Ubiquitous Surveillance Cameras Are Changing Our Understanding of Human Behavior

Surveillance footage is providing new insights into how humans interact in public. But should scientists be able to see it?

Ever since the infamous murder of Kitty Genovese in 1964, it’s been a common assumption that bystanders are unlikely to intervene in a public attack if they witness it as part of a group. But that assumption is now in question—ironically, due to the ubiquitous presence of surveillance cameras in cities around the world.

In most modern city centers, no person or patch of sidewalk goes unwatched. Fixed CCTV cameras dot the urban landscape—more than one for every 14 people in London, by one industry estimate—and police are swiftly increasing their surveillance arsenals with drones and body-worn cameras.


Watchdogs like Tony Porter, the U.K.’s surveillance camera commissioner, warn that governments’ increasing ability to watch everything all the time will lead to both predictable and unforeseen invasions of privacy. But there may be a small silver lining to the constant eye of the camera: it has the potential to open up a new world of social science research and test common assumptions about human behavior.

There are some questions behavioral scientists would really like to answer for which traditional data-gathering techniques are highly ineffective. For example: if you’re attacked in public, how likely is it that a bystander will come to your aid? If you come to someone else’s aid, what’s the likelihood that you will be injured yourself? Simulated situations and questionnaires are poor substitutes for real-world evidence.

But a group of British and Danish researchers took a major step toward answering those questions in a recently published study, using a trove of government CCTV footage from cities in England, Denmark, and South Africa. They determined that in nine out of 10 cases, a bystander intervened to stop a public attack, and the rate of intervention was consistent across cultures.

The results offer a strong counterpoint to the “bystander effect,” which arose from research spurred by the Genovese case. Many of those studies drew data from controlled simulations, interviews, questionnaires, and other traditional data-gathering techniques.


The new study concluded that while any given individual in a crowd witnessing an attack may feel the bystander effect, the presence of more people increases the chance of an intervention and appears to provide support to individuals who do consider stepping in.

“There are a lot of questions that could be addressed with video and I think that not many researchers are aware of the massive amounts of video recordings that are available of all kinds of interactions,” Lasse Liebst, a co-author of the CCTV study and a sociologist at the University of Copenhagen, told Motherboard. “It has a great potential but it’s painfully under-utilized.”

“This is not an argument in favor of expanding the level of surveillance,” he added. “But this technology is out there now, so could this technology give back and benefit the wider public? There is potential, and as researchers I think it’s important that we ask ourselves how we can help the public to regain control over it.”

So far, the use of CCTV footage for behavioral science has been limited, experts say, and primarily employed by academics in Europe and Australia. Researchers must win access to the footage from governments and private companies, a process that can be complicated by bureaucracy and potentially limit research areas. Police, for example, might be less willing to hand over a CCTV database for a study into how officers react to bystander intervention.


Camera footage has drawn more attention—positive and negative—in the field of behavioral medicine, where researchers see wearable devices as potential goldmines for elusive data, according to Camille Nebeker, director of ReCODE Health, a University of California San Diego program that studies ethical practices in digital health research. As with violence studies, academics interested in topics like exercise and dietary habits run into frequent problems with the data returned by questionnaires and other types of sensors.

A 2016 study Nebeker co-authored found 30 studies on PubFacts that drew data from SenseCam, a body-worn camera that can capture as many as 3,000 images daily (As of Jan. 15, PubFacts showed 55 studies using SenseCam). The majority of participants in the study reported overall positive experiences wearing the devices, although nearly a third said they felt uncomfortable, particularly when they were around other people.

“My colleagues want to better understand how people behave in the wild and how we can capture every-day behavior without asking people to self report … [because] that’s a very unreliable way of collecting data,” Nebeker told Motherboard. “What I’m seeing is just a very new landscape and regulations that are not ready.”

Study participants aren’t always digitally literate enough to understand the wide scope of data a camera will collect. Researchers, meanwhile, have to contend with difficult questions, like what to do if a participant’s camera captures them committing an act that must be reported to authorities, especially if the researchers won’t be viewing that footage until months after it was captured.


And while federal rules offer strong informed consent and privacy protections for human participants in studies, the standards are not the same for non-participants who are likely to be caught on camera.

Nebeker said she knows of at least one colleague who has had approval of multiple camera-based studies delayed by an Institutional Review Board (IRB), which must approve research proposals at universities and many companies, pending more research into the technique.

“I don’t think researchers are deterred,” she told Motherboard. “But I do think IRBs are challenged.”

Liebst and his colleagues went to great lengths to anonymize the subjects captured in the CCTV footage they used and secure the data, which was collected before Europe’s General Data Protection Regulation (GDPR) went into effect. He told Motherboard that their methods would likely pass the GDPR’s stricter privacy rules.

The results of the study and follow-up research, Liebst hopes, will lead to improved public education and policy that could potentially make city spaces safer.

Some privacy advocates, though, worry researchers’ altruistic intentions will ultimately be used to bolster surveillance states.

“It’s a tough question,” Cooper Quintin, a senior technologist at the Electronic Frontier Foundation, told Motherboard, adding that he wouldn’t necessarily support regulations that restrict researchers’ access to CCTV footage.

“I think we need to be careful to what degree we normalize and even benefit from the surveillance apparatus,” Quintin said. “Any time we normalize and benefit from an institution it becomes a little more entrenched, little by little, each day. By making use of the data, you are supporting it.”