Indirect Realism and Sense-Data

Indirect realism holds that perception never makes direct contact with the mind-independent world. What we perceive immediately and directly are inner objects (sense-data, phenomenal intermediaries, the “veil of perception”), and the external world is known only by inference from these inner items.

The view has deep historical roots. Its lineage runs from Locke’s primary/secondary quality distinction through Berkeley’s idealism, Hume, the British Empiricists, through Russell’s early work, and through the sense-datum theorists of the early twentieth century (Moore, Price, Ayer). Its central argument, the argument from illusion, remains the most discussed argument in the philosophy of perception. Understanding indirect realism precisely matters because it is not obviously wrong; the argument for it is genuinely compelling, and the history of philosophy of perception is largely the history of attempts to answer it.

The Argument from Illusion

The engine that drives people into indirect realism. Three premises:

The perceptual variation premise: the same external object can produce different perceptual experiences in different conditions (lighting, angle, medium, state of the perceiver). A round coin looks elliptical from an angle; a white shirt looks blue in shadow; a straight stick looks bent in water.
The phenomenal object premise: in these cases, there is something that has the property the experience presents — something that genuinely is elliptical, blue, bent. This something cannot be the external object (the coin is round, the shirt is white, the stick is straight). So it must be an inner object: a sense-datum.
The parity premise: since veridical and illusory experiences can be subjectively indistinguishable, the immediate object of perception in both cases must be the same kind of thing — a sense-datum. In veridical perception, we see sense-data that accurately represent the world; in illusion, sense-data that misrepresent it.

Conclusion: we never perceive the external world directly. We perceive sense-data directly; the world only mediately, by inference.

Berkeley pushes the same lineage one step further — strip out the external world entirely, and only ideas remain.

What Sense-Data Are Supposed to Be

On the standard sense-datum theory:

Private: sense-data exist in the perceiver’s mind, not in the public world
Directly apprehended: we are infallible about their intrinsic properties — if the sense-datum looks elliptical, it is elliptical
Phenomenally characterized: they possess the sensible qualities experience presents (color, shape, texture) as intrinsic properties
Causally produced by but numerically distinct from external objects: the round coin causes an elliptical sense-datum; the external object and the inner object come apart

The inner theater follows immediately from this picture: the mind becomes a private theater in which sense-data perform, and the perceiver is an audience of one, watching an inner show that represents the external world but never is it.

Sellars on the Myth of the Given

Sellars’s “Empiricism and the Philosophy of Mind” (1956) provides the foundational analytic attack on the sense-datum picture, prior to and more general than the contemporary intentionalist critique. The Myth of the Given is the view that empirical knowledge rests on epistemically self-authenticating immediate experiences — items that are simply given to consciousness, presenting themselves directly and incorrigibly, and serving as the foundational evidence base for all further knowledge claims.

Sellars’s attack proceeds in two stages.

First, the structural diagnosis: the Given must do two incompatible jobs. It must be (a) immediate, non-conceptual, available without prior knowledge, or it is not a foundation; and (b) the kind of thing that can serve as a reason for empirical beliefs, or it cannot ground knowledge. But nothing can be both. To serve as a reason, an item must enter the logical space of concepts — must be the kind of thing that can stand in inferential relations to beliefs. And anything that does that has already been conceptualized. The supposedly raw given is a fiction.

Second, the Jonesean myth offers the positive reconstruction. Thought-talk and sensation-talk are not reports of pre-given inner items but theoretical posits modeled on the antecedent practice of public discourse. We do not first encounter inner sensations and then learn to describe them; we first learn the public language of perception (talk of red surfaces, of seeing things correctly or incorrectly) and then learn, via something like Jones’s theoretical move, to apply that vocabulary to inner episodes. The order of explanation runs from world-directed perceptual talk to talk of inner states, not the other way.

The bearing on sense-datum theory is direct: the sense-datum theorist supposes that the inner colored patch (or inner phenomenal item) is the foundational given, the thing perception immediately delivers, prior to all theory. Sellars shows this is incoherent. Whatever the senses deliver, they do not deliver self-authenticating inner objects ready to ground inference. The picture of the inner colored patch as immediate datum is itself a piece of theory — a theory that, on examination, conflates the (legitimate) thought that perception has phenomenal character with the (illegitimate) thought that phenomenal character consists in immediately-presented inner objects.

McDowell’s Mind and World (1996) develops the Sellarsian line into its principal contemporary form: rejection of self-standing inner standpoints, second-nature realism, the unified spontaneity/receptivity picture. The intentionalist case against sense-data inherits Sellars’s structural diagnosis and McDowell’s contemporary deployment.

The Neuroscience Version

The sense-datum picture survived into contemporary neuroscience under different vocabulary. The brain-as-screen model: perception involves the brain constructing an internal model or representation of the world, and what consciousness has access to is this model — not the world itself. Sometimes it travels as “controlled hallucination” (Anil Seth); Dennett attacks a version of it as the Cartesian Theater. The basic structure is unchanged: inner representational objects stand between the perceiver and reality.

The philosophical objection to this version: it runs together (a) the neural processes that realize perception and (b) the objects that perception is about. Describing how the brain produces perceptual states doesn’t establish that what we perceive is a brain state rather than the world.

The Transparency Diagnosis

The central intentionalist move against indirect realism: the argument from illusion rests on the phenomenal object premise, and that premise is false.

