NEURAL-NETWORK MODELS OF PERCEPTUAL-LEARNING OF ANGLE DISCRIMINATION

We study neural network models of discriminating between stimuli with two similar angles, using the two-alternative forced choice (2AFC) par adigm. Two network architectures are investigated: a two-layer percept ron network and a gating network. In the two-layer network all hidden units contribute to the decision at all angles, while in the other arc hitecture the gating units select, for each stimulus, the appropriate hidden units that will dominate the decision. We find that both archit ectures can perform the task reasonably well for all angles. Perceptua l learning has been modeled by training the networks to perform the ta sk, using unsupervised Hebb learning algorithms with pairs of stimuli at fixed angles theta and delta theta. Perceptual transfer is studied by measuring the performance of the network on stimuli with theta' not equal theta. The two-layer perceptron shows a partial transfer for an gles that are within a distance a from theta, where a is the angular w idth of the input tuning curves The change in performance due to learn ing is positive for angles close to theta, but for \theta - theta'\ ap proximate to a it is negative, i.e., its performance after training is worse than before. In contrast, negative transfer can be avoided in t he gating network by limiting the effects of learning to hidden units that are optimized for angles that are close to the trained angle.