Elsevier

Brain and Language

Volume 127, Issue 2, November 2013, Pages 251-263
Brain and Language

Interaction between dorsal and ventral processing streams: Where, when and how?

https://doi.org/10.1016/j.bandl.2012.08.003Get rights and content

Abstract

The execution of complex visual, auditory, and linguistic behaviors requires a dynamic interplay between spatial (‘where/how’) and non-spatial (‘what’) information processed along the dorsal and ventral processing streams. However, while it is acknowledged that there must be some degree of interaction between the two processing networks, how they interact, both anatomically and functionally, is a question which remains little explored. The current review examines the anatomical, temporal, and behavioral evidence regarding three potential models of dual stream interaction: (1) computations along the two pathways proceed independently and in parallel, reintegrating within shared target brain regions; (2) processing along the separate pathways is modulated by the existence of recurrent feedback loops; and (3) information is transferred directly between the two pathways at multiple stages and locations along their trajectories.

Highlights

► A dual stream architecture has been found for vision, audition, and language. ► Some degree of interaction between the two processing networks is necessary. ► How the streams interact, both anatomically and functionally, remains little explored. ► This review examines evidence for the mechanisms underlying dual-stream interaction.

Introduction

The execution of a wide range of higher-order cognitive processes have been postulated to involve a division of labor between two anatomically segregated and functionally specialized parallel processing streams (Arnott et al., 2004, Goodale and Milner, 1992, Hickok and Poeppel, 2004, Hickok and Poeppel, 2007, Mishkin et al., 1983, Rauschecker and Tian, 2000, Ungerleider and Haxby, 1994, Ungerleider and Mishkin, 1982). The ventral pathway, connecting primary sensory cortices with temporal and prefrontal regions, has been functionally conceptualized as the “what” stream, responsible for visual and auditory stimulus identification and object recognition, the mapping of information onto conceptual representations, and the sound-to-meaning (and written word-to-meaning) mapping underlying the comprehension of spoken (and written) language. The dorsal pathway, connecting sensory areas with posterior/inferior parietal and prefrontal regions, constitutes the “where/how” stream, responsible for spatial processing (including location, relative position, and motion), sensorimotor mapping and the guidance of action towards objects in space, and within language production comprises a sound-motor interface responsible for mapping speech sounds onto motor representations for articulation. Although first postulated as an account of the structural and functional organization of processing in the visual domain (Baizer et al., 1991, Goodale and Milner, 1992, Mishkin et al., 1983, Ungerleider and Mishkin, 1982), a dual stream architecture has subsequently been proposed for an increasing number of cognitive functions including audition (Alain et al., 2009, Arnott et al., 2004, Kaas and Hackett, 1999, Rauschecker, 1998a, Rauschecker, 1998b, Rauschecker and Tian, 2000, Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999, Tian et al., 2001), language production and comprehension (DeWitt and Rauschecker, 2012, Friederici, 2009, Gow et al., 2009, Hickok and Poeppel, 2004, Hickok and Poeppel, 2007, Parker et al., 2005, Rauschecker and Scott, 2009, Saur et al., 2008, Saur et al., 2010, Scott and Wise, 2004, Turken and Dronkers, 2011, Ueno et al., 2011, Weiller et al., 2011), and attention (Umarova et al., 2010).

Within the initial conceptualization of the dual stream model postulated by Ungerleider and Mishkin (1982), processing along dual computational streams was argued to proceed independently and in parallel, enabling regions ‘downstream’ to manipulate and use the information supplied via the two pathways based on task requirements. Indeed, there is a substantial amount of empirical support for the anatomical dissociation and functional independence of the two cortical networks. Physiological and tract tracing studies in non-human primates have identified segregated dorsal and ventral projection pathways involving neuronal populations which demonstrate differential response selectivities for spatial and non-spatial stimulus attributes, in both the visual (Baizer et al., 1991, Kaas and Lyon, 2007, Livingstone and Hubel, 1988, Webster et al., 1994, Young, 1992), and auditory (Rauschecker and Tian, 2000, Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999, Tian et al., 2001) domains. For example, neurons within the auditory anterolateral belt cortex of the rhesus monkey have been found to be more selective for the type of monkey call presented, while neurons within the caudolateral belt area demonstrate greater sensitivity to the spatial location of the auditory stimulus, regardless of stimulus type (Tian et al., 2001). Coupled with this contrasting neuronal response selectivity, tract tracer studies have provided evidence for differential connectivity profiles between the two regions, with caudolateral belt areas providing projections dorsally to posterior parietal and dorsolateral prefrontal cortices, while in contrast, anterolateral belt areas project to ventral and orbital prefrontal areas (Rauschecker, 1998a, Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999).

