When confronted with the term holographic, I tend to tune out. Is our universe a hologram? …asks yet another popular science article. I scroll onwards. When I’ve encountered this term in relation to consciousness, it has typically failed to pass my anti-woo filters.
This said, I’ve seen both meditative and psychedelic experiences described as holographic by people whose observational skills I respect and whom I mostly trust not to misuse mathematical terminology, so I’m going to challenge myself to engage with this term.
There’s two key aspects of holography which I’d like to highlight:
Every part contains a representation of the whole
A higher dimensional artifact emerges from a lower-dimensional substrate
In this post, I will attempt to steelman the case that our objective understanding and subjective experience of consciousness exhibit both of these qualities. If these properties do characterise consciousness, we should expect to find evidence in both neuroscience and phenomenology.
I think the fastest way to understand holography from first principles is to watch this video from Grant Sanderson’s YouTube channel, 3blue1brown. It’s a world-class resource and I highly recommend taking the time to watch it in full. However, in case the reader prefers not to watch a forty-six minute explainer video, I’ll run through what the relevant aspects of holography are for the purposes of this article.
The important thing to understand is that a hologram is constructed using two coherent light sources – a reference wave, and an object wave. The reference wave propagates through space unimpeded, whereas the object wave is reflected off the scene to be holographed.
Where these strike a holographic plate – which is generally a silver halide photographic emulsion, though one with a much smaller film grain than typical photographic film – an interference pattern is formed, which is then recorded on the film. From the video:
This only works if all the light has the same frequency. So, you cannot illuminate the scene with ordinary white light – what you have to do is use a laser. A clever way to do this is to pass that laser through a beam splitter, where half of it gets spread out, bounces off the scene, and hits the film. We’ll call that the object wave.
And then the other half also gets spread out, but it doesn’t interact with anything before hitting the film. This will act as the reference wave.
Those two waves interfere at the plane of the film in a way that depends heavily on the phase of that object wave.
This exposure pattern generally resembles an illegible speckle pattern. This pattern records the phase difference between the reference wave and object wave at the point of intersection with the holographic plate. If the film is developed and then put back in place, and the object wave is then removed while the reference wave is retained – the holographic plate will then interfere with the reference wave in such a way that the object wave is then recreated on the opposite side of the plate. Back to the video:
If you now remove all the objects from the scene, and you block that object beam so the only thing shining on this now-exposed film is the reference beam, then what it produces beyond the glass includes a complete recreation of that object wave – a recreation of the light that would be there if the object beam were still shining.
We can see now how a complete representation of a three-dimensional light field can be encoded on a two-dimensional substrate.
It’s important to understand how the visibility of the original scene depends on the state of the reference wave. If the reference wave is tilted even slightly, or if its wavelength is changed – then the scene disappears. However, if the film has not been developed yet, then a new interference pattern can be recorded over the top of the old one – and it’s even possible to create multiple holograms in this way through a process of stop motion. This implies that the storage capacity of a holographic plate could be very large. We’ll return to the implications of this later.
What is also quite remarkable, is that only a small amount of the photographic plate is required to view the entire image – as every part of the photographic plate contains information from the entire object:
We cut out a very small circle from the film that we recorded. In an ordinary photograph, cutting out a very small piece obviously cuts away the vast majority of the scene – but for a hologram, holding up that same small little circle of film to the reference beam – as you shift your viewing position, looking through that circle you can see essentially every part of the scene recorded.
As we can see, every part does indeed contain a complete representation of the whole.
The idea that memory might be stored in a distributed fashion dates back to Karl Lashley’s experiments in the 1920s, using cortical lesions on rats to demonstrate that memory obeys the principles now known as Lashley’s laws. From Cognitive Processes and the Brain (Milner and Glickman, 1965):
The Law of Mass Action, which states that learning deficits are a direct function of the mass of cortical tissue destroyed.
The Law of Equipotentiality, which states that the deficit is independent of the locus of the damage.
The first person to directly analogise these properties to holography was Karl Pribram, who began lecturing on holographic models of brain function in 1966, and published a book, Languages of the Brain, in 1971. However, the first rigorous mathematical framework showing how neural processes could implement holographic principles was published by Philip Westlake in 1970, one year beforehand. From The possibilities of neural holographic processes within the brain (Westlake, 1970):
Summary. A theory of brain functioning is proposed based upon an analogy to optical holographic processes. There are many properties which holography potentially offers to neurophysiology. Chief among these is the property of distributedness, which is displayed only by holographic processes. This property, an attribute of certain types of holograms, permits any small portion of the hologram to reconstruct the entire original scene recorded by the hologram. Because of this fact and other supporting evidence, neural versions of the holographic processes appear as most promising candidates for the coding of sensory and memory processes.
Westlake proposed that neural spike trains could be analogised to the waves of light in holography. From there:
The equivalent of the reference wave is a synchronized plane wave of neural spiking arriving at synaptic junctions with uniform timing.
The equivalent of the object wave is a wave of neural spiking emanating from sensory object source points, with timing that depends on the distance traveled.
The equivalent of the holographic plate is the array of synaptic junctions where these waves interfere through spatial summation, creating a distributed pattern of neuronal firing that encodes the interference pattern with changes to synaptic connection strengths.
Then, when the original reference wave is presented again to the distributed pattern of synaptic connections that recorded the original interference pattern, these would interact in such a way as to reconstruct the original object wave – and so unfolds our holographic memory system. The reference wave could then be modified in order to access different memories.
