Tone Deafness: A New Disconnection Syndrome?

Discussion

The study of vocal pitch abilities as they relate to pitch perception abilities is a relatively neglected area of investigation. Although previous documentation exists of individuals who can sing accurately despite poor perception (Ayotte et al., 2002), a systematic mismatch between perception and production has only recently been reported in a group of tone-deaf individuals (Loui et al., 2008). Using diffusion tensor tractography combined with perception and production psychophysics, we identified neuroanatomical disconnections in the tone-deaf population. Here, we show that people who are tone deaf are affected by a previously unknown disconnection of the AF. Diffusion tensor tractography (Mori et al., 1999) and psychophysical assessments (Patel et al., 2005) were performed on tone-deaf individuals who have impaired pitch perception (Foxton et al., 2004) and vocal pitch production (Loui et al., 2008) and matched controls. We identify a lack of neural connectivity in the tone-deaf brain: robust decreases in AF volume were observed in tone-deaf individuals relative to matched controls. Furthermore, we demonstrate for the first time that the AF is a neural correlate of auditory–motor behavior: its superior and inferior branches predict psychophysically assessed pitch discrimination and sound production-perception abilities, respectively. This newly identified neural abnormality suggests that tone deafness is a new disconnection syndrome resulting in impaired pitch perception and vocal sound production. Results support a dual-stream anatomy of sound production and perception (Hickok and Poeppel, 2007) that may be required for all types of vocal communication. By identifying behavioral correlates of white matter differences observed via diffusion tensor imaging (Jones, 2008), results contribute to our understanding of the role of neural connectivity in human behavior.

Results from the present study converge with previous morphometric studies (Hyde et al., 2007; Mandell et al., 2007) showing frontotemporal gray matter differences between tone-deaf and control individuals. The finding of abnormal projections in the right hemisphere fits with Hyde et al. (2006), showing white matter morphometric abnormalities on the right hemisphere, although voxel-based morphometric studies focusing on gray matter differences have either shown strong left-hemisphere abnormalities in the frontotemporal network (Mandell et al., 2007) or found evidence for cortical thickness abnormalities in both hemispheres (Hyde et al., 2007). Although most of the observed effects in the current study are on the right hemisphere, the observed significance levels may be attributable to fiber volumes being smaller overall on the right (and thus to differences being more easily detectable) and thus may not be indicative of an exclusively right-hemispheric network subserving pitch perception and production. Future studies are needed to further explore the relationships between gray and white matter abnormalities within and between hemispheres in this interesting developmental disorder.

Compared with the normal population, tone-deaf individuals possessed fewer fibers in the arcuate fasciculus, with at least one of the superior AF branches being unidentified with the current imaging parameters. Furthermore, superior AF tract volume was a significant predictor of conscious pitch discrimination ability, whereas inferior AF volume predicted the degree of action-perception mismatch. Both of these correlations were observed even while excluding profoundly tone-deaf individuals with pitch discrimination thresholds of over one semitone so as to minimize the chances of an artificially boosted correlation. This may suggest that the neural correlates of conscious pitch discrimination and action-perception identified here are applicable not only to tone-deaf individuals but can be generalized to the normal population.