Complementary evidence from functional imaging has also provided support for the existence of structurally segregated and functionally independent dorsal and ventral computational networks within the human brain (Barrett and Hall, 2006, Borowsky et al., 2005, Borowsky et al., 2006, Ikkai et al., 2011, Warren and Griffiths, 2003). A meta-analysis of functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) studies identified a preferential selectivity of ventral brain structures for the processing of non-spatial (‘what’) information, and dorsal structures for spatial (‘where/how’) information (Arnott et al., 2004). Specifically, activation co-ordinates for the processing of spatial information revealed activity in the inferior parietal lobe for all but one of the studies reviewed, compared to 41% of non-spatial studies. Within the temporal lobe, spatial processing activations were restricted to posterior regions, while activity was distributed throughout the temporal region for the processing of non-spatial stimulus characteristics. Differential activation patterns for spatial and non-spatial auditory processing was also observed in frontal regions, with the superior frontal sulcus revealing activation more consistently for spatial as opposed to non-spatial tasks (55% vs. 7%), while the inferior frontal region demonstrated the reverse pattern (9% vs. 56%).

Although a full discussion is beyond the scope of the current review, it is important to briefly outline the anatomical and functional models of dual stream processing postulated within each of the domains of vision, audition, and language.

Within vision, extensive empirical evidence regarding the neuroanatomical architecture of the primate visual system has revealed a highly complex system involving multiple brain regions and interconnections (Baizer et al., 1991, Distler et al., 1993, Felleman and Van Essen, 1991, Kaas and Lyon, 2007, Webster et al., 1994). The dorsal and ventral streams emerge from early visual areas (V1, V2, and V3), which provide activating inputs to both computational networks. Entry to the ventral stream begins at region V4, and projects along the temporal lobe via several routes to reach multiple regions including inferior temporal areas TEO and TE, and temporal polar area TG, as well as regions within the ventral prefrontal cortex including lateral prefrontal area 45 and orbitofrontal areas 11, 12, and 13. The dorsal stream begins at area V5/MT (middle temporal area), and projects via numerous connective pathways to regions including areas V3A, MST (medial superior temporal area), FST (fundus of the superior temporal area), and multiple parietal regions including the ventral and lateral intraparietal areas VIP and LIP, terminating in dorsal regions of the prefrontal cortex including area 46. Functionally, the two visual streams are segregated along the classic ‘what/where’ dichotomy, with the ventral ‘what’ stream associated with stimulus identification processes (including pattern and color vision), while the dorsal ‘where’ stream is implicated in visuospatial, visuomotor, and motion detection processes (Baizer et al., 1991, Ungerleider and Mishkin, 1982). The roles of the dual streams within visual processing have subsequently undergone conceptual refinement, with the two pathways defined by their overarching roles in perception and action, that is ‘what/how’ (Goodale, 2001, Goodale, 2011, Goodale et al., 2005, Goodale and Milner, 1992, Goodale and Westwood, 2004, Milner and Goodale, 2008). Within this framework, the ventral stream is engaged in producing a representation of the world for use in subsequent tasks including stimulus identification and memory (vision-for-perception), while the dorsal stream is involved in providing visual guidance for motor action (vision-for-action).

