The gorilla test : why what's obvious is not what's obvious

Aeon may have the occasional duff article about Aztecs being a wonderful and charming people with a group morality we should emulate, but it still has just about the highest ratio of good/bad articles out there. This is definitely one of the good ones.

I've been thinking for a while that, contrary to what's often reported, the vast majority of the time human behaviour is actually very rational and logical. If it wasn't, people would be constantly stabbing themselves in the eye and walking into walls. They don't. Problems with logic mostly only occur with processes of higher reasoning, and while it's interesting to see that they can happen at a more basic level, the importance of this is prone to exaggeration. It would also, as the article points out, perhaps be more useful to study under what situations we perceive things more directly.

It’s a fact that most people who watch the clip miss the gorilla. But it does not necessarily follow that this illustrates – as both the study’s authors and Kahneman argue – that humans are ‘blind to the obvious’. A completely different interpretation of the gorilla experiment is possible.

Imagine you were asked to watch the clip again, but this time without receiving any instructions. You might report that you saw two teams passing a basketball. You are very likely to have observed the gorilla. But having noticed these things, you are unlikely to have simultaneously recorded any number of other things. The clip features a large number of other obvious things that one could potentially pay attention to and report: the total number of basketball passes, the overall gender or racial composition of the individuals passing the ball, the number of steps taken by the participants.

In short, the list of obvious things in the gorilla clip is extremely long. And that’s the problem: we might call it the fallacy of obviousness. There’s a fallacy of obviousness because all kinds of things are readily evident in the clip. But missing any one of these things isn’t a basis for saying that humans are blind. The experiment is set up in such a way that people miss the gorilla because they are distracted by counting basketball passes. Preoccupied with the task of counting, missing the gorilla is hardly surprising. In retrospect, the gorilla is prominent and obvious.

But the very notion of visual prominence or obviousness is extremely tricky to define scientifically, as one needs to consider relevance or, to put differently, obviousness to whom and for what purpose?

The alternative interpretation says that what people are looking for – rather than what people are merely looking at – determines what is obvious. Obviousness is not self-evident. Or as Sherlock Holmes said: ‘There is nothing more deceptive than an obvious fact.’

In other words, there is no neutral observation. The world doesn’t tell us what is relevant. Instead, it responds to questions. When looking and observing, we are usually directed toward something, toward answering specific questions or satisfying some curiosities or problems. ‘All observation must be for or against a point of view,’ is how Charles Darwin put it in 1861. Similarly, the art historian Ernst Gombrich in 1956 emphasised the role of the ‘beholder’s share’ in observation and perception

Quite simply, this human-centric and question-driven view of perception cannot be reconciled with Kahneman’s presumptions concerning obviousness, which make it a function of the actual characteristics of the things in front of us (their size, contrast or colour) without acknowledging the Suchbild, the cognitive orientation or nature of the perceiver.

In essence, Kahneman deems the gorilla to be obvious, but doesn't provide any reason as to why this should be obvious as opposed to any other aspect of the scene. I think he's right to point out that we don't realise our biases, however. But how you interpret data is, as I make it a point to tell students, something that happens in your brain and nowhere else. Data doesn't speak for itself.

The problem, as Sherlock Holmes put it, ‘lay in the fact of there being too much evidence. What was vital was overlaid and hidden by what was irrelevant.’ So, given the problem of too much evidence – again, think of all the things that are evident in the gorilla clip – humans try to hone in on what might be relevant for answering particular questions. We attend to what might be meaningful and useful.

Knowing what to observe, what might be relevant and what data to gather in the first place is not a computational task – it’s a human one. The present AI orthodoxy neglects the question- and theory-driven nature of observation and perception. The scientific method illustrates this well. And so does the history of science. After all, many of the most significant scientific discoveries resulted not from reams of data or large amounts of computational power, but from a question or theory.

Any number of apples, and other objects for that matter, had undoubtedly been observed to have fallen before Newton’s observation. But it was only with Newton’s question and theory that this mundane observation took on new relevance and meaning... Similarly, theories of a heliocentric Universe caused the ‘obvious’ observations of the Sun circling the Earth – or the retrograde loops of planets (as observed from Earth) – to take on completely new meaning and relevance.

Highlighting bias and blindness is certainly catchy and fun. And the argument that humans are blind to the obvious is admittedly far more memorable than an interpretation that simply says that humans respond to questions. But scientists’ own preoccupation with blindness risks driving the type of experiments scientists construct, and what they then observe, look for, focus on, and say. And looking for validations of blindness and bias, they are sure to find them.

Intelligence and rationality are more than just calculation or computation, and have more to do with the human ability to attend to and identify what is most relevant... The human ability to ask new questions, to generate hypotheses, and to identify and find novelty is unique and not programmable. No statistical procedure allows one to somehow see a mundane, taken-for-granted observation in a radically different and new way. That’s where humans come in.

Oh, I bet you could program a computer to generate new hypothesis, but not to invent concepts. Maybe you could do this one day if we understood how we do it ourselves, but at present it's a case of "show me one atom of justice", as a wise man once put it. I don't believe computers will be capable of that kind of thought for a long time yet, at best.

https://aeon.co/essays/are-humans-really-blind-to-the-gorilla-on-the-basketball-court?utm_medium=feed&utm_source=atom-feed