Parametric control of flexible timing through low-dimensional neural manifolds.

Biological brains possess an unparalleled ability to adapt behavioral responses to changing stimuli and environments. How neural processes enable this capacity is a fundamental open question. Previous works have identified two candidate mechanisms: a low-dimensional organization of neural activity and a modulation by contextual inputs. We hypothesized that combining the two might facilitate generalization and adaptation in complex tasks. We tested this hypothesis in flexible timing tasks where dynamics play a key role. Examining trained recurrent neural networks, we found that confining the dynamics to a low-dimensional subspace allowed tonic inputs to parametrically control the overall input-output transform, enabling generalization to novel inputs and adaptation to changing conditions. Reverse-engineering and theoretical analyses demonstrated that this parametric control relies on a mechanism where tonic inputs modulate the dynamics along non-linear manifolds while preserving their geometry. Comparisons with data from behaving monkeys confirmed the behavioral and neural signatures of this mechanism.

A large-scale neural network training framework for generalized estimation of single-trial population dynamics.

Achieving state-of-the-art performance with deep neural population dynamics models requires extensive hyperparameter tuning for each dataset. AutoLFADS is a model-tuning framework that automatically produces high-performing autoencoding models on data from a variety of brain areas and tasks, without behavioral or task information. We demonstrate its broad applicability on several rhesus macaque datasets: from motor cortex during free-paced reaching, somatosensory cortex during reaching with perturbations, and dorsomedial frontal cortex during a cognitive timing task.

Neural implementations of Bayesian inference.

Bayesian inference has emerged as a general framework that captures how organisms make decisions under uncertainty. Recent experimental findings reveal disparate mechanisms for how the brain generates behaviors predicted by normative Bayesian theories. Here, we identify two broad classes of neural implementations for Bayesian inference: a modular class, where each probabilistic component of Bayesian computation is independently encoded and a transform class, where uncertain measurements are converted to Bayesian estimates through latent processes. Many recent experimental neuroscience findings studying probabilistic inference broadly fall into these classes. We identify potential avenues for synthesis across these two classes and the disparities that, at present, cannot be reconciled. We conclude that to distinguish among implementation hypotheses for Bayesian inference, we require greater engagement among theoretical and experimental neuroscientists in an effort that spans different scales of analysis, circuits, tasks, and species.

A precise and adaptive neural mechanism for predictive temporal processing in the frontal cortex.

The theory of predictive processing posits that the brain computes expectations to process information predictively. Empirical evidence in support of this theory, however, is scarce and largely limited to sensory areas. Here, we report a precise and adaptive mechanism in the frontal cortex of non-human primates consistent with predictive processing of temporal events. We found that the speed of neural dynamics is precisely adjusted according to the average time of an expected stimulus. This speed adjustment, in turn, enables neurons to encode stimuli in terms of deviations from expectation. This lawful relationship was evident across multiple experiments and held true during learning: when temporal statistics underwent covert changes, neural responses underwent predictable changes that reflected the new mean. Together, these results highlight a precise mathematical relationship between temporal statistics in the environment and neural activity in the frontal cortex that may serve as a mechanism for predictive temporal processing.

Validating model-based Bayesian integration using prior-cost metamers.

There are two competing views on how humans make decisions under uncertainty. Bayesian decision theory posits that humans optimize their behavior by establishing and integrating internal models of past sensory experiences (priors) and decision outcomes (cost functions). An alternative hypothesis posits that decisions are optimized through trial and error without explicit internal models for priors and cost functions. To distinguish between these possibilities, we introduce a paradigm that probes the sensitivity of humans to transitions between prior-cost pairs that demand the same optimal policy (metamers) but distinct internal models. We demonstrate the utility of our approach in two experiments that were classically explained by Bayesian theory. Our approach validates the Bayesian learning strategy in an interval timing task but not in a visuomotor rotation task. More generally, our work provides a domain-general approach for testing the circumstances under which humans explicitly implement model-based Bayesian computations.

A Network Perspective on Sensorimotor Learning.

What happens in the brain when we learn? Ever since the foundational work of Cajal, the field has made numerous discoveries as to how experience could change the structure and function of individual synapses. However, more recent advances have highlighted the need for understanding learning in terms of complex interactions between populations of neurons and synapses. How should one think about learning at such a macroscopic level? Here, we develop a conceptual framework to bridge the gap between the different scales at which learning operates, from synapses to neurons to behavior. Using this framework, we explore the principles that guide sensorimotor learning across these scales, and set the stage for future experimental and theoretical work in the field.

Bayesian Computation through Cortical Latent Dynamics.

