2011-2012
Philosophic
Exchange
Contents
Virtue and Flourishing in
Our Interpersonal Relationships............. Lorraine Besser-Jones .................. 5
Philosophy of Perception and the
Phenomenology of Visual Space............. Gary Hatfield .............................. 31
Alcibiades and the Politics
of Rumor in Thucydides ......................... C. D. C. Reeve ........................... 67
The Ontological Argument
and Objects of Thought .......................... Edward Wierenga ....................... 81
Gary Hatfield
Philosophy of Perception and
the Phenomenology of Visual Space
Gary Hatfield
In the philosophy of perception, direct realism has come into vogue.
Philosophical authors assert and assume that what their readers want, and what
anyone should want, is some form of direct realism.1 There are disagreements
over precisely what form this direct realism should take. The majority of
positions in favor now offer a direct realism in which objects and their material
or physical properties constitute the contents of perception, either because we
have an immediate or intuitive acquaintance with those objects and properties,
or because our perceptual states have informational content that represents the
properties of those objects (and which is not itself an object of perception and
has no specifically subjective aspect).
My aim in this paper is to consider various forms of perceptual realism,
including, for purposes of comparison, the largely abandoned indirect or
representative realism. After surveying the variety of perceptual realisms and
considering their various commitments, I introduce some considerations
concerning the phenomenology of visual space that cause trouble for most forms
of direct realism. These considerations pertain to the perception of objects in
the distance and, secondarily, to the perception of shapes at a slant. I argue that
one of the lesser known varieties of perceptual realism, critical direct realism,
can meet the challenges offered by the facts of spatial perception.
Varieties of Perceptual Realism
Perceptual realism, as understood here, maintains that we perceive mindindependent physical objects (some prefer to say “material objects” or “physical
entities”). Hence, it holds that physical objects exist, independent of our perception
of them. The manner in which we perceive the objects is differently analyzed
in the various forms of perceptual realism. The terms “perceive” and “physical
object” are sometimes understood differently within the various positions, in ways
that are mentioned as needed. In presenting some standard types of perceptual
realism, I list some benefits and problems. As is common, I focus on vision.
Some of the same questions and problems arise for the other senses, but some
may not.
Traditional representative realism
According to this position, the items that we see have the properties they
34 Gary Hatfield
appear to have: elliptical brown patches when viewing a penny from an angle,
trapezoidally shaped color patches with table tops. From these phenomenally
immediate representations, we infer actually existing physical objects that have
corresponding properties: rectangular for the table top; for its brown color, a
physical basis for reflecting light that causes the directly experienced brown of
the representative item. The immediately perceived items are conceived either as
mind-dependent sensory ideas or as third things. In either case, these mediators
have the phenomenal properties they seem to have and (if everything goes right)
these properties represent the physical objects that are inferred from them.
Because these mediators are the immediate objects of perception and yet provide
a basis for inferring mind-independent objects, this position is a form of realism:
representative or indirect realism.
Here the notion of “physical object” might be equated with an ordinary
notion of a material object but also might, as with Locke, be conceived according
to current science. In either case, the position holds that mind-independent
physical objects are inferred from items that are distinct from them and that
are themselves directly perceived. The notion of direct perception involves an
unmediated conscious acquaintance with the representative item. The classical
sense-data theories of Moore (1909–10) and Russell (1912) held that these items
are non-mental third things: shaped color patches that are distinct from the
surfaces of objects and which are immediately present in consciousness.2 More
recent discussions typically and incorrectly assume that sense data were generally
supposed to be mental items (e.g., Brewer 2011, 63, 139).
Classical arguments for representative realism, as summarized by Russell
(1912), appeal to perceptual relativity. According to this argument, when we look
at a static object in our environment from various viewpoints, we see a variety
of shapes: trapezoids with a rectangular table top viewed at an angle, elliptical
brown patches with a penny. But we correctly believe that the real table top is
rectangular, not trapezoidal. Hence, the item that we see cannot be the table top,
since that top cannot at the same time be both rectangular and also of several
different trapezoidal shapes. Similarly, the flat side of a penny cannot have various
elliptical shapes as well as a circular (non-eccentric elliptical) shape (Russell 1912,
chs. 1–3).
The argument depends on a premise according to which, when we visually
experience or “see” something having a determinate shape and color, there must
be some existing item that has that very shape and color. In Russell’s version, this
item is a sense datum, conceived as a third thing. On one common reading of
Locke (e.g., Brewer 2011, 44–7), these shapes and colors are properties of ideas
and so are in the mind. In either case, the immediate object of perception is the
item (third thing or idea) that has the properties in question.
Some benefits accrue to the representative realist position. The account posits
Philosophy of Perception and the Phenomenology of Visual Space 35
a direct acquaintance with items that have the very properties they are experienced
to have; the colors and shapes are literally present to experience. This accounts
for the experiential presence of phenomenal properties and explains why these
items would be a good foundation for empirical knowledge, on the assumption
that what is immediately present in consciousness can be well known (on which,
see Dicker 1980, ch. 2).
There are some standard problems for the position. One may ask what these
items are that are literally brown and trapezoidal. If mental, then a mental state
is brown, which many find problematic. If the items are non-mental, then there
are three things: (1) perceiver, (2) external material object, and (3) non-mental
representing item. Can’t we get by with only two things, individual perceivers
(with their perceptual states) and objects? The standard sense-datum theory seems
ontologically bloated. Further, representative realism may abet skepticism. If we are
acquainted only with our representations, how can we know that they are caused
by objects having properties that correspond in some way to the immediately
perceived properties?
Naïve direct realism
According to this position, the object itself is directly available to consciousness.
We see the object with its properties without phenomenal or representational
mediation of any kind. The object and its properties are present to consciousness
in a manner similar to the immediate presence of representative items in the
previous position (as Brewer 2011 emphasizes). The table or the penny, which
are rectangular and brown, or circular and brown, are the immediate items of
perceptual awareness. In the normal case, they are what we see. Naïve realists are
typically disjunctivists with respect to hallucination. They hold that in hallucination
there is no object of which we are directly aware, and so they need to provide an
account of what happens in hallucination that differs in kind from what happens
in normal (successful) perception.
Many consider it a benefit that naïve realism avoids saying that we “see” our
representations. We see the objects themselves. This position also avoids the
skeptical problem that can arise from the need to infer material objects from
representations. If there are no representational intermediaries, then the object
itself can be directly known.
Questions arise concerning how naïve realism accounts for the way things
look. If we are directly aware of objects with their properties, we should see them
as they are. However, the table is rectangular but can appear trapezoidal; objects
having the same size appear smaller when perceived at a distance but are not
smaller. So it seems that we don’t experience them as they are. In response to
this kind of worry, recent naïve realists3 such as Noë suggest that what we directly
36 Gary Hatfield
experience is a function both of the object’s properties and of their relations to
us, including distance and orientation. He appeals to “mathematical laws” of
perspective to explain why things look small in the distance, or elliptical when
viewed at a slant (Noë 2004, 83). The sustainability of the naïve realist’s disjunctive
account of hallucination also has been called into question (e.g., Pautz 2010).
Naïve direct realism as applied to color requires that the table has the brownproperty that it appears to have, running counter to the distinction between
primary and secondary qualities and putting itself in opposition to modern
color science. For these reasons, some naïve realists adopt what Price (1932, ch.
3) called a “modified” version of the position: they posit a direct acquaintance
with primary properties but hold that secondary qualities are experienced via
sensations or sensory intermediaries (e.g., Cook Wilson 1926, Prichard 1950).
But some naïve realists adopt Campbell’s (1993) “simple view” of color, according
to which things have the very color properties that we experience phenomenally
(e.g., Brewer 2011, 176 n. 9).
Intentional, pure informational, representational, or “content” direct realism
On this view, our experience of external objects consists of our sensory
system’s being in certain representational or intentional states, or possessing
a content. This position is difficult to name, because the terms “intentional,”
“representation,” and “content” have a variety of uses. But the idea is clear: to
see a rectangular brown table top is to have visual representational content of the
physical properties of the table top. This content may be nonconceptual, in which
case the label “pure information theory” is appropriate (Crane 1992; see, e.g.,
Dretske 1995).4 This view is a form of direct realism because the informational
state of the visual system is not deemed to be the object of perception or to be a
sensation or other subjective aspect of consciousness. Rather, that informational
state is itself the experience of the physical properties of the object of perception.
We see objects by having appropriate sensory content.
