Critical Review of the Perception-Action Model

Do we have independent visual streams for perception and action? Running head: Critical review of the perception-action model 1|Page Abstract: The perception-action model proposes that vision-for-perception and vision-for-action are based on anatomically distinct and functionally independent streams within the visual cortex. This idea can account for diverse experimental findings, and has been hugely influential over the past two decades. The model itself comprises a set of core contrasts between the functional properties of the two visual streams. We critically review the evidence for these contrasts, arguing that each of them has either been refuted or found limited empirical support. We suggest that the perception-action model captures some broad patterns of functional localisation, but that the specialisations of the two streams are relative, not absolute. The ubiquity and extent of inter-stream interactions suggest that we should reject the idea that the ventral and dorsal streams are functionally independent processing pathways. 2|Page 1. Introduction The primate visual cortex is a busy place, estimated to contain more than 40 areas (Tootell, Tsao, & Vanduffel, 2003; Van Essen, 1985, 2005). Neuroscientists have naturally sought a simplifying order to this complexity. Ungerleider and Mishkin (1982) proposed a primary functional division between the ventral and dorsal streams that diverge from extrastriate cortex. They suggested that the ventral stream is specialised for processing visual features, such as shape, pattern, texture and colour, whilst the dorsal stream analyses the spatial aspects of those features. This framework, though still influential, has been superseded by an alternative formulation, introduced by Goodale and Milner (1992), and receiving its fullest exposition in “The visual brain in action” (Milner and Goodale, 1995, 2006). Milner and Goodale argued that a scheme focused on stimulus attributes cannot capture the division of labour between the visual streams. Since natural selection can act only upon real-world outcomes, understanding the design of our visual systems requires that we consider the behavioural goals that vision serves. Vision exists because it improves our actions, but there are more and less direct ways in which it does this. The direct route is via the real-time guidance of action. The less direct route is via the deciphering of visual patterns, enabling us to understand our surroundings and lay down visual memories, and thus to plan actions informed by past experience. Within the perception-action model, the latter role, corresponding better to classical ideas of ‘perception’, is fulfilled by the ventral stream. The direct visual guidance of action is the exclusive province of the dorsal stream. 3|Page The perception-action model proposes no absolute distinction between the visual attributes of our environment which are analysed by the two cortical streams. For instance, the spatial layout of visual scenes and the internal geometry of objects are critical concerns for both pathways. This implies no redundancy, however, because the two streams analyse spatial information in different ways, according to their own behavioural goals. Visual representations for action encode egocentric relationships between the observer and the object of an action. These relationships are fleeting, so they are updated continuously and decay rapidly. Our capacity-limited cognitive selves are shielded from the burdensome details of transforming these representations into motor plans, because the dorsal stream is cognitively impenetrable, and neither requires nor evokes visual awareness. Visual representations for perception, on the other hand, must abstract from egocentric-viewpoints to encode the more stable, viewer-invariant properties of a scene. Ventral stream representations capture important material for visual memory and higher cognition, and furnish our visual awareness. These contrasting properties of the dorsal and ventral streams are often presented as a series of dichotomies (Table 1). (Table 1 about here) Milner and Goodale’s (1995) updating of Ungerleider and Mishkin’s (1982) functional model was accompanied by an anatomical reframing. Ungerleider & Mishkin’s model was mapped onto monkey anatomy, with the ventral stream passing to the inferior temporal lobe, and the dorsal stream to the inferior parietal lobe (IPL). However, Milner and Goodale argued that important differences between human and monkey brains imply that functionally homologous cortical areas will not always be found in anatomically 4|Page similar locations. Therefore they focused on functional properties to identify the human homologues of monkey cortical areas. In particular they proposed that the human dorsal stream does not project to the IPL, but instead to the superior parietal lobe (SPL). In contrast, the human IPL was suggested to fall outside of both visual streams, though having closer functional affinity to the ventral stream. The right IPL, given its association with unilateral neglect and constructional apraxia, may be specialised for supramodal spatial cognition, whereas the left IPL, given its association with limb apraxia, may be specialised for complex action sequences and action planning. Rizzolatti & Matelli (2003), however, have rejected the “rather strange proposal” that monkey IPL functions have relocated across the intraparietal sulcus to the SPL in humans. Less controversial is Milner and Goodale’s reassignment of motion-sensitive area MT (Milner & Goodale, 2006), classically regarded as part of the dorsal stream. Milner and Goodale argued that MT projects heavily to both dorsal and ventral streams, so that its status should be like that of other prestriate areas that feed information to both visual streams. This is now supported by correlated deficits of perception and action in a patient with akinetopsia following bilateral MT lesions (Schenk, Mai, Ditterich, & Zihl, 2000), and in healthy people undergoing transcranial magnetic stimulation (TMS) to area MT (Schenk, Ellison, Rice, & Milner, 2005). In the Epilogue to the second edition of “The visual brain in action”, Milner and Goodale (2006) survey the ample evidence that has accumulated to support and refine their model since its inception. In the present critical review, we lay more emphasis upon empirical and theoretical challenges to the model. We will base our discussion around the core 5|Page dichotomies listed in Table 1, adding an additional section on the topic of illusions-inaction (Section 4). Here, and elsewhere, the relevant data on vision-for-action will come predominantly from studies of reaching and grasping, reflecting the strong bias towards these actions in the literature. Overall, our discussion will support the broad spirit of the perception-action model, but will also identify instances in which its predictions have been refuted, and grounds on which its assumptions may be questioned. A common theme will be that the functional independence of the two visual streams may have been overestimated, and that the specialisations proposed may be relative rather than absolute. In the final section we consider the legacy of the perception-action model, and future directions for the field. 2. Perception versus action: ventral versus dorsal stream The core neuropsychological evidence for the proposed dissociation between vision-forperception and vision-for action is a set of contrasting behavioural impairments resulting from focal lesions to the ventral or dorsal streams. The most intensively studied patient with damage to the ventral stream is DF, a woman who developed visual form agnosia as a consequence of anoxic lesions, destroying the lateral occipital area (LO) bilaterally (James, Culham, Humphrey, Milner, & Goodale, 2003). DF was shown to use visual size and orientation proficiently to guide her hand for reaching and grasping, despite being unable to discriminate these properties explicitly. It was suggested that her preserved visuomotor abilities result from the sparing of her dorsal visual pathway (see Karnath, Rüter, Mandler & Himmelbach, 2009 for an even clearer case anatomically). In support of this view, Milner and Goodale contrasted DF’s condition with that of patients with 6|Page optic ataxia, who show inaccurate visually guided movement following damage to the SPL (Karnath & Perenin, 2005; Perenin & Vighetto, 1988). Naturally occurring lesions do not respect anatomists' boundaries, and neither DF nor most optic ataxic patients have a perfectly selective, let alone complete, lesion of either visual stream (see James et al, 2003, for a detailed analysis of DF’s lesion). Nonetheless, at least to gross analysis, vision-for-action and vision-for-perception are separately susceptible to disruption from brain damage, suggesting that they are segregated in the normal brain. Of course, vision-for-perception and vision-for-action must interact at some level, since we are manifestly capable of deciding a course of action on the basis of what we consciously see. Milner and Goodale (2008) have underscored a distinction between action planning and programming as the critical barrier between dorsal and ventral stream influences on behaviour. Vision-for-perception informs awareness, and influences action planning, but these actions are programmed via the transformation of afferent visual information into motor commands, which is the domain of vision-for-action. As neat as this scheme seems, it will not work without exceptions and qualifications. First, not all metrical parameters of our actions can be programmed on the basis of bottom-up information. When we lift an object, visual size and memory cues influence our fingertip forces (Flanagan & Beltzner, 2000; Gordon, Forssberg, Johansson, & Westling, 1991; Mon-Williams & Murray, 2000), as must visually-based judgements of the object’s material. Force programming thus involves visual recognition, either of a specific object or the stuff of which it is made. A critical role for the perceptual pathway is supported by the finding that DF does not use visual cues to program her fingertip forces (McIntosh, 7|Page 2000). The ventral stream’s influence on action thus reaches down to the programming level, at least for these force parameters. Perhaps the key to ventral stream involvement in action is not high-level planning, but semantic involvement, exemplified by picking up tools, which must be recognized before an appropriate grasp can be selected. DF’s behaviour suggests that this scheme will not work either. She does indeed make semantic errors when interacting with tools, grasping them in non-functional ways (Carey, Harvey, & Milner, 1996), but she also makes errors when grasping neutral blocks, in that she does not select the most comfortable grasp (Dijkerman, McIntosh, Schindler, Nijboer & Milner, 2009). Selection of actions that minimise postural awkwardness might be classed as action planning, but it has no obvious semantic component. DF is unable to make accurate memory-guided actions, which should obviously lie beyond the competence of a dorsal stream evolved for immediate visuomotor guidance (Goodale, Jakobson & Keillor, 1994a), but she also fails at immediate guidance tasks of any significant complexity. She can post a simple plaque through a slot adeptly, but makes systematic errors with a T-shaped object (Goodale, Jakobson, Milner, Perrett, Benson & Hietanen, 1994b), and is poor at determining contact points for grasping an X-shaped object, supporting the idea that the dorsal stream can process simple spatial attributes, such as the orientation of an object's principal axis of elongation, and its height and width with respect to this, but is unable to process more complex shapes (Carey et al, 1996). When presented with objects to be grasped by holes cut into them, DF’s grasping resembles blind groping (Dijkerman, Milner, & Carey, 1998; McIntosh, Dijkerman, Mon-Williams, & Milner, 2004). No simple criterion based 8|Page on planning-level or semantic involvement predicts the visuomotor tasks with which DF struggles and which, by implication, benefit from ventral stream participation in the normal brain. The empirical evidence suggests that DF’s visuomotor competence extends only to the spatially-based programming of rudimentary actions to simple, arbitrary targets. However, even her success in such rudimentary tasks may not be supported by the same combination of visual cues that healthy subjects employ for such tasks. Perturbation studies have revealed that DF is almost exclusively reliant on vergence angle and vertical gaze angle for the programming of reach amplitude (Mon-Williams, McIntosh, & Milner, 2001; Mon-Williams, Tresilian, McIntosh, & Milner, 2001), and on binocular disparities and motion parallax to recover object depth for grasping (Dijkerman & Milner, 1998; Dijkerman, Milner, & Carey, 1996; Dijkerman, Milner, & Carey, 1999). When these cues are perturbed, DF’s performance drops off precipitously. Normal performance is much less vulnerable to these manipulations, suggesting that healthy participants make use of additional depth and distance cues unavailable to DF, presumably mediated by the ventral stream. This implies that the ventral stream participates even in the low-level, spatiallybased programming of rudimentary actions (cf. McIntosh & Lashley, 2008). DF, with her quasi-isolated dorsal stream, succeeds in simple visuomotor tasks, but this testifies to the robustness of the system in adapting to a degraded perceptual analysis in the ventral stream, rather than being a model of visuomotor control in the healthy brain. The evidence points towards ventral stream involvement at all levels of action planning and programming, leaving only fast online updating of movements, and possibly some 9|Page implicit forms of obstacle avoidance, as a potentially ‘pure’ dorsal task (cf. Pisella, Binkofski, Lasek, Toni & Rossetti, 2006; Rossetti, Pisella & Vighetto, 2003; Rizzolatti & Matelli, 2003). We now turn to the complementary claim, that the dorsal stream is not involved in visual perception. This discussion has been informed by findings from patients with optic ataxia following dorsal stream damage. The clinical diagnosis of optic ataxia requires that perceptual deficits be excluded as an explanation for misreaching under visual guidance, but this has often been done only coarsely. More stringent assessments, focusing on discriminations in the peripheral visual field, for which misreaching is most severe, have often observed impaired discrimination of object location or orientation (Michel & Henaff, 2004; Perenin & Vighetto, 1988; Pisella et al, in press), which may be secondary to attentional disorders (Michel & Henaff, 2004; Pisella et al, in press; Striemer et al., 2007; Striemer et al., 2009). Milner and Goodale (1995) recognise that dorsal stream areas play key roles in attention, and that the functional relationship between action programming and spatial attention may be a deep one (Rizzolatti, Riggio & Sheliga, 1994). It is thus consistent with the perception-action model that dorsal stream lesions should influence perception. However, this raises a thorny issue of definition. Some would argue that, where attention plays a necessary role in visual awareness, it should be considered fully part of the perceptual process (Pisella et al, in press). Milner and Goodale, however, class this role as indirect, since attention may filter the contents of awareness, but it is the ventral stream that provides those contents. A true perceptual role for the dorsal stream, on Milner and Goodale’s terms, would require the products of 10 | P a g e dorsal stream processing to enter visual awareness. Possible evidence of this sort will be considered in Section 6, which addresses the issue of awareness in detail. We would conclude that, while the mapping of vision-for-perception and vision-foraction onto the ventral and dorsal stream respectively reflects an important, first-order organising principle, the characterisation is hard to maintain in any strict sense. Part of the problem is that vision-for-perception and vision-for-action are not self-explanatory categories, but theoretical constructs requiring further definition; and part of the success of the perception-action model may owe to a degree of elasticity in these key terms. If vision-for-action is whatever the dorsal stream does, then this definition can evolve to exclude aspects or forms of action that transpire not to have ‘pure’ dorsal status. We have argued that the shrinking island of ‘pure’ dorsal action may now include only certain forms of on-line control, all other visuomotor skills being contaminated by contributions from other areas. At some point, this custom definition of vision-for-action may diverge so far from what we would normally mean by this term that it becomes counterproductive. This notion figures prominently in three substantive criticisms of Milner and Goodale’s account (Jeannerod & Jacob, 2005; Rizzolatti & Matelli, 2003; Rossetti et al, 2003), and similar concerns were noted earlier with respect to vision-for-perception (Pisella et al, in press). However, much of the tension can be relieved simply by relaxing the assumption that the functional domains of the dorsal and ventral stream are independent and mutually exclusive. The broad functional distinction captured by the perception-action model is relative, not absolute. It may be possible to identify some simple visual behaviours at the extreme ends of a ventral-dorsal continuum (e.g. colour 11 | P a g e discrimination vs. online movement correction), but most visual behaviours, especially those of any significant complexity, are likely to involve collaboration between both visual streams, and other brain systems. 3. Spatial codes for perception and action Within the perception-action model, a major rationale for the segregation of vision-forperception from vision-for-action is that spatial coding requirements differ fundamentally between these domains. For action, an object’s position must be coded egocentrically and its spatial dimensions measured in absolute metrics, i.e. in terms of its real physical dimensions. Vision-for-perception, by contrast, is concerned with viewer-invariant relationships in the environment, so spatial features should be coded in allocentric frames of reference. This perceptual strategy applies also to sizes and distances, which will be encoded relative to one another, rather than in absolute metrics. These theoretical considerations underpin the expectation of distinct spatial coding systems in the dorsal and ventral streams. How well do the data conform to this expectation? There is abundant neurophysiological evidence for egocentric coordinate transformations of visual information in the posterior parietal cortex of the monkey, made possible by the convergence of proprioceptive and visual information. For instance, in areas LIP and 7a, visual targets have been found to be represented in retinocentric (Mountcastle, Lynch, Georgopoulos, Sakata, & Acuña, 1975), head-centred (Andersen, Essick, & Siegel, 1985) and body-centred frames of reference (Snyder, Grieve, Brotchie, & Andersen, 1998). Buneo and Anderson (2006) have recently described a chain of transformations in the 12 | P a g e parietal reach region, whereby the coding of visual target and hand positions within a common retinocentric frame allows the extraction of the movement vector required to reach the target (functionally equivalent