The Philosophy of Perception

Proceedings of the 40th International Ludwig Wittgenstein Symposium

Christoph Limbeck-Lilienau, Friedrich Stadler, Christoph Limbeck-Lilienau, Friedrich Stadler

• Publisher: De Gruyter
• Year: 2019
• ISBN:
  9783110654462
• Edition: 1
• Topic: Philosophy
• Subtopic: Modern Philosophy

In this volume the philosophy of perception and observation is discussed by leading philosophers with implications in the philosophy of mind, in epistemology, and in philosophy of science. In the last years the philosophy of perception underwent substantial changes and new views appeared: the intentionality of perception has been contested by relational theories of perception (direct realism), a richer view of perceptual content has emerged, new theories of intentionality have been defended against naturalistic theories of representation (e. g. phenomenal intentionality). These theoretical changes reflect also new insights coming from psychological theories of perception. These changes have substantial consequences for the epistemic role of perception and for its role in scientific observation. In the present volume, leading philosophers of perception discuss these new views and show their implications in the philosophy of mind, in epistemology and in philosophy of science. A special focus is laid on Franz Brentano and Ludwig Wittgenstein. A reference volume for all scholars and students of the history, psychology and philosophy of perception, and cognitive science.

1Objectivity and
Realism

Tyler Burge

Perception: Ground of Empirical Objectivity

Abstract: Several types of objectivity are surveyed. The role of perception as a type of empirical objectivity, and as a source for more sophisticated types of objectivity that it itself does not realize, is discussed. Perceptual representation is distinguished from the sort of representation explained in terms of information theory. Perceiving is also distinguished from sensing. Some threads in the history of philosophy that have taken perception not to be a type or source of objectivity are discussed and criticized. Often inevitable limitations on the types of objectivity that perception can embody have been misconstrued as marks of subjectivity. Often perception has been mis-characterized in the interests of one or another philosophical ideology. The irony of this history is that the ultimate basis for the objectivity of the empirical sciences has commonly been miscast in the philosophical tradition.

Keywords: Perception, objectivity, perceptual constancy, representation

My topic is the objectivity of perception. I begin by surveying various kinds of objectivity. Then I discuss how perception realizes some of these kinds. Finally, I consider some limitations of perception. I try to explain why these limitations should not be taken to undermine the kinds of objectivity that perception in fact has. For the empirical sciences, perception is the ground and sine qua non of all other types of objectivity, even the types that perception cannot itself measure up to.

First, a brief survey of kinds of objectivity.3

The largest division is that between subject-matter types and representation-of-subject- matter types. The broadest, least restrictive example of a subject-matter type is the objectivity of all that is real, or all that exists, or all that has being. The contrast is with the purported, unreal “objects” of fantasy and delusion. Since such “objects” have no being, there is no genuine contrast class for the objective in this broadest sense.

Some subject-matter types of objectivity are more restrictive than the objectivity of everything that is. For example, it is common to count mind-independent aspects of the world as objective – and contrast them with mind-dependent aspects. Here and throughout, I exclude the mind of any deity from consideration. Mud, atoms, stars, and trees are mind-independent, hence objective on this categorization. Pains, beliefs, desires, and theories are non-objective on this categorization. In a sense, not all trees are mind-independent. Many are planned and planted.

To avoid this piece of awkwardness, let us take ‘mind-independence’ to mean ‘constitutive mind-independence’ – independence of mind for being the sort of thing it is. It is not a constitutive aspect of a tree that it be planned. Even with this restriction, quite a lot of things turn out to be mind-dependent, besides mental items themselves. Hammers are constitutively mind-dependent. What it is to be a hammer involves having some function or use, presumably for some being with a mind. Animals such as cats have minds constitutively. Something would not be a cat if it lacked a natural capacity for perception. On many views, colors are constitutively mind-dependent. For example, if colors are secondary qualities – dispositions in objects to produce certain sensations in a class of individuals –, then colors are constitutively mind-dependent. The constitutively-mind-independent type of subject-matter objectivity counts quite a lot of things non-objective: hammers, cats, nations, and perhaps colors. Such consequences are rarely recognized when people put forward notions of objectivity in terms of mind-independence.

A subject-matter type of objectivity that is more restrictive than the whole-real-world type and less restrictive than the constitutively-mind-independent type is the objectivity of things that are not themselves representational perspectives or states of consciousness. This type would include hammers, cats, colors, and nations as in themselves objective, but exclude beliefs, perceptions, pains, theories, and statutes.

I have cited three subject-matter types of objectivity. All are to be distinguished from representation-of-subject-matter types. Most uses of the term ‘objective’ apply to certain types of representation of a subject matter, rather than to subject matters themselves.

