Argument from Illusion

1. Introduction

The Argument from Illusion is a central line of reasoning in the philosophy of perception that uses cases of perceptual error to challenge the idea that we are directly aware of external, mind‑independent objects. The argument focuses on familiar phenomena—such as a straight stick looking bent in water, or a distant object appearing smaller than it is—and asks what, exactly, we are immediately aware of when things look that way.

In its classic form, the argument is used to support sense‑data theories and more broadly indirect realism: the view that perception of the external world is always mediated by some kind of internal item or state. At the same time, it serves as a major target for direct realism and naive realism, which claim that in ordinary, veridical perception we are directly related to objects themselves, not intermediaries.

Philosophers have taken the argument in several directions:

• As a metaphysical claim about the immediate objects of awareness in perception (sense‑data vs. physical objects).
• As an epistemological challenge to the reliability and directness of perceptual knowledge.
• As a source of pressure to distinguish sharply between veridical perception, illusion, and hallucination, or alternatively to emphasize their similarity.

A key feature of the argument is its appeal to subjective indistinguishability: it often proceeds from the idea that certain illusions or hallucinations can be, from the subject’s point of view, phenomenally just like normal perceptions. From this, it draws conclusions about what must be common to all such experiences.

The Argument from Illusion has been developed, refined, and criticized across multiple traditions, from ancient skepticism to 20th‑century analytic philosophy and contemporary debates involving cognitive science and phenomenology. It now functions less as a single, unified argument and more as a family of related strategies and counter‑strategies concerning the nature of perceptual experience.

2. Origin and Attribution

The Argument from Illusion has roots in ancient philosophy but receives its best‑known formulation in early 20th‑century analytic work on perception.

Ancient and Early Modern Antecedents

Ancient Greek skeptics, especially Sextus Empiricus, collected examples of conflicting appearances—objects looking different from different standpoints, different animals perceiving the same thing differently—to question whether perception reveals the world as it really is. These discussions do not use the modern terminology of sense‑data, but they anticipate the argumentative strategy of moving from perceptual variation and illusion to doubts about direct knowledge of external objects.

Early modern philosophers also developed structurally similar arguments. Descartes, Locke, and Berkeley all used cases of illusion and hallucination to motivate theories on which we are immediately aware of ideas or representations rather than external objects themselves.

Sense‑Data Era and Canonical Formulations

The modern, explicitly sense‑data‑oriented Argument from Illusion is mainly associated with early analytic philosophers such as G. E. Moore, C. D. Broad, Bertrand Russell, H. H. Price, and A. J. Ayer. While Moore’s 1903 paper “The Refutation of Idealism” contains elements of the approach, more systematic presentations appear slightly later.

A commonly cited canonical statement is found in:

"

…in all cases alike—whether what we see be a penny or a mirage—we are directly aware of something which is not a material thing.

— H. H. Price, Perception (1932)

A. J. Ayer, in The Foundations of Empirical Knowledge (1940), gives a closely related, highly influential version that explicitly links illusion to the need for sense‑data.

Attribution and Variants

There is no single universally acknowledged “original” formulation. Instead, historians typically attribute the structured, sense‑data‑based argument to the classical sense‑data theorists, with Price and Ayer often cited as providing the clearest textbook expositions.

Later work by philosophers such as Howard Robinson systematizes and labels core components of the argument (e.g., the Phenomenal Principle), clarifying its structure without altering its sense‑data orientation.

3. Historical Context

The Argument from Illusion emerges within several overlapping historical developments in philosophy of perception and epistemology.

Ancient Skepticism to Early Modern Ideas

Greek skeptical traditions used perceptual discordance to undermine claims to certain knowledge about the external world. This general skeptical use of illusion set a template later thinkers could adapt. Early modern philosophers—especially Descartes’s methodical doubt, Locke’s representationalism, and Berkeley’s idealism—explicitly exploited illusion and hallucination to argue that what is immediately present to the mind are ideas rather than external objects.

Late 19th–Early 20th Century: The Rise of Sense‑Data

In the late 19th and early 20th centuries, philosophical attention turned to the phenomenal character of experience and the analysis of perception in terms of immediately given data.

Several trends converged:

Within this environment, philosophers such as Moore, Russell, and Broad articulated the idea that perception involves awareness of sense‑data, which seemed well‑suited to explaining both veridical perception and illusion.

