Most famous psychology demonstrations did not survive the replication crisis. The invisible gorilla did. Simons and Chabris 1999 inattentional blindness held up across hundreds of studies, including a 2013 radiologist replication with an 83% miss rate. The construct is robust; some of the popularizations are not.
If you have read through this hub, you have watched canonical psychology findings dissolve under inspection. Power posing collapsed under its co-author’s own recantation. Ego depletion failed preregistered replication. Bargh’s elderly-priming evaporated. The marshmallow test shrank to a confound for socioeconomic status. The bystander effect’s founding case study turned out to be largely journalistic invention. Most of the famous demonstrations that found their way into TED talks, business books, and undergraduate textbooks did not survive contact with the methodological audit of the last fifteen years.
So when you watch the original “Gorillas in our midst” video --- the one where six players in white and black shirts pass basketballs while a person in a gorilla costume walks through the middle of the frame --- it is reasonable to ask whether this is another classroom demonstration that will turn out to be a methodological artifact, or whether this is one of the few that genuinely holds up.
This one holds up. The basic phenomenon of inattentional blindness --- that observers focused on a primary task can fail to notice unexpected events occurring in plain sight --- has replicated in hundreds of studies, across populations from undergraduates to expert radiologists, across stimulus modalities from video to static images to real-world environments, and across decades of methodological refinement. The 1999 Simons and Chabris paper itself reported a roughly 50% failure rate to notice the gorilla among observers counting basketball passes; the 2013 Drew, Vo, and Wolfe extension to expert radiologists reading CT scans reported an 83% failure rate; and the underlying construct continues to generate productive research on attention, perception, and applied safety problems in 2026.
This article treats the invisible gorilla as an anti-example in this hub’s terms: a finding so well-replicated and mechanistically grounded that it serves as a calibration anchor for evaluating the rest of cognitive psychology. The construct survived. Some of the popular framings around it --- “this proves you cannot multitask,” “this proves you miss the obvious,” “this proves your perception is unreliable” --- are stronger than the data supports. The underlying phenomenon is real and applied implications are genuine, but the marketing language outran the science in the usual way.
Simons and Chabris 1999 --- The Original Study
The canonical citation is Simons, D. J., and Chabris, C. F. (1999). “Gorillas in our midst: Sustained inattentional blindness for dynamic events.” Perception, 28(9), 1059—1074. DOI: 10.1068/p281059.
The study built directly on earlier inattentional-blindness work by Mack and Rock, summarized in their 1998 MIT Press book Inattentional Blindness. Mack and Rock had shown that observers focused on a static primary task --- judging the longer arm of a cross presented at fixation --- frequently failed to notice unexpected stimuli (a small shape) appearing nearby. The Mack and Rock paradigm established the phenomenon for brief, single-trial, fixation-anchored displays. What Simons and Chabris did was demonstrate that the same kind of attentional failure occurs for sustained, dynamic, naturalistic events that the observer is looking directly at for many seconds.
The basic design was elegant. Subjects watched a 75-second video in which two teams of three players each --- one team in white shirts, one team in black shirts --- passed two basketballs among themselves while moving through a constrained space. The instruction was to count the number of passes made by one of the teams. Approximately 45 seconds into the video, a person in a gorilla costume walked into the scene from the right, stopped in the middle of the players, faced the camera, thumped its chest, and walked off to the left. The gorilla was visible on screen for approximately 9 seconds.
After watching, subjects were asked whether they had noticed anything unusual. The headline result was that approximately 46% of subjects --- nearly half --- reported not noticing the gorilla at all. Among the subset of subjects counting passes by the white team (the harder attentional condition because of the visual conflict between the gorilla’s dark form and the black-team players the white-team counters had to filter out), the miss rate was as high as 58%. Among subjects counting passes by the black team, the miss rate dropped to around 27%, consistent with the gorilla being more visually similar to the attended set.
