Can computers learn how to do this? Or, can machines learn to learn how to think and act like people?
It turns out that humans do have some control over how they perceive and interpret reality. In an ideal world, all objects in the world would behave in ways that made sense to the human. But in reality, human behaviors are subject to random, adaptive forces from the world around us. Most of the time, we don’t even recognize ourselves in other people or objects. When we see another human being, we might not be conscious of it, or we may not be conscious of what they’re thinking or experiencing. We’re, essentially, seeing something through someone else’s eyes — so it’s hard to know what to make of our observations.
One experiment with people helped confirm the idea that human behavior is subject to selection:
In a recent study, two psychologists, James Jost and Kevin Harrington, asked college undergraduates to view three different scenes. In the first scene, a man is sitting on a sofa in a dark room. To him, the scene looks normal. When the two are confronted together, though, the man becomes aware of the man next to him. After he does, however, the sight becomes very unsettling. In the second scene — the room where the man actually has dinner with a woman — it’s the exact opposite: The woman sits on a couch in the same room, the man sitting beside her. At first it seems that the two have no relation — neither knows who the other is. But after a while, the man becomes aware of her and starts to act differently. He has trouble focusing on the woman in front of him, and when asked how he felt after watching her, he begins to fidget. Then the experimenter enters the room and the man looks away in horror. The experiments show that people do have control over what they see. This ability seems to emerge in the face of a very strong, adaptive force.
The study also suggests that human behavior isn’t always random and adaptive:
In the current study, the two psychologists gave the people the same scene and asked them to rate it. They then went back to the scene and asked the same people to rate it, and they noticed that the ratings were more consistent in that the one-minute ratings were more consistent with the actual scene than the 15-minute or 40-minute ratings.
This study adds weight to a theory that seems particularly powerful in the age of big data and