Can the Integrated Information Theory Explain Consciousness from Consciousness Itself?

Abstract

In consciousness science, theories often differ not only in the account of consciousness they arrive at, but also with respect to how they understand their starting point. Some approaches begin with experimentally gathered data, whereas others begin with phenomenologically gathered data. In this paper, I analyse how the most influential phenomenology-first approach, namely the Integrated Information Theory (IIT) of consciousness, fits its phenomenologically gathered data with explanatory hypotheses. First, I show that experimentally driven approaches hit an explanatory roadblock, since we cannot tell, at the present stage, which model of consciousness is best. Then, I show that IIT’s phenomenology-first approach implies a self-evidencing explanation according to which consciousness can be explained by starting from consciousness itself. I claim that IIT can take advantage of the virtuous circularity of this reasoning, but it also faces a data-fitting issue that is somehow similar to that faced by experiment-driven approaches: we are not given enough information to decide whether the explanatory hypotheses IIT employs to explain its phenomenological data are in fact best. I call this problem “the self-evidencing problem” for IIT, and after introducing it, I propose a possible way for IIT to solve it.

Decoding Consciousness in Artificial Intelligence

“…• Attention Schema Theory (Wilterson et al, 2020), and • Recurrent Processing Theory (RPT) (Negro, 2023), which are compatible with their computational functionalism framework.…”

“…To correctly place objectivity, Ellia et al (2021) argue that the intrinsic, phenomenal structure of an experience should have a physical explanation and take a cause-effect structure as the physical explanation. However, cause-effect structure does not consider all possible confounders and there is a lack of methods to distinguish causation with association (Negro, 2023). The essential problem is the current theories of consciousness are systematically underdetermined.…”

The Fundamental Tension in Integrated Information Theory 4.0’s Realist Idealism

“…Although previous versions of IIT did not explicitly clarify the logic of this translation (for discussions on this point, see [16][17][18][19]), IIT 4.0 explains that the postulates are inferred via an "inference to a good explanation" from the axioms and a triad of basic assumptions [11,12]. These assumptions are realism, operational physicalism, and atomism.…”