Wen-Hao Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Wen-Hao Zhang has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
Junfeng Zuo
School of Psychology and Cognitive Sciences, Peking University
4 shared publicationsSi Wu
4 shared publications- Si Wu
School of Psychological and Cognitive Sciences, Peking University, Beijing, China
3 shared publications - Ying Nian Wu
UCLA
3 shared publications - Zimei Chen
Lyda Hill Department of Bioinformatics, UT Southwestern Medical Center
3 shared publications Aihua Chen
Key Laboratory of Brain Functional Genomics, Primate Research Center, East China Normal University, Shanghai, China
2 shared publications
Latest Publications
- Open Access
Scale-mixture Langevin sampling in the subspaces of recurrent cortical circuit dynamics
Recurrent cortical circuits are hypothesized to perform Bayesian posterior sampling through their stochastic dynamics. Yet identifying which sampling algorithm a circuit intrinsically implements, as opposed to mapping manually designing algorithms onto circuits, requires rigorous theoretical analysis of the biological circuit dynamics, and remains a fundamental open challenge because of the analytical intractability of nonlinear recurrent dynamics. Here, we tackle this challenge through comprehensive theoretical analysis of analytically tractable nonlinear recurrent circuit dynamics used in neuroscience research— the continuous attractor network (CAN). Surprisingly, we find that the circuit, without any non-Gaussian components in its connectivity or internal variability, intrinsically emerges heavy-tailed subspaces dynamics described as scale-mixture Langevin posterior sampling. The mechanism emerges from the circuit’s own nonlinear dynamics: divisive normalization renders the overall population activation Gamma-distributed which multiplicatively modulates the time constant of the Langevin sampling in the stimulus feature subspace. Eventually, this scale-mixture structure produces a heavy-tailed (Student- t ) sampling step size distribution, significantly accelerating the sampling speed. Moreover, the scale-mixture Langevin sampling approaches a Lévy-like regime in the weak input limit. Our results analytically identify an intrinsic circuit algorithm embedded in subspaces of nonlinear recurrent dynamics, and establish a minimal circuit mechanism for endogenous heavy-tail statistics without any explicit non-Gaussian components.
Open AccessScale-mixture Langevin sampling in the subspaces of recurrent cortical circuit dynamics
Recurrent cortical circuits are hypothesized to perform Bayesian posterior sampling through their stochastic dynamics. Yet identifying which sampling algorithm a circuit intrinsically implements, as opposed to mapping manually designing algorithms onto circuits, requires rigorous theoretical analysis of the biological circuit dynamics, and remains a fundamental open challenge because of the analytical intractability of nonlinear recurrent dynamics. Here, we tackle this challenge through comprehensive theoretical analysis of analytically tractable nonlinear recurrent circuit dynamics used in neuroscience research— the continuous attractor network (CAN). Surprisingly, we find that the circuit, without any non-Gaussian components in its connectivity or internal variability, intrinsically emerges heavy-tailed subspaces dynamics described as scale-mixture Langevin posterior sampling. The mechanism emerges from the circuit’s own nonlinear dynamics: divisive normalization renders the overall population activation Gamma-distributed which multiplicatively modulates the time constant of the Langevin sampling in the stimulus feature subspace. Eventually, this scale-mixture structure produces a heavy-tailed (Student- t ) sampling step size distribution, significantly accelerating the sampling speed. Moreover, the scale-mixture Langevin sampling approaches a Lévy-like regime in the weak input limit. Our results analytically identify an intrinsic circuit algorithm embedded in subspaces of nonlinear recurrent dynamics, and establish a minimal circuit mechanism for endogenous heavy-tail statistics without any explicit non-Gaussian components.
