This paper proposes a model of how uncertainty and ambiguity are represented in neural activity, and validate on hippocampal CA1 recordings, collected on mice being exposed to different environments, It’s well-motivated and sure to be of interest to the NeurIPS neuroscience community. It appears to be well-written (R1, R2), contains excellent figures (R2, R3), and overall the review of prior relevant work is outstanding (R4), although R1 did bring up some missing prior work. The main novel contribution is the interesting theoretical framework of incorporating multiple representations for the same environment within a decoding model (R1, R4). However, R4 wondered about the wider theoretical implications for cognition in general, and whether it offers new insights beyond simply explaining the data. R1 raised some concerns about the correctness of the claims, given that the evidence for rapid fluctuations (one of the claims) is only apparent in single neurons, which can have low SNR, and not at the population level (although their other concerns were addressed and they raised score from 5 to 6). Despite several criticisms raised in review, the AC and SAC found the proposed methods to be highly innovative and likely to be of major interest to the computational neuroscience community focused on neural coding. They agreed that the paper makes a significant contribution, and recommend it for acceptance as a poster presentation.