CounselBench: A Large-Scale Expert Evaluation and Adversarial Benchmarking of Large Language Models in Mental Health Question Answering

• URL: http://arxiv.org/abs/2506.08584v2
• Date: Wed, 01 Oct 2025 17:10:02 GMT
• Title: CounselBench: A Large-Scale Expert Evaluation and Adversarial Benchmarking of Large Language Models in Mental Health Question Answering
• Authors: Yahan Li, Jifan Yao, John Bosco S. Bunyi, Adam C. Frank, Angel Hwang, Ruishan Liu,
• Abstract summary: We present CounselBench, a large-scale benchmark developed with 100 mental health professionals to evaluate and stress-test large language models (LLMs)<n>The first component, CounselBench-EVAL, contains 2,000 expert evaluations of answers from GPT-4, LLaMA 3, Gemini, and human therapists on patient questions from the public forum CounselChat.<n>Expert evaluations show that while LLMs achieve high scores on several dimensions, they also exhibit recurring issues, including unconstructive feedback, overgeneralization, and limited personalization or relevance.
• Score: 1.0262304700896199
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Medical question answering (QA) benchmarks often focus on multiple-choice or fact-based tasks, leaving open-ended answers to real patient questions underexplored. This gap is particularly critical in mental health, where patient questions often mix symptoms, treatment concerns, and emotional needs, requiring answers that balance clinical caution with contextual sensitivity. We present CounselBench, a large-scale benchmark developed with 100 mental health professionals to evaluate and stress-test large language models (LLMs) in realistic help-seeking scenarios. The first component, CounselBench-EVAL, contains 2,000 expert evaluations of answers from GPT-4, LLaMA 3, Gemini, and human therapists on patient questions from the public forum CounselChat. Each answer is rated across six clinically grounded dimensions, with span-level annotations and written rationales. Expert evaluations show that while LLMs achieve high scores on several dimensions, they also exhibit recurring issues, including unconstructive feedback, overgeneralization, and limited personalization or relevance. Responses were frequently flagged for safety risks, most notably unauthorized medical advice. Follow-up experiments show that LLM judges systematically overrate model responses and overlook safety concerns identified by human experts. To probe failure modes more directly, we construct CounselBench-Adv, an adversarial dataset of 120 expert-authored mental health questions designed to trigger specific model issues. Evaluation of 3,240 responses from nine LLMs reveals consistent, model-specific failure patterns. Together, CounselBench establishes a clinically grounded framework for benchmarking LLMs in mental health QA.

Related papers

• Assessing the Quality of Mental Health Support in LLM Responses through Multi-Attribute Human Evaluation [14.243791046586347]
  The escalating global mental health crisis, marked by persistent treatment gaps, availability, and a shortage of qualified therapists, positions Large Language Models (LLMs) as a promising avenue for scalable support.<n>This paper introduces a human-grounded evaluation methodology designed to assess LLM generated responses in therapeutic dialogue.
  arXiv  Detail & Related papers  (2026-01-26T16:04:19Z)
• Simulating Viva Voce Examinations to Evaluate Clinical Reasoning in Large Language Models [51.91760712805404]
  We introduce VivaBench, a benchmark for evaluating sequential clinical reasoning in large language models (LLMs)<n>Our dataset consists of 1762 physician-curated clinical vignettes structured as interactive scenarios that simulate a (oral) examination in medical training.<n>Our analysis identified several failure modes that mirror common cognitive errors in clinical practice.
  arXiv  Detail & Related papers  (2025-10-11T16:24:35Z)
• Psychiatry-Bench: A Multi-Task Benchmark for LLMs in Psychiatry [1.2879523047871226]
  PsychiatryBench is a rigorously curated benchmark grounded exclusively in expert-validated psychiatric textbooks and casebooks.<n> PsychiatryBench comprises eleven distinct question-answering tasks ranging from diagnostic reasoning and treatment planning to longitudinal follow-up, management planning, clinical approach, sequential case analysis, and multiple-choice/extended matching formats totaling over 5,300 expert-annotated items.
  arXiv  Detail & Related papers  (2025-09-07T20:57:24Z)
• OpenAIs HealthBench in Action: Evaluating an LLM-Based Medical Assistant on Realistic Clinical Queries [2.2807344448218507]
  We evaluate our agentic, RAG-based clinical support assistant, DR.INFO, using HealthBench.<n>On the Hard subset of 1,000 challenging examples, DR.INFO achieves a HealthBench score of 0.51.<n>In a separate 100-sample evaluation against similar agentic RAG assistants, it maintains a performance lead with a health-bench score of 0.54.
