Predictive coding in auditory perception: challenges and unresolved questions.

Predictive coding is arguably the currently dominant theoretical framework for the study of perception. It has been employed to explain important auditory perceptual phenomena, and it has inspired theoretical, experimental and computational modelling efforts aimed at describing how the auditory system parses the complex sound input into meaningful units (auditory scene analysis). These efforts have uncovered some vital questions, addressing which could help to further specify predictive coding and clarify some of its basic assumptions. The goal of the current review is to motivate these questions and show how unresolved issues in explaining some auditory phenomena lead to general questions of the theoretical framework. We focus on experimental and computational modelling issues related to sequential grouping in auditory scene analysis (auditory pattern detection and bistable perception), as we believe that this is the research topic where predictive coding has the highest potential for advancing our understanding. In addition to specific questions, our analysis led us to identify three more general questions that require further clarification: (1) What exactly is meant by prediction in predictive coding? (2) What governs which generative models make the predictions? and (3) What (if it exists) is the correlate of perceptual experience within the predictive coding framework?

Children's perception of visual and auditory ambiguity and its link to executive functions and creativity.

The phenomenon of perceptual bistability provides insights into aspects of perceptual processing not normally accessible to everyday experience. However, most experiments have been conducted in adults, and it is not clear to what extent key aspects of perceptual switching change through development. The current research examined the ability of 6-, 8-, and 10-year-old children (N = 66) to switch between competing percepts of ambiguous visual and auditory stimuli and links between switching rate, executive functions, and creativity. The numbers of switches participants reported in two visual tasks (ambiguous figure and ambiguous structure from motion) and two auditory tasks (verbal transformation and auditory streaming) were measured in three 60-s blocks. In addition, inhibitory control was measured with a Stroop task, set shifting was measured with a verbal fluency task, and creativity was measured with a divergent thinking task. The numbers of perceptual switches increased in all four tasks from 6 to 10 years of age but differed across tasks in that they were higher in the verbal transformation and ambigous structure-from-motion tasks than in the ambigous figure and auditory streaming tasks for all age groups. Although perceptual switching rates differed across tasks, there were predictive relationships between switching rates in some tasks. However, little evidence for the influence of central processes on perceptual switching was found. Overall, the results support the notion that perceptual switching is largely modality and task specific and that this property is already evident when perceptual switching emerges.

Similar but separate systems underlie perceptual bistability in vision and audition.

The dynamics of perceptual bistability, the phenomenon in which perception switches between different interpretations of an unchanging stimulus, are characterised by very similar properties across a wide range of qualitatively different paradigms. This suggests that perceptual switching may be triggered by some common source. However, it is also possible that perceptual switching may arise from a distributed system, whose components vary according to the specifics of the perceptual experiences involved. Here we used a visual and an auditory task to determine whether individuals show cross-modal commonalities in perceptual switching. We found that individual perceptual switching rates were significantly correlated across modalities. We then asked whether perceptual switching arises from some central (modality-) task-independent process or from a more distributed task-specific system. We found that a log-normal distribution best explained the distribution of perceptual phases in both modalities, suggestive of a combined set of independent processes causing perceptual switching. Modality- and/or task-dependent differences in these distributions, and lack of correlation with the modality-independent central factors tested (ego-resiliency, creativity, and executive function), also point towards perceptual switching arising from a distributed system of similar but independent processes.

A test of the adaptive network explanation of functional disorders using a machine learning analysis of symptoms.