Mid‑20th Century: Direct Realist and Phenomenological Challenges

By mid‑century, Ayer and Price had codified the Argument from Illusion as a central rationale for sense‑data theories. However, new currents—ordinary language philosophy, phenomenology, and later the rise of cognitive science—began to question whether sense‑data entities were necessary or coherent.

In response to both the Argument from Illusion and its sense‑data conclusion, philosophers developed direct realist, disjunctivist, and representationalist accounts that aimed to preserve the explanatory role of illusion without positing a special class of inner objects.

The argument thus occupies a pivotal place in the transition from classical sense‑data theories to contemporary debates over the metaphysics and epistemology of perceptual experience.

4. The Argument Stated

In its classical analytic form, the Argument from Illusion proceeds from ordinary examples of perceptual error to conclusions about the nature of all perceptual experience.

Illustrative Cases

Common cases include:

• A straight stick partly submerged in water looking bent.
• A coin seen at an angle looking elliptical rather than circular.
• A mirage in the desert that appears as a pool of water where none exists.

In such cases, it is said to appear to the subject that there is something with properties (bentness, ellipticity, wetness) that no corresponding external physical object actually has.

Core Line of Reasoning

The argument is often summarized along the following lines:

1. In illusions (and, in some versions, hallucinations), things appear to have certain sensible qualities that no external object in fact possesses.
2. Nonetheless, the subject is directly aware of something that does possess those qualities as they appear.
3. This “something” is not an external physical object, but a distinct item—often called a sense‑datum.
4. Veridical perceptions can be, from the subject’s point of view, phenomenally indistinguishable from corresponding illusions.
5. When two experiences are subjectively indistinguishable, they must be of the same fundamental kind and involve the same sort of immediate object of awareness.
6. Therefore, even in veridical perception, what we are directly aware of are sense‑data (or similar appearance‑like items), and external objects are perceived only indirectly, if at all.

Different authors emphasize different elements: some stress the indistinguishability of veridical and non‑veridical experience; others foreground the alleged necessity of postulating items that literally have the properties things appear to have. But the overall pattern moves from specific illusory cases to a general conclusion about the basic structure of all perceptual experience.

5. Logical Structure and Premises

The Argument from Illusion is usually treated as a deductive argument with a relatively clear structure. Variants differ in detail, but they share a core set of premises.

Canonical Structure

A widely cited formulation can be presented as:

Key Logical Features

• From particular to general: The argument begins with specific non‑veridical cases and generalizes to all perceptual experiences via P4–P5.
• Bridge principles: P2 and P5 act as bridge principles:
  • P2 embodies the Phenomenal Principle (roughly: if something appears F, there is something F we are aware of).
  • P5 connects subjective indistinguishability with sameness of experiential kind.
• Indirect realism conclusion: The conclusion is not merely that illusions involve sense‑data, but that all perceptions do. This is crucial: many critics accept P1–P3 for illusions while rejecting the generalization via P4–P5.

Variants

Some authors:

• Replace explicit sense‑data with more neutral talk of experiential items or qualia, leaving their metaphysical status open.
• Emphasize hallucinations as stronger cases where no external object corresponds to the experience at all.
• Recast the argument in terms of epistemic access: since we cannot, from within experience, distinguish veridical from non‑veridical cases, perceptual awareness of the external world must be mediated.

Despite these variations, the central logical pattern—using illusion and indistinguishability to challenge direct realism—remains common to most versions.

6. Sense-Data and Appearances

The Argument from Illusion relies heavily on the notion of sense‑data (or closely related entities) to account for the way things appear in perception.

What Are Sense‑Data?

Classical sense‑data theorists characterize sense‑data as:

• Immediately present in experience: they are the items we are directly aware of when we perceive.
• Possessing exactly the sensible qualities that things appear to have: colours, shapes, sizes, textures, spatial locations, and so on.
• Private or non‑public: typically, only the experiencing subject has access to a given sense‑datum.
• Non‑ordinary: they are not usually identified with familiar physical objects, though different theorists offer different metaphysical accounts (mental entities, aspects, or neutral items).

Sense‑data are often introduced precisely to make sense of illusions: when a straight stick looks bent, the sense‑datum is bent, even if the external stick is not.

