Pith. sign in

REVIEW 4 major objections 5 minor 71 references

SCoPE: Shift-Aware Speaker-Conditioned Priors for Emotion Recognition in Conversations

T0 review · 4 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash

Pith's one-line read SCoPE claims that explicitly modeling each speaker's emotional history as a prior, gated by predicted emotion shifts, improves multimodal emotion recognition in conversations and achieves state-of-the-art results on IEMOCAP.

desk verdict SCoPE is a credible integration of shift-gated speaker priors; its IEMOCAP SOTA claim needs a proper held-out validation protocol before the 0.78-point margin can be trusted. read the letter →

arxiv 2607.20445 v1 pith:3AA43FVZ submitted 2026-05-13 cs.CL cs.SDeess.AS

classification cs.CLcs.SDeess.AS
keywords emotionrecognitioninconversationsmultimodalspeaker-conditionedpriorsshiftpredictiontemporaldynamicsGRUproduct-of-expertsIEMOCAP
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that conversational emotion recognition gets better when the model explicitly tracks each speaker's emotional inertia rather than treating every utterance as an isolated event. It introduces a lightweight GRU module, SCoPE, that generates speaker-conditioned priors from the emotional history, and a shift-prediction head that decides how much to trust those priors versus the current multimodal evidence. On the IEMOCAP benchmark the complete model reaches a weighted F1 of 74.68, beating the previous best reported system and its own strong baseline; on MELD it improves over the baseline and does especially well on rare emotion labels. If correct, this means that cheap, speaker-aware temporal memory is a valuable complement to heavy multimodal encoders, and that emotion-shift prediction can act as a dynamic control signal rather than just an auxiliary loss.

What carries the argument

The core mechanism is the shift-aware fusion equation z_post = z_evi + β·(1−p_shift)·z_pri, which adds multimodal evidence logits to speaker-prior logits scaled by the predicted probability that no emotion shift occurs. The prior z_pri is produced by a GRU that consumes only the previous posterior belief and the current speaker embedding, so it never sees the current utterance's evidence; the shift probability comes from a separate classification head trained on binary emotion-shift labels. This lets the model lean on history when continuity is likely and switch to raw evidence when a shift is predicted.

What would settle it

On a version of IEMOCAP where the utterance order within each dialogue is randomly permuted—destroying genuine emotional continuity while preserving all features—SCoPE's weighted F1 should drop back to approximately the SDT baseline of 72.02. If it stays near 74.68, the gains are coming from something other than the speaker-conditioned temporal history.

Watch

Extended reading notes

Core claim

The paper claims that a lightweight GRU module—SCoPE—which tracks each speaker's emotional history as an explicit prior, combined with a shift predictor that estimates whether an emotion change is about to occur, yields more accurate emotion predictions than using multimodal evidence alone. On IEMOCAP's six-way task the full model reaches an average weighted F1 of 74.68±0.46, surpassing the previous best reported system (GS-MCC, 73.90) and clearly beating its own SDT baseline (72.02). On MELD the model does not beat the best published system but consistently improves over its baseline and performs best on the rare Fear and Disgust classes.

Load-bearing premise

The fusion rule adds the evidence logits and the prior logits with a fixed scalar β, implicitly treating them as independent sources of information; both are learned from the same dialogues and labels, so if they are correlated the combination double-counts and the reported gains may not transfer to other datasets.

Editorial extensions

If this is right

  • The SCoPE module improves over the SDT baseline on both IEMOCAP and MELD, with the largest gains on rare emotion labels like Fear and Disgust on MELD.
  • Using emotion shift as a control signal—rather than only as an auxiliary classification target—improves emotion prediction accuracy.
  • The whole architecture runs at about 0.2 ms per utterance and needs only about 3.8 GB of GPU memory, making it plausible for real-time human-robot interaction and on-device affective computing.
  • The prior module is lightweight enough to be added to existing multimodal evidence encoders without major architectural changes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The additive fusion in Eq. 13 is effectively log-linear interpolation; the 'Bayesian-inspired' framing hints that a principled derivation from per-source uncertainty estimates could replace the tuned scalar β and transfer more cleanly to new datasets.
  • The shift-gating idea—using a predicted discontinuity to balance memory versus new evidence—is a general temporal-reasoning module that could be tested in other sequence-prediction tasks such as dialogue state tracking or sentiment over time.
  • The model's weaker showing on MELD, where shifts are frequent and the optimal β is lower, suggests the prior helps most in relatively stable dialogues; extending the prior to long-term speaker persona might help in chaotic, multi-party settings where emotional history is less predictive.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper introduces SCoPE, a lightweight auxiliary module for multimodal emotion recognition in conversations. SCoPE augments an SDT baseline with (i) a GRU-based speaker-conditioned prior that is fed only the previous posterior belief and the current speaker identity, (ii) a binary emotion-shift predictor whose output is converted into a memory-confidence weight, and (iii) a shift-aware fusion z_post = z_evi + β·α·z_pri that combines multimodal evidence logits with prior logits. The authors evaluate on IEMOCAP and MELD, reporting a mean weighted F1 of 74.68±0.46 on IEMOCAP, which they claim is superior to the prior state-of-the-art GS-MCC (73.90), and a MELD WF1 of 65.65, which is below several published systems. The paper also reports per-emotion scores, ablations over β, and inference latency of 0.20–0.25 ms/utterance.

