Where’s the Noise? Key Features of Spontaneous Activity and Neural Variability Arise through Learning in a Deterministic Network

Even in the absence of sensory stimulation the brain is spontaneously active. This background “noise” seems to be the dominant cause of the notoriously high trial-to-trial variability of neural recordings. Recent experimental observations have extended our knowledge of trial-to-trial variability and spontaneous activity in several directions: 1. Trial-to-trial variability systematically decreases following the onset of a sensory stimulus or the start of a motor act. 2. Spontaneous activity states in sensory cortex outline the region of evoked sensory responses. 3. Across development, spontaneous activity aligns itself with typical evoked activity patterns. 4. The spontaneous brain activity prior to the presentation of an ambiguous stimulus predicts how the stimulus will be interpreted. At present it is unclear how these observations relate to each other and how they arise in cortical circuits. Here we demonstrate that all of these phenomena can be accounted for by a deterministic self-organizing recurrent neural network model (SORN), which learns a predictive model of its sensory environment. The SORN comprises recurrently coupled populations of excitatory and inhibitory threshold units and learns via a combination of spike-timing dependent plasticity (STDP) and homeostatic plasticity mechanisms. Similar to balanced network architectures, units in the network show irregular activity and variable responses to inputs. Additionally, however, the SORN exhibits sequence learning abilities matching recent findings from visual cortex and the network’s spontaneous activity reproduces the experimental findings mentioned above. Intriguingly, the network’s behaviour is reminiscent of sampling-based probabilistic inference, suggesting that correlates of sampling-based inference can develop from the interaction of STDP and homeostasis in deterministic networks. We conclude that key observations on spontaneous brain activity and the variability of neural responses can be accounted for by a simple deterministic recurrent neural network which learns a predictive model of its sensory environment via a combination of generic neural plasticity mechanisms.     

Fronto-Central Theta Oscillations Are Related to Oscillations in Saccadic Response Times (SRT): An EEG and Behavioral Data Analysis

The phase reset hypothesis states that the phase of an ongoing neural oscillation, reflecting periodic fluctuations in neural activity between states of high and low excitability, can be shifted by the occurrence of a sensory stimulus so that the phase value become highly constant across trials (Schroeder et al., 2008). From EEG/MEG studies it has been hypothesized that coupled oscillatory activity in primary sensory cortices regulates multi sensory processing (Senkowski et al. 2008). We follow up on a study in which evidence of phase reset was found using a purely behavioral paradigm by including also EEG measures. In this paradigm, presentation of an auditory accessory stimulus was followed by a visual target with a stimulus-onset asynchrony (SOA) across a range from 0 to 404 ms in steps of 4 ms. This fine-grained stimulus presentation allowed us to do a spectral analysis on the mean SRT as a function of the SOA, which revealed distinct peak spectral components within a frequency range of 6 to 11 Hz with a modus of 7 Hz. The EEG analysis showed that the auditory stimulus caused a phase reset in 7-Hz brain oscillations in a widespread set of channels. Moreover, there was a significant difference in the average phase at which the visual target stimulus appeared between slow and fast SRT trials. This effect was evident in three different analyses, and occurred primarily in frontal and central electrodes.     

Persistent neural activity: prevalence and mechanisms

Persistent neural activity refers to a sustained change in action potential discharge that long outlasts a stimulus. It is found in a diverse set of brain regions and organisms and several in vitro systems, suggesting that it can be considered a universal form of circuit dynamics that can be used as a mechanism for short-term storage and accumulation of sensory or motor information. Both single cell and network mechanisms are likely to co-operate in generating persistent activity in many brain areas.     

Background gamma activity in the rabbit electroencephalogram as an index of the level of sustained (tonic) attention in the active oddball paradigm in rabbits

Tonic brain activity substantially affects the character of subjects' responsiveness to sensory stimuli. The dynamics of background gamma-band activity in rabbit electroencephalogram was investigated in the active oddball paradigm modified for animal studies. It was shown that increase in the power and coherence of gamma activity reflects the target stimulus expectancy. Correct responses to stimuli occur at a particular level of background gamma activity, which is likely to correspond to the optimal level of sustained (tonic) attention. Decrease in the level of background gamma activity leads to omissions of responses to target stimuli, while it's excessive level results in erroneous responses to nontarget stimuli (false alarms). The observed dynamics of tonic gamma activity can be interpreted as the result of variations in the level of tonic cholinergic activation of the brain cortex.     

