How to Make a Memorable Portrait

by F.


It’s simple: just distort the veridical image somewhere between 4% and 16%. Go over 16% and things get grotesque; under 4% and it’s forgettable. Clearly this is a principle that Lucien Freud (above) gets.

Here’s a summary of research on this from the Financial Times:

Thanks to a series of experiments carried out over the past decade or so (notably by Professor Gillian Rhodes, based at the University of Western Australia), this key mechanism of memory and visual experience has been clarified. And it carries crucial significance for the secret of portraiture.

Suppose we are presented with a direct image – say a passport photo. This image will be deemed as veridical or true-to-life. and assigned an identity. It might be someone new to us; but the experiment also works when it is someone we already recognise.

The test is as follows. Variations on the veridical image are constructed with digital manipulation. So certain distinctive features of the chosen face will be exaggerated. Bee-stung lips, protruding ears, heavy jowls or a dour frown – such features can be distorted according to a percentage scale. This distortion is a key weapon in the hands of the cartoonist – the device of caricature.

But it is also possible to go the opposite way, and modify the face in a manner that reduces its distinctive features: the lips drawn thin, the ears pinned back, the jowls lifted and the frown smoothed over. This amounts to “anti-caricature” – and again it can be done to a varying degree.

Back to the test. Volunteers are placed in front of a screen and asked to respond to a succession of images. Speed of identification is measured, over a sequence of veridical, caricature and anti- caricature faces.

Thousands of volunteers have been put through such trials, and a pattern of results has emerged. Anti-caricature is a deterrent to recognition. There is, of course, no problem about recognising veridical images. But for sheer speed of response, the most recognisable image is a caricature: to be precise, a degree of distortion quantified at between four and 16 per cent. Too much caricature becomes grotesque and causes a delay in recognition. Turn the face upside-down, and we suffer serious disorientation. But there is no doubt that our brains are programmed to memorise the faces of people by reference to exaggerations of reality.

What’s the cognitive mechanism responsible? Probably peak shift, as described by V.S. Ramachandran of the University of California at San Diego. He gives this example from rat psychology.

Imagine you’re training a rat to discriminate a square from a rectangle. So every time it sees a particular rectangle you give it a piece of cheese. When it sees a square you don’t give it anything. Very soon it learns that the rectangle means food, it starts liking the rectangle – although you’re not supposed to say that if you’re a behaviourist. And it starts going towards the rectangle because it prefers the rectangle to the square.

But now the amazing thing is if you take a longer skinnier rectangle and show it to the rat, it actually prefers the longer skinnier rectangle to the original rectangle that you taught it. And you say: Well that’s kind of stupid. Why does it prefer a longer skinnier rectangle rather than the one you originally showed it? Well it’s not stupid at all because what the rat is learning is a rule – Rectangularity. And of course therefore if you make it longer and skinnier, it’s even more rectangular. So it says: “Wow! What a rectangle!” and it goes towards that rectangle.

The full Financial Times article can be found here. Ramachandran’s wonderful lecture on the cognitive basis of art—from which he above excerpt is taken—is here.