Your brain unscrambles words in the mirror but then switches them back again

We humans can recognise things from different angles and orientations. As Jon Duñabeitia and his colleagues observe in their new paper, a tiger is still a tiger whether you see it facing rightwards or leftwards. When it comes to words, though, this skill largely vanishes - mirror-reversed words are especially tricky to read. It makes sense that the brain becomes sensitive to orientation in this way because, unlike the tiger, a 'd' isn't a 'd' when it faces the other way: 'b' (and the same is true for other letters).

The question that Duñabeitia set out to answer is what happens, in the case of letters, to the brain's usual ability to recognise things regardless of their orientation? Is the automatic reversal process somehow unlearned for letters, or is it merely suppressed at a later stage of processing? Given how recently in our evolutionary history we started reading and writing, the latter seems more likely.

However, a recent brain imaging study using fMRI, led by Stanislas Dehaene, suggested that the automatic reversal process was completely blocked when dealing with letters. Dehaene's team found that mirror-reversed words failed to produce a priming effect, either in terms of brain activity or behavioural performance. That is, the subliminal flash of a mirror-reversed word didn't speed up participants' recognition of that same word when it subsequently re-appeared the right way around. This suggests the mirror-reversed words weren't switched around and processed normally by the brain.

But what if the temporal resolution of fMRI is too poor to detect early mirror reversal processes? Duñabeitia's team performed an experiment in which normal and mirror-reversed words were flashed up subliminally prior to repeated presentations of those same words, but they used electroencephalography (EEG) to measure their participants' brain activity. Unlike fMRI, EEG can measure changes in brain activity over sub-second periods (although its spatial resolution is much poorer).

In contrast with Dehaene, Duñabeitia did observe a priming effect for mirror-reversed words. Although at 150ms after a prime, brain activity was different between mirror-reversed and normally oriented prime words, by 250ms the brain's response to these two kinds of prime was the same. In other words, the brain detects the mirror-reversed orientation but by 250ms it has switched it around the right way. By 400ms (still less than half a second) after the prime, the pattern had changed again, so that now the mirror-reversed prime and normally oriented prime provoked different patterns of activity (located towards the back of the brain). This could be the postulated suppression process in action.

The intriguing implication of this research is that when reading mirror-reversed words your brain automatically flips them the right way around - for an imperceptible instant you have a mirror-reading ability - but then it suppresses that effect, putting the mirror reversal back in place again, hence the words appear as awkward to read. This interpretation is consistent with the finding that many young children are capable of spontaneous mirror-writing and reading, perhaps because they have yet to develop the suppression of the automatic reversal process. There are also reports of brain injury prompting the onset of mirror reading.

This new research is more than just curiosity, it could help further our understanding of dyslexia, which in some cases is associated with the unwelcome automatic rotation of letters and words. 'Now we know that rotating letters is not a problem that is exclusive to some dyslexics, since everybody does this in a natural and unconscious way,' said Duñabeitia. 'But what we need to understand is why people who can read normally can inhibit this, while others with difficulties in reading and writing cannot.'

ResearchBlogging.orgDuñabeitia, J., Molinaro, N., and Carreiras, M. (2011). Through the looking-glass: Mirror reading. NeuroImage, 54 (4), 3004-3009 DOI: 10.1016/j.neuroimage.2010.10.079 [Article pdf via author website].

share