Erkenn (2013) 78:1259–1273 DOI 10.1007/s10670-012-9409-0 ORIGINAL ARTICLE

Is There a Space of Sensory Modalities? Richard Gray

Received: 14 March 2012 / Accepted: 4 November 2012 / Published online: 13 November 2012 Ó Springer Science+Business Media Dordrecht 2012

Abstract Two proposals have recently, and independently, been made about a space of possible sensory modalities. In this paper I examine these different proposals, and offer one of my own. I suggest that there are several spaces associated with distinct kinds of sensory modality.

1 Introduction We use the term ‘(a) space’ in a wide variety of contexts. We talk of logical space, a space of possibilities, mathematical spaces and other kinds of spaces. When we do so, what are we talking about? The first answer to come to mind is probably: something that is, in certain respects, like (a) physical space. After all, (a) physical space is what we mean when we use ‘(a) space’ without further qualification. We think of (physical) space as having three dimensions; those of length, width and height. We also think of (physical) space as something within which things are (1) differentiated from each other and (2) ordered with respect to each other in relation to its dimensions; item a is next to item b in the length dimension, item b is closer to item c than item d in the width dimension, and so on. When we extend the use of ‘(a) space’ to talk of some other kind of space, it is reasonable to assume that in doing so we keep in mind many of the defining features of (physical) space. Why call something ‘a space’ otherwise? One extension of the use of ‘a space’ is to refer to quality spaces. A quality space is constituted by a set of qualitative properties that are ordered along n dimensions according to their relative qualitative similarities. The number of dimensions of a quality space is determined by the number of differentiative/ordering features

R. Gray (&) Cardiff University, Humanities Building, Colum Drive, Cardiff CF10 3EU, UK e-mail: [email protected]

123

1260

R. Gray

possessed by the qualities constitutive of the quality space.1 It makes sense to talk of a quality space as a kind of space because it is something the constituents of which, namely qualities, can be (1) differentiated and (2) ordered along multiple dimensions. Consider colour and the colour quality space. It is constructed by arranging along three dimensions all the colours to which we are normally sensitive according to their relative similarities. Three dimensions are necessary and sufficient to construct colour quality space because colours can be distinguished from each other by means of three differentiative/ordering features, which are the only respects in which the colours can be differentiated and ordered: hue, saturation and brightness.2 The geometric model that represents the colour quality space approximates a threedimensional solid oval. The explanation for its shape lies partly in the way that our ability to discriminate hues and saturation is dependent on brightness (as brightness is increased or decreased from an optimum so fewer colours can be discriminated), and partly in the way that the processing mechanisms underlying the perception of the hues function.3 It should be noted, since it will be of relevance later, that one colour is, nevertheless, differentiated from and ordered with respect to another colour by hue, saturation or brightness independently of its differentiation from and ordering with respect to another colour by another feature. In that this is so, the differentiative/ordering features function independently of each other. Other quality spaces, which are also mapped by reference to qualitative similarity, have different dimensions. The taste quality space, for instance, is geometrically modelled by a tetrahedron with its four vertices representing the four primary tastes (sweet, salty, sour and bitter).4 It is constituted by six dimensions: sweet–salty, sweet–sour, sweet–bitter, salty–sour, salty–bitter, and sour–bitter. All tastes are modelled by their positions on one of these dimensions or in relation to three of the vertices. Hence only the surface of the taste quality space is occupied. It seems reasonable to think of physical space and quality space as different types of space because they share certain features. Whether they are both types of spaces because they have such features in common or whether our thinking of quality space in this way is more a matter of metaphor is harder to determine. Do logical space, a space of possibilities and mathematical spaces all satisfy necessary and sufficient conditions for being a type of space? Do they share a family resemblance? Is the meaning of ‘space’ here metaphorical? Are different explanations applicable to different so-called spaces? Whatever the answers to these questions, there might still be room for disagreement about how exactly to construe a specific type of space, as will become apparent in the following. 1

For further details and discussion of how to determine the structure of a quality space, see Clark (1993, pp. 76–116).

2

Colours can be differentiated into unitary and binary colours, but this does not provide the basis for an ordering.

3

For Clark the main reason behind mapping quality spaces is to provide data about sensory qualities that can then be explained by reference to the neurobiological features of sensory systems.

4

This is sometimes known as the Henning taste tetrahedron. Some models include a fifth primary: umami. I omit this complication here.

