The auditory dorsal pathway: Orienting vision

Abstract

A particularly prominent model of auditory cortical function proposes that a dorsal brain pathway, emanating from the posterior auditory cortex, is primarily concerned with processing the spatial features of sounds. In the present paper, we outline some difficulties with a strict functional interpretation of this pathway, and highlight the recent trend to understand this pathway in terms of one that uses acoustic information to guide motor output towards objects of interest. In this spirit, we consider the possibility that some of the auditory spatial processing activity that has been observed in the dorsal pathway may actually be understood as a form of action processing in which the visual system may be guided to a particular location of interest. In this regard, attentional orientation may be considered a low-level form of action planning. Incorporating an auditory-guided motor aspect to the dorsal pathway not only offers a more holistic account of auditory processing, but also provides a more ecologically valid perspective on auditory processing in dorsal brain regions.

Introduction

The ‘decade of the brain’ brought with it many advances in our understanding of the auditory cortical system. By the early 1990s it had become evident that cells in the auditory cortex were not functionally homogeneous. Single-cell research in non-human primates established that the primary auditory cortex (i.e., AI) was surrounded by a ‘belt’ area, which in turn was bordered laterally by a ‘parabelt’ region (Kosaki et al., 1997, Morel et al., 1993, Rauschecker, 1997). Whereas neurons in the core (i.e., AI) responded best to simple auditory stimuli such as pure tones, those in the belt and parabelt responded best to more complex sounds (e.g., vocalizations and bands of noise, Rauschecker et al., 1995). Most remarkably, caudal but not rostral neurons within these areas, were shown to be particularly sensitive to the location of sounds (Benson et al., 1981, Kaas and Hackett, 2000, Leinonen et al., 1980, Morel et al., 1993, Rauschecker, 1998, Recanzone et al., 2000, Tian et al., 2001, Vaadia et al., 1986), whereas more rostrally positioned neurons in the lateral belt and beyond were more sensitive to non-spatial acoustic qualities such as conspecific vocalizations (Rauschecker and Tian, 2000, Tian et al., 2001). Beyond auditory cortex, connections from the caudal parabelt regions were found to extend to the ventral inferior parietal (VIP) cortex of the monkey (Lewis and Van Essen, 2000) as well as reciprocally into the caudal principal sulcus (area 46) and frontal eye fields (area 8a, Hackett et al., 1999, Romanski et al., 1999a, Romanski et al., 1999b). By comparison, the rostral belt region was found to have connections with the frontal pole (area 10), rostral principal sulcus (area 46), and ventral prefrontal areas (areas 12 and 45, Romanski and Goldman-Rakic, 2002). Taken as a whole, these results were interpreted as forming the basis of a domain-specific model of auditory processing whereby auditory non-spatial (i.e., “what”) and spatial (i.e., “where”) information were processed by ventral and dorsal brain pathways, respectively (i.e., the “what–where” model or WW model, Kaas and Hackett, 1999, Rauschecker and Tian, 2000, Romanski et al., 1999b).

This auditory model bore many similarities to the “what–where” model of the visual cortical system developed over a decade earlier in which two visual pathways were defined as emanating from the striate cortex: a ventral stream crucial for visual identification of objects, and an occipitotemporal–parietal pathway crucial for appreciating the spatial relationships among objects as well as for the visual guidance of movements towards objects in space (Ungerleider and Haxby, 1994, Ungerleider and Mishkin, 1982). Support for that model was derived from research demonstrating that lesions to monkey inferior temporal cortex resulted in deficits on pattern, object, or colour discrimination tasks, but not on visuospatial tasks such as visually guided reaching or relative distance judgments (see Ungerleider and Mishkin, 1982). Lesions to the posterior parietal cortex, on the other hand, did not seem to affect visual discrimination performance but did affect visuospatial performance. Physiological data also supported the distinction (Desimone and Ungerleider, 1989, Maunsell and Newsome, 1987), as did data from humans (Haxby et al., 1991, von Cramon and Kerkhoff, 1993).

Not long after the auditory findings in animals were reported, a similar pattern of auditory processing was revealed in humans. In one of the first experiments to offer double dissociative evidence in neurologically intact humans, we presented functional magnetic resonance imaging (fMRI) participants with noiseburst sounds that varied in pitch and perceived location (Alain et al., 2001). Relative to when listeners were asked to attend to the location features of the sounds, attention to the sounds’ pitch elicited greater blood-oxygen-level dependent (BOLD) hemodynamic activity in the auditory cortex and inferior frontal gyrus. In contrast, attending to the location properties elicited relatively greater activity in the participants’ posterior temporal lobe, superior parietal lobe and superior frontal sulcus. Event-related potentials recorded while the listeners carried out these tasks also revealed differential task effects over the anterior and posterior temporal regions. These, along with many other neuroimaging and patient observations, strongly argued in favour of a dual pathway model of human auditory cortical processing, often termed the auditory ‘what–where’ model (see Fig. 1, Ahveninen et al., 2006, Alain et al., 2008, Altmann et al., 2007, Anourova et al., 2001, Arnott et al., 2004, Arnott et al., 2005, Barrett and Hall, 2006, Belin and Zatorre, 2000, Bushara et al., 1999, De Santis et al., 2007, Degerman et al., 2006, Deouell et al., 2007, Maeder et al., 2001, Schröger and Wolff, 1997, Tardif et al., 2008, Tata and Ward, 2005, Thiran and Clarke, 2003, Weeks et al., 1999, Zatorre et al., 2002).

Despite its popularity, the auditory WW model has not been without criticism (Belin and Zatorre, 2000, Hall, 2003, Recanzone and Cohen, 2010). One concern has been that a distinction based on ‘what’ and ‘where’ features of a sound tends to be an oversimplification of the data and does not provide an adequate functional account of how the auditory system and the brain, in general, operates. Moreover, it has become apparent that non-spatial auditory processing can also elicit activation in the dorsal pathway, albeit often to a lesser extent than does spatial processing (cf. Arnott et al., 2004, Gifford and Cohen, 2005, Husain and Nachev, 2007, Renier et al., 2009), and that, in some cases, the parietal neurons that are activated by auditory spatial tasks can be the same as those activated by non-spatial tasks (Gifford and Cohen, 2005). For example, our meta-analysis of auditory neuroimaging studies demonstrated that unlike regions of the superior frontal sulcus (SFS) that were driven exclusively by auditory spatial processing, over 40% of auditory studies considered to be ‘non-spatial’ in nature (i.e., the auditory stimuli were only ever delivered from one location and the listener's task did not involve any kind of localization or spatial judgment), reported significant functional activity in the inferior parietal lobe (IPL) region in addition to the virtually 100% of auditory ‘spatial’ processing studies that were examined (Arnott et al., 2004). Accordingly, while it is evident that the auditory dorsal pathway, and the IPL in particular, is activated any time a relative auditory spatial judgment has to be made, it is also apparent that the IPL is involved in more than just creating auditory spatial maps.