Significance. If the headline empirical claim is robust, SCoPE is a useful, lightweight contribution: it consistently improves over its SDT baseline on both datasets, it is evaluated with 10-run means and standard deviations, and the per-emotion and latency analyses are informative. The psychological motivation is well grounded. However, the central 'superior performance on IEMOCAP' claim currently rests on an evaluation protocol that may involve test-set selection of hyperparameters and an unpaired comparison with a single published number. The fusion mechanism is also described as 'Bayesian-inspired product-of-experts' without a derivation, and the MELD results undercut the broader wording in the contribution list. The architecture is promising, but the evidence as presented is not yet sufficient to establish the headline claim.

major comments (4)
  1. [§4.2, §5.2, Fig. 6, Table 2] The evaluation protocol for hyperparameters, especially β, is under-specified and is load-bearing for the SOTA claim. §4.2 says Optuna was used 'to have our final values,' and §5.2 says that for different β values the authors 'saved the best models across the runs.' The paper does not state whether β, Optuna hyperparameters, or the 'best models' were selected on a held-out validation split or on the same IEMOCAP test set used in Table 2. If the latter, the reported 74.68 (or 75.82 for the best run) is the maximum over multiple test-set evaluations, not an unbiased estimate of a fixed model. In addition, the comparison with GS-MCC is against a single published number with no significance test; the 0.78 WF1 gap is less than 1.7 standard errors of SCoPE's own mean, so it may be within sampling variation. The authors should describe the exact model-selection protocol and, ideally, report val
  2. [§5.1 and Table 2] The narrative mixes the mean and the best-run results in a way that overstates the result. The text says 'SCoPE achieved 75.72% Accuracy and 75.82% WF1' and then says 'Both results demonstrate SCoPE's superior performance' — but Table 2 clearly labels the 75.82 WF1 as the best of ten runs (†), while the mean is 74.68±0.46. The claim 'superior performance' is supported, if at all, only by the mean; the best run is a selected maximum and should not be used as the primary evidence. The authors should report the mean as the headline number and treat the best run as complementary, or provide a validation-based model-selection procedure that justifies reporting the best run.
  3. [§3.5, Eq. (13)] Equation (13) is described as a 'Bayesian-inspired product-of-experts formulation,' but no probabilistic derivation is given. The formula z_post = z_evi + β·α·z_pri assumes that the evidence and prior logits can be combined by fixed weighted addition, which is exact only if the two sources are conditionally independent and β·α encodes the relative precision (inverse variance) of the prior. Here both logits are trained on the same ground-truth labels, and the prior is produced from the model's own previous posterior, so the independence assumption is questionable. Moreover, β is tuned per dataset (1.5 for IEMOCAP, 1.0 for MELD), which further weakens the 'Bayesian' interpretation. The authors should either provide a proper derivation (e.g., treating each logit as a Gaussian with distinct variance) or describe Eq. (13) as a heuristic weighted sum rather than a product-of-experts. This is n
  4. [Abstract and Contribution list vs. §5.1, Table 3] The paper claims 'superior performance over recent state-of-the-art models on the IEMOCAP dataset' and, in the contribution list, 'superior performance on popular ERC benchmarks.' The MELD results do not support the plural 'benchmarks': in Table 3, SCoPE's WF1 (65.65) is below AdaIGN (66.79), CFN-ESA (67.42), GS-MCC (69.00), and SEDC (66.16). The text does acknowledge 'it does not perform as the best model' on MELD, but the contribution wording and the general framing are misleading. The authors should restrict the 'superior' claim to IEMOCAP and explicitly discuss the MELD shortfall in the conclusion, or provide a statistically supported reason why the MELD comparison is not commensurable.
minor comments (5)
  1. [§3.1] Typo: 'Each utterance u i is has a speaker embedding' should be 'Each utterance u_i has a speaker embedding.'
  2. [Table 2 and Table 3] In the SCoPE rows, the mean and best-run values are run together in the same row (e.g., '66.00±1.083.00±1.1...') with a dagger marker only at the end. This is very hard to read; please split into two rows or use clearer column separation (e.g., separate 'mean' and 'best' blocks).
  3. [§5.1] Use consistent decimal notation: '75,82%' should be '75.82%.' Also 'noticable' is a typo.
  4. [Data Availability] The statement 'The authors declare that this manuscript does not have data generation or analysis' is confusing, since the paper analyzes IEMOCAP and MELD. Presumably the authors mean that no new data are introduced; please rephrase.
  5. [§5.2 and Figure 6] Figure 6 shows WF1 across β values, but the text does not specify how many runs were averaged per β point or whether error bars are provided. Adding mean±std and the number of runs would make the ablation more informative.

Circularity Check

1 steps flagged · score 4.0 of 10

No equation-level circularity, but the IEMOCAP SOTA claim may be partly a fitted result: β was tuned by observing WF1 across runs, and no held-out validation protocol is stated.