- Open Access
Concurrent model evidence computation and posterior sampling in continuous attractor network subspaces
Extensive studies suggest the brain performs Bayesian inference to infer the latent world states. It is a fundamental neuroscience question that how canonical recurrent neural circuits in the brain implement Bayesian inference. Many existing theoretical studies focused on how the recurrent circuits compute the posterior, while largely overlooking how the circuits compute the model evidence (normalization constant in Bayes’ theorem), a key quantity that measures how well a model explains the observed data. Thus, it remains largely unknown about how the recurrent circuits compute the model evidence. The present study performs rigorous theoretical analyses of the continuous attractor networks, a canonical recurrent circuit model, and reveals that the nonlinear circuit dynamics can simultaneously compute the posterior and model evidence in first two dominant subspaces within the circuit dynamics. Specifically, the circuit dynamics in the stimulus feature subspace implements the Langevin posterior sampling, and the circuit dynamics in the subspace of total neuronal activity computes the model evidence in a way analogous to the evidence lower bound in stochastic variational inference. We further extend the circuit model to compute the model evidence of multiple inputs, and simulations validate the computation in the network. Our work for the first time reveals the concurrent model evidence and posterior sampling in subspaces in continuous attractor networks, significantly deepen our understanding of the computational algorithms adopted by the neural circuits.
Open AccessConcurrent model evidence computation and posterior sampling in continuous attractor network subspaces
Extensive studies suggest the brain performs Bayesian inference to infer the latent world states. It is a fundamental neuroscience question that how canonical recurrent neural circuits in the brain implement Bayesian inference. Many existing theoretical studies focused on how the recurrent circuits compute the posterior, while largely overlooking how the circuits compute the model evidence (normalization constant in Bayes’ theorem), a key quantity that measures how well a model explains the observed data. Thus, it remains largely unknown about how the recurrent circuits compute the model evidence. The present study performs rigorous theoretical analyses of the continuous attractor networks, a canonical recurrent circuit model, and reveals that the nonlinear circuit dynamics can simultaneously compute the posterior and model evidence in first two dominant subspaces within the circuit dynamics. Specifically, the circuit dynamics in the stimulus feature subspace implements the Langevin posterior sampling, and the circuit dynamics in the subspace of total neuronal activity computes the model evidence in a way analogous to the evidence lower bound in stochastic variational inference. We further extend the circuit model to compute the model evidence of multiple inputs, and simulations validate the computation in the network. Our work for the first time reveals the concurrent model evidence and posterior sampling in subspaces in continuous attractor networks, significantly deepen our understanding of the computational algorithms adopted by the neural circuits.
- Open Access
From Representation to Action: A Unified Laplacian Framework for Spatial Representation and Path Planning
Navigation in complex environments relies on internal spatial representations that guide action. While the brain employs a diverse repertoire of spatial tuning cells—including grid, place, and head-direction cells—a normative theory linking these static neural codes to the dynamic process of navigation remains elusive. In this work, we propose a Unified Laplacian Framework derived from first principles of representational smoothness and efficiency. We first demonstrate that diverse spatial codes emerge naturally as spectral decompositions of the Laplace operator. Crucially, bridging the gap from representation to action, we derive a biologically plausible navigation policy based on the Green’s function potential. We show that this potential encodes the environment’s intrinsic geometry to enable simple, trap-free gradient ascent, achieving improved sample efficiency and generalization in goal-reaching tasks. Furthermore, we demonstrate that these spectral representations can be learned directly from high-dimensional visual inputs, confirming their plausibility in realistic environments. Our results suggest that the “cognitive map” can be viewed as a spectral embedding of the Laplacian, providing a rigorous foundation for spatial cognition in both biological and artificial agents.
Open AccessFrom Representation to Action: A Unified Laplacian Framework for Spatial Representation and Path Planning
Navigation in complex environments relies on internal spatial representations that guide action. While the brain employs a diverse repertoire of spatial tuning cells—including grid, place, and head-direction cells—a normative theory linking these static neural codes to the dynamic process of navigation remains elusive. In this work, we propose a Unified Laplacian Framework derived from first principles of representational smoothness and efficiency. We first demonstrate that diverse spatial codes emerge naturally as spectral decompositions of the Laplace operator. Crucially, bridging the gap from representation to action, we derive a biologically plausible navigation policy based on the Green’s function potential. We show that this potential encodes the environment’s intrinsic geometry to enable simple, trap-free gradient ascent, achieving improved sample efficiency and generalization in goal-reaching tasks. Furthermore, we demonstrate that these spectral representations can be learned directly from high-dimensional visual inputs, confirming their plausibility in realistic environments. Our results suggest that the “cognitive map” can be viewed as a spectral embedding of the Laplacian, providing a rigorous foundation for spatial cognition in both biological and artificial agents.