123

Sensory Modalities

1261

Regarding quality spaces, it is natural to think that there is some relation between the different quality spaces and the different sensory modalities.5 Indeed, it seems natural to think that the relation is strong. For it seems plausible that two token senses should be thought of as members of different types of sense only if they realize experiences having qualitative features that belong to different quality spaces, and, conversely, should be thought of as members of the same type of sense if they realize experiences having qualitative features that belong to the same quality spaces. Hence what is also determined by a quality space is a condition on being a member of a type of sensory modality.6 Recently, it has been claimed that the sensory modalities can be thought of in terms of a space of their own. In this paper, I explore two different construals of a space of the senses, and offer a proposal of my own. In Sect. 2, I address Brian Keeley’s suggestion of a space of possible modalities that is determined by the different physical energy types that it is possible to sense. In Sect. 3, I turn to Fiona Macpherson’s recent proposal of a multidimensional space in which each sense has a particular location. In Sects. 4–7, I examine four candidate dimensions for such a space of the senses. I conclude, in Sect. 8, by suggesting that there is no space of sensory modalities but that it is plausible to think there are several spaces associated with distinct kinds of sensory modality.

2 Keeley’s Space of Sensory Modalities The idea of a space of possible sensory modalities is first suggested by Brian Keeley. Keeley’s discussion of the idea is only brief. Nevertheless, it plays a significant role in his response to the problem of the individuation of the senses. The individuation of the senses raises a problem because it is insufficiently clear what general principles are used, or should be used, to determine to which particular type of sense any token sense belongs.7 Keeley proposes that for a creature to possess sensory modalities is for it ‘‘to possess appropriately wired-up sensory organs that are historically dedicated to facilitating behavior with respect to an identifiable physical class of energy’’ (Keeley 2002, p. 6).8 He then extracts from this proposal four conditions that, so he argues, can be used to differentiate the distinct types of senses and, thereby, to identify any token sense. The first condition, the physics condition, expands upon the claim that for a creature to possess a sense it has to be sensitive to ‘‘an identifiable physical class of energy’’; part of what makes a token sense a member of a particular type of sense is the type of physical stimulus to which it is sensitive. For instance, on this view, part of what makes a token sense a member of the type visual sense is that it detects 5

Ross (2008, pp. 303–305), following Clark, explicitly makes the connection.

6

Plausible as the view may be, as is noted in §6, there are reasons to doubt that distinct quality spaces and distinct types of senses can be co-related straightforwardly.

7 This assumes that any token sense can be identified, which, as Macpherson (2011b) has pointed out, is a substantive assumption. 8

Keeley (2002, p. 6).

123

1262

R. Gray

differences in electromagnetic stimuli, and part of what makes a token sense a member of the type tactile sense is that it detects differences in pressure stimuli. Having noted how sensory systems respond to different forms of physical stimuli in the environment, Keeley goes on to claim that this can be used as the basis for a non-arbitrary differentiation of the senses: In providing an ontology of possible forms of energy, what physics provides is the space of possible modalities. Whether any animal on earth makes use of a magnetic modality or not, the fact that physics identifies magnetism as a type of energy raises the possibility of a magnetic modality.9 By determining the space of possible modalities, the physics condition is intended to lay down a substantial constraint that is independent of all other constraints, in particular biological and psychological constraints. The physics condition, however, is not all that is required for a type of sensory modality actually to exist. There also have to be creatures that are capable of sensing the physical stimuli. Hence, according to Keeley, a second condition, the neurobiological condition, is required for the existence of a sense. This condition expands upon the claim that for a creature to possess a sense it has ‘‘to possess appropriately wired-up sensory organs’’. So what makes a sense an existing visual sense is the presence of an optical system, and what makes a sense an existing tactile sense is the presence of a tactile system. The two remaining conditions are introduced to rule out problematic cases of two related sorts. The behaviour and dedication conditions build on the claim that sensory systems are ‘‘historically dedicated to facilitating behaviour with respect to an identifiable physical class of energy’’. The behaviour condition is required to rule out vestigial senses; these are systems that possess many of the relevant neurobiological features of a sense but no longer function as sensory systems in so far as they no longer bring about responses to physical stimuli. The dedication condition is required to rule out spurious senses; these would be senses on the basis of their responsiveness to a range of physical stimuli but they are not genuine senses because they have not been selected to respond to these ranges of physical stimuli. It should be evident from this brief summary that the roles of the conditions are distinct: whereas the physics condition, which determines the space of possible senses, provides theidentity condition for the senses, the other three conditions provide existence conditions for the senses. The characterization of Keeley’s space of possible sense modalities as a space seems unobjectionable. After all, like other types of spaces, in this case there is something within which items of a certain kind are differentiated. The more substantive issue is whether this is the right way to think of a space of possible senses. One line of argument indicates that it is open to challenge. The line of argument, in brief, is this: the space of possible senses determines the senses there might be, and the further conditions determine which of the possible senses are actual; however, applied jointly these do not succeed in determining all the actual senses; therefore, either the space of possible senses is too limited, or the additional conditions need to be modified. 9

Keeley (2002, p. 13), emphasis in original.