to hand-centred coding). It is clear that the posterior parietal cortex does not construct a purely sensory representation of space, but computes action-relevant, viewer-dependent, encodings. There is also abundant complementary evidence for viewpoint-invariant spatial coding in monkey inferior temporal cortex (see Logothetis & Sheinberg, 1996, for a review). Recently, fMR adaptation (Grill-Spector et al., 1999), which relies on the reduction of neuronal responses to repeated presentations of equivalent stimuli, has begun to provide insights into representational codes in the human brain. Consistent with expectation, ventral stream area LO adapts to repeated presentations of the same shape, across variations in viewpoint (James, Humphrey, Gati, Menon, & Goodale, 2002; Kourtzi & Kanwisher, 2001; Vuilleumier, Henson, Driver, & Dolan, 2002), including changes in object size (Grill-Spector et al., 1999; Sawamura, Georgieva, Vogels, Vanduffel, & Orban, 2005). Valyear and colleagues (Valyear, Culham, Sharif, Westwood, & Goodale, 2006) contrasted responses in temporo-occipital cortex with responses in parieto-occipito cortex across repeated presentations of objects varying in identity, orientation, or both. They reported doubly dissociated patterns of activation, such that the ventral stream adapted to identity, regardless of orientation, whilst the dorsal stream adapted to orientation, regardless of identity. However, Konen & Kastner (2008) found a remarkably close mirroring between ventral and dorsal streams, with early and intermediate visual areas showing fMR adaptation only to low-level visual features, but some higher-level 13 | P a g e areas adapting to object identity, regardless of changes in viewpoint or size. This viewpoint invariance was observed for area LO in the ventral stream, and for dorsal stream regions in the posterior intraparietal sulcus. This latter result is inconsistent with an absence of allocentric codes in the dorsal stream. A number of neurophysiological findings also show allocentric encoding of space in dorsal stream areas. Snyder et al. (1998) found that, whereas visual targets were represented in monkey LIP predominantly in head-centred coordinates, the majority of neurons they sampled in area 7a reflected world-centred coordinates, coding target position in an environmental frame. Subsequently, in an object-based spatial task, Crowe, Averbeck and Chafee (2008) found that a transformation from retinal to object-centred coordinates arises, or is at least reflected, in area 7a, suggesting that this area may be capable of a variety of allocentric encodings. If the monkey IPL is homologous with the human SPL, as Milner and Goodale suggest, then the allocentric encoding of space in monkey area 7a should map to the human dorsal stream. If the monkey IPL is homologous to the human IPL, as Rizzolatti and Matelli (2003) maintain, then these allocentric encodings would be compatible with a role for the IPL in higher-level spatial representation, but the homology would imply that spatial representations in the human IPL, as in the monkey, are involved in spatial coding for action. In either case, these neurophysiological data are problematic for the perception-action model. Neurophysiological and neuroimaging evidence for allocentric spatial coding in the dorsal stream needs to be reconciled with evidence from lesion studies. Experiments with 14 | P a g e patient DF suggest that she is fundamentally limited in her ability to make use of allocentric spatial relationships, whether making a perceptual judgement or an action response (Schenk, 2006). One implication could be that the allocentric representation of space in the dorsal stream relies on input from the ventral stream. This would be compatible with the perception-action model, in the sense that allocentric coding would be held to depend upon ventral stream processing. However, it would deny the categorical claim that the dorsal stream does not use allocentric codes, and would admit a very close interaction between visual streams in the processing of objects and space, undermining the idea that the two streams are meaningfully independent. Accepting that the dorsal stream uses allocentric representations does not necessarily challenge its visuomotor role. On the contrary, the need for absolute metrics requires that the object representations guiding action must respect size constancy, a cardinal form of viewpoint-invariance. Milner and Goodale (1995) propose that the visuomotor system employs exclusively absolute metrics, whilst perceptual representations are concerned with relative metrics. It is argued that relational coding must be used for perception because “if perceptual representations were to attempt to deliver the real metrics of all objects in the visual array, the computational load would be astronomical” (Goodale & Humphrey, 1998, p195; see also Milner & Goodale, 2006; Milner & Goodale, 2008). However, it is not self-evident that it should be any less onerous to compute size and distance for every object relative to some other object. Nor is it obvious why, if the action system can deliver absolute metrics on demand, the perceptual system could not. We may also question the intuition that egocentric, absolute representations are not useful within 15 | P a g e visual awareness. If a key role of visual awareness is to inform action planning, then such representations may well be relevant (for further discussion, see Section 6). The theoretical rationale for restricting vision-for-perception to relative metrics can be questioned, but what about the complementary claim that relative metrics are not used by the visuomotor system? This claim has engendered the idea that the visuomotor system should ignore relative sources of size and distance information, such as the majority of pictorial depth cues, processing the action target in isolation from its visual context. Against this view, computational analyses of the information available in multi-cue environments do not support strict distinctions between absolute and relative cues: relative depth cues can be ‘promoted’ by other cues in the scene to provide absolute distance information (Landy, Maloney, Johnston & Young, 1995). Moreover, there is a wealth of empirical evidence to show that action systems are sensitive to visual context. Numerous studies have found that visuomotor performance is improved when visual context is available (for example: Bautista & Korienek, 1999; Coello & Iwanow, 2006; Coello & Magne, 2000; Krigolson, Clark, Heath, & Binsted, 2007; Redon & Hay, 2005). Interestingly, these benefits apply not only to action programming (Obhi & Goodale, 2005) but also to online control (Krigolson & Heath, 2004). Action-based responses tend to give proportionately higher weight to binocular and absolute depth and distance cues than do perceptually-based responses (e.g. Knill, 2005), but there seem to be no hard barriers to action systems using pictorial depth cues, even those that depend upon objectknowledge (McIntosh & Lashley, 2008). The most relevant constraints may be imposed 16 | P a g e by the latencies with which different cues can be made available (Greenwald, Knill & Saunders, 2005). 4. Visual illusions in action Proponents of the perception-action model have advocated context-based visual illusions as a simple tool to test the idea that the two visual streams use different spatial representations. Since the relation between a visual target and its surrounding context is presumed to be represented in the ventral, but not the dorsal stream, the action outputs of the dorsal stream should be unaffected by contextual illusions. Initial reports of such dissociations were highly influential (e.g. Bridgeman, Lewis, Heit & Nagle, 1979; Aglioti, DeSouza & Goodale, 1995). More recently, this methodological paradigm has attracted scepticism, on empirical and theoretical grounds, arguably becoming the most controversial pillar in the perception-action framework (for critical reviews, see: Bruno, 2001; Carey, 2001; Franz & Gegenfurtner, 2008; Franz, 2001; Smeets & Brenner, 2006). Rather than providing yet another review of this topic, we will give only a short summary before considering the implications of the debate for the perception-action model. Many studies have now reported that visual illusions do affect actions, though these effects may typically be smaller than those on perceptual judgements (for reviews and meta-analyses, see: Bruno, Bernardis, & Gentilucci, 2008; Bruno & Franz, 2009; Franz & Gegenfurtner, 2008). Differences in illusion strength between response modalities have been ascribed to a host of methodological factors, such as differences in visual feedback (Bruno & Franz, 2009; Franz, Hesse, & Kollath, 2009, Mon-Williams & Bull, 2000), or 17 | P a g e the requirement for relative versus absolute judgments (Pavani, Boscagli, Benvenuti, Rabuffetti, & Farne, 1999; Vishton, Pea, Cutting, & Nunez, 1999) in perception- and action-based responses. In the case of the Ebbinghaus-Titchener illusion, appropriate matching between response conditions yields comparable and correlated effects of the illusion on perception and action (Franz, Gegenfurtner, Bulthoff, & Fahle, 2000; Pavani et al., 1999). The clear influence of methodological factors on illusion strength, within as well as between perception- and action-based varieties of response, suggests that the perception-action distinction does not provide a satisfactory account of the data. Multiple task-specific explanations may be needed (Dassonville & Bala, 2004; Smeets & Brenner, 2006; Smeets, Brenner, de Grave, & Cujpers, 2002). Smeets and colleagues (2002) argue that specific visual illusions affect specific visual attributes but not others, in ways that may be mutually