July 24, 2025 : Issue #95

WONDERCABINET : Lawrence Weschler’s Fortnightly Compendium of the Miscellaneous Diverse

WELCOME

Walter Murch goes on another tear, this time about saccadic cinema, and what it portends that every one of us goes blind every time we move our eyeballs.

Yet Another Spinout with Walter Murch

Several of you will recall how just a few issues back, here, in the lead-up to the imminent release of his manual-manifesto-memoir Suddenly Something Clicked, the legendary polymathic film and sound editor (and longtime friend of this Cabinet) Walter Murch allowed us to preview his chapter on the spliceosome, which is to say the uncanny way that the manner in which dividing cells edit their own DNA as they reproduce virtually mirrors the step-by-step process through which film itself goes from dailies through to the mass-released final product—or is it vice versa?

Well, in celebration of the fact that the book in question, Suddenly Something Clicked, has itself now finally arrived (it saw its American release earlier this week), Murch is allowing us to sample yet another chapter, this one on the implications of the fact that all of us go momentarily blind every time we move our eyeballs, even though we somehow remain blissfully unaware of this fact. How is that even possible? Answering that question led a spate of researchers to uncover the further fact that, as a result, we only see, or rather become aware of what we are seeing, roughly a twentieth of a second behind the actual occurrence of whatever it is that we are seeing (our experience of the present is thus endlessly delayed, even if only slightly). That twentieth-of-a-second delay in turn has everything to do with how cinema, famously pitched to twenty-four frames a second, came into being in the first place and how it in turn gets experienced to this day.

This time out, in the interests of community bandwidth, we have condensed his argument somewhat though we are including the entirety of the chapter on our Google drive, here . We will also be indicating wherever we have made cuts by a […] marker, so all you have to do if you want to know more is reference the wider piece. That wider version is also festooned with citations and footnotes and for that matter with QR codes slotted in the margin that link to a whole range of supplementary video snippets. So do give it a look—and for that matter, do yourself a big favor and track down the whole book: the thing is brimming over with gobsmacking chapters like these two, and others steeped in the love and lore of the cinematic history of the last hundred years, so disconcertingly much of which has been Walter’s own.

And by the way, to reiterate, for those of you in Los Angeles, Walter and I will be in conversation on the evening of Tuesday August 12th in the Billy Wilder Auditorium at the Hammer Museum in Westwood. The event will be free, though advance reservations are strongly advised. For more, see here.

Meanwhile, there’s this:

* * *

The Main Event

WALTER MURCH ON SACCADIC CINEMA

or The Uncanny Persistence of the Persistence of Vision

I begin this chapter with three outrageous facts:

(1) You are blind every time you move your eyeballs.
(2) You experience reality approximately 120 milliseconds (three film frames) after it has happened.
(3) You are not aware of either of these facts.

I will use these strange but scientifically well-established phenomena to urge the final abandonment of the so-called retinal persistence of vision, which is often used (still!), two hundred years after it was first proposed, as an explanation for why we see motion when we watch a motion picture – which is, after all, just a series of still images.

Using the attributes of the saccade – the jump of the eyeball from one focal point to another – I hope to provide a satisfying replacement for retinal persistence. Cutting to the chase, it will amount to this:

The neurology of saccades, which evolved over hundreds of millions of years of vertebrate sight to smooth out the shifts of attention that happen during the sudden movement of eyeballs, was hijacked and put to use when motion pictures were invented.

[…]

Roget, Plateau and their audiences at the time would have been familiar with eighteenth- and nineteenth-century magic-lantern shows, where the lantern would be fitted with two lenses and two slides so that the operator could make a dissolve between one image and the other. It was inferred that on the lantern-show screen, as on the retina, there was a persistence of vision from one image to the next, and that this dissolve between two slightly different images was interpreted by the observer as motion from one frame to the next. This illusion of motion, alternating quickly between two slightly different slides, was performed in many of the magic-lantern shows, as early as the eighteenth century, to great audience-pleasing effect.

