Graphics Reference
In-Depth Information
behavior judged from the perspective of human interaction partners.
This may be due to the fact that individual networks ensure a greater
coherence of the produced behavior. As a consequence, the agent may
appear more coherent and self-consistent which, in turn, may make its
behavior more predictable and easier to interpret for the user.
5.2 Interaction effects of coverbal gestures and
agent appearance over time
The previous section has clearly shown that co-speech gestures are
able to play a beneficial role for how embodied agents are judged
by human recipients. This finding is, however, still limited to one
particular embodied agent and measurements were only taken at one
particular moment of time. So, to elucidate the role of
gesture use
in
embodied agents in relation to the variables
agent appearance
and
point
of measurement
, a follow-up study has been conducted particularly
addressing interaction effects of these variables with regard to the
two major dimensions of social cognition: warmth and competence
(Bergmann et al., 2012).
This question has been investigated in a study with repeated
measures manipulating the agent behavior (gestures absent vs.
gestures present; Bergmann et al., 2012). For the gesture-present
condition, the GNetIc model built from speaker P5 of the SaGA corpus
(the model to be most successful model according to the previous
section) was employed again. This time, two different embodied
agents were used: the child-like agent 'Billie' and the robot-like agent
'Vince' (see Figure 5). The study was conducted in two consecutive
phases. In the first phase, participants were provided with a short
introduction by the embodied agent which took approximately 15
Figure 5.
The two embodied agents employed in the study both showing a gesture for a
square as it occurs in the stimuli participants were provided with: the robot-like agent 'Vince'
(left) and the human-like agent 'Billie' (right).
Search WWH ::
Custom Search