Connectionism and the problem of systematicity: why Smolensky's solution doesn't work

Area of Science:

• Cognitive Science
• Philosophy of Mind
• Artificial Intelligence

Background:

• The Fodor-Pylyshyn challenge questions how connectionist models explain cognitive systematicity without structured representations.
• Paul Smolensky proposes distributed mental representations as a solution.
• Smolensky offers two accounts: weak and strong compositional structure.

Purpose of the Study:

• To evaluate Smolensky's proposed distributed representations for explaining cognitive systematicity.
• To critically analyze the coherence and relevance of weak compositional structure.
• To assess whether strong compositional (tensor product) representations can solve the systematicity problem.

Main Methods:

• Conceptual analysis of Smolensky's arguments.
• Critique of weak compositional structure's relevance and coherence.
• Examination of strong compositional (tensor product) representations regarding constituent structure.

Main Results:

• Weak compositional structure is deemed irrelevant to the systematicity problem and internally questionable.
• Strong compositional representations lack the necessary constituents for structure-sensitive mental processes.
• Neither proposed distributed representation adequately addresses the Fodor-Pylyshyn challenge.

Conclusions:

• Smolensky's distributed representations do not successfully explain cognitive systematicity.
• The critique highlights limitations in current connectionist approaches to structured cognition.
• Further theoretical development is needed to bridge the gap between connectionism and systematicity.