Statistical regularities in the environment create prior beliefs that we rely on to optimize our behavior when sensory information is uncertain. Bayesian theory formalizes how prior beliefs can be leveraged and has had a major impact on models of perception, sensorimotor function, and cognition. However, it is not known how recurrent interactions among neurons mediate Bayesian integration. By using a time-interval reproduction task in monkeys, we found that prior statistics warp neural representations in the frontal cortex, allowing the mapping of sensory inputs to motor outputs to incorporate prior statistics in accordance with Bayesian inference. Analysis of recurrent neural network models performing the task revealed that this warping was enabled by a low-dimensional curved manifold and allowed us to further probe the potential causal underpinnings of this computational strategy. These results uncover a simple and general principle whereby prior beliefs exert their influence on behavior by sculpting cortical latent dynamics.

A two-stage model of concurrent interval timing in monkeys.

Accurate timing is critical for a wide range of cognitive processes and behaviors. In addition, complex environments frequently necessitate the simultaneous timing of multiple intervals, and behavioral performance in concurrent timing can constrain formal models of timing behavior and provide important insights into the corresponding neural mechanisms. However, the accuracy of such concurrent timing has not been rigorously examined. We developed a novel behavioral paradigm in which rhesus monkeys were incentivized to time two independent intervals. The onset asynchrony of two overlapping intervals varied randomly, thereby discouraging the animals from adopting any habitual responses. We found that only the first response for each interval was strongly indicative of the internal timing of that interval, consistent with previous findings and a two-stage model. In addition, the temporal precision of the first response was comparable in the single-interval and concurrent-interval conditions, although the first saccade to the second interval tended to occur sooner than in the single-interval condition. Finally, behavioral responses during concurrent timing could be well accounted for by a race between two independent stochastic processes resembling those in the single-interval condition. The fact that monkeys can simultaneously monitor and respond to multiple temporal intervals indicates that the neural mechanisms for interval timing must be sufficiently flexible for concurrent timing.

A monocular contribution to stimulus rivalry.

When corresponding areas of the two eyes view dissimilar images, stable perception gives way to visual competition wherein perceptual awareness alternates between those images. Moreover, a given image can remain visually dominant for several seconds at a time even when the competing images are swapped between the eyes multiple times each second. This perceptual stability across eye swaps has led to the widespread belief that this unique form of visual competition, dubbed stimulus rivalry, is governed by eye-independent neural processes at a purely binocular stage of cortical processing. We tested this idea by investigating the influence of stimulus rivalry on the buildup of the threshold elevation aftereffect, a form of contrast adaptation thought to transpire at early cortical stages that include eye-specific neural activity. Weaker threshold elevation aftereffects were observed when the adapting image was engaged in stimulus rivalry than when it was not, indicating diminished buildup of adaptation during stimulus-rivalry suppression. We then confirmed that this reduction occurred, in part, at eye-specific neural stages by showing that suppression of an image at a given moment specifically diminished adaptation associated with the eye viewing the image at that moment. Considered together, these results imply that eye-specific neural events at early cortical processing stages contribute to stimulus rivalry. We have developed a computational model of stimulus rivalry that successfully implements this idea.

Dynamical nonstationarity of resting EEGs in patients with attention-deficit/hyperactivity disorder (AD/HD).

This study applied dynamical nonstationarity analysis (DNA) to the resting EEGs of patients with attention-deficit/hyperactivity disorder (AD/HD). We aimed to assess and characterize AD/HD using features based on the local and global duration of dynamical microstate. We hypothesized that AD/HD patients would have difficulties in maintaining stable cognitive states (e.g., attention deficit and impulsivity) and that they would thus exhibit EEGs with temporal dynamics distinct from normal controls, i.e., rapidly and frequently changing dynamics. To test this hypothesis, we recorded EEGs from 12 adolescent subjects with AD/HD and 11 age-matched healthy subjects in the resting state with eyes closed and eyes open. We found that AD/HD patients exhibited significantly faster changes in dynamics than controls in the right temporal region during the eyes closed condition, but slower changes in dynamics in the frontal region during the eyes open condition. AD/HD patients exhibited a disruption in the rate of change of dynamics in the frontotemporal region at rest, probably due to executive and attention processes. We suggest that the DNA using complementary local and global features based on the duration of dynamical microstates could be a useful tool for the clinical diagnosis of subjects with AD/HD.

Increased frontomotor oscillations during tic suppression in children with Tourette syndrome.

This work investigated whether Tourette syndrome patients exhibit alterations in neural oscillations during spontaneous expression and suppression of tics. Electroencephalograms (EEGs) were recorded from 9 medication-naïve children with Tourette syndrome and 10 age-matched healthy subjects in resting conditions and during tic suppression. Their cortical oscillations were examined using the power spectral method and partial directed coherence. The authors found increased oscillations of broad frequency bands in the frontomotor regions of patients during tic expression, suggesting the involvement of aberrant cortical oscillations in Tourette syndrome. More significantly, prominent increases in theta oscillation in the prefrontal area and directed frontomotor interactions in the theta and beta bands were observed during tic suppression. Furthermore, the directed EEG interaction from the frontal to motor regions was positively correlated with the severity of tic symptoms. These findings suggest that the frontal to motor interaction of cortical oscillations plays a significant role in tic suppression.

Dichotomy in perceptual learning of interval timing: calibration of mean accuracy and precision differ in specificity and time course.

