Find Enlightenment

Representationalism and Perceptual Error
by Diana Mertz Hsieh

Date: 12 May 95
Forum: Washington University in St-Louis, Theories of Perception class
Copyright: Diana Mertz Hsieh


In the philosophy of perception, that perceptual errors exist is most frequently regarded as a fundamental fact about perception which must be integrated into any coherent theory. But the position that the existence of perceptual errors ought not serve as a premise in an argument about the nature of perception, since any account of perceptual error presupposes a particular understanding of the nature of perception. In fact, any theory of perceptual error presupposes a particular model of consciousness, one in which there exists a possible correspondence relation between the objects of direct perception and external objects. In other words, the assumption that perceptual errors exist depends upon a representational model of consciousness, which may or may not accurate describe the functions of consciousness.

Representationalism (or indirect realism) with respect to perception is the view that "we are never aware of physical objects, [but rather] we are only indirectly aware of them, in virtue of a direct awareness of an intermediary [mental] object. (Dancy, 145) Because there are both direct and indirect objects of awareness in representationalism, a correspondence relation arises between the mental entities directly perceived and external objects which those mental entities represent. And thus perceptual error occurs when the two objects of awareness do not correspond sufficiently well. In opposition to representationalism, both (direct) realism and idealism agree that perception is direct and unmediated, despite their disagreements about what the object of perception is. (Dancy, 145) In any form of direct perception, no correspondence relationship is possible, since there is only one object of perception. Thus only representationalism will give rise to the view that perceptual errors exist and must be part of a theory of perception. Nevertheless, both idealism and realism must still account for the facts that are referred to as "perceptual errors" by the representationalist.

So in this paper I wish to argue that an account of "perceptual error" is not requirement for a theory of perception, because the view that gives rise to the concept of "perceptual error" is an unstable philosophical position and because the facts referred to by the term can be accounted for in a realist framework. I will argue that representationalism is unstable because it is a hybrid between realism and idealism, but that idealism is no better, since it is self-refuting and implausible, especially as applied to perception. My primary arguments against these positions will be done through a historical analysis of the rise of representationalism out of naive realism and of the subsequent moves towards idealism culminating in Kant. I will then explicitly defend a direct realist account of perception, all the while focussing on accounting for the facts underlying "perceptual errors," given that it was the inability of naive realism to account for perceptual errors which initially lead to the rise of representationalism as an alternative theory. Throughout this paper, I will be particularly drawing on the direct theories of perception put forth by James Gibson and David Kelley, as well as attempting to incorporate some of the methods of analysis of the utilitarian theory of perception.

Historical Progression of Theories of Perception

Throughout history, the progression of the major theories of perception shows a remarkable degree of regularity, for each major philosopher attempted to avoid or correct the contradictions of previous theories, while retaining the presuppositions about how an efficacious consciousness must function which have been nearly universally held Plato. Due to these assumptions about the nature of consciousness, since perception became a philosophical topic worthy of speculation a wide range of views have become dominant and then fallen into disrepute, from naive realism to representationalism to idealism. The instability of all of these views indicates that these assumptions about the nature of consciousness underlying these accounts of perception ought to be made explicit and challenged so that a more complete and stable account of perception can be offered.

Naive Realism:

Naive realism is the position that "unperceived objects are able to retain the properties of all the types we perceive them as having." (Dancy, 147) For the naive realist, the function of perception is to simply be a mirror for the world, directly and perfectly reporting features like temperature, color, and shape as they exist out in reality. Because all the qualities supposedly exist in reality exactly as we perceive them, there is no need for any activity or identity on the part of consciousness; in fact such activity or identity would only interfere with the perception of objects and qualities out in the world. A consciousness that contributed anything at all to perception, i.e. any consciousness that was more than just a passive receptor or that had a specific nature of its own, would distort the image of reality just as a curved mirror or vibrating water does. And so the naive realist concludes that the mind is passive and indefinite.

This presupposition that consciousness must neither be active nor possess a specific identity manifests itself in the way in which the existence of perceptual illusions, the existence different perceptual systems, and the scientific discoveries about the functions of perception quickly dismantle naive realism. Naive realism can not account for such simple perceptual illusions as the "bent" stick in water, because (according to naive realism) the visual information that the stick is bent necessarily means that it is bent, while the tactile information that the stick is straight necessarily means that it is straight. Such contradictions between sense modalities are not all that uncommon, and the naive realists cannot account for these facts while retaining their view that perception is simply an mirror for reality (at least not while retaining the view that reality is non-contradictory).