The classification and etiology of functional disorders is controversial. Evidence supports both psychological and biological (disease) models that show, respectively, that functional disorders should be classified as one (bodily distress syndrome) and many (e.g., irritable bowel syndrome (IBS), fibromyalgia syndrome (FMS), and chronic fatigue syndrome (CFS)). Two network models (symptom network and adaptive network) can explain the specificity and covariation of symptomatology, but only the adaptive network model can explain the covariation of the somatic symptoms of functional disorders. The adaptive network model is based on the premise that a network of biological mechanisms has emergent properties and can exhibit adaptation. The purpose of this study was to test the predictions that symptom similarity increases with pathology and that network connection strengths vary with pathology, as this would be consistent with the notion that functional disorder pathology arises from network adaptation. We conducted a symptom internet survey followed by machine learning analysis. Participants were 1751 people reporting IBS, FMS or CFS diagnosis who completed a 61-item symptom questionnaire. Eleven symptom clusters were identified. Differences in symptom clusters between IBS, FMS and CFS groups decreased as overall symptom frequency increased. The strength of outgoing connections between clusters varied as a function of symptom frequency and single versus multiple diagnoses. The findings suggest that the pathology of functional disorders involves an increase in the activity and causal connections between several symptom causing mechanisms. The data provide support for the proposal that the body is capable of complex adaptation and that functional disorders result when rules that normally improve adaptation create maladaptive change.

Functional brain networks underlying idiosyncratic switching patterns in multi-stable auditory perception.

In perceptual multi-stability, perception stochastically switches between alternative interpretations of the stimulus allowing examination of perceptual experience independent of stimulus parameters. Previous studies found that listeners show temporally stable idiosyncratic switching patterns when listening to a multi-stable auditory stimulus, such as in the auditory streaming paradigm. This inter-individual variability can be described along two dimensions, Exploration and Segregation. In the current study, we explored the functional brain networks associated with these dimensions and their constituents using electroencephalography. Results showed that Segregation and its constituents are related to brain networks operating in the theta EEG band, whereas Exploration and its constituents are related to networks in the lower and upper alpha and beta bands. Thus, the dimensions on which individuals' perception differ from each other in the auditory streaming paradigm probably reflect separate perceptual processes in the human brain. Further, the results suggest that networks mainly located in left auditory areas underlie the perception of integration, whereas perceiving the alternative patterns is accompanied by stronger interhemispheric connections.

The Risky Side of Creativity: Domain Specific Risk Taking in Creative Individuals.

Risk taking is often associated with creativity, yet little evidence exists to support this association. The present article aimed to systematically explore this association. In two studies, we investigated the relationship between five different domains of risk taking (financial, health and safety, recreational, ethical and social) and five different measures of creativity. Results from the first (laboratory-based) offline study suggested that creativity is associated with high risk taking tendencies in the social domain but not the other domains. Indeed, in the second study conducted online with a larger and diverse sample, the likelihood of social risk taking was the strongest predictor of creative personality and ideation scores. These findings illustrate the necessity to treat creativity and risk taking as multi-dimensional traits and the need to have a more nuanced framework of creativity and other related cognitive functions.

Auditory multistability and neurotransmitter concentrations in the human brain.

Multistability in perception is a powerful tool for investigating sensory-perceptual transformations, because it produces dissociations between sensory inputs and subjective experience. Spontaneous switching between different perceptual objects occurs during prolonged listening to a sound sequence of tone triplets or repeated words (termed auditory streaming and verbal transformations, respectively). We used these examples of auditory multistability to examine to what extent neurochemical and cognitive factors influence the observed idiosyncratic patterns of switching between perceptual objects. The concentrations of glutamate-glutamine (Glx) and γ-aminobutyric acid (GABA) in brain regions were measured by magnetic resonance spectroscopy, while personality traits and executive functions were assessed using questionnaires and response inhibition tasks. Idiosyncratic patterns of perceptual switching in the two multistable stimulus configurations were identified using a multidimensional scaling (MDS) analysis. Intriguingly, although switching patterns within each individual differed between auditory streaming and verbal transformations, similar MDS dimensions were extracted separately from the two datasets. Individual switching patterns were significantly correlated with Glx and GABA concentrations in auditory cortex and inferior frontal cortex but not with the personality traits and executive functions. Our results suggest that auditory perceptual organization depends on the balance between neural excitation and inhibition in different brain regions.This article is part of the themed issue 'Auditory and visual scene analysis'.

Computational Models of Auditory Scene Analysis: A Review.

