Neural processing as causal inference

Pattern completion and disruption characterize contextual modulation in the visual cortex

Vision is fundamentally context-dependent, with neuronal responses influenced not just by local features but also by surrounding contextual information. In the visual cortex, studies using simple grating stimuli indicate that congruent stimuli - where the center and surround share the same orientation - are more inhibitory than when orientations are orthogonal, potentially serving redundancy reduction and predictive coding. Understanding these center-surround interactions in relation to natural image statistics is challenging due to the high dimensionality of the stimulus space, yet crucial for deciphering the neuronal code of real-world sensory processing. Utilizing large-scale recordings from mouse V1, we trained convolutional neural networks (CNNs) to predict and synthesize surround patterns that either optimally suppressed or enhanced responses to center stimuli, confirmed by in vivo experiments. Contrary to the notion that congruent stimuli are suppressive, we found that surrounds that completed patterns based on natural image statistics were facilitatory, while disruptive surrounds were suppressive. Applying our CNN image synthesis method in macaque V1, we discovered that pattern completion within the near surround occurred more frequently with excitatory than with inhibitory surrounds, suggesting that our results in mice are conserved in macaques. Further, experiments and model analyses confirmed previous studies reporting the opposite effect with grating stimuli in both species. Using the MICrONS functional connectomics dataset, we observed that neurons with similar feature selectivity formed excitatory connections regardless of their receptive field overlap, aligning with the pattern completion phenomenon observed for excitatory surrounds. Finally, our empirical results emerged in a normative model of perception implementing Bayesian inference, where neuronal responses are modulated by prior knowledge of natural scene statistics. In summary, our findings identify a novel relationship between contextual information and natural scene statistics and provide evidence for a role of contextual modulation in hierarchical inference.

Response sub-additivity and variability quenching in visual cortex

Sub-additivity and variability are ubiquitous response motifs in the primary visual cortex (V1). Response sub-additivity enables the construction of useful interpretations of the visual environment, whereas response variability indicates the factors that limit the precision with which the brain can do this. There is increasing evidence that experimental manipulations that elicit response sub-additivity often also quench response variability. Here, we provide an overview of these phenomena and suggest that they may have common origins. We discuss empirical findings and recent model-based insights into the functional operations, computational objectives and circuit mechanisms underlying V1 activity. These different modelling approaches all predict that response sub-additivity and variability quenching often co-occur. The phenomenology of these two response motifs, as well as many of the insights obtained about them in V1, generalize to other cortical areas. Thus, the connection between response sub-additivity and variability quenching may be a canonical motif across the cortex.

Signal processing in the vagus nerve: Hypotheses based on new genetic and anatomical evidence

Each organism must regulate its internal state in a metabolically efficient way as it interacts in space and time with an ever-changing and only partly predictable world. Success in this endeavor is largely determined by the ongoing communication between brain and body, and the vagus nerve is a crucial structure in that dialogue. In this review, we introduce the novel hypothesis that the afferent vagus nerve is engaged in signal processing rather than just signal relay. New genetic and structural evidence of vagal afferent fiber anatomy motivates two hypotheses: (1) that sensory signals informing on the physiological state of the body compute both spatial and temporal viscerosensory features as they ascend the vagus nerve, following patterns found in other sensory architectures, such as the visual and olfactory systems; and (2) that ascending and descending signals modulate one another, calling into question the strict segregation of sensory and motor signals, respectively. Finally, we discuss several implications of our two hypotheses for understanding the role of viscerosensory signal processing in predictive energy regulation (i.e., allostasis) as well as the role of metabolic signals in memory and in disorders of prediction (e.g., mood disorders).

Learning shapes cortical dynamics to enhance integration of relevant sensory input

Adaptive sensory behavior is thought to depend on processing in recurrent cortical circuits, but how dynamics in these circuits shapes the integration and transmission of sensory information is not well understood. Here, we study neural coding in recurrently connected networks of neurons driven by sensory input. We show analytically how information available in the network output varies with the alignment between feedforward input and the integrating modes of the circuit dynamics. In light of this theory, we analyzed neural population activity in the visual cortex of mice that learned to discriminate visual features. We found that over learning, slow patterns of network dynamics realigned to better integrate input relevant to the discrimination task. This realignment of network dynamics could be explained by changes in excitatory-inhibitory connectivity among neurons tuned to relevant features. These results suggest that learning tunes the temporal dynamics of cortical circuits to optimally integrate relevant sensory input.

Nonequilibrium dynamics of adaptation in sensory systems

Adaptation is used by biological sensory systems to respond to a wide range of environmental signals, by adapting their response properties to the statistics of the stimulus in order to maximize information transmission. We derive rules of optimal adaptation to changes in the mean and variance of a continuous stimulus in terms of Bayesian filters and map them onto stochastic equations that couple the state of the environment to an internal variable controlling the response function. We calculate numerical and exact results for the speed and accuracy of adaptation and its impact on information transmission. We find that, in the regime of efficient adaptation, the speed of adaptation scales sublinearly with the rate of change of the environment. Finally, we exploit the mathematical equivalence between adaptation and stochastic thermodynamics to quantitatively relate adaptation to the irreversibility of the adaptation time course, defined by the rate of entropy production. Our results suggest a means to empirically quantify adaptation in a model-free and nonparametric way.