Learning to use an invisible visual signal for perception

How does the brain construct a percept from sensory signals? One approach to this fundamental question is to investigate perceptual learning as induced by exposure to statistical regularities in sensory signals [1-7]. Recent studies showed that exposure to novel correlations between sensory signals can cause a signal to have new perceptual effects [2, 3]. In those studies, however, the signals were clearly visible. The automaticity of the learning was therefore difficult to determine. Here we investigate whether learning of this sort, which causes new effects on appearance, can be low level and automatic by employing a visual signal whose perceptual consequences were made invisible-a vertical disparity gradient masked by other depth cues. This approach excluded high-level influences such as attention or consciousness. Our stimulus for probing perceptual appearance was a rotating cylinder. During exposure, we introduced a new contingency between the invisible signal and the rotation direction of the cylinder. When subsequently presenting an ambiguously rotating version of the cylinder, we found that the invisible signal influenced the perceived rotation direction. This demonstrates that perception can rapidly undergo "structure learning" by automatically picking up novel contingencies between sensory signals, thus automatically recruiting signals for novel uses during the construction of a percept.     

Predicting the stream of consciousness from activity in human visual cortex

Can the rapid stream of conscious experience be predicted from brain activity alone? Recently, spatial patterns of activity in visual cortex have been successfully used to predict feature-specific stimulus representations for both visible and invisible stimuli. However, because these studies examined only the prediction of static and unchanging perceptual states during extended periods of stimulation, it remains unclear whether activity in early visual cortex can also predict the rapidly and spontaneously changing stream of consciousness. Here, we used binocular rivalry to induce frequent spontaneous and stochastic changes in conscious experience without any corresponding changes in sensory stimulation, while measuring brain activity with fMRI. Using information that was present in the multivariate pattern of responses to stimulus features, we could accurately predict, and therefore track, participants' conscious experience from the fMRI signal alone while it underwent many spontaneous changes. Prediction in primary visual cortex primarily reflected eye-based signals, whereas prediction in higher areas reflected the color of the percept. Furthermore, accurate prediction during binocular rivalry could be established with signals recorded during stable monocular viewing, showing that prediction generalized across viewing conditions and did not require or rely on motor responses. It is therefore possible to predict the dynamically changing time course of subjective experience with only brain activity.     

Sensory information in local field potentials and spikes from visual and auditory cortices: time scales and frequency bands

Studies analyzing sensory cortical processing or trying to decode brain activity often rely on a combination of different electrophysiological signals, such as local field potentials (LFPs) and spiking activity. Understanding the relation between these signals and sensory stimuli and between different components of these signals is hence of great interest. We here provide an analysis of LFPs and spiking activity recorded from visual and auditory cortex during stimulation with natural stimuli. In particular, we focus on the time scales on which different components of these signals are informative about the stimulus, and on the dependencies between different components of these signals. Addressing the first question, we find that stimulus information in low frequency bands (<12 Hz) is high, regardless of whether their energy is computed at the scale of milliseconds or seconds. Stimulus information in higher bands (>50 Hz), in contrast, is scale dependent, and is larger when the energy is averaged over several hundreds of milliseconds. Indeed, combined analysis of signal reliability and information revealed that the energy of slow LFP fluctuations is well related to the stimulus even when considering individual or few cycles, while the energy of fast LFP oscillations carries information only when averaged over many cycles. Addressing the second question, we find that stimulus information in different LFP bands, and in different LFP bands and spiking activity, is largely independent regardless of time scale or sensory system. Taken together, these findings suggest that different LFP bands represent dynamic natural stimuli on distinct time scales and together provide a potentially rich source of information for sensory processing or decoding brain activity.     

