Hypermodal RealismPhenomenal Information and Excluded Possibilities

David Lewis formulated the Hypothesis of Phenomenal Information in 1988: after one has been given all the physical information about an experience, there could be additional information, called phenomenal information. Since information has to exclude possibilities, this phenomenal information would have to exclude possibilities that are not physical. The idea was to force the non-materialist either to accept that there are non-physical possibilities or, by rejecting them, to trivialize their own view.

Since then, however, a limited version of the Hypothesis has gained ground. Unlike Lewis's original framing, this view is compatible with a broadening of physicalism to something beyond the local state: Phenomenal Information Externalism (PIE) claims that systems identical in what they locally do can differ in phenomenal information, because they can differ in what they exclude. But then, excluded by what?

The question can be approached from two sides: from the actual, asking which wider facts supply the exclusions; or from the possible, asking what the excluded alternatives themselves are.

The first is the actual-context problem: is the local state underdetermined because history, environment, or extended embodiment helps determine the mental fact? On this approach, the wider actual facts are what supply the exclusions: history and environment settle which alternatives a state rules out.

The second is the possibility-context problem: what specifies the alternatives in light of which an actual state is phenomenally informative? The candidates on the table are teleo-functions, wide counterfactual roles, quantum potentiae, and intrinsic cause-effect structure. As we will see, the most interesting of these give the exclusions full ontological weight.

The two problems are usually treated separately by different views. The claim here is that wide actualism and realism about possibilities each treat a symptom of the same underlying tension in PIE.

Possibility and actuality, on a minimal reading, are different modes. Possibility is inclusive while the actual is exclusive. For instance, a point in a visual field could be green and it could be red; but when I see it, it has to be one or the other. The contrast is a deep difference in kind, since the point actually becomes green only as the possible alternatives to green are excluded. What the modal (or hypermodal) logic could be beyond this minimal reading (e.g. graded rather than all-or-nothing exclusion, one dimension or many, simple or intricate in shape) will vary across different theories of the nature and process of exclusion.

Since the exclusion is limited, we need a term narrower than possibility in general, so that it picks out the relevant, living possibility. If I only see red and you only see blue, there is no pressure for one to push out the other; but a patch of red in my visual field excludes blue right there. These living possibilities are not free-floating - they have a limited boundary. In addition, these are not just any possibilities, but specific ones. For instance, if I am color blind, I do not exclude green the way you do. We can call these relevant, living possibilities real potentials. If real potentials are real, there is a question of how to cash them out, for instance as capacities of substances, quantum states, or something not located in spacetime. For the actualist, the challenge would be to derive all such potentials from a purely non-modal actual basis.

The natural first instinct might be to reject real potentials altogether. After all, PIE only says that the local state does not fully determine what an experience is - so why not simply include a broader actual context? We might add the causal history, the environment, the extended body, or the functions that history has selected. Wide actualism of this kind had the upper hand in late twentieth-century philosophy: it fit the project of naturalized semantics, which was broadly externalist, and it required no exotic ontology beyond ordinary physicalism. Or so it seemed.

As the externalist naturalization project matured, its central notions had to incorporate modal language. Semantic information was invoked as the exclusion of “relevant alternatives,” where the relevance was functional, environmental, biological, etc. For instance, a teleofunction is individuated by what a state is supposed to do, most of which never actually happens. The most actualist reading - call it hyperactualism - grounds the function exclusively in what ancestral tokens actually did. But even then, history matters to the present experience only through helping determine the modal profile of the current state, i.e. its dispositional power or normative status.

In general, the actual on a wide actualist account usually has two quite different jobs to do: 1) determining what the relevant alternatives are (the modal profile), and 2) determining which one is actual. So if one really wants to avoid realism about possibilities, it is not as easy as simply appealing to wide actualism. One has to either find actualist theories that do not make essential use of modal language in this way, or accept this modal language and deflate it to dispositions or functions solely of the actual (see Section 6).

Suppose we accept PIE: what an experience is depends on more than the local actual state. And suppose, following the last section, that the something more consists not just in extra hidden actuality, but essentially also in the activated modal profile. Then it seems we only have two options: We either deflate these activated modalities (Section 6), or we have to accept some form of real potential associated with the actual state.