The position avoids subject-dependent qualities or subjective intermediaries
by insisting that the content in question represents the physical properties of the
objects of perception (Dretske 1995; Tye 1995, chs. 5–6; 2002).5 The visual system
represents the rectangularity of the table top, perhaps from a point of view. It
represents the physical color property (usually thought of as a surface reflectance
property), perhaps under given lighting conditions. In this manner, we become
visually aware of properties of the table top itself in a direct way.6
This position shares some advantages with each of the previous two. Along
with naïve realism, but for a different reason, it avoids saying that we “see” our
representations. Along with representative realism, it offers an account of the
relativity of perception. Differences in perceptual content explain the differing
Philosophy of Perception and the Phenomenology of Visual Space 37
ways the table top looks from various perspectives, thus avoiding the prima facie
worry that arises for naïve realism. (Whether the position can account for all aspects
of perceptual relativity must remain open.) Further, along with representative
realism, the “content view” holds that in cases of illusion and hallucination,
perceptual states share the type of content with veridical perceptual states. If we
see a stick in water as bent, we do so by representing the physical property of being
bent. In this case, that content misrepresents the actual stick. This view claims
to account for illusions and hallucinations without introducing phenomenally
subjective intermediaries; the representation (allegedly) is transparently “of” an
external property instance. Illusions and hallucinations are misrepresentings of
physical properties (Tye 2002, Pautz 2010).
The difficulties for this position concern technical issues about the perception
of color and spatial properties such as size and shape. As regards color, this
position depends on color being a categorical physical property of surfaces (or
of light itself). That claim is controversial, and we shall not examine it here.7 For
spatial perception, the position appeals to the same sort of “mathematical laws”
to which the naïve realist appeals. That claim is taken up below.
Critical direct realism
According to this position, actually existing material objects are presented
to us through their appearances. We directly perceive the object by means of
mediating appearances or representations of it. These appearances are mental
states, and include subjectively qualified sensory content, including color qualia
and spatial features. Because these mediating contents or representations are
available in consciousness, we can attend to the features of the appearances, but
our normal focus is on the objects themselves. We do not infer objects from the
appearances, rather those appearances directly represent the properties of the
objects themselves (nonconceptually).8 We may cognitively affirm the presence of
an object without making a judgment or inference. We have the appearances, and
we see the objects in virtue of having them. The appearances include determinate
shades of brown and determinate surfaces with a size and shape, but these are in
the appearances only as phenomenal intentional features. The appearances are
not literally brown or rectangular, but they present to consciousness brown and
rectangular qualitative phenomenal contents that represent object properties.9
Like naïve and “content” direct realism, this position avoids saying we “see”
our representations. It also avoids positing third things or mental states that
are literally brown and rectangular. Unlike naïve realism, it avoids rendering
phenomenal color as a property inhering in the surfaces of objects. It explains
apparent or phenomenal properties as mental contents, but these contents are
phenomenally characterized. Hence, it can explain phenomenally available subject-
38 Gary Hatfield
relative aspects of perceptual experience via phenomenal experiences that are
themselves part of the act of seeing. These subject-relative aspects of experience
nonetheless present or make manifest objective properties of external things.
The questions that arise for this position include: What are these appearances?
If they are not literally brown or rectangular, how is it that they phenomenally
present such properties? How is it that we see things directly by means of a mental
content? How can subject-dependent aspects of experience present the objective
properties of things?
Of the positions presented above, critical direct realism is the least well
known. In several recent overviews, it is not mentioned (e.g., Brewer 2011; Smith
2002, Intro.).10 Of positions that have been extensively articulated in the past
two decades, Smith (2002) accords with critical direct realism as I have explained
it thus far, but he calls his position simply “direct realism,” which he distances
from naïve realism. He also separates himself from disjunctivists, who offer
differing analyses of the content of veridical perception in contrast with illusion
or hallucination (and who may be naïve realists or a certain variety of “content”
direct realist).11 Although Smith fits the category of critical direct realist to this
point, I subsequently qualify the position in a way that no longer fits with the
details of Smith’s account. Hence, for now I classify him as a sophisticated direct
realist.
Naïve Realism, the “Content View,” and
Critical Direct Realism Contrasted
In a recent book on perception and its objects, Brewer (2011) uses the device
of a triad of inconsistent theses to characterize the commitments of various
positions on the metaphysics of perception, including: idealism; representative
(or “indirect”) realism; the “content view”; and his own version of naïve realism,
which he calls the “object view.” An exemplary feature of his discussion of the
triad is to show that, while in some cases adherents of a position simply reject
one member of the triad outright, in other cases they reject one member and
reinterpret another. This feature of his analysis can provide a springboard for
my own subsequent elaboration of critical direct realism.
Brewer’s triad
Brewer (2011, 11) formulates his inconsistent triad as follows:
(I) Physical objects are mind-independent.
(II) Physical objects are the direct objects of experience.
(III) The direct objects of experience are mind-dependent.
Philosophy of Perception and the Phenomenology of Visual Space 39
Brewer understands “physical objects,” in the first instance, to be the objects
recognized by common sense, such as “stones, tables, trees, and animals” (ibid., 2).
He takes it that adherents of (I) hold that the “nature” of such objects—what they
are in themselves—is independent of perception but is also empirically available
in perception. In saying that the natures of things are available in perception, he
does not mean that their scientific natures are manifest (so that the fundamental
concepts of physics could be “read off” experience); rather, he means that objects
have accessible natures that can be formulated in such commonsensical terms as
their being “causally integrated, enduring, and spatially extended material unities”
(ibid.). According to (II), these very objects are the direct objects of experience, in
a way to be explained. According to (III), however, the direct object of experience,
or what we perceive directly, is mind-dependent. Obviously, one can’t accept all
three theses (each pairing contradicts the remaining thesis).
According to Brewer’s analysis (2011, ch. 2), Berkeleian idealism rejects (I):
there are no mind-independent physical objects of the sort envisioned in (I). To be
is to be perceived. Adherents of representative (or indirect) realism reject (II). They
affirm that mind-independent physical objects exist, but such objects are perceived
only indirectly, being inferred from the mind-dependent objects countenanced in
(III). Locke provides the primary exemplar of this position (Brewer 2011, ch. 3).
Although Brewer himself rejects Berkeley’s idealism and Locke’s indirect realism,
he accepts what he takes to be their shared conception of perceptual acquaintance: to
perceive something directly is for it to stand before consciousness in an unmediated
fashion. There is a simple and direct relation between a person’s consciousness
and the objects of acquaintance. According to Brewer, Locke and Berkeley were
right about the structure of acquaintance but wrong about what its objects are in
perception: they both chose mind-dependent ideas, while disagreeing over whether
those ideas are further related to mind-independent physical objects.
The two remaining positions that Brewer examines agree in rejecting (III).
They are differing versions of direct realism. According to Brewer’s own object
view (2011, ch. 5), in perception we directly perceive physical objects through the
relation of acquaintance. There is a direct relation between consciousness and
objects such as stones, tables, trees, and animals. By contrast, the content view
does not accept perceptual acquaintance as conceived by Brewer to be the relation
between perceptual consciousness and physical objects. Rather, it holds that we
perceive objects by representing them (Brewer 2011, ch. 4). These representations
are not themselves the objects of perception, hence the content view’s rejection
of (III).
From Brewer’s perspective, although adherents of the content view claim to
accept (II), they don’t accept it under the interpretation that he intends for it, as
involving a relation of acquaintance. In accepting (II), adherents of the content
view redefine the notion of a “direct object” of perception. They describe their
position as a direct realism because it doesn’t posit any intermediaries that are
40 Gary Hatfield
themselves perceived. Although Brewer considers various versions of the content
view, in the version considered herein, that view holds that perceiving is constituted
by the having of certain kinds of nonconceptual content, which represent instances
of the physical properties of things (including, for color perception, properties that
must be described in the language of physics).12 Thus, although the content view
invokes representations, because these representations have only physical content,
there is nothing peculiarly subjective or mind-dependent about that content. If
perception is successful, instances of those properties in the world are what is seen,
and hence are the objects of perception.
Although Brewer does not discuss critical direct realism, we may ask how it
stands with respect to his inconsistent triad. As drawn from R. W. Sellars (1916,
1920, 1961, 1969), critical direct realism accepts (I), reinterprets (II), and replaces
(III). The critical realist accepts the existence of mind-independent physical objects,
although for the purposes of analyzing perception (and especially of secondary
qualities), these are understood in accordance with the findings of the natural
sciences, including physics, chemistry, biology, and psychology. That is part of what
is intended by the adjective “critical.” The critical direct realist reinterprets (II) by
working on the notion of “direct perception.” The position views the notion of
acquaintance as a holdover from an early modern analysis of thought in terms of
intuited or directly perceived mental contents, and as sharing a relational act-object
account of sense perception with the classical sense-data theorists (Sellars 1920,
199; 1969, 46, 78). We may add that, by invoking acquaintance, naïve realism
seems to want to explain seeing by appeal to a primitive notion that itself is best
understood -by analogy with seeing; the position begins and ends with the point
that, phenomenally, seeing just seems to be direct acquaintance with the world.
In place of the commonsensical relation of direct acquaintance, the critical direct
realist offers an analysis of perception as involving two components, a phenomenal
content and its object. In perception, we have mind-dependent phenomenal states.
