1. Pictures, Surfaces, and Depicted Objects

In this paper I want to defend the compatibility between NT and PT, which at first sight seem to be in conflict. Here, I offer a careful analysis and a regimentation of these two theses able to reveal that they are not in opposition. This will allow me to suggest that neuroscientists are justified in their everyday practice with pictures and philosophers are justified in arguing for the peculiar status of pictures as particular objects of perception.

The regimentation I propose is based on a background specification that turns out to be crucial. The fact that when perceiving pictures we perceive their two components, namely, a picture's surface or vehicle and a depicted scene or object, leads us to a very important distinction. It is one thing to talk about picture perception, which refers to both the perception of the surface and of the depicted object. It is another thing to talk about the perception of the depicted object, which is just one aspect of the phenomenon of picture perception. The other aspect is the perception of the surface. Thus, it seems that picture perception is different from face-to-face perception, in which our visual system represents only one thing: the object. This ‘twofold’ aspect of picture perception is uncontroversial in the literature (Hopkins Reference Hopkins2012; Lopes Reference Lopes1996, Reference Lopes2005; Kulvicki Reference Kulvicki2006; Wollheim Reference Wollheim1998; Levinson Reference Levinson1998; Nanay Reference Nanay2011, Reference Nanay2015, Reference Nanay2017; Matthen Reference Matthen2005; Abell Reference Abell2009; Abell and Bantinaki Reference Abell and Bantinaki2010; Briscoe Reference Briscoe, Hill and McLaughlin2016; Hecht et al. Reference Hecht, Schwartz and Atherton2003, see section 2). Naturally, I am not concerned with imagination and nonperceptual representations in picture perception here (Nanay Reference Nanay2017). Of course, during pictorial experience, these two components are not sharply separated from a perceptual point of view. However, much of the philosophical reflection that endorses the perceptual fusion of these two aspects investigates picture perception by analyzing the way in which different portions of our visual system are respectively attuned to the depicted object and to the surface (Nanay Reference Nanay2011: 464, Reference Nanay2015, Reference Nanay2017: sect. 2; Lopes Reference Lopes2005; Hopkins Reference Hopkins2012; Ferretti Reference Ferretti2016a, Reference Ferretti2016c, Reference Ferretti2017a). This is because ‘any account of seeing-in must be able to tell how this experience represents the picture's surface and how it represents the depicted object’ (Nanay Reference Nanay, Abell and Bantilaki2010b: 199, cf. Hopkins Reference Hopkins, Abell and Bantinaki2010). Here I endorse this methodology. Accordingly, vision science shows that the presence of a perceptual fusion between the two aspects, namely, the perception of the surface and the one of the depicted object, does not prevent us from examining, with specific experimental techniques and settings, how our visual brain builds separate conscious or unconscious visual representations of the surface and of the depiction and from studying them in isolation (Nanay Reference Nanay2011: 463, footnote 1; Reference Nanay2017: sect. 2; see section 2 below). This specification leads me to regiment NT in a sharper way. It is now possible to distinguish between two versions of this thesis:

We can endorse the Weak NT without endorsing the Strong NT. Depicted objects and normal objects foster in us the same visual response. This does not conflict with the idea that pictures and normal objects do not set us in the very same perceptual state. Indeed, during the perception of depicted objects our visual system can represent several visual properties it also represents in face-to-face perception. However, a depicted object is embedded in a surface. Therefore, unlike face-to-face perception, in picture perception there are two visual representations, one of the surface and the other one of the depicted object—let us call this the Double Representation Thesis. If we take these two representations separately—and a few lines above we saw that this move is licit—we can say that the visual representation of a normal object can be similar, from a visual point of view, either to that of a depicted object or to that of the picture's surface (Nanay Reference Nanay2011; Ferretti Reference Ferretti2016a, Reference Ferretti2017a), except for at least one crucial difference, described below. However, face-to-face and picture perception are overall different phenomena: the former involves only one visual representation, while the latter involves two representations.

There is a crucial point we have to take into account here. As mentioned above, we can endorse the view that during the perception of depicted objects, our visual system can represent several visual properties we also represent in face-to-face perception. This remains true even if depicted objects cannot foster in us the visual feeling of presence because they are embedded in a visible surface, (Ferretti Reference Ferretti2016c; Matthen Reference Matthen2005; Nanay Reference Nanay2015) - I will get back to this in section 2. The visual feeling of presence is related to the fact that we have a perception of the object as a present one that we can interact with in our action space (something possible with the perception of the surface). Thus, we can take into account this difference by sharpening the Weak NT:

It is easy to note that the Strong NT conflicts with PT. And the Weak NT is not sufficiently specific. But if we select the Sharpened Weak NT, we can reconcile neuroscience and philosophy. Indeed, saying that depicted and normal objects set us in a very similar perceptual state and that depicted objects can make our visual system represent several visual properties as normal objects do, can justify the neuroscientific practice. Also, this does not conflict with the Double Representation Thesis. We are just saying that the visual representation of the depicted object, which is only one of the two representations performed in picture perception—the other one is related to the surface—share many characteristics with the visual representation of a normal object. I will get back to an analysis of the nature of this similarity in section 3. This does not imply that the twofold perceptual processing involved in picture perception is the same perceptual state as the one we are in during face-to-face perception: face-to-face perception does not depend on the representational simultaneity involved in picture perception (section 1). Finally, this takes into account that only face-to-face perception can foster in us the visual feeling of presence. And this is precisely because in the case of picture perception we always represent the picture's surface—the Double Representation Thesis. Crucially, as we shall see (section 2), this representation makes it impossible to represent the depicted object as present. All this is in accordance with PT.

