Explore the vast range of titles available.

Neuroscientist Dean Buonomano illuminates the causes and consequences of the brain's imperfections in terms of its innermost workings and its evolutionary purposes. He then examines how our brains function--and malfunction--in the digital, predator-free, information-saturated, special-effects-addled world we have built for ourselves.

Here the authors describe a recurrent neural network model that tells time on the order of seconds and generates complex spatiotemporal motor patterns in the presence of high levels of noise. Robustness is achieved through the tuning of the recurrent connections, which produces stable patterns in the face of perturbations. The brain's ability to tell time and produce complex spatiotemporal motor patterns is critical for anticipating the next ring of a telephone or playing a musical instrument. One class of models proposes that these abilities emerge from dynamically changing patterns of neural activity generated in recurrent neural networks. However, the relevant dynamic regimes of recurrent networks are highly sensitive to noise; that is, chaotic. We developed a firing rate model that tells time on the order of seconds and generates complex spatiotemporal patterns in the presence of high levels of noise. This is achieved through the tuning of the recurrent connections. The network operates in a dynamic regime that exhibits coexisting chaotic and locally stable trajectories. These stable patterns function as 'dynamic attractors' and provide a feature that is characteristic of biological systems: the ability to 'return' to the pattern being generated in the face of perturbations.

Key Points All forms of sensory processing require sense to be made of the complex spatiotemporal patterns of action potentials that are generated in our sensory organs by external stimuli. Any general model of cortical processing must account for the brain's ability to process both the spatial and the temporal features of stimuli, and thus must account for spatiotemporal processing in general. State-dependent classes of neural network models propose that the temporal information is inherently encoded in the state of the network. The internal state can be divided into the active state, which reflects ongoing neural activity that interacts with incoming external inputs, and the hidden state, which reflects neural properties that change in time even when a network is silent (for example, short-term synaptic plasticity). In vivo electrophysiological recordings show that the neural population response of a network is strongly influenced by preceding activity, and thus that networks behave in a state-dependent manner. A prediction that emerges from the proposed framework is that the neural network response to a given stimulus encodes not only the current stimulus, but also previous stimuli. Most models of sensory processing consider the spatial and temporal aspects of sensory stimuli separately. Here, Buonomano and Maass describe a framework in which spatiotemporal computations emerge from the interaction between incoming stimuli and the internal dynamic state of neural networks. A conspicuous ability of the brain is to seamlessly assimilate and process spatial and temporal features of sensory stimuli. This ability is indispensable for the recognition of natural stimuli. Yet, a general computational framework for processing spatiotemporal stimuli remains elusive. Recent theoretical and experimental work suggests that spatiotemporal processing emerges from the interaction between incoming stimuli and the internal dynamic state of neural networks, including not only their ongoing spiking activity but also their 'hidden' neuronal states, such as short-term synaptic plasticity.

It has been proposed that prediction and timing are computational primitives of neocortical microcircuits, specifically, that neural mechanisms are in place to allow neocortical circuits to autonomously learn the temporal structure of external stimuli and generate internal predictions. To test this hypothesis, we trained cortical organotypic slices on two temporal patterns using dual-optical stimulation. After 24-h of training, whole-cell recordings revealed network dynamics consistent with training-specific timed prediction. Unexpectedly, there was replay of the learned temporal structure during spontaneous activity. Furthermore, some neurons exhibited timed prediction errors as revealed by larger responses when the expected stimulus was omitted compared to when it was present. Mechanistically our results indicate that learning relied in part on asymmetric connectivity between distinct neuronal ensembles with temporally-ordered activation. These findings further suggest that local cortical microcircuits are intrinsically capable of learning temporal information and generating predictions, and that the learning rules underlying temporal learning and spontaneous replay can be intrinsic to local cortical microcircuits and not necessarily dependent on top-down interactions. Because the ability to tell time and make predictions anchor much of cognition, it has been proposed that they are computational primitives. Here, authors directly demonstrated that this is the case by showing that neocortical circuits ex vivo can learn to tell time and make simple predictions.

Converging evidence suggests the brain encodes time in dynamic patterns of neural activity, including neural sequences, ramping activity, and complex dynamics. Most temporal tasks, however, require more than just encoding time, and can have distinct computational requirements including the need to exhibit temporal scaling, generalize to novel contexts, or robustness to noise. It is not known how neural circuits can encode time and satisfy distinct computational requirements, nor is it known whether similar patterns of neural activity at the population level can exhibit dramatically different computational or generalization properties. To begin to answer these questions, we trained RNNs on two timing tasks based on behavioral studies. The tasks had different input structures but required producing identically timed output patterns. Using a novel framework we quantified whether RNNs encoded two intervals using either of three different timing strategies: scaling, absolute, or stimulus-specific dynamics. We found that similar neural dynamic patterns at the level of single intervals, could exhibit fundamentally different properties, including, generalization, the connectivity structure of the trained networks, and the contribution of excitatory and inhibitory neurons. Critically, depending on the task structure RNNs were better suited for generalization or robustness to noise. Further analysis revealed different connection patterns underlying the different regimes. Our results predict that apparently similar neural dynamic patterns at the population level (e.g., neural sequences) can exhibit fundamentally different computational properties in regards to their ability to generalize to novel stimuli and their robustness to noise—and that these differences are associated with differences in network connectivity and distinct contributions of excitatory and inhibitory neurons. We also predict that the task structure used in different experimental studies accounts for some of the experimentally observed variability in how networks encode time.