The paper reported four experiments with systematic manipulations. Experiment 1 compared “easy” (counting bounce passes only) versus “hard” (counting bounce and aerial passes separately) task difficulty, finding that harder primary tasks produced more inattentional blindness. Experiment 2 compared “transparent” versus “opaque” video conditions (transparent: the players and gorilla were filmed separately and overlaid via video transparency; opaque: all filmed in one take). Opaque conditions produced more blindness, which is theoretically interesting because the opaque condition is the more naturalistic one --- the version of the world we actually live in. Experiments 3 and 4 introduced an unexpected woman with an umbrella as an alternative unexpected event, ruling out the possibility that the effect depended on something specific about gorillas.
Several methodological features of the original study are worth highlighting because they distinguish it from the failed replications elsewhere in this hub.
The effect size was enormous. A 46% miss rate is not a marginal d ≈ 0.2 effect that disappears under preregistered replication. It is a binary failure rate where nearly half of all observers, watching a video they were paying attention to, failed to notice a person in a gorilla costume on screen for nine seconds. The signal-to-noise ratio is so high that no plausible statistical-flexibility critique can explain it away.
The phenomenon is self-replicating. You can run this experiment on yourself. You can run it on a friend who has not seen the video. You can run it on a classroom of students. The effect appears reliably. The original video is freely available online, and several variant videos with different unexpected events have been produced to allow second showings to naive observers. The accessibility of the demonstration is in stark contrast to the elaborate experimental setups required to detect (or fail to detect) the typical social-priming finding.
The theoretical framework was mature. The Simons and Chabris paper did not announce a new phenomenon out of nowhere. It extended a well-established line of inattentional-blindness research going back to Mack and Rock in the 1970s and 1980s, and ultimately to even earlier work on selective attention by Neisser, Treisman, and others. The 1999 paper was a refinement and extension of a paradigm that had already been replicated, not a novel claim awaiting confirmation.
Pre-registration and replication followed. The Simons and Chabris paradigm has been replicated in dozens of subsequent studies with variations in task difficulty, unexpected-event type, observer population, and stimulus presentation method. The basic phenomenon survives all of these variations, with the specific miss rate varying as a function of primary-task difficulty and unexpected-event salience in theoretically predictable ways.
The 1999 paper has accumulated approximately 3,800 citations as of 2026, putting it in the top echelon of cognitive psychology papers. More importantly, the citation pattern is not the typical “famous finding cited and then forgotten” pattern --- the paper continues to be actively cited in current research on attention, eye tracking, applied safety, and clinical perception.
Robust Replication --- Hundreds of Studies, Stable Construct
The literature that grew up around inattentional blindness after 1999 is large and broadly consistent. Beyond direct replications of the gorilla paradigm itself, the construct has been investigated using a wide variety of methods, all of which converge on the same basic conclusion: under conditions of focused attention to a primary task, observers reliably fail to detect unexpected events in their visual field, even when those events are large, in plain view, and would otherwise be perceived effortlessly.
A few of the major extensions and confirmations are worth flagging.
Simons and Levin 1998 --- Real-world change blindness. Before the gorilla paper, Simons and Levin had already published one of the most striking demonstrations in this literature: Simons, D. J., and Levin, D. T. (1998). “Failure to detect changes to people during a real-world interaction.” Psychonomic Bulletin and Review, 5(4), 644—649. DOI: 10.3758/BF03208840.
In this study, an experimenter approached pedestrians on a college campus to ask for directions. While the pedestrian and the first experimenter were talking, two confederates carrying a large door walked between them, briefly occluding the pedestrian’s view of the first experimenter. During the occlusion, the first experimenter swapped places with a second experimenter --- a different person, different clothes, different voice --- who continued the conversation. Approximately 50% of pedestrians failed to notice the switch. The Door Study, as it became known, demonstrated change blindness in a richly naturalistic setting and ruled out the possibility that the basic inattentional-blindness phenomenon was an artifact of laboratory video presentation.
Most 2001 and the basic Mack-Rock paradigm. A series of careful replications and extensions by Steven Most and colleagues, including Most, S. B., Simons, D. J., Scholl, B. J., et al. (2001). “How not to be seen: The contribution of similarity and selective ignoring to sustained inattentional blindness.” Psychological Science, 12(1), 9—17, established that the miss rate scales predictably with the visual similarity of the unexpected event to the attended set. Observers tracking white objects are most likely to miss black unexpected events; observers tracking black objects are most likely to miss white unexpected events. The pattern is consistent with current attentional-set theories: focused attention on certain visual features (color, shape, motion) actively suppresses processing of other features, with the suppression being strongest for the most dissimilar items.
