Change blindness, movie magic, and vision

A disappearing head wound is one of more than 400 continuity errors in The Birds. Entire plane engines vanish in The Usual Suspects, a wrecked Porsche is made pristine after Sully is thrown from a cliff in Commando, and The Itchy & Scratchy Show features acoustically implausible skeleton instruments. But for as common as they are, most continuity errors slip by unnoticed—even when they happen in real life (and/or situations contrived by brain scientists):

When changes are obscured—by film edits, physical barriers, or just because we’re looking the other way when they happen—we rarely notice them, a phenomenon called change blindness. People are “blind” to changes in everything from words and letters to real-world scenes. For example, try to spot what’s changing in this image:

Spot the change
Spot the change

William James mentioned the perceptual quirk of change blindness all the way back in 1890, and it’s well-known enough that filmmakers rely on it, as I’ll describe later. But it’s also paradoxical, because our conscious experience of vision is rich and detailed, and our memories seem vivid and precise. We don’t feel change-blind—once you spot the change in that gif, it seems impossible you’d missed it earlier—so how can we reconcile the richness of our experience with the “blind spot” in our abilities?

From one perspective, you can’t reconcile them: change blindness is evidence that our experience of detailed visual memory is a fiction. If we can’t spot changes by memory—if we have to be looking right at the engine as it disappears to notice it—then maybe the brain doesn’t ever “store” those details at all. Maybe the details of visual memories are “offloaded” to the environment, not recalled from in our head but “filled in” by information from out there in the world, the same way we can write down a grocery list rather than memorize it (change blindness researcher Kevin O’ Regan calls this outside memory). In short, by this view, change blindness reveals the true sparseness of our memories.

To others, the idea that visual memory lacks detail is a red herring. For one thing, memory is accurate and almost limitless: people can remember up to ten thousand pictures, including specific visual details like their orientation (also noteworthy: the dedicated and/or well-compensated participants who willingly studied 10,000 pictures). And when people look at changing scenes like that airplane gif, their eyes linger near the changing part seconds before they consciously recognize it, suggesting that changes may actually be identified subconsciously. Change blindness by this view is not the result of “flawed” or undetailed memory, but is more like a laboratory quirk elicited in special circumstances.

• • •

rotated sine-wave grating
rotated sine-wave grating

A recent study helps reconcile the contradicting views, and reveals how change blindness may serve a valuable purpose. In the study, participants viewed a series of rotated sine-wave gratings, presented one at a time, and tried to match the angle of each grating. The important finding was that their performance was influenced by recently-seen gratings. For example, if the last grating they saw was 30° and the current one was 60°, they underestimated the second angle, even though it was no longer visible (and vice versa).

In fact, responses were biased towards the average of the preceding three gratings, corresponding to roughly 15 seconds in real time. In short, their perception of what they were looking at was “pulled” towards, or merged with, what they’d recently seen—up to fifteen seconds of the past blending with the present (a finding now replicated with faces).

Afterimages are common, so the idea that past perceptions and events can influence current ones isn’t novel. But this finding—called either “perceptual serial dependence” or the “continuity field”, depending on your jargon preference—is broader, subtler, and more fundamental than afterimages. Afterimages happen only sometimes and are plainly obvious, like when we’re seeing spots after staring at the sun. The continuity field, in contrast, suggests that what we see is virtually always a composite of the “now” and the “recent”, even when we don’t realize it: that perceptually, the “right now” isn’t cleaved entirely from what came before.

Understanding the continuity field allows for a different perspective on change blindness. Our intuition is that spotting a change means comparing distinct memories of “before” and “after”; that our perceptions of events are discrete, independent, and temporally segregated like frames of a movie. But in reality, our perceptions have blurred edges rather than rigid boundaries, and there aren’t just snapshots at time x and time y that we can compare. If we view changes as getting merged into the continuity field, it means they’re getting averaged out and blended in, and “standing out” less than we think. In some sense, change blindness doesn’t reveal that our memories are undetailed — it mostly reveals how our intuitions about perception are inaccurate.

But what purpose does the continuity field serve? What’s the benefit of merging things together? One simple answer is stability. We move our eyes three times or more every second, only get high-resolution visual data from the point of focus, and don’t process visual input during blinks or eye movements. Thus, incoming visual information reaches the brain not as the continuous stream we perceive it as, but instead as a disjointed and hyperkinetic ping-pong of gaps, blank spots, ephemeral shifting details, and out-of-focus blobs, like the bad adrenochrome trip in Fear and Loathing.

Despite all that discontinuity, we (effortlessly) perceive a stable and richly-detailed visual world. Blending old and new information likely helps bridge that gap between input and perception: noisy data is smoothed out, minor changes are minimized, gaps are filled in with best guesses and predictions. The active, kinetic process of vision is translated into a stable, orderly percept; change blindness is the side effect of achieving that stability.


Another beneficial consequence of the continuity field is the magic of cinema. Stripped of context, the purely visual experience of an edited film is a discontinuous and incoherent sequence of images. Yet when a movie cuts to an exterior shot just as Liam Neeson bursts through a door, we “know” it’s the same time, place, and wise-cracking cop—even if it’s a CGI building, a stunt double, the scar on his face switched sides, and the blood on his shirt disappeared during the cut. Our brain conjures complex narratives from what should be jarring and nonsensical input.

To “understand” a movie, viewers must maintain a sense of permanence and coherence—both perceptual and semantic—across edits and scenes. Discontinuities must be stitched together into a coherent whole. The continuity field is the perceptual side of accomplishing that. Early filmmakers learned by trial-and-error what kinds of edits work (probably those that don’t disrupt the continuity field). Now they’re so in tune with human perceptual abilities that moviegoers are unlikely to even notice edits that are made according to fundamental editing rules. Weirdly, though, if it’s the perceptual blending of past and present that makes edited movies comprehensible, it’s simultaneously what makes continuity errors so difficult to spot — you can’t have one without the other.

Decades ago, John Huston compared blinks and film edits. He was right: both are brief disruptions that are mostly unnoticed. In fact, visual perception as a whole is film-like, even though it doesn’t seem that way: we constantly and surreptitiously cut to new angles and new “shots” when we blink and move our eyes. And just like we easily build a coherent narrative watching a movie, we usually build a coherent perceptual narrative, too.

• • •

Whatever else it is, change blindness is also a good reminder that our cognitive abilities—even basic perceptual ones—are molded by experience. If there weren’t a gray flicker separating the frames in that gif, we’d immediately spot the change, because the visual system automatically orients to quick changes in color, shadow, motion, and the like (the flicker obscures those cues). Our brains have spent a lifetime “learning” to be change-blind; learning that in this universe, objects don’t spontaneously leap into or out of existence, head wounds don’t vanish in a blink, and plane engines don’t teleport. And because the goal of perception is interacting with the world, not just accurately “representing” or recording it, we’ve all spent a lifetime learning to ignore those (im)possibilities.

That said, in our 3D world, four-dimensional objects would appear discontinuous, shape-shifting, and ephemeral—so change blindness may be the real enemy during a 4D alien invasion-themed Twilight Zone.

(Note: this post was updated from a prior version and cross-posted to Medium)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s