I divide representation-of-subject-matter types of objectivity into two large sub-classes – vertical types and horizontal types. The intuitive idea is that vertical types are characterized by a relatively direct representational relation to a subject matter. Horizontal types, by contrast, are characterized by types of representation or relations among representations.

The vertical relations to a subject matter are all aspects of or contributors to veridicality. Such relations as being true of, or accurate of, a subject matter are aspects of veridicality. Referring to a subject matter is a contributor to veridicality. Being veridical – that is, being either accurate or true – is itself a vertical relation to a subject matter, by virtue of its dependence on these sorts of relations.

Different types of subject matters differentiate among types of vertical representation-of- subject-matter objectivity. For example, being true of a real subject matter, being true of a subject matter that is constitutively mind-independent, and being true of a subject matter that consists of things that are not themselves representational perspectives or states of consciousness are different types of vertical representation-of-subject-matter objectivity.

A stricter type of vertical objectivity is representation of laws or of certain structural invariances. This type loomed large in Kant’s work and is central to modern physics.

The horizontal types of representation-of-subject-matter objectivity are a more varied lot. An example of horizontal objectivity is following a procedure that yields representations in a way that is independent of the whims of any particular individual. Following a set procedure in civil law counts as objective in this sense. Perhaps the central procedural type of objectivity in the history of philosophy, again emphasized by Kant, is representation that follows rational procedures, according to some canon or other.

A closely related family of types of objectivity comprises representation in impersonal terms, representation that is independent of first- or second-person pronouns, representation that is independent of demonstratives or indexicals, representation that is independent of individual attitudes or training. These are types of the objectivity of impersonality. Many specifications of laws have striven to be impersonal in some of these ways. The horizontal types can overlap.

A final type of horizontal objectivity is representation that accords with representation by others. Often the others are taken to follow some procedure – such as being rational or being scientific. The intersubjective versions of horizontal objectivity present straightforward contrasts with the subjectivity of idiosyncracy and the subjectivity of privacy.

Some broad points can be made about relations among the horizontal and vertical representation-of-subject-matter types of objectivity. Although many have championed inter- subjectivity as a central type of objectivity, philosophers have wisely tended to qualify intersubjectivity by placing some condition on the subjects to assure that intersubjectivity is not that of agreement among crazies. Kant took intersubjective objectivity to occur among subjects that follow rational or scientific procedures. Frege associated the communicable with law.

There is, of course, evidence that over the long run, a kind of efficiency, sometimes even group rationality, emerges from the aggregate actions of the group. These are instances of the so-called wisdom of crowds.4 I believe that if rationality, as opposed to efficiency or evolutionary success, is at issue, the individuals in a group must have basic rational competencies. Genuine rationality is not an aggregate upshot of efficient behavior of non-rational individuals.

The large point that I want to emphasize is that taking vertical representation-of-subject- matter types of objectivity to be constitutively more basic than horizontal ones is fundamental to a realist view of the world. Peirce defined truth as what rational procedures would lead to in the limit. Kant took following rational and/or scientific procedures to be what objective validity consists in – at least from what he called the transcendental point of view. These are idealist strategies. What I take to be the correct view is that rationality is to be understood partly in terms of being conducive to truth, given certain limitations of information and competence. Good scientific method should be understood partly in terms of truth or approximate truth. It must be method that can be expected to lead toward true or approximately true scientific theory. Statements of laws are often very far from being exactly and literally true, although such statements can serve many useful purposes. But representations of laws are ultimately to be judged by whether they describe, to some approximation, real lawful patterns in the world.

Of course, science must idealize. Few of our methods lead to precise truths. We are constantly finding limitations in our methods and both limitations and imprecision in our theories. But ultimately our procedures are evaluated by how well they describe reality.

Understanding reality in an illuminating way is the basic aim of science. Being veridical is the fundamental idealization that guides our conceptions of rational and scientific procedures.

So horizontal types of objectivity are to be understood as serving vertical types, given a realist attitude toward science and metaphysics. The primary norm for vertical types of objectivity is to be veridical-true or accurate. Veridicality is fidelity to subject matter. So the subject-matter types of objectivity provide the primary basis for assessing the objectivity of vertical representation-of-subject matter types. And as noted, the objectivity of horizontal representation-of-subject-matter types functions to serve vertical types. So subject-matter objectivity – essentially, what is – is in this sense fundamental.

There is a further sense in which the subject-matter types of objectivity are fundamental with respect to representation-of-subject-matter types. The contents of representations and representational states, at least for empirically based representations, are determined to be what they are through interaction with their subject matters. The subject matters of such representations are partial determiners of the natures of our representations.5 One can think, metaphorically, of the world’s stamping itself – primarily through causal interactions – into the very contents of our representations.