Distinct organization of two cortico-cortical feedback pathways

Neocortical feedback is critical for attention, prediction, and learning. To mechanically understand its function requires deciphering its cell-type wiring. Recent studies revealed that feedback between primary motor to primary somatosensory areas in mice is disinhibitory, targeting vasoactive intestinal peptide-expressing interneurons, in addition to pyramidal cells. It is unknown whether this circuit motif represents a general cortico-cortical feedback organizing principle. Here we show that in contrast to this wiring rule, feedback between higher-order lateromedial visual area to primary visual cortex preferentially activates somatostatin-expressing interneurons. Functionally, both feedback circuits temporally sharpen feed-forward excitation eliciting a transient increase-followed by a prolonged decrease-in pyramidal cell activity under sustained feed-forward input. However, under feed-forward transient input, the primary motor to primary somatosensory cortex feedback facilitates bursting while lateromedial area to primary visual cortex feedback increases time precision. Our findings argue for multiple cortico-cortical feedback motifs implementing different dynamic non-linear operations.

Shape-Texture Debiased Training for Robust Template Matching

Finding a template in a search image is an important task underlying many computer vision applications. This is typically solved by calculating a similarity map using features extracted from the separate images. Recent approaches perform template matching in a deep feature space, produced by a convolutional neural network (CNN), which is found to provide more tolerance to changes in appearance. Inspired by these findings, in this article we investigate whether enhancing the CNN's encoding of shape information can produce more distinguishable features that improve the performance of template matching. By comparing features from the same CNN trained using different shape-texture training methods, we determined a feature space which improves the performance of most template matching algorithms. When combining the proposed method with the Divisive Input Modulation (DIM) template matching algorithm, its performance is greatly improved, and the resulting method produces state-of-the-art results on a standard benchmark. To confirm these results, we create a new benchmark and show that the proposed method outperforms existing techniques on this new dataset.

NLGC: Network Localized Granger Causality with Application to MEG Directional Functional Connectivity Analysis

Identifying the directed connectivity that underlie networked activity between different cortical areas is critical for understanding the neural mechanisms behind sensory processing. Granger causality (GC) is widely used for this purpose in functional magnetic resonance imaging analysis, but there the temporal resolution is low, making it difficult to capture the millisecond-scale interactions underlying sensory processing. Magnetoencephalography (MEG) has millisecond resolution, but only provides low-dimensional sensor-level linear mixtures of neural sources, which makes GC inference challenging. Conventional methods proceed in two stages: First, cortical sources are estimated from MEG using a source localization technique, followed by GC inference among the estimated sources. However, the spatiotemporal biases in estimating sources propagate into the subsequent GC analysis stage, may result in both false alarms and missing true GC links. Here, we introduce the Network Localized Granger Causality (NLGC) inference paradigm, which models the source dynamics as latent sparse multivariate autoregressive processes and estimates their parameters directly from the MEG measurements, integrated with source localization, and employs the resulting parameter estimates to produce a precise statistical characterization of the detected GC links. We offer several theoretical and algorithmic innovations within NLGC and further examine its utility via comprehensive simulations and application to MEG data from an auditory task involving tone processing from both younger and older participants. Our simulation studies reveal that NLGC is markedly robust with respect to model mismatch, network size, and low signal-to-noise ratio, whereas the conventional two-stage methods result in high false alarms and mis-detections. We also demonstrate the advantages of NLGC in revealing the cortical network-level characterization of neural activity during tone processing and resting state by delineating task- and age-related connectivity changes.

Circular inference predicts nonuniform overactivation and dysconnectivity in brain-wide connectomes

Schizophrenia is a severe mental disorder whose neural basis remains difficult to ascertain. Among the available pathophysiological theories, recent work has pointed towards subtle perturbations in the excitation-inhibition (E/I) balance within different neural circuits. Computational approaches have suggested interesting mechanisms that can account for both E/I imbalances and psychotic symptoms. Based on hierarchical neural networks propagating information through a message-passing algorithm, it was hypothesized that changes in the E/I ratio could cause a "circular belief propagation" in which bottom-up and top-down information reverberate. This circular inference (CI) was proposed to account for the clinical features of schizophrenia. Under this assumption, this paper examined the impact of CI on network dynamics in light of brain imaging findings related to psychosis. Using brain-inspired graphical models, we show that CI causes overconfidence and overactivation most specifically at the level of connector hubs (e.g., nodes with many connections allowing integration across networks). By also measuring functional connectivity in these graphs, we provide evidence that CI is able to predict specific changes in modularity known to be associated with schizophrenia. Altogether, these findings suggest that the CI framework may facilitate behavioral and neural research on the multifaceted nature of psychosis.

Bayesian causal inference: A unifying neuroscience theory

Understanding of the brain and the principles governing neural processing requires theories that are parsimonious, can account for a diverse set of phenomena, and can make testable predictions. Here, we review the theory of Bayesian causal inference, which has been tested, refined, and extended in a variety of tasks in humans and other primates by several research groups. Bayesian causal inference is normative and has explained human behavior in a vast number of tasks including unisensory and multisensory perceptual tasks, sensorimotor, and motor tasks, and has accounted for counter-intuitive findings. The theory has made novel predictions that have been tested and confirmed empirically, and recent studies have started to map its algorithms and neural implementation in the human brain. The parsimony, the diversity of the phenomena that the theory has explained, and its illuminating brain function at all three of Marr's levels of analysis make Bayesian causal inference a strong neuroscience theory. This also highlights the importance of collaborative and multi-disciplinary research for the development of new theories in neuroscience.

