Consciousness Across Three Worldviews

In late August, the Berggruen Institute’s Future Humans program hosted a global gathering of top thinkers on consciousness at Casa dei Tre Oci in Venice. In this essay, three of those thinkers have sought to synthesize the correspondence of concepts from within three widely divergent perspectives.

This note is an observation on the possibility of a partial — but perhaps illuminating — translation between vastly different worldviews and languages. The three authors are, respectively, the head of the Hindu Vedānta Society of New York, a vice president at Google working on fundamental research in AI and Artificial Life, and a theoretical physicist interested in the foundations of quantum mechanics. From these very different cultural perspectives, we discuss certain similarities and resonances between some central notions in these respective fields. We are, of course, well aware of the profound differences, but we are less interested in these: We are more interested in conceptual cross-fertilization.

In the Advaita Vedānta school, consciousness, in the sense of the individual experiencing self, is recognized as a manifestation of pure awareness, the highest form of reality, described as self-luminous (that is, self-revealing) and universal. This is a non-dualist worldview where reality, knowledge and awareness are not held to be distinct. There is no duality between the experiencing self and Brahman, the Ground of Being, which is pure awareness.

By contrast, among computational neuroscientists and computer scientists, consciousness is typically characterized (by those willing to speculate about it at all) in terms of information processing in the brain. We have evolved to build sophisticated models of the world, of ourselves and of others in order to regulate our internal state and guide our behaviors in a complex world. Consciousness then is simply “what it is like” to have such a self-model. This view of consciousness applies only to large-brained animals (perhaps to sophisticated-enough AIs), which comprise specialized, recently evolved phenomena on Earth.

In quantum theory, we describe all physical systems, which in a physicalist worldview means any aspect of nature, in terms of how they affect other physical systems, for instance (but not only) measuring apparatuses. In the relational interpretation of quantum mechanics, physical systems can have relative, or mutual, information about one another. For two systems to have mutual information, in the sense defined by Claude Shannon, the creator of information theory, means that we can learn something about one of them by observing the other. The information that we ourselves have about the world can be viewed as a special case of this very general notion. Reality — meaning our entire worldview, including our understanding of the world and of ourselves — can be understood as the ensemble of relevant mutual information among all physical systems, including “ourselves” (however defined).

Let us now consider in some detail the meaning of the noun “light.” In everyday language, we use it in various related ways. In its most restricted sense, light comprises electromagnetic waves (or, in the quantum limit, photons) within a spectral band corresponding to wavelengths of roughly 380–750 nanometers. We reserve the term “light” for this narrow window of the electromagnetic spectrum because these are the wavelengths we see; this window provides most of us with our primary source of information about the external world.

Our use of the term “light”, in fact, gives greater priority to this informational role, than to the physical fact of the information being mediated by photons. Hence, radio waves and X-rays are not generally referred to as “light,” while a book or an argument can “shed light” on what was formerly “obscure,” that is, unilluminated. Relatedly, in a conversation, it is common for one to say “I see” to acknowledge understanding — even if the conversation is conducted by telephone.

In the Vedānta tradition, the term “light” is used analogously, as illustrated in the Bṛhadāraṇyaka Upaniṣad, where, according to the sage Yājnãvalkya:

“When the sun has set, the moon becomes one’s light. When the moon has set, fire becomes one’s light. When the fire is out, speech (sound) becomes one’s light. When even speech has ceased, the Self (Ātman) indeed is one’s light.” (Bṛhadāraṇyaka Upaniṣad, 4.3.2–7)

Here “light” refers to the informational or “enlightening” role of an interaction (even, when cut off from the rest of the world, with oneself), not to the specific nature of electromagnetic waves of certain wavelengths. Used in this general sense, the notion of “light” has a central role in the Hindu worldview. In the Bhagavad Gītā, Krishna, the beloved major deity in Hinduism, says:

“That is the Light even of the lights; It is spoken of as beyond darkness. It is Knowledge, the Knowable, and the Known. It exists specially in the hearts of all.” (Bhagavad Gītā 13.18)

Using “light” in this general sense, we can equally say that in its relational interpretation, modern physics is all about the light or enlightenment (that is, the information) one system can have about another, including, as a special case, the complete knowledge we can have about the world.

