Introduction: Linking Brain, Cells, and Morphogenesis

• Researchers propose that the same computational principles used by the brain for perception and action (active inference) also guide how cells build body structures (morphogenesis).
• Active inference is a framework where an agent (brain or cell) predicts its sensory inputs and acts to reduce the difference between its expectations and reality.
• This study connects neuroscience with developmental biology by suggesting that errors in information processing—similar to those seen in mental disorders—can lead to developmental defects.
• It offers a new perspective that sees tissues and cell collectives as “intelligent” systems solving a problem much like a chef following a recipe to create the perfect dish.

Active Inference: The Brain’s (and Cells’) Prediction Machine

• Active inference explains how systems generate predictions about what they should sense and then act to make those predictions come true.
• Analogy: Think of a chef tasting a dish and adjusting spices to match the desired recipe. Similarly, cells adjust their behavior to achieve a target anatomy.
• The process minimizes a quantity called free energy, which mathematically represents the “surprise” or error between what is expected and what is experienced.

Precision in Inference: The Key Factor

• Precision is the weight or confidence assigned to sensory information compared to prior beliefs.
• If sensory inputs are given too much weight (high precision), cells overreact to local signals; if too little, they underreact.
• Too high precision is linked to disorders like autism or schizophrenia, where excessive confidence in noisy signals leads to errors.
• Too low precision means the system ignores important signals, resulting in incomplete adjustments.

Simulations: A Step-by-Step Guide to Morphogenesis

• Normal Morphogenesis:
  • Cells begin in an undifferentiated state with random signaling profiles.
  • They use active inference to sense their environment, infer their position, and differentiate accordingly.
  • The collective minimizes free energy and organizes into the proper target structure.
• High Sensory Precision Simulation:
  • Cells assign excessive confidence to local sensory signals.
  • This causes them to overreact to noise, clustering abnormally and failing to follow the overall body plan.
  • Analogy: Like a radio that picks up too much static, making it hard to tune into the right station.
• High Prior Precision Simulation:
  • Cells are overly convinced of an initial identity (for example, thinking they are intestinal cells).
  • They ignore contradictory sensory information, leading to confusion in migration and differentiation.
  • Analogy: Like stubborn cooks who insist on following a wrong recipe despite feedback.
• Low Sensory Precision Simulation:
  • Cells do not react sufficiently to environmental signals.
  • This results in incomplete differentiation and poor migration to target locations.
  • Analogy: Like a chef who barely tastes the dish and misses important flavor adjustments.
• Rescue Simulation:
  • A simulated biomedical intervention reduces excessive sensitivity in a subset of cells.
  • This adjustment restores proper intercellular communication and allows cells to achieve the correct structure.
  • Analogy: Like adding a corrective ingredient to balance an overly spicy dish.

Experimental Test: Thioridazine and Frog Embryos

• Thioridazine is a dopamine receptor blocker used to reduce sensory precision in biological systems.
• In experiments with Xenopus laevis (frog) embryos, treatment with thioridazine led to developmental defects.
• Observed defects included abnormal pigmentation, kinked body axes, edemas, malformed facial features, and gut abnormalities.
• This supports the model’s prediction that precise regulation of sensory information is crucial for proper development.
• Analogy: Just as a miscalibrated sensor in a machine causes errors, incorrect sensory precision in cells disrupts normal development.

Mathematical and Computational Framework

• The study uses variational free energy minimization, a mathematical method, to model cellular behavior.
• Cells are treated as agents that continuously update their beliefs about the world to minimize prediction errors.
• Tools such as Bayesian inference and the concept of Markov blankets help break down how internal states (beliefs) interact with external signals.
• This framework allows simulation of both normal morphogenesis and pathological conditions.

Implications for Regenerative Medicine and Future Directions

• Understanding morphogenesis through active inference opens new avenues for biomedical intervention.
• Manipulating sensory precision may offer strategies to correct developmental defects and improve regenerative outcomes.
• This interdisciplinary approach bridges computational neuroscience with developmental biology.
• Future research may lead to a “computational somatic psychiatry” that diagnoses and treats developmental disorders by modulating cellular decision-making.

Key Conclusions

• Active inference provides a unifying theory for both neural and non-neural systems, explaining how prediction and error minimization drive behavior.
• Errors in precision—whether too high or too low—can lead to developmental abnormalities, paralleling mechanisms observed in mental disorders.
• The research suggests that approaches from computational psychiatry may be applied to regenerative medicine and developmental repair.
• This work lays the groundwork for future therapies that target information processing at the cellular level rather than just genetic or molecular components.