Within the auditory domain, an analogous anatomical and functional dual stream organization has been postulated, with a division identified between anteroventral and posterodorsal auditory networks. The primate auditory cortex has been divided into core, belt, and parabelt regions, and while regions within the core has been found to preferentially project to belt regions immediately adjacent to them, the dual streams proper have been argued to originate in rostral (ventral stream) and caudal (dorsal stream) belt areas (Rauschecker, 1998a, Rauschecker, 1998b, Rauschecker and Tian, 2000, Tian et al., 2001). The anteroventral stream begins in rostral lateral and medial belt areas, and projects along a pathway involving the rostral parabelt, superior temporal sulcus, and anterior temporal lobe, terminating in multiple frontal lobe regions including the frontal pole (area 10), the rostral principal sulcus (area 46), and ventral prefrontal areas (12 and 45; Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999, Tian et al., 2001). In contrast, the posterodorsal stream begins in caudal lateral and medial belt areas, and projects along a pathway involving regions including the caudal parabelt, posterior temporal regions (including Tpt and TPO), and parietal regions (including areas VIP, AIP, and 7a), terminating in frontal areas including the caudal principal sulcus (area 46), the frontal eye fields (area 8a; Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999, Smiley et al., 2007, Tian et al., 2001). Thus, while the visual and auditory dual streams follow different anatomical pathways, there are a number of regions common to both networks, such as parietal region VIP, which an important region along the dorsal processing stream across both modalities. In addition, it is interesting to note that both the visual and auditory dorsal and ventral streams terminate within the same regions of the prefrontal cortex (i.e., ventrolateral and dorsolateral prefrontal cortices, respectively; Baizer et al., 1991, Rauschecker and Tian, 2000, Romanski, Tian, et al., 1999). In relation to proposed function, akin to vision, a similar ‘what/where’ division has also been proposed for the auditory networks, with the anteroventral stream associated with processing spectrotemporal sound properties involved in auditory pattern/object recognition, and the posterodorsal stream associated with processing the spatial properties of sound and sound source localization (Kusmierek et al., 2012, Rauschecker, 1998a, Rauschecker, 1998b, Rauschecker and Tian, 2000). However, the functioning of the dorsal auditory network has recently been postulated to extend beyond that of spatial processing, and a role in sensorimotor integration has been proposed (Kusmierek et al., 2012, Rauschecker, 2011). Associated with this functional capacity, the dorsal network has been suggested to function as an auditory-motor interface, playing an important role in the mapping of sounds onto motor representations, a function important in speech acquisition and production (Rauschecker, 2011, Scott and Wise, 2004).

As the above would suggest, in addition to the sensory domains of vision and audition, the complex higher-order cognitive processes associated with language have also been postulated to involve a division of labor between dorsal and ventral processing networks, and a number of dual stream language models have recently been proposed (DeWitt and Rauschecker, 2012, Gow, 2012, Gow et al., 2009; Hickok, 2009; Hickok and Poeppel, 2004, Hickok and Poeppel, 2007, Rauschecker and Scott, 2009, Scott and Wise, 2004, Ueno et al., 2011). These models have largely been derived from the models of dual stream auditory processing, most notably that proposed by Rauschecker and Tian (2000), and as such, follow a similar anatomical and functional processing dichotomy. For example, within the dual stream model of Hickok and Poeppel, 2004, Hickok and Poeppel, 2007, the language network is comprised of a ventral stream involving posterior temporal regions (including the middle and inferior temporal gyri), the anterior temporal lobe, and the inferior frontal gyrus, while the dorsal stream which also originates in posterior temporal areas (notably the Sylvian parietal temporal region, Spt), travels via inferior parietal regions (particularly the supramarginal gyrus), to frontal regions including the premotor cortex and inferior frontal gyrus. Within such a model, the ventral stream is involved in sound-to-meaning mappings associated with semantic access and language comprehension, while the dorsal stream is involved in sound-to-motor mappings associated with language production and articulation. In support of this functional dissociation, a recent meta-analysis of spoken word recognition implicated the ventral network in the processing of spectrotemporal features, pattern recognition and the invariant representation of phonetic forms, while the dorsal network was associated with sensorimotor integration and the mapping of speech sounds onto the motor processes required to produce them (DeWitt & Rauschecker, 2012).

