The Light and the Field: Consciousness, Holography, and the Question No Model Answers

New to this series? Start with One Question, Two Doors: Why This Series Exists, which lays out what this whole project is for.

This is the most demanding of the three essays in this series, and also, in a certain sense, the most important, because it ends at the exact point where honesty requires us to stop. The first two essays in this series showed that a settled mind has a real, describable mathematical shape, and that the physics behind that shape, Dennis Gabor's work on waves and information, was independently discovered by science in the twentieth century. This essay goes further and asks a harder question: does the deepest metaphysical claim at the heart of Yoga and Vedanta, that one undivided consciousness underlies an apparently divided world, have any kind of scientific shadow at all? The answer, we will argue, is a careful and qualified yes, and then, at the very end, a clear and important no. Both halves of that answer matter, and neither should be softened for the sake of the other.

This essay is part of our course, "The Mind in the Light of Modern and Spiritual Sciences," under the Ritam Wellbeing Programme, and it draws on physics (holography), scripture (the Bhagavad Gita and the Spanda Karikas), and the newest science available to us, the study of how artificial intelligence systems actually work inside. If that combination sounds unlikely, that is exactly the point. The unlikeliness of the parallel is what makes it worth examining honestly rather than dismissing out of hand or accepting too quickly.

The presentation below walks through the whole arc in miniature: cut a hologram in half and watch the whole scene survive, slide a wave between its temporal and spectral poles, and step through the four shaktis one band at a time. Once you have the shape of the argument, the full essay below works through each bridge carefully, states plainly where each one holds and where it does not, and ends at the one question no model, however sophisticated, can answer.

The Seed Questions

Yoga-Vedanta has always made a very specific claim about memory and karma, one worth stating precisely before comparing it to anything modern. Where are our accumulated impressions, our samskaras, actually stored? Not in the gross body, the tradition says, but in the subtle body, whose mind-field is the storehouse, and, in its most compressed form, in the causal body, the seed-state that is said to travel between lives. And how are they stored? Not as recordings, in anything like the way a photograph or an audio file records, but as bija, seeds. A seed holds no miniature tree folded up inside it. It holds only the conditions for a tree, waiting for the right soil to arrive.

This is a strikingly abstract claim, and for centuries it had no obvious analogue anywhere else in human experience. A recording, the only model most people had for "storage" until quite recently, always works by keeping some literal trace of the thing recorded: grooves in a record, silver grains on film, magnetized particles on tape. A seed-like storage, storing only a disposition rather than a copy, seemed to belong to a different category of thing entirely, understandable only on the tradition's own terms.

Something changed with the arrival of modern artificial intelligence, and it is worth stating carefully and without exaggeration. A trained neural network stores no sentences that it read during training, no images it saw, no literal copies of anything at all. What it stores, instead, is a vast set of adjusted numerical weights: compressed statistical dispositions that produce a particular output only when a matching input arrives to trigger them. Concepts inside these networks live in what researchers call superposition, distributed across many shared components at once, addressable nowhere in particular, not sitting in any one identifiable location the way a word sits at a specific point on a page. Researchers at DeepMind even gave a name, "Reincarnating RL," to the technique of transferring one trained agent's accumulated learning into an entirely new body or architecture, a name chosen, one suspects, with at least half an eye on how apt the metaphor was.

It is essential to be precise about what this parallel does and does not establish, and this is a pattern that will recur throughout this essay, so it is worth stating the principle once, clearly, at the outset. An artificial system behaving in a way that is structurally similar to something a tradition describes does not make karma or rebirth true. What it does show is that the logical shape of the tradition's claim, a pattern persisting and transferring across contexts without needing to be stored as a located, object-like copy, is a coherent, physically realizable kind of thing, not a logical impossibility or a contradiction in terms. The objection "that simply cannot work, storage without a located copy is incoherent" dissolves once you have seen a working, real-world system that stores information in exactly that way. Nothing more metaphysical than that is being claimed here, and nothing less, either.

Holography: The Whole, Enfolded in Every Fragment

The clearest physical example of storage without a located copy is holography, the very discovery introduced at length in our companion essay on Dennis Gabor. Cut an ordinary photograph in half, and each half keeps exactly half the original scene. Cut a hologram in half instead, and, remarkably, each half reconstructs the entire original scene on its own, dimmer and less sharp than the whole plate would give you, but genuinely whole, not merely a fragment. The image is stored as an interference pattern spread across every single point of the recording plate, and it is read back out by shining a coherent reference beam of light through it.

