Sense-Making

In The You On AI Encyclopedia

The concept distinguishes Thompson's enactive framework from functionalist and representationalist theories of cognition. Functionalism holds that cognition is whatever performs the appropriate causal role, regardless of substrate. Representationalism holds that cognition consists in the manipulation of internal symbols that stand for features of the world. Sense-making refuses both. Cognition is not a functional role that can be filled by any system with the right causal structure; it is the specific activity of a living organism whose engagement with its environment is oriented by its own needs. Cognition does not manipulate pre-given representations; it enacts the significance of environmental features through activity.

The concept has immediate application to the collaborations described in You On AI. When Edo Segal describes working with Claude late at night, the enactive analysis reveals the meaning of the exchange as enacted entirely by Segal. Claude generates sequences of tokens that are statistically probable given the input. Segal enacts a world in which those sequences mean something — in which the book matters, in which getting the argument right matters, in which the collaboration serves a project that is embedded in his life, his concerns, his embodied history. The meaning is not in the tokens; it is in the living mind that receives them and finds in them a connection to what it cares about.

Sense-making is graded, not binary. The bacterium's sense-making is minimal — a binary evaluation of sugar or not-sugar — but it is genuine. Human sense-making is extraordinarily rich, shaped by language, culture, emotional history, and intersubjective engagement with other minds. Each level of sense-making is continuous with the levels beneath it; the human's capacity to recognize a friend's face is continuous with the bacterium's capacity to recognize a nutrient, and both are continuous with the autopoietic self-recognition through which the organism maintains its boundary against the environment.

The practical consequence for AI-augmented work is the diagnostic that Thompson's framework provides for fluent fabrication. When an AI system produces a passage that sounds insightful but breaks under examination, the failure is not a bug to be fixed. It is a structural consequence of a system that generates outputs without sense-making. The system has no way to distinguish between a connection that illuminates and a connection that merely sounds as though it does, because distinguishing requires a being that has stakes in the quality of understanding.

Origin

Sense-making was introduced as a technical term in Thompson's Mind in Life (2007), drawing on Varela's earlier work on autopoiesis and Weber's extension of the concept into a theory of biological value (2002). The concept has since been developed by a school of enactive cognitive scientists including Ezequiel Di Paolo, Hanne De Jaegher, and Shaun Gallagher.

Key Ideas

Significance is relational. It lives in the organism-environment relationship, not in the organism alone or the environment alone.

Stakes generate significance. A system has to be something that can win or lose before its environment can carry meaning.

Computation processes; organisms enact. Processing operates on representations; enacting creates the significance that representations presuppose.

Humans supply the sense-making in AI collaborations. The tool generates; the living mind evaluates; the evaluation is where the meaning lives.