Reduced variability of ongoing and evoked cortical activity leads to improved behavioral performance

Sensory responses of the brain are known to be highly variable, but the origin and functional relevance of this variability have long remained enigmatic. Using the variable foreperiod of a visual discrimination task to assess variability in the primate cerebral cortex, we report that visual evoked response variability is not only tied to variability in ongoing cortical activity, but also predicts mean response time. We used cortical local field potentials, simultaneously recorded from widespread cortical areas, to gauge both ongoing and visually evoked activity. Trial-to-trial variability of sensory evoked responses was strongly modulated by foreperiod duration and correlated both with the cortical variability before stimulus onset as well as with response times. In a separate set of experiments we probed the relation between small saccadic eye movements, foreperiod duration and manual response times. The rate of eye movements was modulated by foreperiod duration and eye position variability was positively correlated with response times. Our results indicate that when the time of a sensory stimulus is predictable, reduction in cortical variability before the stimulus can improve normal behavioral function that depends on the stimulus.     

Saccadic Reaction Times to Audiovisual Stimuli Show Effects of Oscillatory Phase Reset

Initiating an eye movement towards a suddenly appearing visual target is faster when an accessory auditory stimulus occurs in close spatiotemporal vicinity. Such facilitation of saccadic reaction time (SRT) is well-documented, but the exact neural mechanisms underlying the crossmodal effect remain to be elucidated. From EEG/MEG studies it has been hypothesized that coupled oscillatory activity in primary sensory cortices regulates multisensory processing. Specifically, it is assumed that the phase of an ongoing neural oscillation is shifted due to the occurrence of a sensory stimulus so that, across trials, phase values become highly consistent (phase reset). If one can identify the phase an oscillation is reset to, it is possible to predict when temporal windows of high and low excitability will occur. However, in behavioral experiments the pre-stimulus phase will be different on successive repetitions of the experimental trial, and average performance over many trials will show no signs of the modulation. Here we circumvent this problem by repeatedly presenting an auditory accessory stimulus followed by a visual target stimulus with a temporal delay varied in steps of 2 ms. Performing a discrete time series analysis on SRT as a function of the delay, we provide statistical evidence for the existence of distinct peak spectral components in the power spectrum. These frequencies, although varying across participants, fall within the beta and gamma range (20 to 40 Hz) of neural oscillatory activity observed in neurophysiological studies of multisensory integration. Some evidence for high-theta/alpha activity was found as well. Our results are consistent with the phase reset hypothesis and demonstrate that it is amenable to testing by purely psychophysical methods. Thus, any theory of multisensory processes that connects specific brain states with patterns of saccadic responses should be able to account for traces of oscillatory activity in observable behavior.     

Task-induced neural covariability as a signature of approximate Bayesian learning and inference

Perception is often characterized computationally as an inference process in which uncertain or ambiguous sensory inputs are combined with prior expectations. Although behavioral studies have shown that observers can change their prior expectations in the context of a task, robust neural signatures of task-specific priors have been elusive. Here, we analytically derive such signatures under the general assumption that the responses of sensory neurons encode posterior beliefs that combine sensory inputs with task-specific expectations. Specifically, we derive predictions for the task-dependence of correlated neural variability and decision-related signals in sensory neurons. The qualitative aspects of our results are parameter-free and specific to the statistics of each task. The predictions for correlated variability also differ from predictions of classic feedforward models of sensory processing and are therefore a strong test of theories of hierarchical Bayesian inference in the brain. Importantly, we find that Bayesian learning predicts an increase in so-called “differential correlations” as the observer’s internal model learns the stimulus distribution, and the observer’s behavioral performance improves. This stands in contrast to classic feedforward encoding/decoding models of sensory processing, since such correlations are fundamentally information-limiting. We find support for our predictions in data from existing neurophysiological studies across a variety of tasks and brain areas. Finally, we show in simulation how measurements of sensory neural responses can reveal information about a subject’s internal beliefs about the task. Taken together, our results reinterpret task-dependent sources of neural covariability as signatures of Bayesian inference and provide new insights into their cause and their function.     

Brain mechanisms associated with top-down processes in perception.

