Por reconocimiento de otras actividades:

It has been known for decades that a general specialism of the left hemisphere for speech processing and production exists (Berker et al. 1986). Since these initial findings, much progress has been made in modelling the systems underly- ing speech and language (Hickok and Poeppel 2007; Hickok 2012). Technological developments in structural and functional brain imaging have enabled detailed

studies of the brain and gross structural hemispheric asymmetries have been found to exist in human cortex (Kong et al. 2018; Chiarello et al. 2016). Auditory and language areas have been specifically examined and these too show structural asymmetries (Abdul-Kareem and Sluming 2008; Moerel et al. 2014; Meyer et al. 2014; Galuske et al. 2000). In fact, auditory area PT has been described as the most asymmetrical cerebral structure in the whole brain (Prete et al. 2016). Investigations measuring structural connectivity have found that the arcuate fas- ciculus (the white matter bundle connecting Wernicke’s and Broca’s areas; key nodes in the language network) is more developed in the left than the right hemi- sphere, and this was related to behavioural differences (Catani et al. 2005, 2007). Another asymmetry of structural connectivity has recently been found where right auditory cortex was generally more integrated into the wider connectome than the left, particularly with respect to interhemispheric connectivity (Miˇsi´c et al. 2018). This result is consistent with functional network connectivity invest- igations and provides evidence that this structural asymmetry is associated with function (Andoh et al. 2015). Functional asymmetries of auditory areas have also been found, such as an asymmetry of sensitivity to temporal structure, with slow temporal modulations preferentially driving the right hemisphere (Boemio et al. 2005). Conversely, evidence has been found for preferential activation of the left hemisphere by rapid temporal modulations, such as those found in speech (Zaehle et al. 2004; Abrams et al. 2008). Multiple behavioural and imaging studies have provided evidence for functional asymmetries (Millman et al. 2011; Saoud et al. 2012; Liem et al. 2014; Han and Dimitrijevic 2015; Hugdahl 2011; Hugdahl and Westerhausen 2016; Schremm et al. 2018). Several theories have been developed to explain these asymmetries and concurrently account for the empirical findings. One prominent theory holds that the left hemisphere is specialised for processing temporal aspects and the right hemisphere is specialised for processing spectral aspects of sound (Zatorre and Belin 2001). In this study metabolic activity in the brain (measured with positron emission tomography) was compared when stimuli that differed in either spectral or temporal content were presented. The

responses showed that activation to rapid temporal changes was weighted towards the left hemisphere while activation to spectral changes was weighted towards the right hemisphere. This theory was later extended to explain music processing as primarily governed by the right hemisphere (Zatorre et al. 2002).

Alternatively, the Asymmetric Sampling in Time (AST) hypothesis proposes that structural asymmetry of auditory cortical regions is due to functional special- isms. It proposes that auditory cortex in each hemisphere preferentially processes aspects of a speech signal based on sampling windows of different lengths. As the brain has limited computational power, it must sample the continuous ana- logue sound signals received by the ear as part of the encoding process. The AST hypothesis accounts for the asymmetry by proposing that the right hemisphere samples and integrates information over 200–250 ms time windows, corresponding to ∼4 Hz and the left hemisphere samples and integrates information on much shorter timescales, ∼25 ms, corresponding to ∼40 Hz. Figure 1.13 shows how sampling windows at these lengths roughly correspond to a speech signal (Poep- pel 2003) This is a simplistic representation, however, as there is evidence for more optimised sampling where the timing of these sampling windows is adjus- ted based on acoustic properties of the signal (Doelling et al. 2014). The AST hypothesis was informed by observed oscillatory patterns in the theta (4–8 Hz) and low gamma (∼40 Hz) ranges during speech perception and there is mounting evidence for the role of this activity (Poeppel 2003; Giraud et al. 2007; Peelle and Davis 2012; Giraud and Poeppel 2012; Obleser et al. 2012; Luo and Poep- pel 2012). This temporal sampling account may reflect a more general principle underlying speech processing as suboptimal temporal sampling is thought to un- derpin language deficits in some disorders, such as dyslexia (Goswami and Leong 2013; Cutini et al. 2016) and autism spectrum disorder (O’Connor 2012). It is widely accepted that speech is processed asymmetrically and that extraction and tracking of the speech amplitude envelope is a key component of this mechanism (Kubanek et al. 2013; Ghitza et al. 2013).

Figure 1.13: Upper: representative sampling windows of 25 ms (40 Hz). Lower: representative sampling windows of 250 ms (4 Hz). The speech sample is the same as in Figure 1.1.

counts are too reductionist; reflecting the general trend in neuroscientific theory (Krakauer et al. 2017). Though their parsimony is compelling, it has been ar- gued that they may not fully account for known properties of the speech signal (McGettigan and Scott 2012). An alternative account suggests that, on closer inspection, there may be a specialism of the right hemisphere for sounds that are longer or change more slowly but there may be no temporal specialism of the left hemisphere (Scott and McGettigan 2013). Further, there is contrasting evidence for speech segmentation using ∼40 Hz sampling windows, as shown by time-reversion studies and the presence of phonetic components longer than the proposed window (Saberi and Perrott 1999; Ueda et al. 2017; McGettigan and Scott 2012).

the auditory cortical regions. These asymmetries are both structural and func- tional, and interhemispheric connectivity may be a crucial underlying component. Multiple accounts have been proposed to explain these asymmetries but there is no current consensus on the exact role they play in speech processing.