Our brain is inexorably confronted with a dynamic environment in which it has to fine-tune spatiotemporal representations of incoming sensory stimuli and commit to a decision accordingly. Among those representations needing constant calibration is interval timing, which plays a pivotal role in various cognitive and motor tasks. To investigate how perceived time interval is adjusted by experience, we conducted a human psychophysical experiment using an implicit interval-timing task in which observers responded to an invisible bar drifting at a constant speed. We tracked daily changes in distributions of response times for a range of physical time intervals over multiple days of training with two major types of timing performance, mean accuracy and precision. We found a decoupled dynamics of mean accuracy and precision in terms of their time course and specificity of perceptual learning. Mean accuracy showed feedback-driven instantaneous calibration evidenced by a partial transfer around the time interval trained with feedback, while timing precision exhibited a long-term slow improvement with no evident specificity. We found that a Bayesian observer model, in which a subjective time interval is determined jointly by a prior and likelihood function for timing, captures the dissociative temporal dynamics of the two types of timing measures simultaneously. Finally, the model suggested that the width of the prior, not the likelihoods, gradually shrinks over sessions, substantiating the important role of prior knowledge in perceptual learning of interval timing.

Disturbances of motivational balance in chronic schizophrenia during decision-making tasks.

AIM: The role of feedback processing in decision-making has been assessed in psychiatric patients using the Iowa Gambling Task (IGT). Although impaired performance on the IGT has been documented extensively in schizophrenia patients, the neuropsychological mechanisms underlying the performance deficits have not yet been elucidated. Therefore, the aim of this study was to investigate the neuropsychological origins of impaired decision-making in schizophrenia patients using various versions of the IGT. METHODS: Thirty chronic schizophrenia patients and 33 healthy subjects underwent computerized versions of the IGT, the Variant Gambling Task (VGT), and the Shuffled Gambling Task (SGT) to assess the contributions of motivational balance and reversal learning on IGT performance. In addition, performance on the Wisconsin Card-Sorting Test (WCST) was assessed. RESULTS: The schizophrenia patients exhibited deficits on the IGT and SGT, particularly in later trials. No significant group difference was detected on the VGT due to the improved performance of schizophrenia patients in the earlier trials. Performance on the gambling tasks in the schizophrenia group did not correlate with performance on the WCST or with the severity of clinical symptoms. CONCLUSION: Deficits in motivational balance, but not reversal learning, play a dominant role in the impaired decision-making of patients with schizophrenia.

Delayed transition from ambiguous to risky decision making in alcohol dependence during Iowa Gambling Task.

It has been demonstrated that alcohol-dependent patients exhibit decision-making deficits, particularly, hypersensitivity to reward and executive dysfunction. Yet, how the impaired motivational process and executive dysfunction in the patients affect decisions under ambiguity and risk with different degrees of uncertainty is little known. To investigate the neuropsychological origin of the impaired decision making under uncertainty in alcohol dependence, we administered the Iowa Gambling Task (IGT), Game of Dice Task (GDT) and Wisconsin Card Sorting Test (WCST) to 23 alcohol-dependent patients and 21 healthy subjects, and calculated the correlations between the task performances. We found that the patients showed poor performance in all three tasks compared with the healthy subjects. Moreover, correlations between performances on the GDT and the later trials of the IGT were delayed in alcohol-dependent patients when compared with healthy subjects. There is also a significant correlation between performances of earlier trials of the IGT and the WCST in the patients. These findings suggest that executive dysfunction in alcohol-dependent patients hampers appropriate estimation of probability distributions of possible alternatives, leading to a delayed transition from ambiguous to risky conditions in the Iowa Gambling Task.

Linear and non-linear EEG analysis of adolescents with attention-deficit/hyperactivity disorder during a cognitive task.

OBJECTIVE: We aimed to investigate whether electroencephalograph (EEG) dynamics differ in adolescents with attention-deficit/hyperactivity disorder (ADHD) compared with healthy subjects during the performance of a cognitive task. METHODS: We recorded EEGs from 19 scalp electrodes in 11 adolescent boys with ADHD and 12 age-matched healthy boys while the subjects were at rest and during a continuous performance test (CPT). The approximate entropy (ApEn), a non-linear information-theoretic measure, was calculated to quantify the complexity of the EEGs. RESULTS: The mean ApEn of the ADHD patients was significantly lower than the healthy subjects over the right frontal regions (Fp2 and F8) during the performance of the cognitive task, but not at rest. The spectral analysis showed significant differences between the two groups in the P3 and T4 regions at rest and the Fp2 and F8 regions during task performance. CONCLUSIONS: The differences in EEG complexity between the two groups suggest that cortical information processing is altered in ADHD adolescents, and thus their levels of cortical activation may be insufficient to meet the cognitive requirements of attention-demanding tasks. SIGNIFICANCE: This study suggests that a non-linear measure such as ApEn is useful for investigating neural dysfunctions in adolescents with ADHD.