The fact that there are significant differences between the perceptual systems of animals and those of humans, and even that there can be differences among humans, also confounds naive realism. Variance in perceptual systems gives rise to variances in perceptual experience, but naive realism must maintain that such variances can only be the result of conflicts in reality. For example, if there are differences between the visual experience of two species because of differences in the ability to perceive color, then perception cannot simply be a mirror for reality, but rather an organism's perceptual system must also contribute to (i.e. distort) the act of perception.

Naive realism could not withstand advances in scientific discovery regarding the workings of perceptual systems, i.e. regarding the specific means by which organisms perceive, since perception isn't actually supposed to occur by any process in naive realism. For example, the discovery that the image on the retina is inverted or of the means by which sound waves are converted into electrical impulses in the ear greatly disturb the naive realist, since any translation on the part of perceptual systems indicates that reality is not being transparently conveyed to the mind.

Given all of these obvious difficulties with naive realism, why was it so widely held previous to Descartes? Naive realism's initial plausibility probably stems from the fact that it captures some basic intuitions about the subjective perspective on perception. From this subjective perspective, we are not aware of the fact that perception occurs by a specific, active means, and so "the awareness of an object seems transparent, the simple presence of an object, a revelation of it." (Kelley, 1986, 37) In other words, in perceiving the world we are not directly aware our saccadic eye movement, of the firing on the neurons below the retina, of the inversion of the image on the retina, etc., but rather we are only aware of the experience of transparently perceiving the objects in the outside world. But investigation into the scientific aspects of perception lead to the realization that perception is not simply a passive process which serves to transparently mirror the world, but rather an active, causal process that is dependent upon the perceiver and not necessarily veridical.

Cartesian Representationalism:

Descartes revolutionized philosophy in numerous respects, and his representational account of perception has shaped the study of perception to this day. Descartes arrives at his representational account of perception through his cognito argument and through the proof for the existence of the material world. The cognito argument presupposes that "consciousness can become the object of awareness, and be identified as consciousness, prior to any awareness of the existence of other things" and thus that "consciousness does not depend on reality for its contents." (Kelley, 1986, 12) Given these assumption about the relation between consciousness and reality, the idea that there is no necessary connection between consciousness and external reality seems plausible, and thus the naive realist position that consciousness transparently reflects reality is simply unjustified. And given the proof for the existence of external objects, the view that the mind just contains representations of reality seems to be the most plausible explanation for our mental activity. According to Descartes, these representations are all that we are aware of, but because a benevolent God exists and because our ideas must have a cause, we can infer a reality external to the mind (although this reality does not necessarily interact with the mind). Thus Descartes' position that "consciousness and reality are two separate and autonomous realms, neither dependent on the other, and although the mind is capable of grasping objects outside it, such knowledge is always mediated by a direct awareness of internal objects -- ideas -- that are its own product" becomes the paradigm case of representationalism. (Kelley, 1986, 10)

Descartes representationalism is clearly a causal account (i.e. the immediate objects of perception are caused by external objects in the world) because the proof for the existence of the material realm is the result of Descartes' view that "ideas must have causes and . . . these causes are material objects," because God would not lead us to believe that these causes were material objects if no such things existed. (Schacht, 1984, 34) Thus Descartes concludes that variations in perceptions when our sense organs remain constant must be the result of changes in the material realm. (Schacht, 1984, 36)

As later philosophers such as Leibnitz and Berkeley will argue, the possibility of establishing this causal relationship between the material realm and our mental representations is slim, since we simply do not have the required access to external objects (the indirect objects of sense) in order to determine whether such a causal relationship exists (or even whether external objects exist). But despite these problems in Descartes' account of perception, he did manage to solve many of the problems which plagued the naive realists. He could account for "perceptual error" by claiming that in such cases there was no correspondence relationship between reality and our representation. He did not have to account for differences in the forms of awareness, provided that the conscious subject was only aware of ideas. And the scientific discoveries about the processes of perception only furthered Descartes' point that we cannot be aware of external objects directly. So despite the difficulties which later philosophers exposed, Cartesian representationalism did serve to give a more scientific account of perception.