123

Sensory Modalities

1263

I have offered this line of argument elsewhere using two problem cases: the pit viper problem and the vampire bat problem.10 Here I will flesh out the line of argument using the former. Given that electromagnetic radiation is a distinct form of physical energy, according to Keeley’s proposal for the determination of the space of possible senses by the set of physical energy types, there will be a corresponding possible sense. Next, if the three existence conditions—the neurobiological, behaviour, and dedication conditions—are met, there will be a corresponding actual sense. This entails that the pit viper, which is receptive to two distinct ranges of electromagnetic radiation by means of two sensory systems, will have two forms of the one sense: vision. However, most neuroethologists (and Keeley’s account was specifically intended to capture the neuroethologist’s practice) say that the pit viper has both a visual sense and a thermal imaging sense. Therefore, the space of possible senses in conjunction with the existence conditions is unable to determine all the existing types of sensory modality. To resolve the problem one could modify the use of the existence conditions. My earlier suggestion was that one could use the neurobiological condition, supported by the other conditions, as an identity condition rather than merely as an existence condition. But there is an alternative response. One could locate the origin of the problem in the way that the space of possible senses has been conceived. Had the space of possible senses included more senses, the problem may not have arisen.

3 Macpherson’s Space of Sensory Modalities Fiona Macpherson’s proposal for a space of possible senses involves a different construal of how such a space is constituted. She proposes a space of senses in order to explicate her rich view of the senses.11 According to Macpherson, a sparse view of the senses holds that (1) there are a limited number of senses, and (2) they are discrete, i.e. they are relatively distinct from each other. Her rich view of the senses denies both. The Aristotelian view of the senses, which is sometimes thought of as the commonsense view of the senses, is an example of the sparse view. It holds that there are (and could be) only five senses: sight, hearing, touch, smell and taste. Macpherson objects, claiming, firstly, that there is good reason to think that there are actually more senses than these five and, secondly, that there could be more senses than these five. In support of the former, she claims that proprioception, equilibrioception and the vomeronasal system already provide evidence that humans possess more than five senses, and bat echolocation, the bee’s capacity to sense in the ultra-violet, the thermal imaging sense of some snakes, the magnetic senses of birds and the electric senses of fish provide evidence of non-human senses. In support of the latter, she cites popular culture, e.g. X-ray vision and a sixth sense. Whether or not these are metaphysical possibilities, that they can be thought of as conceptual possibilities indicates that our commonsense conception of a sense is not committed to a sparse view. 10

Gray (2005).

11

Macpherson (2011a). See Macpherson (2011b) for further background.

123

1264

R. Gray

Exactly how many senses there are depends on how the senses are individuated. Suppose one starts with an easy case: what distinguishes vision from hearing? Common sense might suggest that whether a sense is a visual sense or an auditory sense depends on whether (a) it represents colours or sounds, (b) it realizes experiences having the character of colour experiences or sound experiences, (c) it detects electromagnetic waves or pressure waves, (d) it involves eyes or ears. Following Macpherson, let us call these (A) the representational criterion, (B) the phenomenalcharacter criterion, (C) the proximal stimulus criterion and (D) the sense-organ criterion.12 Much of the recent debate about the individuation of the senses has focused on which of these criteria should be used to individuate the senses.13 Macpherson claims that, in so far as all four criteria can be used equally well to distinguish the five Aristotelian senses, this has been taken to support both the sparse view of the senses and the criterial approach to the individuation of the senses. However, as she points out, problems arise for the criterial approach in the case of certain unusual senses and the unusual function of the standard senses. Macpherson focuses on four examples: bat echolocation, the ultraviolet vision of bees, thermal imaging and tactile-visual sensory substitution (TVSS), where devices that exploit the use of touch are used to replace vision. She argues that, if you take each of the four criteria in turn, none of them is able to tell us unambiguously which Aristotelian sense these senses are; in respect of all the criteria, they each differ in some significant way from the Aristotelian senses. Difficulties arise perhaps most vividly for the criterial approach in two sorts of cases: (1) where a putative sense possesses features of distinct Aristotelian senses according to one of the criteria, and (2) where a putative sense possesses features of distinct Aristotelian senses according to two of the criteria. TVSS is an example of (1). In TVSS light is transformed into a pressure stimulus. Hence TVSS possesses features of distinct Aristotelian senses according to the proximal stimulus criterion. Thus it is unclear whether TVSS should be construed as a visual sense or a tactile sense.14 Thermal imaging is an example of (2). In thermal imaging the heat of objects is detected not by contact but by means of electromagnetic radiation. Hence thermal imaging possesses features of distinct Aristotelian senses according to the representational criterion and the proximal stimulus criterion respectively; it counts as a tactile sense according to the former and a visual sense according to the latter. Thus it is unclear whether thermal imaging should be construed as a tactile sense or a visual sense.15 12

See Grice (1962) for the original suggestion.

