反馈通过降低感官噪音和校准知觉-反应映射来优化视知觉

doi:10.3724/SP.J.1041.2026.2167

摘要/Abstract

摘要： 视知觉的系统性偏差可由 “高效编码—贝叶斯解码—知觉-反应映射” 级联计算框架解释。然而, 该框架主要基于无反馈条件下的开环知觉构建。反馈作为知觉学习的核心要素, 其在此级联过程中的计算靶点尚不明确。本研究采用2 (反馈: 有/无, 组间) × 3 (反应范围: 80°、160°、240°, 组内)混合实验设计, 结合7种备择贝叶斯观测者模型, 系统检验反馈对感官噪声(κ)、先验整合权重(w)及映射缩放因子(α)的调节。行为结果显示, 反馈显著削弱了宽反应范围下的高估偏差并降低了估计变异性。模型比较筛选出最优模型, 其参数揭示反馈显著提高了感官编码精度(κ增大)并降低了映射增益(α减小), 但先验整合权重(w)无显著变化。结论表明, 反馈通过降低感觉编码噪声与校准知觉-反应映射增益双重机制优化视知觉, 而非更新先验信念; 该发现从计算层面支持了先验构建与反馈驱动调节的可分离性, 并将开环知觉框架成功拓展至闭环学习场景。

关键词: 视知觉, 贝叶斯观察者模型, 高效编码, 反馈, 知觉-反应映射

Abstract: Visual perception is often characterized by systematic biases, which have been successfully explained by a three-stage cascade model comprising efficient sensory encoding, Bayesian decoding, and perception-action mapping. However, this framework was developed primarily under no-feedback conditions, leaving it unclear how feedback—an essential component of perceptual learning—modulates these computational stages. The current study investigates how feedback influences visual estimation of self-motion direction (heading) from optic flow, and whether its effects can be explained by changes in sensory noise, prior integration, or perception-action scaling within a unified Bayesian framework.
Forty participants performed a heading estimation task using optic flow stimuli simulating self-motion directions ranging from -33° to 33°. A between-subjects design was employed: one group (n = 20) completed the task without feedback, while the other group (n = 20) received trial-by-trial visual feedback indicating the discrepancy between their estimate and the true heading. Both groups performed the task under three response range conditions (80°, 160°, and 240°), with 720 trials per participant. To uncover the computational basis of feedback effects, we developed seven Bayesian observer models that varied in their assumptions about which components were modulated by feedback: sensory noise (κ), prior integration (weighted combination of natural and experimental priors, w), perception-action scaling (α), and the boundary of the experimental prior (b). Model parameters were jointly fitted to individual participant data using maximum likelihood estimation, and model performance was compared using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC).
Behaviorally, wider response ranges induced a systematic transition from underestimation (at 80°) to overestimation (at 160° and 240°). Feedback significantly reduced overestimation bias and estimation variability, particularly for the wider response ranges. Model comparison revealed that Model 4—which incorporated feedback-dependent sensory noise, weighted integration of natural and experimental priors, and range-specific perception-action scaling—provided the best account of the data, with the lowest AIC and BIC values across conditions. Parameter estimates from Model 4 showed that feedback significantly increased κ (indicating reduced sensory noise) and significantly decreased the scaling factors α₂ and α₃ for the two wider response ranges, while α₁ (80° range) remained unchanged. Critically, the prior weighting parameter w did not differ between feedback conditions, suggesting that feedback did not alter how observers combined long-term natural priors with short-term experimental priors.
Feedback refines visual heading estimation by reducing sensory encoding noise and attenuating the gain of perception-action mapping, rather than by modifying prior integration. These findings reveal dissociable computational roles for feedback across distinct stages of visual processing. By extending a validated cascade model to incorporate feedback, this study provides a unified computational account of how external error signals improve perceptual accuracy. The results bridge behavioral performance with mechanistic modeling, offering theoretical insights into feedback-driven perceptual learning and practical implications for paradigms involving perceptual training, sensorimotor adaptation, and clinical rehabilitation where feedback plays a central role.

Key words: visual perception, Bayesian observer model, efficient coding, feedback, perception-action mapping

中图分类号:

B842

孙琪. (2026). 反馈通过降低感官噪音和校准知觉-反应映射来优化视知觉. 心理学报, 58(11), 2167-2177.

SUN Qi. (2026). Feedback refines visual perception by reducing sensory noise and calibrating perception-action mapping. Acta Psychologica Sinica, 58(11), 2167-2177.

