Emergence of Adaptive Computation by Single Neurons in the Developing Cortex

Rebecca A. Mease, Michael Famulare, Julijana Gjorgjieva, William J. Moody, Adrienne L. FairhallJournal of Neuroscience 2013 · DOI

TL;DR

Developing cortical neurons spontaneously acquire the ability to perform "gain scaling" – a crucial computational strategy for adapting to fluctuating sensory input – solely through the maturation of their intrinsic spike-generating ion channels. This developmental process occurs independently of sensory experience or network activity, suggesting a fundamental, intrinsic mechanism for efficient neural coding.

What Problem Does This Paper Address?

The brain constantly receives a barrage of sensory information that varies in intensity and statistical properties. To make sense of this dynamic environment, neural systems need to adapt their sensitivity. A key mechanism for this adaptation is "gain scaling," where a neuron's response to an input is adjusted based on the typical range of recent inputs. This allows neurons to efficiently represent information across a wide dynamic range without being overwhelmed by strong signals or missing weak ones. While gain scaling is observed in mature neural systems, how this computational ability emerges during development, particularly at the single-neuron level, has been largely unknown. This paper investigates the developmental trajectory of gain scaling in cortical neurons and identifies the underlying biophysical mechanisms.

Methods

The researchers employed a combination of electrophysiology and computational modeling.

  • Electrophysiology: They recorded from single neurons in the developing mouse sensorimotor cortex using whole-cell current-clamp and voltage-clamp techniques. Recordings were made from both organotypic cultures and acute slices at two developmental stages: "immature" (embryonic day 18 to postnatal day 1) and "mature" (postnatal day 6-8). To isolate intrinsic neuronal properties, synaptic blockers were used. They stimulated neurons with Gaussian noise current of varying standard deviations (σ) to probe their input-output relationships.
  • Computational Modeling:
    • Linear-Nonlinear (LN) Models: They characterized neuronal computation by constructing LN models, which represent the neuron's response as a linear filtering of the input followed by a nonlinear transformation. This allowed them to analyze the neuron's "feature selection" (the filter, or Spike-Triggered Average - STA) and "nonlinear encoding" (the input-output relation).
  • Biophysical Models: They used a Hodgkin-Huxley style single-compartment model to simulate neurons with varying maximal sodium (GNa) and potassium (GK) conductances, mimicking developmental changes.
  • Exponential Integrate-and-Fire (EIF) Models: A simplified single-variable model was used for theoretical analysis and to identify the core biophysical conditions for gain scaling.

Datasets

Not discussed in the paper. The study utilized electrophysiological recordings from developing mouse sensorimotor cortex.

Key Insights and Ideas

  • Intrinsic Emergence of Gain Scaling: The most striking finding is that perfect gain scaling emerges in cortical neurons before they are exposed to sensory input or significant network activity. This suggests a built-in developmental program.
  • Role of Spike-Generating Currents: The maturation of the balance between sodium (INa) and potassium (IK) currents, the primary drivers of action potential generation, is identified as the key factor enabling gain scaling. As INa increases relative to IK during development, neurons acquire this adaptive computation.
  • Convergence to a Common Computation: Developing neurons not only gain the ability to perform gain scaling but also converge towards a similar, population-wide input-output function. This suggests a standardization of computational strategy during maturation.
  • Independence from Network Activity: Blocking spontaneous network activity with TTX did not prevent the emergence of gain scaling, decoupling this intrinsic computation from activity-dependent plasticity.
  • Simplified Model Sufficiency: The study demonstrates that even simplified neuron models (EIF) can exhibit gain scaling, highlighting that the core mechanism relies on the fundamental dynamics of spike generation rather than complex, slow adaptation currents.

Key Results

  • Immature vs. Mature Neurons: Immature neurons (E18-P1) showed variable input-output relations and less precise gain scaling compared to mature neurons (P6-P8). Mature neurons exhibited a consistent, population-wide input-output function and near-perfect gain scaling.
  • INa/IK Ratio: The ratio of maximal sodium to potassium currents (INa/IK) increased significantly during early postnatal development, converging to a stable value in mature neurons. This ratio was strongly correlated with improved gain scaling.
  • Pharmacological Manipulation: Reducing transient potassium currents with 4-AP improved gain scaling in mature neurons, while partially blocking sodium channels with TTX disrupted gain scaling, further supporting the role of INa/IK balance.
  • Model Predictions: Biophysical and EIF models successfully replicated the experimental findings, showing that varying the GNa/GK ratio (analogous to INa/IK) directly influenced gain scaling ability and the shape of the input-output relation.

Limitations and Future Work

  • Stimulus Range: While the study explored a range of stimulus standard deviations, the full spectrum of possible input statistics was not exhaustively tested.
  • Synaptic Influences: The study focused on intrinsic properties and used synaptic blockers. The interaction between developing intrinsic gain scaling and the maturation of synaptic inputs remains an open question.
  • Other Ion Channels: While the focus was on INa and IK, the role of other ion channels in modulating gain scaling, particularly slower channels, was not fully explored. The authors suggest that some slower channels might even reduce the inherent gain-scaling property of the spike-generating mechanism.
  • Future Work: The authors suggest investigating how synaptic maturation interacts with intrinsic gain scaling and exploring the role of other ion channels. They also propose that the emergence of gain scaling might play a role in transitioning cortical neurons from participating in spontaneous network activity to becoming efficient, asynchronous information processors.

Summary updated

Abstract

Adaptation is a fundamental computational motif in neural processing. To maintain stable perception in the face of rapidly shifting input, neural systems must extract relevant information from background fluctuations under many different contexts. Many neural systems are able to adjust their input-output properties such that an input's ability to trigger a response depends on the size of that input relative to its local statistical context. This "gain-scaling" strategy has been shown to be an efficient coding strategy. We report here that this property emerges during early development as an intrinsic property of single neurons in mouse sensorimotor cortex, coinciding with the disappearance of spontaneous waves of network activity, and can be modulated by changing the balance of spike-generating currents. Simultaneously, developing neurons move toward a common intrinsic operating point and a stable ratio of spike-generating currents. This developmental trajectory occurs in the absence of sensory input or spontaneous network activity. Through a combination of electrophysiology and modeling, we demonstrate that developing cortical neurons develop the ability to perform nearly perfect gain scaling by virtue of the maturing spike-generating currents alone. We use reduced single neuron models to identify the conditions for this property to hold.
BibTeX
@article{mease2013emergence,
  author = {Rebecca A. Mease and Michael Famulare and Julijana Gjorgjieva and William J. Moody and Adrienne L. Fairhall},
  title = {{Emergence of Adaptive Computation by Single Neurons in the Developing Cortex}},
  journal = {Journal of Neuroscience},
  year = {2013},
  doi = {10.1523/jneurosci.3263-12.2013},
}
Figure