13

This is not universal. Nudds (2003) defends a sparse view without endorsing the application of criteria for the senses. 14 The other criteria do not settle the matter because related clashes arise for them: the property represented appears to be both visual and tactile; the phenomenal character of the experience seems to be both like vision and touch; and both the skin and the artificial eye seem to be relevant sensory organs. 15 At least Ross (2008), who defends a common sense view, suggests these as the two possible senses. He defends a representational criterion by arguing that to which kind of sense thermal imaging belongs will depend on whether it represents hotness or brightness. It is, however, arguable whether this would be enough to override other distinctive features of the sense noted by the other criteria: if hotness is detected, because it is detected distally, it is unlike touch; if brightness is detected, because it is caused by heat, it is unlike vision.

123

Sensory Modalities

1265

Macpherson’s response to the difficulties that emerge in attempting to apply the four criteria to divide up the senses into a limited few is to reject the use of each of the criteria on its own and the sparse view. She takes the lesson from problem cases to be that ‘‘the differences between the senses amounts more to a difference of degree rather that a difference of kind’’.16 She therefore recommends using the four criteria together as dimensions, thereby disclosing the richness of the senses. She summarizes her proposal in the following way: I hold that the four criteria are relatively independent dimensions along which different possible kinds of senses could take different values. We can think of these four criteria as defining a multidimensional space within which we can locate each of the Aristotelian senses, the four examples of unusual senses discussed earlier, and any other sense. Thus, human vision, bee ‘‘vision,’’ snake infrared perception, and TVSS perception would each be located at a different place in the multidimensional space. Indeed, this multidimensional space is a way of delineating the space of all possible senses. All possible senses will occupy a place in the space. (The actual senses will occupy a small number of these places.)17 Unlike Keeley’s proposal, there is no reliance here on the proximal stimuli, supplied by his physics condition, to determine the space of possible senses. Additionally, Macpherson’s space of the senses has other characteristics in common with other types of spaces. In particular, it is constituted by items that are both differentiated and ordered along multiple dimensions by reference to those of the items’ features that can be used to differentiate and order them. So there is every reason to think that this should be thought of as a space of the senses. The substantive question is, again, whether this way of thinking of a space of possible senses will work. In the next four sections I raise some doubts about the capacity of Macpherson’s candidate dimensions to constitute a space of the senses.

4 The Proximal Stimulus Dimension Consider how the proximal stimulus criterion could be used to determine a dimension of a space of the senses in which each sense has a specific location. In order to do this, the proximal stimulus criterion would have to provide a feature that could be used both to differentiate the possible senses and to generate a nonarbitrary ordering of them. Indeed, this is a condition on the adequacy of any feature that might be used to contribute to the determination of such a space of possible senses. Since this is so, it is worth marking out the adequacy condition for future reference:

16

Macpherson (2011a, p. 139).

17

Macpherson (2011a, p. 140).

123

1266

R. Gray

Adequacy Condition: For a feature to be adequate for determining a dimension of the space of possible senses, it must be possible to use it to differentiate the possible senses and to provide a non-arbitrary ordering of the possible senses. Prima facie, it is hard to see how the proximal stimulus criterion is able to specify a feature that can both differentiate and order all the senses. Although it might be able to specify a feature that can differentiate vision from hearing, hearing from touch, touch from taste, and so on, it is not clear how such a feature can be used to generate a non-arbitrary ordering for the possible senses so differentiated. Here is one suggestion that might be made. But, as will become clear, it goes only part of the way to meeting the adequacy condition for a differentiative/ordering feature. To see how energy forms might be used to specify a feature that can provide the basis for a differentiative and ordering feature, consider our visual sense. It can be differentiated from the senses of other animals and ordered with respect to them by reference both to the range of electromagnetic radiation to which the senses are sensitive, and to the highest frequency of electromagnetic radiation to which the senses are sensitive.18 For instance, humans are sensitive to a narrower and lower range of electromagnetic frequencies than butterflies and to a wider and lower range than rats (see Fig. 1). The senses of other animals can be plotted, and thereby related, in a similar fashion. This would also provide a framework for individuating possible senses. Relatedly, our auditory sense and the senses of other animals can be differentiated and ordered by reference both to the range and to the highest frequency of sound waves to which the senses are sensitive.19 For instance, humans are sensitive to a narrower and lower range of sound frequencies than dogs, which are sensitive to a narrower and lower range of sound frequencies than bats (see Fig. 1). Again the actual and possible senses of other animals can be plotted. This proposal would not, however, meet the adequacy condition. For what is being proposed here is not a space of all possible senses but two spaces of two kinds of sensory modality. The positions in the two spaces are occupied by varieties of the kinds. Even if different varieties of the same kind of sense could be ordered by utilizing distinct dimensions in the way suggested, this would remain an incomplete

