IEEE Consumer Electronics Magazine - March 2018 - 98

0.64

Index

0.62
0.6
0.58
0.56
0.54
6

1 11 10 9 8 4 7 2 3
Strategy Number

5 12 13

Moving Average of
the Index over Four Values

FIGURE 7. The index for each strategy for phase 2.

Strategy 1
Strategy 11

0.62

Strategy 6
Strategy 13

0.6
0.58
0.56
0.54
0.52
2

8
6
Throw Number

FIGURE 8. The evolution of a moving average index over four values. The differences between the curves are not significant.

as depicted in Figures 6 and 7. These figures reveal the effect
of sounds on performance.
As the nature of future in-car systems will be closer to
phase 2 than to phase 1, we place significance on the phase 2
results. We first discuss the second-phase outcomes, followed
by those of the first phase.
The first thing to note is that strategy 6-velocity linked
to brightness-seemed to lead to better performance
compared to the other approaches. Considering only the more
simple strategies, strategy 6 was significantly better than
strategies 3 and 5. We can see that temporal strategies of AM
seemed to be misunderstood by the participants, achieving
low performance, as revealed in Figure 7. Concerning the
more complex approaches, strategies 11, 10, 9, and 8 resulted
in good performance. The satisfying results of strategies 10
and 8 may have been due to the velocity-brightness association of strategy 6, but their performance may have been a bit
degraded by the interaction between these parameters. Surprisingly good performance was obtained with strategies 8
and 11, although they were composed of two frequency
98 IEEE Consumer Electronics Magazine

march 2018

parameters, which could have resulted in a complex and confusing association.
In the case of strategy 8, participants might have focused
on only one of the sound parameters, ignoring the other. For
strategies 9 and 11, it is interesting to note that the performance of these strategies was better compared to the corresponding simple strategies-2 and 4 for strategy 9, and 3 and
4 for strategy 11-even if these differences were not significant. There may have been what we might call a symbiosis
effect, in which the strategy associations might have helped
the subjects accomplish the task. At this point, this can be
only a hypothesis, as there were no significant differences.
This kind of symbiosis effect (a significant one) was also
found in the first phase for strategies 11, 3, and 4.
We can also see that the performance of strategies 5 and
12 was close to the control approach, strategy 13, in Figure 7.
This means that, even if certain sounds help participants to
accomplish a task, sounds that are not adapted to the task
might not provide any useful information, leading to the
same results as the absence of sound. It should be noted that,
as with all of the other strategies, sound strategies 5 and 12
contained spatial cues related to the position of the VO
because of the sound spatialization. This should have enabled
the subjects to feel the behavior of the VO and consequently
help them reach the requested zone. As a result, we can formulate two hypotheses: either the spatial information was of
no use, or strategies 5 and 12 disturbed the subjects and
thereby prevented them from using this information.
Finally, we note that each association of a dynamic VO
parameter with a sound parameter leads to different results.
For instance, for the six simple strategies, neither the VO-tohand distance (for strategies 1, 2, and 3) nor the VO velocity
(strategies 4, 5, and 6) seems superior to the others. The
same can be said of the pitch strategy (strategies 1 and 4),
the AM strategy (strategies 2 and 5), and the brightness strategy (strategies 3 and 6). The results do not depend on particular parameters, but on the association between a VO and a
sound parameter.

LEARNING EFFECT IN PHASE 1
In this section, we focus on the learning phenomenon in
phase 1, from the first to the 15th throw. For that purpose, we
considered the mean error distance averaged across zones,
sound strategies, the spring stiffness, and subjects. The
ANOVA revealed a global learning effect, with significant
performance improvement between the first and sixth throw.
After throw six, the performance seemed to stabilize. This
may show that the learning effect was stronger at the beginning of the phase, as might be expected.
We then investigated this learning effect with respect to the
sonification strategies. The evolution of a moving average index
is represented in Figure 8 for the most interesting strategies:
mappings 1, 6, 11 and the control strategy 13. The ANOVA did
not reveal a significance of the (strategy · throw number) interaction or for the index (F (168, 54892) = 1.114, p = 0.149) or for
the distance (F (168, 55542) = 1.033, p = 0.368). Even if the

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - March 2018