High-level Priors Research Articles

A novel model of divergent predictive perception.

Predictive processing theories state that our subjective experience of reality is shaped by a balance of expectations based on previous knowledge about the world (i.e. priors) and confidence in sensory input from the environment. Divergent experiences (e.g. hallucinations and synaesthesia) are likely to occur when there is an imbalance between one's reliance on priors and sensory input. In a novel theoretical model, inspired by both predictive processing and psychological principles, we propose that predictable divergent experiences are associated with natural or environmentally induced prior/sensory imbalances: inappropriately strong or inflexible (i.e. maladaptive) high-level priors (beliefs) combined with low sensory confidence can result in reality discrimination issues, a characteristic of psychosis; maladaptive low-level priors (sensory expectations) combined with high sensory confidence can result in atypical sensory sensitivities and persistent divergent percepts, a characteristic of synaesthesia. Crucially, we propose that whether different divergent experiences manifest with dominantly sensory (e.g. hallucinations) or nonsensory characteristics (e.g. delusions) depends on mental imagery ability, which is a spectrum from aphantasia (absent or weak imagery) to hyperphantasia (extremely vivid imagery). We theorize that imagery is critically involved in shaping the sensory richness of divergent perceptual experience. In sum, to predict a range of divergent perceptual experiences in both clinical and general populations, three factors must be accounted for: a maladaptive use of priors, individual level of confidence in sensory input, and mental imagery ability. These ideas can be expressed formally using nonparametric regression modeling. We provide evidence for our theory from previous work and deliver predictions for future research.

Differential Oriented Image Foresting Transform and Its Applications to Support High-level Priors for Object Segmentation

Differential Oriented Image Foresting Transform and Its Applications to Support High-level Priors for Object Segmentation

A unified model of ketamine's dissociative and psychedelic properties.

Ketamine is an N-methyl-d-aspartate antagonist which is increasingly being researched and used as a treatment for depression. In low doses, it can cause a transitory modification in consciousness which was classically labelled as 'dissociation'. However, ketamine is also commonly classified as an atypical psychedelic and it has been recently reported that ego dissolution experiences during ketamine administration are associated with greater antidepressant response. Neuroimaging studies have highlighted several similarities between the effects of ketamine and those of serotonergic psychedelics in the brain; however, no unified account has been proposed for ketamine's multi-level effects - from molecular to network and psychological levels. Here, we propose that the fast, albeit transient, antidepressant effects observed after ketamine infusions are mainly driven by its acute modulation of reward circuits and sub-acute increase in neuroplasticity, while its dissociative and psychedelic properties are driven by dose- and context-dependent disruption of large-scale functional networks. Computationally, as nodes of the salience network (SN) represent high-level priors about the body ('minimal' self) and nodes of the default-mode network (DMN) represent the highest-level priors about narrative self-experience ('biographical' self), we propose that transitory SN desegregation and disintegration accounts for ketamine's 'dissociative' state, while transitory DMN desegregation and disintegration accounts for ketamine's 'psychedelic' state. In psychedelic-assisted psychotherapy, a relaxation of the highest-level beliefs with psychotherapeutic support may allow a revision of pathological self-representation models, for which neuroplasticity plays a permissive role. Our account provides a multi-level rationale for using the psychedelic properties of ketamine to increase its long-term benefits.

Psychedelic Science of Spirituality and Religion: An Attachment-Informed Agenda Proposal

ABSTRACT In this paper, we set an agenda for a psychedelic science of spirituality and religion based on attachment theory. Attachment theory proposes that people develop internal working models (IWMs) of interactions with others from their relational experiences with caregivers. Such IWMs then function as high-level priors or predictive models, enabling people, for better and for worse, to predict and organize their interpersonal and religious/spiritual relationships. One mechanism by which efficacious psychedelic interventions may work is by relaxing the grip of rigid, defensive priors (e.g. insecure IWMs with regard to others and God), further amplified by corrective relational experiences with the therapist, God, or others. We outline three key proposals to steer future research. First, individual differences in attachment security predict the phenomenology and integration of psychedelic experiences. Second, efficacious psychedelic therapy facilitates increased attachment security as a clinically relevant outcome. Third, attachment-related dynamics (e.g. a sense of connection to others/God/the universe, alleviation of attachment-related worries and defenses) are process-level mechanisms involved in the clinical utility of psychedelic treatment. Finally, we discuss the role of religion and spirituality in psychedelic experiences from an attachment perspective.