Hypermodal Realism takes the second option: the real potentials individuate the actual state, and the actual state in turn constrains the potentials. The actual and the possible reciprocally determine each other, hence hypermodal.

Hypermodal Realism sits between, on the one hand, the hyperactualist view, which deflates all possibilities and potentials to a non-modal actual base; and, on the other, the modal realist or any kind of hyperpossibilist (from Lewisian worlds to Everettian branches), who regards the actual as merely a position within a space of possibilities.

Hypermodal Realism is neutral on where the real potentials live, whether local to the state only or also in the wider context. It is also neutral on the mechanism: how the reciprocal determination is understood or cashed out. Where Hypermodal Realism is firm is in the stipulation that 1) the modes are distinct - more actuality never adds up to possibility (Section 3), 2) there is a real ontological reciprocity, and 3) it is non-deflationary. This last point is what we turn to next.

There are two main ways to deflate the thesis:

First, the semantic or epistemic reading deflates the real potentials as not being objective, but instead merely descriptions, functions, measurement frameworks, or interpretive models. However, if these exclusions do not really happen objectively, but are a result of the way a third party interprets the situation, how can they be real enough to inform our phenomenal experience?

Second, the dispositional or functionalist reading regards the potentials as real, but in a way that is wholly derivative of the actual. The actual determines the possible, while the possible determines the actual only trivially, if at all. But how can such a trivial determination carry real phenomenal information, if these potentials are merely a shadow of the actual state?

These positions make sense in many situations, but realism about phenomenal information puts pressure on both of them. The reason is that the exclusions must be real enough, and precise enough, to individuate phenomenal information. Each deflation gives too little: the first fails to locate the exclusions intrinsically in the experience; the second, by deriving the potentials from the actual, gives us no nontrivial way to use them in turn to individuate the actual.

The criterion for whether a hypermodal realist account is deflationary or not is not just up to philosophical debate. Because the non-deflationary account treats the potentials associated with an actual state as real, any given hypermodal theory will have definite empirical consequences, though not necessarily falsifiable ones. The test is simple in form: Change the real potentials while keeping the actual state constant, and see if the phenomenal information changes.

Some wide-actualist phenomenal externalists famously bit the bullet: According to Dretske, Swampman, a molecule-for-molecule duplicate without the history, would be a zombie. On this theory, just by being conscious, one could know one did not arise this second. Whether one accepts that verdict or refuses it, the consequence is empirical. On the theory, the same local state with a different history-involving space of possibilities really does imply that the experience differs.

On the possibility-context side, Integrated Information Theory (IIT) supplies an operational case: compare a neuron that could fire but stays silent with one clamped so that it cannot fire. The actual activity is the same; on a canonical interpretation, IIT predicts the experience can differ. A silent neuron contributes ontologically to what the experience is.

Arguably, the silent-neuron case might be falsifiable, even if it is difficult with current technology. The difficulty might also lie in the fact that the “clamping” itself alters actual local states, so unless the explanatory asymmetry in favor of the real potentials is strong enough, the test will not settle the matter: the actualist can always attribute the difference to the changed actuality.

Since these theories stake empirical claims, some arguably falsifiable, Hypermodal Realism works in this respect more as a classification of the form some theory of consciousness might take: to rule out the form on a priori grounds would be to rule out the theories with it. However, if these theories are not falsifiable, the actualist can more easily argue that they are just philosophy with quasi-empirical content, which can be dismissed on conceptual grounds.

The simplest way to ensure falsifiability is to make Hypermodal Realism not just about reciprocal determination, but about reciprocal influence. If the real potentials individuate the actual without influencing it, nothing surfaces in the actual record: the difference could at most be subjective, or a matter of explanatory elegance. A theory may accept this and remain on the unfalsifiable side. If the potentials influence the actual, however, it becomes a lot easier to falsify.

Hypermodal Power Realism is the stronger reading of hypermodality, the idea that real potentials not only individuate the actual but influence it. We can read this as simply Hypermodal Realism with the view that phenomenal information is not epiphenomenal (that it has “phenomenal powers,” as Hedda Mørch puts it).