Auditory scene analysis (ASA) refers to the process (es) of parsing the complex acoustic input into auditory perceptual objects representing either physical sources or temporal sound patterns, such as melodies, which contributed to the sound waves reaching the ears. A number of new computational models accounting for some of the perceptual phenomena of ASA have been published recently. Here we provide a theoretically motivated review of these computational models, aiming to relate their guiding principles to the central issues of the theoretical framework of ASA. Specifically, we ask how they achieve the grouping and separation of sound elements and whether they implement some form of competition between alternative interpretations of the sound input. We consider the extent to which they include predictive processes, as important current theories suggest that perception is inherently predictive, and also how they have been evaluated. We conclude that current computational models of ASA are fragmentary in the sense that rather than providing general competing interpretations of ASA, they focus on assessing the utility of specific processes (or algorithms) for finding the causes of the complex acoustic signal. This leaves open the possibility for integrating complementary aspects of the models into a more comprehensive theory of ASA.

Auditory Multi-Stability: Idiosyncratic Perceptual Switching Patterns, Executive Functions and Personality Traits.

Multi-stability refers to the phenomenon of perception stochastically switching between possible interpretations of an unchanging stimulus. Despite considerable variability, individuals show stable idiosyncratic patterns of switching between alternative perceptions in the auditory streaming paradigm. We explored correlates of the individual switching patterns with executive functions, personality traits, and creativity. The main dimensions on which individual switching patterns differed from each other were identified using multidimensional scaling. Individuals with high scores on the dimension explaining the largest portion of the inter-individual variance switched more often between the alternative perceptions than those with low scores. They also perceived the most unusual interpretation more often, and experienced all perceptual alternatives with a shorter delay from stimulus onset. The ego-resiliency personality trait, which reflects a tendency for adaptive flexibility and experience seeking, was significantly positively related to this dimension. Taking these results together we suggest that this dimension may reflect the individual's tendency for exploring the auditory environment. Executive functions were significantly related to some of the variables describing global properties of the switching patterns, such as the average number of switches. Thus individual patterns of perceptual switching in the auditory streaming paradigm are related to some personality traits and executive functions.

Assessing the validity of subjective reports in the auditory streaming paradigm.

While subjective reports provide a direct measure of perception, their validity is not self-evident. Here, the authors tested three possible biasing effects on perceptual reports in the auditory streaming paradigm: errors due to imperfect understanding of the instructions, voluntary perceptual biasing, and susceptibility to implicit expectations. (1) Analysis of the responses to catch trials separately promoting each of the possible percepts allowed the authors to exclude participants who likely have not fully understood the instructions. (2) Explicit biasing instructions led to markedly different behavior than the conventional neutral-instruction condition, suggesting that listeners did not voluntarily bias their perception in a systematic way under the neutral instructions. Comparison with a random response condition further supported this conclusion. (3) No significant relationship was found between social desirability, a scale-based measure of susceptibility to implicit social expectations, and any of the perceptual measures extracted from the subjective reports. This suggests that listeners did not significantly bias their perceptual reports due to possible implicit expectations present in the experimental context. In sum, these results suggest that valid perceptual data can be obtained from subjective reports in the auditory streaming paradigm.

Relative Pitch Perception and the Detection of Deviant Tone Patterns.

Most people are able to recognise familiar tunes even when played in a different key. It is assumed that this depends on a general capacity for relative pitch perception; the ability to recognise the pattern of inter-note intervals that characterises the tune. However, when healthy adults are required to detect rare deviant melodic patterns in a sequence of randomly transposed standard patterns they perform close to chance. Musically experienced participants perform better than naïve participants, but even they find the task difficult, despite the fact that musical education includes training in interval recognition.To understand the source of this difficulty we designed an experiment to explore the relative influence of the size of within-pattern intervals and between-pattern transpositions on detecting deviant melodic patterns. We found that task difficulty increases when patterns contain large intervals (5-7 semitones) rather than small intervals (1-3 semitones). While task difficulty increases substantially when transpositions are introduced, the effect of transposition size (large vs small) is weaker. Increasing the range of permissible intervals to be used also makes the task more difficult. Furthermore, providing an initial exact repetition followed by subsequent transpositions does not improve performance. Although musical training correlates with task performance, we find no evidence that violations to musical intervals important in Western music (i.e. the perfect fifth or fourth) are more easily detected. In summary, relative pitch perception does not appear to be conducive to simple explanations based exclusively on invariant physical ratios.