Cross-modal auditory priors drive the perception of bistable visual stimuli with reliable differences between individuals

It is a widely held assumption that the brain performs perceptual inference by combining sensory information with prior expectations, weighted by their uncertainty. A distinction can be made between higher- and lower-level priors, which can be manipulated with associative learning and sensory priming, respectively. Here, we simultaneously investigate priming and the differential effect of auditory vs. visual associative cues on visual perception, and we also examine the reliability of individual differences. Healthy individuals (N = 29) performed a perceptual inference task twice with a one-week delay. They reported the perceived direction of motion of dot pairs, which were preceded by a probabilistic visuo-acoustic cue. In 30% of the trials, motion direction was ambiguous, and in half of these trials, the auditory versus the visual cue predicted opposing directions. Cue-stimulus contingency could change every 40 trials. On ambiguous trials where the visual and the auditory cue predicted conflicting directions of motion, participants made more decisions consistent with the prediction of the acoustic cue. Increased predictive processing under stimulus uncertainty was indicated by slower responses to ambiguous (vs. non-ambiguous) stimuli. Furthermore, priming effects were also observed in that perception of ambiguous stimuli was influenced by perceptual decisions on the previous ambiguous and unambiguous trials as well. Critically, behavioural effects had substantial inter-individual variability which showed high test–retest reliability (intraclass correlation coefficient (ICC) > 0.78). Overall, higher-level priors based on auditory (vs. visual) information had greater influence on visual perception, and lower-level priors were also in action. Importantly, we observed large and stable differences in various aspects of task performance. Computational modelling combined with neuroimaging could allow testing hypotheses regarding the potential mechanisms causing these behavioral effects. The reliability of the behavioural differences implicates that such perceptual inference tasks could be valuable tools during large-scale biomarker and neuroimaging studies.

Automated Movement Assessment in Stroke Rehabilitation

We are developing a system for long term Semi-Automated Rehabilitation At the Home (SARAH) that relies on low-cost and unobtrusive video-based sensing. We present a cyber-human methodology used by the SARAH system for automated assessment of upper extremity stroke rehabilitation at the home. We propose a hierarchical model for automatically segmenting stroke survivor's movements and generating training task performance assessment scores during rehabilitation. The hierarchical model fuses expert therapist knowledge-based approaches with data-driven techniques. The expert knowledge is more observable in the higher layers of the hierarchy (task and segment) and therefore more accessible to algorithms incorporating high level constraints relating to activity structure (i.e., type and order of segments per task). We utilize an HMM and a Decision Tree model to connect these high level priors to data driven analysis. The lower layers (RGB images and raw kinematics) need to be addressed primarily through data driven techniques. We use a transformer based architecture operating on low-level action features (tracking of individual body joints and objects) and a Multi-Stage Temporal Convolutional Network(MS-TCN) operating on raw RGB images. We develop a sequence combining these complimentary algorithms effectively, thus encoding the information from different layers of the movement hierarchy. Through this combination, we produce a robust segmentation and task assessment results on noisy, variable and limited data, which is characteristic of low cost video capture of rehabilitation at the home. Our proposed approach achieves 85% accuracy in per-frame labeling, 99% accuracy in segment classification and 93% accuracy in task completion assessment. Although the methodology proposed in this paper applies to upper extremity rehabilitation using the SARAH system, it can potentially be used, with minor alterations, to assist automation in many other movement rehabilitation contexts (i.e., lower extremity training for neurological accidents).