The Ecological View of Selective Attention

Accumulating evidence is supporting the hypothesis that our selective attention is a manifestation of mechanisms that evolved early in evolution and are shared by many organisms from different taxa. This surge of new data calls for the re-examination of our notions about attention, which have been dominated mostly by human psychology. Here, we present an hypothesis that challenges, based on evolutionary grounds, a common view of attention as a means to manage limited brain resources. We begin by arguing that evolutionary considerations do not favor the basic proposition of the limited brain resources view of attention, namely, that the capacity of the sensory organs to provide information exceeds the capacity of the brain to process this information. Moreover, physiological studies in animals and humans show that mechanisms of selective attention are highly demanding of brain resources, making it paradoxical to see attention as a means to release brain resources. Next, we build on the above arguments to address the question why attention evolved in evolution. We hypothesize that, to a certain extent, limiting sensory processing is adaptive irrespective of brain capacity. We call this hypothesis the ecological view of attention (EVA) because it is centered on interactions of an animal with its environment rather than on internal brain resources. In its essence is the notion that inherently noisy and degraded sensory inputs serve the animal’s adaptive, dynamic interactions with its environment. Attention primarily functions to resolve behavioral conflicts and false distractions. Hence, we evolved to focus on a particular target at the expense of others, not because of internal limitations, but to ensure that behavior is properly oriented and committed to its goals. Here, we expand on this notion and review evidence supporting it. We show how common results in human psychophysics and physiology can be reconciled with an EVA and discuss possible implications of the notion for interpreting current results and guiding future research.

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

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.

Circular inference in bistable perception

When facing ambiguous images, the brain switches between mutually exclusive interpretations, a phenomenon known as bistable perception. Despite years of research, a consensus on whether bistability is driven primarily by bottom-up or top-down mechanisms has not been achieved. Here, we adopted a Bayesian approach to reconcile these two theories. Fifty-five healthy participants were exposed to an adaptation of the Necker cube paradigm, in which we manipulated sensory evidence and prior knowledge. Manipulations of both sensory evidence and priors significantly affected the way participants perceived the Necker cube. However, we observed an interaction between the effect of the cue and the effect of the instructions, a finding that is incompatible with Bayes-optimal integration. In contrast, the data were well predicted by a circular inference model. In this model, ambiguous sensory evidence is systematically biased in the direction of current expectations, ultimately resulting in a bistable percept.

Learning probabilistic neural representations with randomly connected circuits

We present a theory of neural circuits’ design and function, inspired by the random connectivity of real neural circuits and the mathematical power of random projections. Specifically, we introduce a family of statistical models for large neural population codes, a straightforward neural circuit architecture that would implement these models, and a biologically plausible learning rule for such circuits. The resulting neural architecture suggests a design principle for neural circuit—namely, that they learn to compute the mathematical surprise of their inputs, given past inputs, without an explicit teaching signal. We applied these models to recordings from large neural populations in monkeys’ visual and prefrontal cortices and show them to be highly accurate, efficient, and scalable.

The brain represents and reasons probabilistically about complex stimuli and motor actions using a noisy, spike-based neural code. A key building block for such neural computations, as well as the basis for supervised and unsupervised learning, is the ability to estimate the surprise or likelihood of incoming high-dimensional neural activity patterns. Despite progress in statistical modeling of neural responses and deep learning, current approaches either do not scale to large neural populations or cannot be implemented using biologically realistic mechanisms. Inspired by the sparse and random connectivity of real neuronal circuits, we present a model for neural codes that accurately estimates the likelihood of individual spiking patterns and has a straightforward, scalable, efficient, learnable, and realistic neural implementation. This model’s performance on simultaneously recorded spiking activity of >100 neurons in the monkey visual and prefrontal cortices is comparable with or better than that of state-of-the-art models. Importantly, the model can be learned using a small number of samples and using a local learning rule that utilizes noise intrinsic to neural circuits. Slower, structural changes in random connectivity, consistent with rewiring and pruning processes, further improve the efficiency and sparseness of the resulting neural representations. Our results merge insights from neuroanatomy, machine learning, and theoretical neuroscience to suggest random sparse connectivity as a key design principle for neuronal computation.

Monkeys and humans implement causal inference to simultaneously localize auditory and visual stimuli

The environment is sampled by multiple senses, which are woven together to produce a unified perceptual state. However, optimally unifying such signals requires assigning particular signals to the same or different underlying objects or events. Many prior studies (especially in animals) have assumed fusion of cross-modal information, whereas recent work in humans has begun to probe the appropriateness of this assumption. Here we present results from a novel behavioral task in which both monkeys (Macaca mulatta) and humans localized visual and auditory stimuli and reported their perceived sources through saccadic eye movements. When the locations of visual and auditory stimuli were widely separated, subjects made two saccades, while when the two stimuli were presented at the same location they made only a single saccade. Intermediate levels of separation produced mixed response patterns: a single saccade to an intermediate position on some trials or separate saccades to both locations on others. The distribution of responses was well described by a hierarchical causal inference model that accurately predicted both the explicit "same vs. different" source judgments as well as biases in localization of the source(s) under each of these conditions. The results from this task are broadly consistent with prior work in humans across a wide variety of analogous tasks, extending the study of multisensory causal inference to nonhuman primates and to a natural behavioral task with both a categorical assay of the number of perceived sources and a continuous report of the perceived position of the stimuli.NEW & NOTEWORTHY We developed a novel behavioral paradigm for the study of multisensory causal inference in both humans and monkeys and found that both species make causal judgments in the same Bayes-optimal fashion. To our knowledge, this is the first demonstration of behavioral causal inference in animals, and this cross-species comparison lays the groundwork for future experiments using neuronal recording techniques that are impractical or impossible in human subjects.

