Samerepresenting

Samerepresenting is a term used in cognitive science and artificial intelligence to describe a condition in which two or more agents hold identical or highly aligned internal representations of the same external state or concept. It emphasizes the stability and interoperability of the agents’ interpretive frames, rather than the external data alone.

Formalization of samerepresenting involves considering multiple agents observing the world through possibly different sensors. Each agent i uses a representation function r_i that maps observations to an internal representational space R. Samerepresenting occurs when there exists a consistent alignment across agents such that, for salient states s, the representations r_i(s) are approximately equal or lie within a small, predefined tolerance. Achieving and maintaining samerepresenting typically requires mechanisms to resist perceptual noise, account for differences in sensing or perspective, and sustain alignment over time through calibration, learning, or targeted communication of priors and constraints.

Examples of samerepresenting include multi-robot exploration where agents share a common map representation to coordinate path planning, distributed sensing systems where devices output comparable feature vectors, and human–robot collaboration in which a human and a robot adopt a shared task model to reduce ambiguity and improve joint performance.

Related concepts include shared mental models, representation learning, and alignment in multi-agent systems. Samerepresenting is distinct in its emphasis on convergent internal structure, not merely data exchange, and it highlights how consistent interpretive frames enable coordinated action and interoperability. Evaluation typically combines representation similarity metrics with task performance to assess alignment quality.