Overly Strong Priors for Socially Meaningful Visual Signals Are Linked to Psychosis Proneness in Healthy Individuals.

According to the predictive coding theory of psychosis, hallucinations and delusions are explained by an overweighing of high-level prior expectations relative to sensory information that leads to false perceptions of meaningful signals. However, it is currently unclear whether the hypothesized overweighing of priors (1) represents a pervasive alteration that extends to the visual modality and (2) takes already effect at early automatic processing stages. Here, we addressed these questions by studying visual perception of socially meaningful stimuli in healthy individuals with varying degrees of psychosis proneness (n = 39). In a first task, we quantified participants’ prior for detecting faces in visual noise using a Bayesian decision model. In a second task, we measured participants’ prior for detecting direct gaze stimuli that were rendered invisible by continuous flash suppression. We found that the prior for detecting faces in noise correlated with hallucination proneness (r = 0.50, p = 0.001, Bayes factor 1/20.1) as well as delusion proneness (r = 0.46, p = 0.003, BF 1/9.4). The prior for detecting invisible direct gaze was significantly associated with hallucination proneness (r = 0.43, p = 0.009, BF 1/3.8) but not conclusively with delusion proneness (r = 0.30, p = 0.079, BF 1.7). Our results provide evidence for the idea that overly strong high-level priors for automatically detecting socially meaningful stimuli might constitute a processing alteration in psychosis.

Efficient Hierarchical Multi-Object Segmentation in Layered Graphs

Abstract We propose a novel efficient seed-based method for the multi-object segmentation of images based on graphs, named Hierarchical Layered Oriented Image Foresting Transform (HLOIFT). It uses a tree of the relations between the image objects, with each node in the tree representing an object. Each tree node may contain different individual high-level priors of its corresponding object and defines a weighted digraph, named as layer. The layer graphs are then integrated into a hierarchical graph, considering the hierarchical relations of inclusion and exclusion. A single energy optimization is performed in the hierarchical layered weighted digraph leading to globally optimal results satisfying all the high-level priors. The experimental evaluations of HLOIFT, on medical, natural, and synthetic images, indicate promising results comparable to the related baseline methods that include structural information, but with lower computational complexity. Compared to the hierarchical segmentation by the min-cut/max-flow algorithm, our approach is less restrictive, leading to globally optimal results in more general scenarios, and has a better running time.

A Visual Saliency-Based Approach for Content-Based Image Retrieval

During the past two decades an enormous amount of visual information has been generated; as a result, content-based image retrieval (CBIR) has received considerable attention. In CBIR the image is used as a query to find the most similar images. One of the biggest challenges in CBIR system is to fill up the “semantic gap,” which is the gap between low-level visual features and the high-level semantic concepts of an image. In this paper, the authors have proposed a saliency-based CBIR system that utilizes the semantic information of image and users search intention. In the proposed model, firstly a significant region is identified with the help of method structured matrix decomposition (SMD) using high-level priors that highlight the prominent area of the image. After that, a two-dimensional principal component analysis (2DPCA) is used as a feature, which is compact and effectively used for fast recognition. Experiment results are validated on different image dataset having an extensive collection of semantic classifications.

Dissolving the self

Psychedelic drugs such as psilocybin, LSD and DMT are known to induce powerful alterations in phenomenology. Perhaps of most philosophical and scientific interest is their capacity to disrupt and even “dissolve” one of the most primary features of normal experience: that of being a self. Such “peak” or “mystical” experiences are of increasing interest for their potentially transformative therapeutic value. While empirical research is underway, a theoretical conception of the mechanisms underpinning these experiences remains elusive. In the following paper, psychedelic-induced ego-dissolution is accounted for within an active inference framework, as a collapse in the “temporal thickness” of an agent’s deep temporal model, as a result of lowered precision on high-level priors. The argument here is composed of three moves: first, a view of the self-model is proposed as arising within a temporally deep generative model of an embodied organism navigating an affordance landscape in the service of allostasis. Next, a view of the action of psychedelics as lowering the precision of high-level priors within the generative model is unpacked in terms of a high Bayesian learning rate. Finally, the relaxation of high-level priors is argued to cause a “collapse” in the temporal thickness of the generative model, resulting in a collapse in the self-model and a loss of the ordinary sense of being a self. This account has implications for our understanding of ordinary self-consciousness and disruptions in self-consciousness present in psychosis, autism, depression, and dissociative disorders. The philosophical, theoretical and therapeutic implications of this account are touched upon.

