从计算机科学的角度，脑部发育可被表述为一个工程学问题：单个细胞如何按照基因组编码的程序自我组织成功能性的脑部网络[1]。研究表明，人脑包含约86亿神经元，每个神经元形成多达1万个连接[1]，而人类基因组容量仅约为1千兆字节（3×10^9比特）[1]。这意味着基因组的信息容量远不足以为每个连接单独编码。
面对这一约束条件，科研人员分析了三种可能的神经回路形成策略[1]。其中，使用身份标签法为每个神经元分配独立标识会需要的存储空间超过基因组容量约30万倍[1]，明显不可行。相比之下，基于细胞系谱的坐标法被证明在信息存储需求上可行——细胞在每次分裂时产生两个女细胞，继承编码其在分裂树中位置的二进制地址[1]。经过n轮分裂后，每个细胞获得唯一的n比特地址[1]，而约33比特足以为86亿个神经元编码[1]。这种机制简洁高效，使得复杂的轴突连接规则可以表示为比特位的掩码操作[1]。
此外，神经网络的变异性随脑部规模的增大而增加[1]。线虫C. elegans仅有302个神经元和约7000个突触[1]，不同个体间的神经网络几乎完全相同；而同卵双胞胎人类的脑部接线图则存在显著差异[1]。这种规律反映了基因编程与个体发育过程中随机因素之间的相互作用。
Researchers have approached the problem of brain development as an engineering challenge, examining how individual cells self-organize according to genetic instructions to form functional neural networks [1]. With a genome capacity of approximately one gigabyte (about 3×10^9 bits) and a human brain containing roughly 8.6 billion neurons that each form up to 10,000 connections, the constraints of genetic storage present a fundamental puzzle for how such complexity can be encoded [1].
Three potential strategies for wiring the brain have been analyzed: identity labeling, which would require 300,000 times more bits than the genome contains; landmarks; and a coordinate-based system [1]. The coordinate approach emerges as the most plausible, operating within genomic constraints by assigning each cell a unique binary address inherited through cell divisions [1]. After n rounds of division, each cell receives an n-bit address arranged in lexicographic order, with storage requirements of approximately 33 bits needed to specify connections among 8.6 billion neurons [1]. Genetic rules can encode instructions such as "flip bit 1" as a mask operation to direct axons to their proper targets [1].
The variability in neural wiring grows with brain size: the nervous system of C. elegans, which contains 302 neurons and approximately 7,000 synapses, remains nearly identical across individual animals, whereas identical twins do not share identical brain wiring diagrams [1]. This pattern suggests that while developmental algorithms provide the fundamental scaffold, random activity introduces increasing variation as brains become larger and more complex [1].