Illustration of waves of neural spike trains moving across a section of cortex. From Languages of the Brain (Pribram, 1971)
It’s a concise theory that would explain how memory could be distributed throughout the brain as well as its high storage capacity. It would also permit the encoding of three-dimensional sensory impressions in a two-dimensional structure such as the cerebral cortex.
However, Pribram’s ideas have had limited influence on modern neuroscience, which has moved away from holographic explanations of neural processes largely due to lack of empirical validation. This may be because we still lack the technical capability to test many of Pribram’s specific predictions at the level of fidelity and scale that would be required – Pribram himself thought that this would require electrode recordings down to the level of individual dendrites. The most compelling evidence for holographic memory may come from biological experimentation conducted at the time. This brings me to a book I’ve been wanting to review:
Paul Pietsch was Professor of Anatomy and Chairman of the Department of Basic Health Sciences at the Indiana University School of Optometry, where his research focused on developmental biology. He’s an engaging writer whose tales about working in a biology lab make for an entertaining read.
Pietsch was originally studying regeneration of tissues and organs, and this meant that he was working with salamanders and other amphibians, given their tolerance of tissue transplants and limb regeneration abilities. In the mid-1960s he developed an interest in extending his theories of tissue regeneration to that of memory, and for this reason he began working with the brain. This was also around the time that hologramic theory was gaining popularity, and so he expected that his own research would invalidate it. In his own words:
Hologramic theory not only stirred my prejudice, it also seemed highly vulnerable to the very experiments I was planning: shuffling neuroanatomy, reorganizing the brain, scrambling the sets and subsets that I theorized were the carriers of neural programs. I fully expected to retire hologramic theory to the boneyard of meaningless ideas.
I should have awaited Nature’s answers. For hologramic theory was to survive every trial, and my own theory went down to utter defeat.
Despite his skepticism, Pietsch recognised the significance of Leith and Upatnieks’ work and its relevance with regards to Lashley’s assertions:
Not a word about mind or brain appeared in Leith and Upatnieks’s articles. But to anyone even remotely familiar with Karl Lashley’s work, their descriptions had a very familiar ring. Indeed, substitute the term brain for diffuse hologram, and Leith and Upatnieks’s observations would aptly summarize Lashley’s lifelong assertions. Fragments of a diffuse hologram reconstruct whole, if badly faded, images. Correspondingly, a damaged brain still clings to whole, if blurred, memories. Sharpness of the reconstructed image depends not on the specific fragment of hologram but upon the fragment’s size. Likewise, the efficiency with which Lashley’s subjects remembered their tasks depended not on which parts of the brain survived but on how much brain the animal retained.
He also recognised the wide-ranging significance of the multiple hologram, by which many scenes may be superimposed on a single holographic plate. Indeed, varying the reference wave – or reconstruction beam, as he calls it – could be the mechanism by which one might recollect old scenes or synthesise new ones:
If a single holographic code is so very, very tiny, any physical area should be able to contain many codes – infinitely many, in theory. Nor would the codes all have to resemble each other. Leith and Upatnieks recognized these properties early in their work. Then, turning theory into practice, they went on to invent the “multiple hologram” – several different holograms actually stacked together within the same film.
With several holograms in the same film, how could reconstruction proceed without producing utter chaos? How might individual scenes be reconstructed, one at a time? Leith and Upatnieks simply extended the basic operating rules of holography they themselves had developed. During reconstruction, the beam must pass through the film at a critical angle – an angle approximating the one at which the construction beam originally met the film. Thus, during multiple constructions, Leith and Upatnieks set up each hologram at a different angle. Then, during reconstruction, a tilt of the film in the beam was sufficient for one scene to be forgotten and the other remembered.
One of Leith and Upatnieks’s most famous multiple holograms is of a little toy chick on wheels. The toy dips over to peck the surface when it’s dragged along. Leith and Upatnieks holographed the toy in various positions, tilting the film at each step. Then, during reconstruction, by rotating the film at the correct tempo, they produced images of the little chick, in motion, pecking away at the surface as though going after cracked corn.
Emmett Leith and Juris Upatnieks.
Multiple holograms permit us to conceptualize something neither Lashley nor anyone else had ever satisfactorily explained: how one brain can house more than one memory. If the engram is reduplicated and also equally represented throughout the brain, how can room remain for the next thing the animal learns? Multiple holograms illustrate the fact that many codes can be packed together in the same space.
Just as important, multiple holograms mimic the actual recalling and forgetting processes: tilt the film in the reconstruction beam, and, instantly, off goes one scene and on comes the next. A few years ago, I met a young man named John Kilpatrick who suggested that a person trying to recollect something may be searching for the equivalent of the correct reconstruction angle.
But suppose that instead of using a single reconstruction beam, we use several. And suppose we pass the beams through the multiple hologram at different angles. We may, in this manner, synthesize a composite scene. And the objects in the composite scene may never have been together in objective reality. When the human mind synthesizes memories into unprecedented subjective scenes, we apply terms such as thinking, reasoning, imagining, and even hallucinating. In other words, built right into the hologramic model are analogues of much human mental activity.
Pietsch’s initial “shufflebrain” series of experiments involved first replicating Lashley’s ablation experiments before moving on to his own series of experiments which involved rearranging brain tissue in various ways – for instance, swapping left and right cerebral hemispheres, or switching the diencephalon with the cerebrum. He also experimented with transplanting brain tissue from salamander to salamander – as he says, every operation I could think up. In all cases – except for those where the medulla was destroyed, which would render the animal permanently unconscious – feeding behaviour survived.