Much of the information the brain processes and stores is temporal in nature—a spoken word or a handwritten signature, for example, is defined by how it unfolds in time. However, it remains unclear how neural circuits encode complex time-varying patterns. We show that by tuning the weights of a recurrent neural network (RNN), it can recognize and then transcribe spoken digits. The model elucidates how neural dynamics in cortical networks may resolve three fundamental challenges: first, encode multiple time-varying sensory and motor patterns as stable neural trajectories; second, generalize across relevant spatial features; third, identify the same stimuli played at different speeds—we show that this temporal invariance emerges because the recurrent dynamics generate neural trajectories with appropriately modulated angular velocities. Together our results generate testable predictions as to how recurrent networks may use different mechanisms to generalize across the relevant spatial and temporal features of complex time-varying stimuli.

The discrimination and production of temporal patterns on the scale of hundreds of milliseconds are critical to sensory and motor processing. Indeed, most complex behaviours, such as speech comprehension and production, would be impossible in the absence of sophisticated timing mechanisms. Despite the importance of timing to human learning and cognition, little is known about the underlying mechanisms, in particular whether timing relies on specialized dedicated circuits and mechanisms or on general and intrinsic properties of neurons and neural circuits. Here, we review experimental data describing timing and interval-selective neurons in vivo and in vitro. We also review theoretical models of timing, focusing primarily on the state-dependent network model, which proposes that timing in the subsecond range relies on the inherent time-dependent properties of neurons and the active neural dynamics within recurrent circuits. Within this framework, time is naturally encoded in populations of neurons whose pattern of activity is dynamically changing in time. Together, we argue that current experimental and theoretical studies provide sufficient evidence to conclude that at least some forms of temporal processing reflect intrinsic computations based on local neural network dynamics.

The ability to estimate the passage of time is of fundamental importance for perceptual and cognitive processes. One experience of time is the perception of duration, which is not isomorphic to physical duration and can be distorted by a number of factors. Yet, the critical features generating these perceptual shifts in subjective duration are not understood. We used prospective duration judgments within and across sensory modalities to examine the effect of stimulus predictability and feature change on the perception of duration. First, we found robust distortions of perceived duration in auditory, visual and auditory-visual presentations despite the predictability of the feature changes in the stimuli. For example, a looming disc embedded in a series of steady discs led to time dilation, whereas a steady disc embedded in a series of looming discs led to time compression. Second, we addressed whether visual (auditory) inputs could alter the perception of duration of auditory (visual) inputs. When participants were presented with incongruent audio-visual stimuli, the perceived duration of auditory events could be shortened or lengthened by the presence of conflicting visual information; however, the perceived duration of visual events was seldom distorted by the presence of auditory information and was never perceived shorter than their actual durations. These results support the existence of multisensory interactions in the perception of duration and, importantly, suggest that vision can modify auditory temporal perception in a pure timing task. Insofar as distortions in subjective duration can neither be accounted for by the unpredictability of an auditory, visual or auditory-visual event, we propose that it is the intrinsic features of the stimulus that critically affect subjective time distortions.

Many neural computations emerge from self-sustained patterns of activity in recurrent neural circuits, which rely on balanced excitation and inhibition. Neuromorphic electronic circuits represent a promising approach for implementing the brain’s computational primitives. However, achieving the same robustness of biological networks in neuromorphic systems remains a challenge due to the variability in their analog components. Inspired by real cortical networks, we apply a biologically-plausible cross-homeostatic rule to balance neuromorphic implementations of spiking recurrent networks. We demonstrate how this rule can autonomously tune the network to produce robust, self-sustained dynamics in an inhibition-stabilized regime, even in presence of device mismatch. It can implement multiple, co-existing stable memories, with emergent soft-winner-take-all and reproduce the “paradoxical effect” observed in cortical circuits. In addition to validating neuroscience models on a substrate sharing many similar limitations with biological systems, this enables the automatic configuration of ultra-low power, mixed-signal neuromorphic technologies despite the large chip-to-chip variability. Achieving the same robustness of biological networks in neuromorphic systems remains a challenge due to the variability in their analogue components. Here, the authors apply a biologically-plausible cross-homeostatic rule to balance neuromorphic implementations of spiking recurrent networks.

Fragile X syndrome is a neurodevelopmental disorder associated with a broad range of neural phenotypes. Interpreting these findings has proven challenging because some phenotypes may reflect compensatory mechanisms or normal forms of plasticity differentially engaged by experiential differences. To help minimize compensatory and experiential influences, we used an ex vivo approach to study network dynamics and plasticity of cortical microcircuits. In Fmr1 −/y circuits, the spatiotemporal structure of Up-states was less reproducible, suggesting alterations in the plasticity mechanisms governing network activity. Chronic optical stimulation revealed normal homeostatic plasticity of Up-states, however, Fmr1 −/y circuits exhibited abnormal experience-dependent plasticity as they did not adapt to chronically presented temporal patterns in an interval-specific manner. These results, suggest that while homeostatic plasticity is normal, Fmr1 −/y circuits exhibit deficits in the ability to orchestrate multiple forms of synaptic plasticity and to adapt to sensory patterns in an experience-dependent manner—which is likely to contribute to learning deficits.