inconsistent. Whether or not a given response will be affected by the illusion depends simply upon whether it uses the affected attribute. For example, several investigators have noted that pointing responses to the endpoints of Muller-Lyer figures may be more susceptible to illusory effects when initiated from a point on the shaft as opposed to from outside the figure (Bruno et al, 2008; see also Post and Welch, 1996; Wraga, Creem & Profitt, 2000). Smeets et al (2002) note that the fins of the Muller-Lyer figure change the apparent extent of the central shaft but not the apparent positions of its endpoints (Gillam & Chambers, 1985; Mack, Heuer, Villardi & Chambers, 1985). They thus argue that a movement towards an endpoint may be scaled to the extent of the shaft if it begins from a point on the shaft, but be directed at the endpoint position if it begins elsewhere, being prone to the illusion of extent only in 18 | P a g e the former case. More generally, apparent dissociations between different responses to illusions may always reflect a mismatch between the attributes guiding the different responses. To explain any such pattern requires an understanding of which visual attributes are affected by a specific illusion, and how these attributes are weighted in the guidance of specific responses. These details are not reducible to any general dichotomy. Finally, even if we have a good functional model of how an illusion works, its behavioural consequences cannot inform any anatomical hypothesis, such as the anatomical component of the perception-action model, unless we also know where in the brain the illusion is represented. Until recently, for example, an untested assumption of the illusions-in-action paradigm was that pictorial depth is interpreted in the ventral stream, so that only the ventral stream should represent illusions induced by pictorial depth. However, a recent fMRI study concluded that the perceptual enlargement of objects presented against a receding linear perspective background (Ponzo illusion) is reflected in primary visual cortex (Murray, Boyaci & Kersten, 2006). Even if we allow that these effects may depend on top-down feedback from ventral stream areas, the implications of this finding for the illusions-in-action paradigm are severe. If the Ponzo illusion can affect early visual representations, the perception-action model predicts that the illusory changes should affect perception and action equally, and prior reports of action resistance to the Ponzo illusion (Brenner & Smeets, 1996; Jackson & Shaw, 2000; Stöttinger & Perner, 2006) must have some explanation other than that previously offered by the model. This in turn undermines the idea that dissociated responses to any visual illusions would provide differential support for the perception-action model. 19 | P a g e The illusions-in-action paradigm now looks quite jaded. Its inconsistent results may be better explained by task-specific accounts, and at least some of its key assumptions may be wrong. Consequently, this large literature supports remarkably few firm conclusions regarding the perception-action model, except to reinforce the evidence, reviewed in the previous section, that action responses are not generally immune to the influence of visual context or pictorial depth. Whether such influences stem from early vision, from the ventral stream, or from the dorsal stream itself, is mostly unclear. Context-based illusions remain an unreliable localiser for ventral stream involvement, a fact that has repercussions in the next section, concerning the temporal properties of the two visual streams. 5. Temporal properties of perception and action Vision-for-perception can inform working memory representations that outlast the visual stimulation by several seconds, and enable long-term memories to be formed. Milner and Goodale (1995) propose that visuomotor representations, by contrast, decay rapidly so that only afferent vision can guide action via the dorsal stream. The strongest version of this view proposes that vision-for-action lacks all capacity to store information (Goodale, Westwood & Milner, 2003; Westwood & Goodale, 2003). Only actions to targets visible at the time of movement initiation are guided by visual information in the dorsal stream; any delay between target viewing and movement initiation forces visual control to switch over to stored representations from the ventral stream. Because ventral stream 20 | P a g e representations are optimised for visual recognition rather than action guidance, delayed actions should show predictable changes in character, as well as reductions in quality. Delayed grasping movements show reduced precision (Elliott & Madalena, 1987; Rossetti, 1998; Westwood, Heath, & Roy, 2001), and kinematic changes indicative of greater uncertainty, such as extended movement durations and larger maximum grip apertures (Bradshaw & Watt, 2002; Hu, Eagleson & Goodale, 1999). However, such changes could be explained by a single visual representation which decays upon the removal of visual information. Indeed, the performance decrement observed in grasping with increasing delay is well approximated by an exponential decay function, common across a wide range of memory-based tasks (Hesse & Franz, 2009). To assess ventral stream involvement specifically, Hu and Goodale (2000) compared the influence of a size-contrast illusion on immediate and delayed grasping. They found illusory effects only in delayed actions, taking this as evidence that such actions depend upon ventral stream guidance (see also Westwood & Goodale, 2003). However, this pattern has not been replicated for other illusions. For instance, the effect of Müller-Lyer figures on grasping depends upon whether or not visual feedback is available during execution, not whether initiation is immediate or delayed (Bruno & Franz, 2009; Franz et al, 2009; Westwood, McEachern & Roy, 2001; Heath, Rival, Westwood & Neely, 2005). Moreover, as noted in Section 4, there are sufficient problems with the assumption, that illusion strength indexes ventral stream involvement, that the significance of any such evidence is moot. 21 | P a g e The effect of (non-illusory) visual context provides slightly better quality evidence. Obhi and Goodale (2005; see also Hay & Redon, 2006) found that immediate and delayed reaches both benefited from visual context (environmental structure), but that this benefit was enhanced with delay. An increased weighting of contextual information in delayed reaching is consistent with an increased contribution of allocentric, relative to egocentric, spatial coding. However, as Obhi and Goodale (2005) pointed out, allocentric coding cannot be inferred unambiguously from the influence of visual context. The positions of non-target elements could be computed egocentrically yet influence target localisation, perhaps by supporting a more accurate calibration of available distance cues (Magne & Coello, 2002). Nonetheless, if we are prepared to assume that allocentric spatial coding does underlie these contextual influences, and to accept the idea that allocentric coding requires ventral stream involvement (Section 3), then these findings suggest that delayed actions do have an increased reliance on the ventral stream. More conclusive evidence might be expected from patients with selective damage to one or other visual stream. If the control of delayed actions switches to the ventral stream, then delayed actions should be impaired relative to immediate actions following ventral stream damage, but improved following dorsal stream damage. With respect to ventralstream damage, Goodale et al (1994) confirmed that DF’s good grip scaling was abolished by a 2 second delay. Subsequently, several studies have found the predicted complementary improvement with delay in patients with optic ataxia following dorsal stream lesions (Himmelbach & Karnath, 2005; Milner, Paulignan, Dijkerman, Michel & Jeannerod, 1999; Milner et al, 2001; Rice et al, 2008; Rossetti et al, 2005). However, 22 | P a g e analysis of the time-course of improvement does not support an abrupt switch from dorsal to ventral stream control, but rather a gradual improvement with delay duration (Himmelbach & Karnath, 2005). Moreover, optic ataxic performance remains severely abnormal, even with delay (Milner et al, 1999; 2001; Himmelbach & Karnath, 2005), suggesting that delayed actions do not bypass the dorsal stream. Functional imaging supports this inference, as immediate and delayed actions produced similar activation of surviving dorsal stream tissue in a patient with optic ataxia (Himmelbach et al, 2009). Dorsal stream activation during delayed action is further supported by functional imaging in healthy subjects (Connolly, Andersen, & Goodale, 2003). However, an interesting qualification has been suggested by Singhal and colleagues, who observed re-activation of ventral stream area LO at the time of delayed action (Singhal, Kaufman, Valyear, & Culham, 2006). Consistent with this, Rice-Cohen and colleagues found that TMS over LO affects only delayed action, but TMS over dorsal stream area AIP disrupts both immediate and delayed action (Rice-Cohen, Cross, Tunik, Grafton, & Culham, 2009). Overall the data suggest that the dorsal stream is involved both in the control of immediate and delayed action (see also Rogers, Smith, & Schenk, 2009), but that the ventral stream may provide crucial support for delayed action, so that such actions are severely impaired by damage to the ventral stream (Goodale et al, 1994a). Our conclusion is thus close to the original proposal of the perception-action model, though with important distinctions. First, delayed actions depend on collaboration between the two visual streams, not the substitution of dorsal by ventral stream