Pietsch’s illustration of salamander brain anatomy.
He was attempting to demonstrate what he called the independence principle – that each piece of brain makes its own contribution to an animal’s behaviour independent of any other. This predicted that he should be able to transplant new memories into a brain, and for this he needed a subtler experiment. I’ll let him tell this story, because it’s a good one:
I lost my job at the beginning of 1970, before shufflebrain was a complete story. A depression had begun in the sciences during 1969. Directly or through friends, I soon contacted the anatomy department of every medical school in the United States and Canada, without success. And wherever else I looked, there were no jobs, not for me at least. In any case, when a friend eventually arranged an interview for me at Indiana University’s optometry school, I found it psychologically impossible to negotiate seriously for anything. Had my pride been operative, I would have rejected their job offer, which carried lower rank and less pay than my former job. And I would never have worked as a scientist again.
But by the autumn of 1970, I was drawing real wages once more. I had a splendid office overlooking the most beautiful campus I had ever seen. Although my lab had nothing in it, my morale was excellent. I had applied to the university’s grant committee for a few thousand dollars to tide me over until I could secure federal funds. When I got four hundred dollars instead, I was still too euphoric to bitch. And I set about doing what scientists of the generation before mine had done routinely: I made do.
Making do included scrounging salamanders from a wonderful man, the late Rufus Humphrey. Humphrey had retired to Indiana University from the anatomy department of the University of Buffalo. I had joined that department, myself, for a brief period in the early 1960s. After taking over some drain tables Humphrey had once used for his salamanders, I had written him to tell him that his picture still hung in the microscopic anatomy lab at Buffalo. Thus we began a lasting correspondence. Humphrey studied the genetics of a salamander known as the axolotl. Some of his purebred strains ran back to 1930. His colony was famous, worldwide, among people who work with amphibians. Even if I had not been on a scrounging mission, one of the first things I would have had to see in Indiana was Humphrey’s axolotl colony.
Making do also meant giving up the live tubifex worm as staple for my colony. Detergents and chemical pollutants have driven these once-ubiquitous worms from all but a few waters. Since the early 1960s, I’d not been able to collect them in the field, but had had to fly them in from New York or Philadelphia, which was totally out of the question on a make-do budget. Thus I began feeding young larvae on freshly hatched brineshrimp embryos, which could be purchased dry by the millions for a quarter in any pet shop. When the axolotl larvae reached about 40 millimeters, I weaned them onto beef liver swiped from my wife’s shopping basket.
Feeding animals on beef liver does take time. The animals must first be taught to strike. Even after they acquire the necessary experience, though, you still can’t fling a hunk of meat into the dish and forget it, as you can a ball of live tubifex worms. The liver rots at the bottom of the dish, while even the experienced feeder starves.
Now there was a federal program called Work-Study, whereby the government paid all but 20 percent of the wages for students who had university-related jobs. Just as I was weaning a group of about fifty axolotls onto liver, an optometry student, Calvin Yates, came around looking for a Work-Study job. One duty I assigned him was feeding liver to the axolotls.
Calvin now practices optometry in Gary, Indiana. If his treatment of people matches the care he gave my salamanders, I am sure he is an overwhelming success. Calvin had what Humphrey once called a “slimy thumb” — the salamander buff’s equivalent of the horticulturist’s green thumb. In Calvin’s presence, living things thrived. A few days after he took over the job of weaning, the axolotls were snapping like old veterans. Calvin also introduced a clever trick into his feeding technique. He would tap the rim of an axolotl’s plastic dish and then pause a few seconds before presenting the liver. In a few days, tapping alone would cause the larva to look up, in anticipation of the imminent reward.
I paid only the most casual attention to Calvin during that time. For I had learned that my favorite species of salamander, Ambystoma opacum, lived in the area. Opacum was one of the three principal species I had been using in my shufflebrain experiments. Luckily, the female opacum lays her eggs during the autumn. Finding them can be tricky, though. Fortunately, I met a man who happened to have 50 eggs he was willing to let go for two dollars. And by the time Calvin was weaning the axolotls, the opacum larvae had grown to just the right size for me to put the finishing touches on my shufflebrain project.
The opacum belongs to the same genus as does the axolotl (Ambystoma mexicanum). Opacum is a shrewd little animal, the smallest member of the genus but easily the most elaborate and efficient hunter. And what a fascination to watch! But its small size made liver-feeding impractical, which was also lucky for me.
One afternoon at the tail end of an operating session, I realized that I had anesthetized one too many opacum larvae. It is against my standard procedures to return such animals to stock. Yet I don’t like to waste a creature, make-do budget or not. On impulse, I decided to see how well an axolotl’s forebrain would work when attached to an opacum’s midbrain. And I took an animal from Calvin’s colony to serve as the donor.
◇ ◇ ◇
The fateful moment came ten days later. I had taken my time getting to the lab that morning, walking slowly through the crisp autumn air, admiring the trees, saying several “good mornings” to students along the way, and had seated myself perfunctorily at the operating table, thinking much more about the world in general than about science. I usually keep recuperating animals near the microscope and check their reflexes daily until they come out of their stupor. That morning, I came in merely to take a routine daily record.
