The purpose of this study was to construct a simplified model of previous experiments in which pinnae convolutions were demonstrated to affect the accuracy of sound localisation in the vertical plane.

The Role of Spectral Complexity and Pinna Spectral Transformations in Vertical Sound Localisation

Abstract

The purpose of this study was to construct a simplified model of previous experiments in which pinnae convolutions were demonstrated to affect the accuracy of sound localisation in the vertical plane. Monash University psychology randomly selected from different laboratory classes formed the data pool of 12. Participants were required to verbally report, after listening to either 4kHz or broadband noise with normal or occluded pinnae, the speaker believed to be the source of the sound. Performance results were compared to results expected by chance. The results confirm the pinnae as an important element of vertical sound localisation, as well as illustrating that spectrally complex sounds are far more localizable in the vertical plane than pure tones.

The external human ear (pinna) and its relationship with localisation of sound has been the basis of much study over the last three decades with limited investigation before this time. It is widely agreed that humans posses a remarkably well developed spatial perception system (Wright, Hebrank, & Wilson, 1974; Hartmann, 1983; Chen, Van Veen, & Hecox, 1992); however until recently it was held that the pinna had only a minor effect on the process of hearing sound, merely acting as a sound funnel to the inner ear (Batteau, 1967). Notwithstanding, recent research studies into sound localisation have established that the pinnae play an important role in the localisation of sound (Batteau, 1967; Butler, 1969; Gardner & Gardner, 1973).

The pinnae are considered by many to be especially important when determining the location of a sound source in the vertical plane (Butler, 1969; Gardner & Gardner, 1973). Batteau (1967) was one of the first to demonstrate the effects of the pinnae on localisation accuracy. This was illustrated by recording sounds, received by small microphones inserted into casts of pinnae (either normal pinnae or occluded pinnae, e.g. placing a small tube in each ear canal and filling the pinnae with clay to prevent sound interacting with its convolutions), then playing the recorded sound back to participants via high-fidelity earphones (sound is heard coming from sources outside the head). In this study participants were found to be able to make reasonable judgments as to the horizontal and vertical components of the sound when the pinnae were normal; accuracy decreased significantly when the pinnae were occluded. Batteau came to the conclusion that the convolutions of the pinnae resulted in the production of time delay paths, transforming the incoming signal, allowing the listener to predict the source of the sound in horizontal and vertical planes, depending upon the transformation. Further studies have revealed similar results to Batteau (1967); focusing mainly on the spectral transformation caused by the pinnae (by comparing participant judgment from normal pinnae and occluded pinnae conditions) these results support the important role spectral transformations serve in sound localisation in the vertical plane (Gardner & Gardner, 1973; Wright et al, 1974; Blauert, 1983).

Further studies have identified other factors which influence the accuracy of vertical sound localisation. Many have found that when pure tones are used, localisation of sound in the vertical plane is inhibited, however, when spectrally complex sounds (e.g. broadband noise) containing high frequencies are used, localisation of sound is reasonably accurate (Thurlow & Runge, 1967; Gardiner & Gardiner, 1973; Wright et al, 1974; Musicant & Butler; 1984). Musicant and Butler (1984) found that high frequency sounds were crucial for accurate localisation performance, as high frequency sounds undergo elevation-dependant spectral transformations as a result of the structures of the pinnae, using these spectral transformations to judge localisation of sound in the vertical plane.

The present study is based on a simplified modeled of previous experiments in which pinnae convolutions were demonstrated to affect the accuracy of sound localisation in the vertical plane. The primary aim is to demonstrate that participants will have the ability to judge the localisation of spectrally complex sounds, but not pure tones, in the vertical plane; and that these judgments will yield significantly superior performance than those expected by chance. Furthermore, in line with previous research, the irregularities of the pinna are predicted to facilitate a significantly increased vertical sound localisation performance compared to those results expected by chance. ...

This is a preview of the whole essay

Method

Participants

Members of a third year psychology course at Monash University participated as part of a laboratory class assessment requirement. The data pool consisted of 12 participants each randomly selected from a different laboratory class. Age and sex of participants was not recorded. None of the participants had a known hearing impairment. See Appendices A-C for raw data.