control. Second, this collaboration is not unique to delayed actions, but also underlies immediate control (see 23 | P a g e Section 3), though the ventral stream contribution may be relatively more important when the flow of afferent visual information is turned off. 6. Awareness and the two visual streams Milner & Goodale propose that only the products of ventral stream processing reach visual awareness (Milner, 1995, 1998; Milner & Goodale, 1995, 2006, 2008). The ventral stream is thus necessary for visual awareness. Here, it may be helpful to adopt Koch’s (2004) distinction between necessary neural activity that determines the specific contents of awareness, and activity which does not determine specific contents, but creates enabling conditions for awareness. For instance, the reticular activating system may enable awareness, but its activity does not predict the specific content of awareness. By contrast, lesions within the ventral stream produce content-specific changes in experience, such as form agnosia and prosopagnosia, and neural activity within this stream covaries with perceptual report, even in the face of unchanging stimulation (for instance, during binocular rivalry, e.g. Andrews et al., 2002). Milner and Goodale (2006, Epilogue), provide an overview of neurophysiological and neuroimaging evidence that establishes specific roles for ventral stream sites in visual awareness. An important caveat to the above proposal is that ventral stream activity is not sufficient for awareness. This is true in that we are unaware of much of the processing within the ventral stream, and in that further enabling conditions are required for us to be visually aware of anything. Disorders such as neglect and extinction implicate parietal and frontal regions in visual awareness. This is consistent with brain imaging results which, 24 | P a g e alongside ventral stream sites, have often found frontal and parietal activations that correlate with changes in perceptual report (for reviews, see Rees, Kreiman & Koch, 2002; Rees, 2007). These fronto-parietal activations, including activations within the dorsal stream, might conceivably reflect changes in the specific content of visual awareness. Alternatively, they may be attentional modulations, determining which visual representations enter awareness. This latter interpretation would not challenge Milner and Goodale’s proposal. A true challenge would require evidence that activity outside of the ventral stream can determine the specific content of awareness. In fact, this evidence may already exist. Area MT has all the qualities expected of a specific site for visual awareness. Its stimulation modifies visual awareness of motion (Salzman, Britten & Newsome, 1990), its lesioning (Zihl, von Cramon & Mai, 1983, Schenk & Zihl, 1997) or inactivation (Beckers & Hömberg, 1992) induces selective akinetopsia, and its activity correlates with verbal reports of perceived motion, even when that motion is dissociated from afferent stimulation by being apparent (Kaneoke, Bundou, Koyama, Suzuki & Kakigi, 1997), illusory (Tootell et al, 1995; He, Cohen & Hu, 1998), imagined (Goebel, Khorram-Sefat, Muckli, Hacker & Singer, 1998), or even implied (Kourtzi & Kanwisher, 2000). MT is not within the ventral stream, but is classed by Milner and Goodale as an early visual area, with common feeder status for both visual streams (Section 1). By this reckoning, ventral stream processing does not produce all the contents of visual awareness. 25 | P a g e Leaving MT aside, we can approach Milner and Goodale’s proposal from another angle, and assess its implication that processing within the dorsal stream cannot reach visual awareness. Some of the most-cited evidence for this view comes from blindsight, in which patients may be able to direct saccadic or reaching movements towards stimuli that they do not report seeing (Weiskrantz, 1986). Remarkable as these abilities are, blindsight does not provide strong or specific evidence to dissociate the dorsal stream from visual awareness. First, although many authors emphasise residual action-abilities, there are numerous reports of residual perceptual discriminations in blindsight, for instance of colour, orientation (Weiskrantz, 1986, 1987), and form (Danckert et al, 1998; Trevethan, Sahraie & Weiskrantz, 2007) (for a review, see Danckert & Rossetti, 2005). Second, for actions and discriminations alike, preservation is only relative: above chance but inferior to normal performance. Blindsight does not imply that the normal operation of the dorsal stream is unconscious, since the unconscious abilities are not exclusively visuomotor, and are anyway far from normal. Stronger evidence is found where more fully preserved abilities are dissociated from visual awareness. For instance, a visual extinction patient has been shown to avoid obstacles (McIntosh et al, 2004) and to use brief visual feedback from the hand (Schenk, Schindler, McIntosh & Milner, 2005), regardless of whether he could report the relevant stimuli. Crucially, his ‘unaware’ performance was indistinguishable from his ‘aware’ performance, in contrast to the degraded implicit processing of extinguished stimuli typical in the perceptual domain (for a review, see Driver & Vuilleumier, 2001). Perhaps the strongest evidence, however, comes again from visual form agnosia, and patient DF 26 | P a g e (see also Karnath et al, 2009). DF’s visuomotor repertoire is vastly richer than that of any blindsight patient yet studied, as might be expected given the cortical inputs to her dorsal stream. Her ability to act systematically in accordance with visual attributes that she cannot explicitly report (e.g. size, orientation, shape) suggests that successful dorsal stream processing does not evoke, nor require visual awareness in the normal brain. However, recent evidence shows that DF is not entirely unable to report on the visual features that guide her actions, provided that the question is posed in the right way. Schenk (2006) found that DF’s verbal report in a visual distance-matching task was in the normal range if asked to report on egocentric distance, which presumably underlies her preserved visuomotor performance. Her visual size discrimination also improved significantly if probed just before or during a grasping movement towards the discrimination target (Schenk & Milner, 2006). These findings suggest that the cognitive system can access the information used in action. Whether or not this should qualify as ‘visual’ awareness is a more tricky matter (Clark, 2009). Schenk’s (2006) experiment raises the intriguing notion that the dorsal stream may contribute a heightened visual awareness of the world in relation to oneself. This egocentric quality, and its absence after dorsal stream damage, might be hard to articulate, making it easy to overlook (for a similar suggestion, see Riddoch, Humphreys, Jacobson, Pluck, Bateman & Edwards, 2004). Further possible evidence for this emerged recently, again from patient DF. Although she does not find it easy to describe her visual experiences, it is clear that she is vividly aware of colour and texture, and has stereo 27 | P a g e vision (Milner et al, 1991). Psychophysical investigations show that DF uses absolute binocular disparities to support absolute distance judgements, outperforming many controls, but is selectively impaired in using relative disparities between adjacent surfaces to judge their separation (Read, Phillipson, Milner & Parker, 2008). The prevailing view is that relative depth between surfaces is processed in the ventral stream, whilst absolute disparity is processed preferentially in the dorsal stream (e.g. Neri, 2005). One possible interpretation of these data is that DF’s experience of absolute depth derives from her dorsal stream, although a critical role for spared parts of her ventral stream cannot yet be ruled out. Overall, the exclusive identification of the contents of visual awareness with ventral stream processing is open to debate. A specific role for MT in the awareness of motion may be enough to refute the literal truth of the claim. On the other hand, there is considerable support for the spirit of the proposal, which assumes a prominent role for the ventral stream in furnishing visual awareness. Even so, we should not dismiss the idea that dorsal stream contributions shape the quality of visual awareness in ways that are no less real, though they may be less readily reportable. Finally, even if Milner and Goodale’s proposal is taken as literally true, it should be noted that we only become aware of the products of ventral stream processing given attentional activity in other areas, including the dorsal stream. This echoes our earlier conclusion (Section 2) that any visual function of significant complexity is likely to require the collaboration of multiple brain systems. 