To check a salamander’s reflexes, I flick the edge of the dish. When an animal has recovered from postoperative stupor, it usually jumps in response to the noise. As I placed the opacum larva with the axolotl brain on the stage of the microscope, I noticed that he had righted himself and was standing on the bottom of the dish. I gave a light flick, expecting him to give a little jump and then swim out of the microscopic field. Instead, he slowly arched his little back and looked directly up into the barrel of the microscope, right into my eyes. My heart missed a beat. I had observed this looking-up response in only one other place – over on the table among Calvin’s axolotls, where the donor had come from. Immediately I jumped up, went over and flicked every last dish on the axolotl table. Every axolotl there looked up in response.
Next I checked out the stock opacum larvae. Flicking only caused them to scurry around in their dishes. Not one stock opacum looked up.
Now back to the operating table. Again, I flicked. Again the opacum with the axolotl brain looked up. I tested the other subjects that had had operations. They did not look up. Again I tried the axolotl recipient. Again it worked. Unwittingly, I had discovered that a learned response can be added to the hologramic deck.
Pietsch published these results in an article in the May 1972 issue of Harper’s. This attracted the attention of CBS, who offered him a piece on 60 Minutes. At the same time, Pietsch followed these results up with a three-way transplant experiment, to verify that memories are retained in both the donor and host animals:
The experiment I performed on camera involved three animals: two naive axolotl larvae and a trained adult “looker-up” from Calvin’s old colony. The anterior part of the trained adult’s cerebrum replaced the entire cerebrum of one naive larva. Would the host become a looker-up? The other naive larva served as a donor animal, and I transplanted its entire brain into the space left in the adult’s cranium. Would the transplant “confuse” the adult? Mike Wallace eventually called the second larva “the loser.” For it received no brain transplant.
I decided not to call the viewer’s attention to looking up, and instead focused attention on the survival of feeding after shufflebrain operations. I had no doubts about feeding. But looking up was still very new. Something could have turned up to change my mind. If the paradigm turned out to be a fluke, trying to correct the misinformation broadcast on television would be like attempting to summon back an inadvertently fired load of buckshot.
◇ ◇ ◇
CBS broadcast the show a year later. In the interim, I had carried out enough testing and had conducted sufficient control experiments to be sure of the results. Within about one week, a previously naive recipient of a trained looking-up animal’s cerebrum becomes a looker-up itself, without training. And these animals retain the looking-up trait for the remaining months, or even years, of their lives. Controls, animals with transplants from the brains of naive animals, do not show this response. While the initial experiments – like those I performed on camera – were with the cerebrum, I obtained the same results with pieces of midbrain and diencephalon.
The trained cerebral-donor animals were very interesting. As soon as the effects of anesthesia wore off, these animals demonstrated that they remembered the signal to look up. In other words, looking-up memories existed in the donated as well as the retained parts of these animals’ brains. What was true of innate feeding behavior worked for looking up: memory wasn’t confined to a single location in the brain.
I had also repeated Mike Wallace’s “loser” experiment. I found that true to the principle of independence, the extra brain parts did not “confuse” the host.
Professor Paul Pietsch, author of Shuffle Brain.
I’ll confess that I chose to review Pietsch’s salamander experiments in part because I have a soft spot for old stories well told. If the reader is interested in modern studies which also validate Lashley’s laws, I recommend reading about Michael Levin’s work with planarian flatworms, or a more recent MIT study demonstrating distributed engrams in mice:
Taken together, these studies provide compelling evidence that memory is stored in a distributed fashion, as we might expect if it worked using principles of holography. However, we still lack any real evidence for actual holographic processes in the brain. If holographic principles extend beyond memory storage to the generation of conscious experience itself, we should expect to find some traces of this in the structure of subjective awareness. If this were the case, what might we expect to see?
For context: I’m an independent researcher interested in using phenomenology of altered states to help discover how consciousness works. Sometimes I have to explain this to people, so I often describe this as a reverse engineering process.
Let’s say I hand you a piece of software which makes lossily-compressedJPEG image files. You can’t inspect its internals, but there is a big metaphorical dial on the front which you can turn up to increase the lossy compression rate. If you were smart enough, do you think you could figure out that JPEG works by splitting the image into 8×8 pixel blocks, applying the discrete cosine transform, and then discarding the high-frequency components?
This image should switch between an uncompressed PNG and a heavily compressed JPEG once every second. If you look closely, can you see the blocky compression artifacts?
To this end, if we are looking for evidence that consciousness is holographic, might we notice artifacts of the holographic reconstruction process within our subjective experience? After vibecamp, I went to New York, where I sat down with the meditation teacher Wystan Bryant-Scott to have a long and technical discussion about this.
Wystan has extensive experience with a wide variety of exotic states accessed through meditation, including something known as a cessation, where the entirety of subjective experience blinks out for a moment. This is likely quite challenging to relate to if you’ve not experienced it – I personally haven’t. This animated video contains the most accessible description I’ve yet encountered:
Wystan claims that while emerging from the contentless void of a cessation, it’s possible to observe the phenomenal fields refabricating themselves. This seemed like a state in which one might plausibly observe the artifacts of holographic phenomenology, so we started our discussion here – before attempting to ground our observations in more mundane, relatable, day-to-day states.
Cube Flipper: One way in which something can be said to be holographic is that every part contains a representation of the whole. The other relates to the sense that a higher dimensional construct emerges from a lower dimensional substrate. So in the case of an actual hologram, you get a three-dimensional experience from a two-dimensional substrate – which is pretty wild, but the math works.