Perception arises through an interaction between sensory input and prior knowledge. We propose that at least two brain areas are required for such an interaction: the ''site'' where analysis of afferent signals occurs and the ''source'' which applies the relevant prior knowledge. In the human brain, functional imaging studies have demonstrated that selective attention modifies activity in early visual processing areas specific to the attended feature. Early processing areas are also modified when prior knowledge permits a percept to emerge from an otherwise meaningless stimulus. Sources of this modification have been identified in parietal cortex and in prefrontal cortex. Modification of early processing areas also occurs on the basis of prior knowledge about the predicted sensory effects of the subject''s own actions. Activity associated with mental imagery resembles that associated with response preparation (for motor imagery) and selective attention (for sensory imagery) suggesting that mental imagery reflects the effects of prior knowledge on sensory processing areas in the absence of sensory input. Damage to sensory processing areas can lead to a form of sensory hallucination which seems to arise from the interaction of prior knowledge with random sensory activity. In contrast, hallucinations associated with schizophrenia may arise from a failure of prior knowledge about motor intentions to modify activity in relevant sensory areas. When functioning normally, this mechanism permits us to distinguish our own actions from those of independent agents in the outside world. Failure to make this distinction correctly may account for the strong association between hallucinations and paranoid delusions in schizophrenia; the patient not only hears voices, but attributes (usually hostile) intentions to these voices.     

The influence of early experience on the development of sensory systems

Once sensory stimuli become able to alter firing patterns in the developing brain, they can influence the maturation of neuronal circuits. Recent experimental studies add to our understanding of precisely which developmental events are affected by early experience. In particular, it appears that experience of the external environment can affect the brain earlier in development and at earlier stages of sensory processing than previously thought. These studies emphasise the developmental importance of the patterning of neuronal firing produced either by sensory stimuli or by spontaneous activity. The timing of action potentials is also an important aspect of several exciting studies describing the mechanisms - anatomical, synaptic, and molecular - by which early experience brings about alterations in the maturation of sensory circuitry. Importantly, this kind of approach can lead to predictions concerning the nature of sensory stimulation that is most effective in instructing brain development.     

Biophysical basis of brain activity: implications for neuroimaging

In vivo 13C magnetic resonance spectroscopy (MRS) studies of the brain have quantitatively assessed rates of glutamate-glutamine cycle (Veye) and glucose oxidation (CMRGle(ox)) by detecting 13C label turnover from glucose to glutamate and glutamine. Contrary to expectations from in vitro and ex vivo studies, the in vivo 13C-MRS results demonstrate that glutamate recycling is a major metabolic pathway, inseparable from its actions of neurotransmission. Furthermore, both in the awake human and in the anesthetized rat brain, Veye and CMRGle(ox) are stoichiometrically related, where more than two thirds of the energy from glucose oxidation supports events associated with glutamate neurotransmission. The high energy consumption of the brain measured at rest and its quantitative relation to neurotransmission reflects a sizeable activity level for the resting brain. The high activity of the non-stimulated brain, as measured by cerebral metabolic rate of oxygen use (CMRO2), establishes a new neurophysiological basis of cerebral function that leads to reinterpreting functional imaging data because the large baseline signal is commonly discarded in cognitive neuroscience paradigms. Changes in energy consumption (delta CMRO2%) can also be obtained from magnetic resonance imaging (MRI) experiments, using the blood oxygen level-dependent (BOLD) image contrast, provided that all the separate parameters contributing to the functional MRI (fMRI) signal are measured. The BOLD-derived delta CMRO2% when compared with alterations in neuronal spiking rate (delta v%) during sensory stimulation in the rat reveals a stoichiometric relationship, in good agreement with 13C-MRS results. Hence fMRI when calibrated so as to provide delta CMRO2% can provide high spatial resolution evaluation of neuronal activity. Our studies of quantitative measurements of changes in neuroenergetics and neurotransmission reveal that a stimulus does not provoke an arbitrary amount of activity in a localized region, rather a total level of activity is required where the increment is inversely related to the level of activity in the non-stimulated condition. These biophysical experiments have established relationships between energy consumption and neuronal activity that provide novel insights into the nature of brain function and the interpretation of fMRI data.     

