Binaural fusion: Complexities in definition and measurement

Abstract

Despite the growing interest in studying binaural fusion, there is little consensus over its definition or how it is best measured. This review seeks to describe the complexities of binaural fusion, highlight measurement challenges, provide guidelines for rigorous perceptual measurements, and provide a working definition that encompasses this information. First, it is argued that binaural fusion may be multidimensional and might occur in one domain but not others, such as fusion in the spatial but not the spectral domain or vice versa. Second, binaural fusion may occur on a continuous scale rather than on a binary one. Third, binaural fusion responses are highly idiosyncratic, which could be a result of methodology, such as the specific experimental instructions, suggesting a need to explicitly report the instructions given. Fourth, it is possible that direct ("Did you hear one sound or two?") and indirect ("Where did the sound come from?" or "What was the pitch of the sound?") measurements of fusion will produce different results. In conclusion, explicit consideration of these attributes and reporting of methodology are needed for rigorous interpretation and comparison across studies and listener populations.

FIG. 1.

Schematic of different possible ways that binaural fusion may interact in the spatial and spectral domains. Sound in the left ear is depicted by an upward pointing triangle and in the right ear by a downward pointing triangle. (A) represents when binaural fusion occurs in both the spatial and pitch dimensions, resulting in one perceived sound source. (B) represents when two sources are perceived with one pitch but at two locations. (C) represents when two sources are perceived at one location but with two pitches. (D) represents when two separate images are perceived, each with a separate pitch and location.

FIG. 2.

(Color online) Percentage of fused responses from experiment 1 (fusion) and experiment 2 (lateralization, which included separate ITD and ILD lateralization tasks) in nine bilateral CI listeners from Kan et al. (2013). For the direct fusion measurement of experiment 1 of Kan et al. (2013), response options 1 (on the left), 2 (in the center), 3 (on the right), and 9 (one concentration) were considered as a single fused response (see Fig. 1 of Fitzgerald et al., 2015). For the direct fusion measurement in the lateralization experiment of experiment 2 of Kan et al. (2013), fused responses were trials in which one sound location was reported; indirect fusion measures were calculated from these data but were not shown for clarity. For the indirect fusion measurement using ITD discrimination thresholds that did not include image centering of Kan et al. (2015), electrode pairs with measurable ITD thresholds were plotted as having a fused image (100%), and those with immeasurable ITD thresholds were plotted as having an unfused image (0%). Mismatch in electrodes was introduced by holding a reference electrode fixed in one ear, and the electrode in the other was varied between ±8 electrodes. The reference pair was interaurally pitch matched. Data are from Kan et al. (2013) and Kan et al. (2015).

FIG. 3.

Summary of fusion ranges for adults with normal hearing, bilateral hearing impairment without CI, bimodal (HA and CI), and bilateral CI. Fusion ranges represent the range of tone frequencies (or electrodes for bilateral CI users) that fuse with a reference stimulus (tone for normal hearing and bilateral hearing impairment without CI; electrode for bimodal or bilateral CI) in the contralateral ear. Open and filled symbols indicate spatial and spectral fusion ranges, respectively. Spatial fusion ranges are replotted from four listeners in Buus et al. (1984) with normal hearing at 1 kHz, in Buus et al. (1984) for ten listeners with bilateral hearing impairment without CI at 1 kHz, and in Kan et al. (2013, experiment 1) for bilateral CI (middle electrode); there are no publications explicitly measuring spatial fusion with bimodal CI. Spectral fusion ranges are replotted from Anderson et al. (2020) and Hartling et al. (2020) for normal hearing at 1.6–2 kHz, from Reiss et al. (2017) for bilateral hearing impairment without CI at 1.6 kHz, from Reiss et al. (2014a) for bimodal CI for a mid-apical electrode, and from Reiss et al. (2018) for bilateral CI for a mid-apical electrode. Boxes indicate the interquartile range (50% of data) and horizontal lines depict the median for each group. Fusion ranges in mm for bilateral CI users (right axis) are scaled to acoustic octaves (left axis) based on the Greenwood place-frequency function (Greenwood, 1990). Points experiencing a ceiling effect (i.e., measurements were not performed at sufficiently large differences to reach the 50% fused responses) are shown by arrows for ceiling effects on both sides and “+” symbols show ceiling effects on one side. Data are from Anderson et al. (2020), Buus et al. (1984), Hartling et al. (2020), Kan et al. (2013), Reiss et al. (2014a), Reiss et al. (2017), and Reiss et al. (2018).