Tuesday 10 November 2009

Dretske on 'what we see'

Long-time readers of this blog (as if!) will know I admire Dretske's 'Knowledge and the flow of information'. It turns out that as well as an excellent writer he's a great speaker, as I found from his 2007 Howison Lecture at UC Berkeley, on 'What we see' - embedded below.

He's arguing that we DO see the rich complexity of things, even though we don't realise it. One aspect of his line of argument is that you don't have to think about something actively to know it.


Q: Were there any giraffe's in your bedroom when you got up this morning?

A: You most likely know there weren't even though you didn't actively think about it.

This is interesting, and some other examples he gives - with some nice illustrations using complicated images - probe it further, but I'm not 100% convinced.

In the questions at the end, someone asks about the gorilla in the famous video:

(An earlier version of this idea here: http://viscog.beckman.illinois.edu/flashmovie/15.php)

Dretske argues that those of us who think we didn't see the gorilla, probably did know something about it, which could be revealed if we were asked the right question. Well, maybe, but surely if we'd seen it we would have been actively aware of it, something as out of place as that? It seems a bit of 'clincher' to me!

PS: Here's another one in the same vein:


Allan Jones said...

I'd never seen the 'gorilla' experiment (actually a bear in the clip here)until now but I'd heard a lot about. I'd imagined it as a demonstration of human fallibility, of lack of perception, of virtual 'blindness' in some circumstances.

Having seen it, I interpret it in almost a completely opposite way. I see it as a remarkable demonstration of the skill of human perception - of how it is possible to attend to complexity voluntarily, and of how it is possible to eliminate irrelevance at will when it is necessary to do so. I'd be worried if anyone did spot the bear/gorilla, or the setting changing in the spoof murder mystery. I'd say they had a perception problem.

David Chapman said...

Good point, I'd not thought of it like that before either, but you are surely right - and it very neatly links to something I've been thinking about recently, on the question of 'information overload.

Every day we have alway been in the presence of enough information/data to completely overwhelm us, but most of it we completely ignore. It is only when something or someone draws our attention to a bit of it that it comes to our notice. Thus, could it be that the change, the new problem, is too many people, or too many things, trying to draw our attention to things?

Allan said...

So a feature of modern life is an excessive number of things demanding our attention?

It feels like that, but do we really know that people have to attend to more things now than in the past? How would we check it?

It seems that at every period of history people thought they were subject to unprecedented demands. Maybe they are. It could also be true that the demands remain much the same in quantity, if not in quality.

By the way, with the 'gorilla experiment', while I was counting the passes, I knew perfectly well that all sorts of things could also be happening, and I would miss them. There could have been 10gorillas, 5 lions and an emu. I would have missed them all. I did, however, get the correct number of passes on the first go. I imagine (but don't know) that it would be relatively easy to design an AI system that spotted anomalous interlopers such as gorillas, but much harder to design one that correctly counted the number of passes. The remarkable thing is not that we miss the gorilla, but that we're able to choose to miss it so as to do better something else.