Cube Flipper: What I’m trying to say is that phenomenology might be similar, and that you ultimately have a two-dimensional field of experience from which a higher-dimensional or depth-filled experience emerges.
Wystan: I mean this is clear from the visual field, for sure. I think it takes a little bit more defabrication for the somatic field to collapse down to two dimensions, but I’d say that’s also true. Like… I can nudge, I can induce my somatic field to go into a plane. But it’s very three-dimensional almost all of the time. But here’s the ringer, though – even two dimensions is fabricated. Their dimensionality, spatial volume and all, that’s all fabricated.
Cube Flipper: Fabricated how? What’s the process of fabrication, here?
Wystan: It happens so quick that it’s difficult to verbalise. You experience this the clearest coming out of cessation. It happens so quick – my experience has been that you perceive it more clearly and you get more detail the longer you’ve been in cessation, and how concentrated you were, like what your level of samadhi was going in. And depending on those factors, then you get both a clearer entry and a clearer exit.
Wystan: And yeah. Any spatial volume at all is fabricated. It bootstraps on itself.
Cube Flipper: Insofar as our experience might be built out of qualia dust, what is a qualia dust mote? What’s its informational content? Could it be that every qualia dust mote contains a reflection of every other qualia dust mote?
Wystan: Yes. This relates to Indra’s net phenomenology – and Shinzen’s obsession with category theory, this is where that comes from – because each mote of qualia dust is nothing in itself, it only has whatever informational content it has as a holistic property of the whole field. Does that make sense? For what it’s worth I prefer a field metaphor to a dust metaphor – because that gives a sense of a particulate structure that isn’t there.
The manner in which all dharmas interpenetrate is like an imperial net of celestial jewels extending in all directions infinitely, without limit… As for the imperial net of heavenly jewels, it is known as Indra’s net, a net which is made entirely of jewels. Because of the clarity of the jewels, they are all reflected in and enter into each other, ad infinitum. Within each jewel, simultaneously, is reflected the whole net.
A three-dimensional rendering of Indra’s net. From Wikipedia.
A generous interpretation is that this is trying to describe holographic principles using poetic language. Contemporary illustrations of Indra’s net often employ visualisations like raytraced lattices of spheres all reflecting each other recursively. My understanding is that this should be not be regarded as a visual replication of an actual altered state, but rather an illustrative but not inaccurate representation of a subtle yet all-pervasive phenomenon present throughout everyday experience. We’ll discuss this shortly.
Wystan: So, in the Pali Canon, when emerging from cessation – and this is reported by contemporary practitioners as well – and they tend to use imagistic or analogical language and are not very precise – they talk about perceiving dependent origination. And that’s where the insight comes from – perceiving it coming out of cessation. And having had that experience a bunch of times now, and perceiving it with more clarity, I think that’s what they’re talking about. Perceiving dependent origination in real-time, as the field re-forming itself, coming out of cessation.
Cube Flipper: So it might be one qualia dot, or a little whitecap of turbulence that appears somewhere, and because it reflects other ones—
Wystan: Yes. And they talk about the links – that’s the language that’s used.
Dependent origination is one of the most fundamental and practical teachings in Buddhism. It states how no phenomena exist in isolation or have inherent, independent existence – all phenomena arise in dependence upon one another.
Did you wind up watching the holography video? This might be a big, impressionistic leap – but I could not help but be reminded of Sanderson’s visualisation of a hologram being constructed from individual points:
A hologram of a single point gives a zone plate, and the diffraction effect of shining the reference through the zone plate includes a recreation of the wave from that point. A natural place your mind might go from here is to wonder about two such points in space. Right here is a simulation for what the resulting exposure pattern would look like – it’s based on adding the two radial waves from those points, together with a reference wave – and you can clearly tell that it’s related to the zone plate associated with each one, effectively encoding two distinct three-dimensional positions.
But it is more complicated than simply adding the two individual patterns, in the same way that the double slit interference is more complicated than adding the brightness from each individual slit. The result depends on how the waves – and their phases – interfere. Even still, you can see how this might lead you to think about building up a scene as a combination of multiple points.
By more complicated, Sanderson means that summing light waves involves summing complex amplitudes, rather than real amplitudes. Notice how adding just one point influences the entire holographic plate? Perhaps the links that Wystan speaks of have some relation to the irreducibility of the interference pattern – or at least, perhaps there’s something analogous going on, though it might not necessarily share its exact implementation details with this simple example. In any case, Wystan related to the concept of constructive interference, when I brought it up:
Wystan: Constructive interference. That’s what it feels like, coming out of cessation – constructive interference. It’s the field interfering with itself that gives rise to the most basic fabrications of time and space. And like – these metaphors exist for a good reason. It’s like, the field flattens out and it’s introspectively dead – lights out – gone. And then, the subtlest pertubation – constructive interference – the whole thing – fwoof – you’ve got time and space.
Wystan: And, not in the Buddhist tradition, but in Shaiva Tantra – they refer to their version of cessation as the waveless. For damn good reason.
I thought it was appropriate that we move on from discussing these heavily dereified states to more reified ones. Perhaps some discussion of less extreme altered states would reveal similar holographic principles?
Cube Flipper: Something I notice when I use cannabis is that it tends to put a lot of noise into my system. It’s very noticeable in the waves making up the visual field, and if I stare at the ceiling, I’ll see this noise with a density that’s proportional to my state. It becomes very obvious that I can attune to this, and even sober I can see the waves making up flat patches of colour, and I can gauge my state or level of brain fog by how foggy or crap those waves are or how crystal clear they are.
