Users will develop new interests unrelated to historical interactions and using traditional recommendation methods will form filtering bubbles, resulting in a certain degree of limitation in the range of information received by users. In this paper, we propose data enhancement based on neighborhood effects to steer user interests. Firstly, we design a novel counterfactual attribution neighbor search strategy to obtain neighbor users who are more closely related to the user’s interests. Secondly, we use the causal session reconstruction module to extract the causal dependence and causal differences between neighbor sessions and user sessions. In addition, we propose a causal attention information fusion module to enhance user sessions by utilizing effective information from neighbor sessions. The experimental results on public datasets show that the proposed model consistently outperforms state-of-the-art baselines.