Christian Lains

David Lewis formulated the Hypothesis of Phenomenal Information in 1988: phenomenal experience can exclude possibilities that physical information leaves open. Lewis’ point was to reject this non-physicalist notion. Yet a limited version has gained ground recently. Phenomenal Information Externalism (PIE) claims that systems identical in what they actually do can differ in phenomenal information, because they differ in what they exclude. But excluded by what?

Consider three views: First, on an epistemic reading, exclusions are relative to how states are modeled, not fixed by the states themselves. Second, for the modalist, what is excluded are ontologically real possibilities. Third, for the actualist, the exclusions are wholly derivative of the actual state’s structure, functions, or dispositions. We can set aside the epistemic reading as openly deflationary, but a similar worry confronts the actualist: If the possibilities excluded are solely derived from the actual, how can they turn around to nontrivially inform it?

They cannot, I argue: on an actualist reading, the possibilities cannot individuate in the ontic sense required for experience to be informative. A realist PIE therefore requires Hypermodal Realism: the excluded possibilities must nontrivially help individuate the actual experience itself. If in addition phenomenal information has “phenomenal powers” (Mørch 2017), we get Hypermodal Power Realism: excluded possibilities not only individuate the actual but influence it.

Theories of consciousness as different as Integrated Information Theory (IIT), quantum approaches, process philosophy, enactive views, Mendola’s modal structuralism, and Dretske’s teleofunctional externalism can be read as broadly hypermodal. All look beyond local occurrent activity for the exclusions: on teleofunctional accounts, causal history; on enactive accounts, sensorimotor contingencies; on quantum accounts, real potentiae underlying local actuality; on Mendola’s view, it is “entwined” with possibility in a modal structure; and on IIT, the irreducible cause-effect structure.

Any hypermodal view can be tested: hold local activity constant, vary the real possibilities, and check whether the theory predicts a phenomenal difference. In Dretske’s teleofunctional externalism, history contributes to the possibilities, so a historyless duplicate (Swampman) would not be conscious. In IIT, clamping a silent but activatable unit leaves local activity unchanged while altering the possibilities, so the theory predicts a phenomenal change (Albantakis et al. 2023).

If phenomenal information is epiphenomenal, the test is hard to carry out (Bartlett 2022): How would we know Swampman is a zombie, if it acts the same? We face a circularity objection: If we couple the actual and possible too tightly, the view may collapse into the actualist’s. Adopting Hypermodal Power Realism forces the contrast, but requires frameworks in which possible and actual influence each other, perhaps favoring quantum or process theories (Kastner et al. 2018).

If Hypermodal Realism is true, creating artificial consciousness requires more than replicating the brain’s computations, functions, or its actual activity. Consciousness engineering would need to design the right modal configurations, and, on the “power” reading, the right modal dynamics. Hypermodal interfaces could test the theory by connecting humans across shared spaces of possibility; to one another, to modally rich consciousness-extension devices, and to conscious AIs.

Keywords: phenomenal information; excluded possibilities; modal metaphysics; IIT; artificial consciousness