Fig. 1 Spaces of the senses (not to scale) ordered by range of em radiation and frequency of em radiation detected, and by range of sound waves and frequency of sound waves detected 18 The choice of dimensions is for illustrative purposes. Alternatively, or additionally, the lowest frequency detected could have been used as the basis for a dimension. 19

Again the choice of dimensions is for illustrative purposes.

123

Sensory Modalities

1267

ordering of all the senses. For ordering features whereby each of the possible senses can be ordered with respect to every other possible sense would still be lacking. A way would have to be found of incorporating these two spaces (and any other spaces of kinds of senses there might be) into a single space. Given that the dimensions are based on the values of types of energies, one suggestion might be that the spaces can be united into a single space via their origins at zero energy. However, this cannot be the right way to generate an ordering of all the senses because it would result in the paradoxical outcome that a sense that detects electromagnetic radiation of lower frequencies (for example, human or rat senses in Fig. 1) would be more similar to human hearing than a sense that detects electromagnetic radiation of higher frequencies (for example, butterfly vision in Fig. 1). Macpherson, however, seems to accept this, or, at least, something like it.20 Her idea is that a sense that is receptive to sound waves can be ordered with respect to senses that are receptive to electromagnetic radiation by common reference to their sensitivity to electromagnetic radiation. A sense that is receptive to sound waves would be plotted at zero; senses that are receptive to electromagnetic radiation would be plotted at various points along the dimension, having different degrees of sensitivity to electromagnetic radiation.21 According to Macpherson, this would provide a way of ordering the senses by reference to their proximal stimuli. For a sense with less sensitivity to electromagnetic radiation could then be shown to be more like a sense that is receptive to sound waves than a sense with greater sensitivity to electromagnetic radiation. This suggestion, however, seems counterintuitive. For both senses that are sensitive to electromagnetic radiation are unlike and thus incomparable to the sense that is receptive to sound waves with respect to the stimuli to which they are sensitive. For the required ordering to be possible it would have to be the case that the sense that is receptive to sound waves is also sensitive to the presence of no electromagnetic radiation. But this is not the case; on the contrary, the sense that is receptive to sound waves is not sensitive to the presence of electromagnetic radiation. In short, it seems plausible to think that the proximal stimulus criterion could be used to go some way to ordering the senses by ordering some of them, but it could not order all the senses, as is required by the adequacy condition on a dimension of a space of the senses.

5 The Representational Dimension According to the representational criterion, any token sense can be identified as a type of Aristotelian sense by reference to the representational content it determines. In order for the representational criterion to be used to generate a dimension it would have to provide a feature that meets the adequacy condition for a differentiative/ 20

So she has remarked in conversation and developed in a draft of her (forthcoming).

21

Conversely, senses that are receptive to electromagnetic radiation could be related to senses that are receptive to sound waves by common reference to their sensitivity to sound waves. The senses that are receptive to electromagnetic radiation would be plotted at zero; senses that are receptive to sound waves would be plotted at various points along the dimension, having different degrees of sensitivity to sound waves.

