Radical Embodied Inferentialism and Objectivity in Language

我的研究关注的是，推理主义和激进的认知科学(RECs)的综合是否有助于解决更高层次的认知问题。安东尼·切梅罗将RECs视为生态心理学和动力系统理论(DST)的综合。生态心理学用对环境中的负担和行动的机会的直接感知来解释感知和行动。DST用微分方程组对系统进行建模，描述了它们在连续时间内的演化，并可以描述有机体与其环境的耦合，形成一个不可分解的系统。这种方法与那些认知本质上涉及对符号表征的操作的方法形成了鲜明对比。RECS已经成功地模拟了简单的感觉运动过程。然而，一个悬而未决的问题是，它如何解释似乎需要表征的“更高水平”的认知。现有的回应强调脚手架，复杂的概念认知通过我们进入丰富结构的社会文化环境而得到支撑。特别是，自然语言和科学模型是两种外部认知工具，它们调节着我们与世界的集体互动。因此，区域经济共同体需要对这些看似具有代表性的活动进行强有力的非代表性的描述。我的目标是通过发展一种REC和推理主义的综合体来提供这样的解释，它解释了语言符号的含义，根据可以正确地对它们做出的推论，强调语言的规范性、推理实践，以及罗伯特·布兰登所说的推断或行动的权利和承诺。推理理论已经被毛里西奥·苏亚雷斯等人改编成科学模型，对他们来说，模型的目的是允许他们对被建模的系统进行推断。尽管都强调实践而不是表征，但并不存在完全表达的RECs-推理主义合成。项目的核心将是阐明这些方法之间的概念关系。例如，公共语言的规范性推理结构，以及代理人的权利和承诺，可以用负担能力来表示。该框架既适用于语言，也适用于模型，为此将寻求一个统一的推理者解释。在此基础上，可以明确地解决更高层次的认知问题。现有文献将得到借鉴和发展，包括：区域经济共同体关于科学模型的工作；区域经济共同体邻近的语言方法，如分布式语言，根据这种方法，语言是一个具体的、交互的过程，而不是抽象符号的给定结构；关于语言规则和规范性的存在和性质的工作；以及对更高层次认知的解释，强调人类可获得的负担的复杂社会文化图景。在我的概念性调查中，有一种隐含的张力需要解决：分布式语言优先考虑主体驱动的动态，但对其他人来说，社会文化结构是首要的。我的项目将利用这一划分的两个方面来整合这两个层面，有助于理解它们在核算语言和概念活动方面的相对作用和它们之间的关系。现有的关于推理者和经验主义(个人和集体)代理概念的工作将被用来检验代理人如何与语言负担相互作用并动态地(重建)构建语言负担的图景。我的公众参与活动将主要涉及探索我的研究对机器学习/人工智能研究和实践的影响。例如，它认为人工智能体对语言的理解将取决于具体化(例如，通过机器人或模拟的结构化环境)、多智能体协调和语言社会规范。

My research is concerned with whether a synthesis of inferentialism and radical embodied cognitive science (RECS) can contribute to addressing the higher-level cognition problem. Anthony Chemero articulates RECS as a synthesis of ecological psychology and dynamical systems theory (DST). Ecological psychology explains perception and action together in terms of direct perception of affordances, opportunities for action in the environment. DST models systems with differential equations, characterising their evolution in continuous time, and can describe the coupling of an organism with its environment, forming one nondecomposable system. This approach is in contrast to those for which cognition essentially involves operations over symbolic representations. RECS has successfully modelled simple sensorimotor processes. However, an open problem is how it can explain 'higher-level' cognition, which seemingly requires representations. Existing responses emphasise scaffolding, whereby complex conceptual cognition is undergirded by our access to a richly structured social-cultural environment. In particular, natural language and scientific models are two external cognitive tools which modulate our collective interaction with the world. RECS therefore requires a robust nonrepresentational account of these seemingly representational activities. I aim to provide such an account by developing a synthesis of RECS and inferentialism, which explains the meaning of linguistic signs in terms of the inferences that can correctly be made to and from them, emphasising linguistic normativity, reasoning practises, and what Robert Brandom calls entitlements and commitments to infer or act. Inferentialism has been adapted for scientific models by Mauricio Suárez and others, for whom the purpose of models is their allowing inferences about modelled systems. Despite shared emphasis on practise over representation, no fully articulated RECS-inferentialist synthesis exists. The project core will be articulating the conceptual relations between these approaches. For example, the normative inferential structure of public language, and the entitlements and commitments of agents, might be rendered in terms of affordances. The framework applies to both language and models, for which a unified inferentialist account will be sought. On this basis, the higher-level cognition problem can be explicitly addressed. Existing literature will be drawn on and developed, including: RECS work on scientific models; RECS-adjacent approaches to language, such as distributed language, according to which language is an embodied, interactive process, not a given structure of abstract symbols; work about the existence and nature of linguistic rules and normativity; and explanations of higher-level cognition which emphasises the complex social-cultural landscape of affordances available to humans. There is an implicit tension to be addressed in my conceptual investigations: distributed language prioritises agent-driven dynamics, but for others, social-cultural structure is primary. My project will draw on both sides of this divide to integrate both levels, contributing to understanding the relative role of and relation between them in accounting for linguistic, conceptual activities. Existing work on both inferentialist and embodied conceptions of (individual and collective) agency will be employed to examine how agents interact with and dynamically (re)construct the landscape of linguistic affordances. My public engagement activities will primarily involve exploring the implications of my research for machine learning/AI research and practice. For example, it suggests that artificial agents' understanding of language will lie in embodiment (through, for example, robotics, or simulated structured environments), multi-agent coordination, and linguistic social norms.