IEEE Consumer Electronics Magazine - March 2018 - 97

values were chosen, representing a soft, normal, and stiff spring.
The sonification strategies were tested for each stiffness value in
a random order, resulting in 13 strategies $ 3 stiffness values = 39
conditions.

participants were puzzled because they were certain they performed exactly the same gesture though they obtained different results.

INDEX ANALYSIS
For each of the 39 conditions, the experimental session was
divided into two successive phases: a grouped learning
phase (phase 1) and a distributed learning phase (phase 2).
Each condition was tested, starting with phase 1 and followed by phase 2. Another condition was then randomly
chosen, and the test continued with the two phases being
performed successively.
Phase 1 enabled subjects to understand how the zones
were constructed and how the sound evolved with the VO and
with their gestures. During this phase, subjects were asked to
execute 45 right-handed swipes, i.e., 15 successive swipes in
each zone. Each zone was presented in a randomized order.
In this part, subjects had access to visual feedback on the
accuracy of their throws, presented on an interface (Figure 4).
When the zone was not attained, a too close or too far message appeared, helping the subjects find the correct gesture
for each zone.
Phase 2 was closer to potential in-car use and is of particular interest to us. This phase consisted of effectuating 15
swipes corresponding to five throws toward each randomly
presented zone. This task was harder compared to phase 1
because of the succession of varying gestures required.
Moreover, the visual feedback was less informative than
in phase 1, since it only told whether or not the correct zone
was reached.

To discriminate among the mapping strategies, it is now of
interest to analyze the index, which takes into account more
information than just the success rate. The mean indexes for
each strategy are presented in Figures 6 and 7 for phase 1
and phase 2, respectively. Small indexes indicate good mapping strategies. In these figures, the strategies are ranked in
an increasing order and are grouped by the horizontal lines
in the upper part of the figure according to the Duncan post
hoc analysis.
The analysis reports a significant effect of the strategies
for phase 1 (F (12, 54892) = 1.979, p 1 0.05) and for phase 2
(F (12, 17452) = 2.780, p 1 0.002). The Duncan post hoc
tests reveal several significant differences between strategies,

1.6
1.4
1.2
1
Index

EXPERIMENTAL PHASES

0.8
0.6
0.4
0.2
0

STATISTICAL PROCESSING

RESULTS
PERFORMANCE
We computed the success rates for each sonification strategy
and for each phase, which were between 46% and 50% for
phase 1 and between 41% and 49% for phase 2, considering
all of the strategies. The low success rate obtained can be
explained by the difficulty of the task; the gestures used to
reach each zone were really close to one another. Sometimes

0.2

0.4
0.6
Final VO Position

0.8

Index for Zone 1
Index for Zone 2
Index for Zone 3
FIGURE 5. The index construction.

0.64
0.62
Index

The number of correct throws was collected for each strategy.
Then a success rate (a correct/missed throw ratio) was computed for each strategy. For missed throws, we wanted to take
into account the distance from the center of the requested
zone. We then computed a global index by taking into account
both the success rate and the error distance. To keep the idea
of an error rate, we chose to set the index to 0 if the throw was
in the right zone and to start at 1 if it was outside the zone,
increasing the index with the distance from the zone. Consequently, the smaller the index value, the better the strategy.
The index is represented in Figure 5 for the three zones. We
finally performed an analysis of variance (ANOVA) on this
index with strategy as a factor.

0.6
0.58
0.56
0.54
11 6

8 10 4 9 12 5 13 3
Strategy Number

FIGURE 6. The index for each strategy for phase 1.

march 2018

IEEE Consumer Electronics Magazine

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