Predictive Coding Theory Research Articles

Disentangling predictive processing in the brain: a meta-analytic study in favour of a predictive network

According to the predictive coding (PC) theory, the brain is constantly engaged in predicting its upcoming states and refining these predictions through error signals. Despite extensive research investigating the neural bases of this theory, to date no previous study has systematically attempted to define the neural mechanisms of predictive coding across studies and sensory channels, focussing on functional connectivity. In this study, we employ a coordinate-based meta-analytical approach to address this issue. We first use the Activation Likelihood Estimation (ALE) algorithm to detect spatial convergence across studies, related to prediction error and encoding. Overall, our ALE results suggest the ultimate role of the left inferior frontal gyrus and left insula in both processes. Moreover, we employ a meta-analytic connectivity method (Seed-Voxel Correlations Consensus). This technique reveals a large, bilateral predictive network, which resembles large-scale networks involved in task-driven attention and execution. In sum, we find that: (i) predictive processing seems to occur more in certain brain regions than others, when considering different sensory modalities at a time; (ii) there is no evidence, at the network level, for a distinction between error and prediction processing.

Structural and contextual priors affect visual search in children with and without autism.

Bayesian predictive coding theories of autism spectrum disorder propose that impaired acquisition or a broader shape of prior probability distributions lies at the core of the condition. However, we still know very little about how probability distributions are learned and encoded by children, let alone children with autism. Here, we take advantage of a recently developed distribution learning paradigm to characterize how children with and without autism acquire information about probability distributions. Twenty-four autistic and 25-matched neurotypical children searched for an odd-one-out target among a set of distractor lines with orientations sampled from a Gaussian distribution repeated across multiple trials to allow for learning of the parameters (mean and variance) of the distribution. We could measure the width (variance) of the participant's encoded distribution by introducing a target-distractor role-reversal while varying the similarity between target and previous distractor mean. Both groups performed similarly on the visual search task and learned the distractor distribution to a similar extent. However, the variance learned was much broader than the one presented, consistent with less informative priors in children irrespective of autism diagnosis. These findings have important implications for Bayesian accounts of perception throughout development, and Bayesian accounts of autism specifically. LAY SUMMARY: Recent theories about the underlying cognitive mechanisms of autism propose that the way autistic individuals estimate variability or uncertainty in their perceptual environment may differ from how typical individuals do so. Children had to search an oddly tilted line in a set of lines pointing in different directions, and based on their response times we examined how they learned about the variability in a set of objects. We found that autistic children learn variability as well as typical children, but both groups learn with less precision than typical adults do on the same task.

The encoding of stochastic regularities is facilitated by action-effect predictions

Our brains continuously build and update predictive models of the world, sources of prediction being drawn for example from sensory regularities and/or our own actions. Yet, recent results in the auditory system indicate that stochastic regularities may not be easily encoded when a rare medium pitch deviant is presented between frequent high and low pitch standard sounds in random order, as reflected in the lack of sensory prediction error event-related potentials [i.e., mismatch negativity (MMN)]. We wanted to test the implication of the predictive coding theory that predictions based on higher-order generative models—here, based on action intention, are fed top-down in the hierarchy to sensory levels. Participants produced random sequences of high and low pitch sounds by button presses in two conditions: In a “specific” condition, one button produced high and the other low pitch sounds; in an “unspecific” condition, both buttons randomly produced high or low-pitch sounds. Rare medium pitch deviants elicited larger MMN and N2 responses in the “specific” compared to the “unspecific” condition, despite equal sound probabilities. These results thus demonstrate that action-effect predictions can boost stochastic regularity-based predictions and engage higher-order deviance detection processes, extending previous notions on the role of action predictions at sensory levels.

Limited Evidence for Sensory Prediction Error Responses in Visual Cortex of Macaques and Humans.

A recent formulation of predictive coding theory proposes that a subset of neurons in each cortical area encodes sensory prediction errors, the difference between predictions relayed from higher cortex and the sensory input. Here, we test for evidence of prediction error responses in spiking responses and local field potentials (LFP) recorded in primary visual cortex and area V4 of macaque monkeys, and in complementary electroencephalographic (EEG) scalp recordings in human participants. We presented a fixed sequence of visual stimuli on most trials, and violated the expected ordering on a small subset of trials. Under predictive coding theory, pattern-violating stimuli should trigger robust prediction errors, but we found that spiking, LFP and EEG responses to expected and pattern-violating stimuli were nearly identical. Our results challenge the assertion that a fundamental computational motif in sensory cortex is to signal prediction errors, at least those based on predictions derived from temporal patterns of visual stimulation.

