The over-estimation of distance for self-voice versus other-voice

Wen Wen, Yuta Okon, Atsushi Yamashita, Hajime AsamaScientific 2022 · DOI

Abstract

Self-related stimuli are important cues for people to recognize themselves in the external world and hold a special status in our perceptual system. Self-voice plays an important role in daily social communication and is also a frequent input for self-identification. Although many studies have been conducted on the acoustic features of self-voice, no research has ever examined the spatial aspect, although the spatial perception of voice is important for humans. This study investigated people's distance perception of their own voice when heard from an external position. Participants heard their own voice from one of four speakers located either 90 or 180 cm away, either immediately after speaking (active session) or hearing a replay (replay session). Their voices were either pitch-shifted (other-voice) or unaltered (self-voice). Results showed that self-voice from closer speakers was misattributed to further speakers at a significantly higher proportion than other-voice. This effect was associated with delusional thoughts and distorted body image scores in active sessions, providing insights into how self-related stimuli processing may link to psychosis risk.

Introduction

We are intimately familiar with the sound of our own voices, typically perceived through a combination of air and bone conduction. When we hear a recording of ourselves, it often sounds unfamiliar because the bone conduction component is absent. While researchers have extensively studied the acoustic features of self-voice, such as pitch and timbre, its spatial perception remains largely unexplored. In an era of digital communication and virtual reality, we frequently encounter our "self-voice" originating from locations other than our own mouths. This study seeks to understand how the brain processes this spatial discrepancy and whether our sense of agency—the ability to monitor our own actions—influences this perception.

Methodology

The researchers conducted a controlled auditory localization experiment with 31 participants (24 of whom passed baseline screening). The experimental setup was housed in a sound-insulated booth featuring four speakers positioned at two distinct distances: 90 cm (near) and 180 cm (far).

Experimental setup and trial procedure showing the speaker arrangement and the sequence of events from vocalization to spatial judgment.

Participants uttered a short "ah" sound and then heard voice feedback from one of the speakers. The experiment utilized two primary conditions:

  1. Self-voice: The participant's own voice, unaltered.
  2. Other-voice: The participant's voice pitch-shifted by ±4 semitones, which subjective ratings confirmed sounded like a different person.
Acoustic stimuli comparison between the original voice and pitch-shifted versions used to represent the 'other-voice' condition.

Trials were divided into "Active" sessions (feedback heard 300ms after speaking) and "Replay" sessions (listening to recordings without speaking). Participants identified the source speaker and completed psychological profiling using the Peters Delusion Inventory (PDI), Magical Ideation Scale (MIS), and Perceptual Aberration Scale (PAS).

Key Results

The Self-Voice Distance Bias

The study revealed a significant spatial distortion: participants were much more likely to misattribute their own voice from a near speaker (90 cm) to a far speaker (180 cm). This overestimation was significantly less frequent when hearing the pitch-shifted "other" voice. For the closer speakers, accuracy for self-voice was significantly lower than for other-voice ($p = .020$). Interestingly, when the voice originated at 180 cm, there was no significant difference between conditions, likely because no further speakers were available for misattribution.

Psychological Correlations

Structural Equation Modeling (SEM) indicated that the tendency to overestimate distance was correlated with higher scores in delusional ideation and body image distortion. In the active session, PDI scores (standardized coefficient = 0.405) and PAS scores (0.464) significantly predicted the degree of self-voice distance overestimation ($ps < .05$). Notably, this relationship vanished in the passive replay session, suggesting that the mechanism for monitoring self-generated actions (the "forward model") is the critical link between spatial perception and these psychological traits.

Discussion and Future Directions

The findings suggest that the brain maintains a strong "prior" expectation that our voice should originate from our own head. When it originates 90 cm away, the resulting prediction error causes the brain to over-amplify the perceived distance.

While this study provides a novel perspective on self-voice processing, it is limited by a relatively small sample size (n=24 for behavioral data) and the use of only two distance intervals. Future research could include a wider range of distances to determine if the overestimation effect plateaus. Additionally, testing these parameters in clinical populations could determine if spatial overestimation serves as a behavioral marker for the risk of auditory hallucinations or psychosis. By demonstrating that our "self" is spatially anchored to our expectations, this research offers new insights into how the brain maintains the boundary between the internal and external world.

Summary updated

Abstract

Self-related stimuli are important cues for people to recognize themselves in the external world and hold a special status in our perceptual system. Self-voice plays an important role in daily social communication and is also a frequent input for self-identification. Although many studies have been conducted on the acoustic features of self-voice, no research has ever examined the spatial aspect, although the spatial perception of voice is important for humans. This study proposes a novel perspective for studying self-voice. We investigated people's distance perception of their own voice when the voice was heard from an external position. Participants heard their own voice from one of four speakers located either 90 or 180 cm from their sitting position, either immediately after uttering a short vowel (i.e., active session) or hearing the replay of their own pronunciation (i.e., replay session). They were then asked to indicate which speaker they heard the voice from. Their voices were either pitch-shifted by ± 4 semitones (i.e., other-voice condition) or unaltered (i.e., self-voice condition). The results of spatial judgment showed that self-voice from the closer speakers was misattributed to that from the speakers further away at a significantly higher proportion than other-voice. This phenomenon was also observed when the participants remained silent and heard prerecorded voices. Additional structural equation modeling using participants' schizotypal scores showed that the effect of self-voice on distance perception was significantly associated with the score of delusional thoughts (Peters Delusion Inventory) and distorted body image (Perceptual Aberration Scale) in the active speaking session but not in the replay session. The findings of this study provide important insights for understanding how people process self-related stimuli when there is a small distortion and how this may be linked to the risk of psychosis.
BibTeX
@inproceedings{wen2022over,
  author = {Wen Wen and Yuta Okon and Atsushi Yamashita and Hajime Asama},
  title = {{The over-estimation of distance for self-voice versus other-voice}},
  booktitle = {Scientific},
  year = {2022},
  doi = {10.1038/s41598-021-04437-8},
}
Figure