Fiber tracking with DTI is not directly analogous to studies using tracers such as horseradish peroxidase where the connectivity of axons is assured; here, the real number of axons within a tract relative to the number of reconstructed fibers is unknown (Johansen-Berg and Behrens, 2006). However, fiber tracking can be used to identify sufficiently large tracts of axons traveling in the same general direction, and tract-based statistics have been applied in various normal and diseased populations as a proxy for structural integrity and hemispheric asymmetry of major white matter pathways (Johansen-Berg and Behrens, 2006; Leh et al., 2006; Schaechter et al., 2008; Sundaram et al., 2008). In our results, the diminished number and volume of detectable superior AF fibers may not necessarily imply a complete absence of the tract but rather that the preferred directions of diffusion along the path readily detected in controls are discontinuous in tone-deaf subjects; this may be indicative of reduced fiber volume, decreased myelination along the tract direction, or abnormal development of fiber projections originating from one of the regions. Thus, the origin of this disorder could be in regionally abnormal neuronal migration leading to reduced structural connectivity, similar to what is observed in congenital prosopagnosia (Thomas et al., 2009), dyslexia (Deutsch et al., 2005), or global developmental delay (Sundaram et al., 2008). Furthermore, reduced connectivity in tone-deaf controls may not represent a categorical difference from the normal population but rather a quantitative difference in which the tone-deaf arcuate fasciculus represents the low end of a continuum. Relative to controls, the structural integrity of white matter tracts is so compromised in tone-deaf individuals as to preclude successful tracking of the superior branch of the arcuate fasciculus using currently available neuroimaging and analysis techniques.

Methodologically, the present data suggest that tractography methods can reveal structural integrity not only of white matter in general but of tracts that connect distributed nodes within a structural brain network (Jones et al., 1999). The directionality of these fiber tracts remains to be determined and presents a challenge for current diffusion tractographic methods. Another caveat in this relatively novel technique lies in the statistical inference of fiber volume from diffusion data. In that regard, the present analysis methods, involving the validation of fiber tracts connecting seed regions of interest, parallel those of recent studies comparing diffusion data with dissected postmortem brain tissue or across various neurological populations such as patients of diffuse axonal injury (Wang et al., 2008), children with global developmental delay (Sundaram et al., 2008), congenital prosopagnosics (Thomas et al., 2009), and hemispherectomized patients with and without blindsight (Leh et al., 2006).

Another question concerns directionality of the newly identified disconnection syndrome: whether the abnormality of tone deafness originates from the temporal or frontal regions. As a two-region of interest analysis does not yield information about whether fiber tracts were originating from or projecting to the temporal and frontal regions (Leh et al., 2006), we included a one-region of interest analysis, with seed regions placed at the endpoints of the arcuate fasciculus, to identify all tracts passing to or from the pSTG or pIFG. Resultant fibers showed selective hypoconnectivity from the pSTG to the pIFG, suggesting that the pSTG is the origin of the disorder (Fig. 3); however, as tract volume using one-ROI analyses cannot be constrained to specific branches of the fiber bundle, we report tract volume statistics from two-ROI analyses as the more accurate measure of specific branches of fiber tracts. Test–retest reliability of diffusion tractography has been established by comparing tracts obtained from the same subjects using multiple sets of imaging parameters (Loui and Schlaug, 2009). Combined with threshold data for perception and vocal production obtained from each subject, the present results show for the first time that AF volume can be an effective predictor of vocal behavior.

Although pitch perception and production is a basic ability in humans and other animals, an estimated 17% of the normal population self-identifies as tone deaf (Cuddy et al., 2005). The present results show that this seemingly normal population possesses disrupted pathways bridging perception and action areas in the brain, resulting in an inability to discriminate and produce pitches to sing in tune. The data suggest that action and perception may arise from dissociated dorsal and ventral neural pathways that may be differentially affected in tone deafness. As the posterior inferior frontal gyrus (specifically Broca's area) was shown to be recruited in visuospatial processing involving sequentially manipulating mental objects (Sluming et al., 2007), the present finding of diminished connectivity to the posterior IFG might be related to recent reports of impaired mental rotation among tone-deaf people (Douglas and Bilkey, 2007). Furthermore, the dorsal and ventral pathways may be responsible for automatic, category-based sound analysis and conscious access to perceptual information, respectively, both of which are important in vocal communication. Results parallel existing dual-stream models of speech, language, and auditory processing (Hickok and Poeppel, 2004; Sridharan et al., 2007; Glasser and Rilling, 2008; Griffiths, 2008) and may provide the neurological basis for a domain-general neural framework that subsumes auditory and motor systems in the human brain.