123

1268

R. Gray

ordering feature. It seems clear how such a feature could be used to differentiate the senses. For example, a visual sense can be differentiated from an auditory sense by reference to their distinct representational contents, and an auditory sense can be differentiated from an olfactory sense by reference to their distinct representational contents, and so on for other senses. However, it is less obvious how the kinds of senses so differentiated could also be ordered by reference to their representational content. For the senses would have to be ordered by reference to the degree of the differences in their representational contents. How is a colour more like (or unlike) a sound than a smell? The differences are categorical differences. To resolve this difficulty, at the same time as differing in their representational content, senses would have to have some representational content in common. Varieties of senses might be differentiated and ordered with respect to our senses by reference to more fine-grained variations of content such as differences in the range of colours, sounds, tastes and so on.22 This would also provide another framework for individuating possible senses.23 Like the account of the proximal stimulus criterion in the previous section, however, this would only provide an ordering of the varieties of different kinds of senses. For a complete ordering of the senses, the approach would have to be extended to all senses. What follows is the only way that I can think of to use the representational criterion as a way to differentiate and order the senses. But, again, it still only goes part way to meeting the adequacy condition. Macpherson gives one reason for echolocation being hard to classify by means of the representational criterion: it detects the size and shape of things at a distance by means of sound. Since it detects the size and shape of things at a distance, she claims that we might think of echolocation as somewhat like vision. Granting this, it would be the case that vision is more similar to echolocation than to hearing. Assuming that echolocation represents sound properties, one can also reason that hearing is more similar to echolocation than vision. Hence it might be argued that echolocation can be construed as being located between vision and hearing in so far as echolocation has more representational content in common with vision and hearing than the latter two senses have representational content in common. Similar orderings using common representational content might be established between vision, thermal imaging and heat sensations, and vision, TVSS and touch. Thermal imaging can be construed as being located between vision and feeling heat in so far as thermal imaging has more representational content in common with vision and heat sensations than the latter two senses have representational content in common. TVSS can be construed as being located between vision and touch in so far as TVSS has more representational content in common with vision and touch than the latter two senses have representational content in common. Figure 2 represents this suggestion for the relatedness of these senses by representational content. In this 22 Given the close relationship between the representational content and phenomenal character of experience, the difficulties raised in the next section might also be raised here. 23 We now have two frameworks for organizing the senses: by reference to proximal stimuli and representational content. Both of these could be used for individuating possible senses. One could also use these dimensions to conceive possible senses that occupied the location of an actual sense according to one differentiative feature, e.g. the location of the human visual sense, but took a different value according to the other differentiative feature.

123

Sensory Modalities

1269

Fig. 2 A space of the senses based on the representational criterion

way vision, echolocation, TVSS and thermal imaging can be construed as neighbouring senses. Feeling heat and hearing would be linked more distantly through thermal imaging and echolocation respectively. This model would also have the consequence of distinguishing heat and touch senses. One difficulty with this approach is that it can only extend as far as there are representational contents in common between the senses. So it would not be possible to include smell and taste since they do not, and plausibly could not, represent the shapes and sizes of objects at a distance. But there is a greater problem. On this interpretation of the representational dimension, there is a nexus where different senses represent common sensibles: the size and shape of objects at a distance. The problem with this is that representational contents that are distinctive of vision, echolocation, thermal imaging and TVSS are required to differentiate the senses. In which case vision, echolocation, thermal imaging and TVSS will still be differentiated without being ordered with respect to each other. Their ordering in Fig. 2 is just an arbitrary ordering; for instance, echolocation (and hearing) could have been placed where thermal imaging (and feeling heat) are placed. In discussing echolocation, Macpherson concludes that it is ‘‘like our vision in some respects and like our hearing in others’’.24 However, there is a significant difference in the respects. Putting the present point in another way, echolocation is like our hearing in a way which is distinctive of hearing; it is like our vision in a way which is not distinctive of vision. Hence it seems plausible to think that the representational criterion could be used to go part of the way to ordering the senses by ordering some of them, but, again, that is not to meet the adequacy condition on a dimension of a space of senses.

6 The Phenomenal Character Dimension A similar problem arises when one tries to use the phenomenal character criterion as a way of providing a feature that can be used to generate a dimension: although 24

Macpherson (2011a, p. 134).

123

1270

R. Gray

vision, hearing, taste and other senses can be differentiated by reference to the phenomenal character of experience associated with them, it is unclear how they can be ordered according to the relative differences of the phenomenal characters associated with them. In so far as they can be thought of as belonging to different quality spaces, the characters of experiences associated with vision, audition, gustation and other senses are incommensurable. If we think of the character of experiences by reference to distinct quality spaces that differentiate types of modalities, further difficulties arise. Consider the colour quality space. Its constitution is dependent on the mechanisms that are used by our visual systems to process physical stimuli. Species with similar visual systems, it is plausible to think, would have a similar range of experiences, and thus related quality spaces. Hence senses could be differentiated and ordered by reference to overlapping quality spaces. However, it is unclear whether this could be extended to all the senses that would be related to our visual sense by the other criteria. For the quality spaces associated with the experiences of species that have different light sensitive cells and processing mechanisms would have different dimensions and, in some cases, more of them. The resulting quality spaces would, therefore, be incommensurable with our colour quality space.25 Hence the phenomenal character criterion would provide a feature that differentiates some senses without ordering them as other differentiative/ordering features are able to do.