Roget’s interpretation of retinal persistence was accepted throughout the nineteenth century as the explanation for the illusion of motion in animation devices, and then motion photography when it was perfected in the early 1890s. Amazingly, it is often mentioned even now and can be found on current websites, such as Adobe’s:

As with other motion simulation inventions, the illusions created by the zoetrope rely on the human retina [my italics] retaining an image for roughly a tenth of a second. If a new image appears in that time, the brain merges them and the sequence appears continuous.

Of course, there is such a thing as retinal persistence of vision: when the photoreceptors in your retina are over-stimulated with a sudden flash of bright light, they will continue to generate a current for a few seconds even after you close your eyes, just as your ears will continue to ring after a sudden loud gunshot. But as we all know, it is an imprecise, ‘fuzzy’ image, whose length of persistence depends on the intensity of the original flash.

Nude Descending a Staircase, No. 2, Marcel Duchamp, 1912

When we watch a film, we do not see a cascade of fuzzy, overlapping images piling on top of each other in a visual smear, which is what retinal persistence would imply – something like Marcel Duchamp’s 1912 painting Nude Descending a Staircase, No. 2 or Étienne-Jules Marey’s chronographs.

Chronograph by Étienne-Jules Marey, 1886.

Instead, the frames are individually distinct, the motion between them is precise and the illusion of motion is not dependent upon the intensity of the projector’s light.

*

Enter the Beta Effect

In 1912 (the same year that Duchamp first exhibited his Nude Descending a Staircase, No. 2) Max Wertheimer, one of the founders of the Gestalt school of psychology, published ‘Experimental Studies on the Seeing of Motion’. He proposed a new name for the illusion of apparent motion, which he called the beta effect, moving its origin out of the retina and deeper into the neurological circuitry of the brain. He did not explain what was going on in the brain nor where.

It is not sufficient to draw upon pure peripheral [i.e. retinal] processes in relation to a single eye: we must have recourse to processes which lie behind the retina [i.e. in the processing brain].

{Over the next several pages, Murch provides a detailed history of the gradual sequence of compounding observations and experiments by the likes of Hugo Münsterberg, Joseph and Barbara Anderson, and presently (as late as 1976) and most crucially John Ross that came to flesh out Wertheimer’s initial intuition. — LW}

[…]

What can we conclude from all this? That there must be the equivalent of a RAM cache somewhere in the brain – probably in the visual cortex, located in the occipital lobe at the back of the brain – which preserves a pixel-accurate record of everything we have seen over the previous 50 ms [milliseconds]. This cache is readily available when there is a need to compare those stored images with something currently being seen. In the interests of brevity, clarity and authorship I am going to call this hypothesised cache the Ross cache.

This is literally persistence of vision. But it is not retinal persistence; it is something lying much deeper in the brain, as both Wertheimer and Münsterberg intuited. Much deeper and much, much more accurate than retinal persistence, and limited to a storage capacity of 50 ms. That is one twentieth of a second, and the mathematically inclined reader will have already compared this to the twenty-four- or twenty-five-frame-per-second speed of cinema. Each frame of a film is accordingly held for 41.67 ms (at 24 fps in the US) or 40 ms (at 25 fps in Europe), a shorter time than the hypothesised 50 ms Ross cache. {FN}

{FN} This is related to, but different from, the visual phenomenon of iconic memory. Iconic memory lasts for several hundred milliseconds, instead of a definite 50 ms for the Ross cache. Both caches are probably located in the occipital lobe, particularly the primary visual cortex, which is heavily involved in the initial processing and temporary storage of visual information.

This means that a frame of film and the previous one can be in this cache at the same time, ready to be compared with each other and for the fruits of that comparison to be sent to our conscious awareness. Just as in Ross’s experiment, where the fruitful comparison between two images separated in space resulted in depth, in this case the comparison of two images separated in time yields movement. Both the depth and the movement are apparent; both are creations of the mind.

