4 Perceptual presence

4.1 Ambiguities in Seth’s description of the explanandum

The tomato is not only experienced as perceptually present, it is also perceived as an object in the external world. In a commentary on Seth, Tom Froese (2014, p. 126) has therefore suggested that Seth conflates perceptual presence with experienced objecthood. This proposal has some plausibility, because the tomato is perceived as a real object, whereas afterimages are not experienced as objects (they are more like unstable colored shades). After all, even Seth admits, in his target paper, that it may be important to distinguish presence from objecthood (p. 18). This is one way in which Seth’s definition of the explanatory target is ambiguous: is it about experienced presence or experienced objecthood (cf. also Seth 2014, pp. 105f.)? (This question becomes more pressing still when we consider the ethymology of “realness” or “reality”: the Latin origin of the word is res (thing), which makes it a little confusing that Seth seems to identify perceptual presence with the sense of subjective reality, cf. Seth this collection, p. 2.)

Another ambiguity is related to the notion of a counterfactual model. In his target paper Seth defines a counterfactual model as a model encoding “how sensory inputs (and their expected precisions) would change on the basis of a repertoire of possible actions” (Seth this collection p. 17). On the one hand, one may ask if counterfactual models in the brain necessarily encode SMCs (sensorimotor contingencies). For the perception of a ripe tomato on a bush, it might be equally relevant to encode how sensory signals pertaining to the tomato would change if the wind were to blow the bush or if the tomato were to fall down. On the other hand, it is unclear how explicit a counterfactual representation has to be. Jakob Hohwy (2014) suggests that a rich causal structure could be modeled by extracting higher-order invariants (features that do not change if the tomato is dangling in the wind or has fallen down, for instance). Higher-order invariants are relatively perspective-independent.[14] The degree of perceptual presence would then correspond to the “depth of the inverted model”[15] (Hohwy 2014, p. 128). In his target paper, Seth notes that the depth of the model may indeed play a role (see footnote 13).

Two ambiguities are thus to be found in Seth’s account. One concerns the characterization of the target phenomenon (experienced realness versus experienced objecthood). The other lies in the description of the represented causal structure: counterfactual richness versus perspective-independence of hidden causes. Counterfactual richness and causal “depth” are not completely independent. Below, I will give some examples that may be useful to explore the relationship between these two features. Furthermore, I will suggest that it could be helpful to consider another feature with respect to which the represented causal structure of objects may vary. This feature is the degree of causal encapsulation. For representations not only differ with respect to their counterfactual richness or their degree of perspective-dependence, but also with respect to the extent to which the represented causal structure is encapsulated or integrated. (In what follows, I will use the notion of a counterfactual model mainly in the sense in which Seth uses it: counterfactual models in this sense involve representations of possible bodily actions by the subject of experience.)

A phenomenal representation of a tomato on a plate is not only counterfactually rich and relatively perspective-dependent, the represented causal structure is also causally integrated.[16] It is, for instance, represented as being causally related to the plate, because it is experienced as lying on the plate (that is, it is not hovering above it). Furthermore, it is in possible causal contact with virtually all other objects in its vicinity (e.g., the subject’s hands).

Contrast this with the experience of what is happening in a classical video game—say, a racing game. The player influences how the images on the two-dimensional screen change, because she has control over the vehicle. Hence, we can assume that representations of gaming sequences are (usually) counterfactually rich. At the same time, they are also perspective dependent (although they mainly depend on the virtual perspective from which objects are represented in the game). However, virtual objects in the game are experienced as causally encapsulated: although objects can interact with each other in the virtual world, they do not interact with most other parts of the player’s environment. For instance, they will never break out of the screen and fly around in the room in which the player is sitting. Furthermore, they can only be influenced vicariously through a controller or keyboard. Thus there is not causal encapsulation in every respect (the virtual world is not experienced as completely disambiguated from the rest of the experienced world), but in some respects the encapsulation is rather strong (the virtual world is spatially bounded, e.g., with the screen as the limit). Note that many modern video games are less causally encapsulated, for instance when they are played on a touchscreen (or on devices with a three-dimensional screen, or in an immersive virtual reality).[17]

As mentioned above, causal integration and counterfactual richness are not completely independent. High counterfactual richness implies a certain degree of causal integration (at least in some respects), for it means that at least the subject can interact with the experienced object in some way—regardless of how separate the represented causal structure is from the rest of the subject’s surroundings.

Similarly, highly perspective-invariant representations typically also involve the representation of an encapsulated causal structure. Abstract conscious thoughts, for instance, cannot be touched with the hand or other concrete objects. However, the implied encapsulation only holds in some respects. Sometimes thoughts can evoke strong emotions or a sequence of mental imagery. In certain obsessive-compulsive disorders, for instance, subjects will first have a thought (“My hands are dirty”), presumably followed by a feeling of disgust and the urge to wash the hands, which then leads to motor behavior (washing the hands); this, in turn, may be followed by the thought that the hands are still dirty. The content of the conscious thought is relatively perspective-invariant, and yet it involves, presumably, representations of causal structure that link it to concrete objects in the world.

As long as we interpret counterfactuals only as representations of sensorimotor contingencies, it may also seem that perspective-invariant[18] representations are counterfactually poor. However, if we include representations of possible mental actions and their effects, we can also conceive of counterfactually-rich perspective-invariant representations. A possible example is a philosophical argument or a theory, which someone can contemplate in their mind, being aware that there are several possible ways in which the argument could be probed and attacked, or several important cases to which the theory could be applied.

Bearing in mind that the degree of causal encapsulation is not completely independent from the other two dimensions (counterfactual richness and perspective-invariance), we can depict different types of conscious experiences in a cube, where the three axes stand for the three dimensions described (see Figure 1). The most interesting locations in this cube are, of course, its eight corners, because they depict classes of experiences for which each of the three features is either completely absent or maximally pronounced. Finding examples of these “extremal experiences” is no easy task.[19] Even neural representations of synesthetic concurrents, Seth’s prime example of counterfactually poor models, may, at first sight, seem to be located somewhere in the middle of the perspective-dependence axis.

Figure 1: The figure illustrates how classes of experiences can be located in a cube, according to the extent to which they display counterfactual richness, perspective-independence, and causal integration (see main text for explanations). The cube (without the labels) is adapted from cube figures in Godfrey-Smith (2009); talks by Daniel Dennett brought this style of illustration to my attention.

Grapheme-color concurrents, for instance, are not simply triggered by graphic representations of glyphes, but by representations of abstract objects, i.e., graphemes, associated with certain glyphes (cf. Mroczko et al. 2009). Hence, it may seem that the hidden cause of the concurrent is not simply an object in the world, but also involves an abstract object, i.e., a grapheme, the representation of which is perspective-invariant. This would suggest that synesthetic concurrents cannot conclusively be placed in one of the cube’s corners, because their represented hidden causes involve very high-level invariants.

On the other hand, one could object that the concurrent itself is represented in a rather perspective-dependent way. It may be part of a causal network involving hidden causes that are represented in perspective-invariant ways, but the synesthetic percept itself is not a representation of an abstract hidden cause.[20] Hence, on second thought, it seems that concurrents, as in grapheme-color synesthesia, are in fact located close to the origin of our coordinate system: the representations involved are relatively perspective dependent, and they are counterfactually poor. At the same time, they are causally encapsulated, because they do not interact with physical objects (they cannot be touched, etc.).