Modulatory Effects of Prediction Accuracy on Electroencephalographic Brain Activity During Prediction.

Prediction is essential for the efficiency of many cognitive processes; however, this process is not always perfect. Predictive coding theory suggests that the brain generates and updates a prediction to respond to an upcoming event. Although an electrophysiological index of prediction, the stimulus preceding negativity (SPN), has been reported, it remains unknown whether the SPN reflects the prediction accuracy, or whether it is associated with the prediction error, which corresponds to a mismatch between a prediction and an actual input. Thus, the present study aimed to investigate this question using electroencephalography (EEG). Participants were asked to predict the original pictures from pictures that had undergone different levels of pixelation. The SPN amplitude was affected by the level of pixelation and correlated with the subjective evaluation of the prediction accuracy. Furthermore, late positive components (LPC) were negatively correlated with SPN. These results suggest that the amplitude of SPN reflects the prediction accuracy; more accurate prediction increases the SPN and reduces the prediction error, resulting in reduced LPC amplitudes.

The Predictive Coding Theory of Cognition and Mental Representations

The Predictive Coding Theory of Cognition and Mental Representations

Priors Bias Perceptual Decisions in Autism, But Are Less Flexibly Adjusted to the Context.

According to the predictive coding framework, percepts emerge from combinations of sensory input and prior knowledge, whose relative contributions depend on their reliability. Recent predictive coding theories suggest that Autism Spectrum Disorder (ASD) could be characterized by an atypical weighting of priors. Here, we assessed whether individuals with ASD can flexibly adjust the weight (precision) of the prior to the context. Thirty-one neurotypical adults (NT) and 26 adults with ASD participated in a visual discrimination task designed to elicit a time-order effect (TOE). The TOE reflects the integration of priors with sensory estimates. We used two experimental contexts: a narrow stimulus range (Narrow condition) and a broader range (Broad condition) in order to induce a prior with a higher and lower precision, respectively. Both groups learned a prior that biased their perception, as shown with the TOE. As expected, the NT group had a larger TOE in the Narrow condition than in the Broad condition, revealing a contextual adjustment of the prior precision. In contrast, ASD participants were more inflexible: the extent of the TOE was not modulated by the context. In addition, the accuracy increased when the stimulus range decreased in both group, which may be interpreted as a contextual adjustment of the sensory precision. To conclude, adults with and without ASD implicitly learned a prior mean, but ASD participants failed to flexibly adjust the prior precision to the context. This increased inflexibility in ASD could account for many symptoms, such as their intolerance of uncertainty. LAY SUMMARY: Based on our experience, we have expectations about our environment. Theories suggest that the symptoms encountered in autism could be due to atypical expectations, leading to an impression of an unpredictable world. Using a visual discrimination task, we showed that adults with and without autism were biased by their expectations. Yet, the extent to which expectations biased perception did not depend on the context in autism. This higher inflexibility found in autism may explain symptoms such as resistance to change.

Hierarchical Predictive Coding-Based JND Estimation for Image Compression.

The human visual system (HVS) is a hierarchical system, in which visual signals are processed hierarchically. In this paper, the HVS is modeled as a three-level communication system and visual perception is divided into three stages according to the hierarchical predictive coding theory. Then, a novel just noticeable distortion (JND) estimation scheme is proposed. In visual perception, the input signals are predicted constantly and spontaneously in each hierarchy, and neural response is evoked by the central residue and inhibited by surrounding residues. These two types' residues are regarded as the positive and negative visual incentives which cause positive and negative perception effects, respectively. In neuroscience, the effect of incentive on observer is measured by the surprise of this incentive. Thus, we propose a surprise-based measurement method to measure both perception effects. Specifically, considering the biased competition of visual attention, we define the product of the residue self-information (i.e., surprise) and the competition biases as the perceptual surprise to measure the positive perception effect. As for the negative perception effect, it is measured by the average surprise (i.e., the local Shannon entropy). The JND threshold of each stage is estimated individually by considering both perception effects. The total JND threshold is finally obtained by non-linear superposition of three stage thresholds. Furthermore, the proposed JND estimation scheme is incorporated into the codec of Versatile Video Coding for image compression. Experimental results show that the proposed JND model outperforms the relevant existing ones, and over 16% of bit rate can be reduced without jeopardizing the perceptual quality.