Ross’s conclusion:

It appears that there are records of visual input that can be consulted before anything at all is seen in order to determine the proper framework for perception. What these records are, how many separate repositories there are and what their physical basis is we do not know.

And here he ties his observations about binocular perception together with the subject of this chapter: the illusion of apparent motion.

A number of puzzling visual effects such as apparent motion can be brought together by the assumption that perception must wait on the analysis of independent visual records before we are able to perceive.

At twenty frames per second, each image is held on screen for 50 ms, which is at the limit of the Ross cache. This means that at frame rates slower than 20 fps, with a longer duration for each frame, there will be many moments when there is only one image in the cache, and consequently no ability to compare it with a subsequent frame in order to synthesise motion between them. As a result, perceived motion begins to stagger at frame rates slower than 20 fps.

Comparison with Sound

Just as images begin to stagger below 20 fps, we also begin to lose the ability to hear tonality in music below 20 hertz (cycles per second). We still ‘hear’ sounds below this frequency, but we feel each pulse as a separate impact rather than the tonality we hear above 20 Hz. The parallels between the lack of smooth motion below twenty frames per second and the lack of tonality below twenty cycles per second are intriguing.

You might say that tonality is to sound as motion is to image.

A further comparison can be made between the 50 ms Ross cache for vision and the 50 ms Haas effect for sound. The Haas effect (also known as the precedence effect) says that if two nearly identical sounds are played in quick succession, with less than 50 ms between the leading edge of each one, the listener will hear a single sound with a slightly ‘off-mic’ quality. If the separation between leading edges is greater than 50 ms, the listener will hear two separate sounds, in a distinct echo effect.

But why and how does the synthesis between two still images yield motion? I believe the explanation lies in the sudden movements of the eye known as saccades.

*

Saccades

These sudden, mostly involuntary movements of the eyeball are quick (20–200 ms) and common: we experience on average three saccades every second, for a daily total of well over 150,000. {FN} They are particularly frequent when we are reading, with our attention jumping from phrase to phrase, but they are happening all the time, almost always below our conscious awareness.

{FN} The actual number, for sixteen waking hours, would be 172,800. In an uncanny convergence, this is exactly the number of frames in a two-hour film. Saccades of a sort persist when we are asleep, but now they are called REMs (Rapid Eye Movements). REMs occur during periods of intense dreaming, and it is natural to assume that they are saccades freed from the constraints of external reality. The images of dreams are constructed by us, and we can freely ‘pan and scan’ those images, like shifting our attention across the various parts of the film frame.

It is in exploring saccades that we encounter the first of the three ‘outrages’ listed at the beginning of this chapter: When you move your eyeballs, you are blind. The reason for this is that the motion of saccades is so rapid (around 1,800 arc seconds per millisecond) that it far outstrips the response time of even the fastest photoreceptor cells. {FN}

{FN} It takes cones 20 ms to respond to light: https://tinyurl.com/mrsndwpr. But the ‘dwell time’ of a point of light on the average photoreceptor during the sweep of a saccade is around 20 microseconds, a thousand times slower than the response time of the fastest cone cells. If we could see what the retina ‘sees’ during a saccade, it would be a horizontal smear of different tonal values and colours from the scene in front of us, but with no detail of any kind – like a swish pan in cinema.

This is a remarkable fact, so I will repeat: the moment our eyes start to move in a saccade, the degraded signals coming from the retina are blocked until our eyes come to rest again, perhaps 120 ms (three film frames) later. Consequently, we do not see these blurred images, nor do we see any evidence of their removal.

A vivid demonstration of this is as close as your nearest mirror. Stand about five inches in front of it and ask a friend to watch the goings-on, perhaps making a video at the same time. Now look at your left eye for three seconds, and then suddenly, without moving your head, look at your right eye. What you will experience is . . . nothing, no change. Now look back at your left eye. You will also experience no change. It just seems to you that you have been looking at yourself for six seconds or so, with no movement of your eyeballs. What your friend sees, and what the video will show, however, are your eyeballs moving from left to right and back again. Your visual system has sneakily edited out the movement of your eyeballs and concealed the fact of that edit. This process has a name: saccadic masking.

