神经经济学导论：范畴、历史与核心问题 / Introduction to Neuroeconomics: Scope, History, and Core Questions

神经经济学（Neuroeconomics）是一门旨在揭示决策行为背后神经、激素与计算机制的交叉学科。它并非简单的“用脑成像研究经济学”，而是通过整合神经科学、心理学、经济学与计算机科学的理论工具，构建一个从分子、细胞、脑区到行为的统一解释框架。其核心命题是：决策——无论是选择食物、伴侣、还是投资策略——最终由特定神经回路以可计算的方式执行，这些回路的设计反映了进化压力与个体经验的共同塑造。本章将系统界定该领域的边界，追溯其从行为经济学与认知神经科学土壤中萌芽的历史脉络，并提炼出三个相互关联的核心科学问题：决策的神经实现（neural implementation）、计算原理（computational principles）与进化基础（evolutionary foundations）。随后，我们将深入审视关键研究技术——从功能性磁共振成像（fMRI）到光遗传学——各自的优势与局限，并通过伊阿华赌博任务（Iowa Gambling Task）等经典范例，展示如何将行为范式、神经影像与计算模型编织成解释性叙事。最后，我们将直面该领域引发的根本哲学挑战，特别是关于自由意志、道德责任与神经决定论的辩论，并勾勒出全书从多巴胺信号到社会决策的完整探索路径。

The field of neuroeconomics is an interdisciplinary science aimed at uncovering the neural, hormonal, and computational mechanisms underlying decision-making behavior. It is not merely "using brain imaging to study economics," but rather an endeavor to integrate theoretical tools from neuroscience, psychology, economics, and computer science to build a unified explanatory framework spanning molecules, cells, brain regions, and behavior. Its central thesis is that decision-making—whether choosing food, a mate, or an investment strategy—is ultimately executed by specific neural circuits in a computationally tractable manner, with the design of these circuits reflecting the combined shaping of evolutionary pressures and individual experience. This chapter will systematically define the scope of the field, trace its historical emergence from the soil of behavioral economics and cognitive neuroscience, and distill three interconnected core scientific questions: the neural implementation of decisions, the computational principles they follow, and their evolutionary foundations. We will then scrutinize key research technologies—from functional magnetic resonance imaging (fMRI) to optogenetics—highlighting their respective strengths and limitations. Through classic paradigms like the Iowa Gambling Task, we will demonstrate how behavioral tasks, neuroimaging, and computational models can be woven into an explanatory narrative. Finally, we will confront the fundamental philosophical challenges the field raises, particularly debates on free will, moral responsibility, and neural determinism, and outline the book's full journey from dopamine signals to social decision-making.

1.1 范畴界定：神经经济学的研究边界与核心承诺

神经经济学的定义常被误解，其边界需通过排除法精确定义。它不等同于商业神经科学（Consumer Neuroscience），后者主要应用神经技术（如EEG、眼动）于市场营销与广告测试，虽共享方法论，但理论深度与基本问题不同。它也不是心理学或行为经济学的简单延伸；传统行为经济学（如Kahneman与Tversky的前景理论）描述决策偏差的“是什么”（what），而神经经济学追问“如何可能”（how）——即大脑如何物理地实现这些看似非理性的启发式。更确切地说，神经经济学是计算神经科学（Computational Neuroscience）在决策领域的具体化，其核心承诺包含三个层次：

1. 解释性层级（Levels of Explanation）：遵循David Marr的信息处理层级理论，神经经济学致力于在三个层级上提供一致的解释：

• **计算理论层（Computational Theory）**：决策问题是什么？目标函数（如最大化期望效用、最小化预测误差）是什么？约束条件（如认知资源有限）是什么？
• **算法/表示层（Algorithm/Representation）**：大脑使用何种计算程序（如强化学习中的时序差分学习）？信息如何被编码（如价值信号在腹侧纹状体以群体编码表示）？
• **硬件实现层（Hardware Implementation）**：特定算法由哪些神经元集群、神经递质系统（如多巴胺、血清素）和脑区网络（如前额叶皮层-纹状体环路）执行？

2. 价值中心论（Value-Centricity）：尽管决策涉及感知、记忆、运动等多个认知过程，神经经济学将主观价值（Subjective Value）的神经表征与比较视为其枢纽问题。几乎所有决策最终可归结为对不同选项价值信号的评估与选择。因此，寻找“价值神经元”与“价值脑区”成为该领域的标志性追求。

3. 模型驱动（Model-Based）与数据驱动（Data-Driven）的融合：现代神经经济学拒绝纯粹的现象学描述。它强制要求研究者先验地提出一个计算模型（例如，一个参数化的强化学习模型），然后用该模型的参数或预测来拟合行为数据，并进一步检验这些参数是否与神经活动（如fMRI信号强度、神经元放电率）相关。这种“模型→行为→神经”的三角验证是方法论基石。

The Scope and Core Commitments of Neuroeconomics

The definition of neuroeconomics is often misunderstood; its boundaries require precise delineation through exclusion. It is not synonymous with consumer neuroscience, which primarily applies neural technologies (e.g., EEG, eye-tracking) to marketing and advertising testing. While they share methodologies, their theoretical depth and fundamental questions differ. Nor is it a simple extension of psychology or behavioral economics. Traditional behavioral economics (e.g., prospect theory by Kahneman and Tversky) describes the "what" of decision biases, whereas neuroeconomics asks the "how"—how the brain physically implements these seemingly irrational heuristics. More accurately, neuroeconomics is the application of computational neuroscience to decision-making, with core commitments spanning three levels:

1. Levels of Explanation: Following David Marr's framework, neuroeconomics seeks coherent explanations at three tiers:

• **Computational Theory Level**: What is the decision problem? What is the objective function (e.g., maximizing expected utility, minimizing prediction error)? What are the constraints (e.g., limited cognitive resources)?
• **Algorithm/Representation Level**: What computational procedures does the brain use (e.g., temporal-difference learning in reinforcement learning)? How is information encoded (e.g., population coding of value signals in the ventral striatum)?
• **Hardware Implementation Level**: Which specific neuronal clusters, neurotransmitter systems (e.g., dopamine, serotonin), and brain networks (e.g., prefrontal cortex-striatum loops) execute these algorithms?

2. Value-Centricity: Although decision-making involves perception, memory, motor control, and other cognitive processes, neuroeconomics positions the neural representation and comparison of subjective value as its central problem. Almost all decisions can ultimately be reduced to the evaluation and selection among value signals of different options. Thus, the search for "value neurons" and "value brain areas" has become a signature pursuit of the field.

3. Integration of Model-Based and Data-Driven Approaches: Modern neuroeconomics rejects purely phenomenological description. It mandates that researchers first propose a computational model (e.g., a parametric reinforcement learning model), then use its parameters or predictions to fit behavioral data, and subsequently test whether these parameters correlate with neural activity (e.g., fMRI signal intensity, neuronal firing rates). This "model → behavior → brain" triangulation is a methodological cornerstone.

1.2 历史脉络：从行为经济学到神经经济学的三阶段演进

神经经济学的诞生不是偶然的，而是20世纪末至21世纪初多个学科范式转换交汇的结果。其发展可清晰划分为三个阶段，每一阶段都引入了新的理论工具与实证焦点。

第一阶段：行为经济学的挑战（1950s-1980s）。新古典经济学基于“理性人”假设，认为决策者拥有无限认知能力、稳定偏好，并能进行贝叶斯式概率推理。然而，Kahneman与Tversky的一系列开创性研究（如前景理论，1979）系统性地揭示了系统性的、可预测的决策偏差：损失厌恶、确定性效应、框架效应等。这些发现证明，人类决策是“有限理性”的，受到启发式与情绪的系统性影响。行为经济学由此兴起，但其解释仍停留在描述性心理学层面，未触及大脑机制。关键问题悬而未决：这些启发式是进化模块，还是认知资源限制的副产品？情绪如何具体影响价值计算？

第二阶段：认知神经科学的工具革命（1990s-2000s初）。脑成像技术（特别是fMRI）的成熟与普及，使得在人类受试者执行经济博弈任务时观测全脑活动成为可能。同时，灵长类单神经元记录研究（如 Schultz, Dayan, Montague 1997）发现，中脑多巴胺神经元的活动模式与强化学习理论中的奖励预测误差（Reward Prediction Error, RPE） 信号高度吻合。这一发现是里程碑式的：它首次将一个形式化的计算概念（TD误差）与明确的神经信号（多巴胺神经元放电）联系起来。此时，神经经济学作为一个独立学科身份开始凝聚。研究者开始将fMRI用于测试经典博弈论任务（如最后通牒游戏、信任游戏），识别与社会决策相关的脑区（如前扣带回、岛叶、杏仁核）。

第三阶段：整合与分化（2000s中至今）。学科进入成熟期，呈现两大趋势。整合趋势体现为计算模型成为标准配置。研究不再满足于报告“脑区X在条件Y下激活更强”，而是构建定量模型（如强化学习模型、 drift-diffusion model），提取模型参数（如学习率α、逆温度参数β），并将这些参数与神经信号关联。例如，腹侧纹状体的BOLD信号强度可被解释为预期价值（Expected Value），而中脑多巴胺能活动则与RPE相关。分化趋势体现为子领域涌现：社会神经经济学（研究公平、信任、利他）、情感神经经济学（研究恐惧、焦虑对风险决策的影响）、发展神经经济学（研究决策能力的神经发育轨迹）。同时，因果干预技术（如经颅磁刺激TMS、深部脑刺激DBS、光遗传学在模式动物中的应用）开始用于测试脑区与决策的因果关系，超越了fMRI的的相关性局限。

Historical Trajectory: Three Stages from Behavioral Economics to Neuroeconomics

The emergence of neuroeconomics was not accidental but resulted from the convergence of multiple disciplinary paradigm shifts at the turn of the 21st century. Its development can be clearly divided into three stages, each introducing new theoretical tools and empirical foci.

Stage One: The Challenge of Behavioral Economics (1950s-1980s). Neoclassical economics, based on the "rational agent" hypothesis, assumed decision-makers have unlimited cognitive abilities, stable preferences, and perform Bayesian probabilistic reasoning. However, a series of pioneering studies by Kahneman and Tversky (e.g., Prospect Theory, 1979) systematically revealed systematic, predictable decision biases: loss aversion, certainty effect, framing effect, etc. These findings proved that human decision-making is "bound