A novel perspective for the necessary and sufficient conditions for primary visual perception is proposed based upon relating recent discoveries from the visual sciences to relevant foundational neurological observations. This analysis suggests that a fundamental requirement for the emergence of normal primary visual perception is the coupling between the early visual cortices in the occipital lobe subserving image content with specific areas in the parietal lobe subserving selective attention, representations of extrapersonal space, the body schema, and the initiation of perceptual ownership. Fully intact primary visual perception, which includes the normal placement of image content within extrapersonal space, seems to require at each instant a mutually consistent completeness and corresponding removal of ambiguities in each of the linked neural representations subserving image, space and self. Experimental approaches that could either invalidate or strengthen the proposed framework are suggested as well as opportunities to differentiate some aspects of normal primary visual perception from the severely compromised visual experience that survives bilateral parieto-occipital lesions (Balint's syndrome) when visual experience may persist for single but unlocalizable objects.
I apply the term “primary visual perception” (PVP) to refer to our most basic subjective experiences of brightness and color that are sometimes referred to as “qualia.” Such usage is essentially identical to “phenomenal consciousness” (Block 1995) and “core consciousness” (Damasio 1999). Humphrey (1992) has made a persuasive case that an understanding of primary perceptual experience is at the heart of the consciousness problem if we are ever to understand the basis for any form of higher-order subjective experience. In fully intact PVP, extrapersonal space is the “medium in which focused attention selects which features belong together and conjoins them to represent a single object” (Robertson et al. 1997).
PVP differs from “primary consciousness” (Edelman 2003), which includes the hippocampal formations. However, because their structural integrity is not required for primary visual experience (Damasio 1999), I exclude the hippocampus and its projections as necessary requirements for PVP. However, enabling functions such as wakefulness and thalamocortical integration must remain normally operant (Llinas et al. 1998; Laureys et al. 2004).
PVP also includes the perceptual experience of objects even when their identification has been eliminated as a consequence of an associative agnosia. Such agnosias are generally attributable to lesions in the ventral loop beyond V1/V2 and especially within inferotemporal cortex (IT). Moreover, patients with bilateral parieto-occipital lesions (Balint's syndrome) may experience primary perception, at times of only a single object of finite spatial extent that is unlocalizable in extrapersonal space (Robertson et al. 1997).
Results from clinical studies of parietal neglect syndromes (DeRenzi 1982; Robertson et al. 1997; Vuilleumier et al. 2001; Rizzolatti and Matelli 2003), from masking and priming results (Dehaene et al. 2001) and from disruption of parietal lobe function by transcranial magnetic stimulation in studies of change blindness (Beck et al. 2006) all suggest that selective activations of both ventral and dorsal loops must occur concurrently to engender normal primary visual experience. Moreover, based upon findings that patients with parietal neglect may implicitly process stimuli up to the categorical level of representation but without any accompanying conscious experience, Rizzolatti and Matelli (2003) conclude that “ventral loop processing is not sufficient to obtain perception without parietal spatial processing.” The focus of the present work is to propose a reconceptualization of how different brain areas interact to produce PVP, especially about how spatial maps in different areas must be bound in order to support normal PVP as well as to account for the more limited aspects of visual experience that survive Balint's syndrome.
Cortical Representations of Image Content and Object Recognition
The integrity of the striate cortex (V1) is essential for normal human vision. This cortex, and extrastriate cortical areas along the ventral stream are retinotopically organized (Felleman and Van Essen 1991) and, hence, do not convey the “where” information (Ungerleider and Mishkin 1982) of the location of image content in extrapersonal space with respect to eye, head, or trunk position.
Extrastriate areas subserve modular functions such as the selectivity for motion in V5/middle temporal area (MT) and color constancy in V4 (Zeki 1997, 2001). No lesions within extrastriate cortical areas beyond V1 eliminate static achromatic brightness discriminations for isolated elemental stimuli, that is, a small grating patch of a single preferred orientation and spatial frequency. However, visual deficits for more complex stimuli, especially with respect to texture discriminations, occur according to the extrastriate area destroyed (Pollen 1999).
Bilateral lesions of inferotemporal cortex may impair object recognition and produce associative agnosias. However, in such cases visual fields, visual acuity, and elemental visual discriminations remain intact in macaques (Gross 1992) and brain-damaged human subjects (Damasio 1999). Moreover, when IT is bilaterally ablated, all efferent pathways from this cortex to the prefrontal areas are functionally interrupted as well. Therefore, some aspects of PVP—in contradistinction to higher-order consciousness—are not dependent upon either the functional integrity of IT or the information that IT normally provides to prefrontal cortical areas.
