Search

The ability to manipulate the space-time structure of light waves diversifies light-matter interaction and light-driven applications. Conventionally, metasurfaces are employed to locally control the amplitude and phase of light fields by the material response and structure of small meta-atoms. However, the fixed spatial structures of metasurfaces offer limited opportunities. Here, using quantum control we introduce a new approach that enables the amplitude, sign, and even configuration of the generated light fields to be manipulated in an all-optical manner. Following this approach, we demonstrate the generation of flying doughnut terahertz (THz) pulses. We show that the single-cycle THz pulse radiated from the dynamic semiconductor ring current has an electric field structure that is azimuthally polarized and that the space- and time-resolved magnetic field has a strong, isolated longitudinal component. As a first application, we detect absorption features from ambient water vapor on the spatiotemporal structure of the measured electric fields and the calculated magnetic fields. Quantum control is a powerful and flexible route to generating any structured light pulse in the THz range, while pulse compression of cylindrical vector beams is available for very high-power magnetic-pulse generation from the mid-infrared to near UV spectral region. Pulses such as these will serve as unique probes for spectroscopy, imaging, telecommunications, and magnetic materials.

Firing across populations of neurons in many regions of the mammalian brain maintains a temporal memory, a neural timeline of the recent past. Behavioral results demonstrate that people can both remember the past and anticipate the future over an analogous internal timeline. This paper presents a mathematical framework for building this timeline of the future. We assume that the input to the system is a time series of symbols--sparse tokenized representations of the present--in continuous time. The goal is to record pairwise temporal relationships between symbols over a wide range of time scales. We assume that the brain has access to a temporal memory in the form of the real Laplace transform. Hebbian associations with a diversity of synaptic time scales are formed between the past timeline and the present symbol. The associative memory stores the convolution between the past and the present. Knowing the temporal relationship between the past and the present allows one to infer relationships between the present and the future. With appropriate normalization, this Hebbian associative matrix can store a Laplace successor representation and a Laplace predecessor representation from which measures of temporal contingency can be evaluated. The diversity of synaptic time constants allows for learning of non-stationary statistics as well as joint statistics between triplets of symbols. This framework synthesizes a number of recent neuroscientific findings including results from dopamine neurons in the mesolimbic forebrain., Comment: Fixed a bunch of typos. This is the final version of this paper on arXiv

Observations of power laws in neural activity data have raised the intriguing notion that brains may operate in a critical state. One example of this critical state is "avalanche criticality," which has been observed in various systems, including cultured neurons, zebrafish, rodent cortex, and human EEG. More recently, power laws were also observed in neural populations in the mouse under an activity coarse-graining procedure, and they were explained as a consequence of the neural activity being coupled to multiple latent dynamical variables. An intriguing possibility is that avalanche criticality emerges due to a similar mechanism. Here, we determine the conditions under which latent dynamical variables give rise to avalanche criticality. We find that populations coupled to multiple latent variables produce critical behavior across a broader parameter range than those coupled to a single, quasi-static latent variable, but in both cases, avalanche criticality is observed without fine-tuning of model parameters. We identify two regimes of avalanches, both critical but differing in the amount of information carried about the latent variable. Our results suggest that avalanche criticality arises in neural systems in which activity is effectively modeled as a population driven by a few dynamical variables and these variables can be inferred from the population activity., Comment: 19 pages, 5 figures

Human learners can readily understand speech, or a melody, when it is presented slower or faster than usual. Although deep convolutional neural networks (CNNs) are extremely powerful in extracting information from time series, they require explicit training to generalize to different time scales. This paper presents a deep CNN that incorporates a temporal representation inspired by recent findings from neuroscience. In the mammalian brain, time is represented by populations of neurons with temporal receptive fields. Critically, the peaks of the receptive fields form a geometric series, such that the population codes a set of temporal basis functions over log time. Because memory for the recent past is a function of log time, rescaling the input results in translation of the memory. The Scale-Invariant Temporal History Convolution network (SITHCon) builds a convolutional layer over this logarithmically-distributed temporal memory. A max-pool operation results in a network that is invariant to rescalings of time modulo edge effects. We compare performance of SITHCon to a Temporal Convolution Network (TCN). Although both networks can learn classification and regression problems on both univariate and multivariate time series f(t), only SITHCon generalizes to rescalings f(at). This property, inspired by findings from contemporary neuroscience and consistent with findings from cognitive psychology, may enable networks that learn with fewer training examples, fewer weights and that generalize more robustly to out of sample data.