These are not bare sensations, but are themselves representations of external objects
or object-properties. Sellars (1920, 195) calls them “characters” or “contents” that
are “given” in experience, by which he means that they consist in phenomenally
available content that presents object-properties.13 But these phenomenal states are
not what is perceived in normal sense perception. We perceive object-properties
in virtue of being in these states. In dynamically interacting with objects through
the guidance of these appearances, we affirm the presence of mind-independent
things.14 Thus, Sellars’s reinterpretation of (II) and replacement of (III) might run:
(II′) Physical objects are directly perceived and affirmed via phenomenal mental
contents.
(III′) The phenomenal mental contents have mind-dependent features and yet
mediate the direct perception of physical objects and their properties.
Philosophy of Perception and the Phenomenology of Visual Space 41
Thesis (III′) receives further elaboration in a subsequent section.
Sellars (1920, 196) pairs the appearances with a cognitive act of affirming the
physical object. By contrast, Smith (2002), who thus far is an unwitting friend
of critical direct realism, offers a focus on sensory aspects of perception that are
object-presenting. According to Smith (2002, ch. 6), the perceptual constancies
allow our sensory content to become “perceptual” (object-presenting). The content
of the phenomenal experience of size, shape, and motion presents things as having
constant size and shape and distinguishes subject-motion from object-motion,
despite other changes of appearance. The constant sensory elements in perceptual
experience are therefore object-presenting, rather than being merely subjective.
In a subsequent section, I offer an alternative to Smith’s analysis, one that also
takes sensory rather than cognitive (or conceptual) dimensions of perception as
a starting place but differs in its analysis of those sensory dimensions.
Problems for the object and content views
In the previous section, I mentioned that for the object view and its kin,
a problem arises regarding the phenomenal look of tables, pennies, and other
objects. The table top may appear trapezoidal. But if the table top itself provides
the content of experience by being directly present in consciousness in all its naked
glory, and if it is indeed rectangular, whence comes the trapezoidal appearance?
Similarly, objects at a distance may be phenomenally smaller than when seen
close at hand. Assuming that the object remains the same physical size when
near and far, if the object itself provides the very content of experience, it should
appear with its actual size at both distances.
Naïve direct realists from Cook Wilson to Noë and Brewer have responded to
perceptual relativity by appealing to point of view and lighting conditions. They
suggest that facts about physical environmental conditions and the spatial relations
of perceivers to objects enter into the bare relation between consciousness or
perception and the physical object. Brewer (2011, 96) states his reply to perceptual
relativity as follows:
perceptual experience is a matter of a person’s conscious acquaintance
with various mind-independent physical objects from a given spatiotemporal
point of view, in a particular sense modality, and in certain specific circumstances
of perception (such as lighting conditions in the case of vision). These factors
effectively conjoin to constitute a third relatum of the relation of
conscious acquaintance that holds between perceivers and the mindindependent physical direct objects of their perceptual experience. Thus
the experiential variations noted above, and any others along similar
42 Gary Hatfield
lines, may all perfectly adequately be accounted for by variations within
this third relatum.
The experiential variations include those that come from viewing a coin head on
and from an angle. Brewer’s appeal to a spatiotemporal point of view to explain
the elliptical appearance requires that the physical relation to the penny (the
third relatum) can itself provide the elliptical content of perceptual experience.15
Brewer’s treatment of perceptual variation is rather abstract. It consists
primarily in assertions about how (e.g.) the coin looks at an angle together with
the claim that perceivers may notice a similarity between (a) how the coin looks
and (b) how other objects with differing physical properties look, such as actually
(eccentrically) elliptical objects. There is in fact controversy over how a penny (or
other round object) looks when viewed at a moderate angle, such as 60 or 45
degrees from the plane extended from the flat side of the penny. Some think that
the penny doesn’t look elliptical at all, but rather looks like a circle at a slant (e.g.,
Smith 2002, 172). In any case, Brewer’s “explanation” of how a spatiotemporal
“point of view” helps simply appeals to the (alleged) fact that pennies do look
elliptical when viewed “from an angle” (2011, 163). He does not explain how the
specifics of the perceiver’s spatial relation to the coin should enter in.
By contrast, naïve realist Noë (2004, 82–4) does offer an explanation. Because
the phenomenology of the penny example is disputed, I consider another of Noë’s
examples, which has also been widely discussed, the diminution in apparent size
of objects with increased viewing distance. (The phenomenology here is also
disputed, but the issues are more straightforward.)
Noë distinguishes size and apparent size. Things remain the same size in the
distance, but their apparent size is somewhat smaller. Thus far, he uses language
that other types of perceptual realist might use. But as a naïve direct realist, Noë
must explain apparent size as a matter of mind-independent physical relations.
Here is how he does it (2004, 82–3):
We can distinguish size and apparent size, or size and how things look with
respect to size from here (“size in the visual field”). Size in the visual field
is a distinct property from size. It corresponds to the size of the patch
that one must fill in on a given plane perpendicular to the line of sight
in order to perfectly occlude an object from view. So long as you specify
the plane as located, say, at some distance in front of the perceiver, how
things look with respect to size can be recognized to be a perfectly definite
property of the scene (Armstrong 1961; Harman 1990). Let us call this
property the perspectival size of an object.
Noë recognizes that the important thing for the naïve realist is that apparent
Philosophy of Perception and the Phenomenology of Visual Space 43
size be a property of the scene. It may depend on the perceiver’s relation to the
scene and on the fact the observer is sighted, but, he claims, it does not depend
on subjective factors such as “sensations and feelings.” Instead, it follows from
“mathematical laws” of perspective (2004, 83).
Noë’s claim that, for an object of a given size at a given distance, its “occlusion”
size is fixed (if the occlusion plane is fixed), is correct. But, as I argue in the next
section, this fact alone does not determine “apparent size” or phenomenally
experienced size. What phenomenal size a perceiver experiences for a given object
and distance is determined, I claim, by subjective facts about the perceiver’s visual
system. By this I mean that the mathematical laws of optics and of perspective do
not by themselves specify the size that will be experienced. If things look small in
the distance, then, contrary to common opinion, this is not due merely to their
being at a distance. The laws of optics allow that things in the distance could
appear with the same phenomenal size as they have when close at hand.
Before turning to that argument, we should recall that the issues of perceptual
variability and relativity also come up for the “content view.” Adherents of that
view appear, prima facie, to be better off than naïve realists, as they can appeal
to representational content in their explanations of things looking small at a
distance. However, for veridical perception, they are committed to the claim that
the content must accurately represent the actual physical properties of the object.
They can explain deviations from physical size as a matter of misrepresentation.
So for them the questions become: can they explain phenomenal looks of nearby
objects by appeal to the veridical representation of physical properties alone,
and can they explain the everyday phenomenal looks of objects at moderate
distances, without supposing that these looks are cases of misrepresentation?
On the basis of the same arguments that I use in response to Noë, I contend
that adherents of the content view cannot explain the phenomenal perception
of size close at hand and at moderate distances without greatly expanding the
domain of misperception (which they would be loath to do).
The Geometry of Visual Space, and Things Looking Small at a Distance
In vision, we experience things as being at a distance.16 The world of visual
experience includes depth and distance. Are the spatial structures in visual
experience just the actual physical structures of the things that we see? A little
phenomenal reflection suggests not. Roads, sidewalks, railway tracks, and hallways
often have parallel sides, but, phenomenally, the sides converge. Trees, cars,
and persons retain the same physical height when found at different distances,
but at greater distances they appear in some way smaller. This convergence or
diminished size, however, cannot be exactly the same as that found in perspective
drawings, since those drawings are two-dimensional and are usually seen close
44 Gary Hatfield
up, but experiences of the sidewalks, hallways, trees, and cars clearly involve
depth and the third dimension. Things are phenomenally smaller, and they are
experienced in the distance.
The naïve realist and the “content” direct realist need to explain the facts of
convergence and apparent smallness as the product of merely physical relations,
which are either (a) properties of the objects of direct acquaintance or (b) properties
of the scene that are veridically represented (in the usual case). In order to assess
whether they can do so, we need to examine the relation between distance and
size perception in some detail.
In this section, I will consider the relations among the following three
propositions:
(1) we see things in straight lines according to their direction from us;
(2) we see things at a distance and in the distance; and
(3) the laws of optics require that things physically farther away should
appear smaller.
These propositions all have adherents, but they cannot all to be affirmed advisedly.
(1) is an empirical claim about the relation between physical direction and
experienced visual direction (which we can restrict to the central portion of the
visual field); the claim that experienced visual direction matches physical direction
is fundamental to vision science. (2) is a widely accepted empirical claim, which
is granted by Noë. (3) is a claim about what follows from the laws of optics, that
is, from the mathematical relation that things (including reflected light) have to
our eyes. But, I claim, given (1) and (2), the derivation mentioned in (3) must
be false: that is, it does not follow from the laws of optics and perspective that
things in the distance must appear (phenomenally) smaller than things seen close
at hand. The burden on me is to show that (3) is incompatible with (1) and (2).
Visual angle, perspective size, and optical apparent size
Theories of vision have long been built on the fact that we experience things
along lines of sight, as stated in proposition (1). For our purposes here, we can
take these lines of sight to be visual directions from a given point, and we can
characterize the relations between them in terms of the angle formed by any two
given lines of sight. The geometry of such directions and angles is well known.