Corrigendum: Probing auditory scene analysis.

[This corrects the article on p. 293 in vol. 8, PMID: 25309314.].

A spiking self-organizing map combining STDP, oscillations, and continuous learning.

The self-organizing map (SOM) is a neural network algorithm to create topographically ordered spatial representations of an input data set using unsupervised learning. The SOM algorithm is inspired by the feature maps found in mammalian cortices but lacks some important functional properties of its biological equivalents. Neurons have no direct access to global information, transmit information through spikes and may be using phasic coding of spike times within synchronized oscillations, receive continuous input from the environment, do not necessarily alter network properties such as learning rate and lateral connectivity throughout training, and learn through relative timing of action potentials across a synaptic connection. In this paper, a network of integrate-and-fire neurons is presented that incorporates solutions to each of these issues through the neuron model and network structure. Results of the simulated experiments assessing map formation using artificial data as well as the Iris and Wisconsin Breast Cancer datasets show that this novel implementation maintains fundamental properties of the conventional SOM, thereby representing a significant step toward further understanding of the self-organizational properties of the brain while providing an additional method for implementing SOMs that can be utilized for future modeling in software or special purpose spiking neuron hardware.

Do audio-visual motion cues promote segregation of auditory streams?

An audio-visual experiment using moving sound sources was designed to investigate whether the analysis of auditory scenes is modulated by synchronous presentation of visual information. Listeners were presented with an alternating sequence of two pure tones delivered by two separate sound sources. In different conditions, the two sound sources were either stationary or moving on random trajectories around the listener. Both the sounds and the movement trajectories were derived from recordings in which two humans were moving with loudspeakers attached to their heads. Visualized movement trajectories modeled by a computer animation were presented together with the sounds. In the main experiment, behavioral reports on sound organization were collected from young healthy volunteers. The proportion and stability of the different sound organizations were compared between the conditions in which the visualized trajectories matched the movement of the sound sources and when the two were independent of each other. The results corroborate earlier findings that separation of sound sources in space promotes segregation. However, no additional effect of auditory movement per se on the perceptual organization of sounds was obtained. Surprisingly, the presentation of movement-congruent visual cues did not strengthen the effects of spatial separation on segregating auditory streams. Our findings are consistent with the view that bistability in the auditory modality can occur independently from other modalities.

The effects of rhythm and melody on auditory stream segregation.

While many studies have assessed the efficacy of similarity-based cues for auditory stream segregation, much less is known about whether and how the larger-scale structure of sound sequences support stream formation and the choice of sound organization. Two experiments investigated the effects of musical melody and rhythm on the segregation of two interleaved tone sequences. The two sets of tones fully overlapped in pitch range but differed from each other in interaural time and intensity. Unbeknownst to the listener, separately, each of the interleaved sequences was created from the notes of a different song. In different experimental conditions, the notes and/or their timing could either follow those of the songs or they could be scrambled or, in case of timing, set to be isochronous. Listeners were asked to continuously report whether they heard a single coherent sequence (integrated) or two concurrent streams (segregated). Although temporal overlap between tones from the two streams proved to be the strongest cue for stream segregation, significant effects of tonality and familiarity with the songs were also observed. These results suggest that the regular temporal patterns are utilized as cues in auditory stream segregation and that long-term memory is involved in this process.

Predictive regularity representations in violation detection and auditory stream segregation: from conceptual to computational models.

