The Role of Imagination in Perception

Imagination is a topic of ubiquitous, varied and profoundly existentially significant philosophical reflection in the millennia-spanning heritages of South Asian thought.  In the Brāhmiṇical traditions that grew out of the śramaṇa movements in ancient India, it was often conceived as the “magic” (māyā) through which first the luminous deities and then, in different degrees and ways, consciousness created or manifested the world with all its beautiful but ultimately inessential material bodies and things.  In one branch of South Asian Buddhism, Vijńānavāda, Vasubandhu attributes to the imagination the power to help us see colors, taste flavors, feel textures and love ourselves, in spite of the fact that, in reality, there are no such things as colors, flavors, textures or selves.   It is these marvelous and surely profound conceptions of imagination that most Westerners newly introduced to Asian religious and philosophical histories are presented with.  But are their resources in the Asian traditions that make of imagination something more than the repository of fabrications, illusions and misunderstandings of ourselves?  Might imagination also help us to experience and thrive in the sensuous and physical world that is the home of our embodied lives?

As B.K. Matilal illustrated more than thirty years ago in his landmark English-language study on theories of perception in Indian thought, “imagination” (kalpanā) was given a technical sense and was seen to be a basic mechanism of cognition by classical South Asian thinkers for millennia.  One of the most fraught debates that took place over the role of imagination in cognition between Brāhmiṇical and Buddhist philosophers involved whether or not imagination played a role in direct sensation or perception (pratyakṣa) or whether it was an activity confined to inferential states of reflection (anumāna) or to mental states that were merely fantasies, concoctions of things that were not real.  This issue was placed front and center through the purported definitions given for sensation or perception by the most seminal seventh-century purveyors of Yogācāra-Sautrānta Buddhism, Dińńāga and Dharmakirti.  They insisted, in subtly variant ways, that imagination could not be operative in purely sensory or perceptual cognition.  This precipitated centuries of attempts by Brāhmiṇical opponents to demonstrate that, while “imagined” objects were not themselves directly sensed, the objects of sensation could often not be recognized as the things they are without the aid of imagination.  We do not, they came to insist, only use our eyes to see, but we must often use imagination to see, and by that very token, imagination helps give us access to what is real.

Now, while this debate has often been cast, in both classical South Asian and modern Western hermeneutical terms as turning on the problem of whether or not commonly pervasive properties or “universals” (samānya) should be considered objectively real or natural phenomena or merely conceptual “constructs,” I would like here to focus on how these opposed positions characterize sensation or perception itself.  When we sense, what exactly do we sense, and how do we understand the mechanisms though which sensations and perceptions as specific kinds of cognitions come about?  I will demonstrate in what follows that both Yogācāra-Sautrāntika Buddhists and Naiyāyikas held that imagination enables us to think and act within the world rightly and effectively.  Imagination, that is, helps us to experience what is real.  However, their respective accounts of this are not equally sound and true to our experience.  I believe that what will become apparent as a result of this focus is the degree to which the Yogācāra-Sautrāntika conception of kalpanā impoverishes their depiction of what sensations and perceptions are.  I will in the course of this examination try to explain why I characterize their view in this way in the light of certain Nyāya attempts to defend the role of imagination in perceptual states.  We will then discover the insightful and powerful fashion in which certain Nyāya philosophers argued that imagination intermingles with our sensation and helps us to truly perceive.