However, the conceptualization of the dorsal and ventral language networks, their composite brain regions, and their specific functional roles have not been as easy to characterize as with the sensory domains. Language is a highly complex skill involving a multitude of cognitive processes and cortical regions, and while substantial advances in our understanding have been made, the anatomical and functional organization of language in the brain still remains somewhat of an enigma. Models of language functioning have traditionally focused on identifying linguistic components and processes, independent of neuroanatomy (e.g., Dell et al., 1997, Levelt et al., 1999, Schwartz et al., 2006), with current neurolinguistic models attempting to map these previously defined linguistic concepts onto cortical networks (e.g., Hickok and Poeppel, 2004, Hickok and Poeppel, 2007, Indefrey, 2011, Indefrey and Levelt, 2004). Yet, an important question remains regarding exactly how these linguistic concepts map onto the computational functions carried out by the brain, and a number of researchers have begun to conceptualize the processing of the dual streams in relation to more general functional mechanisms rather than specific linguistic elements. For example, Weiller and colleagues (2011) postulated that the linguistic functions carried out by the dorsal network were a consequence of its more general computational capacity in the analysis of segment sequencing in time and/or space, and the fast on-line integration of external inputs with internal models (predictors) and emulators (controllers). This enabled the dorsal stream to not only map speech sounds onto motor representations for articulation, but also to maintain auditory inputs within working memory, and integrate and maintain the spectral and temporal evolution of (speech) sounds over time. Indeed, a commonality in core dorsal and ventral computational processing is evident across the three domains of vision, audition, and language. For example, the functions ascribed to the ventral stream (that is, stimulus recognition and sound-to-meaning mapping) all rely heavily on time-invariant representations and processing, while those ascribed to the dorsal stream (spatial, sensorimotor/visuomotor processing, and sound-to-motor mapping) are associated with fast-updating, time-variant processing (DeWitt and Rauschecker, 2012, Goodale, 2011, Milner and Goodale, 2008). This leads to an important point in relation to the nature of the dual ‘pathways’ and dual ‘streams’. A ‘pathway’ refers to a distinct anatomical route between brain regions within a given network, while a processing ‘stream’ refers to a computational network underlying a specific function. As such, while the dual pathways across the three domains are composed of distinct anatomical architectures, the underlying organizational principles and computational processes carried out by the dual streams may be the same, reflecting the fundamental functional architecture of the brain. However, exactly how, when, and to what extent these common computational processes are utilized will differ considerably depending on the task being executed.

As proposed by the dual stream model, there is strong evidence that the brain mechanisms associated with processing spatial (or motor/action) and non-spatial information are associated with distinct and dissociable brain networks in vision, audition, and language. Indeed, there is strong evidence to indicate that this dorsal–ventral functional and anatomical segregation continues throughout the brain, from posterior sensory regions all the way through to the prefrontal cortex (Arnott et al., 2004, Cohen et al., 2009, Goldman-Rakic, 1988, O’Reilly, 2010, Petrides and Pandya, 1999, Petrides and Pandya, 2001, Petrides and Pandya, 2006, Petrides and Pandya, 2007, Romanski, Bates, et al., 1999, Romanski, Tian, et al., 1999, Wilson et al., 1993, Yeterian et al., 2012). However, while the separation and relative functional specialization of the dual pathways has gained increasing acceptance, their strict independence has been challenged on both theoretical and empirical grounds.

