{
  "generated": "2026-09-10T20:32:51Z",
  "papers": [
    {
      "slug": "evaluating-iit",
      "title": "IIT and the problem of phenomenal unity: from conscious states to conscious systems",
      "versions": [
        {
          "slug": "evaluating-iit",
          "version": 2,
          "title": "IIT and the problem of phenomenal unity: from conscious states to conscious systems",
          "author": "Felix Noble",
          "tldr": "Integrated Information Theory 4.0 may be able to explain how distinct conscious\ncontents occur together as one unified experience at the level of states - but the\ntheory first has to pick the right physical system to analyse, and currently has no\nindependent justification for that choice. The paper argues this \"boundary problem\"\nis narrower than denying that IIT's phi tracks consciousness at all.",
          "abstract": "Integrated Information Theory (IIT) 4.0 has two different resources for explaining\nconsciousness. Its system-level measures are meant to identify the physical system\nthat exists as a conscious entity. Its state-level unfolding describes the distinctions and\nrelations that correspond to the content of an experience. This paper asks whether\nthose resources can explain phenomenal unity, the fact that distinct contents can occur\ntogether as parts of one conscious state. I argue that IIT's state-level formalism may\nbe well suited to this task. If Sophie sees a red tomato while hearing a bird, IIT\ncan in principle represent the visual and auditory contents as distinct mechanisms\nand represent their unity through a higher-order, irreducible relation. The difficulty\ncomes earlier. IIT applies that analysis only after selecting a physical system, normally\nthe system with maximal system integrated information, or system-level $\\phi$, among\noverlapping candidates. There is no independent reason to think that this system\ncontains the causal relationships responsible for the unity of a particular phenomenal\nstate. A smaller system may omit relevant causes or effects. A larger system may\ninclude those relationships while also measuring causal structure unrelated to the unity\nat issue. The result is a boundary problem. IIT may have a promising account of\nphenomenal structure without yet having a justified account of the physical system\nover which that structure should be calculated. This is a narrower claim than the claim\nthat $\\phi$ is not a correlate of consciousness. System-level $\\phi$ may track global conscious\nstates while remaining insufficient to identify the structure of a particular experience.",
          "keywords": [
            "Integrated Information Theory",
            "IIT 4.0",
            "phenomenal unity",
            "phenomenal binding",
            "system selection",
            "consciousness"
          ],
          "date": "2026-08-19",
          "url": "/papers/evaluating-iit/v2/"
        }
      ],
      "latest": "/papers/evaluating-iit/v2/"
    }
  ]
}