Memmert 2006 --- Eye tracking the gorilla. A frequent objection to the inattentional-blindness literature is that observers might simply be looking somewhere else when the unexpected event occurs --- the failure could be a failure to direct gaze, not a failure of attention while looking. Memmert addressed this directly with eye-tracking data: Memmert, D. (2006). “The effects of eye movements, age, and expertise on inattentional blindness.” Consciousness and Cognition, 15(3), 620—627. DOI: 10.1016/j.concog.2006.01.001. Observers who failed to detect the gorilla spent, on average, approximately 1000 milliseconds with their gaze directly on the gorilla. The failure was not a failure to look. The failure was a failure to perceive what was being looked at, when attention was committed to a competing task.
This is the strongest demonstration in the literature that inattentional blindness is genuinely a perceptual phenomenon and not an artifact of measurement, eye-movement avoidance, or reporting bias. Subjects’ gaze landed on the gorilla; they simply did not see it as a gorilla, because their attention was elsewhere.
Cross-cultural and cross-population replication. The basic inattentional-blindness finding has been replicated in many countries, across age groups from young children to elderly adults, in clinical populations, in expert populations, and in cross-cultural contexts. The size of the miss rate varies as a function of the specific stimulus and the specific task, but the basic phenomenon --- that focused attention produces detection failures for unexpected events --- has not failed to replicate in any major study.
Meta-analytic and review consolidation. The literature has accumulated to the point that multiple review and meta-analytic treatments now exist. Chabris and Simons’ 2010 popular book The Invisible Gorilla synthesizes the academic literature for a general audience; Simons’ subsequent academic reviews update the field. The construct is treated as established in current cognitive-psychology textbooks and in applied attention research.
The picture, in short, is the opposite of what you would see for a failing finding. Effect sizes have not shrunk over decades; the methodology has gotten tighter without weakening the core result; extensions and variants have replicated; eye-tracking has ruled out trivial measurement explanations; and the applied uses have been adopted by practitioners with results that hold up in the field.
Drew, Vo, and Wolfe 2013 --- Radiologists Miss a Gorilla on a CT Scan
The most theoretically and practically important extension of the original Simons-Chabris paradigm is Drew, T., Vo, M. L. H., and Wolfe, J. M. (2013). “The invisible gorilla strikes again: Sustained inattentional blindness in expert observers.” Psychological Science, 24(9), 1848—1853. DOI: 10.1177/0956797613479386.
The setup was deceptively simple. The researchers took chest CT scans --- standard medical-imaging cases that radiologists examine routinely to detect lung nodules --- and digitally inserted an image of a gorilla into one of the slices. The gorilla was approximately 48 times the size of the average nodule that radiologists were searching for. It was clearly visible if you looked at the relevant slice. It was, by the standards of normal visual perception, impossible to miss.
Twenty-four experienced radiologists were asked to read the CT scans for nodules. Of the 24 radiologists, 20 (83%) failed to notice the gorilla. Eye-tracking data showed that the radiologists who missed the gorilla had, on average, looked directly at the gorilla --- their gaze landed on the relevant area of the image. They simply did not see it, because they were attending to lung-nodule features, not to gorilla-shaped artifacts.
The implications of this study are substantial. Inattentional blindness is not a quirk of undergraduates watching a video. It is a robust feature of focused expert attention, surviving in the population that is specifically trained to attend carefully to medical images for the purpose of detecting unexpected abnormalities. The pattern of results suggests that expertise does not protect against inattentional blindness for unexpected events --- if anything, the expertise narrows the attentional set further and may make detection of off-template events more difficult, not less.
