Compositionality: A Connectionist Variation on a Classical Theme

Connectionism has been attacked on the grounds that it does not employ compositionally structured representations (e.g., Fodor & Pylyshyn, 1988). This article develops the response that Connectionist models can, and in fact sometimes do, employ compositionally structured representations without, thereby, simply implementing a Classical “Language of Thought.” Focusing on the mode of combination employed in constructing representations, it distinguishes concatenative compositionality, essential to the Classical approach, from a merely functional counterpart increasingly common in Connectionist research. On the basis of this distinction it is possible to demonstrate that Connectionist representations can be compositional without being Classical, and further, that Fodor and Pylyshyn's supposedly conclusive arguments in favor of the Classical approach do not in fact support that approach over the Connectionist alternative (as opposed to an “ossociotionist” straw man).     

Isolating cognitive modules with the dual-task paradigm: Are speech perception and production separate processes?

A dual-task paradigm is used to investigate whether the auditory input logogen is distinct from the articulatory output logogen. In the first two experiments it is shown that the task of detecting an unspecified name in an auditory input stream can be combined with reading aloud visually presented words with relatively little single- to dual-task decrement. The stimuli for both tasks are independent streams of random words presented at rapid rates. A series of control experiments suggest that the first task places a considerable information processing load on the auditory input logogen, the second a considerable load on the phonological output logogen, and that subjects do not switch between the two tasks. The fact that the two tasks can be combined with ease is therefore interpreted as supporting the view that the systems underlying reading aloud and listening are separate. The ease of performance when the input streams are in different modalities, compared to the difficulties when they are in the same, has implications for general models of attention.