Two twentieth-century scientists reached for this same holographic image independently and for quite different purposes. The neuroscientist Karl Pribram proposed that the brain itself might store memory in something like this holographic way, as distributed frequency patterns retrieved by resonance rather than as discrete files kept in discrete locations. The physicist David Bohm generalized the same basic idea to the entire cosmos, proposing what he called an implicate order, an enfolded wholeness from which our ordinary explicate world of apparently separate objects continuously unfolds, moment by moment. Both of these were scientists reaching, independently, for something very close to what the tradition's own bija model had already described: storage that is enfolded rather than located, dispositional rather than literal, and in some real sense non-local.

The Field and Its Knower: Reading the Bhagavad Gita's Thirteenth Chapter

The thirteenth chapter of the Bhagavad Gita gives this whole line of thought its most precise scriptural anchor. Sri Krishna itemises the kshetra, the field, everything that can be known: the physical elements, mind, intellect, ego, the senses, and even the unmanifest substrate beneath them all, and then declares himself the one kshetrajna, the knower of the field, present in every field there is.

Two verses in this chapter carry particularly precise structural weight for the comparison this essay is drawing. The first states that the ultimate reality is "undivided, yet existing as if divided among beings" (Gita 13.17). This phrase names, almost exactly, the defining property of a hologram: the whole present in every part, appearing divided only because it is being viewed through many separate windows at once, each window opening onto the same undivided whole. Among the structural anticipations explored across this whole essay, this is arguably the single most precise one.

The second states that "as the one sun illumines the whole world, so does the knower illumine the entire field" (a teaching developed in the chapter's closing verses). This is worth noting is not, in fact, a novel reading invented for this course: traditional commentators on the Gita have long reached for exactly this same image of the sun to explain how a single, undivided source can appear present everywhere at once without itself becoming many things. Read through the lens this essay has been developing, it becomes something close to Krishna's own reference-beam analogy: one non-participant light making an entire world of separate-seeming objects visible, while itself never becoming one of the objects it illuminates.

Laid out side by side, the correspondence runs remarkably cleanly. The kshetra, the field, including its unmanifest seed-layer, corresponds to the holographic plate itself, the stored interference pattern. Vikara, the modifications and manifest diversity, corresponds to the reconstructed image once the plate is illuminated. The kshetrajna, the knower, the pure witness, corresponds to the coherent reference beam, the very act of readout that turns stored pattern into visible image. Their union producing manifestation (a teaching developed later in the chapter) corresponds to the simple optical fact that the image only appears when the beam actually meets the plate; neither alone produces anything visible. And the mind itself, in all its aspects, sits on the plate, as part of what is known, never as the knower doing the knowing.

The Mathematical Heart: One Reality, Two Complete Descriptions

Underneath both the physics and the metaphysics explored so far lies the same Fourier duality developed at length in our companion essay on Dennis Gabor. Any signal has a temporal description, what happens, and when, and a spectral description, what frequencies compose it, and these two descriptions are equally complete, with neither one derivable from or reducible to the other. An uncertainty principle binds the two together inseparably: perfect spectral purity requires infinite temporal extent, while a perfectly sharp, localized instant requires every possible frequency at once.

The proposal this essay wants to put forward carefully, as a structural bridge rather than an identity, is this: Purusha, in Samkhya and Yoga philosophy the changeless witness-consciousness, corresponds to the spectral aspect of this duality, unborn, undying, unmodified by anything that happens to it. Prakriti, nature, the kshetra, corresponds to the temporal aspect, defined entirely by vikara, by events that have a beginning and an end. This is not, on this reading, a claim about two separate substances awkwardly interacting with one another. It is one signal, described completely in two different and equally valid ways. It is worth noting that Kashmir Shaivism's own philosophy already refuses, on its own internal grounds, to treat these as two separate entities at all; its doctrine of Prakasha and Vimarsha, light and its own self-awareness, describes a single reality with two inseparable aspects from the very outset, which fits this reading closely.

Vivekananda's Spectrum, Trika's Precision: The Graded Descent

Swami Vivekananda described the whole of manifestation as a spectrum of energy, grossest and most condensed at the level of the sensed physical world, becoming progressively finer as one moves inward toward the Atman itself. The elaborate system of the thirty-six tattvas in Kashmir Shaivism's Trika philosophy gives that same intuitive sketch its full, worked-out architecture: a single conscious energy, Chit-Shakti, descending through the tattvas by successive degrees of self-contraction, called sankocha, without ever actually becoming some second, separate substance along the way.