In his definition, Dińńāga posited that “sensory cognitions are free of imagination” (pratyakṣaṃ kalpanā podhraṃ).  Obviously, it is crucial for us to understand how Buddhist logicians conceived of imagination in order to grasp what immediate sensations or perceptions are free of in lacking it.  The most important sense of kalpanā or vikalpa in Dharmakirti’s commentaries concerns the power of imagination to provide the “establishing basis” or “mark” (sādhana) that serves as the sign for thinking about, remembering or speaking about an object of experience.  That is to say that imagination here does not have the role it has in other systems of thought, in which its objects are distinguished or separated from one another.  Rather, imagination forges connections between one objective form and another by creating commonalities between them, but all of these commonalities are concocted within the imaginative process and are not gleaned from things that are merely sensed.  And it is precisely, for the Yogācāra-Sautrāntikas, their reference to the objects of other cognitions prompted by “marks” that distinguishes “imaginations” from sensory cognitions, for the latter are restricted (niyata) only to that object that appears directly to the sense organs (nirbhāsa).  Unlike other mystical traditions of Indian thought such as those in the lineage of Vedānta, then, Dińńāga and Dharmakīrti did not associate kalpanā as imagination with the concocting of dream or illusory cognitions that could be produced wholesale by the mind.  This was a weighty consideration for Dharmakirti in defending and clarifying Dińńāga’s definition of perception, which had claimed that perceptions were distinct from imaginings because the latter affixed names and a classification to objects.  When the Brāhmiṇical loyalists pointed out that false cognitions and dream states could also attach names and classes to their objects, and so such a distinction did not threaten the possibility that real commonalities could be detected in perceptual experience, Dharmakirti responded.  The kind of imagination he had in mind was free from error (abhrānta) and thus decidedly not associated with dream or illusion.  For the Yogācāra-Sautrāntika position, he insisted, imaginative cognitions could still serve as instruments of knowing, for they could, in the technical epistemological sense, often lead to successful activities such as apprehending or inferring a fire as a fire and using it for warmth.  Instead, imaginations, insofar as they enable people to attribute any commonality (samānyalakṣaṇa) to an experienced thing through memory, association, inference and so on are not drawing any of such commonality from a directly perceptual experience, but from other cognitions.  We do not, that is, see commonality, we imagine it.

This presents at least two serious problems, one for the Buddhist tradition itself and the other of a broader philosophical sort.  The difficulty of Dińńāga’s and Dharmakīrti’s isolation of perception that is internal to Buddhist discourses has been lucidly pointed out by Richard Hayes.  The Sautrāntikas have on the one hand with this model elevated directly perceptual knowledge to the level of paramount truth (paraṃārthasatya) because, unlike inferences that can involve mistaken predication, it is always free from error.  On the other, all the signature teachings associated with Buddhism, the suffering and impermanence of life, the lack of a self or fixed nature in persons and things, the soundness of the eightfold path and so on are all knowable only by inference, by making generalizations about direct experience that are not strictly speaking derived from it, and that don’t ultimately correspond with it.  So much, then, for the claim for Buddhism’s grounding in direct experience if what the Sautrāntikas say about the gap between perceptual and conceptual experience is true.  But, in another implication that is more broadly pertinent, one way to characterize Dińńāga’s and Dharmakīrti’s conception of perceiving is that we really don’t perceive anything to speak of.  We can call the supposed object of nirvikalpapratyākṣa a “self-characterized” momentary experience if we want, but since the content is not perceptually recognized, apprehended, bears no perceptual commonality with other things and cannot be articulated in language, what is the theoretical or practical difference between the so-called svalakṣaṇa object of perception and a mere stimulant of a sense organ that has no cognitive or practical value for the person who is aware of it?