  1. fitted input called prediction [§4.2 Training Configuration; §5.2 Ablation Study; Table 2]
    "We used the Optuna hyperparameter optimization tool to sweep over those parameters to have our final values. ... We performed an additional ablation study in which we tested how different scaling factors (ß) of the shift-aware fusion affect the model performance. We conducted multiple train runs with different ß values, and saved the best models across the runs."

    The headline claim is empirical: SCoPE achieves superior performance on IEMOCAP with mean WF1 74.68 vs GS-MCC 73.90. β is one of the hyperparameters selected via Optuna (Table 1: β=1.5 for IEMOCAP), and the ablation reports WF1 over β values while saving the best models. The paper never states that β was chosen on a held-out validation split rather than on the same IEMOCAP test set used in Table 2. If β was selected by observing test WF1, the reported 74.68 is the best among evaluated β settings on that test set, so the superiority margin is a selected maximum rather than an unbiased prediction of a fixed model.

full rationale

The paper's derivation chain is mostly self-contained: SCoPE, the shift predictor, and the fusion rule are defined by explicit equations, and the comparison against baseline SDT is an empirical evaluation, not a derivation that reduces a predicted quantity to its input by construction. Training the prior on ground-truth labels is normal supervised learning, not circularity. There is no load-bearing self-citation, no imported uniqueness theorem, and no renaming of a known result as a new derivation. The only concrete circularity concern is the fitted hyperparameter β: the central IEMOCAP performance claim could be an artifact of test-set-selected β, but the paper does not provide enough protocol detail to prove that this occurred. This is benchmark tuning rather than an equation-level identity, so it warrants a partial score rather than a high one.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

All listed free parameters are fitted or tuned on the benchmark itself. The axioms include domain assumptions imported from psychology and two ad hoc architectural assumptions: sufficiency of the GRU prior and independence of the two logit sources. No physical entities are invented; the 'prior' is a trained latent module.

free parameters (4)
  • beta fusion scale = 1.5 (IEMOCAP), 1.0 (MELD)
    Directly controls the contribution of the prior in Eq. 13; tuned per dataset via ablation (Fig. 6) and reported in Table 1.
  • shift loss weight = 0.3
    Hand-chosen weight for the auxiliary emotion shift cross-entropy loss in Section 4.3.
  • prior supervision loss weight = 0.2
    Hand-chosen weight for the SCoPE prior loss in Eq. 16.
  • Optuna-selected hyperparameters = learning rate, batch size, dropout, hidden sizes
    Selected per dataset with Optuna on the benchmark; final scores depend on these fitted values, and the search ranges are not reported.
assumptions (4)
  • domain assumption Emotions persist across utterances and evolve smoothly; emotional inertia is speaker-specific.
    Cited psychological literature (Section 2.5) motivates SCoPE, but the paper does not validate this assumption on the benchmark data; the GRU prior encodes it.
  • domain assumption Binary emotion-shift labels can be derived as a change in the ground-truth emotion label between consecutive same-speaker utterances.
    Defined in Section 4.3 and used to supervise the shift head; assumes label changes capture meaningful emotional shifts.
  • ad hoc to paper The prior can be generated from only the previous posterior and current speaker identity (Eq. 11), without current utterance evidence.
    Architectural choice in Section 3.4; no justification that this information is sufficient.
  • ad hoc to paper Evidence and prior logits are conditionally independent and can be combined by weighted logit addition (Eq. 13).
    The 'Bayesian-inspired product-of-experts' fusion is asserted, not derived; the two sources share training labels and dialogue context, so independence is questionable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SCoPE: Shift-Aware Speaker-Conditioned Priors for Emotion Recognition in Conversations." pith.science (2026). https://pith.science/paper/3AA43FVZ

@misc{pith2026260720445,
  author       = {Pith},
  title        = {Pith review of: SCoPE: Shift-Aware Speaker-Conditioned Priors for Emotion Recognition in Conversations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3AA43FVZ}},
  note         = {Machine review of arXiv:2607.20445}
}
read the original abstract

In conversations, human emotions are transient; however, they tend to persist across multiple utterances. For example, we rarely switch instantly between contrasting emotions such as happiness and anger. Instead, emotions tend to evolve smoothly, and these patterns are often speaker-specific. Some people might escalate, while others gradually cool down over time. Furthermore, when emotions change during a conversation, they are often driven by contextual factors, such as newly received information or unexpected events. Even though progress has been made in Emotion Recognition in Conversations (ERC), most existing approaches still rely heavily on overt evidence and do not sufficiently model these non-apparent factors. Especially in multimodal settings, this makes these models fragile when the signals are noisy (e.g., occluded faces, slang expressions, or microphone noise). To address these limitations, we introduce Speaker-Conditioned Priors over Emotions (SCoPE). SCoPE is a light weight module that utilizes the emotional history of each speaker and explicitly models their priors for use in subsequent emotion classification. Second, we incorporate emotion shift prediction, a well-established concept in ERC, to guide the model in balancing the priors from SCoPE and multimodal evidence. Finally, we propose a shift-aware fusion mechanism that performs precision-weighted logit integration between multimodal evidence and the speaker prior, forming a Bayesian-inspired product-of-experts formulation. This dynamic fusion allows the model to rely on historical priors when emotions persist and to prioritize multimodal evidence when shifts are likely. Experimental results show our model achieves superior performance over recent state-of-the-art models on the IEMOCAP dataset in multimodal settings.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