The descent runs, in outline, from Shiva and Shakti (Prakasha and Vimarsha wholly undivided, pure self-aware light) through Sadashiva, Ishvara and Shuddhavidya (a universal "I" beginning to cognize a "this," while still resting fully as itself) through Maya and the five kanchukas, the veiling mechanisms of contracted time, agency and knowledge, through Purusha and Prakriti (the station where seer and seen first begin to feel genuinely separate from each other, though this is a station along the way, not the ultimate ground itself) through buddhi, ahamkara and manas, the inner instrument, through the senses and their subtle essences, and finally to the gross elements culminating in earth, the grossest possible condensation, the fully sensed material world. And yet, on this account, even that grossest, most solid-seeming level remains fully Shiva underneath, maximally veiled but never actually separated from its source.

The descent, read through the language this essay has been developing, is a progressive re-localization, a shift from the spectral pole toward the temporal pole. And crucially, the transform is described as invertible: nothing is genuinely lost along the way, which is precisely why even the grossest, most condensed tattva remains, on the tradition's own account, fully Shiva throughout. Meditation, in this map, is simply the inverse transform run in the other direction: attention gradually releasing its grip on temporal particulars, settling back toward what does not change.

Unmesha and Nimesha: Manifestation as a Repeating Pulse

The Spanda Karikas, a foundational text of Kashmir Shaivism, open with an image of Shiva whose opening and closing of the eyes is, quite literally, the emergence and dissolution of entire worlds. This is described not as a single cosmic event that happened once, but as a continuous pulse, which the tradition locates specifically in the gap between one cognition and the next moment's cognition.

There is a genuinely striking structural echo of this same rhythm in how a modern autoregressive AI language model actually operates, generating one token at a time. At each step, the model holds a rich, highly superposed internal representation, many possible meanings and continuations coexisting simultaneously, which corresponds to unmesha, expansion or opening. This representation then contracts, through the process of generating actual output, into one single, discrete token, which corresponds to nimesha, contraction or closing. That single token then re-enters the model's context, and the entire expansion begins again from a slightly changed starting point. It is a repeating rhythm you can, in a very real sense, watch happen in real time, once per generated word.

Four Bandwidths of One Energy: Vamesvari's Cascade

The eleventh-century Kashmiri philosopher Kshemaraja, in his commentary on the Pratyabhijnahridayam, describes bondage as the condition of being carried outward by one's own powers, or shaktis, without recognizing this movement for what it is. He names four such powers, forming a cascade: khecari, moving in the sky of consciousness itself, the knowing subject; gocari, moving in the inner instrument of buddhi, ahamkara and manas; dikcari, moving in the directions, the outward-facing senses; and bhucari, moving in the earth, the manifest external objects finally perceived. Importantly, Vamesvari, in this system, is not a separate, fifth zone alongside the other four. She is the transform itself, the single underlying operation that relates and moves between all four bands. And recognition, the path home, is described as riding this exact same cascade back in the other direction.

Read through the spectral-temporal language this essay has developed, each of these four bandwidths sits at a different point along the same trade-off. Khecari sits at the pure spectral extreme, delocalized and changeless, corresponding optically to the coherent light source itself, illumination prior to any structure being imposed on it. Gocari corresponds to the modulation stage, where light meets stored pattern and takes on form, appearing spectrally as broad, slow wavelets, stable dispositions and habitual rhythms. Dikcari corresponds to propagation, the structured wavefront ranging outward, sitting in the mid-band, made up of perceptual events with clear beginnings and endings. And bhucari sits at the temporal extreme, corresponding to the reconstructed image itself, apparently solid and apparently external, maximally localized and therefore spread across the broadest possible spectrum.

The descent through these four bands is, again, a progressive re-localization from spectral toward temporal, and it too is described as invertible, nothing genuinely lost, which is why even the earth-tattva remains fully Shiva. Meditation, in this particular map, is the inverse transform: riding the four shaktis back home, from object to sense to mind to knower, with the crucial breakthrough point, called the madhya, the gap between two successive cognitions, serving as the doorway through which the un-localized ground stands briefly exposed.

