November 30, 2004

Matt:

I guess I should introduce myself. I'm Matt Webb, I co-authored the book with Tom. Hello! Mostly I post notes to my personal weblog, Interconnected, and sites that I run across to the Mini Links--that's probably the best way to find out what I'm interested in. What I'm not is a psychologist or neuroscientist (my day job is in social software, internet stuff), but I've been immersed in the academic literature for the past N months. Hopefully that means I'll be able to bridge some of these brain ideas over to the tech world, which is the one I usually inhabit.

It has been great fun writing Mind Hacks. Actually, what was possibly more fun was trying out all the experiments, mucking around with mirrors, or a pendulum, or sitting by the road watching the traffic. I'm aiming to post the same kind of stuff here, things to try at home (or on your friends), and to do the same as in the book: dig a little deeper, see what's really happening in our heads. If there's something you think we should give a go, email in. If it's not too embarrassing I'll put photos up.

—Matt.

Tom:

Hello. And I'm Tom Stafford, the other author of the book. While I was doing my degree and PhD I kept notes on lots of funny little things in psychology and neuroscience, although I never really knew why I was doing it. Then along came Matt and the idea for the book, and I had a place for many of the things I'd jackdawed over the years. But they were just enough to get us going. The rest of Mind Hacks me and Matt discovered during our Summer of Book - and that was a hell of a ride I can tell you. All summer we kept finding out new exciting things, and we knew that we'd only have time or space for a fraction in the book. Hopefully some of them will end up on this weblog, along with other things out there that we missed. There's also room for all the discussions we didn't have in the book, about all the wider issues raised by this stuff, and especially about different implementations of the hacks we suggested. I'm looking forward to finding out some more cool things and having some cool conversations on mindhacks.com...I hope you'll join us.

—tom.

Dragon's Head:

Speaking of eyes following you around the room, this Dragon Optical Illusion is pretty cool. You make it out of paper and sellotape, and move around it with one eye closed. The head seems to move and follow you around. (There's a PDF to make the model, and a video to watch if you can't be bothered.) Here's the one I made:

The head's actually folded inwards, but we're so used to 3D objects bulging outwards that we see the model as if it's moving instead of its true shape. You don't even need to close one eye--from a few feet away it's pretty compelling. A neat instance of the visual system's assumptions dominating our current knowledge.

—Matt.

Hack 101: Make Eyes (or Anything) in Pictures Follow You Round The Room:

The eyes of some pictures seem to follow you around the room, like those of the famous WWI recruitment poster which helped garner almost 3 million volunteers in two years:

Try it. Get up and look at your screen from the side. Is he still looking at you? He should be.

Recently published research in the journal Perception [1] discusses how this effect works. The story was covered in the press (e.g. here). Turned around into a 'how to' rather than a simple 'explanation' it's perfect material for a hack. I saw it too late to include in the book so I'm putting it here.

So here we go: How can you design pictures of faces with eyes that will follow you round the room?

The answer is simple: photograph, or paint, the face looking straight out. If it's a photograph they must look straight at the lens of the camera. In the words of James Todd of Ohio State University, one of authors of the study, 'If a person in a painting is looking straight out, it will always appear that way, regardless of the angle at which it is viewed'

How does it work? First of all, this is only possible because pictures and paintings aren't 3D. They are semblances of 3D on a flat surface. This stops our brains calculating depth by comparing the images in the two eyes (how our brain calculates depth in images is covered in the book). Instead, our brains rely on other cues to depth, such as shading (the use of shadows to imply depth) and movement (all this is also covered in the book).

The explanation lies in how we interpret three-dimensional objects portrayed on a flat surface. Real three-dimensional objects look different depending on the angle because of the changing way light falls across them. But on the flat canvas, shading and light are fixed and the image looks the same from every angle. If the face is looking straight out from one angle, it will appear to be looking straight out at whatever angle it is viewed at.

In fact the only clue that the object in a picture isn't really looking straight out is that the near side of an object should get smaller if you look at it from one side. This doesn't happen in a natural way with a painting. Theoretically your visual system could use this information to figure out that pictures of objects aren't real and thus the eyes aren't really following you around the room, but it appears that they don't. The contradictory information is either overridden or disregarded.

If you want look at the original paper you see that the how-to-make-eyes-follow-you-around-the-room result is actually more of side-issue of the main thrust of the paper - which is a discussion of the visual mechanisms behind and correct interpretation of effect.