Objects that are the same size subtend smaller angles as their distance from the
viewer increases (Fig. 1a). Objects of different sizes at different distances can
subtend the same angle (Fig. 1b). And for objects of different sizes viewed at the
same distance, the smaller object subtends a smaller angle (Fig. 1c).
Philosophy of Perception and the Phenomenology of Visual Space 45
(a)
P
(c)
A
B
P
A
B
P
A
(b)
P
B
Figure 1. Relations among object size, distance, and visual angle. In (a), objects
of the same size at different distances subtend different visual angles; B, which is
farther away, subtends a smaller angle. In (b), objects of different sizes at specified
distances subtend the same visual angle. In (c), objects of different sizes at the same
distance subtend different visual angles; the larger object subtends the larger angle.
These geometrical relations govern sight, but they do not, by themselves,
determine the sizes of things seen in the distance, or indeed at any distance
whatsoever. For something to be seen as having a determinate size, it must be
seen as being at some determinate distance, along determinate lines of sight.
Psychologists call this the size-distance invariance relation.17 All of the diagrams
in Figure 1 can be taken as illustrations of it. Consider Figure 1a. Segments A
and B are the same size, but B is farther away from the perceiver, P. Therefore,
B subtends a smaller visual angle. However, the phenomenal sizes of A and B
depend on the perceiver’s perception of the distance to A and B. If the distances
to both A and B are perceived veridically, then the phenomenal sizes of objects
A and B are perceived veridically. B is farther away than A, but is experienced
with the same phenomenal size. This is called (phenomenal) size constancy.
If size constancy were perfect, then objects would always be seen at their true
distances with their true sizes. But we don’t always experience objects with their
true sizes at various distances. In our actual visual experience, things even at
moderate distances are, phenomenally, smaller than things of the same size that
are yet closer. One can easily become aware of this fact by noticing the phenomenal
sizes of people’s heads in an audience at a lecture; or the phenomenal sizes of
people or cars looking down a sidewalk or street; or indeed the phenomenal
convergence of the edges of a sidewalk or hallway, which is a manifestation of a
diminishing phenomenal size for the width of the sidewalk or hallway.
46 Gary Hatfield
The phenomenal sizes of things fall off regularly with distance. This does not
mean that we think the objects are getting smaller physically; we are aware that
they don’t get smaller. Still, we would be very surprised if a person at twenty feet
away looked exactly the same, as regards phenomenal size, as the same person at
five feet away.
In the previous section, we saw that Noë was sensitive to these facts and
attempted to explain them with the notion of “perspective size” (or occlusion
size). This explanation doesn’t work. In Figure 1b, objects A and B have the same
occlusion size, in the sense that A occludes B. But, manifestly, small objects, such
as toy cars, that would exactly cover a real car of a similar shape at a distance, do
not look to have the same phenomenal size if we shift our gaze from the occluding
toy (now moved a bit to the side) to the distant car.
In order to understand what goes wrong with explanations like Noë’s (which
are fairly widespread), we need to distinguish an optical notion of “apparent size”
from a perceptual or psychological notion. The optical notion simply equates apparent
size with visual angle, as does Noë. Objects A and B in Figure 1b have the same
apparent size, optically defined. This notion of apparent size corresponds to what
is called “extensity” in the field of vision (Rock 1975, 561). Let us define the field
of view as what is included within the angles between the top most perceived
direction and the bottom most, the farthest perceived directions on the left and
right, and so on (to produce an oval field of vision). Thus far, only direction,
not distance, enters into determining the field of view. Optical apparent size
corresponds to the percentage of this field taken up by an object. From where I
am sitting, a hand held up one foot and a half from my face takes up the same
amount of the field of view as does a casement window about eight to ten feet
away. Optical apparent size is phenomenally accessible as the percentage of the
field of view that an object occupies.
This sort of geometrical fact is the kind of thing Noë has in mind with his
“perspective size.” But perspective size is not adequate to the phenomenal facts
and does not cover the vast majority of cases in which things look smaller in
the distance. For one thing, that things look smaller in the distance is easily
noticed. But human perceivers are not very good at estimating visual angles or
at matching visual angles, except when the stimuli are optically contiguous, that
is, are phenomenally next to one another in the field of view (Joynson 1958a, b).
Usually, optical apparent size has little phenomenal salience. More importantly,
the phenomenal structure of visual space includes diminished size in the normal
case (as with sidewalks, roads, hallways, and the heads of people in an audience
toward the back). I contend that in normal visual experience, the experientially
salient phenomenal size is not determined by perspective size (visual angle) alone,
but requires the notion of psychological apparent size.
Philosophy of Perception and the Phenomenology of Visual Space 47
Psychological apparent size and the geometry of visual space
The geometry of size perception has long been of interest in vision science. It
has long been recognized that, in the normal case, the perception of size depends
on both visual angle and distance, in accordance with the size–distance invariance
relation (Ross and Plug 1998). As adumbrated above, this relation specifies that
perceived size varies directly as a function of visual angle and perceived distance.
Two objects that subtend the same visual angle but are perceived as being at
different distances will appear of different sizes. Figure 1b illustrates this relation
in the case of veridically perceived distance. A and B subtend the same visual
angle, but if the endpoints of B are perceived along straight lines (that is, under
their true visual angle) and as being at their true distance, then B will appear
(will be experienced phenomenally as) larger than A.
From the times of Ibn al-Haytham (ca. 1030/1989, 174) and Descartes
(1637/1965, 107), it has been known that perceptual experience of things near
at hand nearly obeys size constancy. However, modern research has produced
results that are at odds on exactly how good size constancy is for things even as
close as 20 feet. Some investigators, such as Gibson (1950, ch. 9), have found
excellent size constancy at distances up to nearly one half mile. Others find a
diminution of phenomenal size constancy at 20 feet, and greater diminution
at 200 feet, under experimental conditions meant to distinguish apparent or
phenomenal size from inferred size (Granrud 2012). That is, they find that, when
an experiment is set up to distinguish inferred or estimated physical size from
apparent or phenomenal size, the latter diminishes with distance, but does so
less rapidly than would be predicted by the diminution in visual angle (optical
apparent size) alone.
Perceptual psychologists have developed experimental techniques to address
the problem of distinguishing between how the sizes of things look to their subjects
and the beliefs or guesses that their subjects may make about the true sizes of things.
Experimentalists have developed different sets of instructions, in order to control
the attitude that subjects take toward their task in a size constancy experiment.
Under “projective” instructions, subjects are supposed to respond to visual
angle; this would mean a match to “perspective size” as defined by Noë (either as
“extensity” or as projective size on a plane at a fixed distance from the perceiver).
Under “apparent” instructions, they are supposed to report the size that objects
appear to have, independent of the subjects’ beliefs about their true size. Under
“objective” instructions, they are to infer or guess the true size of the object.18
The literature studying size constancy under these various instructions is vast.
I focus on some recent studies by Carl Granrud, which are particularly effective at
teasing apart apparent and objective attitudes by studying them developmentally.
Granrud (2004, 2009, 2012) examined differences in the response to objective
48 Gary Hatfield
and apparent instructions in children and adults. It had been an established
finding that children show less size constancy than adults, up to about age 10.
Granrud wanted to test whether these findings reflected a change in perceptual
experience—the “perceptual learning” theory—or a change in cognitive factors
(knowledge and belief)—the “metacognitive” theory.
Granrud established that children’s beliefs about the effects of distance on
the apparent sizes of things follow a bimodal distribution. Some children as
young as 6 years old are aware that things look smaller in the distance, and the
proportion of children who know this increases from age 6 to age 11. He called
such children “strategy users,” on the assumption that they use their knowledge
in making size estimates of objects at a distance. Those who did not show such
knowledge were called “strategy nonusers.”
Granrud tested adults and children at various ages on a size constancy task
with standards at 16 to 20 feet (near distance) and about 200 feet (far distance).
At the near distances, he found slight underconstancy (slightly diminished size
judgments) for all groups under both apparent and objective instructions (that
is, both groups matched the target objects with nearby comparison objects that
were about 10% smaller than the target). At the far distance, he found important
differences between strategy-nonusing children, on the one hand, and strategyusing children together with adults, on the other. Strategy nonusers showed
significant underconstancy at the far distance for both types of instruction
(more than at the near distance). Also at the far distance, strategy users and
adults showed underconstancy with apparent instructions (they chose matches
that were an average of 20% smaller, which shows greater diminution than at
the near distance), and responses ranging from constancy to overconstancy with
objective instructions. Granrud reasoned, no doubt in part based on his own
phenomenal experience, that the objects did not actually look larger than their
physical size at the far distance (as in the response under objective instructions).
Hence, he concluded that, for all groups, the far objects looked phenomenally
smaller at the greater distance, but that, when given objective instructions, strategy
users compensated for this fact in their responses, using judgment or guesses
to provide a value for objective size. Strategy nonusers as well as strategy users
under apparent instructions responded to apparent (phenomenal) size. Granrud
concluded that the increased size constancy for adults and strategy users was a
result of a cognitive strategy, supporting the metacognitive theory.