In what follows, I first report evidence in support of PT. Then, I report evidence in support of the Sharpened Weak NT. Finally, I suggest that if we select the Sharpened Weak NT, philosophy and neuroscience can be reconciled without any problem because this is perfectly compatible with PT.

2. The Philosophical Thesis

Here I suggest that philosophers are justified in arguing for the peculiar status of pictures as particular objects of perception. Thus, we can support PT.

Embracing the Double Representation Thesis (Lopes Reference Lopes2005; Nanay Reference Nanay2011, Reference Nanay2015, Reference Nanay2017; Ferretti Reference Ferretti2016c, Reference Ferretti2017a) leads us to a famous debate about whether we simultaneously visually represent the picture's surface and the depicted object (Wollheim Reference Wollheim and Wollheim1980) or whether, conversely, we visually alternate between them (Gombrich Reference Gombrich1960). Most of the current effort in the literature on picture perception has been geared toward choosing one of these two ideas (Nanay Reference Nanay2011, Reference Nanay2017; Cavedon-Taylor Reference Cavedon-Taylor2011; Chasid Reference Chasid2014; Hopkins Reference Hopkins2012; Kulvicki Reference Kulvicki2006). The balance leans on Wollheim's idea that we always simultaneously visually represent both the surface and the depicted object. However, it is specified that—different from Wollheim's initial insight—we usually consciously represent the depicted object, but we may or may not consciously see/attend to the picture's surface and, indeed, most of the time we do not (Nanay Reference Nanay2011: 463–64, Reference Nanay2017). We can take this simultaneity ‘not as simultaneous attention, but as simultaneous (conscious or unconscious) representations’ (Nanay Reference Nanay2015: 192; see also Nanay Reference Nanay2017). This implies, ipso facto, that we do not need to consciously represent both of them simultaneously (Nanay Reference Nanay2011, Reference Nanay2017; Hopkins Reference Hopkins2012). (This is a weak interpretation of Wollheim's idea of ‘Twofoldness’; Wollheim originally suggested the simultaneity of conscious representation.) I am using ‘conscious vision’ and ‘conscious visual attention’ interchangeably because I do not take part on the bigger debate on their relation when talking about picture perception (see Nanay Reference Nanay2011, Reference Nanay2017). The Double Representation Thesis follows this weaker interpretation of simultaneity according to which, most of the time in picture perception, we consciously visually represent the depiction and, at least unconsciously, visually represent the surface.

Several arguments support this notion of simultaneity. First, it is possible to have unconscious perception of either the surface or of the depicted object (Nanay Reference Nanay2011: footnote 1). If it were necessary to have conscious perception of both, this would rule out evidence about the possibility of unconscious picture perception; conscious perception of both is not necessary (Nanay Reference Nanay2011: 463; Reference Nanay2017). This is also confirmed by the fact that, most of the time, in our everyday experience of pictures, we tend consciously to focus on the depicted object while ignoring, from a conscious/attentional point of view, the surface. Nonetheless, we always, at least unconsciously, visually represent the surface (Nanay Reference Nanay2011: 463; Reference Nanay2015, Reference Nanay2017; Lopes Reference Lopes2005; Levinson Reference Levinson1998). Finally, the picture's surface is visually represented, but not necessarily consciously because perceiving pictures from an oblique angle does not lead to distortions. This phenomenon is explained by invoking the influence of the unconscious perception of the surface/vehicle on the perception of the pictorial content (Nanay Reference Nanay2011: 471; Reference Nanay2017: 5.1; Matthen Reference Matthen2005: 315–317; Hopkins Reference Hopkins2012: 657; for further crucial arguments against the idea of simultaneous conscious representation of both the surface and the depiction see Hopkins [Reference Hopkins2012] and Nanay [Reference Nanay2017]).

There is, however, another crucial argument showing that we always visually represent, at least unconsciously, the picture's surface: the representation of the surface is what prevents us from attributing the visual feeling of presence to the depicted object (Vishwanath Reference Vishwanath2014; Ferretti Reference Ferretti2016c);- there are particular visual features—glossiness, shininess, plasticity, and so on— that we need to detect in order to reach the feeling of presence (see Vishwanath Reference Vishwanath2014; Ferretti Reference Ferretti2016c). Indeed, when we cannot represent the surface, depicted objects look as present as normal objects do. In everyday life, however, depicted objects do not look as present as normal objects do. They look depicted, which is what they really are. Therefore, we are representing the surface.