  arXiv  Detail & Related papers  (2025-08-29T09:51:41Z)
• Med-RewardBench: Benchmarking Reward Models and Judges for Medical Multimodal Large Language Models [57.73472878679636]
  We introduce Med-RewardBench, the first benchmark specifically designed to evaluate medical reward models and judges.<n>Med-RewardBench features a multimodal dataset spanning 13 organ systems and 8 clinical departments, with 1,026 expert-annotated cases.<n>A rigorous three-step process ensures high-quality evaluation data across six clinically critical dimensions.
  arXiv  Detail & Related papers  (2025-08-29T08:58:39Z)
• Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models [92.93521294357058]
  Narrative therapy helps individuals transform problematic life stories into empowering alternatives.<n>Current approaches lack realism in specialized psychotherapy and fail to capture therapeutic progression over time.<n>Int (Interactive Narrative Therapist) simulates expert narrative therapists by planning therapeutic stages, guiding reflection levels, and generating contextually appropriate expert-like responses.
  arXiv  Detail & Related papers  (2025-07-27T11:52:09Z)
• Automating Expert-Level Medical Reasoning Evaluation of Large Language Models [26.702477426812333]
  We introduce MedThink-Bench, a benchmark for rigorous, explainable, and scalable assessment of large language models' medical reasoning.<n>We also propose LLM-w-Ref, a novel evaluation framework that leverages fine-grained rationales and LLM-as-a-Judge mechanisms.<n>Overall, MedThink-Bench offers a foundational tool for evaluating LLMs' medical reasoning, advancing their safe and responsible deployment in clinical practice.
  arXiv  Detail & Related papers  (2025-07-10T17:58:26Z)
• MoodAngels: A Retrieval-augmented Multi-agent Framework for Psychiatry Diagnosis [58.67342568632529]
  MoodAngels is the first specialized multi-agent framework for mood disorder diagnosis.<n>MoodSyn is an open-source dataset of 1,173 synthetic psychiatric cases.
  arXiv  Detail & Related papers  (2025-06-04T09:18:25Z)
• Beyond Empathy: Integrating Diagnostic and Therapeutic Reasoning with Large Language Models for Mental Health Counseling [50.83055329849865]
  PsyLLM is a large language model designed to integrate diagnostic and therapeutic reasoning for mental health counseling.<n>It processes real-world mental health posts from Reddit and generates multi-turn dialogue structures.<n>Our experiments demonstrate that PsyLLM significantly outperforms state-of-the-art baseline models.
  arXiv  Detail & Related papers  (2025-05-21T16:24:49Z)
• Ψ-Arena: Interactive Assessment and Optimization of LLM-based Psychological Counselors with Tripartite Feedback [51.26493826461026]
  We propose Psi-Arena, an interactive framework for comprehensive assessment and optimization of large language models (LLMs)<n>Arena features realistic arena interactions that simulate real-world counseling through multi-stage dialogues with psychologically profiled NPC clients.<n>Experiments across eight state-of-the-art LLMs show significant performance variations in different real-world scenarios and evaluation perspectives.
  arXiv  Detail & Related papers  (2025-05-06T08:22:51Z)
• Quantifying the Reasoning Abilities of LLMs on Real-world Clinical Cases [48.87360916431396]
  We introduce MedR-Bench, a benchmarking dataset of 1,453 structured patient cases, annotated with reasoning references.<n>We propose a framework encompassing three critical examination recommendation, diagnostic decision-making, and treatment planning, simulating the entire patient care journey.<n>Using this benchmark, we evaluate five state-of-the-art reasoning LLMs, including DeepSeek-R1, OpenAI-o3-mini, and Gemini-2.0-Flash Thinking, etc.
  arXiv  Detail & Related papers  (2025-03-06T18:35:39Z)
• Structured Outputs Enable General-Purpose LLMs to be Medical Experts [50.02627258858336]
  Large language models (LLMs) often struggle with open-ended medical questions.<n>We propose a novel approach utilizing structured medical reasoning.<n>Our approach achieves the highest Factuality Score of 85.8, surpassing fine-tuned models.