71 extracted references · 19 canonical work pages

  1. [1]

    IEEE Access7, 100943–100953 (2019) https://doi.org/10.1109/ACCESS.2019.2929050

    Poria, S., Majumder, N., Mihalcea, R., Hovy, E.: Emotion recognition in con- versation: Research challenges, datasets, and recent advances. IEEE Access7, 100943–100953 (2019) https://doi.org/10.1109/ACCESS.2019.2929050

  2. [2]

    In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp

    Poria, S., Hazarika, D., Majumder, N., Naik, G., Cambria, E., Mihalcea, R.: Meld: A multimodal multi-party dataset for emotion recognition in conversations. In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp. 527–536 (2019). https://doi.org/10.18653/v1/P19-1050

  3. [3]

    MIT Press, Cambridge, MA (2000)

    Picard, R.W.: Affective Computing. MIT Press, Cambridge, MA (2000). https: //doi.org/10.7551/mitpress/1140.001.0001

  4. [4]

    International Journal of Human-Computer Studies59(1), 119–155 (2003) https://doi.org/10.1016/ S1071-5819(03)00018-1

    Breazeal, C.: Emotion and sociable humanoid robots. International Journal of Human-Computer Studies59(1), 119–155 (2003) https://doi.org/10.1016/ S1071-5819(03)00018-1

  5. [5]

    IEEE Transactions on Affective Computing1(1), 18–37 (2010) https://doi.org/10.1109/T-AFFC.2010.1

    Calvo, R.A., D’Mello, S.: Affect detection: An interdisciplinary review of models, methods, and their applications. IEEE Transactions on Affective Computing1(1), 18–37 (2010) https://doi.org/10.1109/T-AFFC.2010.1

  6. [6]

    ACM Computing Surveys47(3), 43 (2015) https://doi.org/10.1145/ 2682899

    D’Mello, S.K., Kory, J.: A review and meta-analysis of multimodal affect detec- tion systems. ACM Computing Surveys47(3), 43 (2015) https://doi.org/10.1145/ 2682899

  7. [7]

    In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pp

    Zhong, P., Wang, D., Miao, C.: Knowledge-enriched transformer for emotion detection in textual conversations. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pp. 165–176 (2019). https: //doi.org/10.18653/v1/D19-1016

  8. [8]

    In: Proceedings of the 2019 Conference on Empirical Methods in Natural Lan- guage Processing and the 9th International Joint Conference on Natural Language Processing, pp

    Ghosal, D., Majumder, N., Poria, S., Chhaya, N., Gelbukh, A.: Dialoguegcn: A graph convolutional neural network for emotion recognition in conversation. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Lan- guage Processing and the 9th International Joint Conference on Natural Language Processing, pp. 154–164 (2019). https://doi.org/...

Show all 71 references
  1. [9]

    In: Findings of the Association for Computational Linguistics: EMNLP, pp

    Ghosal, D., Majumder, N., Gelbukh, A., Mihalcea, R., Poria, S.: Cosmic: Com- monsense knowledge for emotion identification in conversations. In: Findings of the Association for Computational Linguistics: EMNLP, pp. 2470–2481 (2020). https://doi.org/10.18653/v1/2020.findings-emnlp.224

  2. [10]

    Proceedings of the AAAI Conference on Artificial Intelligence33(1), 6818–6825 (2019) https: //doi.org/10.1609/aaai.v33i01.33016818 18

    Majumder, N., Poria, S., Hazarika, D., Mihalcea, R., Gelbukh, A., Cambria, E.: Dialoguernn: An attentive rnn for emotion detection in conversations. Proceedings of the AAAI Conference on Artificial Intelligence33(1), 6818–6825 (2019) https: //doi.org/10.1609/aaai.v33i01.33016818 18

  3. [11]

    In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pp

    Hu, D., Wei, L., Huai, X.: Dialoguecrn: Contextual reasoning networks for emo- tion recognition in conversations. In: Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Process...