So you are effectively blind for the 120 ms that you moved your eyes. This fact is of great use to magicians and masters of three-card monte, whose con artistry is to get you to move your eyeballs at the exact same moment that they quickly perform their tricks, which consequently are invisible to you.

We may be effectively blind, but we do see something during a saccade. What is it? And how does our visual system perform the magic of providing us with vision during a saccade? It accomplishes this by taking the last clear image before the start of the saccade, comparing it to first clear image after the saccade, constructing a plausible connection between the two, and then editing that construct into the gap created by saccadic masking. This is exactly Münsterberg’s and Wertheimer’s intuition about the origin of motion between still frames of a movie. As Münsterberg wrote, ‘The movement is not really seen from without, but is superadded, by the action of the mind, to motionless pictures.’

The mirror experiment is dramatic, but there is not much movement in it. So here’s another one, this time with movement:

Sit down and hold your hands out, palms upwards, about a foot away from your eyes and separated by about fifteen inches. Stare straight ahead, wiggle the fingers of both hands and now look at your left hand without moving your head. You see your left fingers wiggling, but you can also see, peripherally, your right fingers wiggling. Now saccade your eyes to look at your right hand, also without moving your head. You see your right hand’s fingers wiggling and, peripherally, you can still see your left hand’s fingers wiggling.

The saccade you just performed lasted around 200 ms (five film frames), during which you were blind, but nonetheless you saw your fingers continuing to wiggle perfectly naturally. Those five frames of wiggling motion were created by your visual cortex, which compared information captured from just before and just after the saccade, and then made best guesses about what would most probably link the two.

This is the result of extremely sophisticated brain software, to say the least, similar but superior to the digital cinema applications that can expand old films shot at 16 fps to 48 fps by making best guesses, using artificial intelligence, about how to interpolate the missing two frames between every one of the original frames. (Peter Jackson’s film They Shall Not Grow Old [2018] is a vivid example of the artistic possibilities of this technology.)

A reasonable question from the reader: But this mental software you are describing can only construct the missing information after it has registered the first clear image at the end of the saccade. And yet our experience of reality is that it moves continuously forwards. This seems to cycle backwards – to add information retrospectively.

An answer: Our experience of reality is forwards, as you say. But recall the second ‘outrage’ at the beginning of the chapter: We are living in the past. More exactly: our experience of reality is of something that has already happened approximately 120 ms previously.

*

Yes, We Live in the Past . . . but Only the Slightly Past

Snap your fingers and you experience a satisfying synchronicity of image, sound and touch. You see, hear and feel the snap at the same moment. And those three things do happen together, but by the time you feel/see/hear the snap, it has already occurred some time earlier. What causes this?

The signals of sight, sound and touch arrive at different times to our central sensory processing unit, the thalamus, located near the centre of our brains. Hearing arrives first, as it is the quickest of our senses; then touch; then sight, which is the slowest because it is the most complex and requires the most processing. The thalamus waits for them all to arrive, puts them in sync, after which they are processed further, and only then finally delivered to our awareness. All of this processing takes time – about 120 ms on average. And this is approximately how much lag there is between reality and our perception of it. The three senses have been put in sync with each other, but they are no longer in sync with the events that triggered them.

David Eagleman, professor of neuroscience at Baylor University, has this to say about our predicament:

The strange consequence of all this is that you live in the past. By the time you think the moment occurs, it’s already long gone. To synchronise the incoming information from the senses, the cost is that our conscious awareness lags behind the physical world. That’s the unbridgeable gap between an event occurring and your experience of it.

This lag gives our visual system the time to generate the interpolated images that will replace the blurred ones suppressed by saccadic masking, supplying them to our awareness just before our lagging consciousness arrives at the first moment of suppression.