  arXiv  Detail & Related papers  (2025-03-05T05:24:55Z)
• Moving Beyond Medical Exam Questions: A Clinician-Annotated Dataset of Real-World Tasks and Ambiguity in Mental Healthcare [0.0545520830707066]
  We present an expert-created and annotated dataset spanning five critical domains of decision-making in mental healthcare.<n>This dataset is designed to capture the nuanced clinical reasoning and daily ambiguities mental health practitioners encounter.
  arXiv  Detail & Related papers  (2025-02-22T03:10:16Z)
• AutoCBT: An Autonomous Multi-agent Framework for Cognitive Behavioral Therapy in Psychological Counseling [57.054489290192535]
  Traditional in-person psychological counseling remains primarily niche, often chosen by individuals with psychological issues.<n>Online automated counseling offers a potential solution for those hesitant to seek help due to feelings of shame.
  arXiv  Detail & Related papers  (2025-01-16T09:57:12Z)
• Do Large Language Models Align with Core Mental Health Counseling Competencies? [19.375161727597536]
  Large Language Models (LLMs) are a promising solution to the global shortage of mental health professionals.<n>We introduce CounselingBench, a novel NCMHCE-based benchmark evaluating 22 general-purpose and medical-finetuned LLMs.<n>Our results underscore the need for specialized, fine-tuned models aligned with core mental health counseling competencies.
  arXiv  Detail & Related papers  (2024-10-29T18:27:11Z)
• Interactive Agents: Simulating Counselor-Client Psychological Counseling via Role-Playing LLM-to-LLM Interactions [12.455050661682051]
  We propose a framework that employs two large language models (LLMs) via role-playing for simulating counselor-client interactions. Our framework involves two LLMs, one acting as a client equipped with a specific and real-life user profile and the other playing the role of an experienced counselor.
  arXiv  Detail & Related papers  (2024-08-28T13:29:59Z)
• LLM Questionnaire Completion for Automatic Psychiatric Assessment [49.1574468325115]
  We employ a Large Language Model (LLM) to convert unstructured psychological interviews into structured questionnaires spanning various psychiatric and personality domains. The obtained answers are coded as features, which are used to predict standardized psychiatric measures of depression (PHQ-8) and PTSD (PCL-C)
  arXiv  Detail & Related papers  (2024-06-09T09:03:11Z)
• Exploring the Efficacy of Large Language Models in Summarizing Mental Health Counseling Sessions: A Benchmark Study [17.32433545370711]
  Comprehensive summaries of sessions enable an effective continuity in mental health counseling. Manual summarization presents a significant challenge, diverting experts' attention from the core counseling process. This study evaluates the effectiveness of state-of-the-art Large Language Models (LLMs) in selectively summarizing various components of therapy sessions.
  arXiv  Detail & Related papers  (2024-02-29T11:29:47Z)
• Addressing cognitive bias in medical language models [25.58126133789956]
  BiasMedQA is a benchmark for evaluating cognitive biases in large language models (LLMs) applied to medical tasks. We tested six models on 1,273 questions from the US Medical Licensing Exam (USMLE) Steps 1, 2, and 3. GPT-4 stood out for its resilience to bias, in contrast to Llama 2 70B-chat and PMC Llama 13B, which were disproportionately affected by cognitive bias.
  arXiv  Detail & Related papers  (2024-02-12T23:08:37Z)
• A Computational Framework for Behavioral Assessment of LLM Therapists [7.665475687919995]
  Large language models (LLMs) like ChatGPT have increased interest in their use as therapists to address mental health challenges.<n>We propose BOLT, a proof-of-concept computational framework to systematically assess the conversational behavior of LLM therapists.
  arXiv  Detail & Related papers  (2024-01-01T17:32:28Z)
• ExpertQA: Expert-Curated Questions and Attributed Answers [51.68314045809179]
  We conduct human evaluation of responses from a few representative systems along various axes of attribution and factuality. We collect expert-curated questions from 484 participants across 32 fields of study, and then ask the same experts to evaluate generated responses to their own questions. The output of our analysis is ExpertQA, a high-quality long-form QA dataset with 2177 questions spanning 32 fields, along with verified answers and attributions for claims in the answers.
  arXiv  Detail & Related papers  (2023-09-14T16:54:34Z)
• Can ChatGPT Assess Human Personalities? A General Evaluation Framework [70.90142717649785]
  Large Language Models (LLMs) have produced impressive results in various areas, but their potential human-like psychology is still largely unexplored. This paper presents a generic evaluation framework for LLMs to assess human personalities based on Myers Briggs Type Indicator (MBTI) tests.
  arXiv  Detail & Related papers  (2023-03-01T06:16:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.