Vision science has clearly shown this perceptual fact. For example, in an experiment subjects are made to look at a depicted scene monocularly through an oval aperture. The aperture is located in front of the eye such that it occludes the rectangular boundary of the image (the frame and its boundaries of the picture's surface). In this case, the depicted object is perceived as a normal present object occupying a portion of the space in front of the subject she/he can interact with (Vishwanath and Hibbard Reference Vishwanath and Hibbard2013: 1674; Vishwanath Reference Vishwanath2014: 153). This result is also drawn from both psychophysical and phenomenological evidence (Vishwanath Reference Vishwanath2014: 174; 2011: 224, 225; Vishwanath and Hibbard Reference Vishwanath and Hibbard2010, Reference Vishwanath and Hibbard2013; Ferretti Reference Ferretti2016c: 2.4; Barry Reference Barry2009).

Therefore, when the surface is not visible—for example, due to the invisibility of the contours—even depicted objects can look present to the observer (Vishwanath Reference Vishwanath2014; Ferretti Reference Ferretti2016c). In the words of Vishwanath and Hibbard:

The reader should note that stereopsis is the peculiar visual process that allows us to perceive an object as vividly present in our action space (Ferretti Reference Ferretti2016c). It depends on the possibility of representing absolute egocentric depth values (Vishwanath Reference Vishwanath2014). In face-to-face object recognition, unlike in picture perception, monocular and binocular depth cues, retinal and expected size, and viewing distance are consistent. Additional shape information is also available from stereopsis, which allows us to represent vergence and disparity correctly (Snow et al. Reference Snow, Pettypiece, MacAdam, McLean, Stroman, Goodale and Culham2011).

Furthermore, egocentric absolute depth concerns the fact that the ‘observer has knowledge of the depth relations scaled in some meaningful way to the actions of the observer’ (Vishwanath Reference Vishwanath, Albertazzi, van Tonder and Vishwanath2011: 222; see also p. 206). Ascribing absolute egocentric depth values to the objects we face in our environment is what leads us to represent visually an object as present in our action space and as offering reliable possibility for motor interaction (Ferretti Reference Ferretti2016c). When we cannot visually represent the surface, absolute egocentric depth values are assigned to the depicted objects. Thus, they look present.

This evidence strongly suggests that we do not visually represent depicted objects as present because the surface's visibility plays a crucial role in avoiding the induction of the visual feeling of their presence: the visual representation involved in the recognition of the vehicle/surface as present hinders the possibility of having the visual impression of the presence of the depicted object (Vishwanath Reference Vishwanath2014: 164; Ferretti Reference Ferretti2016c: 2.2). Thus, when we cannot visually represent the former, the latter looks present. But our usual binocular picture perception can properly visually assign the related distances to both the surface and the depiction. Thus, we visually represent the surface, and the depicted object is represented as such, that is, as a non-present pictorial object. This evidence explains why depicted objects do not look present, in line with the Double Representation Thesis.

Of course, there are genuine situations of picture perception, built without specific experimental settings, in which we cannot visually represent the surface, and thus, what effectively is a depicted object looks like a present one. One common example is the special case of trompe l'oeil pictures. Trompe l'oeil pictures are built in such a way that (even in binocular conditions and without apertures to look through) our visual system cannot properly detect the picture's surface. Thus, it is fooled by the illusory presence of very enhanced visual depth cues (see Ferretti Reference Ferretti2016c) wrongly attributed to the depicted object.

I do not need to focus deeply on this case here since this paper is about normal pictures. But this evidence is crucial because in line with the Double Representation Thesis it suggests that, during picture perception, we do see—or visually represent—both the surface and the depicted object. Following the evidence from vision science, if we were only representing the depicted object but not the surface, the former would look present, like a real object. Accordingly, in the case of trompe l'oeil perception, we also do not have an unconscious representation of the surface. If there were, at least an unconscious representation of the surface, we would not have the feeling of presence of the depicted object.

One might argue that with trompe l'oeil pictures we do perceive the surface unconsciously, and this might explain the difference between normal perception and perception of trompe l'oeil pictures (Nanay Reference Nanay, Abell and Bantilaki2010b, Reference Nanay2011: footnote 11, Reference Nanay2015, Reference Nanay2017: 5.1; Ferretti Reference Ferretti2016c: sec. 3). However, if this were the case, we would not be able to distinguish the trompe l'oeil case from what is taken to be the most usual case of picture perception, according to the literature, in which, following the notion of simultaneity embraced by the Double Representation Thesis, we consciously see the depicted object and perceptually realize it is a depicted object. At the same time, we do not consciously visually represent the surface because, most of the time, our conscious vision is only focused on the depiction (Lopes Reference Lopes2005; Levinson Reference Levinson1998; Nanay Reference Nanay2011, Reference Nanay2017). After all, both of these situations involve an unconscious visual representation of the picture's surface. However, crucially, we know that only one of them, namely, the trompe l'oeil one, fosters in us the feeling of presence. Therefore, invoking the unconscious representation of the surface cannot be the right explanation for trompe l'oeil perception. The only way to explain the difference is by saying that even if only for a moment our visual system cannot ‘find’ the picture's surface and also cannot do so at the unconscious level: not only are we not consciously visually representing the surface, as in the case of usual picture perception mentioned above, but our visual system cannot even build an unconscious representation of a surface (Ferretti Reference Ferretti2016c; Vishwanath Reference Vishwanath2014; Nanay Reference Nanay2017): Footnote ¹ it cannot track that surface's presence.