The Drew, Vo, and Wolfe study was widely cited and has had a real influence on applied radiology training and on the design of imaging workflows. Modern thinking in radiology emphasizes structured search strategies, the use of computer-aided detection as a backstop for human attention, and explicit awareness of inattentional-blindness risks in high-stakes imaging contexts. The applied uptake is one of the strongest validators of the underlying finding: practitioners adopted protocols based on the construct because the construct kept producing real results in real clinical work.
The study has been replicated and extended. Drew, T., Vo, M. L. H., Olwal, A., Jacobson, F., Seltzer, S. E., and Wolfe, J. M. (2013). “Scanners and drillers: Characterizing expert visual search through volumetric images.” Journal of Vision, 13(10), 1—13 characterized the visual search patterns of expert radiologists in more detail, and several subsequent studies in 2014—2024 have investigated specific clinical contexts where inattentional-blindness risks appear --- mammography, chest radiography, dermatoscopy, and others. In each context the basic finding has held: focused expert attention produces detection failures for off-template unexpected events at non-trivial rates.
Change Blindness --- The Sibling Phenomenon
Closely related to inattentional blindness is change blindness, the failure to detect changes between two visual scenes when the change occurs during a brief visual interruption (a blink, a saccade, a flicker, an occlusion). The two phenomena are conceptually distinct --- inattentional blindness is about failure to perceive an unexpected event in real time, while change blindness is about failure to compare a current scene against a remembered scene --- but they emerge from related limits on attention and visual memory and have been investigated together in the literature.
The change-blindness literature is at least as well-replicated as the inattentional-blindness literature. The basic flicker paradigm of Rensink, O’Regan, and Clark in the 1990s and 2000s, in which observers compare two alternating images separated by a brief blank screen, reliably produces detection failures for surprisingly large changes --- a building moves, an object disappears, a person’s clothing changes color --- for many seconds of viewing. The 1998 Simons and Levin door study cited above is the real-world version of change blindness. The two literatures together establish that the failures of attention demonstrated by the gorilla paradigm are not narrow or fragile findings; they are part of a broader, well-characterized set of limits on visual perception.
This matters for the anti-example framing: the invisible gorilla is not an isolated demonstration that happens to replicate. It is one finding within a coherent and well-replicated research program on the limits of visual attention. The robustness of the broader program is part of what makes the specific gorilla finding credible.
Applied Implications --- Medical, Aviation, Security, Driving
The inattentional-blindness construct has produced applied implications in several high-stakes domains. The basic logic is the same in each case: when humans perform a focused primary task in a safety-critical environment, they are vulnerable to missing unexpected events that fall outside their attentional set, even when those events are in plain view and would be obvious to a fresh observer. Practical countermeasures involve structured search protocols, computer-aided detection systems, redundant human observers with different attentional sets, and explicit training about the limits of focused attention.
Medical imaging. The Drew radiologist study is the canonical applied example, and the response in radiology has been substantive. Modern protocols emphasize structured search patterns (e.g., the “U” or “S” pattern for chest radiography), the use of double-reading by independent observers for high-stakes cases (mammography programs in many countries require two-radiologist reads), the integration of computer-aided detection as a backstop, and explicit training about the risks of focused-attention errors. Mammography programs that adopted double-reading have reported measurable improvements in detection rates that are consistent with what an inattentional-blindness analysis would predict.
Aviation. Pilots and air traffic controllers face inattentional-blindness risks routinely. The classic case studies in the aviation human-factors literature involve pilots focused on landing checklists or instrument readings failing to notice traffic, terrain, or runway obstructions in plain view. Crew resource management protocols, structured callouts, and traffic-collision-avoidance systems are all in part responses to the underlying attentional vulnerability. The construct also informs the design of cockpit displays --- there is a long human-factors literature on how to present salient warnings in ways that are detectable even when the pilot’s attention is committed elsewhere.
Security screening. Airport baggage screeners face one of the canonical inattentional-blindness scenarios: focused, sustained attention to a primary task (detecting prohibited items) over long shifts, with very low base rates of targets. The vigilance literature predicts and the inattentional-blindness literature confirms that detection rates decline under these conditions, especially for unusual or unexpected target appearances. The applied response includes computer-aided screening systems, rotation of screener attention, and explicit insertion of fictitious test targets to maintain attentional engagement.