Detecting (Un)seen Change: The Neural Underpinnings of (Un)conscious Prediction Errors.

Detecting changes in the environment is fundamental for our survival. According to predictive coding theory, detecting these irregularities relies both on incoming sensory information and our top–down prior expectations (or internal generative models) about the world. Prediction errors (PEs), detectable in event-related potentials (ERPs), occur when there is a mismatch between the sensory input and our internal model (i.e., a surprise event). Many changes occurring in our environment are irrelevant for survival and may remain unseen. Such changes, even if subtle, can nevertheless be detected by the brain without emerging into consciousness. What remains unclear is how these changes are processed in the brain at the network level. Here, we used a visual oddball paradigm in which participants engaged in a central letter task during electroencephalographic (EEG) recordings while presented with task-irrelevant high- or low-coherence background, random-dot motion. Critically, once in a while, the direction of the dots changed. After the EEG session, we confirmed that changes in motion direction at high- and low-coherence were visible and invisible, respectively, using psychophysical measurements. ERP analyses revealed that changes in motion direction elicited PE regardless of the visibility, but with distinct spatiotemporal patterns. To understand these responses, we applied dynamic causal modeling (DCM) to the EEG data. Bayesian Model Averaging showed visible PE relied on a release from adaptation (repetition suppression) within bilateral MT+, whereas invisible PE relied on adaptation at bilateral V1 (and left MT+). Furthermore, while feedforward upregulation was present for invisible PE, the visible change PE also included downregulation of feedback between right MT+ to V1. Our findings reveal a complex interplay of modulation in the generative network models underlying visible and invisible motion changes.

Effect of Top-Down Connections in Hierarchical Sparse Coding

Hierarchical sparse coding (HSC) is a powerful model to efficiently represent multidimensional, structured data such as images. The simplest solution to solve this computationally hard problem is to decompose it into independent layer-wise subproblems. However, neuroscientific evidence would suggest interconnecting these subproblems as in predictive coding (PC) theory, which adds top-down connections between consecutive layers. In this study, we introduce a new model, 2-layer sparse predictive coding (2L-SPC), to assess the impact of this interlayer feedback connection. In particular, the 2L-SPC is compared with a hierarchical Lasso (Hi-La) network made out of a sequence of independent Lasso layers. The 2L-SPC and a 2-layer Hi-La networks are trained on four different databases and with different sparsity parameters on each layer. First, we show that the overall prediction error generated by 2L-SPC is lower thanks to the feedback mechanism as it transfers prediction error between layers. Second, we demonstrate that the inference stage of the 2L-SPC is faster to converge and generates a refined representation in the second layer compared to the Hi-La model. Third, we show that the 2L-SPC top-down connection accelerates the learning process of the HSC problem. Finally, the analysis of the emerging dictionaries shows that the 2L-SPC features are more generic and present a larger spatial extension.

Shannon + Friston = Content: Intentionality in predictive signaling systems

What is the content of a mental state? This question poses the problem of intentionality: to explain how mental states can be about other things, where being about them is understood as representing them. A framework that integrates predictive coding and signaling systems theories of cognitive processing offers a new perspective on intentionality. On this view, at least some mental states are evaluations, which differ in function, operation, and normativity from representations. A complete naturalistic theory of intentionality must account for both types of intentional state.

The Impact of a Novel Mimicry Task for Increasing Emotion Recognition in Adults with Autism Spectrum Disorder and Alexithymia: Protocol for a Randomized Controlled Trial.