  4. [12]

    In: Proceedings of the 28th ACM International Conference on Multimedia, pp

    Hazarika, D., Zimmermann, R., Poria, S.: Misa: Modality-invariant and -specific representations for multimodal sentiment analysis. In: Proceedings of the 28th ACM International Conference on Multimedia, pp. 1122–1131 (2020). https:// doi.org/10.1145/3394171.3413678

  5. [13]

    IEEE Transactions on Pattern Analysis and Machine Intel- ligence41(2), 423–443 (2019) https://doi.org/10.1109/TPAMI.2018.2798607

    Baltruˇ saitis, T., Ahuja, C., Morency, L.-P.: Multimodal machine learning: A survey and taxonomy. IEEE Transactions on Pattern Analysis and Machine Intel- ligence41(2), 423–443 (2019) https://doi.org/10.1109/TPAMI.2018.2798607

  6. [14]

    Psychological Science21(7), 984–991 (2010) https://doi.org/10

    Kuppens, P., Allen, N.B., Sheeber, L.B.: Emotional inertia and psychological maladjustment. Psychological Science21(7), 984–991 (2010) https://doi.org/10. 1177/0956797610372634

  7. [16]

    Journal of Personality and Social Psy- chology80(6), 1011–1027 (2001) https://doi.org/10.1037/0022-3514.80.6.1011

    Fleeson, W.: Toward a structure-and process-integrated view of personality: Traits as density distributions of states. Journal of Personality and Social Psy- chology80(6), 1011–1027 (2001) https://doi.org/10.1037/0022-3514.80.6.1011

  8. [17]

    Psychological Bulletin128(6), 934–960 (2002) https://doi.org/10.1037/0033-2909.128.6.934

    Robinson, M.D., Clore, G.L.: Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin128(6), 934–960 (2002) https://doi.org/10.1037/0033-2909.128.6.934

  9. [18]

    In: Change in Emotion and Mental Health, pp

    Koval, P., Kuppens, P.: Changing feelings: Individual differences in emotional inertia. In: Change in Emotion and Mental Health, pp. 3–21 (2024). https://doi. org/10.1016/B978-0-323-95604-8.00007-1

  10. [19]

    Social Cognitive and Affective Neuroscience12(9), 1448–1459 (2017) https://doi.org/10.1093/scan/nsx071

    Waugh, C.E., Shing, E.Z., Avery, B.M., Jung, Y., Whitlow, C.T., Maldjian, J.A.: Neural predictors of emotional inertia in daily life. Social Cognitive and Affective Neuroscience12(9), 1448–1459 (2017) https://doi.org/10.1093/scan/nsx071

  11. [20]

    Emotion Review7(4), 355–361 (2015) https://doi.org/10.1177/ 1754073915590617

    Trull, T.J., Lane, S.P., Koval, P., Ebner-Priemer, U.W.: Affective dynamics in psychopathology. Emotion Review7(4), 355–361 (2015) https://doi.org/10.1177/ 1754073915590617

  12. [21]

    Cognition and Emotion23(7), 1307–1351 (2009) https://doi.org/ 10.1080/02699930902928969 19

    Scherer, K.R.: The dynamic architecture of emotion: Evidence for the component process model. Cognition and Emotion23(7), 1307–1351 (2009) https://doi.org/ 10.1080/02699930902928969 19

  13. [22]

    Emotion25(8), 1944–1957 (2025) https://doi.org/10.1037/emo0001541

    Bagnara, L., Moeck, E.K., Kuppens, P., Bianchi, V., Koval, P.: Why do feel- ings persist over time in daily life? investigating the role of emotion-regulation strategies in the process underlying emotional inertia. Emotion25(8), 1944–1957 (2025) https://doi.org/10.1037/emo0001541

  14. [23]

    Emotion (2026) https://doi.org/10.1037/emo0001630

    Shao, S., Ong, A.D.: Seven challenges in affective inertia research. Emotion (2026) https://doi.org/10.1037/emo0001630

  15. [24]

    https://doi.org/10.1093/oso/ 9780195069945.001.0001

    Lazarus, R.S.: Emotion and Adaptation, (1991). https://doi.org/10.1093/oso/ 9780195069945.001.0001

  16. [25]

    https://doi.org/10.1093/oso/9780195130072.001

    Scherer, K.R., Schorr, A., Johnstone, T.: Appraisal Processes in Emotion: The- ory, Methods, Research, (2001). https://doi.org/10.1093/oso/9780195130072.001. 0001

  17. [26]

    In: Proceedings of the Pacific Asia Conference on Language, Information and Computation, pp

    Wang, J., Mine, T.: Multi-task learning for emotion recognition in conversation with emotion shift. In: Proceedings of the Pacific Asia Conference on Language, Information and Computation, pp. 257–266 (2023)

  18. [27]

    Friston, K.: The free-energy principle: A unified brain theory? Nature Reviews Neuroscience11(2), 127–138 (2010) https://doi.org/10.1038/nrn2787

  19. [28]

    Social Cognitive and Affective Neuroscience 12(1), 1–23 (2016) https://doi.org/10.1093/scan/nsw154

    Barrett, L.F.: The theory of constructed emotion: An active inference account of interoception and categorization. Social Cognitive and Affective Neuroscience 12(1), 1–23 (2016) https://doi.org/10.1093/scan/nsw154

  20. [29]

    In: The Value of Emotions for Knowledge, pp

    Wilkinson, S., Deane, G., Nave, K., Clark, A.: Getting warmer: Predictive pro- cessing and the nature of emotion. In: The Value of Emotions for Knowledge, pp. 101–119 (2019). https://doi.org/10.1007/978-3-030-15667-1 5

  21. [30]

    Kube, T., Kirchner, L., Lemmer, G., Glombiewski, J.A.: How the discrepancy between prior expectations and new information influences expectation updating in depression—the greater, the better? Clinical Psychological Science10(3), 430– 449 (2022) https://doi.org/10.1177/2167702...

