Item

Abstract

Introduction: Previous experimental work indicates that biological motion recognition is dependent on learning. In order to determine neural correlates of this learning process we conducted
an fMRI adaptation experiment (cf. [1]). Compared to previous studies using a classical
block design, adaptation experiments have the advantage that they allow to distinguish multiple
functionally distinct neural subpopulations within the same voxel, exploiting the fact that the
BOLD signal decays if the same stimulus is presented repeatedly. Methods: Pairs of unfamiliar
motion stimuli for a discrimination task were generated by motion morphing between triples of
prototypical trajectories of human movements. The morphs were generated by linear combination
of the prototypical trajectories in space-time [2]. Subjects had to discriminate between
two successive stimuli presented as Johansson point light walkers with same or different linear
weights of the prototypes. By choosing appropriate weight vectors the difculty of the discrimination
task can be precisely controlled. This was used to generate four conditions with identical
(SAME), very similar (SIMILAR), moderately similar (DISSIMILAR), and completely
different (DIFFERENT) stimulus pairs. Subjects reported whether they perceived the successive
stimuli as same or different. The subjects were scanned before and after a training period.
Areas relevant for the processing of biological motion (early visual areas, MT+ , KO, FFA,
and STS) were localized using standard techniques. Results: Before the training subjects could
discriminate the stimuli in the DISSIMILAR and in the DIFFERENT condition. After training
they could also discriminate the stimulus pairs in the SIMILAR condition. Comparing the
BOLD signal before and after the training period we found a signicant reduction of the signal
in all localized regions of interest. Before training, we obtained signicant adaptation effects
for the SAME and the SIMILAR condition only in area FFA and the STS. The other areas did
not show selective adaptation. After training, however, no adaptation effect was observed for
the SIMILAR condition any more. This result is consistent with an increased discrimination
capability after training. Conclusion: We have successfully established the fMRI adaptation
paradigm for biological motion experiments. The adaptation effects in area FFA and the STS
are consistent with the discrimination performance of the subjects before and after training.
Conforming with earlier studies, the STS and area FFA seem to be critical for biological motion
recognition. The general decrease in BOLD signal observed after training might be related
to a more efcient encoding of the stimuli after learning.