Even, to press the point in the paragraph above, if we insist as the Sautrāntikas do that the self-characterized object of a perceptual cognition is still somehow a definite thing, how experientially tenable is it to claim that our perceptual encounter with things exhibits no structural relationships at all?  Do we not, for example, genuinely taste both the coolness and flavor of a liquid as belonging to the liquid when we drink it?  Do we not hear both the pitch and duration of a bird’s call as coming from the bird when it is sung?  Do we not see ridges in a tree as inlaid into the tree when we look at it?  This question does not require that we have conceptual ideas of what “coolness,” “pitch” or “ridges” when we have these respective sensations, but rather asks whether we can perceptually identify features as belong to, residing in or qualities of other things.  Do we only sensorally detect one experience at a time with no direct relation to the next, or do we have some genuinely sensory access to the textured layeredness of things?  Do we, to put the above questions in the contrapositive, detect temperature and flavor as coming from outside the glass that holds the liquid I am drinking, or see ridges just happening to appear in proximity to some other sensations of wood?  Speaking of ideas that are counter-intuitive, it seems to strain credulity to suppose that we “infer” things like flavor, pitch and spatial relations after seeing things that are just distinct from one another in some ultimately inexplicable way.  It is maybe for this very reason that the vocabulary of “concept” as a translation for words like kalpanā or vikalpa is particularly unhelpful, because the Buddhists believe that these are in a more literal sense “imaginations” that arise from the sedimented dispositions we have built up as the kinds of beings we are.  And the fact that they are conceived as “imaginings” greatly buttresses the Buddhist theory, for as imaginings, they are presumably subjective to each of us and not features of the physical world.  After all, as Yogācārins like Vasabandu would point out, textures and flavors and spatial relations will appear quite differently to creatures of different kinds, and this would show our qualitative experience of things depends largely on how we are used to taking them.  But, though this reminder is surely a helpful one, it does not quite capture the Naiyāyika concern with the notion of perceptions as being exclusively nirvikalpa.  Nyāya thinkers too acknowledge that our habits and accreted ideational associations (saṃskāra) will cause one person to find a sound beautiful and another person to cringe with annoyance at it.  While we may react to different features of things quite uniquely given our individuated karmic heritages, we still find features, according to Nyāya thinkers, to belong to things in our properly perceptual experiences.  And so, in conceding that imagination does play an active role in our perceptions of structure, Nyāya philosophers wanted to demonstrate how this imaginative process enhances our sensory engagement with things and helps to reveal them, rather than merely distorting perceptual experience and framing it around our subjective desires and plans.

Of course, the contention that we do not directly perceive commonality in the things we encounter strikes at the very heart of the classical Brāhmiṇical conception of realism.  According to Nyāya, our most direct form of knowledge, precisely perception, does present us not just with unique sensations, but also with features that are common to many things.  Naiyāyikas therefore vigorously attempted to refute this contention that commonality was merely a creation of mental activity and not observable in the properties of things.  Now, they often proffered the rebuttal that is built into the classical Nyāya ontology, which asserted that common properties or features are actually in the things we see, and so when we perceive something, we perceive both the individuated thing and its common properties all at once.  But this of course is just a claim; how can we know that it is true?  The demonstration proved to be enormously tricky for Naiyāyikas, in large part because they generally agreed with Dińńāga and Dharmakirti that, in the first moment of sensory cognition, there is no recognition of things as qualified by their common properties (viṣeśyaviśeṣanabhāva).  Indeed, this acknowledgement that an initial sensory cognition does not apprehend commonality in the features sensed seems, as D.N. Shastri pointed out long ago and John Taber has reiterated since, to have been generally accepted among both Mimāmsa and Nyāya thinkers within a century or so of the idea’s introduction by Dińńaga, among Kumārila, Vācaspatimiśra, Śridhāra, Udayana and persisted on into Gańgeśa’s writings.  If, then, imaginings (vikalpaka) that supplied the commonality of certain features of our experience and the relational ties that joined such features with certain objects were not drawn from the first moment of sensation, then in what sense could they be perceptual?  If they are perceptual and not merely produced and projected onto our sensations by mental activity alone, there must be a causal story for how the objects of imagination could be perceptual, or how they could help make perceptions perceptions.

In the next several centuries, Nyāya thinkers made a number of successive attempts to show how imagination enhanced perceptions, and did not merely follow them and distort them in some fundamental fashion as the Buddhists held.  Some, like Vācaspatimiśra, slightly broadened the classical definition of perception itself, saying that it occurred not merely when something comes in direct contact with the sense organs but when something enters into the range of sensory cognition (yad eva indriyajasaya jńānasya gocaras tat pratyakṣaṃ na tv indriya saṃbaddham).  Previous cognitions, memories, have made us familiar with the relational ties between things and their properties.  And so, we remember that certain features are always found in certain kinds of things, and these memories can be relied on for the imaginative content that takes features (dharma) as “marks” of the things in which they are found (dharmin).  Thus, as Vācaspati tells it, while the initial sensory contact with a thing makes us aware of a form (rūpatayā), the memory bringing the imagined relation between presently sensed features and recalled things mingles with the sensation such that the object we finally fully perceive possesses a form (rūpitayā).  This manner of perceiving things comes to be known in the tradition as jńānalakṣaṇa, or a perceptual cognition featuring the possession of marks or determination.  Imagination, on this view, must surely assist, through memory, in the delivery of “marks” or common features through which things are recognized, lest a sensation be only an unnoticed physical interaction between a thing and a sense organ.  But, so long as the features or qualities of things in question are in fact seen in the sensory cognition, then imagination helps us to perceptually recognize what a thing is with the aid of things remembered that resemble the present.  Is this, and other Nyāya representations much like it, sufficient to demonstrate how imagination can aid perception?