This point is very useful for understanding the importance of the presence, in usual picture perception, of an at least unconscious visual representation of the surface, which accompanies our conscious visual representation of the depicted object. The presence of this unconscious representation allows us to focus on/attend to the depicted object consciously without visually representing it as present. However, this representation also guarantees that it is possible to shift our conscious visual attention to the picture's surface, which is, then, consciously visually represented (Nanay Reference Nanay2017; i.e., the unconscious visual representation of the surface can sometimes be consciously accessed; I have little to say about this case here, which is usually called inflection, see Voltolini Reference Voltolini2013, Lopes Reference Lopes2005, Nanay Reference Nanay, Abell and Bantilaki2010b). The reader should note that the notion of shift might recall Gombrich's notion of shift. However, his idea is about visual representation, while here shifting concerns conscious vision. (This is not possible, or at most only momentarily, with trompe l'oeil because we have no visual representation, not even an unconscious one, of the surface of the trompe l'oeil picture.

Explaining pictorial unconscious representations is crucial because ‘philosophers are trying to explain the experience of seeing pictures whereas psychologists are trying to explain the (conscious or unconscious) mechanism of picture perception. In fact, unconscious picture perception is oddly missing from philosophical discussions of picture perception’ (Nanay Reference Nanay2015: 183).

Of course, another difference between perceiving face-to-face and perceiving depicted objects is that, in the latter case we can change spatial position in relation to the picture, but this spatial shift does not reflect any change in our perception of the depicted object (Nanay Reference Nanay2011: 471; Reference Nanay2017: 5.1; Matthen Reference Matthen2005: 315–317; Hopkins Reference Hopkins2012). Following the title of the paper by Vishwanath and colleagues (Reference Vishwanath, Girshick and Banks2005) ‘pictures look right when viewed from the wrong place’. This suggests, as mentioned above, that we represent (even unconsciously) the surface. Indeed, there is no sensorimotor understanding (Noë Reference Noë2004) with depicted objects because there is no movement for which the observer's view of the depiction changes as she/he moves (Nanay Reference Nanay2010a). I will not analyze this evidence because neuroscientists usually use pictures in static experimental settings for discrimination tasks, in which neither the subjects nor the objects presented on the monitor can move (Ferretti Reference Ferretti2016a). I do not deny that this is a crucial feature of perceptual presence in face-to-face seeing, especially concerning perceptual learning of sensorimotor contingencies (Noë Reference Noë2004). However, after learning has properly developed, presence can be perceived even in static cases of face-to-face perception. And the feeling of presence remains a crucial perceptual feature that is missing in static experimental settings with depicted objects but, as we will see, with no crucial problems for several kinds of perceptual attributions.

My account is not in conflict with those who argue that picture perception and face-to-face perception are ‘experiences of the same psychological kind’ (Briscoe Reference Briscoe, Hill and McLaughlin2016) because even they recognize the impossibility of assigning absolute egocentric depth and presence to pictorial objects. It has also been suggested that picture perception is not special because it resembles face-to-face perception: ‘a picture depicts an object by generating in those who view the picture a visual experience of that object’ (Chasid Reference Chasid2014: 471). This view follows Gombrich's stance opposing Wollheim's idea of simultaneous experience of both the surface and the depiction. However, first, I am not committed to this notion of twofoldness: the surface can be unconsciously represented. Second, these authors do not deny that depicted objects cannot solicit the feeling of presence.

Summing up, unlike normal objects, depicted objects are not represented as present. This is because the representation of the surface prevents us from representing the depicted object as present. However, without at least an unconscious representation of the surface, even depicted objects would always look as present as normal objects do. But this would imply that there is no possibility of experiencing usual pictorial perception, but only naturalistic, trompe l'oeil-like perception.

This makes pictures special objects of perception and supports the PT. We now have to examine the arguments in favor of the Sharpened Weak NT.

3. The Sharpened Weak Neuroscientific Thesis

Here I defend the idea that neuroscientists are justified in their everyday practice with pictures. Thus, it is possible to endorse the Sharpened Weak NT. This idea is supported by empirical evidence that depicted objects do, indeed, induce a very similar behavior of our visual system to that induced in face-to-face perception by normal objects (Nanay Reference Nanay2011, Reference Nanay2015; Ferretti Reference Ferretti2016a, Reference Ferretti2016c; Reference Ferretti2017a). This is possible without the need for depicted objects to be visually represented as present.

But what exactly does it mean that depicted objects induce a similar behavior of our visual system with respect to the one induced by normal objects? Or, conversely, what does it mean that face-to-face perception and the perception of depicted objects set us in a very similar perceptual state?