Cognition is a dynamic process, and a flexible interactive system is required to coordinate and modulate activity across cortical networks to enable the adaptation of processing to meet variable task demands. Researchers have argued that the clear division of the dorsal and ventral processing streams is artificial, resulting from experimental situations which do not reflect processing within the natural environment (Weiller et al., 2011). Indeed, although it may be possible to identify some basic cognitive tasks involving primarily spatial/action processing or semantic/identity processing, the successful execution of most visual, auditory, and linguistic behaviors require the complex collaboration and seamless integration of processing between the two systems. For example, while visuo-semantic processing along the ventral stream may allow us to identify an object as a hammer, and visuospatial/visuomotor processing along the dorsal stream may provide information regarding the location and orientation of the object in relation to ourselves, the integration of both types of information is required in order to pick up and grasp the hammer in the correct way for use. This was demonstrated directly in a series of elegant behavioral experiments conducted by Creem and Proffitt (2001), who examined the interaction between semantic and visuomotor information during action execution. In this study, participants were required to pick up familiar hand tools (e.g., toothbrush, hammer, fork), presented with their handles facing away from the observer, whilst performing a concurrent semantic or visuospatial task. When performing the visuospatial task (or no concurrent task), participants grasped the handles of the items in a way congruent with their correct use, even though this required an awkward positioning of the hand and wrist. In contrast, when a concurrent semantic task was performed, although still able to accurately grasp the tools, participants failed to do so in a manner appropriate for their use. Thus, although visuospatial/visuomotor processing along the dorsal stream was unimpaired and able to effect successful grasping, this information was unable to be integrated with semantic object knowledge from the ventral stream, presumably due to the increased processing load induced by the concurrent semantic task.

The study by Creem and Proffitt (2001) highlights two important points. Firstly, it indicates that processing along the dorsal stream is indeed able to proceed successfully without ventral stream input, further underlining the specialized and segregated nature of processing along the dual pathways. However, it also demonstrates that although the dorsal stream is able to process and execute relatively simple visuomotor processing independently, when more complex behavioral responses are required, the integration of semantic information from the ventral stream is essential. This is further highlighted within the domain of language, where it has been noted that semantic information from the ventral stream and phonological sequencing (and articulatory) information from the dorsal stream must interact closely to enable the successful production and comprehension of language (Rolheiser et al., 2011, Rosazza et al., 2009, Weiller et al., 2011). Neuropsychological studies of patients with semantic impairments have consistently demonstrated a deficit in the repetition of short word lists, especially for words which the patients no longer comprehend (Caza et al., 2002, Jefferies et al., 2006, Jefferies et al., 2008, Jefferies et al., 2005, Knott et al., 1997, Patterson et al., 1994). Interestingly, the errors produced by these patients are predominantly phonological misorderings, and are dominated by the migration of phonemes between words in the list (i.e., ‘Spoonerisms’: mint, rug  “rint, mug”; Patterson et al., 1994). It is important to note that these patients have a selective semantic impairment, and their performance is not associated with an underlying phonological deficit, as evidenced by preserved performance on phonological awareness and discrimination tasks (Jefferies et al., 2005). Indeed, a similar pattern of errors can be induced in normal participants when lexical-semantic constraints are reduced (Jefferies, Frankish, & Lambon Ralph, 2006). The striking relationship between semantic impairment and the breakdown of phonological short-term memory led Patterson et al. (1994) to postulate that semantic information from the ventral stream played a critical role in ‘binding’ a word’s phonological elements within the dorsal stream. According to this ‘semantic binding’ account, the co-activation of semantic and phonological information whenever a word is produced establishes a strong link between the word’s pronunciation and its meaning which may be used to help ‘glue’ the phonological elements of a word together. In patients with a semantic impairment, the lack of this semantic ‘glue’ makes the task of maintaining the correct ordering of all the phonological elements more difficult, resulting in a preponderance of phoneme migration errors.

The above studies provide compelling evidence that ventral stream information is required for the successful execution of higher-order behaviors processed predominantly within the dorsal stream, and complementary studies have shown that the reverse is also true. In a series of studies, Almeida and colleagues (Almeida et al., 2010, Almeida et al., 2008) examined the influence of information from the dorsal stream on object recognition processing within the ventral stream, utilizing a semantic priming paradigm in which participants were required to name or categorize pictures of manipulable (tools) or non-manipulable (animals) items. A continuous flash suppression technique was used to selectively mask the picture primes within the ventral stream, while leaving dorsal stream processing unaffected. Across studies, semantic facilitation was consistently observed for the naming and categorization of tools, but not animals. Thus, it appears that information extracted from the prime by the dorsal stream was able to be utilized by ventral stream processing to aid object identification, a finding also observed in other studies (Helbig et al., 2006, Mahon et al., 2007).