In this ongoing contemporary debate about perception and imagination in Nyāya, Chadha has suggested that imagination gives the person a “recognitional capacity” to discriminate between things.  While this conception of a “recognittional capacity” (presumably a translation of the expression samvid samārthya) invoking memory and habits, is helpful in explaining the role that kalpanā plays in our conscious lives, it perhaps requires that we retrain are focus on precisely what kalpanā in this context are.  We have said above that, in this particular philosophical context, imaginings are non-illusory cognitions in which some feature of the world is presented as a common one, that serves as a “mark” or indicator of a relationship with things of certain kinds, and that therefore invokes other cognitions.  Now this would seem, as Chakrabarti has also pointed out, to contravene another basic Nyāya belief about perceptions, namely that they cannot be caused by another cognition but are instead caused by sensory contact with things.  So, if what is being suggested here is that imaginations actually can help precipitate or facilitate perceptions, then Naiyāyikas should not accept that.  Several contemporary scholars have offered technical arguments in defense of Nyāya on this point.  Phillips and Chadha have both argued that an imagining of a structured thing, itself acquired from past experience, though it is a necessary condition for making a mere, unstructured sensation a structured perception, is not the latter’s instrumental or chief cause, which is the sensation itself, and so no incoherence is done to the Naiyāyika position on this matter.  However, it is perhaps the classical authors, the most groundbreaking of which was Vācaspatimiśra, who provide us with a more detailed picture of how this mingling of sensation and imagination can enable us to experience the layerdness and commonality of real things that seems to permeate our awareness.

First of all, he asks whether the Yogācāra conviction that sensations and imaginings really follow two utterly separate causal streams.  If this were the case, Vācaspati argues, then two other postulates that the Buddhists are committed to, namely that some causal relation (the ālambana or “basing causal relation”) does exist between sensations and imaginings and that imaginings can impel us to actually successful activity, could not hold good.  Furthermore, he points out that sensory cognition (indriyavijńāna) can be evoked by a particular image (vikalpa), when, for example, upon seeing a color that reminds us of a similarly-colored candy that tasted sweet, our present sensation of the color is accompanied by a sensory anticipation of sweetness.  In this way, the memory, in the form of a previously sensed image (kalpanā) can enhance our sensory experience, and by so doing, make a merely sensed feature (orange) into a robust perception (this is an orange, sweet M&M).  And when we reach for it and eat it, our sensation, aided by our imagination, has helped us correctly identify a thing sensed.  Not every imagined experience recalled through memory will enable us to correctly identify our sensations through such sensory enhancement, but it is not at all difficult, in turns out, to enumerate many that do.  While the commonality of an immediately sensed feature may not be given through the sensory experience in its first instant, then, previously experienced sensory images preserved in memory which provide “marks” through which commonality is detected can be triggered that add to the sensory experience and help us rightly identify their objects.

One will notice that this narrative also suggests how imaginations can assist us in identifying the structure or layerdness of a given sensation.  To continue with the snacky example of the previous paragraph, the initial sense-organ contact presents us with a color, or perhaps a color that has a roughly round shape.  The sensation triggers a memory in which we enjoyed a piece of candy with the same, or strongly resembling, color or shape, and in addition to the added sensation of sweetness that the memory adds to it, we are able to identify the object of our sensation as an M&M, and we have the robust perception of the complex and predicative sort that we are so familiar with in many of our other experiences.  Again, there is no guarantee that every such structured perception will be correct, will be in accord with the thing before us; we might reach for the treat and discover that it’s an awful Resees Piece instead, , or maybe even something totally unexpected, a large necklace bead.  But there are also many times when the sensation-imagination complex will make us rightly aware of the candy in front of us, will help us identify the thing to which a feature belongs.