Neural mechanisms resolving exploitation-exploration dilemmas in the medial prefrontal cortex

Everyday life often requires arbitrating between pursuing an ongoing action plan by possibly adjusting it versus exploring a new action plan instead. Resolving this so-called exploitation-exploration dilemma involves the medial prefrontal cortex (mPFC). Using human intracranial electrophysiological recordings, we discovered that neural activity in the ventral mPFC infers and tracks the reliability of the ongoing plan to proactively encode upcoming action outcomes as either learning signals or potential triggers to explore new plans. By contrast, the dorsal mPFC exhibits neural responses to action outcomes, which results in either improving or abandoning the ongoing plan. Thus, the mPFC resolves the exploitation-exploration dilemma through a two-stage, predictive coding process: a proactive ventromedial stage that constructs the functional signification of upcoming action outcomes and a reactive dorsomedial stage that guides behavior in response to action outcomes.

Theoretical, and epistemological challenges in scientific investigations of complex emotional states in animals

This review brings to light critical epistemological and theoretical considerations when studying complex emotional states in animals. We discuss anthropomorphic and Umwelt perspectives of nonhuman animals and the ways in which distinct theories of consciousness and neural processing may restrict the potential for the development of knowledge on the topic. Within the same line of argumentation, we consider influences of the debate between monism and dualism and psychology's behaviorism and cognitive theories. Finally, we contrast the affective consciousness, higher-order emotional consciousness, and constructed emotion theories to further our understanding of complex emotional states in animals.

Evaluating the neurophysiological evidence for predictive processing as a model of perception

For many years, the dominant theoretical framework guiding research into the neural origins of perceptual experience has been provided by hierarchical feedforward models, in which sensory inputs are passed through a series of increasingly complex feature detectors. However, the long‐standing orthodoxy of these accounts has recently been challenged by a radically different set of theories that contend that perception arises from a purely inferential process supported by two distinct classes of neurons: those that transmit predictions about sensory states and those that signal sensory information that deviates from those predictions. Although these predictive processing (PP) models have become increasingly influential in cognitive neuroscience, they are also criticized for lacking the empirical support to justify their status. This limited evidence base partly reflects the considerable methodological challenges that are presented when trying to test the unique predictions of these models. However, a confluence of technological and theoretical advances has prompted a recent surge in human and nonhuman neurophysiological research seeking to fill this empirical gap. Here, we will review this new research and evaluate the degree to which its findings support the key claims of PP.

Experience shapes activity dynamics and stimulus coding of VIP inhibitory cells

Cortical circuits can flexibly change with experience and learning, but the effects on specific cell types, including distinct inhibitory types, are not well understood. Here we investigated how excitatory and VIP inhibitory cells in layer 2/3 of mouse visual cortex were impacted by visual experience in the context of a behavioral task. Mice learned a visual change detection task with a set of eight natural scene images. Subsequently, during 2-photon imaging experiments, mice performed the task with these familiar images and three sets of novel images. Strikingly, the temporal dynamics of VIP activity differed markedly between novel and familiar images: VIP cells were stimulus-driven by novel images but were suppressed by familiar stimuli and showed ramping activity when expected stimuli were omitted from a temporally predictable sequence. This prominent change in VIP activity suggests that these cells may adopt different modes of processing under novel versus familiar conditions.

The Generative Adversarial Brain

The idea that the brain learns generative models of the world has been widely promulgated. Most approaches have assumed that the brain learns an explicit density model that assigns a probability to each possible state of the world. However, explicit density models are difficult to learn, requiring approximate inference techniques that may find poor solutions. An alternative approach is to learn an implicit density model that can sample from the generative model without evaluating the probabilities of those samples. The implicit model can be trained to fool a discriminator into believing that the samples are real. This is the idea behind generative adversarial algorithms, which have proven adept at learning realistic generative models. This paper develops an adversarial framework for probabilistic computation in the brain. It first considers how generative adversarial algorithms overcome some of the problems that vex prior theories based on explicit density models. It then discusses the psychological and neural evidence for this framework, as well as how the breakdown of the generator and discriminator could lead to delusions observed in some mental disorders.

