Autistic Traits Modulate Self-Recognition in the Auditory Domain

Background:  Atypical self processing is an emerging theme in autism research, suggested by lower self-reference-effect in memory, and atypical neural responses to visual representations of self face/body in individuals with autism. Most research onphysical self-processing in autism uses visual stimuli (e.g. self-face). However, the self is a multi-modal construct, and therefore it is essential to test self-recognition in other sensory modalities such as the auditory modality. Self recognition in the auditory domain remains relatively unexplored, and has not been tested in relation to autism and related traits.

Objectives:  To develop a metric of auditory self recognition, we created a series of morphs of self-voice with other voices. We then tested if autistic traits modulated this metric of auditory self recognition.

Methods:  In the first session 31 neurotypical adults (7 Males) participants’ voices were recorded as they uttered a train of mono syllables /ba/. The individual participant’s voice was then preprocessed and morphed with two unfamiliar (‘other’) voices, creating a continuum of morphs.  In the second session, normally a few days later, participants performed a ‘self-voice’ identification task, classifying each morph as ‘self’ voice or an ‘other’ voice. A total of 88 morphed voice stimuli were presented, from 100% self to 0% self, in steps of 10%.  All participants also completed the Autism Spectrum Quotient (AQ) online. ‘Self-Voice’ responses for each participant at each morph level were computed and a response curve was generated. The Area Under Curve (AuC) was computed for each participant which represented the area between the maximum self response (maximum stimuli recognised as belonging to self) to minimum self response (minimum stimuli labelled as belonging to self), and was used as a metric of the extent of auditory self recognition (i.e. the higher the AuC, the poorer the discrimination between self and other). 

Results:  

Individuals were able to recognize self voice from a single syllable utterance which contained no semantic information. However we found there are individual differences in judging the shift from self voice to other voice as indicated by the different morph levels at which this shift occurs. (Mean 27.5±15).

Crucially, the metric of auditory self-recognition as represented by the AuC for self-voice recognition was negatively correlated to the AQ scores (r= -.395, p=0.014).

There was no significant difference betweenthe two different unfamiliar voices on overall judgment of self vs other.

Conclusions:  Our data shows individuals with higher autistic traits were better able to discriminate self from other voices.  This may be due to the fact that our ‘self-voice’ stimuli represented the most basic voice utterance without any semantic content. This observation could arise due to both a) enhanced perceptual processing of auditory stimuli (as suggested by (Bonnel et al, 2003)and (Järvinen-Pasley et al, 2008)), as well as b) a narrower extent of self-other overlap in individuals with Autism (Uddin et al., 2008).