Lockean Representationalism:

Although Locke agrees with many of Descartes' conclusions about perception, the means by which he grounds his views is radically different from Descartes methodological skepticism and rationalism. Locke is a thoroughgoing empiricist, and thus his account of the causal connection between ideas and external objects is much less torturous than Descartes' proof.

Locke's basic position on perception is representationalist in that the immediate objects of perception are ideas, not physical objects, and in that these ideas are produced by or resemble actual physical objects. Locke holds the former (that ideas are the objects of direct perception) because physical objects are not in the mind, only ideas are. And thus we cannot be aware of anything but our ideas, even though our ideas may resemble or be caused by physical objects. (Schacht, 1984, 111; see also Locke, 1690, II, VIII, 8) This argument raises the issue of the directness of perception -- what the standards for directness ought to be and whether perception lives up to that standard, which I will examine later.

Like Descartes, Locke also endorses a version of causal representationalism, for external objects are conceived of as having the capacity to cause the ideas in the mind of which we are aware. The action of the objects of external awareness upon our senses produces (by some unspecified means) the ideas of which we are aware. But as later philosophers such as Berkeley will object, there is no justification for positing this causal relationship between external objects and ideas (or even for positing any external objects at all), since we cannot possibly step out of our sense modalities in order to verify or falsify this claim. Thus the idealist element in Locke's representationalism (as is the case with Descartes) is overrun by the idealist element inherent in it, and so in order to resolve the contradictions, later philosophers shifted towards idealism. But just as with Descartes, despite all of the problems exposed in Locke's account, it does address the issues for which naive realism was unable to account.

Responses to Cartesian and Lockean Representationalism:

Philosophers subsequent to Descartes and Locke retained the idealist position that the objects of direct perception must be ideas, while slowly discarding the realist view that external objects exist and cause our ideas. Leibnitz disputed Descartes' view of that which is self-evident and indubitable and paved the way for Kant's intersubjectivistic account of objectivity. Berkeley disputed Locke's view that there must be material objects to cause ideas. Hume's skepticism can be regarded as the necessary culmination of the elimination of the realist element of representationalism until Kant who, despite all of the contradictions in his metaphysics, altered the standards for truth to fit the new idealist model of consciousness. Leibnitz disputes Descartes doubt of the existence of material objects and criticized his proof of its existence, but does not offer any proof of his own. Rather Leibnitz holds that "that the world exists -- that there is something rather than nothing, whatever its nature may be . . . [is] indubitable from the outset." (Schacht, 1984, 50) Thus a lengthy proof of the existence of the material realm is neither necessary nor even desirable. Leibnitz goes further than Descartes in his mind-body dualism, for neither the soul nor the body influence the other, even though there is a correspondence between them. (Schacht, 1984, 59) This further radicalization makes the relation between external objects and ideas even more tenuous than it was in Descartes.

Leibnitz also seems to anticipate Kant in his account of why perceptual experiences between individuals correlate so well. Although no soul can affect another, according to Leibnitz, God correlates everyone's experiences so that they will all be the same. (Schacht, 1984, 65) This is similar to Kant's view that objectivity consists in "universal and permanent recognition" between subjects, which is based on identical structures and functions of human reason. (Kant, 1783, 255)

Berkeley's criticism of representationalism is clearly much more idealist than that of Leibnitz; in fact, Berkeley is, in many ways, the paradigm case of the idealist. He holds that "things exist only as complexes of ideas in the minds of spiritual beings who perceive them." (Schacht, 1984, 150) Berkeley primarily addresses the arguments of Locke, rejecting the key realist premises that there must be a material object to cause ideas and that ideas can resemble physical objects. Regarding the first premise, Berkeley argues that there is no justification for the view that ideas must be in some way caused by material objects. On the contrary, Berkeley contends that since the objects of direct perception can only be ideas and since these ideas exist only in the mind, it makes no sense to say that objects exist outside of the mind. (Schacht, 1984, 152) Berkeley also argues that ideas cannot be said to actually resemble any outside objects, at least not while the former are deemed perceptible while the latter are regarded as imperceptible. (Schacht, 1984, 156-7) This argument, unlike Locke's for the existence of external objects, does not rely upon a particular conception of external objects, but rather simply exposes the contradictions in the representationalist view given all of its premises.