Granrud’s findings are important in the present context, because they confirm
that phenomenal underconstancy is a fact of life, even under full-cue conditions
of observation (two eyes, with head-movements not restrained). They suggest
that reports of full constancy at moderate to far distances results from cognitive
strategy use, rather than from full phenomenal constancy. Thus, psychological
apparent size decreases with distance, measurably at 16 feet and more markedly so
Philosophy of Perception and the Phenomenology of Visual Space 49
at 200 feet. At the same time, with development perceivers become aware of the
fact that objects that appear small in the distance can be the same size as objects
nearby, and they are able with some accuracy to judge or guess the physical size
of distant objects.
The fact of phenomenal diminution paired with cognitive awareness of
physical properties could be known from the simple, everyday observation that
roads and sidewalks exhibit phenomenal convergence and yet we don’t believe
that the road actually narrows. (If the road didn’t phenomenally converge, we
would recognize it as a road that is getting physically wider!)19 This phenomenal
convergence starts from within a yard or so of the perceiver. The normality of
phenomenal diminution in the distance could be known, but hasn’t been widely
recognized, because of a tendency, even by perceptual scientists, to assimilate
the phenomenal convergence to linear perspective (e.g., Rock 1975, 38). If the
assimilation is made, then the diminished apparent size of the sidewalk or hallway
would be accounted for by optical apparent size, or Noë’s perspectival size. It
would not be normal, but the product of a special attitude of noticing or the
accessing of an early, “proximal mode” (retinal-image-like) stage of perception
(Rock 1977, 339–49; 1983, 254–65).
However, this assimilation is in error. Although the geometry of the converging
road is related to linear perspective, it is not equivalent. Indeed, the spatial structure
(in visual space) of the diminishing road is in some ways closer to the physical
value (to size constancy) than it is to the structure exhibited when the road is
projected onto a plane, in linear perspective. The phenomenal convergence is
pervasively present, but is not the same as a linear perspective projection. Let us
see how this works.
Figure 2 shows the geometry of looking down a hallway while standing in the
middle of it at point P and fixating point J.20 The lines radiating from point P are
a representative sample of the lines of sight mentioned in the introduction to this
section, under proposition (1), that “we see things in straight lines according to
their direction from us.” Accordingly, the lines of sight indicate both the physical
direction from the perceiver of points C–J and the direction in which those
points appear to an observer at P. (The figure represents a perceiver’s situation
by taking a God’s eye view from above.)
Proposition (2) above states that “we see things at a distance and in the
distance.” The proposition doesn’t say whether we perceive the distance accurately.
If the distance is accurately registered by the perceptual system,21 then we would
see points C–J in their physically veridical positions. Accordingly, the hallway and
the railway tracks would appear to be parallel; there would be no phenomenal
convergence. This would be full phenomenal constancy; the appearance would
match the physical reality. Our normal experience, however, is of a hallway with
phenomenally converging sides. One instance of convergence is shown by the
50 Gary Hatfield
Figure 2. Phenomenal visual space as a transformation
(contraction) with respect to physical space. A top
view of a perceiver P looking down physical hallway
AGHB, with physically parallel sides. The view shows
the lines of sight in a plane at eye level. With full
phenomenal constancy, the perceiver would experience
the walls AG and BH in their physical location; there
would be no phenomenal convergence. A contracted
phenomenal space with phenomenal convergence is
represented by the dotted lines.
dotted lines in Figure 2. Assuming the size–distance invariance relation obtains,
then the convergence must be produced by an undervaluing of the distance to
the points C–H. This is consistent with proposition (2), which states only that
things are seen at a distance, but doesn’t require that the distance be accurately
registered or phenomenally perceived.
I have claimed that the phenomenal convergence exhibited by the hallway is
not the same as a perspective convergence. This follows immediately from the fact
that a perspective drawing is in two dimensions, and if we perceive it accurately
from a distance of one yard, we recognize the lines in the drawing to be one yard
from us. Yet the hallway phenomenally recedes in distance. The figure represents
Philosophy of Perception and the Phenomenology of Visual Space 51
Figure 3. Hallway with picture, rug,
and doorway. In a two-dimensional
perspective projection, a right angle
formed by the floor and the upright
doorjamb may project as an acute
angle. In this figure, the angle between
the doorjamb and the line formed by
the juncture of the floor and wall is
36 degrees. Phenomenally, the angle
formed in depth when perceivers
observe a real hallway and experience
it in three dimensions is between
80 and 85 degrees (for a given
viewpoint). The hallway as normally
perceived runs away in depth, unlike
a perspective image, which takes the
three-dimensional structure into a
two-dimensional plane.
the fact that points Gɸ and Hɸ appear to the observer at P as being much further
away than do points Cɸ and Dɸ.
By considering Figure 3, we can further explore the relationship between a
perspective image (Noë’s occlusion size) and the actual phenomenal experience
of viewing a hallway. The figure depicts a hallway as drawn in linear perspective.
In the drawing, the lines formed where the ceiling meets the wall and where the
wall meets the floor yield an experience in some ways similar to the convergence
of the hallway. Because such drawings elicit a depth response, they are partial
simulations of the experience of an actual hallway.22 But the fact that the drawing
also exists on the surface of the page and can be measured provides a convenient
basis for exploring the relationship between a three-dimensional phenomenally
converging hallway and Noë’s occlusion size (exhibited by the physical perspective
rendering).
The phenomenal values of angles perceived in a real hallway can be used
to demonstrate the difference between 3D convergence and a 2D drawing. The
observation can be made by standing in a hallway with two rulers or other straight
items in hand. Position yourself in the center of the hallway, about a yard back
52 Gary Hatfield
from the edge of a doorway (so, somewhere in the nearer panel of the rug shown
in Fig. 3). Hold one ruler so it is phenomenally aligned with the near vertical
edge of the doorway. It will be held about parallel to the doorframe (hence,
about 90 degrees in relation to the physical floor). Hold the second ruler so it
touches the bottom of the first ruler (it can cross it by an inch or so) and extends
away from you. Now adjust that ruler so that it phenomenally matches the rise
of the floor or the converging line formed by wall and floor. Measure the angle
formed by the two rulers. For a hallway a couple of yards wide, it should be in
the neighborhood of 80 to 85 degrees. Now put a protractor on Figure 3. The
corresponding angle measures about 36 or 37 degrees.
These very different values demonstrate that the geometry of the 3D hallway
as experienced to be converging is very different from that of a 2D perspective
rendering (in this context, 80 is very different from 37). Indeed, the value of 80
to 85 degrees is much closer to the physical value of 90 degrees for the angle
between doorframe and floor than it is to the perspective drawing.
Nonetheless, the 3D converging space is in a family of projective geometric
structures created along the lines of sight, as in Figure 4. This figure adds another
contracted or converging 3D phenomenal structure to that in Figure 2, illustrating
different convergences when distance is undervalued by different amounts. It
also shows how the lines of sight would project onto a perspectival picture plane
at KL. The family relation between 2D and 3D projected structures is manifest:
they are all projections along the lines of sight in which straight lines remain
straight and verticals remain vertical (for further details on the geometry, see
Hatfield 2012).
Figure 4 illustrates that proposition (3), that “the laws of optics require that
things physically farther away should appear smaller,” is false. That is, things
do not appear smaller in the distance simply in virtue of the fact that they are
more distant and so project a smaller retinal image (or subtend a smaller visual
angle). Given that we see things at a distance in straight lines and that the size–
distance invariance relation holds, it is left open whether things appear smaller
in the distance (and so is not required). Full phenomenal constancy—that things
in the distance would be phenomenally of the same size as when close at hand—is
logically and geometrically possible. Psychologically, it doesn’t hold. In the normal
case, we phenomenally experience sidewalks and hallways as converging. It is an
empirical matter to determine the exact extent of underconstancy or contraction
at various physical distances. Granrud found a 10 percent diminution at 16 to 20
feet and a 20 percent diminution at 200 feet, under apparent instructions. Other
techniques have found a more radical contraction, with values above 50 percent
even at moderate distances (Wagner 2005, ch. 7).
For the purposes of my argument, I need to have established the fact that
phenomenal diminution with distance is normal. For it shows that both naïve
direct realism as articulated by Noë and the “content view” cannot be maintained.
Philosophy of Perception and the Phenomenology of Visual Space 53
Figure 4. Several contracted spaces in
relation to physical hallway AGHB, shown
in a plane through the eyes of observer P. Two
3D to 3D contracted spaces are shown with
differing degrees of phenomenal convergence,
as well as a slice of a 3D to 2D perspective
transformation projected onto plane KL
(which is perpendicular to plane AGHB).
