*3.1. Development of Sound Localization Accuracy Is Driven by Time since Activation of Bilateral Implants*

Simple linear regression analyses of cross-sectional data (first test, *n* = 18) indicated that increasing time since activation of BiCI was associated with increasing sound localization accuracy (EI = 0.83–0.19 time since BiCI, r = −0.47, *p* = 0.05). A linear mixed model for the entire longitudinal dataset (*n* = 42) showed a significant main effect of time since BiCI on the EI, whereas no effect of random intercept (i.e., of subject) or interaction with number of trials existed. The final linear model, thus, included time since activation of BiCI, which explained 25% of the variance in the EI (R<sup>2</sup> = 0.25, F = 13.6, *n* = 42, *p* = 0.0007) (Figure 2). According to the model equation, which was similar to the linear fit from the cross-sectional analysis, the EI decreased by 0.21/year with an intercept of 0.82.

**Figure 2.** Children with congenital severe-to-profound hearing loss develop horizontal sound localization abilities with increasing time since activation of bilateral cochlear implants. The black open circles depict localization accuracy (Error Index) for 18 infants measured at 1 to 5 occasions (*n* = 42, mean follow-up time = 0.9 years (0.09 years–1.7 years); mean age at test = 1.5 years (0.58 years–3.6 years). Localization accuracy increased as a function of time since activation of bilateral cochlear implants (R2 = 0.25, F = 13.6, *n* = 42, *p* = 0.0007, linear mixed model). The black solid line is the linear fit from a linear mixed model analysis, and the dotted lines depict the 95% confidence interval of the fit.

Individual perceived versus presented azimuths were plotted for the child with the longest time since activation of BiCI (Figure 3), to visualize development of behavioral response patterns. With increasing time since activation of BiCI (4 measurements over 1.4 years follow-up), perceived azimuths were approaching target azimuths.

**Figure 3.** Perceived as a function of target sound-source azimuth at 4 occurrences (one per panel) for an individual child. In the left panel, this child had bilateral cochlear implants activated for 0.33 years. As experience with bilateral cochlear implants increases (from left to right), datapoints approach the line of equality corresponding to perfect sound localization accuracy in this task.