These moves towards idealism by Berkeley and Leibnitz expose the basic instability of representationalism. But Hume moved representationalism further into idealism by assuming that we can only perceive ideas and then questioning the very notion of cause that both Descartes and Locke had so heavily relied upon. Hume's attack on the concept of cause is well known, and I only wish to mention it here in order to indicate how much it undermined the realist element in representationalism. Before Hume, whether it could be determined that a causal relationship existed between external objects and ideas was brought into question. But Hume's challenge of the very validity of the concept of cause made the representationalist account of perception all the more difficult to swallow.

The proximal cause of Hume's skepticism is clearly his failure to ground the concept of cause. But the distal cause was, I think, the result of the slow elimination of realism in the philosophy of perception, compounded by a lack of change in the epistemological standards (e.g. objectivity, truth) that were affected by the move away from realism. In other words, "Hume retains the realist standard or truth and objectivity: the principle that ideas must correspond to external objects in order to be valid" but rejects the position that we can determine if such correspondence exists. (Kelley, 1986, 21)

According to Kelley's analysis, after Hume, "the next obvious step is to reject this last element of realism and bring the epistemological standards back into like with the metaphysical nature of consciousness," which is exactly what Kant's metaphysics did. (Kelley, 1986, 21) Kant made these epistemological standards dependent upon what humans could know, i.e. the appearance of objects, rather than on things in themselves, which are unknowable in principle. Thus, according to Kant, "a judgment must be considered objective when it conforms, not to the real object outside consciousness, but to the internal nature of consciousness itself." (Kelley,1986, 26) This step by Kant is partially a means of revising the account of perceptual error so that perceptual errors both exist and so that we can detect them when they do. It is the latter which is so crucial to any concept of perceptual error, since without the ability to know what is erroneous and what is correct, we could be deluded with normal, coherent perceptions most of the time, while only the abnormalities would be an accurate reflection of reality.

But Kant's metaphysics suffered from severe self-contradictions, such as his assertion that a realm of things in themselves exists when it is supposed to be unknowable in principle and that the realm of things in themselves causes the realm of appearances, when the law of cause and effect is supposed to only apply to appearance (being that it is an a priori principle). But eliminating the realm of things in themselves would only compound Kant's difficulties rather than solve them. It would effectively prevent the differentiation of the mind from external objects (since the mind is all that exists) in order to form the concept of mind which is at the foundation of idealism. (Kelley, 1984, 32; see Kelley, 1984, 31-36 for a full discussion of the unintelligibility of idealism)

The basic historical progression from naive realism to representationalism to idealism illustrates the major weaknesses of all of these perspectives on perception. The vast body of scientific discovery which contradicts naive realism cannot be used to coherently uphold representationalism, because representationalism is the product of a mix of premises from realism (about external objects) and from idealism (about the objects of perception). But idealism does not solve the problems of a theory of perception either, because it is unable perform some of the necessary basic functions of any philosophy of perception, such as individuating the mind. Today, most philosophers (with the notable exceptions of Gibsonians) speak of perception as some form of representation, despite the problems that such accounts led to in previous years. Precisely because of those problems, I think that it is absolutely vital to attempt a new form of realism, one which does not make the basic errors of naive realism, i.e. one that can account for what is termed "perceptual errors."

Gibson's Direct Realism and the Utilitarian Theory of Perception:

In the latter half of this century, James Gibson revived direct perception with his ecological approach, which seeks to explain the function of perception for an organism through a series of new concepts and ideas relating to the intimate relationship between an organism and its environment. There are, I think, two fundamental ideas of Gibson's that are vital for any defense of realism: that perception is an inherently active process and that all necessary information for the (visual) awareness of the environment is contained in the ambient light array. The former is necessary to prevent a reversion into some form of naive realism, while the latter serves as a defense against the representationalist view that there is a poverty of stimulus in perception and thus that organisms could not experience all that they do without additional information supplied by the mind.