Emotion words, emotion concepts, and emotional development in children: A constructionist hypothesis

In this article, we integrate two constructionist approaches-the theory of constructed emotion and rational constructivism-to introduce several novel hypotheses for understanding emotional development. We first discuss the hypothesis that emotion categories are abstract and conceptual, whose instances share a goal-based function in a particular context but are highly variable in their affective, physical, and perceptual features. Next, we discuss the possibility that emotional development is the process of developing emotion concepts, and that emotion words may be a critical part of this process. We hypothesize that infants and children learn emotion categories the way they learn other abstract conceptual categories-by observing others use the same emotion word to label highly variable events. Finally, we hypothesize that emotional development can be understood as a concept construction problem: a child becomes capable of experiencing and perceiving emotion only when her brain develops the capacity to assemble ad hoc, situated emotion concepts for the purposes of guiding behavior and giving meaning to sensory inputs. Specifically, we offer a predictive processing account of emotional development. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Causal inference accounts for heading perception in the presence of object motion

The brain infers our spatial orientation and properties of the world from ambiguous and noisy sensory cues. Judging self-motion (heading) in the presence of independently moving objects poses a challenging inference problem because the image motion of an object could be attributed to movement of the object, self-motion, or some combination of the two. We test whether perception of heading and object motion follows predictions of a normative causal inference framework. In a dual-report task, subjects indicated whether an object appeared stationary or moving in the virtual world, while simultaneously judging their heading. Consistent with causal inference predictions, the proportion of object stationarity reports, as well as the accuracy and precision of heading judgments, depended on the speed of object motion. Critically, biases in perceived heading declined when the object was perceived to be moving in the world. Our findings suggest that the brain interprets object motion and self-motion using a causal inference framework.

The power of predictions: An emerging paradigm for psychological research

The last two decades of neuroscience research has produced a growing number of studies that suggest the various psychological phenomena are produced by predictive processes in the brain. When considered together, these studies form a coherent, neurobiologically-inspired research program for guiding psychological research about the mind and behavior. In this paper, we briefly consider the common assumptions and hypotheses that unify an emerging framework and discuss its ramifications, both for improving the replicability and robustness of psychological research and for innovating psychological theory by suggesting an alternative ontology of the human mind.

Single-neuron perturbations reveal feature-specific competition in V1

The computations performed by local neural populations, such as a cortical layer, are typically inferred from anatomical connectivity and observations of neural activity. Here we describe a method-influence mapping-that uses single-neuron perturbations to directly measure how cortical neurons reshape sensory representations. In layer 2/3 of the primary visual cortex (V1), we use two-photon optogenetics to trigger action potentials in a targeted neuron and calcium imaging to measure the effect on spiking in neighbouring neurons in awake mice viewing visual stimuli. Excitatory neurons on average suppressed other neurons and had a centre-surround influence profile over anatomical space. A neuron's influence on its neighbour depended on their similarity in activity. Notably, neurons suppressed activity in similarly tuned neurons more than in dissimilarly tuned neurons. In addition, photostimulation reduced the population response, specifically to the targeted neuron's preferred stimulus, by around 2%. Therefore, V1 layer 2/3 performed feature competition, in which a like-suppresses-like motif reduces redundancy in population activity and may assist with inference of the features that underlie sensory input. We anticipate that influence mapping can be extended to investigate computations in other neural populations.

Historical pitfalls and new directions in the neuroscience of emotion

In this article, we offer a brief history summarizing the last century of neuroscientific study of emotion, highlighting dominant themes that run through various schools of thought. We then summarize the current state of the field, followed by six key points for scientific progress that are inspired by a multi-level constructivist theory of emotion.

Concepts dissolve artificial boundaries in the study of emotion and cognition, uniting body, brain, and mind

Theories of emotion have often maintained artificial boundaries: for instance, that cognition and emotion are separable, and that an emotion concept is separable from the emotional events that comprise its category (e.g. "fear" is distinct from instances of fear). Over the past several years, research has dissolved these artificial boundaries, suggesting instead that conceptual construction is a domain-general process-a process by which the brain makes meaning of the world. The brain constructs emotion concepts, but also cognitions and perceptions, all in the service of guiding action. In this view, concepts are multimodal constructions, dynamically prepared from a set of highly variable instances. This approach obviates old questions (e.g. how does cognition regulate emotion?) but generates new ones (e.g. how does a brain learn emotion concepts?). In this paper, we review this constructionist, predictive coding account of emotion, considering its implications for health and well-being, culture and development.

Concepts, Goals and the Control of Survival-Related Behaviors

Scientists have long studied the actions that impact basic survival in various domains of life, such as defense, foraging, reproduction, thermoregulation, and so on, as if such actions will reveal the nature of emotion. Each domain of survival came to be characterized by a repertoire of distinct actions, and each action was thought to be caused by a dedicated neural circuit, called a survival circuit. Survival circuits are thought to be triggered by sensory events in the world, quickly producing obligatory, stereotypic reflexes as well as more flexible, deliberate responses. In this paper, we consider recent evidence from behavioral ecology that even so-called "reflexes" are better understood as purposeful, flexible actions that unfold across a range of temporal trajectories. They are highly context-dependent and tailored to the requirements of the situation. We then consider evidence from the neuroscience of motor control that motor actions are assembled by neural populations, not triggered by simple circuits. We end by considering the value of these suggestions for understanding the species-general vs. species-specific contributions to emotion.

Primary Interoceptive Cortex Activity during Simulated Experiences of the Body

Studies of the classic exteroceptive sensory systems (e.g., vision, touch) consistently demonstrate that vividly imagining a sensory experience of the world-simulating it-is associated with increased activity in the corresponding primary sensory cortex. We hypothesized, analogously, that simulating internal bodily sensations would be associated with increased neural activity in primary interoceptive cortex. An immersive, language-based mental imagery paradigm was used to test this hypothesis (e.g., imagine your heart pounding during a roller coaster ride, your face drenched in sweat during a workout). During two neuroimaging experiments, participants listened to vividly described situations and imagined "being there" in each scenario. In Study 1, we observed significantly heightened activity in primary interoceptive cortex (of dorsal posterior insula) during imagined experiences involving vivid internal sensations. This effect was specific to interoceptive simulation: It was not observed during a separate affect focus condition in Study 1 nor during an independent Study 2 that did not involve detailed simulation of internal sensations (instead involving simulation of other sensory experiences). These findings underscore the large-scale predictive architecture of the brain and reveal that words can be powerful drivers of bodily experiences.