We perceive a lot of things in the visual scene. Following the usual jargon in the philosophical literature, when we see, for example, a shape, we are in a perceptual state that is responsible for such a shape perception. This allows us to individuate a visual state on the basis of what it represents (colors, shapes, texture, etc.). Usually, representation is described in terms of property attribution: a visual state represents a certain shape in the sense that through this representation we attribute a given shape property (e.g., the property of being cubic) to a given object. We can also take into account the results from vision science and say that a given visual state depends on the activity of a specific portion of the visual cortex. I am endorsing here a common view about perception, which can be used also for picture perception and according to which perception is described in terms of representations that attribute properties to normal objects (Siegel Reference Siegel2007, Reference Siegel2010; Nanay Reference Nanay2013) and depicted objects (Nanay Reference Nanay2011; Ferretti Reference Ferretti2016c, Reference Ferretti2017a).

That said, there is compelling evidence that our visual system sets us in a very similar perceptual state during both face-to-face perception and the perception of depicted objects. In other words, it attributes very similar properties (color properties, shape properties, etc.) to both normal and depicted objects. The Sharpened Weak NT perfectly takes into account this perceptual fact. Why very similar and not the very same perceptual state? Or, to put it in other words, why very similar but not the very same property attribution?

On the one hand, overall, the perception of depicted objects and face-to-face perception are very similar: we attribute several similar properties to both normal and depicted objects. But as we saw in section 2, there is at least one property that our visual system cannot attribute to depicted objects but only to normal objects: the property of being present (Ferretti Reference Ferretti2016c; Matthen Reference Matthen2005; Nanay Reference Nanay2015).

This distinction underlines the number of properties we attribute in these two perceptual situations. There is a set of properties we equally visually attribute to both normal and depicted objects. And there are properties visually attributed only to normal objects. However, the first set is much richer than the second (Briscoe Reference Briscoe, Hill and McLaughlin2016; Ferretti Reference Ferretti2016a, Reference Ferretti2016c, Reference Ferretti2017a; see also below). And the most evident and crucial different property within this second set is the property of being presentFootnote ² —this does not conflict with the idea of pictorial presence that ‘one visually perceives something present in the realm of the picture when experiencing a picture as a depiction’ (Aasen Reference Aasen2015: 188). The possibility of this kind of property attribution distinguishes these two perceptual processes. This is not a problem for the Sharpened Weak NT, which perfectly takes into account this perceptual fact.

Furthermore, we can address a perceptual hierarchy with respect to both visual representations and property attributions in relation to the distinction between face-to-face perception, picture perception, and the perception of depicted objects. First, only picture perception entails the Double Representation Thesis. Second, face-to-face perception and the perception of the depicted objects are, overall, very similar, but not precisely the very same perceptual process; though they attribute very similar properties to their respective objects, only the former but not the latter can attribute the property of being present to its proper perceptual objects. However, and this is an important point, concerning those properties we can attribute in both face-to-face perception and the perception of depicted objects—apart from the feeling of presence—the property attribution is, from a perceptual point of view, the same. This means that, for example, we can equally visually attribute the property of being red to both a depicted and a normal object; this holds for all the properties we can visually attribute to both normal and depicted objects I will list below (I cannot develop this view in full detail here, but see Nanay Reference Nanay2011; Briscoe Reference Briscoe, Hill and McLaughlin2016; Ferretti Reference Ferretti2016a, 2107; Hecht et al. Reference Hecht, Schwartz and Atherton2003). Crucially, this is possible even if depicted objects do not look present.

Summing up, this explanation perfectly captures the idea that we can attribute the same properties to both normal and depicted objects, except for the property of being present. Now we can examine, step by step, the evidence about the similarity between face-to-face perception and the perception of depicted objects, and this evidence shows that we can visually attribute most of the properties we usually attribute to normal objects also to depicted objects and that we can do so even without attributing the property of being present to depicted objects.

First of all, neuroscience teaches us that our visual system is divided, from an anatomo-functional point of view, into two main visual pathways (Milner and Goodale [1995] Reference Milner and Goodale2006; Jacob and Jeannerod Reference Jacob and Jeannerod2003): one ventral stream for conscious visual recognition in an allocentric frame of reference and one dorsal stream for visually guided action in an egocentric frame of reference. Both streams respond to depicted objects in the same manner as they respond to normal objects. Therefore, most of the perceptual properties usually attributed by each stream to normal objects can also be attributed to depicted objects (Nanay Reference Nanay2011, Reference Nanay2015; Ferretti Reference Ferretti2016a, Reference Ferretti2016b, Reference Ferretti2017a). Here is the list of properties we can visually represent depicted objects as having:

(a) (Ventral) Recognition of Visual Properties: When we look at depicted objects, we recognize what they depict. The ventral stream serves conscious or unconscious visual recognition. Thus, the ventral stream is activated during depicted object recognition (Nanay Reference Nanay2011: 467). First, by being activated even during the perception of the depicted object, the ventral stream attributes color properties, space properties, texture, shape, and size properties to the depicted object in the same way it does with normal objects (Grill-Spector and Weiner Reference Grill-Spector and Weiner2014). Second, cases of cortical damage show that subjects with an impaired ventral stream cannot see things in a picture (Turnbull et al. Reference Turnbull, Driver and McCarthy2004; Westwood et al. Reference Westwood, Danckert, Servos and Goodale2002; Nanay Reference Nanay2011: 468). For example, in the case of achromatopsia (the inability to see colors) and prosopagnosia (face blindness) due to specific damage to specific portions of the ventral stream, subjects cannot appreciate depicted colors and faces (Jacob and Jeannerod Reference Jacob and Jeannerod2003: 9, see also pp. 138, 173; for a full account see Nanay Reference Nanay2011).