Sustained neural rhythms reveal endogenous oscillations supporting speech perception

Rhythmic sensory or electrical stimulation will produce rhythmic brain responses. These rhythmic responses are often interpreted as endogenous neural oscillations aligned (or “entrained”) to the stimulus rhythm. However, stimulus-aligned brain responses can also be explained as a sequence of evoked responses, which only appear regular due to the rhythmicity of the stimulus, without necessarily involving underlying neural oscillations. To distinguish evoked responses from true oscillatory activity, we tested whether rhythmic stimulation produces oscillatory responses which continue after the end of the stimulus. Such sustained effects provide evidence for true involvement of neural oscillations. In Experiment 1, we found that rhythmic intelligible, but not unintelligible speech produces oscillatory responses in magnetoencephalography (MEG) which outlast the stimulus at parietal sensors. In Experiment 2, we found that transcranial alternating current stimulation (tACS) leads to rhythmic fluctuations in speech perception outcomes after the end of electrical stimulation. We further report that the phase relation between electroencephalography (EEG) responses and rhythmic intelligible speech can predict the tACS phase that leads to most accurate speech perception. Together, we provide fundamental results for several lines of research—including neural entrainment and tACS—and reveal endogenous neural oscillations as a key underlying principle for speech perception.

Just as a child on a swing continues to move after the pushing stops, this study reveals similar entrained rhythmic echoes in brain activity after hearing speech and electrical brain stimulation; perturbation with tACS shows that these brain oscillations help listeners to understand speech.     

A Bayesian Attractor Model for Perceptual Decision Making

Even for simple perceptual decisions, the mechanisms that the brain employs are still under debate. Although current consensus states that the brain accumulates evidence extracted from noisy sensory information, open questions remain about how this simple model relates to other perceptual phenomena such as flexibility in decisions, decision-dependent modulation of sensory gain, or confidence about a decision. We propose a novel approach of how perceptual decisions are made by combining two influential formalisms into a new model. Specifically, we embed an attractor model of decision making into a probabilistic framework that models decision making as Bayesian inference. We show that the new model can explain decision making behaviour by fitting it to experimental data. In addition, the new model combines for the first time three important features: First, the model can update decisions in response to switches in the underlying stimulus. Second, the probabilistic formulation accounts for top-down effects that may explain recent experimental findings of decision-related gain modulation of sensory neurons. Finally, the model computes an explicit measure of confidence which we relate to recent experimental evidence for confidence computations in perceptual decision tasks.     

Dynamic construction of stimulus values in the ventromedial prefrontal cortex

Signals representing the value assigned to stimuli at the time of choice have been repeatedly observed in ventromedial prefrontal cortex (vmPFC). Yet it remains unknown how these value representations are computed from sensory and memory representations in more posterior brain regions. We used electroencephalography (EEG) while subjects evaluated appetitive and aversive food items to study how event-related responses modulated by stimulus value evolve over time. We found that value-related activity shifted from posterior to anterior, and from parietal to central to frontal sensors, across three major time windows after stimulus onset: 150-250 ms, 400-550 ms, and 700-800 ms. Exploratory localization of the EEG signal revealed a shifting network of activity moving from sensory and memory structures to areas associated with value coding, with stimulus value activity localized to vmPFC only from 400 ms onwards. Consistent with these results, functional connectivity analyses also showed a causal flow of information from temporal cortex to vmPFC. Thus, although value signals are present as early as 150 ms after stimulus onset, the value signals in vmPFC appear relatively late in the choice process, and seem to reflect the integration of incoming information from sensory and memory related regions.     

Sensory systems

Our understanding of sensory systems has grown impressively in recent years as a result of intense efforts to characterize the mechanisms underlying perception. A large body of evidence has accrued regarding the processes through which sensory information at the biochemical, electrophysiological, and systems levels contributes to the conscious experience of a stimulus. Our efforts to understand the function of sensory systems have been aided by the development of new techniques, including powerful methods of molecular biology, refined short- and long-term approaches to recording from single and multiple neurons, and non-invasive neuroimaging techniques that allow us to study activity within the human brain while subjects perform a variety of cognitive tasks. In future research, the last approach is likely to form a bridge between the large body of electrophysiological knowledge acquired in animal experiments and that currently being obtained in human imaging research.     