Are Prefrontal Cortices Essential for PVP?
Attended visual stimuli increase functional magnetic resonance imaging (fMRI) recorded activity in the parietal lobe and secondarily in the right dorsolateral prefrontal cortex (Dehaene and Naccache 2001; Rees et al. 2002; Dehaene and Changeux 2004; Lau and Passingham 2006). Such activation is frequently followed by top-down amplification of activity in posterior stimulus-specific areas (Dehaene et al. 2006). However, these results do not inform as to whether such prefrontal activations are correlates of higher-order consciousness, PVP, or both. Although prefrontal activity can secondarily modify activity in posterior brain regions, longstanding neurological evidence suggests that such prefrontal activity is not essential for PVP.
The lack of essentiality for any unilateral area within prefrontal and frontal lobes for PVP has been known since Penfield (Penfield and Evans 1935) surgically “amputated” the entire right frontal lobe (including prefrontal cortex) from the brain of his sister to remove a malignant tumor. Her basic visual capacities remained unimpaired. Left frontal lobe ablations also spare basic visual capacities (Penfield and Evans 1935).
Such capacities in humans are also preserved after extensive bilateral lesions of the prefrontal cortex (Brickner 1936; Damasio 1999). It might be argued that such lesions were not complete enough to exclude a role of these cortices, particularly the dorsolateral prefrontal cortex (Brodmann area 46), in PVP. However, that possibility appears remote given the sparing of the basic visual capacities of psychiatric patients following selective bilateral lesions of prefrontal cortex made with the intent to alleviate severe psychiatric symptoms without impairing the special senses (Heath et al. 1949). Bilateral prefrontal ablations were made in some subjects in Brodmann areas 45 and surrounding cortices, in others in areas 46 and surrounding cortices, and in still other subjects both areas 45 and 46 were bilaterally ablated. No bilateral ablation of prefrontal cortices disrupted basic visual experience for luminance or color and visual fields remained intact.
The above results supported the longstanding neurological views that prefrontal cortical areas are concerned with the abstract and symbolic significance of a stimulus not with its basic sensory qualities (Goldstein 1936). Also lost after frontal lesions is the organized exploratory visual scanning activity with respect to a complex object rather than any disturbance of basic visual perception (Luria et al. 1966). The ability to manipulate the meaning of sensory data, to organize it appropriately with respect to the past history of the individual and his/her future needs is also lost after prefrontal lobotomy (Milner 1982). These functions of prefrontal cortex are attributes of higher-order consciousness related to symbolic content not of PVP.
Studies on “frontal neglect” are also consistent with the above interpretations. Frontal lesions in humans, particularly those in the dorsolateral prefrontal cortex and frontal eye-fields, may sometimes produce a predominantly “intentional neglect” (Heilman and Valenstein 1972a; Watson et al. 1978; Damasio et al. 1980; Bisiach et al. 1990; Heilman 2004). Such neglect is generally transitory and characterized by a decreased ability to direct motor attention toward the visual representations in the contralateral visual field.
Frontal neglect may also occur after lesions in the frontal eye-fields in macaque monkeys, especially for brief low contrast stimuli in the peripheral visual fields contralateral to the lesion. Even so, foveal and near-foveal flashes are responded to correctly (Latto and Cowey 1971). Transcranial magnetic stimulation of the frontal eye-fields in humans enhances perceived contrast for peripheral relative to central visual stimuli with corresponding increase in fMRI activity in the peripheral contralateral representations of retinotopic cortical areas V1–V4 (Ruff et al. 2006). These results argue against a primary sensory deficit as a major contributant to frontal neglect and support explanations for the latter based upon intentional deficits and/or top-down attentional modulations of activity especially for peripheral representations in early visual cortical areas (Heilman 2004).
Moreover, prefrontal cortical areas are not required for all types of reportability. Conscious reportability of visual experience in the form of saccadic eye movements, reaching toward and/or grasping attended visual targets can be initiated by networks within the posterior parietal lobes (Grüsser and Landis 1991; Goodale and Milner 1992; Andersen et al. 1997).