Materials

Five speakers were arranged in a vertical arc, numbered ‘1’ to ‘5’ from top to bottom, to provide a 12 degree separation between speakers equidistant, at a distance of 2 metres, to the participants head. A height adjustable chair was used to counterbalance different heights amongst participants, positioning speaker 3 at the participants’ interaural axis (directly ahead at ear level). The sound generator employed produced sound bursts, of either 4-kHz tone or broadband noise, with durations of 400msec, a rise time of approximately 40ms, and of approximately 70dB (A) SPL at the position of the participants head. The sound stimulus was able to be presented through any one of the 5 speakers on a particular trial though the utilization of a selector switch on the sound generator.

Lengths of rubber tubing (~1.5cm) were used to provide an uninterrupted path to the ear canal when “Otoform” silicon ear impression compound was employed to occlude the pinnae. There was optional use of a blindfold.

Procedure

The participant, seated in the chair, was administered trials under each of the sound stimulus conditions (4kHz, Noise Normal Pinna, then Noise Pinna Occluded) and verbally reported which speaker they judged the sound stimulus to originated from (“1” to “5”), as each of the five speakers presented the stimulus, in turn, in pseudo-random order (see Response Sheet, Appendix D); their responses were noted on the response sheet. It is important to note that the experimenter started at a different trial number in each condition to prevent the participant from learning the sequence.

Under each condition the participant was given 15 practice trials in which the experimenter gave feedback by indicating whether the response was “correct” or “incorrect” (identifying the correct speaker number if incorrect); practice trial responses were not recorded. Fifty test trials followed for each condition where no feedback was given.

In the Noise Pinna Occluded condition the tubing was placed into the ear canals of the participant and the convolutions of the pinnae, around the tubing, filled with Otoform. The trials were then administered as above.

Results

Means and standard deviations as a function of number of correct responses are presented in Table 1 to test whether spectrally complex sounds with pinna normal can be localized in the vertical plane better than pinna occluded and pure tones, relative to chance.

Table 1

Means and Standard Deviations as a Function of Number of Correct Responses

From Table 1 it can be seen that the number of correct responses (NCR) were much higher for NPN than both NPO and 4 kHz conditions. An independent measures t-test used on each condition determined the relationship between the obtained NCR and the NCR expected by chance (10), evaluated against an alpha level of .01. The difference was significant between NPN and chance, t(11) = 10.05, p<.01, two-tailed. However, there was no significant difference in both 4kHz and NPO conditions and chance; t(11) = .52, p>.01, two-tailed, and t(11) = .31, p>.01, two-tailed, respectively. Details of t-test calculations are presented in Appendix E.

Further analysis carried out examined the distribution of errors from the NCR and the influence it exuded on the significance of the three conditions relative to chance. This analysis is presented in Table 2; means and standard deviations are displayed as a function of the mean squared deviation (MSD) from NCR.

Table 2

Means and Standard Deviations as a Function of the Mean Squared Deviation from Correct Responses

From Table 2 it can be seen that the MSD from NCR was much lower for NPN than both NPO and 4 kHz conditions. An independent measures t-test used on each condition determined the relationship between obtained MSD from NCR and the MSD from NCR expected by chance (4), evaluated against an alpha level of .01. Difference from chance was significant in both NPN and NPO conditions; t(11) = -29.54, p<.01, two-tailed, and t(11) = -4.70, p<.01, two-tailed, respectively. However, there was no significant difference between the 4kHz condition and chance; t(11) = -.27, p>.01, two-tailed. Details of t-test calculations are presented in Appendix E.

Unexpected significance, of NPO in the MSD from NCR data, prompted additional analysis to examine whether there was a significant difference between NPN and NPO, in the MSD from NCR data, a paired samples t-test was performed. This difference was significant; t(11) = -12.58, p<.01, two-tailed. Details of additional t-test calculations are presented in Appendix F.

Discussion

The aim of this study was to reproduce aspects of previous experiments, in which the convolutions of the pinnae were demonstrated to affect the accuracy of sound localisation in the vertical plane, in a simpler design. As hypothesized, it was found that participants had better performance judging vertical sound localisation of spectrally complex sounds than expected by chance. Also, as predicted, participants performed no better than expected by chance when judging the vertical sound localisation of pure tones. However, the third and final hypothesis, that the irregularities of the pinna would facilitate significantly increased vertical sound localisation performance compared to those results expected by chance, was rejected.