Driving. Distracted-driving research draws heavily on the inattentional-blindness literature. The basic finding --- that drivers focused on a phone conversation or a navigation display can fail to notice pedestrians, cyclists, or other vehicles in plain view --- is well-replicated and has informed legal restrictions on phone use while driving, hands-free vs. handheld policy debates, and the design of in-vehicle infotainment systems. The recent literature on autonomous-vehicle handoffs --- where a human driver must reassume control after a period of automated driving --- is a contemporary application of the same attentional dynamics.
In each of these domains the underlying logic is the same and the practical countermeasures rhyme: assume focused attention will produce inattentional-blindness errors at non-trivial rates; design systems with redundancy, structured search, and machine-assisted backstops; train humans explicitly on the limits of their own perception; and resist the intuition that attentive experts will reliably notice obvious things.
Where the Popularization Overstates --- A Necessary Caveat
Like every famous psychology finding, the invisible gorilla has produced popularizations that go beyond what the data supports. The construct is robust; the popular framings around it are sometimes stronger than the underlying science.
“This proves you cannot multitask.” The inattentional-blindness literature does not directly establish this. What it establishes is that focused attention on a primary task produces detection failures for unexpected events. This is consistent with, but not equivalent to, claims about multitasking. There is a separate literature on multitasking and task switching, with its own methodology and findings, and the conclusions of that literature --- including some genuinely strong findings about task-switching costs --- should be sourced to that literature, not to the gorilla study. Using inattentional blindness to argue against multitasking is borrowing the rhetorical weight of one finding for the conclusions of a different finding.
“This proves your perception is fundamentally unreliable.” The construct establishes a specific kind of perceptual failure under specific conditions: when attention is committed to a primary task and an unexpected, unattended event occurs. It does not establish that perception is unreliable in general. Perception of attended events under reasonable conditions is, in fact, highly reliable; the gorilla finding works precisely because we expect perception to be reliable, which is why the failure is striking. The popular framing that “you cannot trust what you see” is too strong; the accurate framing is that “you should not trust that focused attention will pick up unattended events.”
“If you focus on X, you will miss everything else.” This is a stronger version of the previous overstatement. The actual finding is that some unexpected events are missed by some observers some of the time, with the rate depending on task difficulty, event salience, and the similarity of the event to the attended set. A 46% miss rate is not a 100% miss rate. The popular framing that focused attention produces total blindness to unattended events overstates a graded effect into a binary one.
“Eyewitness testimony is fundamentally unreliable because of this finding.” Inattentional-blindness research does have implications for eyewitness reliability, but those implications are more nuanced than the popular framing suggests. The construct establishes that witnesses can fail to perceive events that they were not actively attending to. It does not establish that all witnessed events are unreliable, that focused witnesses cannot reliably perceive what they were attending to, or that any specific high-profile case can be dismissed on inattentional-blindness grounds. The eyewitness-reliability literature is its own substantial body of work, with its own well-replicated findings (some of which the popular Loftus-style work has been challenged on methodological grounds); citing the gorilla finding as the foundation of broad eyewitness-skepticism conclusions is rhetorically powerful but evidentially mismatched.
“This is why people miss the obvious in business meetings, group decisions, etc.” This is a borrowing of the visual-attention finding into the domain of attention to information in conversational or decision-making contexts. There is a separate and much weaker literature on these analogical applications, and the inferential leap from a 9-second video gorilla to a business meeting blind spot is not warranted by the original data. The construct can plausibly be motivating for the broader claim, but the broader claim should be sourced to the literature on managerial perception, motivated reasoning, and information processing in groups --- not to a basketball-passing video.
The discipline here matters because the credibility of the underlying construct is one of the more durable cognitive-psychology findings available. Overselling it into adjacent domains where the evidence is weaker is the same pattern that produced the replication crisis for less well-grounded findings. The strategy that protects the credibility of the strong findings is to cite them precisely for what they actually show, and to source adjacent claims to the appropriate adjacent literatures.
