Saltar al contenido ForoHackers

Computational neuroscience: start here

Neurociencia computacional
#747 · Ophuscado · · Denunciar

Computational neuroscience uses mathematics and simulation to understand how nervous systems process information: how a neuron fires, how circuits compute, how brains learn and how signals can be decoded. Its models range from one cell's electrical behaviour to networks of thousands, and they are judged by whether they explain, and predict, recorded data.

Key concepts

  • Neuron models: from the simple integrate-and-fire unit to conductance-based models of ion channels.
  • Synapses and plasticity: how connection strengths change with activity, the basis of learning in models.
  • Network dynamics: how recurrent circuits give rise to oscillations, memory and decisions.
  • Neural coding: how spike rates and timing carry information, and how to decode it from recordings.
  • Data analysis: spike sorting, filtering of signals and statistics on noisy, high-dimensional recordings.

Vocabulary

  • Membrane potential: the voltage across a neuron's membrane, which drives its spikes.
  • Spike train: the list of times at which a neuron fires.
  • Tuning curve: how a neuron's firing rate depends on a stimulus.
  • Hodgkin-Huxley model: the classic set of equations for how a neuron's ion channels generate a spike.

What to study, in order

  1. Calculus, differential equations, probability and basic neuroscience.
  2. Simulate a leaky integrate-and-fire neuron and plot its response to different input currents.
  3. Implement a conductance-based model, and see how its spikes arise from channel dynamics.
  4. Build a small network of spiking neurons, and observe oscillations and synchrony.
  5. Analyse a public recording: compute firing rates, tuning curves and an estimate of what the neurons encode.

Related tags

#computational-biology #simulation #machine-learning #signal-processing #neuromorphic

Inicia sesión o únete para responder.