Gibson conceives of perception as an inherently active process, especially with respect to the pickup of information. The organism's movements, from the saccadic movements of the eye to whole-body actions, help it to gather information about its environment which is "specific to the environment . . . and to self-movements." (Turvey, 1992, 91). According to Gibson, perception is "not an appearance in the theater of [an individual's] consciousness" but rather "a keeping-in-touch with the world, an experiencing of things, rather than a having of experiences." (Gibson, 1979, 239) The perceived information does not need to be constructed out of sensory inputs, but rather "the centers of the nervous system, including the brain, resonate to information." (Turvey, 1992, 93)

Gibson does rightly reject the representationalist view that the mind must construct the objects of direct perception out of sense data on the grounds that this view does not properly account for the function of perception and that it is not necessary to explain the facts of perception. Gibson holds that the purpose of perception is guide an organism's actions, and thus any theory of perception ought to account for "an organism's apprehension of its environment and how it controls its acts with respect to that environment." (Turvey et al., 1981, 244). And since organisms do not live among mental representations, but rather among actual physical objects, a representational model of perception cannot properly account for the interaction between an organism and its environment. (Turvey et al., 1981, 239)

Gibson also does not need to appeal to any form of representationalism because he holds that the light (or more specifically the "ambient optic array") contains all the information necessary to unambiguously specify an environment, since any information is "specific to its environmental sources." (Reed, 1986, 78) Gibson takes this position in opposition to the "poverty of stimulus" argument, which states that there is not enough information in any act of perception to give rise to the richness of our perceptual experiences, and so there must be some background information (or inferences) which help give rise to our experiences. But Gibson's concept of the visual field allows him to reject this argument, since the visual field contains much more information than the perception of a single object could, such as information about color, texture, and angles, all of which occur in "lawful regularities."

But there is perhaps one aspect of the poverty of stimulus argument that the concept of the visual field by itself cannot overcome -- that there are parts of the visual field which do not supply any information at all (e.g. the blind spot), but which seem continuous to the perceiver. I believe that Daniel Dennett has a plausible explanation for this phenomena of "filling in" the gaps in the visual field which avoids the serious hazards of representationalism, and even is compatible with Gibson's approach to perception in many respects. In addressing this issue of "filling in" the gaps in the visual field, Dennett does speak loosely of "representations," although I do not think that his analysis is entirely antithetical to Gibson's views. Rather it seems to provide a valuable method of thinking about a direct view of perception.

Dennett objects to the idea that the brain "fills in" the information missing in the blind spot on the grounds that it presupposes the view that the visual field, plus the filled in spots, is going to eventually be presented to someone watching events in the brain. (Dennett, 1992, 44) Dennett postulates that when the brain is presented with a repeating pattern (such as the sky, a manicured lawn, or Marilyn Monroe wallpaper), it generalizes from one instance of the pattern to the assumption that all instances which are relevantly similar are exactly the same. So, with the case of the Marilyn Monroe wallpaper (Dennett's example) the brain focuses on one Marilyn, recognizes it as a Marilyn, and then assumes that all the other Marilyn-shaped blobs are actually Marilyns until it receives evidence to the contrary. (Dennett, 1992, 46) Thus the brain isn't "filling in" any information but is rather omitting information, i.e. taking the omission of information that no information is being conveyed in the blind spot as information that there is something being conveyed (that is the same as all the information around it). (Dennett, 1992, 48)

I take Dennett here to be endorsing something close to the view that perception is strategy-laden, i.e. that the brain uses certain strategies in processing information which serve to compress the amount of information which needs to be processed in the visual field. Although this focus on perceptual strategy is often accompanied by a form of representationalism, I do not think it necessary to postulate the existence of representations in order to say that perception is strategy-laden. In fact, Ramachandran's utilitarian theory of perception, in which perception is regarded as "essentially a 'bag of tricks'; that through millions of years of trial and error the visual system has evolved numerous shortcuts, rules-of-thumb, and heuristics which were adopted . . . because they worked" can effectively be used in a direct theory of perception as a partial account of perceptual error. (Ramachandran, 348)

The paradigm case of strategy-ladenness in perception is the method by which the visual system handles the spots on a jumping leopard. The visual system significantly cuts down on the amount of processing necessary by assuming that the spots of a leopard move with the outline of the animal, rather than correlating the shortest distance between pairs of spots in successive visual snapshots. (Ramachandran, 349) As Ramachandran points out, this "the visual system can only afford to use this trick or 'short-cut' because in the real world spots do not usually fly off leopards." (Ramachandran, 349) Thus the utilitarian theory of perception can be seen as endorsing a type of "evolutionary attentional mechanism," in which the routine information gets processed by certain strategies so that more interesting (and more important) information in the visual field can be focussed on.

