Datensatz

Zusammenfassung

It has been shown, that humans are able to learn to discriminate between dierent styles
of natural movements (e.g., gaits or sports movements). However, it remains unknown
whether this learning is based on innate templates for biological movement patterns, or
if humans can learn representations of new arbitrary complex movements. We address
this question by investigating whether subjects can learn articial biological movement
stimuli. Methods: We generated biological motion stimuli by linear combination of
prototypical trajectories. Spatio-temporal linear combinations were computed using
special algorithm, spatio-temporal morphable model (Giese Poggio, 2000, International
Journal of Computer Vision, 59-73). The following two classes of stimuli were
generated: (A) stimuli derived by linear combination of dissimilar natural movements
(e.g., walking, kicking and dancing). (B) Stimuli generated by animation of an articial
skeleton model that is highly dissimilar from naturally occurring body structures. The
joint angle trajectories of the skeleton were given by linear combinations of synthetic
trajectories. These trajectories were sinusoidal functions. Their amplitudes and frequencies
were approximately matched with the joint trajectories of human actors during
natural movements. Subjects had to discriminate between pairs of these stimuli that
were dened by linear combinations with slightly dissimilar weights. The trajectories
were presented as normal point light walkers (PLW), and as point light walker with position
jitter (PLWJ). The PLWJ were generated by adding random displacements of the
dots along the skeleton of the walker for in each frame. Each subject took part in two
training and three test blocks. Feedback was provided only during training. Results:
Subjects trained with stimuli derived from natural movements (group A) learned the
discrimination between novel patterns very quickly (about 8 repetitions). For rotation
of the test stimuli against the training stimuli we found transfer only for the normal
PLW, but not for the PLWJ stimuli. Subjects were able to learn the completely arti-
cial stimuli (group B) presented as PLWJ equally fast as the stimuli from group A.
Conclusions: (1) New templates for movement recognition can be learned very quickly.
(2) Learning aects at least two dierent levels of representation (local and holistic).
(3) The learned holistic representations seem to be view-dependent. (4) There seems
to be no signicant dierence in the learning process between stimuli derived from
articial and natural movements.