What This Anti-Example Tells Us
Compare the inattentional-blindness construct’s profile against the profiles of the canonical replication failures in this hub, and the same pattern as for the Stroop effect emerges.
The original effect size was large. A 46% miss rate is not a marginal d ≈ 0.2 effect. It is a binary detection failure where nearly half of subjects, looking at a video they were paying attention to, failed to see a gorilla on screen for nine seconds.
The phenomenon is self-replicating and demonstrable. You can show the original Simons-Chabris video to a naive observer right now and observe the effect. You can run variants on yourself, friends, students, or colleagues. The accessibility of the demonstration is in stark contrast to the elaborate setups required for many failed social-psychology findings.
The mechanism converges across methods. The inattentional-blindness construct is supported by behavioral data, eye-tracking data showing that observers were looking directly at the missed events, neuroimaging evidence about the role of attention in modulating early visual processing, and clinical evidence from patients with attention disorders. The mechanism story is grounded in well-established theories of selective attention going back to Treisman, Broadbent, and others.
Extensions and variants replicate. The original gorilla paradigm has been extended to umbrella-carrying women, real-world door-swap studies, expert radiologists, airport security contexts, driving contexts, and many other variants. Each extension preserves the basic phenomenon while varying the specific stimulus, task, or population. Findings that survive this kind of variation are usually real; findings that work only in one specific setup are usually fragile.
Effect sizes have been stable across decades. The 46% miss rate from the original 1999 study is similar in magnitude to the miss rates reported in subsequent replications of the same paradigm. The 83% miss rate in the Drew radiologist study is in line with what an inattentional-blindness analysis would predict for that population and task. The effect has not shrunk over time as samples have grown and methodology has tightened, which is one of the strongest indicators of original-finding robustness.
Applied uses survive practical scrutiny. Radiology protocols, aviation crew resource management, security screening procedures, and driving-safety regulations have all incorporated implications from the inattentional-blindness literature, and those applications have produced measurable improvements in real-world outcomes. Practitioners would have abandoned an unreliable construct decades ago; instead the construct continues to inform applied work in 2026.
The decision-useful conclusion is that the inattentional-blindness construct is in the same reliability tier as the Stroop effect, the default effect, and the other findings that have survived the replication crisis intact. The popular framings around it sometimes overstate the implications, but the underlying construct is one of the most robustly established findings in cognitive psychology.
Strategist Implications --- What to Take Away
For an executive evaluating attention-related claims --- in safety-critical work, in human-factors design, in security operations, or in the more analogical “what is your organization not seeing” framings --- the invisible gorilla is a useful anchor for what a real attention-failure finding looks like.
Distinguish the construct from the marketing. When a consultant or vendor cites “the invisible gorilla” to motivate a claim, ask whether the claim is about a perceptual attention failure under focused primary-task conditions (within the scope of the construct) or about a broader organizational, cultural, or cognitive blind-spot phenomenon (analogical use that may or may not be supported by other literatures). The construct is robust for the narrow claim; the broader claims need their own evidence.
Take the safety-critical applied implications seriously. Inattentional blindness is a real and quantifiable risk in radiology, aviation, security screening, and driving. Practitioners in these fields have adopted protocols based on the construct, and the protocols work. If your organization operates in any of these high-stakes attention-dependent domains, the inattentional-blindness literature has direct, validated implications for how you design workflows, training, and oversight.
Resist the “you cannot multitask” overstatement. The construct establishes inattentional blindness for unexpected events under focused-attention conditions. It does not establish a general claim about the impossibility of multitasking or the unreliability of attention more broadly. There is a separate literature on those questions, with its own findings (some of which are robust and some of which are not), and conclusions in that domain should be sourced to that literature.
Use eye-tracking-equivalent evidence as a credibility marker. The Memmert eye-tracking demonstration that observers who missed the gorilla were looking directly at it is one of the strongest pieces of evidence in this literature, because it rules out the trivial measurement explanation. When evaluating other perception or attention findings, look for analogous evidence that rules out trivial explanations. Findings that survive this kind of stress test are more credible than findings whose proponents have not done the work.
