Expecting the unexpected

When the midcentury psychologists Jerome S Bruner and Leo Postman presented test subjects with brief views of playing cards, including some non-standard varieties – such as a red two of spades, or a black ace of diamonds – many people never called out the incongruities. They reported that they felt uneasy for some reason but often couldn’t identify why, even though it was literally right before their eyes.

So, crucially, some understanding of the expected signal usually exists prior to its detection: to be able to see, we must know what it is we’re looking for, and predict its appearance, which in turn influences the visual experience itself. The process of perception is thus a bit like a Cubist painting, a jumble of personal visual archetypes that the brain enlists from moment to moment to anticipate what our eyes are presenting to us, thereby elaborating a sort of visual theory. Without these patterns we are lost, adrift on a sea of chaos, with a deeply unsettling sense that we don’t know what we are looking at, yet with them we risk seeing only the familiar. How do we learn to see something that is truly new and unexpected?

Because of the complexity of both visual experience and scientific observation, it is clear that while seeing might be believing, it is also true that believing affects our understanding of what we see. The filter we bring to sensory experience is commonly known as cognitive bias, but in the context of a scientific observation it is called prior knowledge. To call it prior knowledge does not imply that we are certain it is true, only that we assume it is true in order to get to work making predictions.

If we make no prior assumptions, then we have no ground to stand on. The quicksand of radical and unbounded doubt opens beneath our feet and we sink, unable to gain purchase. We remain forever at the base of the sheer rock face of the world, unable to begin our climb. Yet, while we must start with prior knowledge we take as true, we must also remain open to surprise; else we can never learn anything new. In this sense, science is always Janus-headed, like the ancient Roman god of liminal spaces, looking simultaneously to the past and to the future. Learning is essentially about updating our biases, not eliminating them. We always need them to get started, but we also need them to be open to change, otherwise we would be unable to exploit the new vistas that our advancing technology opens to view... The iterative bootstrapping of learning-to-see, then seeing-to-learn, continues apace.

https://aeon.co/essays/seeing-is-not-simple-you-need-to-be-both-knowing-and-naive