Causal inference and temporal predictions in audiovisual perception of speech and music

Improved emotion differentiation under reduced acoustic variability of speech in autism

BACKGROUND

Socio-emotional impairments are among the diagnostic criteria for autism spectrum disorder (ASD), but the actual knowledge has substantiated both altered and intact emotional prosodies recognition. Here, a Bayesian framework of perception is considered suggesting that the oversampling of sensory evidence would impair perception within highly variable environments. However, reliable hierarchical structures for spectral and temporal cues would foster emotion discrimination by autistics.

METHODS

Event-related spectral perturbations (ERSP) extracted from electroencephalographic (EEG) data indexed the perception of anger, disgust, fear, happiness, neutral, and sadness prosodies while listening to speech uttered by (a) human or (b) synthesized voices characterized by reduced volatility and variability of acoustic environments. The assessment of mechanisms for perception was extended to the visual domain by analyzing the behavioral accuracy within a non-social task in which dynamics of precision weighting between bottom-up evidence and top-down inferences were emphasized. Eighty children (mean 9.7 years old; standard deviation 1.8) volunteered including 40 autistics. The symptomatology was assessed at the time of the study via the Autism Diagnostic Observation Schedule, Second Edition, and parents' responses on the Autism Spectrum Rating Scales. A mixed within-between analysis of variance was conducted to assess the effects of group (autism versus typical development), voice, emotions, and interaction between factors. A Bayesian analysis was implemented to quantify the evidence in favor of the null hypothesis in case of non-significance. Post hoc comparisons were corrected for multiple testing.

RESULTS

Autistic children presented impaired emotion differentiation while listening to speech uttered by human voices, which was improved when the acoustic volatility and variability of voices were reduced. Divergent neural patterns were observed from neurotypicals to autistics, emphasizing different mechanisms for perception. Accordingly, behavioral measurements on the visual task were consistent with the over-precision ascribed to the environmental variability (sensory processing) that weakened performance. Unlike autistic children, neurotypicals could differentiate emotions induced by all voices.

CONCLUSIONS

This study outlines behavioral and neurophysiological mechanisms that underpin responses to sensory variability. Neurobiological insights into the processing of emotional prosodies emphasized the potential of acoustically modified emotional prosodies to improve emotion differentiation by autistics.

TRIAL REGISTRATION

BioMed Central ISRCTN Registry, ISRCTN18117434. Registered on September 20, 2020.

Perceptual uncertainty explains activation differences between audiovisual congruent speech and McGurk stimuli

Face-to-face communication relies on the integration of acoustic speech signals with the corresponding facial articulations. In the McGurk illusion, an auditory /ba/ phoneme presented simultaneously with a facial articulation of a /ga/ (i.e., viseme), is typically fused into an illusory 'da' percept. Despite its widespread use as an index of audiovisual speech integration, critics argue that it arises from perceptual processes that differ categorically from natural speech recognition. Conversely, Bayesian theoretical frameworks suggest that both the illusory McGurk and the veridical audiovisual congruent speech percepts result from probabilistic inference based on noisy sensory signals. According to these models, the inter-sensory conflict in McGurk stimuli may only increase observers' perceptual uncertainty. This functional magnetic resonance imaging (fMRI) study presented participants (20 male and 24 female) with audiovisual congruent, McGurk (i.e., auditory /ba/ + visual /ga/), and incongruent (i.e., auditory /ga/ + visual /ba/) stimuli along with their unisensory counterparts in a syllable categorization task. Behaviorally, observers' response entropy was greater for McGurk compared to congruent audiovisual stimuli. At the neural level, McGurk stimuli increased activations in a widespread neural system, extending from the inferior frontal sulci (IFS) to the pre-supplementary motor area (pre-SMA) and insulae, typically involved in cognitive control processes. Crucially, in line with Bayesian theories these activation increases were fully accounted for by observers' perceptual uncertainty as measured by their response entropy. Our findings suggest that McGurk and congruent speech processing rely on shared neural mechanisms, thereby supporting the McGurk illusion as a valid measure of natural audiovisual speech perception.