Treat applied uptake as confirmation. When radiology programs adopt structured search protocols, when aviation regulators mandate crew resource management, when driving-safety researchers establish phone-use restrictions, all on the basis of an underlying behavioral-science construct, that adoption is informative. Practitioners with skin in the game and decades of opportunity to discard ineffective interventions have kept the protocols, which is itself a strong validation of the underlying construct.
Be skeptical of “this proves the obvious thing I already believed” applications. The invisible-gorilla finding gets cited for an enormous range of claims about organizational blind spots, market-research insights, leadership blind spots, and other domains far removed from the original 9-second video. Many of these analogical applications are post-hoc justifications for conclusions the speaker reached on other grounds. The construct is genuinely strong for the narrow finding; it is weaker when stretched to fit any conclusion the user wants to motivate.
The most useful frame, for an executive trying to calibrate behavioral-science claims, is to put the invisible gorilla in the same tier as the Stroop effect and the default effect: a robust, well-replicated finding with genuine applied implications, surrounded by a fringe of popularized overstatements that should be evaluated separately. The underlying construct will still be in the textbook in 2075. Some of the management-book framings will not.
Sources
Simons, D. J., & Chabris, C. F. (1999). Gorillas in our midst: Sustained inattentional blindness for dynamic events. Perception, 28(9), 1059—1074. DOI: 10.1068/p281059.
Mack, A., & Rock, I. (1998). Inattentional Blindness. Cambridge, MA: MIT Press.
Chabris, C. F., & Simons, D. J. (2010). The Invisible Gorilla: How Our Intuitions Deceive Us. New York: Crown.
Drew, T., Vo, M. L. H., & Wolfe, J. M. (2013). The invisible gorilla strikes again: Sustained inattentional blindness in expert observers. Psychological Science, 24(9), 1848—1853. DOI: 10.1177/0956797613479386.
Simons, D. J., & Levin, D. T. (1998). Failure to detect changes to people during a real-world interaction. Psychonomic Bulletin & Review, 5(4), 644—649. DOI: 10.3758/BF03208840.
Most, S. B., Simons, D. J., Scholl, B. J., Jimenez, R., Clifford, E., & Chabris, C. F. (2001). How not to be seen: The contribution of similarity and selective ignoring to sustained inattentional blindness. Psychological Science, 12(1), 9—17. DOI: 10.1111/1467-9280.00303.
Memmert, D. (2006). The effects of eye movements, age, and expertise on inattentional blindness. Consciousness and Cognition, 15(3), 620—627. DOI: 10.1016/j.concog.2006.01.001.
Related
- The Stroop Effect: Cognitive Psychology’s Most-Replicated Finding --- The other major cognitive-psychology anti-example in this hub. Like the invisible gorilla, the Stroop effect has held up across decades of methodological refinement and serves as a calibration anchor for evaluating other claims.
- Mere Exposure Effect --- A well-replicated social-cognition finding with a similar profile: robust construct, broad applied uptake, occasional popularization overstatement.
- Availability Heuristic --- A foundational Tversky-and-Kahneman finding that has held up far better than most heuristics-and-biases applied claims.
- Confirmation Bias --- A robust finding with a different profile: well-replicated as a phenomenon but contested at the mechanism level.
- Halo Effect --- A century-old social-perception finding with a profile somewhere between the cognitive-psychology anti-examples and the failed social-psychology cluster.
FAQ
Did the invisible gorilla study actually replicate?
Yes, robustly. The basic Simons-Chabris paradigm has been replicated in dozens of subsequent studies with variations in task difficulty, unexpected-event type, observer population, and stimulus presentation method. The specific miss rate varies as a function of primary-task difficulty (harder tasks produce more blindness) and unexpected-event salience (more similar events are easier to detect), in theoretically predictable ways. The basic phenomenon --- that focused attention produces detection failures for unexpected events --- has not failed to replicate in any major study, and the eye-tracking work of Memmert and others has ruled out the trivial explanation that observers were simply not looking at the unexpected events.
Why did the radiologists miss the gorilla on the CT scan?
