Cause and effect

Having put it down a few years ago, yesterday I picked up Jakob Hohwy's important book on the predictive mind. All about how the mind is a hierarchically organised prediction engine based on Bayesian principles, the evolution of which must have started aeons before Bayes was around. With part of the idea being that you see what you expect to see. So if, for example, you see something rustling in the bushes by the side of your path through a tiger infested jungle, your brain is apt to turn that something into a fully fledged tiger. Or, in the jargon of the snap left, if your prior expectation (p(H)) is high, your posterior (p(H|D)), your evidence based reasoning, is apt to be high too.

One of the examples he adduces being the clouds in the sky, which seen from below are rather flat and two dimensional, while seen from above from an aeroplane are full of interest. For Hohwy, this is an example of the brain knowing that the upper half of the visual field is usually not very interesting and so not trying very hard. Perhaps a consequence of our having spent a lot of evolutionary time wandering about the savannahs of Africa under a bright blue, but rather featureless sky. Much more important to keep an eye on all the dangerous bugs slithering around on the ground.

For me, there are at least three complications. First, my understanding is that, perhaps because of said wandering about the savannahs, there are more receptive cells in the upper retina, corresponding to the lower part of the visual field, that there are in the lower retina. So one is getting a stronger signal from the lower part of the visual field. Second, I would have thought that the clouds viewed from above, from an aeroplane, are generally much closer than those viewed from the ground. Furthermore, one is moving quite fast relative to the clouds and the brain is able to extract lots of information from the changes in the signals from the clouds arising from that movement. Third, the side of the clouds one is looking at is being illuminated by the sun, giving rise to plenty of shadows for the brain to work on.

None of which blocks the Hohwy example, but which does make it more complicated than might at first appear. An illustration of the care needed when drawing conclusions from observations.

No doubt someone somewhere has devised a natty experiment that demonstrates that the effects of my three complications are indeed amplified by prior expectations, as per Hohwy.

Maybe someone else can find the source of the snap. Both Bing and Google clearly know about it, but I have failed to track it to its source, despite the clue  bottom right. With the most entertaining result being a more or less pornographic offering from pinsdaddy.

Reference 1: The predictive  mind – Jakob Hohwy – 2013. Page 33 in my copy.

Reference 2: Bing knows all about this sort of thing, turning up another example at https://www.ucl.ac.uk/swc/about/street-displays/illusion.