(b) (Dorsal) Ascription of Action Properties: The dorsal stream also represents depicted objects. There are specific dorsal portions that are responsible, during face-to-face perception, for the visuomotor transformation of an object's geometrical attributes into action properties (anterior intraparietal area) and, then, of action properties into motor commands (the most rostral part of the ventral premotor cortex, named F5; Castiello Reference Castiello2005; Sakata et al. Reference Sakata, Taira, Murata and Mine1995; Murata et al. Reference Murata, Gallese, Luppino, Kaseda and Sakata2000; Raos et al. Reference Raos, Umilta, Murata, Fogassi and and2006; Ferretti Reference Ferretti2016a, Reference Ferretti2016b, Reference Ferretti2016c, Reference Ferretti2017b). The visuomotor transformation is accompanied by a neurophysiological motor simulation of the motor act related to the action properties (Ferretti Reference Ferretti2016a: sect. 4, Reference Ferretti2017b). This is a form of covert mental action and is a mechanism of motor preparation for potential action that is simultaneously blocked and can be recruited if the person decides to act. Note that dorsal computations are triggered without the subject wanting or attempting to grasp the object: observation in static conditions directly triggers these dorsal visuomotor responses (Ferretti Reference Ferretti2016a: sect. 4). Now, dorsal processing responds not only to real objects but also to depicted objects by attributing to them action properties and representing the related motor commands. Indeed, dorsal processing is triggered regardless of whether an object is real or depicted (Chao and Martin Reference Chao and Martin2000; Proverbio et al. Reference Proverbio, Adorni and D'Aniello2011; Buccino et al. Reference Buccino, Sato, Cattaneo, Rodà and Riggio2009; Costantini et al. Reference Costantini, Ambrosini, Tieri, Sinigaglia and Committeri2010; Grézes and Decety Reference Grézes and Decety2002; Zipoli Caiani Reference Zipoli Caiani2013; Romero et al. Reference Romero, Pani and Janssen2014; Pani et al. Reference Pani, Theys, Romero and Janssen2014; Sakata et al. Reference Sakata, Tsutsui and Taira2003). The evidence for this perceptual fact is well established in both the neuroscientific (ibid.) and the philosophical literature (Zipoli Caiani Reference Zipoli Caiani2013; Ferretti Reference Ferretti2016a, Reference Ferretti2016b, Reference Ferretti2016c, Reference Ferretti2017a) and is compatible (Ferretti Reference Ferretti2016a: 3811–12; Reference Ferretti2017a: sect. 3.1) with other investigations of the behavior of the visual streams in picture perception (Nanay Reference Nanay2011, Reference Nanay2015).

I am not saying that the motor behavior of the subject in general is the same with both real and depicted objects: we do not attempt to grasp objects in a picture. I am saying that the representational behavior of the dorsal stream is the same with both of them. Dorsal visuomotor responses are very specific: the same triggered motor response that encodes a specific motor act (e.g., a power grip) in the case of a specific real object (a real handle) is activated in the case of a specific depiction of the very same object (e.g., a depicted handle). The difference in relation to the perception of a depiction and of a real object, is that the result of dorsal computations, related to visuomotor transformation and simulated covert action, remains stored in the motor memory and cannot be used for overt action with depictions. It can be exercised only with real objects when we decide to act (Ferretti Reference Ferretti2016a; Zipoli Caiani Reference Zipoli Caiani2013). Moreover, both the visuomotor transformation and the simulation are automatic unconscious subpersonal processes that are not consciously accessed by the subject (Ferretti Reference Ferretti2016a, Reference Ferretti2017b).

Thus, saying that our dorsal stream can represent depicted objects as offering action possibilities and, thus, that the representational behavior of dorsal processing in particular, not the motor behavior of the subject in general, is the same with respect to both normal and depicted objects does not mean, of course, that we have the conscious impression that we can act upon those possibilities. It only means that our dorsal brain behaves at the level of automatic subpersonal visuomotor processing that is not consciously accessible as if the action properties pertaining to the geometrical configuration of the depicted object pertained to a real 3-D object (Ferretti Reference Ferretti2016b: 3804). And this is possible because, as I pointed out, dorsal processing is active regardless of whether an object is real or depicted. At the conscious level, ventral processing informs us that we are in front of a depiction, not in front of a present object. And this is because, at the unconscious level, ventral processing tracks the presence of the surface.