引言：连接大脑、细胞与形态发生

• 研究人员提出，大脑用于感知与行动的主动推理机制同样指导细胞构建身体结构的过程（形态发生）。
• 主动推理是一种框架，系统预测它应感知的内容，并采取行动以减少预期与实际之间的差距。
• 本研究将神经科学与发育生物学联系起来，表明信息处理中的错误——类似于精神疾病中观察到的问题——也可能导致发育缺陷。
• 这种观点将组织和细胞集体视为具备“智能”的系统，像厨师按菜谱烹饪出理想菜肴一样完成目标形态的构建。

主动推理：大脑和细胞的预测机器

• 主动推理解释了系统如何生成关于应感知内容的预测，并采取行动使这些预测成真。
• 类比：就像厨师品尝菜肴并调整调料以达到预期效果，细胞也会调整自身行为以实现目标解剖结构。
• 这一过程通过最小化“自由能”来实现，自由能在数学上衡量预期与实际之间的“惊讶”或误差。

推理精度：关键因素

• 精度指的是系统对感官信息与先验信念赋予的信心权重。
• 若感官输入被赋予过高的权重，细胞便会对局部信号反应过度；反之，则反应不足。
• 过高的精度与自闭症或精神分裂症中的感官过敏有关，即对噪声反应过度。
• 过低的精度则导致系统忽略重要信号，调整不完全。

模拟：形态发生的分步指南

• 正常形态发生：
  • 细胞最初处于未分化状态，信号分布随机。
  • 它们利用主动推理感知环境，推断自身位置，并做出适当分化。
  • 集体行为使系统自由能降低，最终达到正确的目标形态。
• 高感官精度模拟：
  • 细胞对局部感官信号赋予过高信心。
  • 这种过度反应导致细胞异常聚集，无法遵循整体身体规划。
  • 类比：如同收音机接收到太多杂音，难以调到正确的频道。
• 高先验精度模拟：
  • 细胞过于坚信自身初始身份（例如，都认为自己是肠细胞）。
  • 它们忽略与之矛盾的感官信息，导致迁移和分化混乱。
  • 类比：就像固执的厨师拒绝接受反馈，坚持错误的菜谱。
• 低感官精度模拟：
  • 细胞对环境信号反应不足。
  • 结果导致分化不完全和迁移不达标。
  • 类比：如同厨师几乎不尝味道，错失重要的风味调整机会。
• 救治模拟：
  • 通过模拟生物医学干预，降低部分细胞的过高敏感性。
  • 此举恢复了细胞间正常的通讯，使它们能够达到正确的结构配置。
  • 类比：如同加入调味料以平衡过辣的菜肴。

实验测试：硫利达嗪与非洲爪蟾胚胎

• 硫利达嗪是一种多巴胺受体拮抗剂，用于降低生物系统中感官精度。
• 在非洲爪蟾胚胎实验中，硫利达嗪处理导致发育缺陷。
• 观察到的缺陷包括色素减退、身体轴弯曲、水肿、面部异常和肠道畸形。
• 这些结果支持了模型预测：感官信息的精确调控对正常发育至关重要。
• 类比：正如传感器校准错误会导致机器运作出错，不正确的感官精度会破坏细胞正常发育。

数学与计算框架

• 论文采用变分自由能最小化方法来模拟细胞行为。
• 细胞被视为不断更新其环境信念以最小化预测误差的智能体。
• 利用贝叶斯推理、马尔科夫毯等工具分解内部状态与外部信号的相互作用。
• 这一框架为模拟正常与异常的形态发生提供了量化依据。

对再生医学的启示及未来方向

• 通过主动推理理解形态发生，为设计新的生物医学干预措施提供了新思路。
• 调控细胞的感官精度可能成为纠正发育缺陷和改善再生效果的方法。
• 这种跨学科的方法将计算神经科学与发育生物学紧密结合。
• 未来可能发展出“计算体细胞精神病学”，通过调控细胞决策来诊断和治疗发育障碍。

主要结论

• 主动推理为理解神经与非神经系统的运作提供了统一理论，解释了预测与误差最小化如何驱动行为。
• 感官精度的异常（过高或过低）可导致发育异常，其机制与精神疾病中的信息处理错误相似。
• 研究表明，计算精神病学中的治疗策略可能对再生医学和发育修复产生重要启示。
• 这一工作为未来针对细胞信息处理调控的疗法奠定了基础，突破了单纯针对基因或分子层面的干预。