Appearances vs. External Reality

The notion of appearance is central:

• In ordinary language, one might say “it appears as if the coin is elliptical.”
• Sense‑data theorists understand this as reporting a relation to a datum that is in fact elliptical in the subject’s experience.

Other philosophers prefer to speak of ways things appear without reifying them into distinct items. The Argument from Illusion traditionally treats these appearances as at least quasi‑objects of awareness, to which predicates like “red,” “bent,” or “elliptical” literally apply.

Alternative Conceptions

Not all versions of the argument use the term “sense‑data”; some adopt more cautious labels such as:

• Sense‑impressions (Broad)
• Perceptual particulars (Price)
• Qualia or phenomenal characters (later usage)

Proponents hold that some such category is needed to explain:

• How illusory properties can be genuinely instantiated in experience.
• How the phenomenal character of perception can remain the same across veridical and illusory cases.

Critics argue that appearances can be handled in terms of representational content or modes of presentation without positing special inner objects. Nonetheless, within the classical Argument from Illusion, sense‑data serve as the primary theoretical vehicles for capturing how things appear in perception.

7. Illusion, Hallucination, and Veridical Perception

The Argument from Illusion turns on systematic comparisons between three types of perceptual episode: illusion, hallucination, and veridical perception.

Distinguishing the Types

A common taxonomy is:

The argument in its narrower form focuses on illusions, since they arise in roughly normal circumstances and thus seem more troubling for direct realism. In expanded forms, hallucinations further reinforce the central line of thought.

Role in the Argument

Illusions and hallucinations are used to motivate two core claims:

1. Non‑physical objects of awareness: In non‑veridical cases, the argument claims, there must be something that genuinely has the properties that things appear to have, even when no external object does. This is where sense‑data are invoked.
2. Commonality with veridical perception: Since veridical perceptions can be indistinguishable “from the inside” from illusions or hallucinations, it is inferred that all three kinds of experience share a common object or structure.

On this view, all three are fundamentally episodes of awareness of appearances (sense‑data or similar entities), which may or may not correspond to external reality.

Disputes Over Classification

Different philosophical theories interpret these categories differently:

• Some disjunctivists insist that veridical perceptions and hallucinations belong to distinct fundamental kinds, despite their phenomenological similarities.
• Some representationalists claim that all three are states with intentional content; the difference lies in whether the content is accurate, not in whether special objects (sense‑data) are present.

However, within the traditional Argument from Illusion, illusion and hallucination serve as paradigmatic cases used to draw conclusions about the nature of all perceptual awareness, including apparently straightforward veridical perception.

8. The Phenomenal Principle and Common Factor

Two key theoretical components underlie many formulations of the Argument from Illusion: the Phenomenal Principle and the Common Factor Thesis.

The Phenomenal Principle

As labelled and discussed by Howard Robinson, the Phenomenal Principle states, roughly:

"

Whenever something appears to a subject to be F, there is something of which the subject is directly aware that is F.

In the context of illusions:

• If a stick appears bent, there is something bent that the subject is directly aware of.
• If a patch appears red, there is something red that is immediately present in experience.

Within the argument, this principle supports the move from talk of appearances to the existence of sense‑data:

1. It appears to S as if there is an F‑object.
2. By the Phenomenal Principle, S is directly aware of something that is F.
3. Since no material object is F, this must be a non‑material item (a sense‑datum).

Critics challenge whether appearances require such objects, but the principle is pivotal for classical sense‑data formulations.

The Common Factor Thesis

The Common Factor Thesis concerns the relationship between veridical and non‑veridical experiences. It holds that:

"

Veridical perceptions, illusions, and hallucinations can share a core experiential factor—such as the same kind of mental state, object of awareness, or phenomenal character—when they are subjectively indistinguishable.

Applied to the argument, this yields:

• If an illusion and a veridical perception are subjectively indistinguishable, they involve the same type of immediate object (e.g., sense‑data with the same properties).
• Hallucinations that are subjectively like veridical perception also share this common factor.

Together with the Phenomenal Principle, the Common Factor Thesis underwrites the generalization from non‑veridical to veridical perception: what we find in illusions (awareness of sense‑data) is taken to be present in all perceptual episodes.

Interplay of the Two Principles

In combination:

• The Phenomenal Principle introduces objects of appearance (sense‑data).
• The Common Factor Thesis extends their presence from abnormal cases to normal perception.