V1 and PVP
Crick and Koch (1995) hypothesized that neurons in V1 do not directly participate in conscious experience. They based their conclusion primarily upon 2 arguments; first, that V1 did not explicitly represent any visual stimulus dimension and second, that V1 does not project directly to anterior motor or decision-making areas. I proposed an alternative hypothesis that because V1 and V2 are targets of back projections from several extrastriate and inferotemporal cortices as well as processors of afferent information that “dynamic iterative interactions between recognition spaces and perceptual spaces may well be a minimal requirement for visual awareness” (Pollen 1995). Several related issues are considered below.
Selective activation of neurons in V1 along its afferent or feedforward pathway is not in general sufficient for PVP. In “crowding” a test stimulus that is readily perceived when presented in isolation can be eliminated from conscious experience when that stimulus is flanked by the nearby placement of selective stimuli. The nonperceived “crowded” stimulus nevertheless continues to activate neurons in V1 (He et al. 1996). Feedforward activity in V1 and the ventral loop may also be evoked by an unseen second stimulus in the attentional blink paradigm (Marois et al. 2000).
The above results may be interpreted consistent with neurological findings, physiological results, and matched psychophysical observations that all suggest that the finest grained neural representation for both the spatial detail and retinotopic localization that underlies local static, achromatic brightness discriminations is expressed by the activity of certain neurons in V1 (Pollen 1999). Psychophysical results assessing the properties of the “channels” selective to orientation and spatial frequency match well the properties of V1 neurons. These and other findings support the hypothesis that certain neurons in V1 do play a direct role in at least some aspects of PVP but only as components of recursive neuronal networks (Pollen 1995, 1999; Supèr et al. 2001).
Several lines of evidence further support recursive hypotheses. Primary perception of visual motion is abolished when V1 is inactivated by transcranial magnetic stimulation following activation of MT/V5 within a time window consistent with the time required for recursive activity from MT/V5 to project back to V1 (Pascuale-Leone and Walsh 2001; Silvanto et al. 2005). The perception of isoluminant colored stimuli may also require completion of a feedback loop to V1 and perhaps to V2 and V3 as well (Ro et al. 2003). Moreover, there is a secondary activation of neurons in V1 following the initial afferent response by some 100 ms only when macaques report that a salient texture is evident (Supèr et al. 2001).
These results are also consistent with the “reverse hierarchy theory” (Ahissar and Hochstein 2004) of perceptual learning which is largely guided by a gradual increase of first high- then lower-level tasks with relevant information reaching down as far as V1 when precise serial conjunctions and/or improved signal to noise ratios are required. Moreover, perception of the “gist” of a scene may precede that of its finest details because the latter may require more scrutiny and processing time (Ahissar and Hochstein 2004).
For some classes of stimuli, visibility requires a dynamically associated coupling of activity between primary visual cortex and the fusiform cortex (Haynes et al. 2005). In other experimental paradigms, fMRI responses in early visual cortices, particularly in V1, may be predictive of whether or not a human subject perceives a stimulus (Ress and Heeger 2003).
There are other examples of perceptual transitions correlating with changes in fMRI activity in early visual cortical areas. Such activity in V1/V2 can predict which of 2 overlapping orientation-selective patterns an observer is attending (Kamitawi and Tong 2005). Furthermore, changes in fMRI activity in V1 during binocular rivalry correlate with the perceptual transition from one rivalrous stimulus to the other (Lee et al. 2005).
Thus, recursive activity back to V1 within the ventral loop and between V1 and MT/V5 appears likely to comprise a necessary condition for some types of perceptual experience. However, such recursive activity within the ventral loop alone without concurrent recursive links to the parietal lobe is not likely to serve as a sufficient condition for PVP.
For example, 1 function of the parietal lobe is to provide the neural basis for the localization of objects within extrapersonal space which appears as an essential requirement for normal perceptual experience. These results derive from 1) neurological studies of humans with spatial hemineglect syndromes following unilateral parietal lobe injuries, 2) from attempts to reduce hemineglect by induced changes in egocentric reference frames, and 3) from electrophysiologic studies of single neurons in the parietal lobe of awake macaque monkey trained to perform cognitive tasks. I shall briefly summarize pertinent findings.