Implications for Direct Realism
Naïve direct realism and the “content view” share the notion that the content
of perceptual experience, at least in the veridical case, can be explained by the
physical properties of objects or physical relations between the perceiver and
the object. These positions could not rely on physical facts about objects alone,
because of phenomenal facts such as that things look smaller in the distance. So
they appeal to the physical relation between the perceiver and the object. They
need the content to be explained by a physical (not mind- or subject-dependent)
relation in order to make good on their respective claims that there are no
54 Gary Hatfield
subjective aspects to perceptual experience.
In the case of things looking smaller in the distance, the relation in question
would be the fact that, with increasing distance from the perceiver, an object
that retains the same size subtends an ever decreasing visual angle. As mentioned
in the previous section, by taking a special attitude we can become aware of
visual angle as a matter of visual field “extensity” (though we are not very good
at it). However, the case I have focused on is normal perception of things at a
distance, as this is the paradigm case for the argument. For the normal case,
visual angle alone does not determine phenomenal or apparent size; only visual
angle plus distance. If the distance were always accurately registered or perceived,
then the object would appear with its true physical size. But things normally
look smaller in the distance, as converging sidewalks attest. The smaller size of
objects at a distance is a function of a systematic undervaluing of distance by
the visual system.23 In accordance with the size–distance invariance relation, if
the distance registered by the visual system is systematically smaller than the true
physical distance of objects, then the objects will appear phenomenally smaller
at a distance, and will appear even smaller the greater the physical distance. The
amount of undervaluing must be determined empirically. It is a subjective feature
of visual experience. It is in no way required by the geometry of projection or
by the physical relationship between objects and the eye. Hence, the claim of
these two theories to account for the phenomenal experience of normal cases of
perception by appealing only to physical properties and mathematical relations
(or their veridical representation) must be rejected.24
At this juncture, it is open to the “content view” to treat normal cases of
perception, such as the converging sidewalk, as illusions. Because the “content
view” endorses perceptual representation, it can suggest that the undervaluing
of distance by the visual system is a matter of misrepresentation. However, this is
not a happy outcome for the content view. It wants to explain visual experience
by the representation of physical properties, but it now ends up appealing to
their pervasive misrepresentation, defeating its main purpose. Moreover, it has
no context for theorizing why such misrepresentation should be so pervasive.
Smith’s direct realism is closer in spirit to critical direct realism, but it, too,
runs afoul of the arguments in the previous section, at least in part. In order to
explain how perception can both have a mind-dependent phenomenal aspect and
be object-presenting, Smith (2002, ch. 6) appeals to the perceptual constancies.
I have mentioned his claim that the penny viewed from an angle does not look
elliptical but rather appears as a tilted circle. This claim allows him to grant that
the penny looks different when seen from different angles (its tilt is different, it
also produces different “proximal mode” sensations), while also claiming that
perception presents the penny as a circle and so is object-presenting. (It is objectpresenting because it tracks an object-property, circularity, rather than subjective
Philosophy of Perception and the Phenomenology of Visual Space 55
sensations, such as elliptical appearances). In his view, the perception of a circle
at a slant is the direct perception of the penny (as a shaped object).
Smith implies that both size and shape constancy are “typical” of perception. I
am leaving aside the shape case, but with respect to size, I have argued that perfect
phenomenal constancy is not typical. Although phenomenal size perception is
closer to constancy than it is to projective values, underconstancy is a pervasive
phenomenal fact. Hence, Smith’s appeal to constancy per se in order to establish
that perception is object-presenting does not work.
Nonetheless, from the standpoint of critical direct realism, Smith is on
the right track. Phenomenal experience does present object-properties and
perceptually segregates things that are objects, making them available for cognitive
affirmation as individuals. Critical direct realism handles perceptual relativity
not by positing intermediate objects that have the varying properties, nor by
attempting to account for the variation wholly in terms of the changing physical
relations between things and perceivers, but by positing appearances that are
object-presenting.25 These appearances needn’t copy the object (as with Smith’s
allegedly perfect phenomenal constancy) in order to be presentations of objects
and their properties. Rather, they present object properties in a functionally
adequate manner, that is, in a manner that can guide action and serve to make
individual objects and their properties perceptually available.
In the case of objects at a distance, the contracted spatial appearances
described in the previous section present the objects accurately in several respects.
Directions from the perceiver to all visible portions of the object are veridical
(within very close tolerances). The relations among the visible parts of objects are
veridical. The proportionate relations among the parts of the object and between
the object and other things at the same distance are preserved; this is known
as proportional or relative size constancy (Palmer 1999, 319). Up to some point
(perhaps somewhere near the horizon), things appear at a phenomenal distance
that is proportionate to physical distance, so that if one thing is nearer than
another, it appears so (within tolerances). Metacognitive skill of the kind Granrud
tracked developmentally allows individual perceivers to recognize cognitively the
constant physical size of objects that look smaller with increasing distance. Figure/
ground organization tends to segregate things that are objects as perceptual units.
These segregated volumes can then serve the purposes of object recognition and
identification (see Hatfield 2009, ch. 7). Nonetheless, the appearances, even in
the case of primary qualities, bear a subjective and mind-dependent aspect.26
As regards their spatial relations, objects are not only presented from a point of
view, but also with a contracted spatiality that is peculiar to perceivers.
Some questions do remain for the critical direct realist. What are these
appearances? If they are not literally brown or rectangular, how is it that they
phenomenally present such properties? The critical direct realist does not posit
56 Gary Hatfield
a relation of acquaintance between the perceiver and the object of perception
and does not require that perceived properties such as color, shape, and size
be literally present in the appearances. In this way, critical direct realism
differs from naïve realism.27 But it does require that phenomenal brown and
phenomenal size and shape be present in consciousness as phenomenal qualities
that present object properties. As with the “content view,” the represented
content accounts for phenomenal experience. But it does so by using subjective
qualitative phenomenality to represent external properties. For size and shape,
it does so by presenting subjectively qualified sizes and shapes as phenomenal
content. To say more about the status of these qualitative states of experience is
desirable. They have the status of Brentano (1874/1995, 88) intentionality: they
are phenomenally characterized intentional contents.28 They make phenomenal
brownness and phenomenal shape present to consciousness. In so doing, they
represent physically brown and physically shaped things in the environment—not
by copying them, but by presenting them in a subjectively phenomenal way that
allows for successful action and cognition.
One might also ask how it is that we see things directly by means of a mental
content. Here the critical direct realist joins with some adherents of the “content
view” to endorse nonconceptual content as the initial vehicle of perception. This
content, however, has a subjective aspect and is introspectible. The phenomenal
appearance of a trapezoidal brown expanse just is the phenomenal presentation
of the table top. The subject can become aware of the trapezoidal shape of the
top as it runs away in the distance (the table top is trapezoidal in 3D), and this is
a subjective feature. But such features are object-presenting. Again, it is desirable
to fill out this account, but that must await another occasion.
For the nonce, I have offered some reflections on the phenomenology of visual
spatial perception that extend and articulate the notion of perceptual relativity.
Things appear small in the distance not simply as a function of the mathematical
laws of perspective, but in virtue of subject-dependent undervaluing of registered
or perceived distance in the visual system. Naïve realism and the “content view”
can’t handle these newly prominent facts. Critical direct realism can account for
these facts of perceptual relativity, while offering a realism that is direct enough.
University of Pennsylvania
Philosophy of Perception and the Phenomenology of Visual Space 57
Notes
I am grateful to audiences at SUNY Brockport and at the University of
Pennsylvania for questions and comments on the arguments of the paper, and
to Georges Dicker and Alistair Isaac for helpful comments on the penultimate
draft.
1
E.g., Noë (2005), 236. On this point, see also Smith (2002), 3.
2
In addition to indirect or representative realism, Moore considered a position
that equated sense data with directly given portions of the surfaces of things
(a direct realism), but he also held that, with such a position, he would need
to violate a tenet he wished to preserve: that sense data are perceived as they
are. For, if a round penny is seen at a slant and appears elliptical, then the
actually circular surface of the penny, which is present in perception as a
sense datum, must nonetheless appear elliptical (Moore 1918–19, 24). On the
notion of sense data in the first decades of the twentieth century, see Hatfield
(2013).
3
Naïve realists often protest against the term, as they do not wish to be labeled
“naïve” (see, e.g., Alexander 1909–10, 2). Noë (2005), 258, characterizes both
himself and Campbell (2002) as “naïve realists.” Campbell (2002, 117–18)
uses that label for his position as well, though he more commonly calls it a
“relational view,” to highlight that (successful) perception consists in a direct
relation between conscious experience and the surrounding environment.
Cambell (ibid.) also aligns his view with disjunctivism (that there is no
common content between veridical perception and hallucination).
4
I do not further discuss conceptual versions of the content theory, such as
McDowell (1996), which hold that phenomenal experience is constituted
of belief-like contents. In the extreme, this sort of approach eliminates
phenomenal experience (Dennett 1988).
5
In this way, adherents of the “content” view distinguish themselves from
representative or indirect realism. They hold that the phenomenal brown
or phenomenal ellipse are not present as properties of a directly experienced
item (whether mental, third thing, or external); rather, perceptual states
represent physical properties of things (whether veridically or not), and that
represented content (somehow) constitutes the phenomenal experience of a
color or shape. One may wonder how this is supposed to work, but I do not
focus on this potential problem for the content view.