The theory of constructed emotion: an active inference account of interoception and categorization

The science of emotion has been using folk psychology categories derived from philosophy to search for the brain basis of emotion. The last two decades of neuroscience research have brought us to the brink of a paradigm shift in understanding the workings of the brain, however, setting the stage to revolutionize our understanding of what emotions are and how they work. In this article, we begin with the structure and function of the brain, and from there deduce what the biological basis of emotions might be. The answer is a brain-based, computational account called the theory of constructed emotion.

Comprehensive review: Computational modelling of schizophrenia

Computational modelling has been used to address: (1) the variety of symptoms observed in schizophrenia using abstract models of behavior (e.g. Bayesian models - top-down descriptive models of psychopathology); (2) the causes of these symptoms using biologically realistic models involving abnormal neuromodulation and/or receptor imbalance (e.g. connectionist and neural networks - bottom-up realistic models of neural processes). These different levels of analysis have been used to answer different questions (i.e. understanding behavioral vs. neurobiological anomalies) about the nature of the disorder. As such, these computational studies have mostly supported diverging hypotheses of schizophrenia's pathophysiology, resulting in a literature that is not always expanding coherently. Some of these hypotheses are however ripe for revision using novel empirical evidence. Here we present a review that first synthesizes the literature of computational modelling for schizophrenia and psychotic symptoms into categories supporting the dopamine, glutamate, GABA, dysconnection and Bayesian inference hypotheses respectively. Secondly, we compare model predictions against the accumulated empirical evidence and finally we identify specific hypotheses that have been left relatively under-investigated.

Mixed emotions in the predictive brain

Understanding complex or mixed emotions first requires an exploration of the human nervous system underlying emotions, and indeed all experience. We review current research in neuroscience, which describes the brain as a predictive, internal model of the world that flexibly combines features from past experience to construct emotions. We argue that "mixed emotions" result when these features of past experience correspond to multiple emotion categories. Integrating event perception and cognitive linguistic theories, we propose that "mixed emotions" are perceived as an episode of distinct, linked emotional events due to attentional shifts which update the predicted model of experience. These proposed mechanisms have profound implications for the study of emotion; we conclude by suggesting methodological improvements for future research.

Know thy agency in predictive coding: Meta-monitoring over forward modeling

Though the computation of agency is thought to be based on prediction error, it is important for us to grasp our own reliability of that detected error. Here, the current study shows that we have a meta-monitoring ability over our own forward model, where the accuracy of motor prediction and therefore of the felt agency are implicitly evaluated. Healthy participants (N=105) conducted a simple motor control task and SELF or OTHER visual feedback was given. The relationship between the accuracy and confidence in a mismatch detection task and in a self-other attribution task was examined. The results suggest an accuracy-confidence correlation in both tasks, indicating our meta-monitoring ability over such decisions. Furthermore, a statistically identified group with low accuracy and low confidence was characterized as higher schizotypal people. Finally, what we can know about our own forward model and how we can know it is discussed.

Experimental evidence for circular inference in schizophrenia

Schizophrenia (SCZ) is a complex mental disorder that may result in some combination of hallucinations, delusions and disorganized thinking. Here SCZ patients and healthy controls (CTLs) report their level of confidence on a forced-choice task that manipulated the strength of sensory evidence and prior information. Neither group's responses can be explained by simple Bayesian inference. Rather, individual responses are best captured by a model with different degrees of circular inference. Circular inference refers to a corruption of sensory data by prior information and vice versa, leading us to ‘see what we expect' (through descending loops), to ‘expect what we see' (through ascending loops) or both. Ascending loops are stronger for SCZ than CTLs and correlate with the severity of positive symptoms. Descending loops correlate with the severity of negative symptoms. Both loops correlate with disorganized symptoms. The findings suggest that circular inference might mediate the clinical manifestations of SCZ.

Schizophrenia is a mental disorder characterized by hallucinations and delusions. Here the authors report a novel probabilistic inference task in which compared to healthy subjects, schizophrenia patients show greater degree of circular inference that matches the severity of their clinical symptoms.

Computational Account of Spontaneous Activity as a Signature of Predictive Coding

Spontaneous activity is commonly observed in a variety of cortical states. Experimental evidence suggested that neural assemblies undergo slow oscillations with Up ad Down states even when the network is isolated from the rest of the brain. Here we show that these spontaneous events can be generated by the recurrent connections within the network and understood as signatures of neural circuits that are correcting their internal representation. A noiseless spiking neural network can represent its input signals most accurately when excitatory and inhibitory currents are as strong and as tightly balanced as possible. However, in the presence of realistic neural noise and synaptic delays, this may result in prohibitively large spike counts. An optimal working regime can be found by considering terms that control firing rates in the objective function from which the network is derived and then minimizing simultaneously the coding error and the cost of neural activity. In biological terms, this is equivalent to tuning neural thresholds and after-spike hyperpolarization. In suboptimal working regimes, we observe spontaneous activity even in the absence of feed-forward inputs. In an all-to-all randomly connected network, the entire population is involved in Up states. In spatially organized networks with local connectivity, Up states spread through local connections between neurons of similar selectivity and take the form of a traveling wave. Up states are observed for a wide range of parameters and have similar statistical properties in both active and quiescent state. In the optimal working regime, Up states are vanishing, leaving place to asynchronous activity, suggesting that this working regime is a signature of maximally efficient coding. Although they result in a massive increase in the firing activity, the read-out of spontaneous Up states is in fact orthogonal to the stimulus representation, therefore interfering minimally with the network function.