Perceptual oddities: assessing the relationship between film editing and prediction processes

During film viewing, humans parse sequences of individual shots into larger narrative structures, often weaving transitions at edit points into an apparently seamless and continuous flow. Editing helps filmmakers manipulate visual transitions to induce feelings of fluency/disfluency, tension/relief, curiosity, expectation and several emotional responses. We propose that the perceptual dynamics induced by film editing can be captured by a predictive processing (PP) framework. We hypothesise that visual discontinuities at edit points produce discrepancies between anticipated and actual sensory input, leading to prediction error. Further, we propose that the magnitude of prediction error depends on the predictability of each shot within the narrative flow, and lay out an account based on conflict monitoring. We test this hypothesis in two empirical studies measuring electroencephalography (EEG) during passive viewing of film excerpts, as well as behavioural responses during an active edit detection task. We report the neural and behavioural modulations at editing boundaries across three levels of narrative depth, showing greater modulations for edits spanning less predictable, deeper narrative transitions. Overall, our contribution lays the groundwork for understanding film editing from a PP perspective. This article is part of the theme issue 'Art, aesthetics and predictive processing: theoretical and empirical perspectivess'.

Interdependence of "What" and "When" in the Brain

From a brain's-eye-view, when a stimulus occurs and what it is are interrelated aspects of interpreting the perceptual world. Yet in practice, the putative perceptual inferences about sensory content and timing are often dichotomized and not investigated as an integrated process. We here argue that neural temporal dynamics can influence what is perceived, and in turn, stimulus content can influence the time at which perception is achieved. This computational principle results from the highly interdependent relationship of what and when in the environment. Both brain processes and perceptual events display strong temporal variability that is not always modeled; we argue that understanding-and, minimally, modeling-this temporal variability is key for theories of how the brain generates unified and consistent neural representations and that we ignore temporal variability in our analysis practice at the peril of both data interpretation and theory-building. Here, we review what and when interactions in the brain, demonstrate via simulations how temporal variability can result in misguided interpretations and conclusions, and outline how to integrate and synthesize what and when in theories and models of brain computation.

Spoken language processing activates the primary visual cortex

Primary visual cortex (V1) is generally thought of as a low-level sensory area that primarily processes basic visual features. Although there is evidence for multisensory effects on its activity, these are typically found for the processing of simple sounds and their properties, for example spatially or temporally-congruent simple sounds. However, in congenitally blind individuals, V1 is involved in language processing, with no evidence of major changes in anatomical connectivity that could explain this seemingly drastic functional change. This is at odds with current accounts of neural plasticity, which emphasize the role of connectivity and conserved function in determining a neural tissue's role even after atypical early experiences. To reconcile what appears to be unprecedented functional reorganization with known accounts of plasticity limitations, we tested whether V1's multisensory roles include responses to spoken language in sighted individuals. Using fMRI, we found that V1 in normally sighted individuals was indeed activated by comprehensible spoken sentences as compared to an incomprehensible reversed speech control condition, and more strongly so in the left compared to the right hemisphere. Activation in V1 for language was also significant and comparable for abstract and concrete words, suggesting it was not driven by visual imagery. Last, this activation did not stem from increased attention to the auditory onset of words, nor was it correlated with attentional arousal ratings, making general attention accounts an unlikely explanation. Together these findings suggest that V1 responds to spoken language even in sighted individuals, reflecting the binding of multisensory high-level signals, potentially to predict visual input. This capability might be the basis for the strong V1 language activation observed in people born blind, re-affirming the notion that plasticity is guided by pre-existing connectivity and abilities in the typically developed brain.