Performance Analysis of Structured Matrix Decomposition with Contour Based Spatial Prior for Extracting Salient Object from Complex Scene

Automatically extracting most conspicuous object from an image is useful and important for many computer vision related tasks. Performance of several applications such as object segmentation, image classification based on salient object and content based image editing in computer vision can be improved using this technique. In this research work, performance of structured matrix decomposition with contour based spatial prior is analyzed for extracting salient object from the complex scene. To separate background and salient object, structured matrix decomposition model based on low rank matrix recovery theory is used along with two structural regularizations. Tree structured sparsity inducing regularization is used to capture image structure and to enforce the same object to assign similar saliency values. And, Laplacian regularization is used to enlarge the gap between background part and salient object part. In addition to structured matrix decomposition model, general high level priors along with biologically inspired contour based spatial prior is integrated to improve the performance of saliency related tasks. The performance of the proposed method is evaluated on two demanding datasets, namely, ICOSEG and PASCAL-S for complex scene images. For PASCAL-S dataset precision recall curve of proposed method starts from 0.81 and follows top and right-hand border more than structured matrix decomposition which starts from 0.79. Similarly, structural similarity index score, which is 0.596654 and 0.394864 without using contour based spatial prior and 0.720875 and 0.568001 using contour based spatial prior for ICOSEG and PASCAL-S datasets shows improved result.

A Visual Saliency Detection Approach by Fusing Low-Level Priors With High-Level Priors

Saliency detection has always been a challenging and interesting research area for researchers. The existing methodologies either focus on foreground regions or background regions of an image by computing low-level features. However, considering only low-level features did not produce worthy results. In this paper, low-level features, which are extracted using super pixels, are embodied with high-level priors. The background features are assumed as the low-level prior due to the similarity in the background areas and boundary of an image which are interconnected and have minimum distance in between them. High-level priors such as location, color, and semantic prior are incorporated with low-level prior to spotlight the salient area in the image. The experimental results illustrate that the proposed approach outperform the sate-of-the-art methods.

REBUS and the Anarchic Brain: Toward a Unified Model of the Brain Action of Psychedelics.

This paper formulates the action of psychedelics by integrating the free-energy principle and entropic brain hypothesis. We call this formulation relaxed beliefs under psychedelics (REBUS) and the anarchic brain, founded on the principle that—via their entropic effect on spontaneous cortical activity—psychedelics work to relax the precision of high-level priors or beliefs, thereby liberating bottom-up information flow, particularly via intrinsic sources such as the limbic system. We assemble evidence for this model and show how it can explain a broad range of phenomena associated with the psychedelic experience. With regard to their potential therapeutic use, we propose that psychedelics work to relax the precision weighting of pathologically overweighted priors underpinning various expressions of mental illness. We propose that this process entails an increased sensitization of high-level priors to bottom-up signaling (stemming from intrinsic sources), and that this heightened sensitivity enables the potential revision and deweighting of overweighted priors. We end by discussing further implications of the model, such as that psychedelics can bring about the revision of other heavily weighted high-level priors, not directly related to mental health, such as those underlying partisan and/or overly-confident political, religious, and/or philosophical perspectives.Significance StatementPsychedelics are capturing interest, with efforts underway to bring psilocybin therapy to marketing authorisation and legal access within a decade, spearheaded by the findings of a series of phase 2 trials. In this climate, a compelling unified model of how psychedelics alter brain function to alter consciousness would have appeal. Towards this end, we have sought to integrate a leading model of global brain function, hierarchical predictive coding, with an often-cited model of the acute action of psychedelics, the entropic brain hypothesis. The resulting synthesis states that psychedelics work to relax high-level priors, sensitising them to liberated bottom-up information flow, which, with the right intention, care provision and context, can help guide and cultivate the revision of entrenched pathological priors.