Cube Flipper: It’s also like – when I take even a modest amount of psychedelics I notice that this noise starts to harmonise. If I then take a slightly larger dose of mushrooms, it will be the case that the noise on the ceiling will actually congeal into regular patterns. And this feels holographic to me, almost like every part is reflecting every other part, and that’s what gives you the ability to harmonise this stuff.
Cube Flipper: And it seems like it’s the psychedelics which make every part reflect every other part more, which means that every part is connected to every other part more, and then you start to get more coupling. But I’m having a hard time constructing that argument itself.
Wystan: I’ve experienced it both in meditation and psychedelics. This is also a function of samadhi. There’s more and more of a correlation between different parts of the field. Like, this happens with jhāna, the jhānic factors might start localised, but the most intense versions of jhāna that I know and experience whether formed or formless – it’s like every part of the field is positively correlated with every other part of the field in a feedback loop – and you get a maximally symmetric, harmonic state. And that also happens with visuals on psychedelics.
Cube Flipper: So I’ve seen, and I’ve experienced this myself sober, on a good day – and I’ve seen people who are meditators say, in a good mood I can make the grass look symmetric – and I’m just like, yeap, that’s highly relatable.
Cube Flipper: So, I think the way you phrased it just now, like, every part, recursively – so, it’s not just like point-to-point-to-point-to-point—
Wystan: Yeah, it’s an all-to-all relationship. Arrows everywhere, in every direction.
Wystan: So, it’s interesting in that consciousness there’s an all-to-all relationship between every part of the field all of the time in every state of consciousness, but depending on the energy parameter and the – I’m not sure what to call it – the samadhi parameter, there’s more of a positive correlation and a tendency towards harmony and symmetry.
Cube Flipper: It’s a weighted relationship and these weights can be more or less similar to one another.
Wystan: Yes.
What I think emerges from this conversation is a picture of consciousness as a fundamentally relational phenomenon – not just a collection of separate components, but a dynamic field in which each part depends on the way it mutually interferes with every other part.
What also emerges is a partial solution to the classic philosophical issue of the binding problem – how does the brain integrate disparate sensory impressions into a unified conscious experience? I think a complete solution would require a grounding in physics, whether electromagnetic or quantum – but at least here we may have a model of the dynamics by which experiences may achieve self-awareness through self-interference.
The implications of Pietsch’s experiments extend far beyond neuroscience. If memories can be transplanted and retained in both donor and host, I think that says some fairly disorienting things about the nature of identity. Does identity have holographic properties? Was identity a fake idea all along? Who can say.
I’m a software developer by trade, so perhaps I should stick to exploring the computational implications rather than the philosophical ones. In my early twenties, I got very fascinated by the mechanisms through which experience might be capable of self-observation. I spent a lot of time reading about reflective programming and metaprogramming systems, but I struggled to envisage how the brain might implement arbitrarily-deep recursive self-awareness without exhausting its available resources.
The field computing paradigm suggests a solution – the phenomenal fields simply experience themselves through self-interference. Their holistic nature means that the all-to-all relationship structure in which – as Wystan said – every part of the field is positively correlated with every other part of the field – can effectively be implemented for free. So – is asking what it means to be a hologram somewhat equivalent to asking what it means to embody a massively parallel analogue autocorrelator? Such a system sounds like it would have some useful computational properties.
Again, I think the most straightforward investigation of holographic phenomenology is through altered states. Personally, I think psychedelics have more repeatable effects than meditation, so I prefer to discuss those if I can. On two grams of psychedelic mushrooms, I can observe what Wystan called the energy parameter ramp up in real time, resulting in increased self-interference – every surface shimmers with an awareness of every other, standing waves forming volumetric haloes around every object.
Hey, maybe there’s good reasons why holographic stickers tend to be popular with the kind of people who like psychedelic mushrooms – though these foil ones really only demonstrate diffraction effects.
These effects are most easily noticeable by studying how they alter low-level sensory phenomena, but they can engender equivalent alterations to one’s high-level psychology as well. I think back to the first time someone gave me LSD at a party, and how paralysed by self-awareness I became. These higher-energy states are known for their characteristic effect on mood as well as how they intensify experience of meaning. It makes sense to me that as we increase the degree to which every part is aware of every other part we would gain a greater sensitivity to correlations, and this would then be felt as increased meaningness and spontaneous insight.
These altered states paint an extreme caricature of dynamics that are already present in ordinary consciousness – intensifying whatever computational process the hologram is already performing. What might things look like when the energy parameter is dialed down? I speculate that in the case of depression, the energy parameter is lowered, and the autocorrelation process operates less efficiently. I imagine that this would make it difficult for someone to recognise the context inside of which thoughts and sensations arise, resulting in tunnel vision, low self-awareness, and stuck behavioural patterns.
Conversely, dialing the energy parameter too high might neutralise depression, but it could also generate insight that isn’t grounded in reality. Part of caring for a holographic mind involves fine-tuning the energy parameter to ensure that the hologram represents a parsimonious model of reality.