Exploring Memory Representations with Activity-Based Genetics

The brain is thought to represent specific memories through the activity of sparse and distributed neural ensembles. In this review, we examine the use of immediate early genes (IEGs), genes that are induced by neural activity, to specifically identify and genetically modify neurons activated naturally by environmental experience. Recent studies using this approach have identified cellular and molecular changes specific to neurons activated during learning relative to their inactive neighbors. By using opto- and chemogenetic regulators of neural activity, the neurons naturally recruited during learning can be artificially reactivated to directly test their role in coding external information. In contextual fear conditioning, artificial reactivation of learning-induced neural ensembles in the hippocampus or neocortex can substitute for the context itself. That is, artificial stimulation of these neurons can apparently cause the animals to “think” they are in the context. This represents a powerful approach to testing the principles by which the brain codes for the external world and how these circuits are modified with learning.A central feature of nervous systems is that, to function properly, specific neurons must become active in response to specific stimuli. The nature of this selective activation and its modification with experience is the focus of much neuroscience research, ranging from studies of sensory processing in experimental animals to disorders of thought such as schizophrenia in humans. The central dogma of neuroscience is that perceptions, memories, thoughts, and higher mental functions arise from the pattern and timing of the activity in neural ensembles in specific parts of the brain at specific points in time. Until quite recently, the investigation of these “circuit”-based questions has primarily been limited to observational techniques, such as single unit recording, functional magnetic resonance imagery (fMRI), and calcium imaging, to document the patterns of neural activity evoked by sensory experience or even complex psychological contingencies in human fMRI studies. These techniques have been enormously successful and created a framework for understanding information processing in the brain. For example, recordings in the visual system have indicated that, in the primary visual cortex, neurons are tuned to the orientation of linear stimuli (Hubel and Wiesel 1962). In contrast, neurons in higher brain areas can respond to discrete items. The most striking example of this specificity comes from in vivo recording in the human medial temporal lobe in which single units have been identified that respond to photos of the actress Halle Berry as well as her written name (Quiroga et al. 2005). This highly selective tuning of neural activity is suggestive of function, but how can this be directly tested? What would be the effect of stimulating just this rare population of neurons, a memory of the actress, a sensory illusion of her image? How does this type of specific firing arise? Do these neurons differ from their nonresponsive neighbors in terms of biochemistry, cell biology, or connectivity? Do they undergo molecular alterations when new information is learned about this individual and are these changes required for the learning? These types of questions have recently become accessible to study in mice through the use of activity-based genetic manipulation, in which neurons that are activated by a specific sensory stimulus can be altered to express any gene of experimental interest. These studies and approaches will be the focus of this work.     

Probing the cortical neuronal correlates of a sensory discrimination process

Key to understanding perception is the form of how sensory stimuli are represented in the evoked activity of the brain. Here, we addressed the question of which components of the evoked neuronal activity in the somatosensory cortex represent the stimulus features while trained monkeys discriminated the difference in frequency between two vibrotactile stimuli. We probed whether these cortical neuronal representations are essential to perception. The results show a strong link between the cortical representation of the stimulus and perception.     

Abstract coding of audiovisual speech: beyond sensory representation

Is there a neural representation of speech that transcends its sensory properties? Using fMRI, we investigated whether there are brain areas where neural activity during observation of sublexical audiovisual input corresponds to a listener's speech percept (what is "heard") independent of the sensory properties of the input. A target audiovisual stimulus was preceded by stimuli that (1) shared the target's auditory features (auditory overlap), (2) shared the target's visual features (visual overlap), or (3) shared neither the target's auditory or visual features but were perceived as the target (perceptual overlap). In two left-hemisphere regions (pars opercularis, planum polare), the target invoked less activity when it was preceded by the perceptually overlapping stimulus than when preceded by stimuli that shared one of its sensory components. This pattern of neural facilitation indicates that these regions code sublexical speech at an abstract level corresponding to that of the speech percept.