Spatial Hemineglect Syndromes
Following the end of World War I, European neurologists reported upon the dramatic reductions of spatial awareness to visual stimuli in the left hemispatial field in wounded veterans with traumatic right parietal lobe injuries. A hemispatial field refers to space on 1 side of the midline of the body and is not identical to a visual hemifield (Heilman and Valenstein 1978). Right-sided hemineglect, although it is less prevalent than left-sided neglect, may occur after injury to the left parietal lobe. These deficits were recognized as impairments in the brain's construction of extrapersonal space as referenced to personal space and the body schema (Brain 1941). The latter term was taken as the proprioceptive map of a subject's body whether at rest or in motion (Head and Holmes 1911–1912).
Hemineglect syndromes have often been considered hemispatial visual inattentional disorders (Brain 1941; DeRenzi 1982; Kooistra and Heilman 1989). The neuronal sources of such impaired attention to image representations may themselves be mapped within specific spatial reference frames.
In milder cases of neglect, the presence and severity of inattentional deficits may vary from moment to moment depending upon the subject's attention and the severity and location of the lesion. However, it is my objective here to emphasize, when appropriate, the effects of the most devastating parietal lesions as those most revealing of severe dorsal loop dysfunction and conversely the extent of normal function in its absence.
In many cases of parietal neglect, subjects may detect single isolated stimuli on the left side of the center of gaze but have deficient or absent awareness of the left side of attended objects regardless of object location (Grüsser and Landis 1991; Driver and Mattingly 1998). A subject may remain totally unaware of food placed on the left side of a plate until the plate is rotated 180°. In other cases, hemispatial neglect may “masquerade” as a hemianopsia when the subject's eyes are directed straight ahead (Kooistra and Heilman 1989). However, such left-sided deficits may abate when the subject's eyes are directed to the normal right hemispace.
In the most severe cases of parietal neglect, patients behave as if that half of the universe contralateral to the side of parietal lobe injury has ceased to exist (Mesulam 1981). Similarly, Bisiach and Vallar (1988) regard total neglect as an “absolute representational vacuum”. Presumably, the networks damaged in such severe cases normally subserve the establishment of reference frames for the localization of and attention to objects in extrapersonal space, representations of the contralateral body schema and the latter's relationship to the neural correlates for the sense of self of the contralateral side.
Other studies of unilateral neglect suggest that visual objects may not be subjectively experienced unless an attended object's location in extrapersonal space is referable to at least 1 intact egocentric reference frame (Karnath et al. 1991, 1993). Such coordinate frames specify the position of the eyes in the orbits with respect to the head and the position of the head with respect to the trunk (Karnath et al. 1991). Orientation of the head in space is specified by the vestibular system, whereas the position of the head with respect to the trunk is specified by neck proprioceptors.
Spatial shifts in the localization of subjective experiences of after-images in the direction of saccadic eye movements have been documented since Aristotle. Our subjective interpretation of “where” such after-images are perceived also shifts compensatorily if we turn our head or rotate our head and body together. These results further support the link between egocentric reference frames and the localization of objects in extrapersonal space.
The localization of the principal lesion responsible for neglect has been a matter of current controversy. Karnath et al. (2001) proposed that lesions in the mid-portion of the superior temporal gyrus play the principal role. Their conclusion disagreed with the longstanding view that lesions in the inferior parietal lobe are especially critical for neglect (Heilman and Valenstein 1972b; Vallar and Perani 1986).
However, in a more recent fMRI study utilizing high spatial resolution, Mort et al. (2003) confirmed that the angular gyrus (Brodmann area 39) within the inferior parietal lobe was the critical area for neglect in all patients experiencing this symptom as a consequence of a stroke in the territory of the middle cerebral artery. Additionally, the most severe cases of visuospatial neglect are also associated with lesions of the supramarginal gyrus (Brodmann area 40) also within the inferior parietal lobe (Vallar 2001). Subcortical and limbic areas to which these inferior parietal areas project may also contribute to neglect syndromes (Marshall et al. 2002).
Parietal hemineglect may apply to visual imagery as well as to PVP (Bisiach and Luzzatti 1978). These findings are germane to both a common locus within the parietal lobe for ownership of both visual imagery and PVP and to the related relevance of the position of the eyes, head, and trunk whether a subject engages in either PVP or visual imagery.