  22. [31]

    In: Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pp

    Poria, S., Cambria, E., Hazarika, D., Majumder, N., Zadeh, A., Morency, L.-P.: Context-dependent sentiment analysis in user-generated videos. In: Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pp. 873–883 (2017). https://doi.org/10.186...

  23. [32]

    In: Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp

    Hazarika, D., Poria, S., Mihalcea, R., Cambria, E., Zimmermann, R.: Icon: Inter- active conversational memory network for multimodal emotion detection. In: Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 2594–2604 (2018). https://doi...

  24. [33]

    In: Proceedings of 20 the 28th International Conference on Computational Linguistics, pp

    Li, J., Ji, D., Li, F., Zhang, M., Liu, Y.: Hitrans: A transformer-based context- and speaker-sensitive model for emotion detection in conversations. In: Proceedings of 20 the 28th International Conference on Computational Linguistics, pp. 4190–4200 (2020). https://doi.org/10....

  25. [34]

    IEEE Transactions on Multimedia26, 776–788 (2024) https://doi.org/10.1109/ TMM.2023.3271019

    Ma, H., Wang, J., Lin, H., Zhang, B., Zhang, Y., Xu, B.: A transformer-based model with self-distillation for multimodal emotion recognition in conversations. IEEE Transactions on Multimedia26, 776–788 (2024) https://doi.org/10.1109/ TMM.2023.3271019

  26. [35]

    Proceedings of the AAAI Conference on Artificial Intelligence38(17), 19089–19097 (2024) https://doi.org/10.1609/aaai

    Tu, G., Xie, T., Liang, B., Wang, H., Xu, R.: Adaptive graph learning for multi- modal conversational emotion detection. Proceedings of the AAAI Conference on Artificial Intelligence38(17), 19089–19097 (2024) https://doi.org/10.1609/aaai. v38i17.29876

  27. [36]

    arXiv, 2309–11911 (2023) https://doi.org/10.48550/arXiv.2309

    Lei, S., Dong, G., Wang, X., Wang, K., Qiao, R., Wang, S.: Instructerc: Reforming emotion recognition in conversation with multi-task retrieval-augmented large language models. arXiv, 2309–11911 (2023) https://doi.org/10.48550/arXiv.2309. 11911

  28. [37]

    Cognitive Computation18, 14 (2026) https://doi.org/10.1007/s12559-026-10553-2

    Wu, S., Li, W., Lin, J., Jiang, D.: Reinforcement learning-driven adaptive emotion-cause analysis in conversation. Cognitive Computation18, 14 (2026) https://doi.org/10.1007/s12559-026-10553-2

  29. [38]

    In: Proceedings of the 31st International Conference on Computational Linguistics, pp

    Fu, Y., Wu, J., Wang, Z., Zhang, M., Shan, L., Wu, Y., Liu, B.: Laerc-s: Improving llm-based emotion recognition in conversation with speaker characteristics. In: Proceedings of the 31st International Conference on Computational Linguistics, pp. 6748–6761 (2025)

  30. [39]

    In: Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp

    Lee, J., Lee, W.: Compm: Context modeling with speaker’s pre-trained memory tracking for emotion recognition in conversation. In: Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp....

  31. [40]

    In: Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp

    Song, X., Huang, L., Xue, H., Hu, S.: Supervised prototypical contrastive learning for emotion recognition in conversation. In: Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp. 5197–5206 (2022). https://doi.org/10.18653/v1/2022.emnlp-main.347

  32. [41]

    In: Findings of the Association for Computational Linguistics: EMNLP, pp

    Wu, C., Cai, Y., Liu, Y., Zhu, P., Xue, Y., Gong, Z., Hirschberg, J., Ma, B.: Multimodal emotion recognition in conversations: A survey of methods, trends, challenges and prospects. In: Findings of the Association for Computational Linguistics: EMNLP, pp. 6257–6274 (2025). htt...

  33. [42]

    In: International Conference on Artificial Neural Networks, pp

    Xue, J., Nguyen, M.-P., Matheny, B., Nguyen, L.-M.: Bioserc: Integrating biog- raphy speakers supported by llms for erc tasks. In: International Conference on Artificial Neural Networks, pp. 277–292 (2024). https://doi.org/10.1007/ 21 978-3-031-72344-5 19

  34. [43]

    Kaplan, B.C., Carneiro, H.C.D.C., Wermter, S.: Can large language models generate effective datasets for emotion recognition in conversations? Procedia Computer Science264, 346–355 (2025) https://doi.org/10.1016/j.procs.2025.07. 145

  35. [44]

    Language Resources and Evaluation42(4), 335–359 (2008) https://doi.org/10.1007/s10579-008-9076-6

    Busso, C., Bulut, M., Lee, C.-C., Kazemzadeh, A., Mower, E., Kim, S., Chang, J.N., Lee, S., Narayanan, S.S.: Iemocap: Interactive emotional dyadic motion capture database. Language Resources and Evaluation42(4), 335–359 (2008) https://doi.org/10.1007/s10579-008-9076-6

