Mechanistic and representational explanations in cognitive neuroscience

认知神经科学家通过描述现象背后的神经机制来解释认知现象，如感知、记忆或解决问题。在这样做的过程中，他们通常假设这些机制的某些组件具有表示属性。例如，视觉皮层中的神经元被认为代表某些刺激特征，这解释了生物体如何能够感知世界并与之互动。然而，机械解释和表征解释的结合产生了一种张力：神经科学的机械解释可以初步地只涉及大脑内的因素。然而，表征性属性取决于生物体与其外部世界和/或过去的关系。这就提出了我们所谓的兼容性挑战：认知神经科学的解释能否同时具有机械性和表征性？兼容性挑战还没有在哲学上得到充分的检验，尽管它与20世纪80年代心灵哲学中熟悉的一个问题——经典挑战——有关。该项目可以被理解为根据科学哲学、认知哲学和认知神经科学的最新发展，对经典挑战进行必要的、早该进行的修订和重新评估。我们将通过与实证研究人员密切合作，并采用一种称为“对抗性合作”的新方法来检验两组工作假设，从而解决兼容性挑战。第一个是：A.认知神经科学可以不需要表征性的解释，而仅仅依靠机械的解释。B.计算解释在认知神经科学中的突出地位解释了为什么科学家仍然使用表征词汇，同时表明计算机制解释是足够的。第二组工作假设是：C.计算解释为改进表征内容和解释提供了第一步。D.根据计算交通工具的功能信息特性，有可能发展一种表征的说明。单凭广泛的解释还不能使表征性解释与机械性解释相容。F.机械框架可以扩展，以便它允许计算车辆的功能信息属性在机械解释中得到体现。该项目将为表征在心理现象的机械解释中所扮演的角色提供新的见解。通过其方法，它显然是开放式的。这些结果将有助于对我们心灵的普遍理解和对它的科学解释，并将显著地有助于表征主义者和反表征主义者之间辩论的根本重新定位。

Cognitive neuroscientists explain cognitive phenomena such as perception, memory, or problem solving by describing the neural mechanisms underlying the phenomena. In doing so, they usually assume that some components of these mechanisms have representational properties. For example, neurons in the visual cortex are thought to represent certain stimulus features, which explains how the organism is able to perceive and interact with the world. However, the combination of mechanistic and representational explanations yields a tension: neuroscientific mechanistic explanations can prima facie refer exclusively to factors within the brain. Representational properties, though, supervene on the organism’s relations to its external world and/or past. This raises, what we dub, the compatibility challenge: can explanations in cognitive neuroscience be simultaneously mechanistic and representational? The compatibility challenge has not been sufficiently examined philosophically, though it is related to a problem familiar from philosophy of mind in the 1980sthe classical challenge. The project can be understood as the necessary and long overdue revision and reassessment of the classical challenge in light of recent developments in philosophy of science, philosophy of cognition, and cognitive neuroscience. We will approach the compatibility challenge by working in close collaboration with empirical researchers and by applying a novel method called “adversarial collaboration” to examine two sets of working hypotheses. The first one is: A. Cognitive neuroscience can do without representational explanations and rely solely on mechanistic explanation. B. The prominence of computational explanation in cognitive neuroscience explains why scientists still use representational vocabulary while at the same time showing that computational-mechanistic explanations are sufficient. The second set of working hypotheses is: C. Computational explanation provides the first step towards an improved account of representational content and -explanation. D. It is possible to develop an account of representations in terms of function-informational properties of computational vehicles. E. Wide explananda alone do not yet render representational explanation compatible with mechanistic explanation. F. The mechanistic framework can be extended so that it allows for function-informational properties of computational vehicles to figure in mechanistic explanation. The project will provide new insights into the role representations can play in mechanistic explanations of mental phenomena. Through its methodology, it is explicitly open-ended. The results will contribute to the general understanding of our mind and the scientific explanation of it, and will significantly contribute to a fundamental reorientation of the debate between representationalists and anti-representationalists.