Robust Multiview Subspace Learning With Nonindependently and Nonidentically Distributed Complex Noise.

Multiview Subspace Learning (MSL), which aims at obtaining a low-dimensional latent subspace from multiview data, has been widely used in practical applications. Most recent MSL approaches, however, only assume a simple independent identically distributed (i.i.d.) Gaussian or Laplacian noise for all views of data, which largely underestimates the noise complexity in practical multiview data. Actually, in real cases, noises among different views generally have three specific characteristics. First, in each view, the data noise always has a complex configuration beyond a simple Gaussian or Laplacian distribution. Second, the noise distributions of different views of data are generally nonidentical and with evident distinctiveness. Third, noises among all views are nonindependent but obviously correlated. Based on such understandings, we elaborately construct a new MSL model by more faithfully and comprehensively considering all these noise characteristics. First, the noise in each view is modeled as a Dirichlet process (DP) Gaussian mixture model (DPGMM), which can fit a wider range of complex noise types than conventional Gaussian or Laplacian. Second, the DPGMM parameters in each view are different from one another, which encodes the "nonidentical" noise property. Third, the DPGMMs on all views share the same high-level priors by using the technique of hierarchical DP, which encodes the "nonindependent" noise property. All the aforementioned ideas are incorporated into an integrated graphics model which can be appropriately solved by the variational Bayes algorithm. The superiority of the proposed method is verified by experiments on 3-D reconstruction simulations, multiview face modeling, and background subtraction, as compared with the current state-of-the-art MSL methods.

A Dynamic-Shape-Prior Guided Snake Model with Application in Visually Tracking Dense Cell Populations.

This work proposes a dynamic-shape-prior guided snake model (DSP G-snake) that is designed to improve the overall stability of the point-based snake model. The dynamic shape prior is first proposed for snakes, that efficiently unifies different types of high-level priors into a new force term. To be specific, a global-topology regularity is first introduced that settles the inherent self-intersection problem with snakes. The problem that a snake's snaxels tend to unevenly distribute along the contour is also handled, leading to good parameterization. Unlike existing methods that employ learning templates or commonly enforce hard priors, the dynamic-template scheme strongly respects the deformation flexibility of the model, while retaining a decent global topology for the snake. It is verified by experiments that the proposed algorithm can effectively prevent snakes from self-crossing, or automatically untie an already selfintersected contour. In addition, the proposed model is combined with existing forces and applied to the very challenging task of tracking dense biological cell populations. The DSP G-snake model has enabled an improvement of up to 30% in tracking accuracy with respect to regular model-based approaches. Through experiments on real cellular datasets, with highly dense populations and relatively large displacements, it is confirmed that the proposed approach has enabled superior performance, in comparison to modern active-contour competitors as well as state-of-the-art cell tracking frameworks.

Sparse Gradient Based Structured Matrix Decomposition for Salient Object Detection

In the salient object detection, the given image can be decomposed into background regions (low-rank part) and salient regions (sparse part). In this paper, we present a novel sparse gradient-based structured matrix decomposition model for salient object detection. We use the ${l_{1}}$ norm of logistic function on the singular values to approximate the rank function, which avoid over-penalized problem of the nuclear norm. And a group sparsity induced norm regularization is imposed on the salient part to explore the relationship among superpixels. In order to widen the gap between salient regions and background regions in feature space, we suggest a sparse gradient regularization to replace the conventional Laplacian regularization. Finally, the model is solved through an augmented Lagrange multipliers method, and high-level priors are embedded into our model to promote the performance. Experiments indicate that the proposed method performs better in terms of various evaluation metrics than the state-of-the-art methods.

