In Sci-Fi TV shows such as Westworld, it can be difficult to tell the difference between androids and humans. According to new research done at UC Berkeley, the human brain takes less than a second to distinguish between fantasy and reality in real life. The study’s results show that humans are wired visually to take in information extremely fast and make a snap judgment about what’s real or not.
The scientists have discovered a visual mechanism they call “ensemble lifelikeness perception”. This mechanism determines how we recognize groups of objects and people in virtual or artificial, and real worlds.
Allison Yamanashi Leib, a postdoctoral scholar in psychology at UC Berkeley and study lead author, notes that the unique visual mechanism needs just 250 milliseconds to allow us to perceive what’s alive and what’s simulated. The same mechanism guides us to define the overall level of activity in a scene.
It has long been assumed by vision scientists that humans need to consider multiple details carefully before they can judge if an object or person is lifelike. David Whitney, a UC Berkeley psychology professor and study senior author, explains that the participants in the study made animacy decisions without conscious deliberation. They also agreed on what was lifelike and what was not.
Whitney finds it surprising that we immediately share in our impressions of lifelikeness, even without deliberating about it together, or talking about it. He adds that study participants used ensemble perception and, without focusing on all the individual details, were able to make snap judgments about the liveliness of groups of people or objects, or even entire scenes.
Leib pointed out that shoppers, tourists and other party goers all use visual cues processed through ensemble perception in real life to gauge where the action is.
Whitney added that if humans did not possess the ability to determine lifelikeness fast, our world would be very confusing, and every object, person, or animal we see would appear to be equally alive.
A total of 68 healthy adults with normal vision were used in 12 separate experiments conducted by the scientists.
In the majority of tests, participants were asked to view up to a dozen images of random people, animals and objects. The images included a hockey player, an ice cream sundae, a toy car carrying toy passengers, a caterpillar, a guinea pig wearing a shirt, a statue of a wooly mammoth and more.
After quickly viewing groups of images, participants rated them on a scale of 1 to 10 according to their average lifelikeness. Participants were able to assess the average lifelikeness of the groups accurately, even if they were displayed for less than 250 milliseconds.
To test participants’ memory for details in another experiment, researchers flashed selected images and then showed participants ones that they had seen, as well as ones they had not.
The results indicated that while participants often forgot many details, their “ensemble perception” of what had been lifelike remained sharp. Whitney suggests that this shows that the visual system is better with abstract global impressions such as lifelikeness at the expense of the fine details. Although we don’t perceive the trees, we do notice the forest and how alive it is.