BackgroundImpaired facial emotion expression recognition (FEER) has typically been considered a correlate of autism spectrum disorder (ASD). Now, the alexithymia hypothesis is suggesting that this emotion processing problem is instead related to alexithymia, which frequently co-occurs with ASD. By combining predictive coding theories of ASD and simulation theories of emotion recognition, it is suggested that facial mimicry may improve the training of FEER in ASD and alexithymia.ObjectiveThis study aims to evaluate a novel mimicry task to improve FEER in adults with and without ASD and alexithymia. Additionally, this study will aim to determine the contributions of alexithymia and ASD to FEER ability and assess which of these 2 populations benefit from this training task.MethodsRecruitment will primarily take place through an ASD community group with emphasis put on snowball recruiting. Included will be 64 consenting adults equally divided between participants without an ASD and participants with an ASD. Participants will be screened online using the Kessler Psychological Distress Scale (K-10; cut-off score of 22), Autism Spectrum Quotient (AQ-10), and Toronto Alexithymia Scale (TAS-20) followed by a clinical interview with a provisional psychologist at the Federation University psychology clinic. The clinical interview will include assessment of ability, anxiety, and depression as well as discussion of past ASD diagnosis and confirmatory administration of the Autism Mental Status Exam (AMSE). Following the clinical interview, the participant will complete the Bermond-Vorst Alexithymia Questionnaire (BVAQ) and then undertake a baseline assessment of FEER. Consenting participants will then be assigned using a permuted blocked randomization method into either the control task condition or the mimicry task condition. A brief measure of satisfaction of the task and a debriefing session will conclude the study.ResultsThe study has Federation University Human Research Ethics Committee approval and is registered with the Australian New Zealand Clinical Trials. Participant recruitment is predicted to begin in the third quarter of 2021.ConclusionsThis study will be the first to evaluate the use of a novel facial mimicry task condition to increase FEER in adults with ASD and alexithymia. If efficacious, this task could prove useful as a cost-effective adjunct intervention that could be used at home and thus remove barriers to entry. This study will also explore the unique effectiveness of this task in people without an ASD, with an ASD, and with alexithymia.Trial RegistrationAustralian New Zealand Clinical Trial Registry ACTRN12619000705189p; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=377455International Registered Report Identifier (IRRID)PRR1-10.2196/24543

Visual Sensory Cortices Causally Contribute to Auditory Word Recognition Following Sensorimotor-Enriched Vocabulary Training.

Despite a rise in the use of “learning by doing” pedagogical methods in praxis, little is known as to how the brain benefits from these methods. Learning by doing strategies that utilize complementary information (“enrichment”) such as gestures have been shown to optimize learning outcomes in several domains including foreign language (L2) training. Here we tested the hypothesis that behavioral benefits of gesture-based enrichment are critically supported by integrity of the biological motion visual cortices (bmSTS). Prior functional neuroimaging work has implicated the visual motion cortices in L2 translation following sensorimotor-enriched training; the current study is the first to investigate the causal relevance of these structures in learning by doing contexts. Using neuronavigated transcranial magnetic stimulation and a gesture-enriched L2 vocabulary learning paradigm, we found that the bmSTS causally contributed to behavioral benefits of gesture-enriched learning. Visual motion cortex integrity benefitted both short- and long-term learning outcomes, as well as the learning of concrete and abstract words. These results adjudicate between opposing predictions of two neuroscientific learning theories: While reactivation-based theories predict no functional role of specialized sensory cortices in vocabulary learning outcomes, the current study supports the predictive coding theory view that these cortices precipitate sensorimotor-based learning benefits.

Segregation, connectivity, and gradients of deactivation in neural correlates of evidence in social decision making

Functional imaging studies of sensory decision making have detected a signal associated with evidence for decisions that is consistent with data from single-cell recordings in laboratory animals. However, the generality of this finding and its implications on our understanding of the organization of the fMRI signal are not clear. In the present functional imaging study, we investigated decisions in an elementary social cognition domain to identify the neural correlates of evidence, their segregation, connectivity, and their relationship to task deactivations. Besides providing data in support of an evidence-related signal in a social cognition task, we were interested in embedding these neural correlates in models of supramodal associative cortex placed at the top of a hierarchy of processing areas. Participants were asked to decide which of two depicted individuals was saddest based on information rich in sensory features (facial expressions) or through contextual cues suggesting the mental state of others (stylized drawings of mourning individuals). The signal associated with evidence for the decision was located in two distinct networks differentially recruited depending on the information type. Using the largest peaks of the signal associated with evidence as seeds in a database of connectivity data, these two networks were retrieved. Furthermore, the hubs of these networks were located near or along a ribbon of cortex located between task activations and deactivations between areas affected by perceptual priming and the deactivated areas of the default network system. In associative cortex, these findings suggest gradients of progressive relative deactivation as a possible neural correlate of the cortical organization envisaged by structural models of cortical organization and by predictive coding theories of cortical function.