I turn from these vertiginously general reflections to more specific points. I focus on ways in which perception is objective and on the role of its objectivity in the objectivity of more sophisticated types of representation.

To understand the role of perception in this welter of types of objectivity, it is crucial to distinguish perceiving from non-perceptual sensing. Sensing is an extremely broad phenomenon. Plants are sensitive to light and respond to it. I say that they do not sense anything, because they do not act and sensing serves action. Still, they are sensitive to the environment. Even laying aside the distinction between sensing and being sensitive, numerous organisms sense their environment without perceiving it. Bacteria sense light and swim away from it. Rotifers sense food through their cilia. Ticks sense warmth and crawl toward it.

No science explains these instances of sensing in terms of states that have conditions for being accurate. Sensing is causally based, statistically significant interaction with the environment that has a function for the organism. Relevant functions here are biological functions and functions associated with action. Fulfilling such functions are broadly, and in many cases richly, practical successes. We can speak of the accuracy of such sensing. But we are thereby describing nothing more than fulfillment of a practical function. In fulfilling a biological function, an organism or state contributes to fitness for survival long enough to mate. In fulfilling a function of an action – in reaching the action’s target –, an organism commonly contributes to such fitness. Fulfilling such functions is not being accurate or inaccurate. Talk of these sensory states as being accurate or inaccurate is metaphorical.

An organism has perception only when it has states that are accurate or inaccurate, where having accuracy conditions is a real, non-metaphorical feature of the states’ natures. Finding that having accuracy conditions is a feature of a state’s nature is discovering that strong causal explanations appeal to states with accuracy conditions. In such cases accuracy is not merely a matter of biological functional success. Such success is constitutively a practical matter – being useful for survival long enough to reproduce, including fulfilling or contributing to fulfilling biological functions of action and reaction for the animal or animal species. Accuracy in perception or belief is not constitutively a practical matter. A perceptual state can be accurate but practically deleterious, or practically useful but inaccurate.6

The term ‘representation’ is often used, even in science, in two importantly different ways. One way applies to all sensing, perceptual or not. The other way applies to a distinctively psychological capacity. This latter is the type of representation that underlies the main types of representation-of-subject-matter objectivity, vertical and horizontal.

To understand the former type of representation, one must first understand the basic idea of Shannon information theory, a theory of statistical correlation. State X provides (Shannon) information about state Y if X and Y are statistically correlated to some relevant degree.

Providing information is a symmetrical relation.7 Clearly, providing information in this sense is not in any way equivalent to the sort of representation that we evaluate for veridicality.

Shannon information is a component element in applications of the term ‘representation’ that require meeting more conditions than providing information. These applications have been useful in understanding animal behavior. The conditions for state kind X’s informationally registering state kind Y are (a) that state kinds X and Y are statistically correlated (provide information with respect to one another) to some significant degree; (b) that instances of Y commonly cause instances of state X; and (c) th...

Table of contents

• Cover
• Title Page
• Copyright
• Content
• Editorial
• 1 Objectivity and Realism
• Perception: Ground of Empirical Objectivity
• Objectivity: How is it Possible?
• Realism’s Kick
• The Good, The Bad, and The Naïve
• 2 Content and Intentionality
• How to Think About the Representational Content of Visual Experience
• Structure, Intentionality and the Given
• Brentano on Perception and Illusion
• The Problem with J. Searle’s Idea That ‘all Seeing is Seeing-as’ (or What Wittgenstein did not Mean With the Duck-Rabbit)
• 3 Perception, Cognition and Images
• The Perception/Cognition Divide: One More Time, With Feeling
• Why Verbal Understanding is Unlikely to be an Extended Form of Perception
• Sound and Image
• 4 The Cognitive Penetrability of Perception
• Bias-Driven Attention, Cognitive Penetration and Epistemic Downgrading
• Pre-Cueing, Early Vision, and Cognitive Penetrability
• Predictions do not Entail Cognitive Penetration: “Racial” Biases in Predictive Models of Perception
• 5 Epistemology of Perception
• Boundless
• The Manifest and the Philosophical Image of Perceptual Knowledge
• The Co-Presentational Character of Perception
• Knowledge Without Observation: Body Image or Body Schema?
• 6 Perception and the Sciences
• Scheinbewegungen. Wahrnehmung zwischen Wissensgeschichte und Gegenwartskunst
• Zur Analogie von Wittgensteins Konzept des Aspektwechsels und der wissenschaftlichen Metapher als Vehikel der Innovation
• 7 Wittgenstein
• The Structure of Tractatus and the Tractatus Numbering System
• Wittgensteins Welt
• Index of Names