scholarly journals A Predictive Processing Account of Bottom-Up Visual Saliency Using Cross-Predicting Autoencoders

2018 ◽  
Author(s):  
Beren Millidge ◽  
Richard Shillcock
Keyword(s):  
Prediction Error ◽  
Visual Saliency ◽  
Generative Models ◽  
Predictive Processing ◽  
Bottom Up ◽  
Low Level ◽  
Sense Data ◽  
Saliency Maps ◽  
Human Participants ◽  
The Brain

We propose a novel predictive processing account of bottom-up visual saliency in which salience is simply the low-level prediction error between the sense-data and the predictions produced by the generative models in the brain. We test this with modelling in which we use cross-predicting deep autoencoders to create salience maps in an entirely unsupervised way. The resulting maps closely mimic experimentally derived human saliency maps and also spontaneously learn a centre bias, a robust viewing behaviour seen in human participants.

scholarly journals Toward the unity of pathological and exertional fatigue: A predictive processing model

2021 ◽  
Author(s):  
A. Greenhouse-Tucknott ◽  
J. B. Butterworth ◽  
J. G. Wrightson ◽  
N. J. Smeeton ◽  
H. D. Critchley ◽  
...  
Keyword(s):  
Prediction Error ◽  
Conscious Experience ◽  
Physical Exertion ◽  
Generative Models ◽  
The Body ◽  
Predictive Processing ◽  
Research Fields ◽  
Common Foundation ◽  
Health And Disease ◽  
Sensory Evidence

AbstractFatigue is a common experience in both health and disease. Yet, pathological (i.e., prolonged or chronic) and transient (i.e., exertional) fatigue symptoms are traditionally considered distinct, compounding a separation between interested research fields within the study of fatigue. Within the clinical neurosciences, nascent frameworks position pathological fatigue as a product of inference derived through hierarchical predictive processing. The metacognitive theory of dyshomeostasis (Stephan et al., 2016) states that pathological fatigue emerges from the metacognitive mechanism in which the detection of persistent mismatches between prior interoceptive predictions and ascending sensory evidence (i.e., prediction error) signals low evidence for internal generative models, which undermine an agent’s feeling of mastery over the body and is thus experienced phenomenologically as fatigue. Although acute, transient subjective symptoms of exertional fatigue have also been associated with increasing interoceptive prediction error, the dynamic computations that underlie its development have not been clearly defined. Here, drawing on the metacognitive theory of dyshomeostasis, we extend this account to offer an explicit description of the development of fatigue during extended periods of (physical) exertion. Accordingly, it is proposed that a loss of certainty or confidence in control predictions in response to persistent detection of prediction error features as a common foundation for the conscious experience of both pathological and nonpathological fatigue.

scholarly journals Active interoceptive inference and the emotional brain

2016 ◽  
Vol 371 (1708) ◽  
pp. 20160007 ◽  
Author(s):  
Anil K. Seth ◽  
Karl J. Friston
Keyword(s):  
Functional Anatomy ◽  
Generative Models ◽  
Predictive Processing ◽  
Clinical Neuroscience ◽  
Autonomic Reflexes ◽  
Embodied Self ◽  
Bodily States ◽  
Theoretical Issues ◽  
The Brain

We review a recent shift in conceptions of interoception and its relationship to hierarchical inference in the brain. The notion of interoceptive inference means that bodily states are regulated by autonomic reflexes that are enslaved by descending predictions from deep generative models of our internal and external milieu. This re-conceptualization illuminates several issues in cognitive and clinical neuroscience with implications for experiences of selfhood and emotion. We first contextualize interoception in terms of active (Bayesian) inference in the brain, highlighting its enactivist (embodied) aspects. We then consider the key role of uncertainty or precision and how this might translate into neuromodulation. We next examine the implications for understanding the functional anatomy of the emotional brain, surveying recent observations on agranular cortex. Finally, we turn to theoretical issues, namely, the role of interoception in shaping a sense of embodied self and feelings. We will draw links between physiological homoeostasis and allostasis, early cybernetic ideas of predictive control and hierarchical generative models in predictive processing. The explanatory scope of interoceptive inference ranges from explanations for autism and depression, through to consciousness. We offer a brief survey of these exciting developments. This article is part of the themed issue ‘Interoception beyond homeostasis: affect, cognition and mental health’.