Neuro-aesthetics and the iconography in photography.

Can neurosciences explain art? No, but it can help us to understand why some images are more memorable and, thus, more successful than others. This article aims to identify certain factors that may influence the artistic success of photographic images. These factors are discussed within the contexts of basic neuropsychological concepts, visual perception, and visual memory. A new computational and neuroscientifically based model, the predictive coding theory, provides a powerful framework for integrating social and individual factors that influence aesthetic experience and activity. A case study of Dorothea Lange's iconic photograph Migrant Mother demonstrates the importance of identifiable factors that influence and determine a photograph's potential success. We are convinced that a future systemic approach will enable the complementary integration of neuroscientific, perceptual, cognitive, emotional, and sociopsychological insights through the framework of predictive coding theory with socioscientific, art-theoretical, and art-historical as well as neuro- and behavioral-economical models.

Coronavirus : cerveau prédictif et gestion de la terreur

Emerging infectious diseases like Covid-19 cause a major threat to global health. When confronted with new pathogens, individuals generate several beliefs about the epidemic phenomenon. Many studies have shown that individual protective behaviors largely depend on these beliefs. Due to the absence of treatment and vaccine against these emerging pathogens, the relation between these beliefs and these behaviors represents a crucial issue for public health policies. In the premises of the Covid-19 pandemic, several preliminary studies have highlighted a delay in the perception of risk by individuals, which potentially holds back the implementing of the necessary precautionary measures: people underestimated the risks associated with the virus, and therefore also the importance of complying with sanitary guidelines. During the peak of the pandemic, the salience of the threat and of the risk of mortality could then have transformed the way people generate their beliefs. This potentially leads to upheavals in the way they understand the world. Here, we propose to explore the evolution of beliefs and behaviors during the Covid-19 crisis, using the theory of predictive coding and the theory of terror management, two influential frameworks in cognitive science and in social psychology.

T64. LINKING SUBCLINICAL PERSECUTORY IDEATION TO INFLEXIBLE SOCIAL INFERENCE UNDER UNCERTAINTY

BackgroundPersecutory delusions (PD) are a prominent symptom in first episode psychosis and psychosis patients. PD have been linked to abnormalities in probabilistic reasoning and social inference (e.g., attribution styles). Predictive Coding theories of delusion formation suggest that rigid delusional beliefs could be formalized as precise (i.e. held with certainty) high-level prior beliefs, which were formed to explain away overly precise low-level prediction errors (PEs). Rigid reliance on high-level prior beliefs would in turn lead to diminished updating of high-level PEs, i.e. decreased learning and updating of high-level beliefs.MethodsWe tested the prediction that subclinical PD ideation is related to altered social inference and beliefs about others’ intentions. To that end, N=1’145 participants from the general population were pre-screened with the Paranoia Checklist (PCL) and assigned to groups of high (“high PD”) or low PD tendencies (“low PD”). Participants with intermediate scores were excluded, participants assigned to either group filled in the PCL again after four weeks, only individuals whose score remained inside the cut-offs for either group were subsequently invited to the study. We invited 162 participants and included 151 participants in the analyses based on exclusion criteria defined in an analysis plan, which was time-stamped before the conclusion of data acquisition. Participants performed a probabilistic advice-taking task with dynamic changes in the advice-outcome mapping (volatility) under one of two experimental frames. These frames differentially emphasised possible reasons behind unhelpful advice: (i) the adviser’s possible intentions (dispositional frame) or (ii) the rules of the game (situational frame). Our design was thus 2-by-2 factorial (high vs. low delusional ideation, dispositional vs. situational frame). Participants were matched regarding age, gender, and education in years. In addition to analyses of variance on participants’ behaviour, we applied computational modeling to test the predictions regarding prior beliefs and belief updating mentioned above.ResultsWe found significant group-by-frame interactions, indicating that in the situational frame high PD participants took advice less into account than low scorers (df = (1,150), F = 5.77, p = 0.018, partial η2= 0.04). This was also reflected in the model parameters of the model explaining participants’ learning under uncertainty best in comparison to other learning models (e.g. tonic evolution rate omega2: df = (1,150), F = 4.75, p = 0.03).DiscussionOur findings suggest that social inference in individuals with subclinical PD tendencies is shaped by rigid negative prior beliefs about the intentions of others. High PD participants were less sensitive to the attributional framing and updated their beliefs less vs. low PD participants thereby preventing them to make adaptive use of social information in “safe” contexts.