Ramachandran's utilitarian theory of perception presumably does not fully reject representationalism, since he argues for a weak form of it in other papers. (Churchland et al., 24) But I do not think that representationalism is particularly important to Ramachandran's account of the strategies adopted by perceptual systems, since the form of representationalism that he adopts is extremely weak and since many aspects of his utilitarian theory of perception are compatible with direct realism. Ramachandran argues for the notion of visual semi-worlds in which "only immediately relevant information is explicitly represented" as opposed the "fully elaborated representation of a visual scene" which is at the core of representationalism. (Churchland et al., 25, 24) In fact, the way in which the term "representation" with respect to these visual semi-worlds is used often suggests that it means nothing more than "awareness." (Churchland et al., 26)

The fact that both direct realism and the utilitarian theory of perception focus on the need to filter out extraneous information (rather than on adding information or making inferences), it seems plausible that the strategies of perception that Ramachandran discusses could be used in a direct theory of perception. In fact, adopting the view that the brain uses strategies in processing information from the visual field serves to aid the direct realist in accounting for the facts referred to as "perceptual errors." At least some "perceptual errors" could be regarded as the application of a specific strategy of information processing out of its proper scope in such a way that pertinent information is ignored (unless perhaps directly focussed upon). Because these strategies were only initially adopted because they compressed information without sacrificing veridicality, e.g. because spots do not fly off of leopards, the only way to discover these strategies would be in artificially constructed situations. And so because these strategies do work in nearly all situations and because the brain must use them in order to process perceptual information, to call them "perceptual errors" would seem to be a misnomer. To "correct" the errors would prevent us from perceiving at all.

This analysis of some "perceptual errors" as strategies of the brain which occasionally ignore information relevant to the organism can be widened into a more general account of the facts underlying "perceptual errors" which does not rely on a form of representationalism. Obviously, our sense modalities have certain limits on the type and scope of information that they can detect; one cannot, for example, taste light or get an absolute temperature reading in Kelvin from touch. This fact necessitates that the senses have certain limits on what sort of discriminations can be made and how finely they can be made. For example, the fact that Turvey's shark initially acts identically towards the natural (edible) bioelectric field of a flatfish and the artificially produced (inedible) field is the result of its inability to differentiate between artificial and natural bioelectric fields. The shark was not in error by investigating the artificial field, because in the shark's environment, bioelectric fields and edible things are correlated by the "physics of the ecological world" (as opposed to by inference). On the contrary, "the wrong action for the shark, given its niche and appetite, would be not investigating the source of the field." (Turvey, et al., 1981, 277) The fact that the natural and the made-made electric fields are treated identically by the shark is the result of the fact that he does not have the necessary equipment to discriminate between the two.

In many cases, an organism does have the capacity to discriminate, but has not yet focussed on the differences between two similar objects of perception. For example, the ability to discriminate between types of wine is an acquired skill, since it takes time to actually notice the differences. The fact that an organism has not yet acquired the capacity to make such distinctions between two objects does not make identical actions towards both an "error." Considering the organism's level of knowledge in such a case, identical action is completely appropriate; only asymmetric action would be an error, since the organism perceives the objects as similar in all relevant respects.

The strategies that are referred to in the utilitarian theory of perception could actually be regarded as the mind's taking advantage of the lawful regularities of any environment in order to reduce the computational resources of perceptual awareness. Just as the strategy of making the dots of the leopard follow the outline utilizes the fact that spots stick to leopards, the shark's strategy of investigating bioelectric fields utilizes the fact that (until humans set up these experiments at least) there was a physical connection between bioelectric fields and edible things in the shark's niche.

I think that this analysis of "perceptual error" as a failure of ability to differentiate between relevantly similar objects of perception is extremely useful. It does not involve any form of representationalism, but at the same time avoids the basic problems that the naive realist faced. It does not account for simple perceptual illusions like the stick that looks bent in water, but considering our knowledge of physics, and particularly how light behaves in water in this example, "there is no intelligible sense in which it can be claimed that the stick ought to appeal straight in perception if perception were free of error and perception were direct." (Turvey, 1992, 275) To see the stick as straight would rather be an "error" on the part of the perceptual system, since light does actually move slower (on average) through water. It also does not account for verbal misidentification of an object of perception, but those sorts of errors do not seem to be perceptual, but rather conceptual in nature. To conflate the two is a commitment to some sort of "metaphysical essences" that are directly perceived, a position which I suspect few philosophers would want to endorse.

