple trials. It is possible, however, that context is encoded by network dynamics that emerge from the activity of a large number of neurons. To examine this possibility, we analyzed local field potential (LFP) data recorded from the amygdala of two macaque monkeys presented with alternating blocks of social and non-social tactile stimuli (Gothard Lab, UA). We trained a deep convolutional neural network to classify spectrograms of baseline (inter-stimulus) LFP activity by experimental block type (social vs. non-social context).   This talk will cover the results of this classification task, which show that context can be reliably decoded from baseline LFP in amygdala, regardless of nuclei region, providing evidence of distinct context-dependent brain states. Additionally, I will discuss recent work in analyzing the context-dependent changes in synchrony between spike trains and LFP as well as ongoing and future plans for improving interpretability of the network.