Diversity induced matrix decomposition model for salient object detection

Over the past decade, salient object detection has attracted a lot of interests in computer vision. Although many models have been proposed to detect the salient object in an arbitrary image, this problem is still plagued with complex backgrounds and scattered objects. To address this issue, in this paper, we explore the information in cross features via a diversity-induced multi-view regularization under the Hilbert-Schmidt Independence Criterion (HSIC). Based on the diversity term, a new matrix decomposition based model is proposed for salient object detection. Furthermore, S1/2 regularizer is introduced to constrain the background part. This regularizer will make the background much cleaner in the saliency map. A group sparsity induced norm is imposed on the salient part in order to involve the potential spatial relationships of image patches. Our method is solved through an augmented Lagrange multipliers method, and high-level priors are also integrated to boost the performance. Experiments on the four widely used datasets show that our method outperforms the state-of-the-art models.

Salient Object Detection via Structured Matrix Decomposition.

Low-rank recovery models have shown potential for salient object detection, where a matrix is decomposed into a low-rank matrix representing image background and a sparse matrix identifying salient objects. Two deficiencies, however, still exist. First, previous work typically assumes the elements in the sparse matrix are mutually independent, ignoring the spatial and pattern relations of image regions. Second, when the low-rank and sparse matrices are relatively coherent, e.g., when there are similarities between the salient objects and background or when the background is complicated, it is difficult for previous models to disentangle them. To address these problems, we propose a novel structured matrix decomposition model with two structural regularizations: (1) a tree-structured sparsity-inducing regularization that captures the image structure and enforces patches from the same object to have similar saliency values, and (2) a Laplacian regularization that enlarges the gaps between salient objects and the background in feature space. Furthermore, high-level priors are integrated to guide the matrix decomposition and boost the detection. We evaluate our model for salient object detection on five challenging datasets including single object, multiple objects and complex scene images, and show competitive results as compared with 24 state-of-the-art methods in terms of seven performance metrics.

Global feature integration based salient region detection

The goal of saliency detection is to locate the regions which are most likely to capture human׳s attention without prior knowledge of their contents. Visual saliency detection has been widely used in image processing, but it is still a challenging problem in computer vision. In this paper, we propose a salient region detection algorithm by integrating global features namely uniqueness and spatial distribution. Two measures of contrast are computed in pixel and superpixel level respectively. In order to suppress background noise, Low-level features are refined by High-level priors which are computed with the Gaussian model based on salient region. We formulate salient region detection as a binary labeling problem that separates salient region from the background. A Conditional Random Field is learned to effectively combine these refined features for salient region detection. Experimental results on the large benchmark database demonstrate the proposed method performs well when against fifteen state-of-the-art methods in terms of precision and recall.

Low-level and high-level prior learning for visual saliency estimation

Visual saliency estimation is an important issue in multimedia modeling and computer vision, and constitutes a research field that has been studied for decades. Many approaches have been proposed to solve this problem. In this study, we consider the visual attention problem blue with respect to two aspects: low-level prior learning and high-level prior learning. On the one hand, inspired by the concept of chance of happening, the low-level priors, i.e., Color Statistics-based Priors (CSP) and Spatial Correlation-based Priors (SCP), are learned to describe the color distribution and contrast distribution in natural images. On the other hand, the high-level priors, i.e., the relative relationships between objects, are learned to describe the conditional priority between different objects in the images. In particular, we first learn the low-level priors that are statistically based on a large set of natural images. Then, the high-level priors are learned to construct a conditional probability matrix blue that reflects the relative relationship between different objects. Subsequently, a saliency model is presented by integrating the low-level priors, the high-level priors and the Center Bias Prior (CBP), in which the weights that correspond to the low-level priors and the high-level priors are learned based on the eye tracking data set. The experimental results demonstrate that our approach outperforms the existing techniques.