theear,and ofcourse thevisual syste mand auditory localization.Theresultis a unified sensationofbody positionand
orientation, which in the present framework can be seen (for afirst approximation) as the consequence of a single
integrative processor reconciling the inputs fro ma large nu mber of independent interface modules. Here, if anywhere,
is a modular system in Fodor's sense: it is certainlya fast, mandatory, and cognitivelyimpenetrable input system—but
with multiple sources of input.
These are cases where multiple processors converge on a single“blackboard.”The converse also exists: multiple
interface processors that read fro mthe sa me“blackboard.”One case, discussed in the previous section, is the four
interface modules that all read different aspects of auditory input: interfaces that feed general-purpose audition,
phonology, voice recognition, and affect recognition. Similarly, conceptual structure is used as input for speaking (via
theconceptual structure–syntax interface),drawing inferences (a process creatingnewconceptual structures fro mold),
and acting (i.e. converting an intention—a kind of conceptual structure—into a structural format that can drive body
motion such as navigating or reaching). Through the interface to the visual system, conceptual structure can also be
used tocheck a belief, an inference, or a verbalclaimagainst a visual input. Thus conceptual structurefulfillsitscentral
function by virtue of its“blackboard”being accessed by interfaces to many different capacities.
A different sort of situation arises in reading aloud. Here visual input drives phonological structure through an
interface, but then the informationflow goes in both directions: toward pronunciation via the phonology–motor
interface, but also toward understanding via the phonology–syntax interface (except when reading aloud in a language
one does not know!).
7.5.5 Informational encapsulation among levels of structure
As observedearlier,a totallyinformationallyencapsulated modulewouldbeisolatedfrom therestofthef-mind, hence
useless. The interface modules are what prevent such isolation. But Fodor's idea of informationally encapsulated
modules still reflects an intuition that is worth capturing.
It seems to me that we can't talk properly abouttheinformational encapsulation of a module. Rather, we have to talk
about informational encapsulation in relative terms: to what degree does a distinction in one level of structure have a
bearing on distinctionsin some otherlevel?For instance, wordboundaries in phonology havea rather directreflection
insyntacticstructure; so syntax is notat allencapsulated withrespect tothisaspect ofphonology. Similarly, embedding
in syntactic structure is closely related to embedding in corresponding conceptual structure, so there is little
encapsulation here.