ACC
A Spiking Neural Network for Visual Causal Inference with the Hidden Markov Model
TOTAL VIEWS: 659
Abstract
Causal inference serves as a fundamental cortical function that inherently involves neural communication across various cortical areas. However, the computational principles through which neurons communicate with each other to implement causal inference remain unclear. To address this question, this paper presents a two-layer spiking neural network under the hidden Markov model (HMM) framework to investigate basic visual causal inference. A hidden cause determines whether directions of paired visual stimuli share a common source, randomly. Receiving stimuli, the network incorporates neural communication to generate random responses and employs a distance-based method to infer the hidden cause. With rewards from inference, the network updates its parameters during training. The trained network achieves acceptable performance in designed visual causal inference while reproducing key neurodynamic phenomena of sparse coding and neural variability quenching. The spiking neural network integrates neural communication into an HMM-based unified framework for sparse coding, neural variability quenching, and visual causal inference.
Keywords
Causal inference; Clustering; Spiking neural network; Hidden Markov model
References
[1] Shams L, Beierholm U. Bayesian causal inference: A unifying neuroscience theory. Neurosci Biobehav Rev. 2022;137:104619.
[2] Mohl JT, Pearson JM, Groh JM. Monkeys and humans implement causal inference to simultaneously localize auditory and visual stimuli. J Neurophysiol. 2020;124:715-27.
[3] Fang Y, Yu Z, Liu JK, Chen F. A unified neural circuit of causal inference and multisensory integration. Neurocomputing. 2019;358:355-68.
[4] French RL, DeAngelis GC. Multisensory neural processing: from cue integration to causal inference. Curr Opin Physiol. 2020;16:8-13.
[5] Cao Y, Summerfield C, Park H, Giordano BL, Kayser C. Causal inference in the multisensory brain. Neuron. 2019;102:1076-87.
[6] Huo H, Liu X, Tang Z, Dong Y, Zhao D, Chen D, et al. Interhemispheric multisensory perception and Bayesian causal inference. iScience. 2023;5:106706.
[7] Ma WJ, Rahmati M. Towards a neural implementation of causal inference in cue combination. Multisens Res. 2013;26:159-76.
[8] Liu W, Pan X. Effects of neural assembles in causal inference based on an entropy-maximization Bayesian neural network. IEEE Access. 2024;12:184442-55.
[9] Cant JS, Arnott SR, Goodale MA. fmr-adaptation reveals separate processing regions for the perception of form and texture in the human ventral stream. Exp Brain Res. 2009;192:391-405.
[10] Cant JS, Large ME, McCall L, Goodale MA. Independent processing of form, colour, and texture in object perception. Perception. 2008;37:57-78.
[11] Liu W, Ren R. Disinhibition maintains network performances in concept learning through regulating neural responses. Neurocomputing. 2025;649:130760.
[12] Liu W, Liu X. Pre-stimulus network responses affect information coding in neural variability quenching. Neurocomputing. 2023;531:1-20.
[13] Liu W, Liu X. The effects of eye movements on the visual cortical responding variability based on a spiking network. Neurocomputing. 2021;436:58-73.
[14] Pokorny C, Ison MJ, Rao A, Legenstein R, Papadimitriou C, Maass W. STDP forms associations between memory traces in networks of spiking neurons. Cereb Cortex. 2020;30:952-68.
[15] Jonke Z, Legenstein R, Habenschuss S, Maass W. Feedback inhibition shapes emergent computational properties of cortical micro-circuit motifs. J Neurosci. 2017;37:8511-23.
[16] Wolff A, Chen L, Tumati S, Golesorkhi M, Gomez-Pilar J, Hu J, et al. Prestimulus dynamics blend with the stimulus in neural variability quenching. Neuroimage. 2021;238:118160.
[17] Field DJ. What is the goal of sensory coding? Neural Comput. 1994;6:559-601.
[18] Enroth-Cugell C, Robson JG. The contrast sensitivity of retinal ganglion cells of the cat. J Physiol. 1966;187:517-52.
[19] Segal IY, Giladi C, Gedalin M, Rucci M, Ben-Tov M, Kushinsky Y, et al. Decorrelation of retinal response to natural scenes by fixational eye movements. Proc Natl Acad Sci U S A. 2015;112:3110-5.
[20] Rueckert E, Kappel D, Tanneberg D, Pecevski D, Peters J. Recurrent spiking networks solve planning tasks. Sci Rep. 2016;6:21142.
[21] Kappel D, Nessler B, Maass W. STDP installs in winner-take-all circuits an online approximation to hidden markov model learning. PLoS Comput Biol. 2014;10:e1003511.
[22] Heinerman J, Haasdijk E, Eiben A. Unsupervised identification and recognition of situations for high-dimensional sensori-motor streams. Neurocomputing. 2017;262:90-107.
[23] Kloosterman NA, de Gee JW, Werkle-Bergner M, Lindenberger U, Garrett DD, Fahrenfort JJ. Humans strategically shift decision bias by flexibly adjusting sensory evidence accumulation. Elife. 2019;8:e37321.
[24] Olshausen BA, Field DJ. Sparse coding of sensory inputs. Curr Opin Neurobiol. 2004;14:481-7.
How to cite this paper
A Spiking Neural Network for Visual Causal Inference with the Hidden Markov Model
How to cite this paper: Ruihuan Ren, Pengcheng Cui, Jinping Yuan, Jiaheng Song, Yiran Li, Yutong Lu, Zixuan Huang, Jianyu Wang, Weisi Liu. (2025) A Spiking Neural Network for Visual Causal Inference with the Hidden Markov Model. Advances in Computer and Communication, 6(5), 310-316.
DOI: http://dx.doi.org/10.26855/acc.2025.12.009
More Articles
- Research on the Teaching Reform and Practice of C Language Course Based on Project-driven
- A Spiking Neural Network for Visual Causal Inference with the Hidden Markov Model
- Research on Generative AI Psychological Intervention Models Based on Multimodal Emotion Recognition
- AI-enabled Methods for Business Data Analytics Accuracy Enhancement and Emerging Advances
- Research on the Application of Digital Healthcare Platforms in Chronic Disease Management
- Research on Lifecycle Configuration and Reclamation Strategies for Edge Nodes Based on Microservice Architectures