58 Gary Hatfield
6
Dretske (1995) holds that our awareness of the properties is abstract, that is,
we are aware of a particular type of color quality and shape, but that content
does not represent a singular object (see also Tye 1995, ch. 5). Object reference
arises through the object’s relation (whether causal or informational) to our
sensory system (Dretske 1995, ch. 1.4).
7
I raise problems for the pure information or content view of color in Hatfield
(2009), chs. 8–9. For a representative sample of various positions on color
perception and color ontology, see Mausfeld and Heyer (2003).
8
An appearance of a red and round area in front of an observer nonconceptually
represents (imagistically or depictively) the redness and roundness of an
object; conceptual representation occurs when the perceiver (or visual system)
classifies the object as red and round.
9
The type of perceptual realism described here has gone under several names,
including “critical realism” as in R. W. Sellars (1916) and Drake (1920). It
has been described as a “critical realism” that is “direct” (R. W. Sellars 1961,
6) and as a “direct realism” that is “critical” (W. Sellars 1963, 90); these can
be joined so as to name a “critical direct realism” (Hatfield 2009, chs. 1,
11; Levine 2007). Mandelbaum (1964), Essay 4, develops a “radical critical
realism” that shares some of the outlook of critical direct realism, in that it
draws closely on physical and perceptual science; nonetheless, for my tastes,
in the end his position makes the objects of perception too far removed from
their presentations in experience for it to be a type of direct realism. More
generally, the reader may find similarity between R. W. Sellars’s critical direct
realism and the adverbial position of Chisholm (1957). Chisholm (1957, 157)
prominently cites Sellars, and he previously had written an article (Chisholm
1954) endorsing the main tenets of Sellars’s analysis of perception (while
withholding assent to Sellars’s version of materialism). Many of the points
made in favor of Sellars’s position also hold of an adverbial position.
10
Smith (2002, 69 n. 14) mentions “Critical Realism” in a footnote, and
subsequently ascribes the position to W. Sellars and Reid. He does not
reference the self-avowed critical realists of Drake (1920).
11
Although some authors equate disjunctivism with versions of naïve realism
(e.g., Pautz 2010), there are disjunctivisms built on the “content view” (e.g.,
Byrne and Logue 2008).
12
The nonconceptualist versions of the content view must rely on something
akin to Dretske’s (1995) notion of (natural, physical) information in order
to explain how nonconceptual representations can have sophisticated
informational content. They must of course hold that phenomenal color just
is the representation of a physics-based color property in objects.
Philosophy of Perception and the Phenomenology of Visual Space 59
13
Sellars (1920, 194) describes the directness of perceptual knowledge as follows:
“While knowledge is mediate both in the sense that it is not intuition [i.e.,
acquaintance] and in the sense that there is much constructive activity at work
in the mind, yet it is direct. We mean independent objects and we interpret
these objects in terms of ideas. The fact that we can dwell upon ideas for their
own sake should not be allowed to confuse us with respect to the knowledgeclaim.”
14
Sellars (1920, 195) distinguishes affirming from inferring: “the critical
realist holds that we must distinguish between the givenness of content and
knowledge of the physical thing, and that we do not infer a realm of existents
co-real with ourselves, but instead affirm it through the very pressure and
suggestion of our experience.” By “given” he means phenomenally available,
not as objects of knowledge or incorrigible foundations, but as aspects of
phenomenal experience. Also, note that Sellars’s analysis, like Dretske’s
(1995, ch. 1.4), distinguishes perceived properties or features from, in his
case, affirmed or “meant” individuals. He differs from Dretske in offering
a phenomenal characterization of property-representations and a seemingly
deictic account of affirming individuals. For a discussion of these issues and
development of a position similar to that of Sellars, see Hatfield (2009), ch. 7.
15
Brewer’s (2011, 96) appeal to sense modality is intended to account for some
aspects of, e.g., visual phenomenology, but it can’t be used to introduce
aspects of perceptual content that are not reducible to physical properties of
external objects, including their geometrical relations to the eye.
16
Most philosophers and vision scientists agree with this phenomenal claim,
but not all do. I defend it in Hatfield (2009), ch. 6. It receives an implicit
defense below, in my discussions of how things such as roads and hallways
look.
17
Wagner (2006, ch. 6; 2012) reviews a century of research on size constancy
and the size–distance invariance relation (or “hypothesis”), and finds that, for
unfamiliar stimuli, it is well-supported by the data.
18
For review and critical discussion of these various instructional sets, see
Carlson (1977), Hatfield (2012), Hatfield and Epstein (2012), and Wagner
(2012).
19
In an important set of studies conducted more than one hundred years ago,
Hillebrand (1902) and Blumenfeld (1913) studied the physical structure
necessary to yield phenomenally parallel lines receding from the observer at
near distance, on or above a table top. They found that lines or walls must
diverge in order to appear parallel.
60 Gary Hatfield
20
Fixation of J provides a reference point for the geometry of the sides. The
geometry changes only slightly if the perceiver’s eyes move to other points,
such as E, G, H, or F. The conditions of perception are “full cue”; the
perceiver’s head is unrestrained, and he or she is using both eyes and so has
good depth perception. Further, although this and the subsequent figures
depict a rectilinear space, the contraction can also be observed in open space,
such as a field or meadow. Whether the geometry of the contraction is the
same in all instances is an empirical matter.
21
On the notion of “registered” vs. phenomenally accessible information in the
visual system, see Epstein (1973), 276–7, and Hatfield (2009, ch. 2), 66.
22
Drawings and paintings in linear perspective, including line drawings, yield
a reliable depth response (Smith, Smith, and Hubbard 1958; Rogers 1995).
Derksen (2005) argues that drawings and paintings in linear perspective
are “faithful” to the scene to the extent that they achieve a response of full
phenomenal constancy, and that they aren’t fully faithful because they yield
underconstancy. According to my argument, such drawings could be faithful
to visual experience if they achieved a partial constancy response, which includes
roads that converge as they run away in depth.
23
The “registered” distance is undervalued. This value is subpersonal, and
presumably enters into nonconscious processes of perception that combine
the registered value for distance with a registered visual angle to yield a
phenomenally perceived distance that is available to consciousness.
24
In addition to Noë’s position, this account tells against Schellenberg’s (2008)
claim that, for two trees of the same size at different distances, the smaller
apparent size of the more distant tree can be accounted for entirely by mindindependent situational properties. Schellenberg seems to assume that things
subtending a smaller visual angle must appear smaller, which is false. (She
rejects an equation of her situationally dependent smaller tree with visual
field size or extensity.)
25
Or at least object-property presenting. See note 14.
26
Adherents of naïve realism and the content view often assume that if some
feature of phenomenal experience is subjective, it therefore is not and cannot
be object presenting (e.g., Schellenberg 2008). Smith (2002, chs. 5–6) pushes
back against this attitude by developing the notion of perceptual sensations. I
have argued (Hatfield 2009, ch. 9) that subject-dependent aspects of color
perception can nonetheless be objective (fact-supporting), and have developed
a notion of functional veridicality for spatial perception that does not require
strict congruence between phenomenal spatial structures and the object
spatial structures that they represent (Hatfield 2012, 57–60).
Philosophy of Perception and the Phenomenology of Visual Space 61
27
And accords with an adverbial view, see note 9.
28
Sellars (1920, 190) writes that “in the act of knowledge, the idea which
gives the content of the knowledge (the esse intentionale of the scholastics)
is other than the object of knowledge. In what sense it is ‘other than’ the
object affirmed is obviously one of our problems.” Such intentional content
phenomenally presents properties that may or may not be externally present.
A full accounting of phenomenal intentional content would require a solution
to the mind–body or phenomenality–brain problem, something that none of
the positions herein can uncontroversially claim to have achieved. For further
discussion, see Hatfield (2009), chs. 10–11.
62 Gary Hatfield
References
Alexander, Samuel (1909–10). “On Sensations and Images,” Proceedings of the
Aristotelian Society 10: 1–35.
Armstrong, David M. (1961). Perception and the Physical World. New York: Humanities
Press.
Blumenfeld,Walter (1913). “Untersuchungen über die scheinbare Grösse im
Sehraume,” Zeitschrift für Psychologie und Physiologie der Sinnesorgane 65: 241–404.
Brentano, Franz Clemens (1874). Psychologie vom empirischen Standpunkt. Leipzig:
Duncker and Humblot.
—— (1995). Psychology from an Empirical Standpoint, trans. Antos C. Rancurello, D. B.
Terrell, and Linda L. McAlister. London: Routledge.
Brewer, Bill (2011). Perception and Its Objects. Oxford: Oxford University Press.
Byrne, Alex, and Heather Logue (2008). “Either / Or,” in A. Haddock and F.
Macpherson (eds.), Disjunctivism: Perception, Action, and Knowledge, 57–94. Oxford:
Oxford University Press.