[…]

*

Fast Reaction Times

You can easily imagine the difficulty that sensory latency creates for athletes who play tennis, baseball, cricket, football and other ‘fast-ball’ sports. Pitching and bowling speeds in professional baseball and cricket can exceed 100 miles an hour, and aggressively fast serves in tennis can attain 150 mph. At that speed, a tennis ball travels approximately 18 feet in 80 ms, so by the time the receiver ‘sees’ it consciously, the ball is no longer where it appears to be.

One of the solutions to this problem is that athletes can apparently learn to bypass sophisticated consciousness and rely on instinctual ‘knee jerk’ reflex arc responses processed in the spinal cord, which are many times faster than ‘conscious’ perception routed through the brain – think of how we instinctively yank our hand away from unexpected contact with a hot stove before we are even aware of its heat. Also, after years of experience, athletes become expert at making predictions about where the ball might be, even though they may not be able to ‘see’ it in the normal sense of the word. As tennis player Ben Shelton said at the 2023 US Open, ‘Sometimes you just have to shut off the brain, close your eyes, and swing.’

Musicians also go through similar ‘bypassing of consciousness’ training, and it is essential in Formula One racing, as well as other activities where reaction times need to be as short and ‘predictive’ as possible.

[…]

*

Conclusion: Cinema Is Saccadic

The analogy between cinema and saccades was especially precise in cinema’s first mechanical super-century (1894–2004). Every camera and projector had to have a shutter to mask the blurring that would occur when the film was rapidly jerked from one frame to the next. And this shutter performed the same job that saccadic masking does when it prevents us from seeing the blurry result of the rapid jerking of the eyeball during a saccade. {FN}

{FN} Birds are an interesting example of vertebrates whose eyes do not move in their sockets, so they do not have saccades in the human sense. But they compensate by moving their heads in quick ‘saccade-like’ jerks, familiarly stereotyped as being characteristically ‘bird-like’.

This ‘behind the curtain’ operation of the saccade is so extremely close to what happens when we watch a film that a saccadic explanation for why we see motion in motion pictures seems inescapable, circling us back to the objective expressed at the beginning of this chapter:

The intricate neurology of vertebrate sight, which evolved over hundreds of millions of years to deal, in part, with the rapid eye motion of saccades, was simply hijacked and immediately put to use when motion pictures were invented 190 years ago.

So persistence of vision has persisted, after all, but in a much more sophisticated form than retinal persistence: sophisticated both in its pinpoint accuracy, as Ross has demonstrated, as well as in its crucial role in enabling our visual system to create the illusion of movement, which is how the brain resolves the differences between one image and the next. Whether it is the intermittent nature of the movement of our eye during saccades or the intermittent nature of the modern motion picture film, the process is the same. {FN}

{FN} What I have called the ‘Ross cache’ for simplicity’s sake is actually a multilayered part of the visual cortex known as extrastriate visual areas V1 to V5. Specific neurons in regions like V5 are tuned to detect motion. These neurons specialise in comparing changes in position between adjacent frames, effectively ‘stitching’ together the differences between still images to create the perception of motion. While there isn’t a literal ‘frame storehouse’, as implied by the term ‘Ross cache’, the visual cortex and interconnected areas do maintain a dynamic, continuously updated sequence of visual ‘snapshots’ of everything that has been seen in the last 50 ms.

* * *

ANIMAL MITCHELL

Cartoons by David Stanford, from the Animal Mitchell archive

animalmitchellpublications@gmail.com

***

OR, IF YOU WOULD PREFER TO MAKE A ONE-TIME DONATION, CLICK HERE.

*

Thank you for giving Wondercabinet some of your reading time! We welcome not only your public comments (button above), but also any feedback you may care to send us directly: weschlerswondercabinet@gmail.com.

Here’s a shortcut to the COMPLETE WONDERCABINET ARCHIVE.