Top-Down Predictions Determine Perceptions

2017 ◽  
Author(s):  
Martin V. Butz ◽  
Esther F. Kutter
Keyword(s):  
Graphical Models ◽  
Information Integration ◽  
Visual Processing ◽  
Visual Information ◽  
Generative Models ◽  
Sources Of Information ◽  
Multiple Sources ◽  
Top Down ◽  
Bottom Up ◽  
The Brain

While bottom-up visual processing is important, the brain integrates this information with top-down, generative expectations from very early on in the visual processing hierarchy. Indeed, our brain should not be viewed as a classification system, but rather as a generative system, which perceives something by integrating sensory evidence with the available, learned, predictive knowledge about that thing. The involved generative models continuously produce expectations over time, across space, and from abstracted encodings to more concrete encodings. Bayesian information processing is the key to understand how information integration must work computationally – at least in approximation – also in the brain. Bayesian networks in the form of graphical models allow the modularization of information and the factorization of interactions, which can strongly improve the efficiency of generative models. The resulting generative models essentially produce state estimations in the form of probability densities, which are very well-suited to integrate multiple sources of information, including top-down and bottom-up ones. A hierarchical neural visual processing architecture illustrates this point even further. Finally, some well-known visual illusions are shown and the perceptions are explained by means of generative, information integrating, perceptual processes, which in all cases combine top-down prior knowledge and expectations about objects and environments with the available, bottom-up visual information.

ARTISTIC COMPLEXITY AND SALIENCY: TWO FACES OF THE SAME COIN?

2013 ◽  
Vol 09 (02) ◽  
pp. 1350010 ◽  
Author(s):  
MATTEO CACCIOLA ◽  
GIANLUIGI OCCHIUTO ◽  
FRANCESCO CARLO MORABITO
Keyword(s):  
Information Content ◽  
Visual Saliency ◽  
Relevant Information ◽  
Human Observer ◽  
Top Down ◽  
Complexity Measures ◽  
Bottom Up ◽  
Content Description ◽  
Brain Processing ◽  
The Brain

Many computer vision problems consist of making a suitable content description of images usually aiming to extract the relevant information content. In case of images representing paintings or artworks, the information extracted is rather subject-dependent, thus escaping any universal quantification. However, we proposed a measure of complexity of such kinds of oeuvres which is related to brain processing. The artistic complexity measures the brain inability to categorize complex nonsense forms represented in modern art, in a dynamic process of acquisition that most involves top-down mechanisms. Here, we compare the quantitative results of our analysis on a wide set of paintings of various artists to the cues extracted from a standard bottom-up approach based on visual saliency concept. In every painting inspection, the brain searches for more informative areas at different scales, then connecting them in an attempt to capture the full impact of information content. Artistic complexity is able to quantify information which might have been individually lost in the fruition of a human observer thus identifying the artistic hand. Visual saliency highlights the most salient areas of the paintings standing out from their neighbours and grabbing our attention. Nevertheless, we will show that a comparison on the ways the two algorithms act, may manifest some interesting links, finally indicating an interplay between bottom-up and top-down modalities.

scholarly journals Gender fluidity as affordance negotiation

2020 ◽  
Author(s):  
Mahault Albarracin ◽  
Pierre Poirier
Keyword(s):  
Prediction Error ◽  
Probabilistic Models ◽  
Predictive Processing ◽  
Gender Fluidity ◽  
Social Scripts ◽  
Sensorimotor Interaction ◽  
Binary State ◽  
Gender Scripts ◽  
Cultural Niche ◽  
The Brain

Gender is often viewed as static binary state for people to embody, based on the sex they were assigned at birth. However, cultural studies increasingly understand gender as neither binary nor static, a view supported both in psychology and sociology. On this view, gender is negotiated between individuals, and highly dependent on context. Specifically, individuals are thought to be offered culturally gendered social scripts that allow them and their interlocutors the ability to predict future actions, and to understand the scene being set, establishing roles and expectations. We propose to understand scripts in the framework of enactive-ecological predictivism, which integrates aspects of ecological enactivism, notably the importance of dynamical sensorimotor interaction with an environment conceived as a field of affordances, and predictive processing, which views the brain as a predictive engine that builds its probabilistic models in an effort to reduce prediction error. Under this view, script-based negotiation can be linked to the enactive neuroscience concept of a cultural niche, as a landscape of cultural affordances. Affordances are possibilities for action that constrain what actions are pre-reflectively felt possible based on biological, experiential and cultural multisensorial cues. With the shifting landscapes of cultural affordances brought about by a number of recent social, technological and epistemic developments, the gender scripts offered to individuals are less culturally rigid, which translates in an increase in the variety of affordance fields each individual can negotiate. This entails that any individual has an increased possibility for gender fluidity, as shown by the increasing number of people currently identifying outside the binary.