Because expertise narrows attention to the specific features being searched for (lung nodules), and the gorilla --- while large and visually striking to a naive observer --- did not match the perceptual template the radiologists were searching for. The eye-tracking data showed that the radiologists who missed the gorilla did, on average, fixate the relevant area of the image; the failure was perceptual, not visual. The study is the canonical demonstration that expertise does not protect against inattentional blindness for off-template unexpected events, and it has had a real influence on radiology training and on the design of computer-aided detection systems.
Does the invisible gorilla prove that I cannot multitask?
No. The inattentional-blindness literature establishes that focused attention on a primary task produces detection failures for unexpected unattended events. This is related to but distinct from claims about multitasking, which have their own literature with their own methodology. There are genuinely robust findings about task-switching costs and about the difficulty of dividing attention across complex tasks, but those findings come from a different research program. Using the invisible gorilla to argue against multitasking is borrowing rhetorical weight from one finding for the conclusions of another.
How is inattentional blindness different from change blindness?
Inattentional blindness is the failure to perceive an unexpected event occurring in real time while attention is committed to a primary task --- you do not notice the gorilla walk through the basketball game because you are counting passes. Change blindness is the failure to detect changes between two visual scenes when a brief interruption (a blink, a flicker, an occlusion) prevents direct visual comparison --- you do not notice that the building in the background moved between two pictures because the flicker between them disrupted your motion perception. The two phenomena are conceptually distinct but related: both reflect the limits of attention and visual memory, and both are well-replicated. The 1998 Simons-Levin door study is a real-world demonstration of change blindness.
What are the applied implications I should actually take seriously?
The strongest applied implications are in safety-critical attention-dependent domains: medical imaging (especially radiology, mammography, and dermatoscopy, where double-reading and computer-aided detection have been adopted in part because of inattentional-blindness considerations), aviation (where crew resource management and structured callouts address the same underlying vulnerability), security screening (where computer-aided systems and rotation of attention are standard practice), and driving (where distracted-driving research and phone-use restrictions are partially grounded in the same construct). These are domains where practitioners have adopted protocols based on the construct, and the protocols have produced measurable improvements in real-world outcomes. The broader analogical applications --- “this is why your organization misses opportunities,” “this is why leaders have blind spots” --- are weaker and should be sourced to literatures on managerial perception, organizational learning, or whatever the relevant domain is, not to the basketball video.
Is there a version of the experiment I can run on myself?
Yes. The original Simons-Chabris gorilla video is freely available online, and several variant videos with different unexpected events have been produced to allow second showings to naive observers (since you cannot show the same video to the same person twice and get the same effect). You can also run a simpler version of the Mack-Rock paradigm: have a friend perform a focused visual task (judging the longer arm of a cross presented briefly) and insert an unexpected stimulus on one trial. The basic phenomenon will appear. The accessibility of the demonstration --- that you can verify it yourself in a few minutes --- is one of the features that distinguishes a robust cognitive finding from a fragile one.
Is the invisible gorilla finding ever challenged in the academic literature?
There are productive theoretical debates about the precise mechanisms of inattentional blindness (the relative contribution of attentional set, perceptual load, working memory limits, etc.), about boundary conditions (how event salience and task difficulty interact), and about specific applied questions (the magnitude of inattentional-blindness risk in particular real-world contexts). But the basic phenomenon --- that focused attention produces detection failures for unexpected events at meaningful rates --- is not in serious dispute and has not been since the original 1999 paper. The construct is one of the few major cognitive-psychology findings that you can confidently expect to still be in the textbook in 2075.
What is the practical takeaway for someone evaluating behavioral-science claims?
Use the invisible gorilla --- along with the Stroop effect --- as a calibration anchor for what a robust cognitive finding looks like. Large original effect size, easy self-replication, mechanism convergence across methods (behavioral, eye-tracking, neural), successful extensions and variants, stable effect sizes across decades, and applied uses that have been adopted by practitioners and have survived practical scrutiny. Findings that look like the invisible gorilla on these dimensions are worth betting on. Findings that look like power posing or elderly priming are not. And in either case, be careful to separate the underlying construct from the popularized framings around it --- the construct may be solid even when the marketing language outruns the science.