It was found, contrary to prediction that both the NPN and NPO conditions performed significantly better than expect by chance in the MSD from NCR data (which took error responses into account). This result initially suggests that the convolutions of the pinnae had no effect on vertical sound localisation. Further analysis was carried out to examine this unexpected occurrence, investigating whether there was a significant difference between vertical sound localisation performance in NPN and NPO conditions. This analysis yielded that there was significantly superior performance in the NPN compared to the NPO condition; establishing that the convolutions of the pinnae did have a significant effect on vertical sound localisation performance, even though the NPO condition performed better than expected by chance.

There are several possible explanations for why the NPO data performed better than expected by chance. In a study undertaken by Hebrank & Wright (1974) it was found that participants who had an earplug placed in their ear canal (before being administered vertical sound localisation trials) did not retain impaired localisation ability for an extended period of time, instead participants became accustom to the device and after several trials were yielding results similar to if they were not wearing the plug at all. These results suggest a practice effect. Even though the participants in the current study were required to have both pinnae totally occluded, the presence of a practice effect as a result of the 50 consecutive trials with the pinnae occluded cannot be overlooked as a possibility.

Another explanation, although few studies relating to vertical localisation have been published on it, is that the 400ms duration of the sound bursts may have been simply too long, overriding the pinnae in its vertical sound localisation role and giving way to other spatial cues such as reflections from interior aspects of the room, such as tables, chairs, walls, floor, ceiling and onlookers (Hartmann, 1983). Because the rooms in which the tests were administered in (university class rooms) were not under anechoic conditions this is a plausible explanation.

The current study possessed several methodological shortcomings which, if managed appropriately in future studies,

may alter and add additional insight to the results. These shortcomings included non-anechoic conditions as discussed previously, possibility of practice effects, and the fixed testing positions of the speakers. The possibility of a practice effect as described earlier could have been eliminated if the conditions were administered in a random mixed order (e.g. trial one 4kHz, trial 2 NPN, trial 3 NPO, trial 4 NPN etc.); although this randomization would depend on the type of pinnae occlusion (it would be troublesome to be placing/removing silicon ear impression compound from the pinnae for each trial). Additionally the fixed position of the speakers may have served as a localisation clue, influencing results; as demonstrated in Gardiner and Gardiner (1973) the speakers could be interchanged at random intervals to counteract this effect.

The results of this study confirm previous research that present the pinnae as an important element of vertical sound localisation, as well as illustrating the observation that complex sounds are far more localizable in the vertical plane than pure tones. These finding have the potential to be used in real life situations, for example the use of high frequency sounds in emergency devices such as whistles, alarms and sirens may be more effective than use of low frequency sounds, however further investigation into the horizontal sound localisation plane in addition to the vertical sound localisation plane would be needed.

References

Batteau, D. W. (1967). The role of the pinna in human localization. Proceedings of the
Royal Society of London, 168, 158-180.

Blauert, J. (1983). Spatial hearing: The psychophysics of human sound localization. Cambridge: MIT Press.

Butler, R. A. (1969). Monaural and binaural localization of noise bursts vertically in

the median saggital plane. Journal of Auditory Research, 3, 230-235.

Chen, J., Van Veen, B. D., & Hecox, K. E. (1992). External ear transfer function modeling: A beamforming approach. Journal of the Acoustical Society of America, 92, 1933- 1944.

Gardner, M. B., & Gardner, R. S. (1973). Problem of localization in the median plane: effect of pinnae cavity occlusion. Journal of the Acoustical Society of America, 53, 400-408.

Hartmann, W. M. (1983). Localization of sound in rooms. Journal of the Acoustical Society of America, 74, 1380-1391.

Hebrank, J., & Wright, D. (1974). Are two ears necessary for localization of sound sources on the median plane? Journal of the Acoustical Society of America, 56, 935-938.

Musicant, A., & Butler, R. (1984). The influence of pinnae-based spectral cues on sound

localization. Journal of the Acoustical Society of America, 75, 1195-1200.

Roffler, S., & Butler, R. A. (1968). Factors that influence the localisation of sound in the vertical plane. Journal of the Acoustical Society of America, 43, 1255–1259.

Thurlow, W. R., & Runge, P. S. (1967). Effect of Induced Head Movement on Localisation. Journal of the Acoustical Society of America, 42, 480-488.

Wright, D., Hebrank, J. H., & Wilson, B. (1974). Pinna reflections as cues for localization. Journal of the Acoustical Society of America, 56, 957-962.