scholarly journals Adjudicating Between Local and Global Architectures of Predictive Processing in the Subcortical Auditory Pathway

2021 ◽  
Vol 15 ◽  
Author(s):  
Alejandro Tabas ◽  
Katharina von Kriegstein
Keyword(s):  
Cerebral Cortex ◽  
Prediction Error ◽  
Sensory Input ◽  
Generative Models ◽  
Predictive Processing ◽  
Auditory Modality ◽  
Processing Stage ◽  
Information Encoding ◽  
Global View ◽  
Local View

Predictive processing, a leading theoretical framework for sensory processing, suggests that the brain constantly generates predictions on the sensory world and that perception emerges from the comparison between these predictions and the actual sensory input. This requires two distinct neural elements: generative units, which encode the model of the sensory world; and prediction error units, which compare these predictions against the sensory input. Although predictive processing is generally portrayed as a theory of cerebral cortex function, animal and human studies over the last decade have robustly shown the ubiquitous presence of prediction error responses in several nuclei of the auditory, somatosensory, and visual subcortical pathways. In the auditory modality, prediction error is typically elicited using so-called oddball paradigms, where sequences of repeated pure tones with the same pitch are at unpredictable intervals substituted by a tone of deviant frequency. Repeated sounds become predictable promptly and elicit decreasing prediction error; deviant tones break these predictions and elicit large prediction errors. The simplicity of the rules inducing predictability make oddball paradigms agnostic about the origin of the predictions. Here, we introduce two possible models of the organizational topology of the predictive processing auditory network: (1) the global view, that assumes that predictions on the sensory input are generated at high-order levels of the cerebral cortex and transmitted in a cascade of generative models to the subcortical sensory pathways; and (2) the local view, that assumes that independent local models, computed using local information, are used to perform predictions at each processing stage. In the global view information encoding is optimized globally but biases sensory representations along the entire brain according to the subjective views of the observer. The local view results in a diminished coding efficiency, but guarantees in return a robust encoding of the features of sensory input at each processing stage. Although most experimental results to-date are ambiguous in this respect, recent evidence favors the global model.

scholarly journals A tale of two densities: active inference is enactive inference

2019 ◽  
Vol 28 (4) ◽  
pp. 225-239 ◽  
Author(s):  
Maxwell JD Ramstead ◽  
Michael D Kirchhoff ◽  
Karl J Friston
Keyword(s):  
Predictive Coding ◽  
Generative Models ◽  
Predictive Processing ◽  
Active Inference ◽  
Energy Principle ◽  
Free Energy Principle ◽  
The One ◽  
And Control ◽  
The Brain ◽  
Selection Of

The aim of this article is to clarify how best to interpret some of the central constructs that underwrite the free-energy principle (FEP) – and its corollary, active inference – in theoretical neuroscience and biology: namely, the role that generative models and variational densities play in this theory. We argue that these constructs have been systematically misrepresented in the literature, because of the conflation between the FEP and active inference, on the one hand, and distinct (albeit closely related) Bayesian formulations, centred on the brain – variously known as predictive processing, predictive coding or the prediction error minimisation framework. More specifically, we examine two contrasting interpretations of these models: a structural representationalist interpretation and an enactive interpretation. We argue that the structural representationalist interpretation of generative and recognition models does not do justice to the role that these constructs play in active inference under the FEP. We propose an enactive interpretation of active inference – what might be called enactive inference. In active inference under the FEP, the generative and recognition models are best cast as realising inference and control – the self-organising, belief-guided selection of action policies – and do not have the properties ascribed by structural representationalists.

scholarly journals Visual Saliency via Multiscale Analysis in Frequency Domain and Its Applications to Ship Detection in Optical Satellite Images

2022 ◽  
Vol 15 ◽  
Author(s):  
Ying Yu ◽  
Jun Qian ◽  
Qinglong Wu
Keyword(s):  
Wavelet Transform ◽  
Frequency Domain ◽  
Multiscale Analysis ◽  
Satellite Images ◽  
Visual Saliency ◽  
Bottom Up ◽  
Ship Detection ◽  
Saliency Maps ◽  
Proposed Model ◽  
Optical Satellite Images