参考文献

[1] Adams W. J., Graf E. W., & Ernst M. O. (2004). Experience can change the 'light-from-above' prior.Nature Neuroscience, 7(10), 1057-1058.
[2] Attneave, F. (1954). Some informational aspects of visual perception.Psychological Review, 61(3), 183-193.
[3] Barlow, H. B. (1961). Possible principles underlying the transformations of sensory messages.Sensory Communication, 1(01), 217-233.
[4] Berens, P. (2009). CircStat: a MATLAB toolbox for circular statistics.Journal of Statistical Software, 31(10), 1-21.
[5] Crane, B. T. (2012). Direction specific biases in human visual and vestibular heading perception.PLoS One, 7(12), e51383.
[6] Cuturi, L. F., & MacNeilage, P. R. (2013). Systematic biases in human heading estimation.PLoS One, 8(2), e56862.
[7] De Gardelle V., Kouider S., & Sackur J. (2010). An oblique illusion modulated by visibility: Non-monotonic sensory integration in orientation processing.Journal of Vision, 10(10), 6.
[8] De Nil, L. F., & Lafaille, S. J. (2002). Jaw and finger movement accuracy under visual and nonvisual feedback conditions.Perceptual and Motor Skills, 95(3_suppl), 1129-1140.
[9] De Valois R. L., Yund E. W., & Hepler N. (1982). The orientation and direction selectivity of cells in macaque visual cortex.Vision Research, 22(5), 531-544.
[10] Ernst, B., & Steinhauser, M. (2018). Effects of feedback reliability on feedback-related brain activity: A feedback valuation account.Cognitive, Affective, & Behavioral Neuroscience, 18, 596-608.
[11] Gekas N., Chalk M., Seitz A. R., & Seriès P. (2013). Complexity and specificity of experimentally-induced expectations in motion perception.Journal of Vision, 13(4), 8.
[12] Gu Y., Fetsch C. R., Adeyemo B., DeAngelis G. C., & Angelaki D. E. (2010). Decoding of MSTd population activity accounts for variations in the precision of heading perception.Neuron, 66(4), 596-609.
[13] Hürlimann F., Kiper D. C., & Carandini M. (2002). Testing the Bayesian model of perceived speed.Vision Research, 42(19), 2253-2257.
[14] Knill D. C.,& Richards, W. (Eds.). (1996). Perception as Bayesian inference. Cambridge University Press.
[15] Li B., Peterson M. R., & Freeman R. D. (2003). Oblique effect: a neural basis in the visual cortex.Journal of Neurophysiology, 90(1), 204-217.
[16] Mao, J., & Stocker, A. A. (2024). Sensory perception is a holistic inference process.Psychological Review, 131(4), 858-890.
[17] Noel J. P., Zhang L. Q., Stocker A. A., & Angelaki D. E. (2021). Individuals with autism spectrum disorder have altered visual encoding capacity.PLoS Biology, 19(5), e3001215.
[18] Stocker, A. A., & Simoncelli, E. P. (2006). Noise characteristics and prior expectations in human visual speed perception.Nature Neuroscience, 9(4), 578-585.
[19] Sun Q., Dong L. L., You F. H., & Chang D. H. (2024). Estimations of biological motion walking direction are affected by observer and walker genders.BMC Psychology, 12(1), 698.
[20] Sun Q., Gong X. M., & Sun Q. (2025). Impact of conflicts between long-and short-term priors on the weighted prior integration in visual perception.Cognition, 254, 106006.
[21] Sun Q., Wang J. Y., & Gong X. M. (2024). Conflicts between short-and long-term experiences affect visual perception through modulating sensory or motor response systems: Evidence from Bayesian inference models.Cognition, 246, 105768.
[22] Sun Q., Xu L. H., & Stocker A. A. (2025). A linear perception-action mapping accounts for response range-dependent biases in heading estimation from optic flow.PLoS Computational Biology, 21(6), e1013147.
[23] Sun Q., Zhan L. Z., You F. H.,& Dong, X. F.(2024). Attention affects the perception of self-motion direction from optic flow. iScience, 272024.109373
[24] Treue, S., & Trujillo, J. C. M. (1999). Feature-based attention influences motion processing gain in macaque visual cortex.Nature, 399(6736), 575-579.
[25] Varrier R. S., Rothkirch M., Stuke H., Guggenmos M., & Sterzer P. (2020). Unreliable feedback deteriorates information processing in primary visual cortex.NeuroImage, 214, 116701.
[26] Varrier R. S., Stuke H., Guggenmos M., & Sterzer P. (2019). Sustained effects of corrupted feedback on perceptual inference.Scientific Reports, 9(1), 5537.
[27] Wei, X. X., & Stocker, A. A. (2012). Efficient coding provides a direct link between prior and likelihood in perceptual Bayesian inference.Advances in Neural Information Processing Systems, 25, 1313-1321.
[28] Wei, X. X., & Stocker, A. A. (2015). A Bayesian observer model constrained by efficient coding can explain 'anti-Bayesian' percepts.Nature Neuroscience, 18(10), 1509-1517.
[29] Wei, X. X., & Stocker, A. A. (2017). Lawful relation between perceptual bias and discriminability.Proceedings of the National Academy of Sciences, 114(38), 10244-10249.
[30] Xue C., Kaping D., Ray S. B., Krishna B. S., & Treue S. (2017). Spatial attention reduces burstiness in macaque visual cortical area MST.Cerebral Cortex, 27(1), 83-91.