This perceptual fact is possible due to the different and peculiar computational characteristics of each stream. Ventral vision is the one involved in the visual selection of targets for reliable action, that is, it is involved in ‘action planning’. This is because it is the one involved in ‘response selection’, a process of high-level cognition that can be conscious or unconscious and that allows us to detect which objects are real and present (that is, nonpictorial), and thus offer reliable interaction and which are not (Westwood et al. Reference Westwood, Danckert, Servos and Goodale2002; Ferretti Reference Ferretti2016a, Reference Ferretti2016c). Only ventral vision can perform both pictorial recognition and recognition of presence. As I said, dorsal vision, alone, cannot distinguish whether an object is real or pictorial, and on the basis of the geometrical properties detected, it responds to the depicted objects that are apparently located in the peripersonal space of the observer and, thus, can be represented as apparently reachable (for a complete account see Ferretti Reference Ferretti2016a, Reference Ferretti2016c).Footnote ³ For this reason and because of the fact that due to the magnocellular advantage dorsal computations for motor programming are triggered before ventral processing for recognition occurs (Milner and Goodale [1995] Reference Milner and Goodale2006; Barrett and Bar Reference Barrett and Bar2009), dorsal visuomotor attribution of action properties and motor simulation occur before ventral response selection and action planning track presence and remain stored in our motor memory for any possible recruitment related to overt action.

Accordingly, only when ventral processing performs response selection and recognizes the object as being present, does it select that object for action and triggers action planning toward it. And only at this point can the thin motor parameters computed by the dorsal stream and stored in our visuomotor quiver be used for overt visuomotor interaction (Milner and Goodale [1995] Reference Milner and Goodale2006; Zipoli Caiani and Ferretti Reference Zipoli Caiani and Ferretti2017; Briscoe and Schwenkler Reference Briscoe and Schwenkler2015; Briscoe Reference Briscoe2009; Goodale and Milner Reference Goodale and Milner2004; Chinellato and del Pobil Reference Chinellato and del Pobil2016; Ferretti Reference Ferretti2016b, Reference Ferretti2016c).

On this matter, the literature distinguishes between ‘ventral action planning’, which establishes whether the object we face is a real one or merely a depiction and, thus, whether it can be selected for action (and the general way to act with respect to our intentions), and ‘dorsal motor processing’, which computes the specific, thin motor parameters to be used for overt interaction (Dijkerman et al. Reference Dijkerman, McIntosh, Schindler, Nijboer and Milner2009; Milner and Goodale [1995] Reference Milner and Goodale2006: 244; Briscoe and Schwenkler Reference Briscoe and Schwenkler2015: 1437), triggered independently of whether the object is real (Ferretti Reference Ferretti2016a).

The interplay between these computational characteristics of the streamsFootnote ⁴ leads our visuomotor system to respond also to depicted objects without the risk that we attempt to grasp them. Thanks to its computational characteristics, ventral response selection is attuned both to the surface and the depiction (compare (a) above and see below). It recognizes the depicted object as such, namely, as a nonpresent pictorial object. This representation is usually conscious in picture perception. But it also recognizes the surface as a real and present object with which we can interact. This representation is usually unconscious in picture perception (Nanay Reference Nanay2011; Ferretti Reference Ferretti2016c). And following what I said above, only in the case of the recognition of presence, given by ventral processing, which in picture perception is related to the surface, is it possible to make use of the motor parameters computed by dorsal processing (Ferretti Reference Ferretti2016c). An exception is made for pictorial illusions (e.g., trompe l'oeil pictures, see section 2 and footnote 1) able to deceive our ventral visual recognition that there is no surface. In this case, what effectively is a depiction is perceived by ventral processing as a present object toward which action planning can be triggered and to which motor execution, whose parameters are computed by dorsal processing, can be subsequently directed. All this suggests that, in order not to be fooled, at the conscious level, that the depicted object is present, as in the case of trompe l'oeil, we need at least an unconscious representation attuned to the surface of a particular kind: one served by ventral processing and involved in visual recognition of presence, response selection, and action planning (cf. section 2, footnote 1). As explained above, when we can rely on such a representation following the evidence reported in this section, ventral vision allows us to consciously represent the depiction as such, that is, as a pictorial nonpresent object. This is in line with the idea that unconscious recognition of the surface modulates the conscious representation of the depiction (Nanay Reference Nanay2017: sect. 2) and is perfectly compatible with the weaker notion of simultaneity in picture perception endorsed here following the Double Representation Thesis (section 1).

All this suggests that the main differences between real and depicted objects are tracked by ventral recognition (Ferretti Reference Ferretti2016c). This point is compatible with the claim that both dorsal and ventral vision represent both the depiction and the surface (Nanay Reference Nanay2011; Ferretti Reference Ferretti2016a, Reference Ferretti2016c, Reference Ferretti2017a, see also point (a) above), but only ventral vision represents them, respectively, as a pictorial object and as a real and present one; dorsal vision cannot distinguish between these two different kinds of perceptual objects. But this does not prevent dorsal vision from representing the different motor acts respectively related to the different geometrical arrangements of the depiction (e.g., a depicted handle recalls a power grip) and of the surface (which usually recalls a precision grip) (see Ferretti Reference Ferretti2016a, Reference Ferretti2016c, Reference Ferretti2017a for this point and for the explanation, in line with Nanay's account, of how ventral and dorsal vision are respectively attuned to both the surface and the depiction).