Novelty Manipulations, Memory Performance, and Predictive Coding: the Role of Unexpectedness.

Novelty is central to the study of memory, but the wide range of experimental manipulations aimed to reveal its effects on learning produced inconsistent results. The novelty/encoding hypothesis suggests that novel information undergoes enhanced encoding and thus leads to benefits in memory, especially in recognition performance; however, recent studies cast doubts on this assumption. On the other hand, data from animal studies provided evidence on the robust effects of novelty manipulations on the neurophysiological correlates of memory processes. Conceptualizations and operationalizations of novelty are remarkably variable and were categorized into different subtypes, such as stimulus, context, associative or spatial novelty. Here, we summarize previous findings about the effects of novelty on memory and suggest that predictive coding theories provide a framework that could shed light on the differential influence of novelty manipulations on memory performance. In line with predictive coding theories, we emphasize the role of unexpectedness as a crucial property mediating the behavioral and neural effects of novelty manipulations.

Predictability of higher-order temporal structure of musical stimuli is associated with auditory evoked response

Sound predictability resulting from repetitive patterns can be implicitly learned and often neither requires nor captures our conscious attention. Recently, predictive coding theory has been used as a framework to explain how predictable or expected stimuli evoke and gradually attenuate obligatory neural responses over time compared to those elicited by unpredictable events. However, these results were obtained using the repetition of simple auditory objects such as pairs of tones or phonemes. Here we examined whether the same principle would hold for more abstract temporal structures of sounds. If this is the case, we hypothesized that a regular repetition schedule of a set of musical patterns would reduce neural processing over the course of listening compared to stimuli with an irregular repetition schedule (and the same set of musical patterns). Electroencephalography (EEG) was recorded while participants passively listened to 6–8 min stimulus sequences in which five different four-tone patterns with temporally regular or irregular repetition were presented successively in a randomized order. N1 amplitudes in response to the first tone of each musical pattern were significantly less negative at the end of the regular sequence compared to the beginning, while such reduction was absent in the irregular sequence. These results extend previous findings by showing that N1 reflects automatic learning of the predictable higher-order structure of sound sequences, while continuous engagement of preattentive auditory processing is necessary for the unpredictable structure.

Stereoscopic Omnidirectional Image Quality Assessment Based on Predictive Coding Theory

Objective quality assessment of stereoscopic omnidirectional images is a challenging problem since it is influenced by multiple aspects such as projection deformation, field of view (FoV) range, binocular vision, visual comfort, etc. Existing studies show that classic 2D or 3D image quality assessment (IQA) metrics are not able to perform well for stereoscopic omnidirectional images. However, very few research works have focused on evaluating the perceptual visual quality of omnidirectional images, especially for stereoscopic omnidirectional images. In this article, based on the predictive coding theory of the human vision system (HVS), we propose a stereoscopic omnidirectional image quality evaluator (SOIQE) to cope with the characteristics of 3D 360-degree images. Two modules are involved in SOIQE: predictive coding theory based binocular rivalry module and multi-view fusion module. In the binocular rivalry module, we introduce predictive coding theory to simulate the competition between high-level patterns and calculate the similarity and rivalry dominance to obtain the quality scores of viewport images. Moreover, we develop the multi-view fusion module to aggregate the quality scores of viewport images with the help of both content weight and location weight. The proposed SOIQE is a parametric model without necessary of regression learning, which ensures its interpretability and generalization performance. Experimental results on our published stereoscopic omnidirectional image quality assessment database (SOLID) demonstrate that our proposed SOIQE method outperforms state-of-the-art metrics. Furthermore, we also verify the effectiveness of each proposed module on both public stereoscopic image datasets and panoramic image datasets.