Much of the contemporary debate about the Argument from Illusion focuses on whether these principles are justified, how they should be interpreted, or whether they can be replaced by alternative explanations of appearances and phenomenological similarity.

9. Direct and Indirect Realist Responses

The Argument from Illusion directly challenges direct realism and is often used to motivate indirect realism. Contemporary discussion features a range of responses from both camps.

Direct Realist Strategies

Direct realists maintain that, at least in veridical perception, we are immediately aware of external objects themselves.

Common strategies include:

• Rejecting the Phenomenal Principle: Direct realists argue that experiences can present the world as having certain properties without there being any entity that literally has those properties in experience. The “bent‑looking” aspect of the stick need not correspond to a bent sense‑datum.
• Challenging the Common Factor Thesis: Some (often disjunctivist) direct realists deny that veridical perception and illusion share a common fundamental kind, even if they can seem alike to the subject.
• Reinterpreting illusions: Illusions are treated as misperceptions of real objects, often explained in terms of viewing conditions, perspective, or cognitive processing. The direct object of awareness is still the external object, though it is misperceived.

On these views, illusions do not require positing inner objects of awareness; instead, they are cases where our direct relation to the world is distorted or misleading.

Indirect Realist Approaches

Indirect realists accept that awareness of external objects is mediated by internal states or items, but they diverge on how to understand these mediating factors.

Two broad patterns appear:

Indirect realists generally embrace the argumentative pressure from illusion and hallucination: since we can have experiences with the same phenomenal character whether or not the world matches them, what we are immediately aware of must be some internal factor (either sense‑data or representational content).

While indirect realist positions differ significantly in metaphysics, they converge in seeing the Argument from Illusion as supporting the view that perception of external objects is only indirect or mediate.

10. Disjunctivist Critiques

Disjunctivism offers one of the most prominent contemporary critiques of the Argument from Illusion, especially its reliance on the Common Factor Thesis.

Core Disjunctivist Idea

Disjunctivists maintain that:

"

Veridical perceptions and non‑veridical experiences (illusions or hallucinations) belong to fundamentally different kinds, even if they are subjectively indistinguishable.

On this view:

• In veridical perception, one’s experience is constituted by a direct relation to external objects and their properties.
• In hallucination (and often in certain illusions), there is no such relation; instead, a different kind of mental event occurs.

These are seen not as two instances of a common type (e.g., awareness of sense‑data) but as disjuncts in an inclusive description such as “either seeing an F object or merely having an F‑like experience.”

Critique of the Argument from Illusion

Disjunctivists typically target:

1. P5 / Common Factor Thesis: They deny that subjective indistinguishability entails sameness of fundamental experiential kind or object. Two experiences can feel the same while being metaphysically different.
2. Generalization from illusion to perception: Since veridical and non‑veridical cases are of different kinds, disjunctivists argue that one cannot infer from what is true of illusions (awareness of an internal item, if any) to what is true of veridical perception.

Some forms of disjunctivism also question the need for sense‑data in illusions, but their central aim is to protect direct realism about veridical perception.

Positive Disjunctivist Accounts

Rather than only critiquing the argument, disjunctivists propose alternative models:

• Relational view of experience: In veridical perception, phenomenal character is partly constituted by the very external objects perceived.
• Different explanatory roles: Illusions and hallucinations may be explained in terms of brain events or internal states, but these states are not taken to define what is essential to veridical perception.

Influential proponents—such as J. M. Hinton, John McDowell, and M. G. F. Martin—develop versions of disjunctivism that attempt to accommodate the phenomenology of illusion while rejecting the key inferential moves of the Argument from Illusion.

11. Representationalist and Intentionalist Approaches

Representationalist (or intentionalist) theories offer another major response to the Argument from Illusion. They accept many of its starting points—especially the importance of phenomenal character and subjective indistinguishability—but reinterpret them in terms of representational content rather than sense‑data.

Core Representationalist Claim

Representationalists hold that:

"

Perceptual experiences are fundamentally states that represent the world as being a certain way; their phenomenal character is determined by their intentional content.

Accordingly:

• A visual experience can represent there being a red, round object in front of one, whether or not such an object exists.
• Illusions and hallucinations are cases where the content is misrepresentative or lacks a corresponding external object.