Thus, all-encompassing awareness, or undifferentiated self-luminous consciousness, appears to play a role in the conceptual system of the Vedānta tradition that corresponds to the overall field of relative information in the relational interpretation of modern physics.

There is a point of apparent divergence between the relational physics and Vedānta perspectives, however. The pure awareness and self-luminously conscious nature of Vedānta “information” implies a vibrant saliency, or phenomenal character, that seems hard to reconcile with the cold, quantitative nature of Shannon’s mutual information. This is where evolutionary and computational perspectives, which we are now going to discuss, can offer additional insight, and perhaps a conceptual bridge.

In the Vedānta tradition, the individual, experiencing self, is built by constructing the (fundamentally illusory) distinction between a subject and a perceived object. In an analogous manner, evolution can be understood as a dynamical process that gives rise to living subjects or “selves”: processes capable of persisting and replicating. This involves constructing a distinction (albeit a semipermeable one) between “self” and “world.”

Such a self-world interface is both physical and conceptual. Physically, it is a literal boundary — in its most primordial form, a cell membrane. Conceptually, the interface is not only about what cannot pass through, but about what can. It defines an umwelt, or perspective-dependent universe of salient information about the outside world.

Although biologist Jakob von Uexküll coined the term umwelt to describe only what an organism cares about outside itself, we need to consider that every organism also has an internal umwelt, including behaviorally relevant “inner” senses such as hunger, pain and satiety.

Gregory Bateson, an influential thinker in systems theory and psychology, described a bit of information as “a difference that makes a difference,” emphasizing that salient information is not just a statistical distinction, but one that has causal or functional consequences for some system. We can call this semantic information, or meaningful information, as distinct from Shannon’s purely combinatorial definition of information. The DNA sequence of a live bacterium and that of a dead one carry the same amount of Shannon information if the sequences are identical; the difference lies in the context, since only in the live cell can that sequence have causal effects on the bacterium’s future — and be passed on to subsequent generations.

As shown by Turing and von Neumann in their foundational work in informatics, any process involving such causal manipulation of information can be understood abstractly as computation, or the running of code. By the same token, only data, that is, information that is processed by such a computation, can be behaviorally salient. A far-infrared photon and a visible photon may both carry Shannon information, but from the perspective of a human observer, only the visible photon carries semantic information — meaning it can be perceived and acted upon and thus “make a difference.” Semantic information, in other words, is something’s — or someone’s — umwelt.

Hence, everything a living subject experiences, whether internally or externally, is, relative to it, semantic information, or in the Vedānta sense, light which “exists specially in the hearts of all,” per the words of Krishna quoted above. Further, everything a living subject persistently is (that is, which is preserved by its homeostatic processes and/or replicated during growth or reproduction) can also be described as semantic information, hence “light.”

In this surprisingly literal sense, we are indeed all made of light, as is everything we perceive or understand.

Yet this same formulation also allows us to understand the perspective-dependence, or, to put it another way, the illusoriness of the self/other distinction. Under different perspectives (or, in physical language, coarse grainings) one can consider an individual immune cell as a living entity, or the bee in which this cell lives as an entity, or the entire hive as a living entity. In each case, the umwelt will differ, as will the information preserved by the entity through time, otherwise known as its “identity” — as the differing lifespans of these nested processes make clear.

Thus, we can articulate a unified perspective in which central concepts in three different domains — Hindu tradition, computer science and quantum physics — find analogies and reflect one another.

We do not want to overplay these similarities and relations; the differences among the traditions we have referred to are wide. But we believe that the world admits diverse kinds of accounts, and that rather than being taken as conflicting worldviews, these can often be understood as incomplete, complementary and perhaps even compatible perspectives on reality.

In any case, it is our hope that highlighting convergences, searching for approximate translations and finding structural analogies might lead to more useful cross-fertilization than debating differences between a priori assumptions.