This paper explores the causal perspective of graph contrastive learning (GCL) and addresses the issue that traditional GCL methods may not effectively learn invariant representations due to the presence of non-causal information in graphs. The authors propose a novel GCL method called Graph Contrastive Invariant Learning (GCIL), which aims to enhance the learning of invariant representations by simulating causal interventions and designing invariance and independence objectives. The invariance objective ensures that the encoder captures invariant information from causal variables, while the independence objective reduces the influence of confounders on causal variables. Experimental results on five node classification datasets demonstrate the effectiveness of GCIL, showing superior performance compared to both semi-supervised and self-supervised baselines. The paper also includes ablation studies and hyper-parameter sensitivity analysis to validate the contributions of each component of the proposed method.This paper explores the causal perspective of graph contrastive learning (GCL) and addresses the issue that traditional GCL methods may not effectively learn invariant representations due to the presence of non-causal information in graphs. The authors propose a novel GCL method called Graph Contrastive Invariant Learning (GCIL), which aims to enhance the learning of invariant representations by simulating causal interventions and designing invariance and independence objectives. The invariance objective ensures that the encoder captures invariant information from causal variables, while the independence objective reduces the influence of confounders on causal variables. Experimental results on five node classification datasets demonstrate the effectiveness of GCIL, showing superior performance compared to both semi-supervised and self-supervised baselines. The paper also includes ablation studies and hyper-parameter sensitivity analysis to validate the contributions of each component of the proposed method.