7 The Sense-Organ Dimension The sense-organ criterion seems a more promising place to look for features by means of which different senses could be differentiated and ordered. One method for differentiating and ordering the senses is suggested by the way in which species of organisms can be differentiated and ordered. Just as species can be related by reference to the relative divergence of their constitutive properties, so might the senses be differentiated and ordered by reference to the relative divergence of sensory organs. The divergence of sensory organs would be measured in respect of the divergence of their constitutive properties. What these properties are, and how they diverge, would be the subject of empirical investigation. It is possible to conceive a multidimensional space of the senses related to our visual sense that mapped a path from the simplest sense organ to the most complex variations on that simple sense organ through all the intermediate variations using the properties of the sense organs that vary as the differentiative/ordering features.26 While human eyes are very different from bird, goldfish and bee eyes, some indication of their degree of difference might be determined in this way. This would provide further reason to support the view that bee vision, one of the examples Macpherson uses to cast doubt on the criterial approach, should be construed as related to our visual sense. 25

See Allen (2009).

26

The presence of pigments in a wide variety of organisms provides some evidence of a point of origin for our visual sense in the Cambrian period. See Schwab (2011).

123

Sensory Modalities

1271

Employing such differentiative/ordering features, the space of actual senses could also be situated in a space of possible visual senses. The actual and possible varieties of the senses that are related to our auditory sense could be mapped in a similar way. In so doing, it would become clearer to what degree our auditory sense is related to bat echolocation. However, despite the possibility of such links being made between the senses by reference to the sense-organ criterion, such links do not seem to exist between all sensory organs. In particular no such link can be made between our visual and auditory sense organs.27 Hence ultimately it is also unclear how the sense-organ criterion, as modelled in this way, could provide a way of meeting the adequacy condition.

8 Are There Spaces of Sensory Modalities? Even if the criteria have not been used quite as Macpherson might have intended, it is hard to see how other relevant interpretations could resolve the problem that none of the criteria provide a feature that could be used to generate a non-arbitrary ordering of all the senses. It might at this point be argued that a key feature of Macpherson’s proposal has been omitted: it is the ‘‘relatively independent’’ dimensions taken jointly that constitute the space of sensory modalities. After all, each dimension of physical space only provides an incomplete ordering of the spatial points, and each dimension of the colour quality space only provides an incomplete ordering of the colours; other independent dimensions are required to constitute the respective spaces. However, this would not resolve the present problem, which is that none of the differentiative/ordering features order all of the senses with respect to each other. Still, it does raise another query to Macpherson’s original proposal. In the case of a quality space, qualities are differentiated from and ordered with respect to neighbouring qualities by one feature independently of their differentiation from and ordering with respect to another feature. The same goes for the items located in physical space. Yet this does not seem to be so for the senses. Although there is much dispute about the exact relationship between the phenomenal character of experience and the representational content of experience, what is not in dispute is the evidence from a variety of forms of perceptual experience that indicates how the variation of phenomenal character generally depends on the variation of representational content. Hence there is little reason to think that the phenomenal character criterion could provide a differentiative/ ordering feature that was, to any significant degree, independent of the differentiative/ordering feature based on the representational criterion. The relationship that holds between the differentiative/ordering features given by the representational criterion and the proximal stimulus criterion is complex. Nevertheless, of what we know about the various forms of perception there is a 27 This is not to dispute that important processing connections exist between the senses. It is only to claim that the sense-organ criterion does not provide a relevant way in which all the senses can be related such that they can be ordered.