The conclusion that PVP requires conjoined processing of early ventral and posterior parietal cortical areas is further supported by parietal lobe extinction studies (Vuilleumier et al. 2001). For example, in extinction a stimulus is visible when tested in isolation in either visual field. However, when stimuli are tested simultaneously in both fields, the percept in the field contralateral to the injured parietal lobe may sometimes be extinguished. When bilateral trials failed to extinguish the left-sided stimulus, there was greater fMRI activity in both the normal right primary visual cortex and intact left parietal cortex than in trials when the percept was extinguished. Recent studies of the right parietal lobe utilizing transcranial magnetic stimulation suggest that transient disruption of the right supramarginal and superior parietal lobule support a role for interhemispheric attentional competition as a basis for extinction (Chambers et al. 2006).
Parietal activation also occurs in the attentional blink paradigm. When the second stimulus becomes visible it may also secondarily activate dorsolateral prefrontal and anterior cingulate cortical areas (Marois et al. 2000; Sergent et al. 2005). However, such secondary activations do not establish whether such activations are selective for working memory, decision making, a task-specific report, or PVP. Nor do such anterior activations distinguish whether the limited access stage primarily contributing to the lack of visibility of the unseen second stimulus is attributable to intrinsic functions of the parietal lobe as a “gatekeeper” or to its reciprocal connections with prefrontal areas.
Compensations for Hemineglect by Induced Changes in Egocentric Reference Frames
Results that partial recovery from neglect syndromes can occur by an induced change in a spatial coordinate system provide some of the most compelling evidence that the visibility and localization of a given visual stimulus is dependent both upon an image representation specified in the ventral loop and an egocentric reference frame in the dorsal loop for the spatial location of percepts in extrapersonal space. These results attain even for shifts in spatial coordinates induced by nonvisual stimuli as discussed below.
Vibration of posterior cervical muscles activates muscle afferents that provide proprioceptive signals (Goodwin et al. 1972) that, in turn, can modify the central representation of the instantaneous direction of gaze for stationary visual targets (Lackner and Levine 1979; Biguer et al. 1988). Left-sided hemineglect may be reduced by rotating a patient's trunk to the left such that stimuli to the left hemifield now become projected to the right side of the midline of “trunk space” (Karnath et al. 1991). Thus, trunk orientation is the “physical anchor” of the internal representation of body orientation in space. The localization of this representation is independent of any concurrent attentional shifts. Vibration of posterior left neck muscles also modifies head-on-trunk proprioceptive signals and may reduce left hemineglect by shifting the subjective localization of the midsagittal plane to the left side (Karnath et al. 1993).
Balint's syndrome (DeRenzi 1982; Robertson et al. 1997) refers to the deficits found in patients who have sustained bilateral parieto-occipital infarcts generally involving the angular gyri and generally sparing the supramarginal gyri. Patients can frequently perceive a single object of any size but cannot discriminate any individual feature within the object or scene without losing perception of the rest of the stimulus, a condition called “simultanagnosia” (Luria 1959).
Moreover, Robertson et al. (1997) confirm early reports on Balint's that subjects sustain a severe loss of the ability to localize objects in space, either verbally or by pointing or looking in the object's direction. This impairment of perceptual space also precludes the subject from individuating letters within words and correctly binding colors to shapes. Consequently, subjects with Balint's may frequently report illusory conjunctions of shape and color (Robertson et al. 1997). In normal subjects, individual features of objects are encoded initially as independent entities and bound together by serial direction of attention (Treisman and Schmidt 1982). However, even in such normal subjects, illusory conjunctions may occur when attention is diverted or overloaded. Thus, these conjunctional deficits found in Balint's may occur in normal subjects when presumably the same type of spatial attentive mechanisms are strained by the brevity of stimulus exposures.
The restriction of attention to a single perceived object (DeRenzi 1982) is less well understood. Alternative hypotheses, including the inability of an attentional source to disengage attention from the targeted object, are considered in detail within Robertson et al. (1997). Nevertheless, there is general agreement that the experienced target has been subject to some attentional resource.
Furthermore, patients with Balint's syndrome have both intact egocentric reference frames and a normal sense of perceptual ownership (DeRenzi 1982; Robertson et al. 1997). Thus, I assume that the neural source of residual attention and the feedforward return of attended sense data from image representations continue to activate neural representations of perceptual ownership and sense of self.
