Inferring hidden structure in multilayered neural circuits

Abstract

A central challenge in sensory neuroscience involves understanding how neural circuits shape computations across cascaded cell layers. Here we develop a computational framework to reconstruct the response properties of experimentally unobserved neurons in the interior of a multilayered neural circuit. We combine non-smooth regularization with proximal consensus algorithms to overcome difficulties in fitting such models that arise from the high dimensionality of their parameter space. Our methods are statistically and computationally efficient, enabling us to rapidly learn hierarchical non-linear models as well as efficiently compute widely used descriptive statistics such as the spike triggered average (STA) and covariance (STC) for high dimensional stimuli. For example, with our regularization framework, we can learn the STA and STC using 5 and 10 minutes of data, respectively, at a level of accuracy that otherwise requires 40 minutes of data without regularization. We apply our framework to retinal ganglion cell processing, learning cascaded linear-nonlinear (LN-LN) models of retinal circuitry, consisting of thousands of parameters, using 40 minutes of responses to white noise. Our models demonstrate a 53% improvement in predicting ganglion cell spikes over classical linear-nonlinear (LN) models. Internal nonlinear subunits of the model match properties of retinal bipolar cells in both receptive field structure and number. Subunits had consistently high thresholds, leading to sparse activity patterns in which only one subunit drives ganglion cell spiking at any time. From the model's parameters, we predict that the removal of visual redundancies through stimulus decorrelation across space, a central tenet of efficient coding theory, originates primarily from bipolar cell synapses. Furthermore, the composite nonlinear computation performed by retinal circuitry corresponds to a boolean OR function applied to bipolar cell feature detectors. Our general computational framework may aid in extracting principles of nonlinear hierarchical sensory processing across diverse modalities from limited data. Author SummaryComputation in neural circuits arises from the cascaded processing of inputs through multiple cell layers. Each of these cell layers performs operations such as filtering and thresholding in order to shape a circuit's output. It remains a challenge to describe both the computations and the mechanisms that mediate them given limited data recorded from a neural circuit. A standard approach to describing circuit computation involves building quantitative encoding models that predict the circuit response given its input, but these often fail to map in an interpretable way onto mechanisms within the circuit. In this work, we build two layer linear-nonlinear cascade models (LN-LN) in order to describe how the retinal output is shaped by nonlinear mechanisms in the inner retina. We find that these LN-LN models, fit to ganglion cell recordings alone, identify filters and nonlinearities that are readily mapp...

Recurrent dynamics of prefrontal cortex during context-dependent decision-making

“…In lieu of estimating a projection subspace using the unitary integration axis, we projected the fixed points we found into the subspace spanned by the top 3 principal components over a collection of recorded activations. This method of lower-dimensional analysis is common in the reverse-engineering literature [16, 18, 19]. In our case, 99% of the variance in network trajectories is accounted for in the top three principal components.…”

“…The computational mechanism governing our optimized network complicates the schemes proposed in related work that employs task-based computational models [4, 13, 19]. The spectral distributions of the network around fixed points, unlike the corresponding distributions of task-based networks, suggest that when artificial neurons are held to simulate experimental ones, integration of input information in the network state space necessarily evolves along hundreds of dimensions.…”

“…One possible critique of the set-up and conclusions drawn here rests on the number of degrees of freedom in constructing artificial models, from choosing a non-linearity to tuning the sparsity of recurrent connections between neurons [19]. Recent work, however, has demonstrated at least partial universality in the ways in which RNNs learn to simulate complex dynamics in spite of respective differences in training methods and other internal features [19].…”

“…This work is largely task oriented: The dynamics of neurons from a network trained to simulate a task of interest are studied in order to gain insight into how the brain may solve the same task [4, 13]. Task-based methodology affords a hypothesis of how the brain performs some computation [4, 13, 14, 19], but leaves open a critical question of whether or not the trained artificial neurons accomplish the task of interest using the same—or even similar—computational techniques as those employed by their supposed biological analogues.…”

Receptive field center-surround interactions mediate context-dependent spatial contrast encoding in the retina

“…First, nonlinear processing in the retina can contribute more to decorrelating retinal ganglion cell (RGC) responses to natural scenes than the RF surround [10] (see also [11,12]). Second, human fixational eye movements can remove spatial correlations in natural inputs before any neural processing takes place [13,14,15,16].…”