Studies of individuals with schizophrenia have also revealed evidence of subtle visual object recognition impairments in these patients, despite intact processing along the ventral stream (Doniger et al., 2002, Sehatpour et al., 2010). Doniger et al. (2002) found that patients with schizophrenia showed an impairment in the ability to recognize complete objects based on fragmentary information (perceptual closure). Electrophysiological recordings indicated a reduction in amplitude of the P1 component over dorsal stream areas, coupled with intact ventral N1 generation. These researchers concluded that ventral stream processing remained intact in schizophrenia, and the visual object recognition impairments observed were the consequence of a deficit to dorsal stream processes which resulted in a subsequent dysregulation of processing along the ventral stream. Nevertheless, it is important to note that, like the dorsal stream, ventral stream processing is able to proceed successfully without cross-stream input, and while the ventral stream is essential for successful object recognition, the dorsal stream is not. Higher-order tasks of increasing complexity may utilize additional information from the dorsal stream to aid object recognition, however, accurate identification via the ventral stream is still observed when dorsal stream inputs are impaired or absent. Indeed, within the study of Doniger et al. (2002), perceptual recognition impairments were only observed in the patients with schizophrenia when stimuli were degraded, and no deficits were identified when whole pictures were presented. This is echoed in the performance of individuals with optic ataxia, a deficit in visuomotor skills resulting from damage to dorsal stream structures, who demonstrate intact recognition processes despite their visuomotor impairments (Goodale, 2011, Goodale, Meenan, et al., 1994, Milner and Goodale, 2008).

The studies above are a small sample of the ever-growing body of literature which has provided strong evidence for the existence of, and indeed necessity for, the close collaboration and interaction of the dorsal and ventral processing streams (see also e.g., Bartolo et al., 2007, Koshino et al., 2005, Lewald et al., 2008, McIntosh and Lashley, 2008, Mondor et al., 1998). While the evidence in support of the existence of dissociable dorsal and ventral processing streams is overwhelming and almost universally accepted, these and other similar findings have led researchers to call for a rejection of the idea of a strict functional (and anatomical) independence between the dual pathways. Instead, it is postulated that although segregation and specialization does indeed exist, a dynamic interplay between the two networks is essential for the successful execution of complex behavior (Almeida et al., 2010, Callaway, 2005, Goodale, 2001, Goodale and Milner, 1992, Goodale and Westwood, 2004, Goodale et al., 2005, Weiller et al., 2011; see Schenk and McIntosh (2010) for a review of the issue within the visual domain).

However, while it is acknowledged that there must be some degree of interaction between the two processing streams, how they interact, both anatomically and functionally, is a question which remains little explored. Within most current models of dual stream processing, the issue of interaction is largely absent (e.g., Hickok & Poeppel, 2004). Indeed, even within models which have expressly stipulated the existence and importance of interaction between the two networks, potential mechanisms underlying this ‘cross-talk’ have not been postulated, and continue to be “one of the important issues that remain to be solved” (Goodale et al., 2005, p. 279).

The question then remains, if a degree of interaction between the dorsal and ventral processing streams is required for the successful execution of complex behavior, where, when, and how does this happen? While no definitive answers to this question are currently available, based on current models of dual stream processing and existing anatomical and empirical evidence, three potential models of interaction (not exhaustive, and not necessarily mutually exclusive) may be postulated: (1) computations along the two pathways proceed strictly independently and in parallel, reintegrating at some ‘terminal’ stage of processing within a shared target brain region (the ‘independent processing’ account); (2) processing along the separate pathways is modulated by the existence of feedback loops which transmit information from ‘downstream’ brain regions, including information processed along the complementary stream (the ‘feedback’ account); and (3) information is transferred between the two systems at multiple stages and locations along their processing pathways (the ‘continuous cross-talk’ account). Before examining the evidence as it relates to each of these specific accounts (‘how’), it is important to first explore the potential neuroanatomical basis for any interaction between the two networks (‘where’), as well as the time course of dorsal and ventral processing (‘when’). In order to explore these issues, the current review will evaluate evidence across the three domains of vision, audition, and language. As noted above, there are obvious differences between these domains in relation to the specific tasks carried out by the dorsal and ventral stream, and the brain regions involved. However, commonalities in underlying neuroanatomical/functional organization are likely to exist. This is not to infer that what is true in one domain will necessarily transfer to another, but rather the aim is to identify and evaluate potential mechanisms of interaction between the dorsal and ventral streams which may be relevant to one or all of the three domains explored.

