Neuroscience Research Skill

KEY PRINCIPLES: LOOK UP, DON'T GUESS — use PubMed/EuropePMC for neuroanatomy facts, WormBase for C. elegans connectome data, UniProt for neural protein properties. Verify claims with literature before answering. Use Python computation for quantitative neuroscience problems.

LOOK UP, DON'T GUESS

When uncertain about any neuroscience fact — brain region function, neural circuit connectivity, ion channel properties, neurotransmitter receptor subtypes — SEARCH databases first. A PubMed-verified answer is always more reliable than reasoning from memory. This is especially critical for neuroanatomy, where structures have precise boundaries and connectivity patterns that are easy to confuse.

1. Computational Neuroscience Reasoning

Rate-Based Models

• Firing rate of a neuron: r = f(I - theta), where I = total synaptic input, theta = threshold, f = transfer function (sigmoid, ReLU, or threshold-linear)
• Balanced excitation/inhibition: in cortical networks, excitatory and inhibitory inputs are large but nearly cancel, leaving a small net drive
• Population rate equations: tau * dr/dt = -r + f(W*r + I_ext), where W = connectivity matrix
• Steady-state analysis: set dr/dt = 0, solve r = f(W*r + I_ext) — use fixed-point iteration or Newton's method