- Open Access
- 28 January 2026
- Published by Springer Nature in Nature
- Vol. 652 (8108) , 157-166
Holistic motor control of zebra finch song syllable sequences
How brain circuits are organized to skillfully produce learned sequences of behaviours is still poorly understood. Here we functionally examined how the cortical song premotor region HVC, which is necessary for zebra finch song1, controls the sequential production of learned song syllables. We found that HVC could generate the complete sequence of learned song syllables independently of its main synaptic input pathways. Thalamic input to HVC was needed for song initiation, but it was not required for transitions between syllables or for song completion. We showed that excitation of HVC neurons during song reliably caused vocalizations to skip back to the beginning of the song, in a manner reminiscent of a skipping record. This restarting of syllable sequences could be induced at any moment of the song and relied on local circuits within HVC. We identified and computationally modelled a synaptic network, including intratelencephalic premotor and corticostriatal neurons within HVC that are essential for completing song syllable sequences. Together, our results show that the learned zebra finch song is controlled by a cortical sequence-generating network in HVC that, once started, can sustain production of all song syllables independent of major extrinsic input pathways. Thus, sequential neuronal activity can be organized to fuse well-learned vocal motor sequences, ultimately achieving holistic control of this naturally learned behaviour.
- 28 January 2026
- Published by Springer Nature in Nature
- Vol. 652 (8108) , 157-166
Open AccessHolistic motor control of zebra finch song syllable sequences
How brain circuits are organized to skillfully produce learned sequences of behaviours is still poorly understood. Here we functionally examined how the cortical song premotor region HVC, which is necessary for zebra finch song1, controls the sequential production of learned song syllables. We found that HVC could generate the complete sequence of learned song syllables independently of its main synaptic input pathways. Thalamic input to HVC was needed for song initiation, but it was not required for transitions between syllables or for song completion. We showed that excitation of HVC neurons during song reliably caused vocalizations to skip back to the beginning of the song, in a manner reminiscent of a skipping record. This restarting of syllable sequences could be induced at any moment of the song and relied on local circuits within HVC. We identified and computationally modelled a synaptic network, including intratelencephalic premotor and corticostriatal neurons within HVC that are essential for completing song syllable sequences. Together, our results show that the learned zebra finch song is controlled by a cortical sequence-generating network in HVC that, once started, can sustain production of all song syllables independent of major extrinsic input pathways. Thus, sequential neuronal activity can be organized to fuse well-learned vocal motor sequences, ultimately achieving holistic control of this naturally learned behaviour.
- Open Access
Towards a mental programming neural circuit: Insights from working memory sequence manipulation
Cognitive reasoning, also known as mental programming, is fundamental to our intelligence. A central function in mental programming is working memory (WM), involving both temporarily maintaining the information and manipulating it based on rules. While the neural mechanisms of WM maintenance have been extensively studied, those governing WM manipulation remain largely unknown. To bridge this gap, the present study focuses on an elementary operation in WM manipulation: re-ordering items in WM sequences. We propose a functional, biologically plausible neural circuit model that consists of two interconnected modules: a memory module composed of continuous attractor-based memory slots that store item features, and a control module sending gain-modulating commands to orchestrate specific operations in the memory module. The model successfully implements two-item swapping in a WM sequence, generating neuronal responses similar to recent primate experiments of WM sequence manipulation. By incorporating principles from the algebraic permutation group, we generalize the circuit model to accommodate more complex sequence manipulations. This math foundation reveals how arbitrary permutations can be decomposed into sequences of elementary swapping operations, which can be generated by a hierarchical tree-structured control circuit module. And the mutual inhibition within the control tree ensures that only one program is being executed at the same time. Our study establishes overarching connections among mental programming neural circuit models, neuro-science experiments, and abstract algebraic structure. These insights enhance our understanding of the neural underpinnings of cognitive reasoning and inspire the design of artificial cognitive systems.