(c) Ascription of Spatial Characteristics: We can also represent several spatial features in the pictorial space that resemble those represented during face-to-face perception. We can perceive ‘metric relative depth relations within and among objects, including 3-D surface shape, slant, depth order, and so on, all of which imply an understanding of the ratios of distances or depths among points in the visual field’ (Vishwanath Reference Vishwanath, Albertazzi, van Tonder and Vishwanath2011: 222). Accordingly, we can perceive orientation, apparent 3-D configuration, layout and volumes (Gonzalez et al. Reference Gonzalez, Ganel, Whitwell, Morrissey and Goodale2008; Snow et al. Reference Snow, Pettypiece, MacAdam, McLean, Stroman, Goodale and Culham2011; Berryhill and Olson Reference Berryhill and Olson2009; Marotta and Goodale Reference Marotta and Goodale2001; Cutting Reference Cutting, Hecht, Schwartz and Atherton2003; Koenderink and van Doorn Reference Koenderink, van Doorn, Hecht, Schwartz and Atherton2003; Koenderink Reference Koenderink2012; Nelissen et al. Reference Nelissen, Joly, Durand, Todd, Vanduffel and Orban2009; James et al. Reference James, Humphrey, Gati, Menon and Goodale2002; Taira et al. Reference Taira, Nose, Inoue and Tsutsui2001). In addition, ventral vision allows us to quasi-egocentrically perceptually represent depicted objects (Briscoe Reference Briscoe2009: 447, sect. 6.1–6.4; Grush Reference Grush2000; Pani et al. Reference Pani, Theys, Romero and Janssen2014; Bruzzo et al. Reference Bruzzo, Borghi and Ghirlanda2008; Vishwanath Reference Vishwanath2014; Cutting Reference Cutting, Hecht, Schwartz and Atherton2003; Hecht et al. Reference Hecht, Schwartz and Atherton2003): the subject ‘non-reflectively assumes the perspective of the camera’ (Briscoe Reference Briscoe2009: 447). This is in line with the possibility of (apparent) localization of depicted objects in our peripersonal space by the dorsal stream (Ferretti Reference Ferretti2016a) and with the fact that absolute egocentric localization is not possible with depictions (see section 2).

(d) Ascription of Emotional Properties: The two streams are connected to two different portions of the orbitofrontal cortex, which is involved in the emotional encoding of the visual scene: the dorsal stream is connected to the medial orbitofrontal cortex, while the ventral stream is connected to the lateral orbitofrontal cortex. The former pathway encodes emotional properties salient for visually guided action; the latter encodes emotional properties salient for visual recognition (Barrett and Bar Reference Barrett and Bar2009; O'Reilly Reference O'Reilly2010; Elliott et al. Reference Elliott, Dolan and Frith2000). There is compelling evidence that both pathways respond to depicted objects by attributing salient emotional properties concerning both recognition and (apparent) action (Ferretti Reference Ferretti2017a). The lateral orbitofrontal cortex and its related cortical projections related to the ventral pathway respond to pictorial contents with high emotional value (Wright et al. Reference Wright, He, Shapira, Goodman and Liu2004; Schienle et al. Reference Schienle, Schäfer, Hermann, Walter, Stark and Vaitl2006; Phillips et al. Reference Phillips, Marks, Senior, Lythgoe, O'Dwyer and Meehan2000; Shapira et al. Reference Shapira, Liu, He, Bradley, Lessig, James, Stein, Lang and Goodman2003). Concerning the orbitofrontal/dorsal link, depictions that represent graspable objects that would be, in principle, approachable without any risk, activate a facilitating motor response, while depicted dangerous objects evoke a motor resonance concerning aversive action possibilities that recalls an interference-effect between the information about the object's potential risks and the action possibilities detected (Anelli et al. Reference Anelli, Borghi and Nicoletti2012; for a complete review see Ferretti Reference Ferretti2017a: 5.1).

These results suggest that depicted objects do, indeed, induce an equal behavior of our visual system as face-to-face perception does. We can equally attribute several properties to both of these perceptual situations. Thus, this evidence supports the stance used by neuroscientists, who are justified in their everyday practice with pictures: a depicted object—which is just one of the two components of a picture—and a normal object are, in this sense, very similar objects of perception.

The equal attribution of these specific properties is possible even if, unlike normal objects, depicted objects cannot foster any visual feeling of presence (section 2). The representation of the difference concerning presence is mainly due to the computational behavior of the ventral stream, which detects, at least at the unconscious level of visual processing, the presence of the surface (see point (b) above). This is in line with the fact that we usually focus on the depiction, while unconsciously representing the surface.

But the fact that depicted objects induce an equal behavior of our visual system with respect to the properties mentioned above, even if they do not look present, is sufficient in order to rely on experimental settings that make use of depicted objects rather than normal objects. Indeed, if our visual system can attribute, say, a color property, a shape property, or the like to both a normal and a depicted object, the neuroscientific practice of using depicted objects when aiming at discovering the neural correlates of vision is reliable. This supports the Sharpened Weak NT.