Spontaneous bursts of activity, commonly observed in the brain, can be understood in terms of error-correcting computation within a neural network. Bursts arise automatically in a network that is inefficiently correcting its internal representation.

A Hierarchical Predictive Coding Model of Object Recognition in Natural Images

Predictive coding has been proposed as a model of the hierarchical perceptual inference process performed in the cortex. However, results demonstrating that predictive coding is capable of performing the complex inference required to recognise objects in natural images have not previously been presented. This article proposes a hierarchical neural network based on predictive coding for performing visual object recognition. This network is applied to the tasks of categorising hand-written digits, identifying faces, and locating cars in images of street scenes. It is shown that image recognition can be performed with tolerance to position, illumination, size, partial occlusion, and within-category variation. The current results, therefore, provide the first practical demonstration that predictive coding (at least the particular implementation of predictive coding used here; the PC/BC-DIM algorithm) is capable of performing accurate visual object recognition.

An active inference theory of allostasis and interoception in depression

In this paper, we integrate recent theoretical and empirical developments in predictive coding and active inference accounts of interoception (including the Embodied Predictive Interoception Coding model) with working hypotheses from the theory of constructed emotion to propose a biologically plausible unified theory of the mind that places metabolism and energy regulation (i.e. allostasis), as well as the sensory consequences of that regulation (i.e. interoception), at its core. We then consider the implications of this approach for understanding depression. We speculate that depression is a disorder of allostasis, whose myriad symptoms result from a 'locked in' brain that is relatively insensitive to its sensory context. We conclude with a brief discussion of the ways our approach might reveal new insights for the treatment of depression.This article is part of the themed issue 'Interoception beyond homeostasis: affect, cognition and mental health'.

Causal Inference for Cross-Modal Action Selection: A Computational Study in a Decision Making Framework

Animals try to make sense of sensory information from multiple modalities by categorizing them into perceptions of individual or multiple external objects or internal concepts. For example, the brain constructs sensory, spatial representations of the locations of visual and auditory stimuli in the visual and auditory cortices based on retinal and cochlear stimulations. Currently, it is not known how the brain compares the temporal and spatial features of these sensory representations to decide whether they originate from the same or separate sources in space. Here, we propose a computational model of how the brain might solve such a task. We reduce the visual and auditory information to time-varying, finite-dimensional signals. We introduce controlled, leaky integrators as working memory that retains the sensory information for the limited time-course of task implementation. We propose our model within an evidence-based, decision-making framework, where the alternative plan units are saliency maps of space. A spatiotemporal similarity measure, computed directly from the unimodal signals, is suggested as the criterion to infer common or separate causes. We provide simulations that (1) validate our model against behavioral, experimental results in tasks where the participants were asked to report common or separate causes for cross-modal stimuli presented with arbitrary spatial and temporal disparities. (2) Predict the behavior in novel experiments where stimuli have different combinations of spatial, temporal, and reliability features. (3) Illustrate the dynamics of the proposed internal system. These results confirm our spatiotemporal similarity measure as a viable criterion for causal inference, and our decision-making framework as a viable mechanism for target selection, which may be used by the brain in cross-modal situations. Further, we suggest that a similar approach can be extended to other cognitive problems where working memory is a limiting factor, such as target selection among higher numbers of stimuli and selections among other modality combinations.

Goal-Directed Decision Making with Spiking Neurons

Behavioral and neuroscientific data on reward-based decision making point to a fundamental distinction between habitual and goal-directed action selection. The formation of habits, which requires simple updating of cached values, has been studied in great detail, and the reward prediction error theory of dopamine function has enjoyed prominent success in accounting for its neural bases. In contrast, the neural circuit mechanisms of goal-directed decision making, requiring extended iterative computations to estimate values online, are still unknown. Here we present a spiking neural network that provably solves the difficult online value estimation problem underlying goal-directed decision making in a near-optimal way and reproduces behavioral as well as neurophysiological experimental data on tasks ranging from simple binary choice to sequential decision making. Our model uses local plasticity rules to learn the synaptic weights of a simple neural network to achieve optimal performance and solves one-step decision-making tasks, commonly considered in neuroeconomics, as well as more challenging sequential decision-making tasks within 1 s. These decision times, and their parametric dependence on task parameters, as well as the final choice probabilities match behavioral data, whereas the evolution of neural activities in the network closely mimics neural responses recorded in frontal cortices during the execution of such tasks. Our theory provides a principled framework to understand the neural underpinning of goal-directed decision making and makes novel predictions for sequential decision-making tasks with multiple rewards.