Thus, although surviving but compromised perceptual experience in Balint's syndrome seems to have a plausible explanation, it seems more difficult to account for how an object can be perceived in the absence of explicit knowledge of its location within extrapersonal space. One possibility may be that intact object representations may be the attentional source directed upon image representations even when such sources are disconnected from spatial attentional mechanisms or when the latter are impaired or overwhelmed (Dehaene and Cohen 1994; Robertson et al. 1997). The possibility that intact object-based attentional mechanisms may link image representations to those for image ownership even in the absence of spatial localization has enhanced relevance given the discovery that representations of single objects may be encoded within the parietal lobe (Sereno and Maunsell 1998). Indeed, even normal subjects may sometimes be able to identify an object but not localize it when spatial attentional resources have been severely overloaded (Treisman and Schmidt 1982). However, such compromised unlocalizable residual visual experience can scarcely be considered normal PVP nor an example of the organization of a normally functioning visual system.
Electrophysiological Studies of Normal Parietal Lobe Function
Neurons in the inferior and posterior parietal lobe of the macaque encode the spatial coordinates of reference frames for common locations of visual, auditory and somatosensory inputs with respect to eye-, head-, body-, and world-centered coordinates frames (Andersen et al. 1997). It is presumably analogous groups of neurons in the human inferior parietal lobe that are rendered inoperative in neglect syndromes.
The contribution of these neurons to visual awareness under normal conditions probably includes both the expression of spatial attention upon image representations within the ventral loop that are bound to their spatial localizations in extrapersonal space encoded with the dorsal loop (Bisley and Goldberg 2003). Goldberg et al. (2006) propose that lateral intraparietal (LIP) neurons provide a map of the salience or attention-worthiness of a stimulus without specifying how the map will be used. The information can be interpreted by the oculomotor system as a saccade goal when a saccade is appropriate and/or used by the visual system to determine the locus of attention. Thus, the various functions of attention, intention (Snyder et al. 1997; Andersen and Buneo 2002), and decision making (Platt and Glimcher 1999; Roitman and Shadlen 2002) may build upon the primacy of the salience map (Bisley and Goldberg 2003).
The full expressions of attention, intention, and a decision to initiate voluntary movement may contribute to the long-recognized ever-constant sense of ownership that accompanies perceptual experience (Gallagher 2000). Moreover, the basis for any meaningful model of intentionality requires its linkage to a preexistent internal model of the external world (Denton 2006).
Space, Self, and Perceptual Ownership
Kant (1781) considered it self evident that the act of perception requires a perceiver. William James (1890) concurred and further emphasized the necessity for the perceiver to distinguish “self” from “nonself.” Humphrey (1992) assumed an evolutionarily early origin for the organism's ability to make such distinctions. Both superior parietal areas (Shimada et al. 2005) and medial parietal areas (Vogt and Laureys 2005) together with posterior and inferior parietal cortex may participate in the distinction between self and nonself. However, it is not yet known whether further referential relationships between the posterior parietal lobe and these other 2 parietal lobe regions are required for PVP. Nor is it known whether ownership of visual experiences involves the right posterior insula and right frontal operculum as recently reported for body ownership (Tsakiris et al. 2007).
A sense of self has long been recognized to include both a sense of ownership of a percept and a sense of agency to act based upon some attribute of a percept or its memory. The findings that the activity of certain neurons in the posterior parietal lobe may express the neural correlates of spatial attention (Goldberg et al. 2006), motor intention (Snyder et al. 1997), and decision making (Platt and Glimcher 1999) encourage efforts to develop further a neural framework that binds together neural representations of images, their locations in extrapersonal space, and their sense of ownership. However, these parietal correlates of agency and ownership, may yet remain early stages that may require further processing, perhaps as far forward as the anterior cingulate cortices as suggested by Damasio (1999) based upon his studies of akinetic mutism.
Approaches to link a sense of self to ownership of perceptual experience have been widely appreciated within the philosophical (Gallagher 1995), psychological (Gibson 1979), and neurological (Damasio 1999) literatures. Gibson (1979) hypothesized that perceiving involves the coperception of self and environment, and Neisser (1988) proposed the “ecological self” as the most primary sense of self that is “directly perceived with respect to the immediate physical environment.” Damasio (1999) considered “core consciousness” to depend upon “sense of self in the act of knowing.”
A Functional Model for PVP
The most parsimonious interpretation of the evidence considered herein suggests that normal PVP (percept, its spatial localization, and associated ecological self) requires, at a minimum, recursive neuronal interactions within and between the early visual cortices in the ventral (occipito-temporal) and dorsal (occipito-parietal) loops. This model likely requires explicit neural representations for stimulus-dependent image content in the early visual pathways including at the very least (V1, V2, V3 complex, V4, V5/MT). Representations for extrapersonal space with respect to personal space are likely implemented within a number of parietal cortical areas that are also task dependent and that include at the very least LIP, anterior intraparietal, the parietal reach areas, and the angular and supramarginal gyri. Whether representations of perceptual ownership and a sense of self involve these same parietal areas or still unspecifiable projections to higher cortical areas remains unknown.
