A Rose From Any Other Direction is Still a Rose, But Its Name is Not

Tuesday, January 12, 2010

Imagine a cup. Now rotate it 180 degrees. What is it now? It’s still a cup. Surprising? Not really. We live in a three dimensional world and take for granted that something flipped around an axis doesn’t change into something else. This generalization across mirror images is called mirror invariance and holds true for most visual stimuli.

But what about the letter b? Draw it on something transparent, turn it around – and lo and behold, it changes into the letter d! Writing systems are one of the few domains where an object’s orientation matters. Our brains have trouble with this, and this difficulty becomes apparent when children learn to read and write. It is well documented that children often flip letters when writing, and sometimes even write entire words spontaneously flipped. It is only after years of practice that they get over this tendency.

Neuroscientist Stanislas Dehaene and colleagues recently found evidence of mirror invariance (and the lack thereof for words) in the brain.


They used a technique known as fMRI repetition suppression. The basic logic of the technique is this: brain regions responding to a repeated stimulus will show a decreased activation. This is thought to be due to neuron adaptation or fatigue. Therefore, by showing someone different visual stimuli and looking for repetition suppression, we can get an idea of which stimuli the brain categorizes as a the same (a repeat), and which are different.

In this particular study, reported in Neuroimage ( 29, (2010) 1837-1848), participants saw words and pictures that were preceded either by a normal or mirror reversed image of the same category. (The participants were asked to perform an unrelated size judgment task.) The researchers found that picture and word processing regions both show repetition suppression to repeated identical images. This was not surprising. It’s the basic repetition suppression effect.
Dehaene and colleagues then looked for areas that showed repetition suppression for words/pictures and their mirror images. They found a region in the left fusiform gyrus that showed mirror repetition suppression for pictures, but no such region that showed mirror repetition suppression for words. Therefore, they found evidence for brain regions that considered pictures equivalent to their mirror images, but no such regions for words.

Surprisingly, the region that showed the strongest mirror invariance effect for objects was an area of the left fusiform known as the Visual Word Form Area, an area that has been shown in many studies to be active during word processing. Therefore, it’s possible that the same exact brain region that processes pictures in a mirror invariant way knows to behave differently for words. With the low resolution of fmri, we can’t rule out the possibility that the word network and picture network are in fact separate, but it should be an interesting question to pursue.

Is this lack of mirror invariance specific to writing systems we’ve learned, or does it hold for any script-like stimulus? A follow up behavioral study suggests the latter. In this study, people saw words, tools, faces, and scripts. Each image was presented for a quick flash of 200ms each, and was preceded by another image of the same category, either normal orientation or flipped. Their task was to say whether the two stimuli were the same, where a mirror image was also considered the same.

As measured by reaction time, people were relatively quick to judge an object and its mirror reflection to be the same, but slower to judge two mirror image scripts to be the same. This difficulty held true even for unfamiliar or false scripts, suggesting that the brain generalized the importance of orientation to unfamiliar, unlearned scripts.

Ideas for future research? The obvious next step would be to do this with children. Also, does anyone know of any studies of word to object cross categorical priming? That would help answer the question of whether the word network and object networks are the same.

Blog Carnival Coverage:
Psychology Articles Carnival

9 comments:

Jason Black January 12, 2010 at 12:18 PM  

"As measured by reaction time, people were relatively quick to judge an object and its mirror reflection to be the same, but slower to judge two mirror image scripts to be the same."

I remain skeptical, wondering whether they also tested mirror-images of random scribble drawings; ones that mimic the visual properties of script (high-contrast, monochrome, formed of lines and curves) but that were in no way actual examples of scripts.

I would expect people to have similar difficulty in assessing whether such scribbles were true mirror images as with script forms.

Livia January 12, 2010 at 12:22 PM  

Jason -- a random scribble in that case would just be like another false script, wouldn't it? I think you're arguing the same point as the paper authors -- that script-like images (including false scripts) give equal difficulty to scripts.

Livia January 12, 2010 at 12:28 PM  

Er, lemme rephrase that. The paper authors are saying that people have as much difficulty with script-like images as with real scripts, which is what I think you're saying too?

Meghan January 27, 2010 at 8:11 PM  

Very interesting. Were there any comparisons of reaction times of those presented with a learned word vs. reaction times of those presented with words in a foreign script? Similarly, were there any comparisons of the severity of repetition suppressions for those presented with a learned word vs. those presented with a foreign script?

Livia January 28, 2010 at 12:44 PM  

Meghan - -good quesion. Behaviorally, I don't think there was a significant difference (skimming the paper again now). They didn't compare words adn false fonts in the scanning experiment, so I dunno about that.

debbie January 28, 2010 at 11:14 PM  

It seems that Dehaene found an interesting discovery concerning the activity of the left fusiform area when a person is viewing 'mirrored objects,' and while there is not as much activation with scripts/languages, does that necessarily assume that the vwfa exists and is specialized in identifying words? I guess I'm not seeing the relation in what you are describing in the article - perhaps its a wording misunderstanding. But it is possible to conclude that the left fusiform gyrus is responsible for word/script/language identification just because of a slower reaction time?

Livia January 29, 2010 at 8:07 AM  

Deb -- this paper isn't enough to establish word specialization. The paper is relying on a large body of previous literature that has shown the VWFA to be specialized for word processing. This paper is more to characterize that region more closely.

Ken January 30, 2010 at 12:40 AM  

A very interesting blog with many pertinent questions raised.

Jemi Fraser February 7, 2010 at 7:19 PM  

I just taught my kids about learning disabilities by using a calculator and turning it to different orientations on Friday. Then I wrote b, p & d on the board and pointed out that sometimes orientation matters. They were fascinated that some brains wouldn't see it the way their brains did.

Post a Comment

  © Blogger template The Professional Template II by Ourblogtemplates.com 2009

Back to TOP