Section snippets

Where could dorsal–ventral interaction take place?

As noted above, tract tracing and electrophysiological recording studies in non-human primates have provided evidence for a strong degree of neuroanatomical segregation between the dorsal and ventral processing streams, and the selectivity of neuronal populations within these pathways for spatial and non-spatial information, respectively (Baizer et al., 1991, Kaas and Lyon, 2007, Livingstone and Hubel, 1988, Rauschecker, 1998a, Rauschecker and Tian, 2000). However, the anatomical dissociation

When could dorsal–ventral interaction take place?

When attempting to elucidate the nature and extent of communication between the dorsal and ventral processing streams, information regarding the time course of processing along the two pathways may be particularly informative. Electrophysiological (EEG and MEG) studies with both humans and non-human primates have provided evidence regarding the time windows in which regions along the dorsal and ventral pathways are activated, and have revealed some interesting patterns of activation and

How could dorsal–ventral interaction take place?

The review of the neuroanatomical basis and time course of processing within the dorsal and ventral streams has provided some initial insights into the potential mechanisms underlying interaction between the dorsal and ventral streams. Numerous reciprocal connections between regions across the dual networks, and connections to shared target brain regions, provides a neuroanatomical substrate for interaction to occur. In addition, while the time course of activation along the dorsal and ventral

Conclusions

Neurophysiological, neuroimaging, and lesion studies in human and non-human primates have provided converging evidence for the importance of a dual stream architecture in the execution of a range of higher cognitive functions including vision, audition, and language. However, while much has been learned regarding the anatomical, temporal, and functional processes carried out by the two complementary computational networks, there are still many questions left unanswered. While a degree of

References (184)

  • E.M. Callaway

    Neural substrates within primary visual cortex for interactions between parallel visual pathways

    Progress in Brain Research

    (2005)
  • D.P. Carey et al.

    Visuomotor sensitivity for shape and orientation in a patient with visual form agnosia

    Neuropsychologia

    (1996)
  • S. Clarke et al.

    Auditory agnosia and auditory spatial deficits following left hemispheric lesions: Evidence for distinct processing pathways

    Neuropsychologia

    (2000)
  • L.L. Cloutman et al.

    The variation of function across the human insula mirrors its patterns of structural connectivity: Evidence from in vivo probabilistic tractography

    NeuroImage

    (2012)
  • N.U.F. Dosenbach et al.

    A dual-networks architecture of top-down control

    Trends in Cognitive Sciences

    (2008)
  • N.U.F. Dosenbach et al.

    A core system for the implementation of task sets

    Neuron

    (2006)
  • A. Ellison et al.

    Time course of the involvement of the ventral and dorsal visual processing streams in a visuospatial task

    Neuropsychologia

    (2007)
  • R. Farivar

    Dorsal–ventral integration in object recognition

    Brain Research Reviews

    (2009)
  • A.D. Friederici

    Pathways to language: Fiber tracts in the human brain

    Trends in Cognitive Sciences

    (2009)
  • M.A. Goodale

    Different spaces and different times for perception and action

    Progress in Brain Research

    (2001)
  • M.A. Goodale

    Transforming vision into action

    Vision Research

    (2011)
  • M.A. Goodale et al.

    Dual routes to action: Contributions of the dorsal and ventral streams to adaptive behavior

    Progress in Brain Research

    (2005)
  • M.A. Goodale et al.

    Separate neural pathways for the visual analysis of object shape in perception and prehension

    Current Biology

    (1994)
  • M.A. Goodale et al.