Relevant Standards for Directness:

The final issue which I wish to address about realism is what a relevant standard for the directness of perception is. Robert Schwartz has argued that the direct-indirect issue in perception is "a swamp of issues that it is best to avoid" because "the debate . . . hinge[s] on resolving some tricky metaphysical, epistemological, and linguistic problems." (Schwartz, 1994, 143) Although I agree with Schwartz that the debate can be likened to a "swamp of issues," I think that there are ways to make sense of this issue, primarily by ensuring that plausible standards for the directness or indirectness of perception are upheld.

It is has been argued on a priori grounds, particularly in response to Gibson, that (certain types of) perception cannot possibly be direct. (Schwartz, 1994, 143) Although it is unclear why one should call perception indirect if it cannot possibly be direct, the fundamental problem with this argument is an unattainable standard for the directness of perception. Such a priori arguments rely on the idea that if perception occurs in any form, by any means, then it cannot be direct, which is simply a variation on the naive realist assumption that perception must be passive imprinting on the mind in order to be veridical. But that perception occurs in a certain form and by a certain means (whatever that may be) does not mean that we are not directly aware of the external world. To clarify this point, let me make an analogy to digestion. Humans (and lots of other organisms) gather food from the environment, eat it, and digest it, all by a certain specific process. But the fact that there is a specific process does not imply that we gain the necessary nutrition to live "indirectly." If we tried to feed ourselves without any specific means, we would not be able to feed ourselves at all. And so in perception the means by which we are aware of external objects and the form in which we are aware of them is intrinsically necessary to any act of perception at all. To assert that direct perception must not be by any specific process or in any particular form that is determined by the perceiver is to say that perception can only be direct if it does not occur at all. Thus our standard for directness must be attainable, and so it ought to be something akin to the idea that we are directly aware of the external world, even if there is a specific form and process to that awareness.


The failure of both naive realism, representationalism, and idealism to properly account for known facts about perception indicates that a whole different approach to perception must be put forth. I have argued that a form of realism is the only possible alternative, but that it must be one which can properly account for the facts underlying the idea of "perceptual error." Gibson's direct realism in many ways provides the foundations for this new form of realism, and the integration of his views with some of the utilitarian theory of perception yields a plausible account of "perceptual error." This integration of views into a form of direct realism can account for scientific aspects of perception, as well as accord with internal perspective of perceivers in a way that representationalism fails to do.


Churchland, Patricia, Ramachandran, V.S., Sejnowski, Terrence. A Critique of Pure Vision. Theories of Perception Course Packet. 23-60.

Dancy. Theories of Perception Course Packet.

Dennett, Daniel. (1992) "Filling In" versus Finding Out. In Herbert Ficke, Paulus Van Den Broek, David Knill (eds.), Cognition: Perceptual and Methodological Issues. American Psychological Association, Washington, DC. 33-49.

Fodor, J.A. and Pylyshyn, Z. (1981) How Direct is Visual Perception? Cognition, 9, 139-196.

Gibson, Eleanor. (1992) How to think About Perceptual Learning. In Herbert Ficke, Paulus Van Den Broek, David Knill (eds.), Cognition: Perceptual and Methodological Issues. American Psychological Association, Washington, DC. 215-237.

Gibson, J.J. (1979) The Ecological Approach to Visual Perception. Houghton-Mifflin, Boston.

Kelley, David. (1986) The Evidence of the Senses. Louisiana State UP, Baton Rouge.

Kant, Immanuel. (1783) Prolegomena to Any Future Metaphysics. Hackett, Indianapolis.

Locke, John. (1690) An Essay Concerning Human Understanding. Dover Publications, New York.

Ramachandran, V.S. The Utilitarian Theory of Perception. Theories of Perception Course Packet. 346-360.

Reed, Edwards. (1986) James J. Gibson's Revolution in Perceptual Psychology. Studies in the History and Philosophy of Science. 65-98.

Schacht, Richard. (1984) Classical Modern Philosophers. Rutledge, New York.

Schwartz, Robert. (1994) Vision. Blackwell, Oxford.

Turvey, M.T., Shaw, R.E., Reed, E.S., Mace, W.M. (1981) Ecological Laws of Perceiving and Acting Cognition, 9, 237-304.

Turvey, M.T. (1992) Ecological Foundations of Cognition. In Herbert Ficke, Paulus Van Den Broek, David Knill (eds.), Cognition: Perceptual and Methodological Issues. American Psychological Association, Washington, DC. 85-117.

Find Enlightenment at