Neural correlates of audiovisual speech synchrony perception and its relationship with autistic traits

The anterior insula (AI) has the central role in coordinating attention and integrating information from multiple sensory modalities. AI dysfunction may contribute to both sensory and social impairments in autism spectrum disorder (ASD). Little is known regarding the brain mechanisms that guide multisensory integration, and how such neural activity might be affected by autistic-like symptoms in the general population. In this study, 72 healthy young adults performed an audiovisual speech synchrony judgment (SJ) task during fMRI scanning. We aimed to investigate the SJ-related brain activations and connectivity, with a focus on the AI. Compared with synchronous speech, asynchrony perception triggered stronger activations in the bilateral AI, and other frontal-cingulate-parietal regions. In contrast, synchronous perception resulted in greater involvement of the primary auditory and visual areas, indicating multisensory validation and fusion. Moreover, the AI demonstrated a stronger connection with the anterior cingulate gyrus (ACC) in the audiovisual asynchronous (vs. synchronous) condition. To facilitate asynchrony detection, the AI may integrate auditory and visual speech stimuli, and generate a control signal to the ACC that further supports conflict-resolving and response selection. Correlation analysis, however, suggested that audiovisual synchrony perception and its related AI activation and connectivity did not significantly vary with different levels of autistic traits. These findings provide novel evidence for the neural mechanisms underlying multisensory temporal processing in healthy people. Future research should examine whether such findings would be extended to ASD patients.

Enriched learning: behavior, brain, and computation

The presence of complementary information across multiple sensory or motor modalities during learning, referred to as multimodal enrichment, can markedly benefit learning outcomes. Why is this? Here, we integrate cognitive, neuroscientific, and computational approaches to understanding the effectiveness of enrichment and discuss recent neuroscience findings indicating that crossmodal responses in sensory and motor brain regions causally contribute to the behavioral benefits of enrichment. The findings provide novel evidence for multimodal theories of enriched learning, challenge assumptions of longstanding cognitive theories, and provide counterevidence to unimodal neurobiologically inspired theories. Enriched educational methods are likely effective not only because they may engage greater levels of attention or deeper levels of processing, but also because multimodal interactions in the brain can enhance learning and memory.

Time in schizophrenia: a link between psychopathology, psychophysics and technology

It has been widely demonstrated that time processing is altered in patients with schizophrenia. This perspective review delves into such temporal deficit and highlights its link to low-level sensory alterations, which are often overlooked in rehabilitation protocols for psychosis. However, if temporal impairment at the sensory level is inherent to the disease, new interventions should focus on this dimension. Beyond more traditional types of intervention, here we review the most recent digital technologies for rehabilitation and the most promising ones for sensory training. The overall aim is to synthesise existing literature on time in schizophrenia linking psychopathology, psychophysics, and technology to help future developments.

Multisensory correlation computations in the human brain identified by a time-resolved encoding model

Neural mechanisms that arbitrate between integrating and segregating multisensory information are essential for complex scene analysis and for the resolution of the multisensory correspondence problem. However, these mechanisms and their dynamics remain largely unknown, partly because classical models of multisensory integration are static. Here, we used the Multisensory Correlation Detector, a model that provides a good explanatory power for human behavior while incorporating dynamic computations. Participants judged whether sequences of auditory and visual signals originated from the same source (causal inference) or whether one modality was leading the other (temporal order), while being recorded with magnetoencephalography. First, we confirm that the Multisensory Correlation Detector explains causal inference and temporal order behavioral judgments well. Second, we found strong fits of brain activity to the two outputs of the Multisensory Correlation Detector in temporo-parietal cortices. Finally, we report an asymmetry in the goodness of the fits, which were more reliable during the causal inference task than during the temporal order judgment task. Overall, our results suggest the existence of multisensory correlation detectors in the human brain, which explain why and how causal inference is strongly driven by the temporal correlation of multisensory signals.

Surmising synchrony of sound and sight: Factors explaining variance of audiovisual integration in hurdling, tap dancing and drumming