Open AccessTowards a mental programming neural circuit: Insights from working memory sequence manipulation
Cognitive reasoning, also known as mental programming, is fundamental to our intelligence. A central function in mental programming is working memory (WM), involving both temporarily maintaining the information and manipulating it based on rules. While the neural mechanisms of WM maintenance have been extensively studied, those governing WM manipulation remain largely unknown. To bridge this gap, the present study focuses on an elementary operation in WM manipulation: re-ordering items in WM sequences. We propose a functional, biologically plausible neural circuit model that consists of two interconnected modules: a memory module composed of continuous attractor-based memory slots that store item features, and a control module sending gain-modulating commands to orchestrate specific operations in the memory module. The model successfully implements two-item swapping in a WM sequence, generating neuronal responses similar to recent primate experiments of WM sequence manipulation. By incorporating principles from the algebraic permutation group, we generalize the circuit model to accommodate more complex sequence manipulations. This math foundation reveals how arbitrary permutations can be decomposed into sequences of elementary swapping operations, which can be generated by a hierarchical tree-structured control circuit module. And the mutual inhibition within the control tree ensures that only one program is being executed at the same time. Our study establishes overarching connections among mental programming neural circuit models, neuro-science experiments, and abstract algebraic structure. These insights enhance our understanding of the neural underpinnings of cognitive reasoning and inspire the design of artificial cognitive systems.
- Open Access
Canonical cortical circuits: A unified sampling machine for static and dynamic inference
A bstract: The brain lives in an ever-changing world and needs to infer the dynamic evolution of latent states from noisy sensory inputs. Exploring how canonical recurrent neural circuits in the brain realize dynamic inference is a fundamental question in neuroscience. Nearly all existing studies on dynamic inference focus on deterministic algorithms, whereas cortical circuits are intrinsically stochastic, with accumulating evidence suggesting that they employ stochastic Bayesian sampling algorithms. Nevertheless, nearly all circuit sampling studies focused on static inference with fixed posterior over time instead of dynamic inference, leaving a gap between circuit sampling and dynamic inference. To bridge this gap, we study the sampling-based dynamic inference in a canonical recurrent circuit model with excitatory (E) neurons and two types of inhibitory interneurons: parvalbumin (PV) and somatostatin (SOM) neurons. We find that the canonical circuit unifies Langevin and Hamiltonian sampling to infer either static or dynamic latent states with various moving speeds. Remarkably, switching sampling algorithms and adjusting model’s internal latent moving speed can be realized by modulating the gain of SOM neurons without changing synaptic weights. Moreover, when the circuit employs Hamiltonian sampling, its sampling trajectories oscillate around the true latent moving state, resembling the decoded spatial trajectories from hippocampal theta sequences. Our work provides overarching connections between the canonical circuit with diverse interneurons and sampling-based dynamic inference, deepening our understanding of the circuit implementation of Bayesian sampling.
Open AccessCanonical cortical circuits: A unified sampling machine for static and dynamic inference
A bstract: The brain lives in an ever-changing world and needs to infer the dynamic evolution of latent states from noisy sensory inputs. Exploring how canonical recurrent neural circuits in the brain realize dynamic inference is a fundamental question in neuroscience. Nearly all existing studies on dynamic inference focus on deterministic algorithms, whereas cortical circuits are intrinsically stochastic, with accumulating evidence suggesting that they employ stochastic Bayesian sampling algorithms. Nevertheless, nearly all circuit sampling studies focused on static inference with fixed posterior over time instead of dynamic inference, leaving a gap between circuit sampling and dynamic inference. To bridge this gap, we study the sampling-based dynamic inference in a canonical recurrent circuit model with excitatory (E) neurons and two types of inhibitory interneurons: parvalbumin (PV) and somatostatin (SOM) neurons. We find that the canonical circuit unifies Langevin and Hamiltonian sampling to infer either static or dynamic latent states with various moving speeds. Remarkably, switching sampling algorithms and adjusting model’s internal latent moving speed can be realized by modulating the gain of SOM neurons without changing synaptic weights. Moreover, when the circuit employs Hamiltonian sampling, its sampling trajectories oscillate around the true latent moving state, resembling the decoded spatial trajectories from hippocampal theta sequences. Our work provides overarching connections between the canonical circuit with diverse interneurons and sampling-based dynamic inference, deepening our understanding of the circuit implementation of Bayesian sampling.
