Developing Intelligence

The claim that language processing can be carried out by purely “general purpose” information processing mechanisms in the brain – rather than relying on language-specific module(s) – may seem contradicted by a slew of recent neuroimaging studies demonstrating what appears to be a visual “word form” area in the left fusiform gyrus of the temporal lobe. By all appearances, this region is highly specialized for word processing. But this evidence causes a predicament for more than just domain-generalists; those who advocate an evolved language module may also be challenged by these results, since writing emerged too recently (~5400 years ago) for the brain to have already evolved a specialized mechanism to accomodate it.

Only some very powerful evidence could result in such strange theoretical bedfellows, as reviewed by McCandliss et al. in TiCS:

– Even purely behavioral evidence supports the idea that word recognition is carried out by extremely specialized and dedicated hardware. Short words are extracted in parallel from their constituent letters (the speed of word recognition is unaffected by the number of letters in a 3-6 letter word), in some cases fast enough to aid in the identification of those letters themselves (the word superiority effect, which has even been demonstrated for well-formed nonwords such as “lort”)

– Neuroimaging metaanalyses identify an area in the left occipitotemporal sulcus, bordering the fusiform gyrus, that is so consistently active in word reading studies that it includes the peak area of activation within a given individual’s data in reading studies over 90% of the time

– Damage to this region of cortex causes deficits in visual but not auditory word recognition

– Words can be presented to subjects below the threshold of conscious awareness (e.g., for 29 ms) and yet still show differential activation of this region and associated priming effects, such that subjects are faster to identify that same word later – even if the word itself appears different in font, case or size – indicating that this area truly responds to some abstract and high-level visual representation of the words.

How is it possible that we have such seemingly “dedicated hardware” for visual word recognition if writing hasn’t been around long enough for this mechanism to evolve? And how can we reconcile the apparent specificity of this region with the idea that the brain uses parallel distributed processing instead of modular organization?

One answer comes from the observation that such apparent modularity of function can emerge as the end-product of the domain-general distributed mechanisms underlying neural development. McCandliss et al. argue that this “visual word form area” is the product of long developmental process whereby we all become word reading experts, drawing analogies to evidence demonstrating similar expertise effects in the right fusiform gyrus, where face-selective activity is observed alongside with bird-selective activity among bird experts, and car-selective activity among car experts.

A variety of evidence supports this claim that ontogenetic development can account for this apparent specialization of function. Obviously, we’re not born knowing how to read, and first-graders accordingly show the kinds of word length effects you’d expect: they’re slower to identify longer words than shorter ones. This appears to be accompanied by electrical activity on the scalp known as the N200, which comes to show adult-like responses to visual reading of real words by early adolescence, but doesn’t show adult-like responses to nonwords until adulthood. (Apparently, adults are so expert at word reading that they show automatic word recognition even for never-before seen words!) All of these phenomena have been associated with the left fusiform area, suggesting that it does indeed follow a protracted developmental trajectory.

Similarly, developmental dyslexics show reduced activity in the visual word form area, leading the authors to suggest that dyslexics fail to develop the same form of word reading expertise that healthy word readers take for granted. Of course, this could as easily be a cause of dyslexia as an effect of a deeper underlying cause.


  1. #1 David Harmon
    September 14, 2007

    It seems to me they’d want to take a look at otherwise healthy people who didn’t learn to read until adulthood.

    A more general resolution of the dilemma might run as follows:

    If a putative writing system didn’t key into some processing capability which was available to early humans, then people wouldn’t have been able to use it! Therefore, any writing system which did take hold, must necessarily be suited to human brains, even if the low-level subsystems or functions involved were never “intended” for any such task. In short, tracing the “functional regions” of reading/writing is actually examining the aftermath of a co-opting of those regions to a task which may be formally “novel” as far as evolution, but which was “chosen” to be workable.

  2. #2 Al Fin
    September 17, 2007

    If very young children suffer brain damage of the left fusiform gyrus and occipitotemporal sulcus, then go on to read normally, that would suggest the specialisation comes about during early childhood neural development.

    Too bad there are no good animal models for reading.

    Reading phonetic alphabets should be somewhat different, neurologically, than reading ideograms or logographs, such as in written Chinese dialects or Egyptian hieroglyphics.

    The pattern recognition neural net architecture would have been present prior to the development of written language. Reading would represent an intensive specialisation of visual pattern recognition, with auditory feed forward/feedback to the temporal lobe.

New comments have been disabled.