**Summary:**

The paper introduces a multimodal contrastive learning method for neurophysiological data, NEMO, which can be pre-trained using large amounts of unlabeled paired data and fine-tuned for different downstream tasks such as cell-type and brain region classification. NEMO utilizes a contrastive learning framework to jointly embed individual neurons' activity autocorrelations and average extracellular waveforms, aiming to capture shared information while discarding modality-specific noise. The method is validated on three different datasets: an opto-tagged mouse visual cortex dataset, a mouse cerebellum dataset, and an IBL Brain-wide Map dataset, demonstrating its superiority over current unsupervised and supervised methods, especially in label-limited regimes.

**Strengths:**

- **Innovative Approach:** The use of multimodal contrastive learning for neurophysiological data is innovative, aiming to integrate information from both recorded EAPs and spiking activity.
- **Performance:** NEMO outperforms current unsupervised (PhysMAP and VAEs) and supervised methods in cell-type and brain region classification tasks.
- **Flexibility:** The method is adaptable for different datasets and tasks, and can be fine-tuned for improved performance.
- **Label Efficiency:** NEMO shows promise in requiring less labeled data for fine-tuning, which is particularly valuable in neuroscience studies where ground truth data are costly or difficult to obtain.

**Weaknesses:**

- **Complexity of Application:** The method might require substantial computational resources and expertise to implement and fine-tune.
- **Assumption of Shared Information:** The effectiveness of NEMO relies on the assumption that the shared information between different modalities is most informative for cell identity or anatomical location. This might not always be the case, as each modality might contain unique features relevant to cell identity or location.
- **Independence of Neurons:** The method treats neurons independently, ignoring the potential benefits of encoding population-level features that can help in distinguishing cell types.

**Questions:**

- How does the multimodal contrastive learning framework specifically improve the classification performance compared to using modalities in isolation?
- What are the limitations of treating neurons independently, and how might these be addressed in future iterations of the method?
- How robust is NEMO to variations in data quality and quantity across different datasets, and what are the implications for its generalizability?

**Soundness:**
**Presentation:**
**Contribution:**
Given the novelty of the approach, the clear methodology, the validation on multiple datasets, and the demonstrated improvements over existing methods, the paper presents a sound contribution to the field of neurophysiological data analysis.

**Rating:**
Considering the innovative approach, the method's performance, and the potential for practical application in neuroscience research, the paper is rated as "strongly acceptable" with the potential for high impact in the field. However, the method's limitations and the questions raised above suggest that further research is needed to fully understand its performance across different scenarios and to address the potential benefits of modeling both shared and modality-specific information.

**Paper Decision:**
- **Decision:** Accept
- **Reasons:** The paper presents a novel method with significant potential for advancing the field of cell-type and brain region classification in neuroscience. While there are areas for improvement and further exploration, the method's demonstrated performance and flexibility justify its acceptance for publication, with encouragement for future work to address identified limitations and to explore broader applications.