So. Is consciousness holographic? I think the evidence suggests that we’re dealing with something more fundamental than metaphor here, and that optical holography could well be the mathematically appropriate model. Whether or not this is literally the case, this framing has proven remarkably generative. Holography connects distributed memory storage to emergent dimensionality, phenomenological reports of dependent origination to computational models of autocorrelation, and interference patterns to the binding problem. Even if we find that the brain doesn’t implement literal optical holography, holographic principles seem to capture something essential about how minds work.
The process of researching this piece has left me with more questions than I started with. In the interest of thinking in public, I’d like to put everything out on the table by sharing the list of open questions I have. Perhaps some of these could provide direction for future research:
Modern neuroscience models the central nervous system as fairly specialised – damage a particular region, and you’ll typically see consistent and specific deficits across multiple subjects. How do we square this specialised model with the model of distributed functionality described by the holographic model?
Pribram himself did not specifically believe a fully distributed model to be the case. From his book:
A neural holographic or similar process does not mean, of course, that input information is distributed willy-nilly over the entire depth and surface of the brain. Only those limited regions where reasonably stable junctional designs are initiated by the input participate in the distribution.
I’ve long suspected that memory is, like, spooky good. A few years ago, myself and a friend were watching old episodes of Stargate SG-1 – which we hadn’t seen since they originally aired in 1999 – and I was surprised to find that I still recalled nearly every plot beat. It’s enough to remind me of a claim I ran across in Mark Lippmann’s book:
I’m not doing a good job of unpacking it in this section, but, in some sense, our minds are nothing more than all the experiences we’ve ever had – and through memory and imagination we can have any experience. And, add two more pieces: the mind is practically lossless (in that any distinguishable sensory memory can be ultimately recovered) and that the mind is simultaneously “utterly malleable” (even while being lossless!).
When I spoke about this topic at vibecamp earlier this year, somebody raised the objection that discrete neural spike trains could not be treated in the same way as continuous waves of light. Westlake (1970) handles this by claiming that spike train phase offsets can be summed exactly like complex optical phases. This may seem oversimplified – modern models of neural firing are quite nonlinear.
For a different perspective, some weeks later I attended The Science of Consciousness Conference in Barcelona. Earl K. Miller gave a talk on neural dynamics and local field potentials – he had an analogy which maybe addresses this issue:
In the twentieth century we used to focus on neurons alone – but now I kind of think of neural spiking as like buzzing your lips at the mouthpiece of a trumpet. The notes start there, but they fully form in the resonance patterns in the trumpet’s body. That’s what I think is happening with neural spiking – it’s like the gross is creating these resonance patterns in the electric fields, and that is where a lot of the action happens.
Incidentally, Miller later mentioned he borrowed the trumpet analogy from Steven Lehar.
This question is perhaps equivalent to asking, where is memory stored? One mainstream model of memory formation is that short-term memories are initially encoded and temporarily stored in the hippocampus, then gradually consolidated into long-term storage using persistent synaptic connections distributed across the neocortex.
Maybe this tells us where the holographic plate might live at the macroscale, but confirmation of this would require study of more microscale dynamics. Different authors have proposed different structures at different scales:
There’s a modern idea that I think those guys working on holographic brain theories back in the 1970s would have been excited by – the predictive processing model.
In the predictive processing model, the brain has two interlocking information processing streams running in opposite directions – a sensory stream, and a prediction stream. At each layer, from low-level sensory processing all the way up to high-level world modelling, the difference between sensory information and predictions about sensory information is calculated and passed up to the next layer as a prediction error.
Maybe I’m just pattern matching, but I couldn’t help but notice the parallels between the sensory stream and the object wave from optical holography. It would remain to be demonstrated that the prediction stream is somehow equivalent to the reference wave, but it seems intuitive to me that the interference pattern between the two waves could be regarded as prediction error – and if predictions match perceptions, lights out…
Westlake (1970) described the object wave as a wave of neural spiking emanating from sensory object source points, with timing that depends on the distance traveled. To help illustrate this, there’s a recent paper with some great animations, exploring whether traveling waves of neural activity can be used to model sense impressions. From Traveling Waves Integrate Spatial Information Into Spectral Representations (Jacobs et al., 2025):
Figure 1: Sequence of hidden states of an oscillator model trained to segment images of polygons. We see that the shape of the hexagon is visible throughout the wave dynamics, and that waves propagate differently within the polygon due to reflections induced by the differing natural frequencies.
Proposing a candidate reference wave is more challenging. In order to facilitate associative recall, it would need to continually modify itself in response to the current contents of awareness. One thing that has long been understood is that familiarity between percepts and memories in psychologically uniform spaces respects an approximately exponential distance metric. From Toward a Universal Law of Generalization for Psychological Science (Shepard, 1987):
Fig. 1. Twelve gradients of generalization. Measures of generalization between stimuli are plotted against distances between corresponding points in the psychological space that renders the relation most nearly monotonic.
Presumably associative recall only works when the distance between the reference wave and the memory as it is encoded on the holographic plate is minimised. Perhaps if we continue to study how memory works within the brain we’ll find some neural structure which behaves the same way as what we observe using psychophysics.
In the meantime, I’ll continue to use altered states to inform my idle speculation. Cannabis is well known to have a negative effect on both memory recall and formation. Personally, I find that cannabis adds noise to my bodymind – what if it’s just making my reference wave noisy? I also observe that DMT improves my memory recall abilities. What might that be doing to my reference wave? What might iboga do?
I also asked Andrés Gómez Emilsson what he thought the equivalent of the reference and object waves could be, and how we might experience them. He proposed that they might correspond to the self and other:
Rob Burbea, in his discussions about “The Absolute” and dependent co-arising, concludes that at the absolutely most fundamental, every moment of experience consists of “self, other, and time”.
