Which statement about a connectionist model is false?
A) The connectionist model assumes that a given node representing a feature can be connected to multiple nodes representing categories and that a given category node can be connected to multiple different feature nodes.
B) Networks may contain a layer of "hidden units" that enable particular compounds of feature nodes to predict something different from either single inputs or inputs from other feature compounds.
C) The most efficient connectionist models assume that a novel "configural input" is formed to represent each unique set of feature compounds
D) Connectionist models that are similar to the Rescorla-Wagner model cannot predict the extraction of a rule that describes a general outcome when the subject has merely been exposed to a series of learning trials.

Question

Quizplus · Accepted Answer

The statement in option C is false because connectionist models do not typically assume that a novel "configural input" is formed for each unique set of feature compounds. Instead, they rely on the distributed representation of features across nodes.

Which Statement About a Connectionist Model Is False