  36. [45]

    Knowledge-Based Systems248, 108861 (2022) https://doi.org/10.1016/ j.knosys.2022.108861

    Gao, Q., Cao, B., Guan, X., Gu, T., Bao, X., Wu, J., Liu, B., Cao, J.: Emotion recognition in conversations with emotion shift detection based on multi-task learning. Knowledge-Based Systems248, 108861 (2022) https://doi.org/10.1016/ j.knosys.2022.108861

  37. [46]

    In: Interspeech 2022, pp

    Chen, C.-Y., Lin, Y.-S., Lee, C.-C.: Emotion-shift aware crf for decoding emotion sequence in conversation. In: Interspeech 2022, pp. 1148–1152 (2022). https:// doi.org/10.21437/Interspeech.2022-10438

  38. [47]

    IEEE Transactions on Affective Computing15(4), 1919–1933 (2024) https://doi.org/10.1109/TAFFC

    Li, J., Wang, X., Liu, Y., Zeng, Z.: Cfn-esa: A cross-modal fusion network with emotion-shift awareness for dialogue emotion recognition. IEEE Transactions on Affective Computing15(4), 1919–1933 (2024) https://doi.org/10.1109/TAFFC. 2024.3389453

  39. [48]

    Complex & Intelligent Systems11, 287 (2025) https://doi.org/10.1007/s40747-025-01903-y

    Yang, J., Wei, P., Du, X., Shen, J.: Graph attention based on contextual reasoning and emotion-shift awareness for emotion recognition in conversations. Complex & Intelligent Systems11, 287 (2025) https://doi.org/10.1007/s40747-025-01903-y

  40. [49]

    In: Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval), pp

    Kumar, S., Akhtar, M.S., Cambria, E., Chakraborty, T.: Semeval 2024 - task 10: Emotion discovery and reasoning its flip in conversation (ediref). In: Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval), pp. 1933– 1946 (2024). https://doi.org/10.1865...

  41. [50]

    Knowledge-Based Systems240, 108112 (2022) https://doi.org/10

    Kumar, S., Shrimal, A., Akhtar, M.S., Chakraborty, T.: Discovering emotion and reasoning its flip in multi-party conversations using masked memory network and transformer. Knowledge-Based Systems240, 108112 (2022) https://doi.org/10. 1016/j.knosys.2021.108112

  42. [51]

    In: Findings of the Association for Com- putational Linguistics: EMNLP, pp

    Sun, Y., Yu, N., Fu, G.: A discourse-aware graph neural network for emotion recognition in multi-party conversation. In: Findings of the Association for Com- putational Linguistics: EMNLP, pp. 2949–2958 (2021). https://doi.org/10.18653/ v1/2021.findings-emnlp.252

  43. [52]

    In: Proceedings of the 59th Annual Meeting 22 of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pp

    Shen, W., Wu, S., Yang, Y., Quan, X.: Directed acyclic graph network for con- versational emotion recognition. In: Proceedings of the 59th Annual Meeting 22 of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Process...

  44. [53]

    arXiv preprint arXiv:2108.12009 (2021) https://doi.org/10

    Kim, T., Vossen, P.: Emoberta: Speaker-aware emotion recognition in conver- sation with roberta. arXiv preprint arXiv:2108.12009 (2021) https://doi.org/10. 48550/arXiv.2108.12009

  45. [54]

    In: Proceedings of the Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL), pp

    Wang, Y., Zhang, J., Ma, J., Wang, S., Xiao, J.: Contextualized emotion recogni- tion in conversation as sequence tagging. In: Proceedings of the Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL), pp. 186–195 (2020). https://doi.org/10.18653/v1/2...

  46. [55]

    Liang, C., Xu, J., Lin, Y., Yang, C., Wang, Y.: S+page: A speaker and position- aware graph neural network model for emotion recognition in conversation. In: Proceedings of the 2nd Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the ...

  47. [56]

    In: 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp

    Song, X., Zang, L., Zhang, R., Hu, S., Huang, L.: Emotionflow: Capture the dialogue level emotion transitions. In: 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 8542–8546 (2022). https: //doi.org/10.1109/ICASSP43922.2022.9746464

  48. [57]

    In: Pro- ceedings of the NAACL-HLT Student Research Workshop, pp

    Saxena, P., Huang, Y.J., Kurohashi, S.: Static and dynamic speaker modeling based on graph neural network for emotion recognition in conversation. In: Pro- ceedings of the NAACL-HLT Student Research Workshop, pp. 247–253 (2022). https://doi.org/10.18653/v1/2022.naacl-srw.31

  49. [58]

    In: Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp

    Hazarika, D., Poria, S., Zadeh, A., Cambria, E., Morency, L.-P., Zimmermann, R.: Conversational memory network for emotion recognition in dyadic dialogue videos. In: Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: H...

  50. [59]

    In: Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pp

    Wang, Y., Wang, B., Zhao, Y., Zhao, D., Jin, X., Zhang, J., He, R., Hou, Y.: Emotion recognition in conversation via dynamic personality. In: Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 202...