This formulation of the intermingling of sensation and imagination that helps give rise to robust perceptions was criticized by D.N. Shastri long ago, and echoes of his critique still reverberate in some quarters of modern scholarship.  His biggest complaint seems to have been that, with this explanation, Vācaspati has abandoned the already established Nyāya position (siddhānta) that perception arises from sense-organ contact with the object alone and is not brought about by another cognition.  Surely while abandoning a previously held scholastic position could be cause for censure and defeat in a classical Indian debate, there are several issues that mitigate this criticism.  While early Naiyāyikas did seem to believe, as noted above, that veridical sensory cognitions  were determinate from the beginning since their objects were in fact structured substances, we have also seen that their original definitions did continence the possibility that indefinite, vague sensations could occur that required aid  from additional sensory cognitions to properly identify.  This allowance may have left enough room for not only Vācaspati but for others to refine the theory of how perceptions are brought about without smacking of necessary systematic incoherence.  However, even if Vācaspati’s thematization of the mutually supportive activities of perception and imagination may not in the end be perfectly reconcilable with Gautama’s and Vātsyāyana’s, insisting that original definitions be either immovably committed to lest one surrender one’s entire worldview would obviously make classical Indian thought in its practice nothing more than a face-off of fixed doctrines rather than a heritage of substantive philosophical growth that was and is bolstered by refinements and improvements rather than undermined by them.  We should no more hold Buddhist philosophers, whose tradition itself exhibits extraordinary and powerful intellectual hybridity, to such an inflexible precedent than we should Naiyāyikas or any other seekers of truth.  What ought to matter is whether or not the arguments, and the representation of the world and our own nature they support, are cogent or not.  And on this score, it seems to me, Vācaspati’s explanation of perception, and how sensations can be aided by imaginings in correctly identifying things, is superior to that of his successor Gaṇgeśa, who, on this matter, obfuscates the issue of how we can cognize unlocalized features of the world more than he clarifies it.  In Vācaspati, we find an exemplary instance of an Indian philosophical theory that empowers imagination not just to project its desirous obsessions through Vasubandhu’s mere intentions or construct the world out of Śańkarācārya’s cognitive magic, but to enrich and enable our experience of living in the actual world.

About Douglas Berger

Douglas Berger is Professor of Global and Comparative Philosophy at Leiden University in the Netherlands. He is also the former president of the Society for Asian and Comparative Philosophy and the current chief editor of Dimensions of Asian Spirituality, a book series published by University of Hawai'i Press

7 Replies to “The Role of Imagination in Perception”

  1. A lot of great stuff to think about here, Doug. A few initial ideas in continuation of your fecund reflections:

    1. You illustrate just how vague the boundaries are between perception and inference (etc.), and, I think that it is hopeless to create official, absolutely clear boundaries between them. The prAcIna-naiyAyikas recognize that perception is “informed” by prior cognition (if not officially caused by it in a strict sense), and it is this fact which allows error to happen in the first place, when the “informing” cognition leads to a perceptual misfire (e.g., seeing that post in the distance as a man).

    The sort of cognition to which you call our attention seems to be at the boundary of perception and inference, as the target of cognition is indeed perceptual, as by definition it is scoped by one’s eyes. But the content exceeds what is given in perception proper (the sweetness of the candy, which is somehow “there” as one sees it.) We know the other examples of seeing snow as cold.

    The richer one’s background information, and the more one is able to bring it to bear in perception allows more inferentially-rich perceptual content, which yet remains perception. That is, since perception is informed by cognition, etc., then the more sophisticated our conceptual background is, the more we can inform perceptual content, while allowing it to be perceptual, and not mere inference.

    2. To underscore a point you hint at: In the background of your candy example is the fact that for nyAya, we can see things. We can see substances, which is a central feature of nyAya realism (c.f. sutras 2.1.33-6, embedded in the discussion on perception). This point is the main reason, I would guess, that imagination is helpful. Imagination helps us get things right, because there are things we can get right! If we couldn’t see objects like candy, then the fact that the rich cluster of cognitive an affective states that we call imagination causes us to see something as candy would be unhelpful for nyAya.

