Human visual motion perception shows hallmarks of Bayesian structural inference

Yang, Sichao; Bill, Johannes; Drugowitsch, Jan; Gershman, Samuel J.

doi:10.1101/2020.11.05.370452

Cited by 3 publications

(10 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Having qualitatively replicated motion structure inference in common motion displays, we next asked if our online model could quantitatively explain human motion structure perception. To address this question, we reevaluated behavioral data from Yang et al 17 , where participants had to categorize the latent structure of short motion displays (see Fig. 3a).…”

Section: Online Inference Outperforms Ideal Observers In Explaining H...mentioning

confidence: 99%

“…The resulting set of interconnected differential equations decomposes a scene's velocities with the goal of minimizing prediction errors for subsequent observations. Beyond capturing human percepts in many psychophysics experiments qualitatively, the model explains human motion structure classification quantitatively with higher fidelity than a previous ideal observer-based model 17 . Furthermore, the model provides a normative explanation for the putative origin of human illusory motion perception, and yields testable predictions for future psychophysics experiments.…”

mentioning

confidence: 92%

“…Motion structure thus carries essential information about the spatial and temporal evolution of the environment, and aids behaviors such as navigation, tracking, prediction, and pursuit [1][2][3][4][5][6][7][8] . It remains, however, unclear how the visual system identifies a scene's underlying motion structure and exploits it to turn noisy, unstructured, sensory impressions into meaningful motion percepts.In recent years, Bayesian inference has provided a successful normative perspective on many aspects of visual motion perception [9][10][11][12][13][14][15][16][17] . Human perception of motion stimuli spatially constrained by an aperture is well-explained by Bayesian statistical inference [9][10][11]14 , and neural circuits that integrate local retinal input into neural representations of motion have been identified [18][19][20][21][22][23] .…”

mentioning

confidence: 99%

“…Recent theoretical work 15 has introduced a representation of tree structures for the mental organization of observed velocities into nested hierarchies. Theory-driven experiments subsequently demonstrated that the human visual system indeed makes use of hierarchical structure when solving visual tasks 16 , and that salient aspects of human motion structure perception can be explained by normative models of Bayesian inference over tree structures 17 . Because these studies were restricted to modeling motion integration only with regard to the perceptual outcome-they analyzed presented visual scenes offline using ideal Bayesian observer models-it remained unclear how the visual system solves the chicken-and-egg problem of parsing (in real time) instantaneous motion in a scene while simultaneously inferring the scene's underlying structure.We address this question by formulating visual motion perception as online hierarchical inference in a generative model of structured motion.…”

mentioning

confidence: 99%

“…In recent years, Bayesian inference has provided a successful normative perspective on many aspects of visual motion perception [9][10][11][12][13][14][15][16][17] . Human perception of motion stimuli spatially constrained by an aperture is well-explained by Bayesian statistical inference [9][10][11]14 , and neural circuits that integrate local retinal input into neural representations of motion have been identified [18][19][20][21][22][23] .…”

mentioning

confidence: 99%

See 4 more Smart Citations

Visual motion perception as online hierarchical inference

2022

View full text Add to dashboard Cite

Identifying the structure of motion relations in the environment is critical for navigation, tracking, prediction, and pursuit. Yet, little is known about the mental and neural computations that allow the visual system to infer this structure online from a volatile stream of visual information. We propose online hierarchical Bayesian inference as a principled solution for how the brain might solve this complex perceptual task. We derive an online Expectation-Maximization algorithm that explains human percepts qualitatively and quantitatively for a diverse set of stimuli, covering classical psychophysics experiments, ambiguous motion scenes, and illusory motion displays. We thereby identify normative explanations for the origin of human motion structure perception and make testable predictions for future psychophysics experiments. The proposed online hierarchical inference model furthermore affords a neural network implementation which shares properties with motion-sensitive cortical areas and motivates targeted experiments to reveal the neural representations of latent structure.

show abstract

Section: Online Inference Outperforms Ideal Observers In Explaining H...mentioning

confidence: 99%

mentioning

confidence: 92%

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Visual motion perception as online hierarchical inference

2022

View full text Add to dashboard Cite

show abstract

Structure in motion: visual motion perception as online hierarchical inference

Bill

Gershman

Drugowitsch

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Identifying the structure of motion relations in the environment is critical for navigation, tracking, prediction, and pursuit. Yet, little is known about the mental and neural computations that allow the visual system to infer this structure online from a volatile stream of visual information. We propose online hierarchical Bayesian inference as a principled solution for how the brain might solve this complex perceptual task. We derive an online Expectation-Maximization algorithm that explains human percepts qualitatively and quantitatively for a diverse set of stimuli, covering classical psychophysics experiments, ambiguous motion scenes, and illusory motion displays. We thereby identify normative explanations for the origin of human motion structure perception and make testable predictions for new psychophysics experiments. The algorithm furthermore affords a neural network implementation which shares properties with motion-sensitive cortical areas and motivates a novel class of experiments to reveal the neural representations of latent structure.

show abstract

Causal inference during closed-loop navigation: parsing of self- and object-motion

Noel

Bill

Ding

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

A key computation in building adaptive internal models of the external world is to ascribe sensory signals to their likely cause(s), a process of Bayesian Causal Inference (CI). CI is well studied within the framework of two-alternative forced-choice tasks, but less well understood within the cadre of naturalistic action-perception loops. Here, we examine the process of disambiguating retinal motion caused by self- and/or object-motion during closed-loop navigation. First, we derive a normative account specifying how observers ought to intercept hidden and moving targets given their belief over (i) whether retinal motion was caused by the target moving, and (ii) if so, with what velocity. Next, in line with the modeling results, we show that humans report targets as stationary and steer toward their initial rather than final position more often when they are themselves moving, suggesting a misattribution of object-motion to the self. Further, we predict that observers should misattribute retinal motion more often: (i) during passive rather than active self-motion (given the lack of an efference copy informing self-motion estimates in the former), and (ii) when targets are presented eccentrically rather than centrally (given that lateral self-motion flow vectors are larger at eccentric locations during forward self-motion). Results confirm both of these predictions. Lastly, analysis of eye-movements show that, while initial saccades toward targets are largely accurate regardless of the self-motion condition, subsequent gaze pursuit was modulated by target velocity during object-only motion, but not during concurrent object- and self-motion. These results demonstrate CI within action-perception loops, and suggest a protracted temporal unfolding of the computations characterizing CI.

show abstract

Human visual motion perception shows hallmarks of Bayesian structural inference

Cited by 3 publications

References 57 publications

Visual motion perception as online hierarchical inference

Visual motion perception as online hierarchical inference

Structure in motion: visual motion perception as online hierarchical inference

Causal inference during closed-loop navigation: parsing of self- and object-motion

Contact Info

Product

Resources

About