But the best thing, and the thing which I didn't see picked up by any of the press, is that you can do the trick with any object which has shading, and that for their investigation the authors used a statue showing a woman's bum.

Hilarious! Why did no one mention it in the press?

And it works, too, look:

They've made a picture of a bottom that follows you round the room. Ain't science great. (I'm absolutely convinced that more psychology should involve the analysis of naked bottoms.)

Refs:

1. Koenderink J.J., Doorn A.J. van, Kappers A.M.L., Todd J.T. (2004). Pointing out of the Picture. Perception, 33, 513-530. Here for subscribers

—tom.

Little men and their discontinuities:

There are little men inside my head that tell me what to do.

No.

Really.

Here's one of them:

The little man is a model according to a sketch of the human body on the surface of our brain.

The different parts are sized according to how much space the brain gives to processing information about that part of the body. This particular model - called the sensory homunuculus because it is constructed based on the sensory cortex - is where sensory signals from the body converge. The sensory cortex is a physical map of the body on the surface of the skin, also called the 'somatosensory' cortex (mean 'body sensing'). Each part of your body has a different position on the surface of the map, and mostly, but not always, bits of the body that are next to each other are processed next to each other in the cortex.

There are similar maps in both sensory cortex and in motor cortex (where motor plans - and even the perception of other people's movements - convert before becoming actions). They are not completely identical - one of the applications discussed in the book lets you compare the differences between the two - but one similarity is that the major breaks in the continuity of the map are the same. The hand areas are next to the face areas, and the foot areas are next to the genitals area.

You can mostly see this from this image of the motor cortex (which i took from PBS- thanks PBS!)

Rather that appearing on the map near the hip, the genitals appear below the toes, in the crevice at the top of the brain (rather prudishly the genitals are not shown on this image, which isn't uncommon). The fact that the genitals are next to feet are responsible for the 'whole leg orgasm' phenomenon in amputees (which the neuroscientist VS Ramachandran talks about here. Following injury to the body the brain's maps can reorganise - unused areas becoming canibalised to help with processing of areas which are still being using. VS reports of leg amputees whose somatosensory cortex had reorganised to devote the areas previously used to represent the leg to processing the sensory input from the genitals. When the patients had sex they had an orgasm - but with two or three times the amount of brain space devoted to representing the feeling!

Wow.

Anyway, this is a bit of distraction from the thing I wanted to talk about in this post: why are there breaks in the map of the body as it appears on the surface of the cortex?

Martha J. Farah of the University of Pennsylvania wrote a paper [1] which gives one suggestion: the maps are like that not because they are innately fixed that way, but because they self-organise that way as brain develops in the womb. We know that a lot of the structure of the brain isn't entirely pre-specified, but arises due to what activity happens during development. It's called activity dependent development, and similar processes are responsible for the re-wiring the happens during adult learning. Martha Farah suggest that, because of the way a baby is tucked up in the womb - with the hands curled up near the face and the feet tucked in by the genitals - these two pairs of locations are most likely to be co-activated by any movement experienced by the baby. And it is this activation - and co-activation - by random movements that is used to seed the structure of the sensory map.

Neat, eh?

This goes to explaining the organisation of the sensory map, and perhaps the motor map is based on the initial template of the sensory map and so develops that way - an initial bias towards the face-hands, feet-genitals discontinuity becoming locked in during the process of map self-organisation.

A couple of points Prof. Farah doesn't make are:

Firstly, the four areas mentioned in the title of the paper are the four with the largest representational resources dedicated to them. Surely it is not coincidence that the four areas where the map continuity is broken are the four most important areas?

Secondly, the standard diagram shows the cortical map as being, effectively, 1 dimensional (ie an ordered scale of areas). Is there a reason for this (perhaps due to the four aforementioned regions requiring so much representation that they require the whole width of somatosensory cortex, preventing proper 2 dimensional representation of the body)? When mapping a 3D body to a 1D map discontinuities would have to occur, wouldn't they? So the existence of discontinuities itself isn't at all surprising, but the self-organisation of representational maps suggests why they are where they are.

Refs

1. Farah, M.J. (1998). Why Does the Somatosensory Homunculus Have Hand Next to Face and Feet Next to Genitals?: An Hypothesis. Neural Computation, 10 (8), 1983-1985.

—tom.

November 29, 2004

It's arrived!:

Well, well, it looks like Mind Hacks is shipping. I was only expecting a couple of CDs today, and look what the postman brought.

—Matt.