Extracting temporal relationships over a range of scales is a hallmark of human perception and cognition -- and thus it is a critical feature of machine learning applied to real-world problems. Neural networks are either plagued by the exploding/vanishing gradient problem in recurrent neural networks (RNNs) or must adjust their parameters to learn the relevant time scales (e.g., in LSTMs). This paper introduces DeepSITH, a network comprising biologically-inspired Scale-Invariant Temporal History (SITH) modules in series with dense connections between layers. SITH modules respond to their inputs with a geometrically-spaced set of time constants, enabling the DeepSITH network to learn problems along a continuum of time-scales. We compare DeepSITH to LSTMs and other recent RNNs on several time series prediction and decoding tasks. DeepSITH achieves state-of-the-art performance on these problems.

Understanding the activity of large populations of neurons is difficult due to the combinatorial complexity of possible cell-cell interactions. To reduce the complexity, coarse-graining had been previously applied to experimental neural recordings, which showed over two decades of scaling in free energy, activity variance, eigenvalue spectra, and correlation time, hinting that the mouse hippocampus operates in a critical regime. We model the experiment by simulating conditionally independent binary neurons coupled to a small number of long-timescale stochastic fields and then replicating the coarse-graining procedure and analysis. This reproduces the experimentally-observed scalings, suggesting that they may arise from coupling the neural population activity to latent dynamic stimuli. Further, parameter sweeps for our model suggest that emergence of scaling requires most of the cells in a population to couple to the latent stimuli, predicting that even the celebrated place cells must also respond to non-place stimuli.

Advances in neural recording methods enable sampling from populations of thousands of neurons during the performance of behavioral tasks, raising the question of how recorded activity relates to the theoretical models of computations underlying performance. In the context of decision making in rodents, patterns of functional connectivity between choice-selective cortical neurons, as well as broadly distributed choice information in both excitatory and inhibitory populations, were recently reported [1]. The straightforward interpretation of these data suggests a mechanism relying on specific patterns of anatomical connectivity to achieve selective pools of inhibitory as well as excitatory neurons. We investigate an alternative mechanism for the emergence of these experimental observations using a computational approach. We find that a randomly connected network of excitatory and inhibitory neurons generates single-cell selectivity, patterns of pairwise correlations, and indistinguishable excitatory and inhibitory readout weight distributions, as observed in recorded neural populations. Further, we make the readily verifiable experimental predictions that, for this type of evidence accumulation task, there are no anatomically defined sub-populations of neurons representing choice, and that choice preference of a particular neuron changes with the details of the task. This work suggests that distributed stimulus selectivity and patterns of functional organization in population codes could be emergent properties of randomly connected networks.

The ability to measure and control the carrier envelope phase (CEP) of few-cycle laser pulses is of paramount importance for both frequency metrology and attosecond science. Here, we present a phase meter relying on the CEP-dependent photocurrents induced by circularly polarized few-cycle pulses focused between electrodes in ambient air. The new device facilitates compact single-shot, CEP measurements under ambient conditions and promises CEP tagging at repetition rates orders of magnitude higher than most conventional CEP detection schemes as well as straightforward implementation at longer wavelengths.

