**4.1 Study selection overview**

Several studies of the impact of HRTF training on localisation accuracy have been selected from existing literature, for which authors graciously provided raw participant data used in the comparative analysis. A short description of each study is provided in the next section, reporting only those elements that concern the present analysis.

Common to most of the presented studies is the notion of HRTF *perceptual quality*. This term refers to the perceptual matching, localisation wise, between a participant and an HRTF. A low quality HRTF is one that results in bad localisation accuracy. Inversely, the higher the quality, the better the localisation accuracy, the highest quality match corresponding in theory to one's own HRTF. Replicating the potential outcomes of selecting an HRTF from an existing database, three degrees of perceptual matching are considered in these studies in addition to individual HRTF: *worst-match*, *random-match*, and *best-match* HRTF. Best and worst-match HRTFs represent respectively a best and worst case outcome, typically obtained by asking participants to perform a localisation task with, or a perceptual ranking of, an existing set of HRTFs.