- Open Access
Natural gradient Bayesian sampling automatically emerges in canonical cortical circuits
A bstract: Accumulating evidence suggests the canonical cortical circuit, consisting of excitatory (E) and diverse classes of inhibitory (I) interneurons, implements Bayesian posterior sampling. However, most of the identified circuits’ sampling algorithms are simpler than the nonlinear circuit dynamics, suggesting complex circuits may implement more advanced algorithms. Through comprehensive theoretical analyses, we discover the canonical circuit innately implements natural gradient Bayesian sampling, which is an advanced sampling algorithm that adaptively adjusts the sampling step size based on the local geometry of stimulus posteriors measured by Fisher information. Specifically, the nonlinear circuit dynamics can implement natural gradient Langevin and Hamiltonian sampling of uni- and multi-variate stimulus posteriors, and these algorithms can be switched by interneurons. We also find that the non-equilibrium circuit dynamics when transitioning from the resting to evoked state can further accelerate natural gradient sampling, and analytically identify the neural circuit’s annealing strategy. Remarkably, we identify the approximated computational strategies employed in the circuit dynamics, which even resemble the ones widely used in machine learning. Our work provides an overarching connection between canonical circuit dynamics and advanced sampling algorithms, deepening our understanding of the circuit algorithms of Bayesian sampling.
Open AccessNatural gradient Bayesian sampling automatically emerges in canonical cortical circuits
A bstract: Accumulating evidence suggests the canonical cortical circuit, consisting of excitatory (E) and diverse classes of inhibitory (I) interneurons, implements Bayesian posterior sampling. However, most of the identified circuits’ sampling algorithms are simpler than the nonlinear circuit dynamics, suggesting complex circuits may implement more advanced algorithms. Through comprehensive theoretical analyses, we discover the canonical circuit innately implements natural gradient Bayesian sampling, which is an advanced sampling algorithm that adaptively adjusts the sampling step size based on the local geometry of stimulus posteriors measured by Fisher information. Specifically, the nonlinear circuit dynamics can implement natural gradient Langevin and Hamiltonian sampling of uni- and multi-variate stimulus posteriors, and these algorithms can be switched by interneurons. We also find that the non-equilibrium circuit dynamics when transitioning from the resting to evoked state can further accelerate natural gradient sampling, and analytically identify the neural circuit’s annealing strategy. Remarkably, we identify the approximated computational strategies employed in the circuit dynamics, which even resemble the ones widely used in machine learning. Our work provides an overarching connection between canonical circuit dynamics and advanced sampling algorithms, deepening our understanding of the circuit algorithms of Bayesian sampling.
- Open Access
Holistic Motor Control of Zebra Finch Song Syllable Sequences
How brain circuits are organized to skillfully produce learned sequences of behaviors is still poorly understood. Here, we functionally examine how the ‘cortical’ song premotor region HVC, which is necessary for zebra finch song, controls the sequential production of learned song syllables. We find that HVC can generate the complete sequence of learned song syllables independently of its main synaptic input pathways. Thalamic input to HVC is permissive for song initiation but it is not required for transitions between syllables or completing song. We show that excitation of HVC neurons during song reliably causes vocalizations to skip back to the beginning of song, reminiscent of a skipping record. This restarting of syllable sequences can be induced at any moment in song and depends on local circuits within HVC. We identify and computationally model a synaptic network including intratelencephalic premotor and corticostriatal neurons within HVC that are essential for completing song syllable sequences. Together, our results show that the learned zebra finch song is controlled by a ‘cortical’ sequence-generating network in HVC that, once started, can sustain production of all song syllables independent of major extrinsic input pathways. Thus, sequential neuronal activity can be organized to fuse well-learned vocal-motor sequences, ultimately achieving holistic control of this naturally learned behavior.
Open AccessHolistic Motor Control of Zebra Finch Song Syllable Sequences
How brain circuits are organized to skillfully produce learned sequences of behaviors is still poorly understood. Here, we functionally examine how the ‘cortical’ song premotor region HVC, which is necessary for zebra finch song, controls the sequential production of learned song syllables. We find that HVC can generate the complete sequence of learned song syllables independently of its main synaptic input pathways. Thalamic input to HVC is permissive for song initiation but it is not required for transitions between syllables or completing song. We show that excitation of HVC neurons during song reliably causes vocalizations to skip back to the beginning of song, reminiscent of a skipping record. This restarting of syllable sequences can be induced at any moment in song and depends on local circuits within HVC. We identify and computationally model a synaptic network including intratelencephalic premotor and corticostriatal neurons within HVC that are essential for completing song syllable sequences. Together, our results show that the learned zebra finch song is controlled by a ‘cortical’ sequence-generating network in HVC that, once started, can sustain production of all song syllables independent of major extrinsic input pathways. Thus, sequential neuronal activity can be organized to fuse well-learned vocal-motor sequences, ultimately achieving holistic control of this naturally learned behavior.