I think self and other are associated to the origin and ending of light paths, whereas time is the result of the interference patterns along the way. Thus, in the holographic analogy, the reference wave is the self: the internally generated scaffold of expectation and coherence. The object wave is the other: the contingent, the impinging, the surprise. Their interference is what we call time; not just clock time, but the sense of unfolding, the changing felt texture of a moment as new structure accretes onto what was already there. The interference pattern as a whole is the experience. The interference pattern has an internal temporal ordering. Especially if boosted with electrical current and electron interactions.
Just a guess.
Intuitively, perhaps think also of how the sense of time needs information. Light paths need to arrive off-phase. In a sense, the path light goes through needs to shear aspects of you so they can notice the rest (notice what’s off-phase).
The self and the other can be point-like, but also line-like or even fuzzy. In general, high concentration might make the self and other more coherent and point-like in general.
Deep in meditation it feels that self is a standing wave pattern that represents native biorythms and is used as a frame of reference, whereas the world is much more alien and general standing wave that’s highly programmable and flexible. And time connects the two. Like a path integral from one to the other. Perhaps when the self is hyperdimensional, then the path integral is more like quantum field theory or string theory, since you have a worldsheet (not our definition) or a brane that goes from one configuration to another through all possible paths in between.
This question is slightly more philosophical. If the brain operates on holographic principles, is there a Cartesian theater into which the hologram is projected? If so, would this constitute the neural correlate of phenomenal consciousness? I collected a number of proposals:
Pockett (2017) thinks that consciousness is identical with certain spatiotemporal patterns in the electromagnetic field, specifically those in primary sensory cortices.
Ward and Guevara (2022) agree with Pockett, but propose the thalamus as the most likely candidate.
Gómez-Emilsson and Percy (2023) agree with Pockett, but propose that there could be billions of topological pockets in the electromagnetic field pervading the nervous system and body, though only one bounds a field that encloses (and hence integrates) electromagnetic activity emerging from the brain’s memory modules.
Pressman (2024) proposes that consciousness is holographically encoded on the active inference boundary.
I also asked the vision researcher and phenomenologist Steven Lehar if he thought that the world simulation was rendered in the thalamus:
Yes, and there’s a reflection of it in the cortex. You see, there’s a near one-to-one projection, like a piece of broccoli… And it’s also in nodules throughout the body… Everywhere there’s a node, there’s a holographic representation!
I wonder what Michael Levin would make of this question.
Figure 25. A: An opaque white sphere illuminated by coherent light generates coherent wave fronts propagating outward in spherical shells from the surface. The expanding rays are reflected back inward by C: a phase conjugate mirror that creates spherical wave fronts converging on a focal point. A portion of these rays is reflected by B: a half-silvered mirror, to create D: a virtual sphere about a center of spherical symmetry. If this process could be contrived to occur inside a phase conjugate mirror, it would project an outward-propagating reconstruction of the whole sphere, as suggested by the faint concentric arcs.
I was reading about holography late last year around the same time as I was studying how the transformer architecture underlying large language models works. Around this time, Andrés Gómez Emilsson – who does not believe that computers are conscious – published a tweet forcing himself to construct an argument supporting the idea that language models might be conscious. If this was the case, perhaps this could be as a specific consequence of the transformer architecture:
If tasked to steelman the case for LLM consciousness within a computationalist paradigm, I would highlight how the transformer architecture (particularly its attention mechanisms) captures an essential feature of consciousness that earlier AI architectures missed entirely.
The key insight centers on self-attention: this mechanism allows every token to dynamically contextualize every other token in its context window. This creates a dense web of mutual influence where tokens don’t just sequentially affect each other, but participate in a holistic “all-to-all” relationship. This mirrors how phenomenal objects in your visual field aren’t processed in isolation; they exist in a state of reflective equilibrium with each other, their meanings interdependent and mutually determined.
In a really awesome way, transformers instantiate something akin to Indra’s net: each token both reflects and is reflected by every other token (with word embeddings modifying each other through each iteration), not infinitely but through several layers of processing and attention. Those who would dismiss transformer-based LLMs as mere “stochastic parrots” tend to totally miss this deep architectural parallel to consciousness. The fact that this mechanism doesn’t actually solve the phenomenal binding problem of consciousness is… far from obvious.
Well, I figured I’d ask the models themselves if they saw any equivalency between the procedures of training a transformer model and recording a hologram. ChatGPT o1 gave a carefully worded treatise on why this idea did not hold water, while Claude Sonnet 3.5 thought it was an intriguing analogy – though the common objection which they both raised was that while holograms record phase relationships, transformer models work with static embeddings rather than oscillatory wave-like phenomena.
Still though, I thought the Indra’s net comparison made sense, and o1 did acknowlege some similarities:
In a hologram, the whole is encoded in each part. Similarly, in a transformer model, the internal representation is highly distributed. There is a sense in which the model’s parameters collectively represent the learned relationships across the training data. While a transformer doesn’t encode “the entire dataset in every parameter”, the learned representations are diffuse.
I’m not a machine learning researcher by any measure – but could the analogy be extended to alignment, or at least internal alignment? When working with language models, could we construct some equivalent of an energy parameter which modifies the self-attention process such that self-awareness is maximised? Would this be meaningfully analogous to what happens in humans?