28 | P a g e 7. Conclusions The ability of the perception-action model to give a unifying and plausible account of diverse findings in neurophysiology, neuropsychology, functional imaging and experimental psychology has made it enormously influential over two decades. The conceptual simplicity and counterintuitive appeal of the model have promoted its wide dissemination and discussion within and beyond the field of cognitive neuroscience, and its assumptions and predictions have been targeted by hundreds of empirical papers. This brief review has necessarily been very selective. However, alongside support for the spirit of the model, we have found mixed evidence for its specific claims, especially for the high degree of functional independence assumed between the two visual streams. Milner and Goodale’s characterisations of the ventral and dorsal streams (Table 1) may be broadly correct, but the specialisations are relative, not absolute. The ventral stream is necessary for object and scene comprehension, and intimately involved in forging the contents of visual awareness. The dorsal stream is critical for the normal guidance of action. These functional roles are secure, but it is doubtful that they can be parsed into mutually exclusive domains of vision-for-perception and vision-foraction. Section 2 concluded that ventral stream influences on action guidance are not restricted to higher planning levels but extend even to the spatial programming of simple actions. The subset of visuomotor behaviours that do not benefit from ventral stream participation in the normal brain may be vanishingly small, perhaps restricted to fast online updating and implicit obstacle avoidance. Conversely, Section 6 proposed that the 29 | P a g e dorsal stream contributes egocentric spatial aspects to our visual awareness, participating directly in vision-for-perception. Sections 3 and 5 found some evidence to suggest that only the ventral stream can encode allocentric relationships between objects, or support sustained visual representations, consistent with Milner and Goodale’s view. On the other hand, the data imply that the dorsal stream can access and employ these ventrally-derived representations, again emphasising extensive inter-stream interactions. This highlights a recurrent theme of the present review: visual functions of any significant complexity are likely to involve collaboration between the two visual streams, and other brain systems. To take a prominent example, even if we were to accept an exclusive role for the ventral stream in providing the contents of visual awareness, awareness of these contents would still require attentional modulation from other brain areas. Some would argue that descriptive labels with such broad connotations as ‘perception’ or ‘action’ should not be applied to the specific contribution of any sub-division of the total participating network (cf. Jeannerod & Jacob, 2003; Pisella et al, in press; Rossetti et al, 2003). A major mission of the perception-action model has been to displace the entrenched classical conception of a unitary, all-purpose, visual representation of the world, so it is perhaps inevitable that the independence of the two streams has been emphasised over their interaction. However, one of the ironies of the model’s success is that the classical view has, to some extent, been replaced by the notion that our brains house two separate, unitary visual representations. The reality is much more fluid and task-specific than this. 30 | P a g e For instance, the various depth cues available in viewing a scene may be extracted in different brain areas, albeit with different latencies, then be generally available to inform any responses for which they are useful. Visuomotor responses tend to weight binocular and extra-retinal cues relatively more, and pictorial cues relatively less heavily than do perceptual responses, but evidence for categorical distinctions between the guiding representations is lacking. In defining and defending their model, Milner and Goodale’s (1995, 2008) stated goal has been to provide a theoretical framework to stimulate research. This goal has been met many times over. More fundamentally, their model has redefined the way that cognitive neuroscience views the functional specialisation of sensory systems, by prioritising the role of behavioural output requirements in determining where in the brain, and how, sensory information is analysed. The perception-action framework encompasses diverse experimental findings, and it would be self-defeating to reject such explanatory scope outright. On the other hand, this review suggests that each of the specific claims of the model regarding the specialisations of the two visual streams (Table 1), has either been refuted (Section 2, Section 6) or found limited empirical support (Sections 3, 4 & 5). Perhaps, then, we should view the model not as a formal hypothesis, but as a set of heuristics to guide experiment and theory. The differing informational requirements of visual recognition and action guidance still offers a compelling explanation for the broad relative specialisations of dorsal and ventral streams. However, to progress the field, we may need to abandon the idea that these streams work largely independently of one other, 31 | P a g e and to address the dynamic details of how the many visual brain areas arrange themselves from task-to-task into novel functional networks. References Aglioti, S., Desouza, J. F. X., & Goodale, M. A. (1995). Size-contrast illusions deceive the eye but not the hand. Current Biology, 5, 679-685. Andersen, R. A., Essick, G. K., & Siegel, R. M. (1985). Encoding of spatial location by posterior parietal neurons. Science, 230(4724), 456-458. Andrews, T. J., Schluppeck, D., Homfray, D., Matthews, P., & Blakemore, C. (2002). Activity in the fusiform gyrus predicts conscious perception of Rubin's vase-face illusion. Neuroimage, 17, 890-901. Bautista, A. B., & Korienek, G. G. (1999). Manual aiming in the presence of visual frames of reference. Journal of Human Movement Studies, 36(4), 153-166. Beckers, G., & Homberg, V. (1992). Cerebral visual-motion blindness - transitory akinetopsia Induced by Transcranial Magnetic Stimulation of Human Area V5. Proceedings of the Royal Society of London Series B-Biological Sciences, 249, 173178. Bradshaw, M. F., & Watt, S. J. (2002). A dissociation of perception and action in normal human observers: the effect of temporal-delay. Neuropsychologia, 40, 1766-1778. Brenner, E., & Smeets, J. B. J. (1996). Size illusion influences how we lift but not how we grasp an object. Experimental Brain Research, 111, 473-476. Bridgeman, B., Lewis, S., Heit, G., & Nagle, M. (1979). Relation between cognitive and motor-oriented systems of visual position perception. Journal of Experimental Psychology-Human Perception and Performance, 5, 692-700. Bruno, N. (2001). When does action resist visual illusions? Trends in Cognitive Sciences, 5, 379-382. Bruno, N., Bernardis, P., & Gentilucci, M. (2008). Visually guided pointing, the MullerLyer illusion, and the functional interpretation of the dorsal-ventral split: Conclusions from 33 independent studies. Neuroscience and Biobehavioral Reviews, 32, 423-437. 32 | P a g e Bruno, N., & Franz, V. H. (2009). When is grasping affected by the Muller-Lyer illusion? A quantitative review. Neuropsychologia, 47, 1421-1433. Buneo, C. A., & Andersen, R. A. (2006). The posterior parietal cortex: Sensorimotor interface for the planning and online control of visually guided movements. Neuropsychologia, 44, 2594-2606. Carey, D. P. (2001). Do action systems resist visual illusions? Trends in Cognitive Sciences, 5, 109-113. Carey, D. P., Harvey, M., & Milner, A. D. (1996). Visuomotor sensitivity for shape and orientation in a patient with visual form agnosia. Neuropsychologia, 34, 329-337. Clark, A. (2009). Perception, action, and experience: Unraveling the golden braid. Neuropsychologia, 47, 1460-1468. Coello, Y., & Iwanow, O. (2006). Effect of structuring the workspace on cognitive and sensorimotor distance estimation: No dissociation between perception and action. Perception & Psychophysics, 68, 278-289. Coello, Y., & Magne, P. (2000). Determination of target distance in a structured environment: Selection of visual information for action. European Journal of Cognitive Psychology, 12, 489-519. Connolly, J. D., Andersen, R. A., & Goodale, M. A. (2003). FMRI evidence for a 'parietal reach region' in the human brain. Experimental Brain Research, 153, 140145. Crowe, D. A., Averbeck, B. B., & Chafee, M. V. (2008). Neural ensemble decoding reveals a correlate of viewer- to object-centered spatial transformation in monkey parietal cortex. Journal of Neuroscience, 28, 5218-5228. Danckert, J., Maruff, P., Kinsella, G., de Graaff, S., & Currie, J. (1998). Investigating form and colour perception in blindsight using an interference task. Neuroreport, 9, 2919-2925. Danckert, J., & Rossetti, Y. (2005). Blindsight in action: what can the different sub-types of blindsight tell us about the control of visually guided actions? Neuroscience and Biobehavioral Reviews, 29, 1035-1046. Dassonville, P., & Bala, J. K. (2004). Perception, action, and Roelofs effect: A mere illusion of dissociation. Plos Biology, 2, 1936-1945. 33 | P a g e Dijkerman, H. C., McIntosh, R. D., Schindler, I., Nijboer, T. C. W., & Milner, A. D. (2009). Choosing between alternative wrist postures: Action planning needs perception. Neuropsychologia, 47, 1476-1482. Dijkerman, H. C., & Milner, A. D. (1998). The perception and prehension of objects oriented in the depth plane. II. Dissociated orientation functions in normal subjects. Experimental Brain Research, 118, 408-414. Dijkerman, H. C., Milner, A. D., & Carey, D. P. (1996). The perception and prehension of objects oriented in the depth plane. I. Effects of visual form agnosia. Experimental Brain Research, 112, 442-451. Dijkerman, H. C., Milner, A. D., & Carey, D. P. (1998). Grasping Spatial Relationships: Failure to Demonstrate Allocentric Visual Coding in a Patient with Visual Form Agnosia,. Consciousness and Cognition, 7, 424-437. Dijkerman, H. C., Milner, A. D., & Carey, D. P. (1999). Prehension of objects oriented in depth: motion parallax restores performance of a visual form agnosic when binocular vision is unavailable. Neuropsychologia, 37, 1505-1510. Driver, J., & Vuilleumier, P. (2001). Perceptual awareness and its loss in unilateral neglect and extinction. Cognition, 79, 39-88. Elliott, D., & Madalena, J. (1987). The influence of premovement visual information on manual aiming. Quarterly Journal of Experimental Psychology, 39A, 541-559. Flanagan, J. R., & Beltzner, M. A. (2000). Independence of perceptual and sensorimotor predictions in the size-weight illusion. Nature Neuroscience, 3, 737-741. Franz, V., & Gegenfurtner, K. (2008). Grasping visual illusions: Consistent data and no dissociation. Cognitive Neuropsychology, 25, 920-950. Franz, V. H. (2001). Action does not resist visual illusions. Trends in Cognitive Sciences, 5, 457-459. Franz, V. H., Gegenfurtner, K. R., Bulthoff, H. H., & Fahle, M. (2000). Grasping visual illusions: No evidence for a dissociation between perception and action. Psychological Science, 11, 20-25. Franz, V. H., Hesse, C., & Kollath, S. (2009). Visual illusions, delayed grasping, and memory: No shift from dorsal to ventral control. Neuropsychologia: Perception and Action, 47, 1518-1531. 