We introduce a new technique for the generation of magnetic impulses. This technique is based on coherent control of electrical currents using cylindrical laser beams with azimuthal polarization. When used to ionize a medium, in this case atomic hydrogen is considered, an azimuthal current impulse is driven. The spatial distribution of this current bears close resemblance to that of a solenoid, and produces a magnetic field impulse. The excitation and relaxation dynamics of this current temporally confine the resulting magnetic field to a Tesla-scale, terahertz bandwidth impulse. Importantly, the magnetic fields are spatially isolated from electric fields. This all-optical approach will enable ultrafast time-domain spectroscopy of magnetic phenomena., Comment: 5 pages, 4 figures

Natural learners must compute an estimate of future outcomes that follow from a stimulus in continuous time. Widely used reinforcement learning algorithms discretize continuous time and estimate either transition functions from one step to the next (model-based algorithms) or a scalar value of exponentially-discounted future reward using the Bellman equation (model-free algorithms). An important drawback of model-based algorithms is that computational cost grows linearly with the amount of time to be simulated. On the other hand, an important drawback of model-free algorithms is the need to select a time-scale required for exponential discounting. We present a computational mechanism, developed based on work in psychology and neuroscience, for computing a scale-invariant timeline of future outcomes. This mechanism efficiently computes an estimate of inputs as a function of future time on a logarithmically-compressed scale, and can be used to generate a scale-invariant power-law-discounted estimate of expected future reward. The representation of future time retains information about what will happen when. The entire timeline can be constructed in a single parallel operation which generates concrete behavioral and neural predictions. This computational mechanism could be incorporated into future reinforcement learning algorithms., Comment: 25 pages, 10 figures

In both the human brain and any general artificial intelligence (AI), a representation of the past is necessary to predict the future. However, perfect storage of all experiences is not feasible. One approach utilized in many applications, including reward prediction in reinforcement learning, is to retain recently active features of experience in a buffer. Despite its prior successes, we show that the fixed length buffer renders Deep Q-learning Networks (DQNs) fragile to changes in the scale over which information can be learned. To enable learning when the relevant temporal scales in the environment are not known *a priori*, recent advances in psychology and neuroscience suggest that the brain maintains a compressed representation of the past. Here we introduce a neurally-plausible, scale-free memory representation we call Scale-Invariant Temporal History (SITH) for use with artificial agents. This representation covers an exponentially large period of time by sacrificing temporal accuracy for events further in the past. We demonstrate the utility of this representation by comparing the performance of agents given SITH, buffer, and exponential decay representations in learning to play video games at different levels of complexity. In these environments, SITH exhibits better learning performance by storing information for longer timescales than a fixed-size buffer, and representing this information more clearly than a set of exponentially decayed features. Finally, we discuss how the application of SITH, along with other human-inspired models of cognition, could improve reinforcement and machine learning algorithms in general., Comment: Preprint for submission to Neural Computation. Submitted to Neural Computation - Update 12/18/2018: revised based on reviewer comments, resubmitted to Neural Computation on 15 December, 2018. Restructured introduction and discussion, combined figures, added section for SITH parameterization

To compensate for sensory processing delays, the visual system must make predictions to ensure timely and appropriate behaviors. Recent work has found predictive information about the stimulus in neural populations early in vision processing, starting in the retina. However, to utilize this information, cells downstream must in turn be able to read out the predictive information from the spiking activity of retinal ganglion cells. Here we investigate whether a downstream cell could learn efficient encoding of predictive information in its inputs in the absence of other instructive signals, from the correlations in the inputs themselves. We simulate learning driven by spiking activity recorded in salamander retina. We model a downstream cell as a binary neuron receiving a small group of weighted inputs and quantify the predictive information between activity in the binary neuron and future input. Input weights change according to spike timing-dependent learning rules during a training period. We characterize the readouts learned under spike timing-dependent learning rules, finding that although the fixed points of learning dynamics are not associated with absolute optimal readouts, they convey nearly all the information conveyed by the optimal readout. Moreover, we find that learned perceptrons transmit position and velocity information of a moving bar stimulus nearly as efficiently as optimal perceptrons. We conclude that predictive information is, in principle, readable from the perspective of downstream neurons in the absence of other inputs, and consequently suggests that bottom-up prediction may play an important role in sensory processing., Comment: 36 pages, 5 figures, 3 supplemental figures