| Computational
neuroscience, including neural network modeling, is being conducted
by the Barto and Moore
laboratories. This figure depicts a neural network architecture
that evolved with a genetic/neurodevelopmental algorithm and
that can learn a discrimination between stimuli that activate
different sensory channels (e.g., visual and auditory stimuli)
but not between stimuli within the same sensory channel (e.g.,
between a red and an orange visual stimulus). The network consists
of excitatory units in the cortex (unfilled ovals) with inputs
from hippocampal units (filled squares) that modulate synaptic
efficacies between units in sensory-association cortex and ventral
tegmental units (filled circles) that modulate synaptic efficacies
between units in motor-association cortex. |