Auditory and visual percepts are integrated even when they are not perfectly temporally aligned with each other, especially when the visual signal precedes the auditory signal. This window of temporal integration for asynchronous audiovisual stimuli is relatively well examined in the case of speech, while other natural action-induced sounds have been widely neglected. Here, we studied the detection of audiovisual asynchrony in three different whole-body actions with natural action-induced sounds–hurdling, tap dancing and drumming. In Study 1, we examined whether audiovisual asynchrony detection, assessed by a simultaneity judgment task, differs as a function of sound production intentionality. Based on previous findings, we expected that auditory and visual signals should be integrated over a wider temporal window for actions creating sounds intentionally (tap dancing), compared to actions creating sounds incidentally (hurdling). While percentages of perceived synchrony differed in the expected way, we identified two further factors, namely high event density and low rhythmicity, to induce higher synchrony ratings as well. Therefore, we systematically varied event density and rhythmicity in Study 2, this time using drumming stimuli to exert full control over these variables, and the same simultaneity judgment tasks. Results suggest that high event density leads to a bias to integrate rather than segregate auditory and visual signals, even at relatively large asynchronies. Rhythmicity had a similar, albeit weaker effect, when event density was low. Our findings demonstrate that shorter asynchronies and visual-first asynchronies lead to higher synchrony ratings of whole-body action, pointing to clear parallels with audiovisual integration in speech perception. Overconfidence in the naturally expected, that is, synchrony of sound and sight, was stronger for intentional (vs. incidental) sound production and for movements with high (vs. low) rhythmicity, presumably because both encourage predictive processes. In contrast, high event density appears to increase synchronicity judgments simply because it makes the detection of audiovisual asynchrony more difficult. More studies using real-life audiovisual stimuli with varying event densities and rhythmicities are needed to fully uncover the general mechanisms of audiovisual integration.

Auditory and Semantic Cues Facilitate Decoding of Visual Object Category in MEG

Sounds (e.g., barking) help us to visually identify objects (e.g., a dog) that are distant or ambiguous. While neuroimaging studies have revealed neuroanatomical sites of audiovisual interactions, little is known about the time course by which sounds facilitate visual object processing. Here we used magnetoencephalography to reveal the time course of the facilitatory influence of natural sounds (e.g., barking) on visual object processing and compared this to the facilitatory influence of spoken words (e.g., "dog"). Participants viewed images of blurred objects preceded by a task-irrelevant natural sound, a spoken word, or uninformative noise. A classifier was trained to discriminate multivariate sensor patterns evoked by animate and inanimate intact objects with no sounds, presented in a separate experiment, and tested on sensor patterns evoked by the blurred objects in the 3 auditory conditions. Results revealed that both sounds and words, relative to uninformative noise, significantly facilitated visual object category decoding between 300-500 ms after visual onset. We found no evidence for earlier facilitation by sounds than by words. These findings provide evidence for a semantic route of facilitation by both natural sounds and spoken words, whereby the auditory input first activates semantic object representations, which then modulate the visual processing of objects.

A Survey on Probabilistic Models in Human Perception and Machines

Extracting information from noisy signals is of fundamental importance for both biological and artificial perceptual systems. To provide tractable solutions to this challenge, the fields of human perception and machine signal processing (SP) have developed powerful computational models, including Bayesian probabilistic models. However, little true integration between these fields exists in their applications of the probabilistic models for solving analogous problems, such as noise reduction, signal enhancement, and source separation. In this mini review, we briefly introduce and compare selective applications of probabilistic models in machine SP and human psychophysics. We focus on audio and audio-visual processing, using examples of speech enhancement, automatic speech recognition, audio-visual cue integration, source separation, and causal inference to illustrate the basic principles of the probabilistic approach. Our goal is to identify commonalities between probabilistic models addressing brain processes and those aiming at building intelligent machines. These commonalities could constitute the closest points for interdisciplinary convergence.

Weak observer-level correlation and strong stimulus-level correlation between the McGurk effect and audiovisual speech-in-noise: A causal inference explanation

The McGurk effect is a widely used measure of multisensory integration during speech perception. Two observations have raised questions about the validity of the effect as a tool for understanding speech perception. First, there is high variability in perception of the McGurk effect across different stimuli and observers. Second, across observers there is low correlation between McGurk susceptibility and recognition of visual speech paired with auditory speech-in-noise, another common measure of multisensory integration. Using the framework of the causal inference of multisensory speech (CIMS) model, we explored the relationship between the McGurk effect, syllable perception, and sentence perception in seven experiments with a total of 296 different participants. Perceptual reports revealed a relationship between the efficacy of different McGurk stimuli created from the same talker and perception of the auditory component of the McGurk stimuli presented in isolation, both with and without added noise. The CIMS model explained this strong stimulus-level correlation using the principles of noisy sensory encoding followed by optimal cue combination within a common representational space across speech types. Because the McGurk effect (but not speech-in-noise) requires the resolution of conflicting cues between modalities, there is an additional source of individual variability that can explain the weak observer-level correlation between McGurk and noisy speech. Power calculations show that detecting this weak correlation requires studies with many more participants than those conducted to-date. Perception of the McGurk effect and other types of speech can be explained by a common theoretical framework that includes causal inference, suggesting that the McGurk effect is a valid and useful experimental tool.