34 | P a g e Gillam, B., & Chambers, D. (1985). Size and Position Are Incongruous - Measurements on the Muller-Lyer Figure. Perception & Psychophysics, 37, 549-556. Goebel, R., Khorram-Sefat, D., Muckli, L., Hacker, H., & Singer, W. (1998). The constructive nature of vision: direct evidence from functional magnetic resonance imaging studies of apparent motion and motion imagery. European Journal of Neuroscience, 10, 1563-1573. Goodale, M. A., & Humphrey, G. K. (1998). The objects of action and perception. Cognition, 67, 181-207. Goodale, M. A., Jakobson, L. S., & Keillor, J. M. (1994a). Differences in the visual control of pantomimed and natural grasping movements. Neuropsychologia, 32, 1159-1178. Goodale, M. A., Jakobson, L. S., Milner, A. D., Perrett, D. I., Benson, P. J., & Hietanen, J. K. (1994b). The nature and limits of orientation and pattern processing supporting visuomotor control in a visual form ggnosic. Journal of Cognitive Neuroscience, 6, 46-56. Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in Neuroscience, 15, 20-25. Goodale, M. A., Milner, A. D., Jakobson, L. S., & Carey, D. P. (1991). A neurological dissociation between perceiving objects and grasping them. Nature, 349, 154-156. Goodale, M. A., Westwood, D. A., & Milner, A. D. (2003). Two distinct modes of control for object-directed action. In: C.A. Heywood, A.D. Milner, C. Blakemore (Eds.) Roots of Visual Awareness - a Festschrift in Honor of Alan Cowey. Progress of Brain Research, 144, 131-144. Gordon, A. M., Forssberg, H., Johansson, R. S., & Westling, G. (1991). Visual size cues in the programming of manipulative forces during precision grip. Experimental Brain Research, 83, 477-482. Greenwald, H. S., Knill, D. C., & Saunders, J. A. (2005). Integrating visual cues for motor control: a matter of time. Vision Research, 45, 1975-1989. Grill-Spector, K., Kushnir, T., Edelman, S., Avidan, G., Itzchak, Y., & Malach, R. (1999). Differential processing of objects under various viewing conditions in the human lateral occipital complex. Neuron, 24, 187-203. 35 | P a g e Hay, L., & Redon, C. (2006). Response delay and spatial representation in pointing movements. Neuroscience Letters, 408, 194-198. He, S., Cohen, E. R., & Hu, X. P. (1998). Close correlation between activity in brain area MT/V5 and the perception of a visual motion aftereffect. Current Biology, 8, 12151218. Heath, M., Rival, C., Westwood, D. A., & Neely, K. (2005). Time course analysis of closed- and open-loop grasping of the Muller-Lyer illusion. Journal of Motor Behavior, 37, 179-185. Hesse, C., & Franz, V.H. (2009). Memory mechanisms in grasping. Neuropsychologia: Perception and Action, 47, 1532-1545. Himmelbach, M., & Karnath, H. O. (2005). Dorsal and ventral stream interaction: Contributions from optic ataxia. Journal of Cognitive Neuroscience, 17, 632-640. Himmelbach, M., Nau, M., Zündorf, I., Erb, M., Perenin, M.-T., & Karnath, H.-O. (2009). Brain activation during immediate and delayed reaching in optic ataxia. Neuropsychologia: Perception and Action, 47, 1508-1517. Hu, Y., Eagleson, R., & Goodale, M. (1999). The effects of delay on the kinematics of grasping. Experimental Brain Research, 126, 109-116. Hu, Y., & Goodale, M. A. (2000). Grasping after a delay shifts size-scaling from absolute to relative metrics. Journal of Cognitive Neuroscience, 12, 856-868. Jackson, S. R., & Shaw, A. (2000). The Ponzo illusion affects grip-force but not gripaperture scaling during prehension movements. Journal of Experimental Psychology-Human Perception and Performance, 26, 418-423. James, T. W., Culham, J., Humphrey, G. K., Milner, A. D., & Goodale, M. A. (2003). Ventral occipital lesions impair object recognition but not object-directed grasping: an fMRI study. Brain, 126, 2463-2475. James, T. W., Humphrey, G. K., Gati, J. S., Menon, R. S., & Goodale, M. A. (2002). Differential effects of viewpoint on object-driven activation in dorsal and ventral streams. Neuron, 35, 793-801. Jeannerod, M., & Jacob, P. (2005). Visual cognition: a new look at the two-visual systems model. Neuropsychologia, 43, 301-312. 36 | P a g e Kaneoke, Y., Bundou, M., Koyama, S., Suzuki, H., & Kakigi, R. (1997). Human cortical area responding to stimuli in apparent motion. Neuroreport, 8, 677-682. Karnath, H. O., & Perenin, M. T. (2005). Cortical control of visually guided reaching: Evidence from patients with optic ataxia. Cerebral Cortex, 15, 1561–1569. Karnath, H. O., Ruter, J., Mandler, A., & Himmelbach, M. (2009). The Anatomy of Object Recognition-Visual Form Agnosia Caused by Medial Occipitotemporal Stroke. Journal of Neuroscience, 29, 5854-5862. Knill, D. C. (2005). Reaching for visual cues to depth: The brain combines depth cues differently for motor control and perception. Journal of Vision, 5, 103-115. Koch, C. (2004) The quest for consciousness: A neuroscientific approach. Colorado: Roberts & Co. Konen, C. S., & Kastner, S. (2008). Two hierarchically organized neural systems for object information in human visual cortex. Nature Neuroscience, 11, 224-231. Kourtzi, Z., & Kanwisher, N. (2000). Activation in human MT/MST by static images with implied motion. Journal of Cognitive Neuroscience, 12, 48-55. Kourtzi, Z., & Kanwisher, N. (2001). Representation of perceived object shape by the human lateral occipital complex. Science, 293(5534), 1506-1509. Krigolson, O., Clark, N., Heath, M., & Binsted, G. (2007). The proximity of visual landmarks impacts reaching performance. Spatial Vision, 20, 317-336. Krigolson, A., & Heath, M. (2004). Background visual cues and memory-guided reaching. Human Movement Science, 23, 861-877. Landy, M. S., Maloney, L. T., Johnston, E. B., & Young, M. (1995). Measurement and modeling of depth cue combination - in defense of weak fusion. Vision Research, 3, 389-412. Logothetis, N. K., & Sheinberg, D. L. (1996). Visual object recognition. Annual Review of Neuroscience, 19, 577-621. Mack, A., Heuer, F., Villardi, K., & Chambers, D. (1985). The Dissociation of position and extent in Muller-Lyer figures. Perception & Psychophysics, 37, 335-344. Magne, P., & Coello, Y. (2002). Retinal and extra-retinal contribution to position coding. Behavioural Brain Research, 136, 277-287. 37 | P a g e McIntosh, R. D. (2000). Seeing size and weight. Trends in Cognitive Sciences, 4, 442444. McIntosh, R. D., Dijkerman, H. C., Mon-Williams, M., & Milner, A. D. (2004). Grasping what is graspable: Evidence from visual form agnosia. Cortex, 40, 695-702. McIntosh, R. D., & Lashley, G. (2008). Matching boxes: familiar size influences action programming. Neuropsychologia, 46, 2441–2444. McIntosh, R. D., McClements, K. I., Schindler, I., Cassidy, T. P., Birchall, D., & Milner, A. D. (2004). Avoidance of obstacles in the absence of visual awareness. Proceedings Of The Royal Society Of London. Series B. Biological Sciences, 271, 15-20. Michel, F., & Henaff, M. A. (2004). Seeing without the occipito-parietal cortex: Simultagnosia as a shrinkage of the attentional visual field. Behavioural Neurology, 15, 3-13. Milner, A. D. (1995). Cerebral correlates of visual awareness. Neuropsychologia, 33, 1117-1130. Milner, A. D. (1998). Streams and consciousness: visual awareness and the brain. Trends in Cognitive Sciences, 2, 25-30. Milner, A. D., Dijkerman, H. C., Pisella, L., McIntosh, R. D., Tilikete, C., Vighetto, A., et al. (2001). Grasping the past: delay can improve visuomotor performance. Current Biology, 11, 1896-1901. Milner, A. D., & Goodale, M. A. (1995). The visual brain in action. Oxford: Oxford Press. Milner, A. D., & Goodale, M. A. (2006). The visual brain in action (2nd ed.). Oxford: Oxford University Press. Milner, A. D., & Goodale, M. A. (2008). Two visual systems re-viewed. Neuropsychologia, 46, 774-785. Milner, A. D., Paulignan, Y., Dijkerman, H. C., Michel, F., & Jeannerod, M. (1999). A paradoxical improvement of misreaching in optic ataxia: new evidence for two separate neural systems for visual localization. Proceedings of the Royal Society of London Series B-Biological Sciences, 266, 2225-2229. 38 | P a g e Mon-Williams, M., & Bull, R. (2000). The Judd illusion: evidence for two visual streams or two experimental conditions? Experimental Brain Research, 130, 273-276. Mon-Williams, M., McIntosh, R. D., & Milner, A. D. (2001). Vertical gaze angle as a distance cue for programming reaching: insights from visual form agnosia II (of III). Experimental Brain Research, 139, 137-142. Mon-Williams, M., & Murray, A. H. (2000). The size of visual size cue used for programming manipulative forces during precision grip. Experimental Brain Research, 135, 405-410. Mon-Williams, M., Tresilian, J. R., McIntosh, R. D., & Milner, A. D. (2001). Monocular and binocular distance cues: insights from visual form agnosia I (of III). Experimental Brain Research, 139, 127-136. Mountcastle, V. B., Lynch, J. C., Georgopoulos, A. P., Sakata, H., & Acuña, C. (1975). Posterior parietal association cortex of the monkey: command function of operations within extrapersonal space. Journal of Neurophysiology, 38, 871-908. Murray, S. O., Boyaci, H., & Kersten, D. (2006). The representation of perceived angular size in human primary visual cortex. Nature Neuroscience, 9, 429-434. Neri, P. (2005). A stereoscopic look at visual cortex. Journal of Neurophysiology, 93, 1823-1826. Obhi, S. S., & Goodale, M. A. (2005). The effects of landmarks on the performance of delayed and real-time pointing movements. Experimental Brain Research, 167, 335344. Pavani, F., Boscagli, I., Benvenuti, F., Rabuffetti, M., & Farne, A. (1999). Are perception and action affected differently by the Titchener circles illusion? Experimental Brain Research, 127, 95-101. Perenin, M. T., & Vighetto, A. (1988). Optic ataxia: a specific disruption in visuomotor mechanisms. I. Different aspects of the deficit in reaching for objects. Brain: a Journal of Neurology, 111, 643-674. Pisella, L., Binkofski, F., Lasek, K., Toni, I., & Rossetti, Y. (2006). No doubledissociation between optic ataxia and visual agnosia: Multiple sub-streams for multiple visuo-manual integrations. Neuropsychologia, 44, 2734-2748. 