Search papers, labs, and topics across Lattice.
This study investigates the impact of retrieval-augmented generation (RAG) on single-turn mental health question answering, revealing that its benefits are context-dependent. By implementing a selective retrieval policy based on psychoeducational need, coping need, and response specificity, the authors demonstrate that this approach can mitigate the safety-sensitive failures associated with unconditional retrieval. The results indicate that while always using retrieval enhances specificity, it can degrade overall response quality and introduce safety risks, highlighting the importance of controlled retrieval in sensitive applications.
Selective retrieval can enhance mental health QA by preserving response quality and safety while still improving specificity.
Retrieval-augmented generation (RAG) can improve the specificity and grounding of large language model responses, but its effect is not uniformly beneficial in single-turn mental-health question answering, where user queries often combine emotional distress, treatment concerns, and safety-sensitive needs. We study when retrieval helps or hurts mental-health QA, and whether a lightweight selective retrieval policy can better control this trade-off. We operationalize retrieval need using three draft-conditioned utility dimensions: psychoeducational need, coping need, and response specificity, together with a rule-based safety trigger. Following psychotherapy-grounded RAG systems such as coTherapist, we construct a compact and controllable guideline corpus comprising coping-strategy, psychoeducational, and safety resources. We fine-tune an instruction-tuned generator on MentalChat16K using QLoRA and compare Closed-book, Always Retrieval, and Selective Retrieval settings on CounselBench-Eval and CounselBench-Adv. Experiments show that retrieval is not uniformly beneficial in this domain. Always Retrieval improves specificity but lowers overall quality and introduces additional safety-sensitive failures. Selective Retrieval preserves closed-book behavior for low-need cases while avoiding the additional degradation caused by unconditional retrieval, supporting the view that retrieval activation is a safety-sensitive control decision.