In ActInfLab, we are directing our attention to phenomena across multiple scales, and trying to implement changes across different scales. In Active Inference and Systems Engineering, this is equivalent to hierarchical or nested modeling.

The three scales that ActInfLab models are Personal scale (Agent is Participant), Lab scale (Agent is ActInfLab), and Community scale (Agent is ActInfLab Niche).

  • Personal scale

    • Participants are interacting subunits of the Lab generative model. From Personal perspective, Lab active states are observations.

    • For Participants, Lab provides affordances & opportunity for generative model update via niche modification.

  • Lab scale

    • This is where we can engage in Lab-level policy selection, reducing our uncertainty about realizing our expectations and preferences.

  • Community scale

    • Community is the epistemic niche of Lab, and external generative process.

    • From Community perspective, Lab active states are observations (that may modify niche & affordances).

This multi-scale approach to strategy is applied across the key aspects of ActInfLab:

  1. Education

  2. Research

  3. Outreach & Engagement

  4. Methods

  5. Open-endedness


  • ActInfLab is an educational non-profit.

    1. Personal scale. The current and future community of Active Inference learners, researchers, and practitioners is diverse in many ways (time zone, background, familiarity with Science/Philosophy/Engineering/etc). Thus our educational materials and strategy should reflect this plurality of individual priors, journeys, and preferences.

    2. Community scale. ActInfLab works on education and reducing research debt for a broad community from academia, industry, and beyond. This education work takes multiple forms, such as courses and ontology development.


    1. ActInfLab organizes, catalyzes, and participates in Active Inference research.

    2. Personal scale. We provide affordances to scaffold and launch research and development projects by participants. Some research projects, especially those with long-term impact or public goods features, will be stewarded directly by the lab. Other projects will be driven by lab participants outside of our organization, in which case the lab will provide relevant support.

    3. Community scale. We recognize that Research must be carried out alongside discussions of Responsibility and Ownership. We seek to learn, apply, and improve upon practices from open science.


    1. ActInfLab works on a global scale with a participatory model.

    2. Personal scale. ActInfLab recognizes the tremendous diversity of participant initial conditions (e.g. familiarity with language, science, philosophy, computers) and preferences. We believe in the power of participatory community-based approaches for Active Inference Education and Research, meaning that entry and advancement within ActInfLab should be accessible and legible.

    3. Community scale. We perceive a deficit in our niche regarding high-quality science communication and inclusion around cutting-edge research topics such as Active Inference.


    1. ActInfLab applies state-of-the-art methods across domains.

    2. Personal scale. We educate participants so that they grow their expertise level in different disciplines while also developing as a person. As disciplinary expertise entails competency in practices and tools, we take a holistic approach to participant professional development. We construct a niche for participants such that their personal affordances are clear, and role-based task performance is effective and aligned.

    3. Lab scale. As an online participatory lab, communication and infrastructural design are our fundamental organizational afford. We apply Active Inference & Systems Thinking to the Lab operation and approach. This approach clarifies the difference between causal engagement with the Lab, and role-based allocation of practices, cases, responsibilities, and concerns.


    1. ActInfLab is on the path of open-endedness.

    2. Personal scale. We recognize and support the open-endedness of the personal journey of learning and applying Active Inference.

    3. Lab scale. Our strategy considers changes in the niche and the effect of our actions over multiple nested timescales. We recognize the importance of short-term implementation and project-level specificity, while also thinking more broadly about what/where/how Active Inference can be in the deeper future. We draw on the insight and contributions of our Lab Advisory Board as a vital part of the Lab.

    4. Community scale. In principle, ActInfLab can participate officially in various kinds of activities beyond Education, Research, and Outreach/Engagement. However this must be assessed situationally, and opportunities such as consulting may be better fulfilled by individual-level partnerships, or future peer organizations.