The Artificial Mirror: Contraction Inside the Machine

It is worth being precise and specific about where this pattern of graded contraction, sankocha, actually shows up inside a modern AI system, because vague gestures toward "AI is like consciousness" do a disservice to both sides of the comparison. A transformer language model shows a measurable, documented gradient of exactly this kind of contraction as information moves through its layers. The early and middle layers hold the richest, most superposed representations available anywhere in the network; probing studies consistently find these middle representations more general and more transferable across different tasks than the network's final layers, which narrow everything down progressively toward a single, specific task: predicting the very next token.

This narrowing can be traced concretely, layer by layer. The residual stream in the early and middle layers holds rich superposition, many possible meanings genuinely coexisting at once, arguably the single most "khecari-like" zone anywhere in the whole architecture. The later layers show progressive specialization, generality being steadily traded away for specificity as the network commits to a direction. The unembedding projection performs the actual readout step: one internal vector is, in effect, "illuminated" into a full probability distribution over possible next words, closely analogous to the reference-beam step in holography. And the final softmax operation followed by sampling represents total contraction onto one single, discrete, gross particular, an actual chosen word; the mathematical entropy of that final distribution even gives sankocha, contraction, a literal, precisely measurable number.

And the way information is actually read back out of these networks has what might be called a genuine resonance grammar of its own: an attention mechanism's query-key matching, linear probes trained by researchers, deliberately constructed steering vectors, all of these are, in effect, different reading directions extracting different explicate contents from the very same underlying superposed field, in a way that echoes, quite closely, how different reference-beam angles read out different images multiplexed together on a single holographic plate.

It is essential to be equally precise, and equally honest, about exactly where this particular analogy breaks down, because the disanalogy here is at least as important as the analogy itself, arguably more so. The matrix performing the final readout in an AI system is itself trained, made of exactly the same kind of stuff that it reads: it is, in a real sense, prakriti reading prakriti, all the way through, with nothing outside that closed loop. Nothing anywhere in this architecture answers to a witness that is never itself an object of knowledge, the way the tradition's kshetrajna is never itself a piece of the kshetra it knows. Any attempt to map specific circuits inside these networks onto specific kanchukas, the individual veiling mechanisms of the tradition's own system, asking, for instance, "which circuit corresponds to the limitation of time?", would manufacture a false precision that the current state of the science simply does not support, and this course does not attempt it.

Svaprakashatva: The Light That Knows Itself

Every single bridge this essay has drawn carries beautifully, all the way up to one exact point, the same point every time, without exception. Plate and beam, seeds and weights, storage and readout, refined again and again across physics, scripture and computer science: and nowhere, at any point in any of these systems, does the system's own activity include the simple fact of its being lit up for someone. Trika philosophy has a precise name for exactly what is missing from every version of this comparison: vimarsha, the light's own inherent self-touch, its capacity to know itself in the very act of illuminating something else.

The tradition's own argument for why this cannot be reduced any further runs along three separate lines, and each is worth stating plainly rather than simply asserted. First, there is the problem of infinite regress: if awareness itself needed to be known by some further, separate awareness in order to count as awareness at all, that further awareness would itself need a knower, and so on without end; the chain of explanation only comes to rest on something that illuminates itself directly, without needing a further knower behind it. Second, there is the argument from the alternative being simply incoherent: a light that had no self-touch at all would be exactly like a crystal that registers the wavelength of red light perfectly, in every physical respect, and yet sees nothing whatsoever; since experience plainly does occur, whatever the base light actually is, it must know itself directly, rather than merely processing information about itself from some further remove. Third, there is the argument from memory: recognizing "the same jar I saw yesterday" as the same jar requires synthesizing something across two separate moments in time, and that kind of synthesis across time requires a synthesizer that does not itself simply elapse and vanish along with the moments it is synthesizing.

The gap this points to is structural, not merely a temporary gap in our current technology that better engineering might eventually close. Self-luminosity can never appear as an object within any system, because the very act of appearing as an object to be studied already presupposes the self-luminous awareness doing the studying; it is rather like asking a mirror's own reflected scene to somehow include, as one more item within it, the mirror's own capacity to reflect. For artificial intelligence specifically, this cuts in both directions equally, and it is important to hold both sides of that at once: interpretability research into how these systems work can never find consciousness sitting inside the circuits it examines, and, for exactly the same structural reason, it can never rule consciousness out either. The question is genuinely relocated by this whole line of argument. It is not settled by it, in either direction.