- 26 January 2024
- Published by Springer Nature
- Vol. 1437, 1-21
Decentralized Neural Circuits of Multisensory Information Integration in the Brain
The brain combines multisensory inputs together to obtain a complete and reliable description of the world. Recent experiments suggest that several interconnected multisensory brain areas are simultaneously involved to integrate multisensory information. It was unknown how these mutually connected multisensory areas achieve multisensory integration. To answer this question, using biologically plausible neural circuit models we developed a decentralized system for information integration that comprises multiple interconnected multisensory brain areas. Through studying an example of integrating visual and vestibular cues to infer heading direction, we show that such a decentralized system is well consistent with experimental observations. In particular, we demonstrate that this decentralized system can optimally integrate information by implementing sampling-based Bayesian inference. The Poisson variability of spike generation provides appropriate variability to drive sampling, and the interconnections between multisensory areas store the correlation prior between multisensory stimuli. The decentralized system predicts that optimally integrated information emerges locally from the dynamics of the communication between brain areas and sheds new light on the interpretation of the connectivity between multisensory brain areas. Keywords Multisensory integration Decentralized architecture Sampling-based Bayesian inference Coupled network dynamics
- 26 January 2024
- Published by Springer Nature
- Vol. 1437, 1-21
Decentralized Neural Circuits of Multisensory Information Integration in the Brain
The brain combines multisensory inputs together to obtain a complete and reliable description of the world. Recent experiments suggest that several interconnected multisensory brain areas are simultaneously involved to integrate multisensory information. It was unknown how these mutually connected multisensory areas achieve multisensory integration. To answer this question, using biologically plausible neural circuit models we developed a decentralized system for information integration that comprises multiple interconnected multisensory brain areas. Through studying an example of integrating visual and vestibular cues to infer heading direction, we show that such a decentralized system is well consistent with experimental observations. In particular, we demonstrate that this decentralized system can optimally integrate information by implementing sampling-based Bayesian inference. The Poisson variability of spike generation provides appropriate variability to drive sampling, and the interconnections between multisensory areas store the correlation prior between multisensory stimuli. The decentralized system predicts that optimally integrated information emerges locally from the dynamics of the communication between brain areas and sheds new light on the interpretation of the connectivity between multisensory brain areas. Keywords Multisensory integration Decentralized architecture Sampling-based Bayesian inference Coupled network dynamics
- Open Access
- 4 November 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-19
Sampling-based Bayesian inference in recurrent circuits of stochastic spiking neurons
Two facts about cortex are widely accepted: neuronal responses show large spiking variability with near Poisson statistics and cortical circuits feature abundant recurrent connections between neurons. How these spiking and circuit properties combine to support sensory representation and information processing is not well understood. We build a theoretical framework showing that these two ubiquitous features of cortex combine to produce optimal sampling-based Bayesian inference. Recurrent connections store an internal model of the external world, and Poissonian variability of spike responses drives flexible sampling from the posterior stimulus distributions obtained by combining feedforward and recurrent neuronal inputs. We illustrate how this framework for sampling-based inference can be used by cortex to represent latent multivariate stimuli organized either hierarchically or in parallel. A neural signature of such network sampling are internally generated differential correlations whose amplitude is determined by the prior stored in the circuit, which provides an experimentally testable prediction for our framework.
- 4 November 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-19
Open AccessSampling-based Bayesian inference in recurrent circuits of stochastic spiking neurons
Two facts about cortex are widely accepted: neuronal responses show large spiking variability with near Poisson statistics and cortical circuits feature abundant recurrent connections between neurons. How these spiking and circuit properties combine to support sensory representation and information processing is not well understood. We build a theoretical framework showing that these two ubiquitous features of cortex combine to produce optimal sampling-based Bayesian inference. Recurrent connections store an internal model of the external world, and Poissonian variability of spike responses drives flexible sampling from the posterior stimulus distributions obtained by combining feedforward and recurrent neuronal inputs. We illustrate how this framework for sampling-based inference can be used by cortex to represent latent multivariate stimuli organized either hierarchically or in parallel. A neural signature of such network sampling are internally generated differential correlations whose amplitude is determined by the prior stored in the circuit, which provides an experimentally testable prediction for our framework.