Predictive accounts of perception have received increasing attention in the past 20 years. Detecting violations of auditory regularities, as reflected by the Mismatch Negativity (MMN) auditory event-related potential, is amongst the phenomena seamlessly fitting this approach. Largely based on the MMN literature, we propose a psychological conceptual framework called the Auditory Event Representation System (AERS), which is based on the assumption that auditory regularity violation detection and the formation of auditory perceptual objects are based on the same predictive regularity representations. Based on this notion, a computational model of auditory stream segregation, called CHAINS, has been developed. In CHAINS, the auditory sensory event representation of each incoming sound is considered for being the continuation of likely combinations of the preceding sounds in the sequence, thus providing alternative interpretations of the auditory input. Detecting repeating patterns allows predicting upcoming sound events, thus providing a test and potential support for the corresponding interpretation. Alternative interpretations continuously compete for perceptual dominance. In this paper, we briefly describe AERS and deduce some general constraints from this conceptual model. We then go on to illustrate how these constraints are computationally specified in CHAINS.

Multimodal integration of micro-Doppler sonar and auditory signals for behavior classification with convolutional networks.

The ability to recognize the behavior of individuals is of great interest in the general field of safety (e.g. building security, crowd control, transport analysis, independent living for the elderly). Here we report a new real-time acoustic system for human action and behavior recognition that integrates passive audio and active micro-Doppler sonar signatures over multiple time scales. The system architecture is based on a six-layer convolutional neural network, trained and evaluated using a dataset of 10 subjects performing seven different behaviors. Probabilistic combination of system output through time for each modality separately yields 94% (passive audio) and 91% (micro-Doppler sonar) correct behavior classification; probabilistic multimodal integration increases classification performance to 98%. This study supports the efficacy of micro-Doppler sonar systems in characterizing human actions, which can then be efficiently classified using ConvNets. It also demonstrates that the integration of multiple sources of acoustic information can significantly improve the system's performance.

Modelling the emergence and dynamics of perceptual organisation in auditory streaming.

Many sound sources can only be recognised from the pattern of sounds they emit, and not from the individual sound events that make up their emission sequences. Auditory scene analysis addresses the difficult task of interpreting the sound world in terms of an unknown number of discrete sound sources (causes) with possibly overlapping signals, and therefore of associating each event with the appropriate source. There are potentially many different ways in which incoming events can be assigned to different causes, which means that the auditory system has to choose between them. This problem has been studied for many years using the auditory streaming paradigm, and recently it has become apparent that instead of making one fixed perceptual decision, given sufficient time, auditory perception switches back and forth between the alternatives-a phenomenon known as perceptual bi- or multi-stability. We propose a new model of auditory scene analysis at the core of which is a process that seeks to discover predictable patterns in the ongoing sound sequence. Representations of predictable fragments are created on the fly, and are maintained, strengthened or weakened on the basis of their predictive success, and conflict with other representations. Auditory perceptual organisation emerges spontaneously from the nature of the competition between these representations. We present detailed comparisons between the model simulations and data from an auditory streaming experiment, and show that the model accounts for many important findings, including: the emergence of, and switching between, alternative organisations; the influence of stimulus parameters on perceptual dominance, switching rate and perceptual phase durations; and the build-up of auditory streaming. The principal contribution of the model is to show that a two-stage process of pattern discovery and competition between incompatible patterns can account for both the contents (perceptual organisations) and the dynamics of human perception in auditory streaming.

A robust sound perception model suitable for neuromorphic implementation.

We have recently demonstrated the emergence of dynamic feature sensitivity through exposure to formative stimuli in a real-time neuromorphic system implementing a hybrid analog/digital network of spiking neurons. This network, inspired by models of auditory processing in mammals, includes several mutually connected layers with distance-dependent transmission delays and learning in the form of spike timing dependent plasticity, which effects stimulus-driven changes in the network connectivity. Here we present results that demonstrate that the network is robust to a range of variations in the stimulus pattern, such as are found in naturalistic stimuli and neural responses. This robustness is a property critical to the development of realistic, electronic neuromorphic systems. We analyze the variability of the response of the network to "noisy" stimuli which allows us to characterize the acuity in information-theoretic terms. This provides an objective basis for the quantitative comparison of networks, their connectivity patterns, and learning strategies, which can inform future design decisions. We also show, using stimuli derived from speech samples, that the principles are robust to other challenges, such as variable presentation rate, that would have to be met by systems deployed in the real world. Finally we demonstrate the potential applicability of the approach to real sounds.