Reinterpretation of Illusion and Hallucination

Within this framework:

• The bent‑looking stick involves an experience that represents a bent object. The experience need not involve any bent sense‑datum; rather, the world is presented in a bent‑like way.
• In hallucinations, experiences have content as if an object with certain properties were present, even though no such object exists.

This allows representationalists to accept that appearances can be the same across veridical and non‑veridical cases, without concluding that they involve awareness of a special class of inner objects.

Responses to Key Premises

Representationalists typically:

• Reject P2 (Phenomenal Principle as object‑requiring): They deny that when something appears F, there must be some entity in experience that literally is F. Instead, they analyze this in terms of the experience’s content presenting F‑ness.
• Reinterpret the Common Factor: They often accept a kind of common factor, but as shared content or functional role, not as a shared object of awareness. Thus, the same representational content can occur in both veridical and non‑veridical experiences.

Philosophers such as Fred Dretske, Michael Tye, and William Lycan develop versions where the phenomenal character of experience is either identical to, or supervenes on, its representational content.

Relation to Indirect Realism

Some representationalists are indirect realists, emphasizing that we are only directly aware of contents or informational states. Others combine representationalism with direct realism, arguing that in veridical cases the content is partly constituted by the external objects themselves and that experiences thereby present those objects directly.

In all cases, representationalist approaches seek to preserve the explanatory role of illusion and hallucination while avoiding commitment to sense‑data as immediate objects of awareness.

12. Skeptical Uses of the Argument

Beyond its metaphysical implications, the Argument from Illusion has long been employed in skeptical arguments about perceptual knowledge.

From Illusion to Epistemic Doubt

The skeptical use proceeds roughly as follows:

1. Perceptual fallibility: Illusions and hallucinations show that experiences can be misleading while appearing subjectively just like veridical perceptions.
2. Indistinguishability: Since, from the inside, we cannot always tell whether we are in a veridical or non‑veridical state, our experiences do not by themselves guarantee that the world is as they present it.
3. Insecurity of perceptual belief: Therefore, beliefs based solely on perception may lack the kind of certainty or justification often ascribed to them.

On this reading, the emphasis is less on sense‑data and more on the epistemic unreliability or insufficiency of perceptual appearances as foundations for knowledge.

Connections to Classical Skepticism

The skeptical use of illusion arguments connects with:

• Ancient Pyrrhonian strategies: where conflicting appearances motivate suspension of judgment.
• Cartesian doubt: where systematic deception (dreaming, evil demon) extends the logic of illusion to all sensory experience.

In these contexts, the Argument from Illusion contributes to a broader case that perceptual evidence alone cannot conclusively establish the existence or nature of the external world.

Moderate vs. Radical Skepticism

Not all uses are fully radical:

• Moderate skeptics draw on the argument to insist that perceptual justification is always fallible, defeasible, and in need of corroboration (e.g., by coherence with other beliefs, reliability considerations, or inference to the best explanation).
• Radical skeptics deploy it, particularly in conjunction with hallucination and dream arguments, to claim that for all we know we may be systematically deceived, undermining ordinary claims to knowledge of the external world.

Philosophers responding to such skepticism often must address the indistinguishability point: they either deny that knowledge requires ruling out all such possibilities, or argue that perception can still provide adequate justification despite the theoretical possibility of illusion.

In this way, the Argument from Illusion plays a dual role: as a tool for theorizing about the structure of experience, and as a key component in traditional and contemporary skeptical challenges concerning perception.

13. Contemporary Status and Debates

In contemporary philosophy, the Argument from Illusion functions less as a settled argument for sense‑data and more as a shared reference point in debates about perception.

From Sense-Data to Pluralism

Few philosophers now defend classical sense‑data theories in their original form. However, the core phenomena highlighted by the argument—illusion, hallucination, and subjective indistinguishability—remain central.

Current positions include:

The Argument from Illusion is thus not abandoned but redistributed: different components are accepted, rejected, or reformulated depending on the theory.

Ongoing Points of Dispute

Key live issues include:

• Status of the Phenomenal Principle: Whether appearances require objects that truly instantiate the properties presented, or can be understood purely in representational or intentional terms.
• Extent of the common factor: Whether phenomenological similarity across veridical and non‑veridical cases implies a deep metaphysical or only a superficial commonality.
• Nature of phenomenal character: Whether the “what‑it‑is‑like” of experience is fully explained by representational content, or whether additional, possibly non‑representational, factors (e.g., qualia, relations to objects) are needed.
• Epistemic implications: How seriously the possibility of illusion should undermine claims to perceptual knowledge.