  51. [60]

    Proceedings of the AAAI Con- ference on Artificial Intelligence37(11), 13121–13129 (2023) https://doi.org/10

    Li, W., Zhu, L., Mao, R., Cambria, E.: Skier: A symbolic knowledge integrated model for conversational emotion recognition. Proceedings of the AAAI Con- ference on Artificial Intelligence37(11), 13121–13129 (2023) https://doi.org/10. 1609/aaai.v37i11.26541

  52. [61]

    Current Opinion in Psychology 23 17, 22–26 (2017) https://doi.org/10.1016/j.copsyc.2017.06.004

    Kuppens, P., Verduyn, P.: Emotion dynamics. Current Opinion in Psychology 23 17, 22–26 (2017) https://doi.org/10.1016/j.copsyc.2017.06.004

  53. [62]

    Cognition and Emotion12(3), 307–330 (1998) https://doi.org/10

    Davidson, R.J.: Affective style and affective disorders: Perspectives from affective neuroscience. Cognition and Emotion12(3), 307–330 (1998) https://doi.org/10. 1080/026999398379628

  54. [63]

    Emotion11(1), 20–26 (2011) https://doi.org/10.1037/a0021239

    Verduyn, P., Van Mechelen, I., Tuerlinckx, F.: The relation between event pro- cessing and the duration of emotional experience. Emotion11(1), 20–26 (2011) https://doi.org/10.1037/a0021239

  55. [64]

    PLOS ONE9(3), 92410 (2014) https://doi.org/10.1371/journal.pone.0092410

    Brans, K., Verduyn, P.: Intensity and duration of negative emotions: Comparing the role of appraisals and regulation strategies. PLOS ONE9(3), 92410 (2014) https://doi.org/10.1371/journal.pone.0092410

  56. [65]

    Sci- ence China Information Sciences67(8), 182103 (2024) https://doi.org/10.1007/ s11432-023-3908-6

    Liu, Y., Li, J., Wang, X., Zeng, Z.: Emotionic: Emotional inertia and contagion- driven dependency modeling for emotion recognition in conversation. Sci- ence China Information Sciences67(8), 182103 (2024) https://doi.org/10.1007/ s11432-023-3908-6

  57. [66]

    Emotion15(5), 625–636 (2015) https://doi.org/10.1037/emo0000059

    Koval, P., Brose, A., Pe, M.L., Houben, M., Erbas, Y., Champagne, D., Kuppens, P.: Emotional inertia and external events: The roles of exposure, reactivity, and recovery. Emotion15(5), 625–636 (2015) https://doi.org/10.1037/emo0000059

  58. [67]

    In: Affect Dynamics, pp

    Verduyn, P.: Emotion duration. In: Affect Dynamics, pp. 3–18 (2021). https: //doi.org/10.1007/978-3-030-82965-0 1

  59. [68]

    In: Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pp

    Li, D., Wang, Y., Funakoshi, K., Okumura, M.: Joyful: Joint modality fusion and graph contrastive learning for multimodal emotion recognition. In: Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pp. 16051–16069 (2023). https://doi.org/10...

  60. [69]

    IEEE Transactions on Neural Networks and Learning Systems36(3), 4908–4921 (2025) https://doi.org/10.1109/TNNLS.2024.3367940

    Ai, W., Shou, Y., Meng, T., Li, K.: Der-gcn: Dialog and event relation-aware graph convolutional neural network for multimodal dialog emotion recognition. IEEE Transactions on Neural Networks and Learning Systems36(3), 4908–4921 (2025) https://doi.org/10.1109/TNNLS.2024.3367940

  61. [70]

    Pro- ceedings of the AAAI Conference on Artificial Intelligence39(11), 11418–11426 (2025) https://doi.org/10.1609/aaai.v39i11.33242

    Ai, W., Zhang, F., Shou, Y., Meng, T., Chen, H., Li, K.: Revisiting multimodal emotion recognition in conversation from the perspective of graph spectrum. Pro- ceedings of the AAAI Conference on Artificial Intelligence39(11), 11418–11426 (2025) https://doi.org/10.1609/aaai.v39...

  62. [71]

    IEEE Transactions on Affec- tive Computing16(3), 1885–1902 (2025) https://doi.org/10.1109/TAFFC.2025

    Yang, Z., Zhang, Z., Cheng, Y., Zhang, T., Wang, X.: Semantic and emotional dual channel for emotion recognition in conversation. IEEE Transactions on Affec- tive Computing16(3), 1885–1902 (2025) https://doi.org/10.1109/TAFFC.2025. 3544608

  63. [72]

    IEEE Transactions on Affective Computing16(3), 2261–2273 (2025) https://doi.org/ 10.1109/TAFFC.2025.3558222 25

    Shen, S., Liu, F., Wang, H., Zhou, A.: Towards speaker-unknown emotion 24 recognition in conversation via progressive contrastive deep supervision. IEEE Transactions on Affective Computing16(3), 2261–2273 (2025) https://doi.org/ 10.1109/TAFFC.2025.3558222 25

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.