This model uses a 16-degree of freedom robotic hand, a simple visual system, and Higher order Hopfield network (HOPP) to test the hypothesis of basic forms of imitation could emerge from self-observation. The results from the model indicates HOPP is capable of storing learned (i.e. self-observed) gestures, playback similar gestures after observing external patterns (i.e. imitation), and sometimes can even generalize to form new gestures.
This model assess the extent of imitation features that can be bootstrapped by combining
a robotic human-like hand with a minimal visual retina using biologically inspired associative network to reproduce an infant’s early visuomotor experiences.
This model combines video capture, low-level hand control server and high-level coordinator to provide a visuomotor associated Hebbian-like learning resulting from self-observation. The model is tested in terms of generalization between agents and generalization between gestures.
The role of the left inferior frontal gyrus in imitation. Imitation
typically or invariably occurs via activation
of the left IFG which plays an important
role in translating perceptual input from
observation of a model’s movement into
matching motor output