When I see the bent-looking stick, there is indeed an experience with a particular character. But that character consists in how the stick is represented, not in the properties of an inner object. The experience represents the stick as bent — that’s what it is for the experience to have bent-stick character. No inner bent thing exists; there is only an outer (straight) thing represented incorrectly.

The Transparency Thesis dissolves the argument from illusion by denying that introspection reveals a second inner object. When you attend carefully to your visual experience of the red apple, awareness returns to the apple — its redness, roundness, and position — while also allowing you to notice the visual mode in which those features are presented. That mode belongs to the complete intentional structure of the experience; it is not an inner colored surface or non-intentional residue.

On this diagnosis, the phenomenal object premise mistakes the representational character of experience (how things are presented) for properties of an inner object (what things are present to awareness). Reject that premise, and the argument from illusion doesn’t establish sense-data — it establishes only that perceptual experiences can misrepresent their objects. Misrepresentation requires no inner objects; it requires only that a state with a certain representational content fails to match the world it represents.

Adverbial Theory

An intermediate attempt to preserve the argument’s grip while avoiding inner objects. Instead of saying “S perceives a red sense-datum,” say “S perceives red-ly” or “S senses red-ly.” The adverb modifies the perceptual act, not a perceived object.

The many-property problem (Jackson): if I simultaneously experience a red triangle and a green circle, the adverbial theory must distinguish “experiencing red-triangularly and green-circularly” from “experiencing red-circularly and green-triangularly.” The object-theory handles this by individuating the objects; the adverbial theory must introduce complex adverbial modifiers that replicate object-theory structure without the objects. The simplicity gain disappears.

After Jackson’s objection, few remained adverbialists; in the field’s retrospect the view reads as a way-station, not a destination.

Disjunctivism

The direct-realist alternative that acknowledges the parity premise’s bite while rejecting its conclusion. McDowell, Hinton, Snowdon, Martin:

The disjunctivist move: veridical perception and subjectively indistinguishable hallucination are not instances of the same fundamental kind of experience. In veridical perception, the subject stands in a direct relation to the mind-independent object — the object itself is a constituent of the experience. In hallucination, no such relation obtains; the experience has only an internal cause. These are disjuncts, not instances of a common kind.

The parity premise fails: the subjective indistinguishability of perception and hallucination doesn’t establish that they share the same immediate object. Two situations can be indistinguishable from the inside without being fundamentally the same type of situation.

A further pressure case, due to Tye: a hand hidden behind a screen causes, by a deviant route, a visual experience as of a hand. Naive realism struggles to say why this is not perception of that hand; intentionalists answer such cases by specifying the right kind of causal-representational relation. Deviant causal chains remain a standing test for both camps.

The Inner Theater Picture

What indirect realism entrenches, its critics charge, is a whole picture of mind:

Mind as container: the mind is a space inside which mental objects reside
Experience as object: experiences are things we encounter, not relations we stand in
Perception as two-stage: first inner object, then inference to outer world
Privileged access as infallibility: we cannot be wrong about inner objects because they are constituted by how they appear

On the critics’ diagnosis, each element generates philosophical problems that wouldn’t exist without the picture: skepticism about the external world, the problem of other minds, the problem of phenomenal properties. All presuppose that inner objects exist and must somehow bridge to a world they stand apart from.

My View

The argument from illusion fails at its second step: the phenomenal object premise is false. Illusion is the experience getting the world wrong, not the mind being presented with a different, inner object. Misrepresentation replaces the sense-datum, and the whole apparatus of inner items becomes unnecessary. I read Sellars as having shown the deeper point a generation earlier: the self-authenticating inner given was never coherent to begin with.

The problems indirect realism generates aren’t solved by the picture; they’re created by it. Skepticism about the external world, the puzzle of other minds, the mystery of phenomenal properties — each presupposes the inner objects I deny. The fix isn’t a better bridge from theater to world; it’s the abandonment of the theater.

On disjunctivism, my sympathy is real but bounded. It rightly rejects indirect realism, but it does so by denying the parity premise, where I deny the phenomenal object premise — and its constituent claim makes even veridical misrepresentation hard to explain. Intentionalism keeps what disjunctivism wanted, the world itself in view, without that cost: directness consists in transparency, not in the object being a component of the mental state.

Transparency — the positive thesis that replaces indirect realism; introspection returns us to the world, not to inner objects; the argument from illusion’s phenomenal object premise is false
Direct Realism — the view indirect realism denies; intentionalist direct realism answers all five classical objections without positing sense-data
Intentionalism — the representational account that explains how experience is world-directed without inner intermediaries; misrepresentation replaces the sense-datum
Qualia — some qualia theories preserve the inner-picture pressure by positing intrinsic mental paint, but qualia realism is not simply indirect realism: a sophisticated realist may treat qualia as properties of experiential states without treating them as perceived inner objects
The Hard Problem of Consciousness — the hard problem partly derives from the indirect realist picture; if phenomenal properties are intrinsic inner properties, they obviously resist functional explanation
Intentionality — Brentano’s directedness points the right way; his language of immanent content can be misread as an inner-object ontology, but the relational reading does not require a second object inside the mind
Identity Theory — Smart’s reply to the spatial objection (experiences aren’t spatially located inner things) directly attacks the sense-datum picture
Embodiment — indirect realism severs mind from world; embodiment reconnects them at the level of causal-historical grounding