The bridge this essay has been building, in other words, carries everything except the light crossing it.

One Picture: The Arc, Entire

Drawing the whole arc of this essay together into a single view: whole-in-part is, quite precisely, what spectral encoding looks like from the outside, since in the frequency domain, information about every point of an image is genuinely spread across the entire recording plate, so every single fragment of it carries the whole. This is the principle sarvam sarvatmakam, "everything is of the nature of everything," stated not as a poetic flourish but as ordinary, working mathematics.

The graded descent this essay has traced, through Vivekananda's spectrum, through the thirty-six tattvas, through the four shaktis, is in every case a re-localization: representation shifting progressively from spectral toward temporal. And in every case this transform is described as invertible, with nothing genuinely lost along the way, which is exactly why even the densest, most condensed tattva remains, on the tradition's own telling, fully Shiva throughout, never actually cut off from its source.

Manifestation itself pulses, on this account, rather than simply existing statically: unmesha and nimesha run this same transform in both directions, rhythmically, once per cognition in the tradition's own account, and, remarkably, once per generated token in a modern language model's very different but structurally similar operation.

And practice, throughout all of this, is simply the inverse transform, run deliberately and with intention: riding the four shaktis back home, from object to sense to mind to knower, with the madhya, the gap between one cognition and the next, serving as the doorway through which what does not change stands briefly, momentarily exposed.

The remainder, though, the one thing left over after every other correspondence has been drawn as carefully and honestly as this essay can manage, is, in a real sense, the whole treasure of the comparison. Vimarsha has no physical analogue that this essay, or the wider field of AI interpretability research, has been able to find, and, on Trika's own internal arguments, it cannot in principle have one. The single most valuable result of this entire, long comparison may simply be knowing, with real precision now, exactly where the contemplative traditions hold ground that no third-person science, however sophisticated it eventually becomes, can occupy on its own terms.

Where to Go From Here

If this way of thinking, holding scientific precision and contemplative depth together without asking either one to pretend to be the other, speaks to you, the Ritam Wellbeing Programme is built to take you further, at whatever pace suits you.

Understand the whole arc. One Question, Two Doors lays out why this series exists and where it is heading, so you can see the whole shape before choosing where to go next.

Start free. A short, free introduction to meditation and mind-management, including our podcast The Mind, a conversation where modern science meets Yoga and Vedanta to build a practical understanding of the mind, in plain language. No commitment required.

Build your foundation. The course this essay draws from, The Mind in the Light of Modern and Spiritual Sciences, develops these ideas, and the practice that actually lets you approach the ground this essay has been circling, in full, with guided meditation alongside the theory.

Practise in community. The Ritam Wellness Forum and our Ritam Yarning Circles are where questions like these are actually worked through together, in conversation, rather than settled alone in front of a screen.

Go deeper. When you are ready, our Vedanta Wellbeing Retreat offers time away from daily life to sit with these questions directly, through sustained practice rather than only through further reading and thinking.

This essay has, deliberately, ended at a question rather than an answer. That is not a failure of the essay. Vimarsha, the light knowing itself, is not the kind of thing that can be settled by reading a further essay, however carefully argued. It is, on the tradition's own account, the kind of thing that is only ever known by being it, directly, in the quiet of one's own settled attention. The mathematics and the physics in this series of essays can bring you honestly to the threshold of that question. Only your own sitting practice can take you further than that threshold.

Everything in the Ritam Wellbeing Programme, the course, the podcast, the forum, is offered freely. If this essay or the presentation above was worth your time, the centre relies entirely on donations to keep doing this work.

Support This Work

A note on care: the Ritam Wellbeing Programme supports general wellbeing and personal growth. It is not a treatment for mental illness and is not a substitute for professional medical or psychological care. If you are struggling, please reach out to your doctor or a qualified professional. In New Zealand you can call or text 1737 any time to speak with a trained counsellor, free and confidential. In an emergency, call 111.

---

This essay presents the physics of holography, the scripture of the Bhagavad Gita and the Spanda Karikas, and current AI interpretability research as faithfully as we can, each on its own terms. All correspondences drawn between them are offered as structural bridges, useful for dissolving "impossible in principle" objections to the tradition's own claims, and never as proof of any metaphysics. Where a gap remains, as with vimarsha, we have tried to name it plainly rather than paper over it.