Integration with Other Debates

The argument now interacts with:

• Discussions of perceptual justification in epistemology.
• Work on consciousness, especially debates about the transparency of experience.
• Empirical research on visual processing and cognitive penetration.

Rather than a self‑standing, decisive argument, the Argument from Illusion is treated as a test case for theories of perception, a constraint any adequate theory must address by explaining how illusions are possible without mischaracterizing ordinary veridical perception.

14. Connections to Cognitive Science and Phenomenology

Recent work has explored how the Argument from Illusion relates to findings in cognitive science and themes from phenomenology.

Cognitive Science of Perception

Empirical research on perception provides detailed models of how illusions arise:

• Visual illusions (e.g., Müller‑Lyer, Ponzo) are often explained in terms of heuristics, prior assumptions, or neural processing biases.
• Studies of change blindness and inattentional blindness show that perception can be strikingly incomplete or distorted while still subjectively compelling.
• Neuropsychological conditions (e.g., Charles Bonnet syndrome, certain psychoses) offer real‑world instances of hallucination.

These findings are used in different ways:

• Some theorists argue that such data support representationalist models, where perception is understood as an inferential or predictive process generating internal models of the world.
• Others note that cognitive science explanations can be made consistent with direct realist or relational views by treating illusions as systematic misapplications of generally reliable perceptual mechanisms.

Cognitive science thus enriches the catalogue of illusions and hallucinations that any philosophical theory must accommodate, but it does not by itself settle the metaphysical issues.

Phenomenology and the Structure of Appearance

Phenomenological traditions (Husserl, Merleau‑Ponty, Sartre) approach illusion from the standpoint of lived experience, focusing on:

• The intentional structure of perception: experiences are directed at objects as existing in the world, even in illusion.
• The role of embodiment and context: how bodily orientation, movement, and practical engagement shape what appears.
• The dynamics of correction and revelation: how shifting perspectives reveal illusions as such.

Phenomenologists tend to resist reifying appearances into inner objects (sense‑data), instead emphasizing horizons of givenness, profiles, or modes of presentation. Illusion is seen as a modification within the overall intentional relation to the world rather than evidence that perception is always of inner items.

Some contemporary philosophers synthesize these strands, using cognitive science to model the mechanisms behind illusion while drawing on phenomenology to characterise how those illusions show up in experience. The Argument from Illusion, in this interdisciplinary context, becomes a crossroads where empirical and experiential analyses of perception meet, each informing but not decisively determining the philosophical conclusions.

15. Legacy and Historical Significance

The Argument from Illusion has had a lasting impact on both the development of analytic philosophy and broader debates about perception.

Shaping 20th-Century Philosophy of Perception

Historically, the argument:

• Provided a central rationale for sense‑data theories, influencing figures such as Moore, Russell, Broad, Price, and Ayer.
• Helped to define the agenda of analytic philosophy of perception by foregrounding questions about:
  • The immediate objects of awareness.
  • The relation between phenomenal character and external reality.
  • The epistemic role of perception.

Its structure and terminology became standard tools in mid‑century textbooks and discussions, shaping generations of philosophical training.

Catalyst for Alternative Theories

At the same time, the Argument from Illusion served as a foil against which later theories defined themselves:

Thus, even where the argument’s original sense‑data conclusion is rejected, its problems and examples continue to structure the field.

Broader Philosophical Influence

The argumentative pattern of moving from error to claims about underlying structure has influenced:

• Epistemology, through discussions of perceptual justification, skepticism, and the nature of evidence.
• Philosophy of mind, by prompting reflections on the relationship between conscious experience and representational content.
• Metaphysics, concerning the status of appearances, properties, and the mind–world relation.

The Argument from Illusion also connects historical and contemporary work: it provides a bridge from ancient skepticism and early modern theories of ideas to present‑day debates involving cognitive science and phenomenology.

While few philosophers now endorse the classic sense‑data version in full, the Argument from Illusion remains historically significant as a key driver of theoretical innovation and as an enduring touchstone for evaluating competing accounts of perception.