Carlson, V. R. (1977). “Instructions and Perceptual Constancy Judgments,” in
William Epstein (ed.), Stability and Constancy in Visual Perception: Mechanisms and
Processes, 217–54. New York: Wiley.
Campbell, John (1993). “A Simple View of Colour,” in John Haldane and Crispin
Wright (eds.), Reality, Representation, and Projection, 257–68. New York: Oxford
University Press.
—— (2002). Reference and Consciousness. Oxford: Clarendon Press.
Chisholm, Roderick M. (1954). “Sellars’ Critical Realism,” Philosophy and
Phenomenological Research 15: 33–47.
—— (1957). Perceiving: A Philosophical Study. Ithaca: Cornell University Press.
Cook Wilson, John (1926). “Letter in Criticism of a Paper on Primary and
Secondary Qualities,” in Statement and Inference, with Other Philosophical Papers, 2
vols., pp. 764–800. Oxford: Clarendon Press.
Philosophy of Perception and the Phenomenology of Visual Space 63
Crane, Tim (1992). “Introduction,” in Tim Crane (ed.), The Contents of Experience:
Essays on Perception, 1–17. Cambridge: Cambridge University Press.
Dennett, Daniel C. (1988). “Quining Qualia,” in Anthony J. Marcel and Edoardo
Bisiach (eds.), Consciousness in Contemporary Science, 42–77. Oxford: Oxford
University Press.
Derksen, Anthony A. (2005). “Linear Perspective as a Realist Constraint,” Journal
of Philosophy 102: 235–58
Descartes, René (1637). Discours de la methode pour bien conduire sa raison et chercher
la verité dans les sciences: plus la dioptrique, les meteores, et la geometrie, qui sont des essais
de cete methode. Leiden: Maire.
—— (1965). Discourse on Method, Optics, Geometry, and Meteorology, trans. Paul J.
Olscamp. New York: Bobbs-Merrill.
Dicker, Georges (1980). Perceptual Knowledge: An Analytical Historical Study.
Dordrecht: Reidel.
Drake, Durant (ed.) (1920). Essays in Critical Realism. London: Macmillan.
Epstein, William (1973). “The Process of 'Taking into Account' in Visual
Perception,” Perception 11: 75–83.
Dretske, Fred (1995). Naturalizing the Mind. Cambridge: MIT Press.
Gibson, James J. (1950). The Perception of the Visual World. Boston: Houghton
Mifflin.
Granrud, Carl E. (2004). “Visual Metacognition and the Development of Size
Constancy,” in Daniel T. Levin (ed.), Thinking and Seeing: Visual Metacognition in
Adults and Children, 75–95. Cambridge: MIT Press.
—— (2009). “Development of Size Constancy in Children: A Test of the
Metacognitive Theory,” Attention, Perception, and Psychophysics 71: 644–54.
—— (2012). “Judging the Size of a Distant Object: Strategy Use by Children and
Adults,” in Gary Hatfield and Sarah Allred (eds.), Visual Experience: Sensation,
Cognition, and Constancy, 13–34. Oxford: Oxford University Press.
64 Gary Hatfield
Harman, Gilbert (1990). “The Intrinsic Quality of Experience,” in James E.
Tomberlin (ed.), Philosophical Perspectives, vol. 4, Action Theory and Philosophy of
Mind, 31–52. Atascadero, Calif.: Ridgeview.
Hatfield, Gary (2009). Perception and Cognition: Essays in the Philosophy of Psychology.
Oxford: Clarendon Press.
—— (2012). “Phenomenal and Cognitive Factors in Spatial Perception,” in Gary
Hatfield and Sarah Allred (eds.), Visual Experience: Sensation, Cognition, and
Constancy, 35–62. Oxford: Oxford University Press.
—— (2013). “Perception and Sense Data,” forthcoming in Oxford Handbook of the
History of Analytical Philosophy, ed. M. Beaney. Oxford: Oxford University Press.
Hatfield, Gary, and William Epstein (2012). “Epilogue: Advances and Open
Questions,” in Gary Hatfield and Sarah Allred (eds.), Visual Experience: Sensation,
Cognition, and Constancy, 232–41. Oxford: Oxford University Press.
Hillebrand, Franz (1902). “Theorie der scheinbaren Grösse beim binokularen
Sehen,” Denkschrift der Kaiserlichen Akademie der Wissenschaften Wien, MathematischNaturwissenschaftliche Classe 72: 255–307.
Ibn al-Haytham (1989). The Optics of Ibn al-Haytham: Books I–III, On Direct Vision,
Vol. 1, Translation, ed. and trans. Abdelhamid I. Sabra. London: Warburg Institute.
(Originally written ca.1030.)
Joynson, Robert B. (1958a). “An Experimental Synthesis of the Associationist and
Gestalt Accounts of the Perception of Size: Part I,” Quarterly Journal of Experimental
Psychology 10: 65–76.
—— (1958b). “An Experimental Synthesis of the Associationist and Gestalt Accounts
of the Perception of Size: Part II,” Quarterly Journal of Experimental Psychology 10:
142–54.
Levine, Steven M. (2007). “Sellars’ Critical Direct Realism,” International Journal of
Philosophical Studies 15: 53–76.
McDowell, John (1996). Mind and World. Cambridge: Harvard University Press.
Mandelbaum, Maurice (1964). Philosophy, Science, and Sense Perception: Historical and
Critical Studies. Baltimore: Johns Hopkins Press.
Philosophy of Perception and the Phenomenology of Visual Space 65
Mausfeld, Rainer and Dieter Heyer (eds.) (2003). Colour Perception: Mind and the
Physical World. Oxford: Oxford University Press.
Moore, G. E. (1909–10). “The Subject-Matter of Psychology,” Proceedings of the
Aristotelian Society 10: 36–62.
—— (1918–19). “Some Judgments of Perception,” Proceedings of the Aristotelian Society
19: 1–29.
Noë, Alva (2004). Action in Perception. Cambridge: MIT Press.
—— (2005). “Real Presence,” Philosophical Topics 33: 235–64.
Palmer, Stephen E. (1999). Vision Science: Photons to Phenomenology. Cambridge:
MIT Press.
Pautz, Adam (2010). “Why Explain Visual Experience in Terms of Content?” in
Bence Nanay (ed.), Perceiving the World, 254–309. New York: Oxford University
Press.
Price, H. H. (1932). Perception. London: Methuen.
Prichard, H. A. (1950). “Perception,” in Knowledge and Perception: Essays and Lectures,
52–68. Oxford: Clarendon Press.
Rock, Irvin (1975). Introduction to Perception. New York: Macmillan.
—— (1977). “In Defense of Unconscious Inference,” in William Epstein (ed.),
Stability and Constancy in Visual Perception: Mechanisms and Processes, 321–73. New
York: John Wiley.
—— (1983). The Logic of Perception. Cambridge: MIT Press.
Rogers, Sheena (1995). “Perceiving Pictorial Space,” in William Epstein and
Sheena Rogers (eds.), Perception of Space and Motion, 119–63. San Diego: Academic
Press.
Ross, Helen E., and Cornelis Plug (1998). “The History of Size Constancy and Size
Illusions,” in Vincent Walsh and Janusz Kulikowski (eds.), Perceptual Constancy:
Why Things Look as They Do, 499–528. Cambridge: Cambridge University Press.
66 Gary Hatfield
Russell, Bertrand (1912). Problems of Philosophy. London: Williams and Norgate.
Schellenberg, Susanna (2008). “The Situation-Dependency of Perception,” Journal
of Philosophy 105: 55–84.
Sellars, Roy Wood (1916). Critical Realism. Chicago: Rand McNally.
—— (1920). “Knowledge and Its Categories,” in Durant Drake (ed.), Essays in Critical
Realism, 187–219. London: Macmillan.
—— (1961). “Referential Transcendence,” Philosophy and Phenomenological Research
22: 1–15.
—— (1969). Reflections on American Philosophy from Within. Notre Dame: University
of Notre Dame Press.
Sellars, Wilfrid (1963). “Phenomenalism,” in Science, Perception and Reality, 60–105.
London: Routledge and Kegan Paul.
Smith, A. D. (2002). The Problem of Perception. Cambridge: Harvard University
Press.
Smith, Olin W., Patricia Cain Smith, and Dorothy Hubbard (1958). “Perceived
Distance as a Function of the Method of Representing Perspective,” American
Journal of Psychology 71: 662–74.
Tye, Michael (1995). Ten Problems of Consciousness: A Representational Theory of the
Phenomenal Mind. Cambridge: MIT Press.
—— (2002). “Representationalism and the Transparency of Experience,” Noûs 36:
137–51.
Wagner, Mark (2006). Geometries of Visual Space. Mahwah, N.J.: Erlbaum.
—— (2012). “Sensory and Cognitive Explanations for a Century of Size Constancy
Research,” in Gary Hatfield and Sarah Allred (eds.), Visual Experience: Sensation,
Cognition, and Constancy, 63–86. Oxford: Oxford University Press.