The power reading makes third-person scientific tests possible, because influences leave physical traces, not just subjective experiences. The difficulty of this view is of course that we need real potentials to influence actual local states in some non-trivial way. Whether this power view can be cashed out via extensions of IIT, or if a quantum theory with live potentiae is needed, or in some other way, is still uncertain. It may raise deep questions about the completeness of quantum theory, conservation laws, computational functionalism, and the closure of physics.

Even granting Hypermodal Power Realism, the ability to falsify theories postulating such influences will likely depend on our capacity for hypermodal engineering: varying what a system could do while only marginally changing what it actually does. The goal is to produce effects due to the modal configuration rather than to the actual details of the intervention.

To build artificially conscious systems, hypermodal engineering would be essential on this view, and it is also where the question of deflation becomes most acute: For a given machine, is it really conscious, or does it simply behave and function as if it were? If a machine's exclusions live only in our description of its computation, the semantic deflation returns: the phenomenal information belongs to the interpreter, not the machine. If its exclusions are merely dispositions or functions of its actual states, the dispositional deflation returns. The live question for machine consciousness is therefore neither behavioral nor architectural in the usual sense. It is whether the system really excludes real potentials when it actualizes particular states, and whether its actual states and these potentials determine each other in the way phenomenal information requires.

Put plainly: can what a system could have done, but did not, help determine what it experiences? If it can, then creating artificial consciousness just is, in part, a problem of hypermodal engineering. Moreover, to test such theories we might need “hypermodal interfaces,” connecting conscious beings across shared spaces of possibility: to one another, to modally rich consciousness-extension devices, and to conscious AIs.

This synopsis is a compressed summary of work in progress. The full article contains more details.

1. Lewis, D. (1990) What Experience Teaches in W. Lycan (ed.), Mind and Cognition, Blackwell; first published 1988, Proceedings of the Russellian Society
2. Davidson, D. (1987) Knowing One's Own Mind Proceedings and Addresses of the American Philosophical Association 60(3)
3. Boghossian, P. (1989) Content and Self-Knowledge Philosophical Topics 17(1)
4. McKinsey, M. (1991) Anti-Individualism and Privileged Access Analysis 51(1)
5. Dretske, F. (1995) Naturalizing the Mind MIT Press
6. Dretske, F. (1996) Phenomenal Externalism, or If Meanings Ain't in the Head, Where Are Qualia? Philosophical Issues 7
7. O'Regan, J. K. & Noë, A. (2001) A Sensorimotor Account of Vision and Visual Consciousness Behavioral and Brain Sciences 24(5)
8. Whitehead, A. N. (1929) Process and Reality: An Essay in Cosmology Macmillan (Gifford Lectures 1927–28)
9. Penrose, R. (1994) Shadows of the Mind: A Search for the Missing Science of Consciousness Oxford University Press
10. Balduzzi, D. & Tononi, G. (2009) Qualia: The Geometry of Integrated Information PLOS Computational Biology 5(8)
11. Kremnizer, K. & Ranchin, A. (2015) Integrated Information-Induced Quantum Collapse Foundations of Physics 45(8)
12. Tononi, G. & Koch, C. (2015) Consciousness: Here, There and Everywhere? Philosophical Transactions of the Royal Society B 370(1668)
13. Mørch, H. H. (2018) Does Dispositionalism Entail Panpsychism? Topoi 39(5)
14. Kastner, R., Kauffman, S. & Epperson, M. (2018) Taking Heisenberg's Potentia Seriously International Journal of Quantum Foundations 4(2)
15. Chalmers, D. & McQueen, K. (2021) Consciousness and the Collapse of the Wave Function arXiv 2105.02314
16. Albantakis, L., … & Tononi, G. (2023) Integrated Information Theory (IIT) 4.0: Formulating the Properties of Phenomenal Existence in Physical Terms PLOS Computational Biology 19(10)
17. Seth, A. (2025) Conscious Artificial Intelligence and Biological Naturalism Behavioral and Brain Sciences