SIGNIFICANCE STATEMENT Goal-directed actions requiring prospective planning pervade decision making, but their circuit-level mechanisms remain elusive. We show how a model circuit of biologically realistic spiking neurons can solve this computationally challenging problem in a novel way. The synaptic weights of our network can be learned using local plasticity rules such that its dynamics devise a near-optimal plan of action. By systematically comparing our model results to experimental data, we show that it reproduces behavioral decision times and choice probabilities as well as neural responses in a rich set of tasks. Our results thus offer the first biologically realistic account for complex goal-directed decision making at a computational, algorithmic, and implementational level.

Hebbian Wiring Plasticity Generates Efficient Network Structures for Robust Inference with Synaptic Weight Plasticity

In the adult mammalian cortex, a small fraction of spines are created and eliminated every day, and the resultant synaptic connection structure is highly nonrandom, even in local circuits. However, it remains unknown whether a particular synaptic connection structure is functionally advantageous in local circuits, and why creation and elimination of synaptic connections is necessary in addition to rich synaptic weight plasticity. To answer these questions, we studied an inference task model through theoretical and numerical analyses. We demonstrate that a robustly beneficial network structure naturally emerges by combining Hebbian-type synaptic weight plasticity and wiring plasticity. Especially in a sparsely connected network, wiring plasticity achieves reliable computation by enabling efficient information transmission. Furthermore, the proposed rule reproduces experimental observed correlation between spine dynamics and task performance.

Evaluation of ambiguous associations in the amygdala by learning the structure of the environment

Recognizing predictive relationships is critical for survival, but an understanding of the underlying neural mechanisms remains elusive. In particular, it is unclear how the brain distinguishes predictive relationships from spurious ones when evidence about a relationship is ambiguous, or how it computes predictions given such uncertainty. To better understand this process, we introduced ambiguity into an associative learning task by presenting aversive outcomes both in the presence and in the absence of a predictive cue. Electrophysiological and optogenetic approaches revealed that amygdala neurons directly regulated and tracked the effects of ambiguity on learning. Contrary to established accounts of associative learning, however, interference from competing associations was not required to assess an ambiguous cue-outcome contingency. Instead, animals' behavior was explained by a normative account that evaluates different models of the environment's statistical structure. These findings suggest an alternative view of amygdala circuits in resolving ambiguity during aversive learning.

Things happen: Individuals with high obsessive-compulsive tendencies omit agency in their spoken language

The study examined the prediction that obsessive-compulsive tendencies are related to an attenuated sense of agency (SoA). As most explicit agency judgments are likely to reflect also motivation for and expectation of control, we examined agency in sentence production. Reduced agency can be expressed linguistically by omitting the agent or by using grammatical framings that detach the event from the entity that caused it. We examined the use of agentic language of participants with high vs. low scores on a measure of obsessive-compulsive (OC) symptoms, using structured linguistic tasks in which sentences are elicited in a conversation-like setting. As predicted, high OC individuals produced significantly more non-agentic sentences than low OC individuals, using various linguistic strategies. The results suggest that OC tendencies are related to attenuated SoA. We discuss the implications of these findings for explicating the SoA in OCD and the potential contribution of language analysis for understanding psychopathology.

Bayesian Estimation and Inference Using Stochastic Electronics

In this paper, we present the implementation of two types of Bayesian inference problems to demonstrate the potential of building probabilistic algorithms in hardware using single set of building blocks with the ability to perform these computations in real time. The first implementation, referred to as the BEAST (Bayesian Estimation and Stochastic Tracker), demonstrates a simple problem where an observer uses an underlying Hidden Markov Model (HMM) to track a target in one dimension. In this implementation, sensors make noisy observations of the target position at discrete time steps. The tracker learns the transition model for target movement, and the observation model for the noisy sensors, and uses these to estimate the target position by solving the Bayesian recursive equation online. We show the tracking performance of the system and demonstrate how it can learn the observation model, the transition model, and the external distractor (noise) probability interfering with the observations. In the second implementation, referred to as the Bayesian INference in DAG (BIND), we show how inference can be performed in a Directed Acyclic Graph (DAG) using stochastic circuits. We show how these building blocks can be easily implemented using simple digital logic gates. An advantage of the stochastic electronic implementation is that it is robust to certain types of noise, which may become an issue in integrated circuit (IC) technology with feature sizes in the order of tens of nanometers due to their low noise margin, the effect of high-energy cosmic rays and the low supply voltage. In our framework, the flipping of random individual bits would not affect the system performance because information is encoded in a bit stream.

Causal Inference and Explaining Away in a Spiking Network

While the brain uses spiking neurons for communication, theoretical research on brain computations has mostly focused on non-spiking networks. The nature of spike-based algorithms that achieve complex computations, such as object probabilistic inference, is largely unknown. Here we demonstrate that a family of high-dimensional quadratic optimization problems with non-negativity constraints can be solved exactly and efficiently by a network of spiking neurons. The network naturally imposes the non-negativity of causal contributions that is fundamental to causal inference, and uses simple operations, such as linear synapses with realistic time constants, and neural spike generation and reset non-linearities. The network infers the set of most likely causes from an observation using explaining away, which is dynamically implemented by spike-based, tuned inhibition. The algorithm performs remarkably well even when the network intrinsically generates variable spike trains, the timing of spikes is scrambled by external sources of noise, or the network is mistuned. This type of network might underlie tasks such as odor identification and classification.