This article proposes a bottom-up visual saliency model that uses the wavelet transform to conduct multiscale analysis and computation in the frequency domain. First, we compute the multiscale magnitude spectra by performing a wavelet transform to decompose the magnitude spectrum of the discrete cosine coefficients of an input image. Next, we obtain multiple saliency maps of different spatial scales through an inverse transformation from the frequency domain to the spatial domain, which utilizes the discrete cosine magnitude spectra after multiscale wavelet decomposition. Then, we employ an evaluation function to automatically select the two best multiscale saliency maps. A final saliency map is generated via an adaptive integration of the two selected multiscale saliency maps. The proposed model is fast, efficient, and can simultaneously detect salient regions or objects of different sizes. It outperforms state-of-the-art bottom-up saliency approaches in the experiments of psychophysical consistency, eye fixation prediction, and saliency detection for natural images. In addition, the proposed model is applied to automatic ship detection in optical satellite images. Ship detection tests on satellite data of visual optical spectrum not only demonstrate our saliency model's effectiveness in detecting small and large salient targets but also verify its robustness against various sea background disturbances.

scholarly journals Dopamine modulates prediction error forwarding in the nonlemniscal inferior colliculus

2019 ◽  
Author(s):  
Catalina Valdés-Baizabal ◽  
Guillermo V. Carbajal ◽  
David Pérez-González ◽  
Manuel S. Malmierca
Keyword(s):  
Inferior Colliculus ◽  
Prediction Error ◽  
Sensory Input ◽  
Population Level ◽  
Auditory Pathway ◽  
Synaptic Activity ◽  
Predictive Processing ◽  
Bottom Up ◽  
Perceptual Model ◽  
Dopaminergic Systems

AbstractThe predictive processing framework describes perception as a hierarchical predictive model of sensation. Higher-level neural structures constrain the processing at lower-level structures by suppressing synaptic activity induced by predictable sensory input. But when predictions fail, deviant input is forwarded bottom-up as ‘prediction error’ to update the perceptual model. The earliest prediction error signals identified in the auditory pathway emerge from the nonlemniscal inferior colliculus (IC). The drive that these feedback signals exert on the perceptual model depends on their ‘expected precision’, which determines the postsynaptic gain applied in prediction error forwarding. Expected precision is theoretically encoded by the neuromodulatory (e.g., dopaminergic) systems. To test this empirically, we recorded extracellular responses from the rat nonlemniscal IC to oddball and cascade sequences before, during and after the microiontophoretic application of dopamine or eticlopride (a D2-like receptor antagonist). Hence, we studied dopaminergic modulation on the subcortical processing of unpredictable and predictable auditory changes. Results demonstrate that dopamine reduces the net neuronal responsiveness exclusively to unexpected input, without significantly altering the processing of expected auditory events at population level. We propose that, in natural conditions, dopaminergic projections from the thalamic subparafascicular nucleus to the nonlemniscal IC could serve as a precision-weighting mechanism mediated by D2-like receptors. Thereby, the levels of dopamine release in the nonlemniscal IC could modulate the early bottom-up flow of prediction error signals in the auditory system by encoding their expected precision.

scholarly journals Attenuated sensory precision and contextual prediction error in persistent depression

2021 ◽  
Author(s):  
Anna Bevan ◽  
Caitlin Hitchcock ◽  
Daniel Mitchell ◽  
Tim Dalgleish
Keyword(s):  
Prediction Error ◽  
Covert Attention ◽  
Predictive Processing ◽  
Attention Task ◽  
Sensory Data ◽  
New Information ◽  
Sensory Modalities ◽  
Persistent Depression ◽  
The Brain ◽  
High Uncertainty

Chronic and recurrent forms of clinical depression can persist for a lifetime and often respond poorly to intervention. Psychological formulations implicate rigid, negative expectations of self and world which are resistant to updating with new information, a phenomenology consistent with a Bayesian account of brain function. Bayesian predictive processing models suggest that sensory data which is represented with low precision (high uncertainty) in the brain cannot exert much influence on existing beliefs, giving rise to the hypothesis that persistent forms of depression may be characterised by disturbances in sensory precision optimization. We optimized a computational model with data from a cross-modal (visual, auditory, somatic) covert attention task to estimate sensory precision in persistently depressed participants relative to healthy controls. Results suggested that both sensory precision and the salience of attentional targets were attenuated in depressed participants across sensory modalities, contributing to a suppression of contextual prediction error in this group. These outcomes provide support for a novel theoretical account of depression chronicity and suggest avenues for enhancing the effectiveness of psychological interventions for this population.

Sign in / Sign up

Export Citation Format

Share Document