Zeki (2001) challenges us to consider the proposition that processing nodes, that is, cortical areas that compute explicit representations such as MT/V5 for motion and V4 for color, or serial combinations of such nodes may also, in themselves, be sufficient to serve as perceptual nodes to express a particular “microconsciousness.” This hypothesis appears quite plausible when such nodal representations are linked to representations for selective attention, extrapersonal space, and perceptual ownership as expressed by neurons within the parietal lobe. However, this model, if taken to the most reductionistic level when such nodal representations would be isolated from the rest of the brain, appears not presently testable because such complete isolation would preclude reportability.
Damasio (1999) has emphasized the need for neuronal representations of image content to be accessed by representations for sense of self. The evidence presented here supports an extension of Damasio's model in normal PVP to include the linkage of image representations to representations of self and ownership though the intermediary of representations of extrapersonal space.
I do not suggest that the briefest primary perceptual experiences tied to Neisser's equally minimal expression of “ecological self” (Neisser 1988) can be any more than transitory phenomenological frames that persist only as long as the stimulus-driven underlying recursive networks remain active. Higher-order mechanisms are necessary to provide the continuity of changing visual and visuomotor experience and to compare past with present and to eventually achieve the later stages of full self-awareness largely dependent upon the frontal lobes (Stuss and Levine 2002). Even so, I suggest that we may have identified a set of posterior cortical areas and some integrating principles that set the bounds for the neural correlates of PVP in functional terms consistent with plausible anatomic bases.
One tentative framework for further analysis may be that even the most basic primary visual percepts require distributed information processing that links a series of cortical areas—and perhaps their thalamic dependencies as well—even beyond the confines of a single cortical lobe. A second plausible hypothesis may be that each set of recursive linkages must eliminate ambiguities before perceptual experience can emerge. For example, for virtually every instant of the active waking day, posterior cortical areas assess whether the visual stimulus or the viewer has moved relative to the visual world. Indeed, we seem to dynamically perceive each visual moment only when there is a mutually consistent completeness and corresponding removal of ambiguities in the linkages of neural representations subserving image, space, and self.
Future Directions and Tests of the Model
The present model would be invalidated if it could be shown that the surviving perceptual experiences in Balint's syndrome (see above) or in any other neglect syndrome can exist in the absence of linkage of their image representations to attentional sources whether space based or object-representation based that originate outside the early ventral loop. Second, the collapse of extrapersonal perceptual space in Balint's syndrome with the preservation of an unlocalizable single percept provides an opportunity to explore by fMRI the neural correlates of perceptual ownership of such spatially compromised perceptual experiences and to compare such results with those that attain in subjects with intact extrapersonal space. Furthermore, given the existence of bimodal visuotactile neurons in macaques (Graziano et al. 1997) and inferences for their existence in humans (Schendel and Robertson 2004), we should not exclude the possibility that the impaired localization of percepts in Balint's might be modifiable by activation of selective visuotactile representations.
Third, we do not yet know whether the most severe neglect syndromes differ from the milder forms by virtue of more spatial reference frames being involved in the former or whether the more severe syndromes require a specific sequence of linkages across a series of spatial coordinates systems. This remains a challenging problem because few neglect syndromes respect the anatomic cortical boundaries.
Finally, even as neuroscience reaches closer to defining the neural and functional correlates of PVP, we seem no nearer to formulating any conceptual approach that accounts for the existence of subjective experience. Moreover, it still remains unknown whether multistage recursive linkages express qualities of subjective experiences not necessarily emergent within individual cortical areas. Perhaps the framework proposed here will provide both some impetus and constraints for the requisite theoretical advances.
I am very grateful for the constructive comments of Drs Arash Afraz, Moshe Bar, Ned Block, David Drachman, Charles Jennings, Ken Nakayama, Deepak Pandya, David Paydarfar, Peter Grigg, and Andrzej Przybyszewski. I am also greatly indebted to 2 anonymous referees whose constructive comments have lead to a considerable expansion of the depth and breadth of the originally submitted manuscript. Conflict of Interest: None declared.