123

1272

R. Gray

significant degree of dependence. For instance, seeing different colours is dependent on detecting different frequencies of electromagnetic radiation and hearing different sounds is dependent on detecting different frequencies of soundwaves. Hence it is unclear how the representational criterion could provide a differentiative/ordering feature that was, to any significant degree, independent of the differentiative/ ordering feature based on the proximal stimulus criterion. It is also hard to see how the differentiative/ordering features based on the proximal stimulus criterion and the sense-organ criterion are not co-related in some way. Sensory systems evolve in a certain way because they equip those creatures that possess such senses to detect certain stimuli that it is advantageous for them to detect. For instance, the human visual system has evolved to detect specific ranges of electromagnetic radiation, and the human auditory system has evolved to detect specific ranges of sound waves. Hence it is difficult to see how the proximal stimulus criterion could provide a differentiative/ordering feature that was, to any significant degree, independent of the differentiative/ordering feature based on the sense-organ criterion. Given that a case can be made for thinking that, in general, there is not a change in phenomenal character without a change in representational content, that there is not a change in representational content without a change in proximal stimulus, and that there is not a change in proximal stimulus detected without a change in type of sensory organ (including differences in the processing mechanisms of a type of sensory organ), it is hard to see how the four criteria could be used as the basis for four relatively independent differentiative/ordering features, and thus four relatively independent dimensions. When the four criteria that have standardly been used for determining whether a token sense is a certain kind of sense are used to provide features that generate dimensions, the outcome is not a multidimensional space constituted by independent dimensions in which each sense is located at a specific position relative to all the other senses in virtue of taking a value along each dimension. The outcome is four, more or less similar, orderings of the sensory modalities, none of which, however, orders each of the senses in relation to every other sense. Macpherson claims that differences between the senses amounts not to a difference of kind but to a difference of degree. Given that there are orderings but also that they are incomplete, there is evidence both for and against the claim. There is evidence for the claim that the differences between the senses involve differences of degree in so far as some senses can be ordered. But there is also evidence against the claim that the differences between the senses involve differences of degree in so far as some senses resist an ordering. The senses that can be ordered, and thus that differ by degree, can be regarded as varieties of the kinds of senses that cannot be so ordered. On this view, senses can be thought of as belonging to both a kind and variety. A kind of sense would be characterized by the possession of features that enable its differentiation from other kinds of senses without enabling its ordering with respect to them, and a variety of sense would be characterized by the possession of features that enable not only its differentiation from other senses but also its ordering with respect to them. Although our senses belong to different kinds, it might be misleading to think of those kinds in terms of our senses because some varieties of

123

Sensory Modalities

1273

the kind may be distantly related according to various of their features. It would be less misleading to construe a kind of sense simply as that which is constituted by those senses that can be ordered by reference to their differentiative/ordering features. In that a distinction can be drawn between varieties and kinds of senses, there is also some support for both a sparse view of the senses and a rich view of the senses. A sparse view is supported by the fact that the criteria differentiate some senses without ordering them. A rich view is supported in two ways: by the fact that the criteria differentiate more kinds of sense than the five senses recognized by common sense, and by the fact that the criteria differentiate and order many varieties of the kinds of senses. One final point should be made. Although the differentiations and orderings produced by the differentiative/ordering features that are derived from the familiar criteria are, in many ways, similar, they are not the same. The similarities arise because the features that are used to differentiate and order the senses depend on the presence of the other features that are used to differentiate and order the senses. The differences arise because the differentiative features are distinct in nature and depend in different ways on the presence of the other differentiative features. For instance, thermal imaging senses can be related to one set of senses that are receptive to electromagnetic radiation and another set of senses that have thermal properties as their contents. In so much as an understanding of the senses requires an understanding of the nature of these features of the senses and the various dependence relations that hold between them, thinking of the senses dimensionally does not just pertain to the issue of how we classify the senses, it is the basis for any understanding of the senses.28

References Allen, K. (2009). Inter-species variation in colour perception. Philosophical Studies, 142, 197–220. Clark, A. (1993). Sensory qualities. Oxford: Clarendon. Gray, R. (2005). On the concept of a sense. Synthese, 147, 461–475. Grice, P. (1962). Some remarks about the senses. In R. J. Butler (Ed.), Analytical philosophy (pp. 133–153). Oxford: Blackwell. Keeley, B. L. (2002). Making sense of the senses: Individuating modalities in humans and other animals. The Journal of Philosophy, 99, 5–28. Macpherson, F. (2011a). Taxonomizing the senses. Philosophical Studies, 153, 123–142. Macpherson, F. (2011b). Introduction: Individuating the senses. In F. Macpherson (Ed.), The senses: Classic and contemporary philosophical perspectives. Oxford: OUP. Macpherson, F. (forthcoming) The space of sensory modalities. In D. Stokes, S. Biggs & M. Matthen (Eds.), Perception and its modalities. Oxford: OUP. Nudds, M. (2003). The significance of the senses. Proceedings of the Aristotelian Society, 102, 31–51. Ross, P. (2008). Common sense about qualia and the senses. Philosophical Studies, 138, 299–316. Schwab, I. (2011). Evolution’s witness: How eyes evolved. New York: OUP. 28

I am grateful to Fiona Macpherson for a number of enjoyable and informative discussions on the issues raised here stretching back several years now. Many thanks also go to Ivan Ivanov and Debbie Goldgaber for their invitation to a very pleasant and profitable workshop at Northwestern University, where a first version of this paper was presented. I am also grateful for the advice provided by a referee for this journal.

123