39 | P a g e Pisella, L., Sergio, L., Blangero, A., Torchin, H., Vighetto, A. & Rossetti, Y. (in press). Optic ataxia and the function of the dorsal stream: contribution to perception and action. Neuropsychologia. Post, R. B., & Welch, R. B. (1996). Is there dissociation of perceptual and motor responses to figural illusions? Perception, 25, 569-581. Read J. C. A., Phillipson G. P., Milner A. D. & Parker A. J. (2008). Binocular depth perception in a human subject with bilateral lesions of the lateral occipital cortical area. FENS Abstracts, 4, A190.13. Redon, C., & Hay, L. (2005). Role of visual context and oculomotor conditions in pointing accuracy. Neuroreport, 16, 2065-2067. Rees, G. (2007). Neural correlates of the contents of visual awareness in humans. Philosophical Transactions of the Royal Society B-Biological Sciences, 362, 877886. Rees, G., Kreiman, G., & Koch, C. (2002). Neural correlates of consciousness in humans. Nature Reviews Neuroscience, 3, 261-270. Rice, N. J., Edwards, M. G., Schindler, I., Punt, T. D., McIntosh, R. D., Humphreys, G. W., et al. (2008). Delay abolishes the obstacle avoidance deficit in unilateral optic ataxia. Neuropsychologia, 46, 1549-1557. Rice-Cohen, N., Cross, E. S., Tunik, E., Grafton, S. T., & Culham, J. C. (2009). Ventral and dorsal stream contributions to the online control of immediate and delayed grasping: A TMS approach. Neuropsychologia: Perception and Action, 47, 15531562. Riddoch, M. J., Humphreys, G. W., Jacobson, S., Pluck, G., Bateman, A., & Edwards, M. (2004). Impaired orientation discrimination and localisation following parietal damage: On the interplay between dorsal and ventral processes in visual perception. Cognitive Neuropsychology, 21, 597-623. Rizzolatti, G., & Matelli, M. (2003). Two different streams form the dorsal visual system: anatomy and functions. Experimental Brain Research., 153, 146-157. Rizzolatti, G., Riggio, L., & Sheliga, B. M. (1994). Space and selective attention. In C. Umiltà & M. Moscovitch (Eds.), Attention and Performance XV. Conscious and Nonconscious Information Processing (pp. 231-265). Cambridge, MA: MIT Press. 40 | P a g e Rogers, G., Smith, D., & Schenk, T. (2009). Immediate and delayed actions share a common visuomotor transformation mechanism: A prism adaptation study. Neuropsychologia: Perception and Action, 47, 1546-1552. Rossetti, Y. (1998). Implicit short-lived motor representations of space in brain damaged and healthy subjects. Consciousness and Cognition, 7, 520-558. Rossetti, Y., Pisella, L., & Vighetto, A. (2003). Optic ataxia revisited: visually guided action versus immediate visuomotor control. Experimental Brain Research., 153, 171-179. Rossetti, Y., Revol, P., McIntosh, R., Pisella, L., Rode, G., Danckert, J., et al. (2005). Visually guided reaching: bilateral posterior parietal lesions cause a switch from fast visuomotor to slow cognitive control. Neuropsychologia, 43, 162-177. Salzman, C. D., Britten, K. H., & Newsome, W. T. (1990). Cortical Microstimulation Influences Perceptual Judgments of Motion Direction. Nature, 346, 174-177. Sawamura, H., Georgieva, S., Vogels, R., Vanduffel, W., & Orban, G. A. (2005). Using functional magnetic resonance imaging to assess adaptation and size invariance of shape processing by humans and monkeys. Journal of Neuroscience, 25, 4294-4306. Schenk, T. (2006). An allocentric rather than perceptual deficit in patient D.F. Nature Neuroscience, 9, 1345-1457. Schenk, T., Ellison, A., Rice, N. J., & Milner, A. D. (2005). The role of V5/MT+ in the control of catching movements: an rTMS study. Neuropsychologia, 43, 189-198. Schenk, T., Mai, N., Ditterich, J., & Zihl, J. (2000). Can a motion-blind patient reach for moving objects? European Journal of Neuroscience, 12, 3351-3360. Schenk, T., & Milner, A. D. (2006). Concurrent visuomotor behaviour improves form discrimination in a patient with visual form agnosia. European Journal of Neuroscience, 24, 1495-1503. Schenk, T., Schindler, I., McIntosh, R. D., & Milner, A. D. (2005). The use of visual feedback is independent of visual awareness: evidence from visual extinction. Experimental Brain Research, 167, 95-102. Schenk, T., & Zihl, J. (1997). Visual motion perception after brain damage: I. Deficits in global motion perception. Neuropsychologia, 35, 1289-1297. 41 | P a g e Singhal, A., Kaufman, L., Valyear, K., & Culham, J. C. (2006). fMRI reactivation of the human lateral occipital complex during delayed actions to remembered objects. Visual Cognition, 14, 122-125. Smeets, J. B. J., & Brenner, E. (2006). 10 Years of Illusions. Journal of Experimental Psychology: Human Perception and Performance, 32, 1501-1504. Smeets, J. B. J., Brenner, E., de Grave, D. D. J., & Cujpers, R. H. (2002). Illusions in action: consequences of inconsistent processing of spatial attributes. Experimental Brain Research, 147, 135-144. Snyder, L. H., Grieve, K. L., Brotchie, P., & Andersen, R. A. (1998). Separate body- and world-referenced representations of visual space in parietal cortex. Nature, 394, 887-891. Stottinger, E., & Perner, J. (2006). Dissociating size representation for action and for conscious judgment: Grasping visual illusions without apparent obstacles. Consciousness and Cognition, 15, 269-284. Striemer, C., Blangero, A., Rossetti, Y., Boisson, D., Rode, G., Vighetto, A., et al. (2007). Deficits in peripheral visual attention in patients with optic ataxia. Neuroreport, 18, 1171-1175. Striemer, C., Locklin, J., Blangero, A., Rossetti, Y., Pisella, L., & Danckert, J. (2009). Attention for action? Examining the link between attention and visuomotor control deficits in a patient with optic ataxia. Neuropsychologia, 47, 1491-1499. Tootell, R. B. H., Reppas, J. B., Dale, A. M., Look, R. B., Sereno, M. I., Malach, R., et al. (1995). Visual-motion aftereffect in human cortical area MT revealed by functional magnetic-resonance-imaging. Nature, 375, 139-141. Tootell, R. B. H., Tsao, D., & Vanduffel, W. (2003). Neuroimaging weighs in: humans meet macaques in "primate" visual cortex. Journal of Neuroscience, 23, 3981-3989. Trevethan, C. T., Sahraie, A., & Weiskrantz, L. (2007). Form discrimination in a case of blindsight. Neuropsychologia, 45, 2092-2103. Ungerleider, L. G., & Mishkin, M. (1982). Two cortical visual systems. In D. J. Ingle, M. A. Goodale & R. J. Mansfield (Eds.), Analysis of visual behavior. Cambridge, M.A.: MIT Press. 42 | P a g e Valyear, K. F., Culham, J. C., Sharif, N., Westwood, D., & Goodale, M. A. (2006). A double dissociation between sensitivity to changes in object identity and object orientation in the ventral and dorsal visual streams: A human fMRI study. Neuropsychologia, 44, 218-228. Van Essen, D. C. (1985). Functional organization of primate visual cortex. In A. Peters & E. G. Jones (Eds.), Visual Cortex (Vol. 3, pp. 259-329). New York: Plenum Press. Van Essen, D. C. (2005). Corticocortical and thalamocortical information flow in the primate visual system. Progress in Brain Research, 149, 173-185. Vishton, P. M., Pea, J. G., Cutting, J. E., & Nunez, L. N. (1999). Comparing effects of the horizontal-vertical illusion on grip scaling and judgment: Relative versus absolute, not perception versus action. Journal of Experimental Psychology-Human Perception and Performance, 25, 1659-1672. Vuilleumier, P., Henson, R. N., Driver, J., & Dolan, R. J. (2002). Multiple levels of visual object constancy revealed by event-related fMRI of repetition priming. Nature Neuroscience, 5, 491-499. Weiskrantz, L. (1986). Blindsight: A Case Study and Implications. Oxford: Oxford University Press. Weiskrantz, L. (1987). Residual Vision in a Scotoma - a Follow-up-Study of Form Discrimination. Brain, 110, 77-92. Westwood, D. A., & Goodale, M. A. (2003). Perceptual illusion and the real-time control of action. Spatial Vision, 16, 243-254. Westwood, D. A., Heath, M., & Roy, E. A. (2001). The Accuracy of Reaching Movements in Brief Delay Conditions. Canadian Journal of Experimental Psychology, 55, 304-310. Westwood, D. A., McEachern, T., & Roy, E. A. (2001). Delayed grasping of a MullerLyer figure. Experimental brain research, 141, 166-173. Wraga, M., Creem, S. H., & Proffitt, D. R. (2000). Perception-action dissociations of a walkable Muller-Lyer configuration. Psychological Science, 11, 239-243. Zihl, J., Von Cramon, D., & Mai, N. (1983). Selective Disturbance of Movement Vision after Bilateral Brain-Damage. Brain, 106, 313-340. 43 | P a g e Table 1. Core characteristics of the perception-action model. Ventral stream Dorsal stream Behavioural function vision for perception vision for action Spatial Properties allocentric coding/relative metrics egocentric coding/absolute metrics Temporal Properties sustained representations transient representations Visual Awareness critically linked to awareness independent of awareness 44 | P a g e