    Separate visual pathways for perception and action

    Trends in Neurosciences

    (1992)
  • M.A. Goodale et al.

    An evolving view of duplex vision: Separate but interacting cortical pathways for perception and action

    Current Opinion in Neurobiology

    (2004)
  • D.W. Gow

    The cortical organization of lexical knowledge: A dual lexicon model of spoken language processing

    Brain and Language

    (2012)
  • D.W. Gow et al.

    Parallel versus serial processing dependencies in the perisylvian speech network: A Granger analysis of intracranial EEG data

    Brain and Language

    (2009)
  • T.A. Hackett

    Information flow in the auditory cortical network

    Hearing Research

    (2011)
  • D.A. Hall

    Auditory pathways: Are ‘what’ and ‘where’ appropriate?

    Current Biology

    (2003)
  • O. Hauk et al.

    Somatotopic representation of action words in human motor and premotor cortex

    Neuron

    (2004)
  • C. Hesse et al.

    Visuomotor performance based on peripheral vision is impaired in the visual form agnosic patient DF

    Neuropsychologia

    (2012)
  • G. Hickok

    The functional neuroanatomy of language

    Physics of Life Reviews

    (2009)
  • G. Hickok et al.

    Dorsal and ventral streams: A framework for understanding aspects of the functional anatomy of language

    Cognition

    (2004)
  • P. Indefrey et al.

    The spatial and temporal signatures of word production components

    Cognition

    (2004)
  • E. Jefferies et al.

    Lexical and semantic binding in verbal short-term memory

    Journal of Memory and Language

    (2006)
  • E. Jefferies et al.

    The impact of semantic impairment on verbal short-term memory in stroke aphasia and semantic dementia: A comparative study

    Journal of Memory and Language

    (2008)
  • J.H. Kaas et al.

    Pulvinar contributions to the dorsal and ventral streams of visual processing in primates

    Brain Research Reviews

    (2007)
  • A. Klistorner et al.

    Separate magnocellular and parvocellular contributions from temporal analysis of the multifocal VEP

    Vision Research

    (1997)
  • V.A.F. Lamme et al.

    The distinct modes of vision offered by feedforward and recurrent processing

    Trends in Neurosciences

    (2000)
  • R. Laycock et al.

    A role for the ‘magnocellular advantage’ in visual impairments in neurodevelopmental and psychiatric disorders

    Neuroscience and Biobehavioral Reviews

    (2007)
  • J. Ahveninen et al.

    Task-modulated “what” and “where” pathways in human auditory cortex

    Proceedings of the National Academy of Sciences USA

    (2006)
  • C. Alain et al.

    “What” and “where” in the human auditory system

    Proceedings of the National Academy of Sciences USA

    (2001)
  • C. Alain et al.

    Spatiotemporal analysis of auditory “what” and “where” working memory

    Cerebral Cortex

    (2009)
  • J. Almeida et al.

    The role of the dorsal visual processing stream in tool identification

    Psychological Science

    (2010)
  • J. Almeida et al.

    Unconscious processing dissociates along categorical lines

    Proceedings of the National Academy of Sciences USA

    (2008)
  • J.S. Baizer et al.

    Organization of visual inputs to the inferior temporal and posterior parietal cortex in macaques

    Journal of Neuroscience

    (1991)
  • D. Bendor et al.

    Neural response properties of primary, rostral, and rostrotemporal core fields in the auditory cortex of marmoset monkeys

    Journal of Neurophysiology

    (2008)
  • R. Borowsky et al.

    FMRI of ventral and dorsal processing streams in basic reading processes: Insular sensitivity to phonology

    Brain Topography

    (2006)
  • R. Borowsky et al.

    Modularity and intersection of “what”, “where” and ‘how’ processing of visual stimuli: A new method of fMRI localization

    Brain Topography

    (2005)
  • V. Boulenger et al.

    Cross-talk between language processes and overt motor behavior in the first 200 msec of processing

    Journal of Cognitive Neuroscience

    (2006)
  • Cited by (0)

    View full text