Audiovisual temporal integration: Cognitive processing, neural mechanisms, developmental trajectory and potential interventions

To integrate auditory and visual signals into a unified percept, the paired stimuli must co-occur within a limited time window known as the Temporal Binding Window (TBW). The width of the TBW, a proxy of audiovisual temporal integration ability, has been found to be correlated with higher-order cognitive and social functions. A comprehensive review of studies investigating audiovisual TBW reveals several findings: (1) a wide range of top-down processes and bottom-up features can modulate the width of the TBW, facilitating adaptation to the changing and multisensory external environment; (2) a large-scale brain network works in coordination to ensure successful detection of audiovisual (a)synchrony; (3) developmentally, audiovisual TBW follows a U-shaped pattern across the lifespan, with a protracted developmental course into late adolescence and rebounding in size again in late life; (4) an enlarged TBW is characteristic of a number of neurodevelopmental disorders; and (5) the TBW is highly flexible via perceptual and musical training. Interventions targeting the TBW may be able to improve multisensory function and ameliorate social communicative symptoms in clinical populations.

Musical Expertise Affects Audiovisual Speech Perception: Findings From Event-Related Potentials and Inter-trial Phase Coherence

In audiovisual speech perception, visual information from a talker's face during mouth articulation is available before the onset of the corresponding audio speech, and thereby allows the perceiver to use visual information to predict the upcoming audio. This prediction from phonetically congruent visual information modulates audiovisual speech perception and leads to a decrease in N1 and P2 amplitudes and latencies compared to the perception of audio speech alone. Whether audiovisual experience, such as with musical training, influences this prediction is unclear, but if so, may explain some of the variations observed in previous research. The current study addresses whether audiovisual speech perception is affected by musical training, first assessing N1 and P2 event-related potentials (ERPs) and in addition, inter-trial phase coherence (ITPC). Musicians and non-musicians are presented the syllable, /ba/ in audio only (AO), video only (VO), and audiovisual (AV) conditions. With the predictory effect of mouth movement isolated from the AV speech (AV-VO), results showed that, compared to audio speech, both groups have a lower N1 latency and P2 amplitude and latency. Moreover, they also showed lower ITPCs in the delta, theta, and beta bands in audiovisual speech perception. However, musicians showed significant suppression of N1 amplitude and desynchronization in the alpha band in audiovisual speech, not present for non-musicians. Collectively, the current findings indicate that early sensory processing can be modified by musical experience, which in turn can explain some of the variations in previous AV speech perception research.

"Seeing Rain": Integrating phenomenological and Bayesian predictive coding approaches to visual hallucinations and self-disturbances (Ichstörungen) in schizophrenia

We present a schizophrenia patient who reports "seeing rain" with attendant somatosensory features which separate him from his surroundings. Because visual/multimodal hallucinations are understudied in schizophrenia, we examine a case history to determine the role of these hallucinations in self-disturbances (Ichstörungen). Developed by the early Heidelberg School, self-disturbances comprise two components: 1. The self experiences its own automatic processing as alien to self in a split-off, "doubled-I." 2. In "I-paralysis," the disruption to automatic processing is now outside the self in omnipotent agents. Self-disturbances (as indicated by visual/multimodal hallucinations) involve impairment in the ability to predict moment-to-moment experiences in the ongoing perception-action cycle. The phenomenological approach to subjective experience of self-disturbances complements efforts to model psychosis using the computational framework of hierarchical predictive coding. We conclude that self-disturbances play an adaptive, compensatory role following the uncoupling of perception and action, and possibly, other low-level perceptual anomalies.