working memory; working memory for manipulable vs. nonmanipulable objects,
working memory for faces vs. houses; working memory for olfaction... In
the domain of language, it is possible to dissociate between working memory
for phonological vs. semantic vs. syntactic content (Shivde a nd Thompson-Schill
2004), and, within syntax, some of the subregions of Broca’s area have been
proposed to be specialized for processing basic syntax, while others would handle
the short-term memory requirements of complex syntactic structures such as
wh-movement (Fiebach et al. 2005, Fedorenko et al. 2012). How can we rec-
oncile these findings? Where does short-term storage happen within state-based
accounts of WM?
The domain-generality of prefrontal cortex functions does not necessarily
preclude the emergence and measurement of domain-specific subregions that
guarantee an efficient cognitive control of specific tasks. Language in particular
could benefit from the existence of dedicated channels of control, since it requires
the precise combination across multiple cognitive cycles of very distinct material
(widely distributed instances of sound and meaning). Fedorenko et al.’s (2012)
language-specific regions of Broca’s area show individual-specific variability,
preferentially activate to linguistic input, and recruit more domain-general areas
in the prefrontal cortex whenever processing demands increase. Instead of
constituting an actual buffer where linguistic information is put on hold, they
can be considered as a case of neuronal recycling by extensive exposure to lan-
guage since infancy (Matchin 201 4). Neuronal recycling was first proposed by
Dehaene and Cohen (2007) as a way to account for the existence of visual
regions of the brain that seem to be specialized for the recognition of letters
when reading one’s native script. Given the recent invention of writing systems,
the researchers concluded that their existence has to be a result of the environ-
ment shaping the brain. According to Matchin (201 4), the term neuronal
recycling should perhaps be replaced by neuronal retuning, stressing the idea
that the areas that specialize do so only if they are functionally related to the
specialized task to perform. For language-specific regions, this would indicate
that the study of cognitive control functions, which take place adjacently to
these language-specific areas, can still inform us when trying to assign their
function.
All in all, the question of where short-term storage takes place is ultimately
a question about the format of those short-term storage representations. The
current view is that modality-specific storage capacities are an automatic result
of processing modality-specific information. Applied to the multicomponent
model, this view entails that the visuospatial sketchpad and the phonological
loop can no longer be considered dedicated buffers localized in a distinct short-
term memory area, but the visual and auditory systems themselves, which activate
either in the presence of stimuli or as a result of top-down signals. What we
perceive from the outside as the distinct categories of perception, imagination
and memory actually take place in a much related manner, using the same
physical locations (posterior brain regions) to create phenomenological experi-
ences through the MD network.
Language and working memory 109