    In relation to this last point, one of the most disappointing parts of Dignaga’s PramANasamuccaya is his response to this issue. A realist says “wait, we can see that the same thing can be the locus of various sense presentations (e.g., we can touch and see a wooden block, and feel the edge as hard and flat), and thus we see substances (we indeed see a block, not just a collection of sense presentations which we then cluster together in imagination).”

    His response is that by definition, this is impossible, since difference sense modalities by definition cannot be the “same thing”. End of story. Maybe I’m just a simple-minded realist, but this can’t be the end of the story.

    3. I will overlook your aparAdha toward Reese’s Pieces this time, but next time will require prAyazchitta.

  2. Thanks for reading, Matthew, and all good points I agree with. The definitionally strict separation between perception and inference, with perception being completely uninformed by imagination, is indeed what I think gets Dignaga’s position into great difficulty. Dharmakirti sometimes tries to soften it by arguing that imagination constructs its commonalities through repeated perception of similar but not identical objective content. But, being a simple-minded realist myself, I think Dharmakirti’s revisions are an improved defense of a still untenably strict division between perception and imagination.

  3. very interesting post here. just regarding the “realism” of naiyāyikas that you (and matthew) seem sympathetic to: does it consist in the affirmation that general (conceptually structured and linguistically identifiable) properties are present in the object, or that the processes by which our imagination imputes these properties to the object (on the basis of memories, etc.) are reliable and deserve to be included under “perception” as much as the collection of adventitious sense-data? or some combination thereof—i.e., our imagination is responsible for “putting together” (einbilden, if you’re a Kantian) representations which we ultimately derive from our experience of objects?

    • The former. And yes, I am sympathetic to it, or at least to its motivations and core rationale.

      In fact, I would say that for naiyAyikas in general, the affirmation of pre-conceptual cognition is actually an affirmation of their realism, not at odds with it. If I may just rip something out from something I’ve written before (in a book review for PEW):

      “Ironically, perhaps, Nyāya’s acceptance of non-conceptual perception is thus meant to support the view that our perceptually-relevant concepts ultimately come “from the outside”—serving the opposite role of that given to non-conceptual perception by its Buddhist champions. Phillips’s further discussions of recognition, perceptual error, and apperception underscore just how powerful Nyāya takes perception to be: we have the capacity to perceptually recognize an occurently perceived object as the same object seen yesterday. We have the capacity to perceive absences, which, like recognition, are informed by, but are not reducible to, memory of previously experienced objects. And it is because of this profound capacity of perception to directly apprehend things, properties, our own mental states (through apperception, a subspecies of perception proper), substances, and indeed, relationships of various sorts, that Nyāya empiricism, as Phillips points out, does not collapse into idealism, as have other empiricisms, both Indian and European (44).”

      That perception of absences is also something richly informed by prior concepts/imaginings goes back to Vatsyayana’s opening comments on the nyAyasUtra itself.

      –Sorry if I preempted you, Doug. Please step in and if needed, nuance/correct what I said!

  4. I realize this is outside the precise scope of the post (which was very well done), but it’s an intriguing fact that within the four major schools of Sanskrit poetics (Alaṅkāra, Rīti, Dhvani, and Rasa), according to V.K. Chari, imagination (pratibhā, ‘poetic genius’) is not used in the definition of poetry, “although nearly all critics paid homage to it.” Indeed, pratibhā is simply cited as “only one of the causes of poetry, together with training (śikṣā) and understanding of the world (vyutpatti).”* Chari himself thinks what others see here as a failure to do justice to the intuitive or imaginative parts of poetic creation is rather an analytic virtue of the scholasticism of Sanskrit critics, for imagination “is at best a dubious concept, and its usefulness for criticism has not been proved.” The second half of the coordinating conjunction is likely true, although I disagree with the proposition that imagination is “at best a dubious concept.”

    Is this lack of philosophical interest in imagination here owing to different conceptions of the concept as employed in philosophy and “Sanskrit criticism”?

    * Sanskrit manuals, writes Chari, make a firm distinction between “the cause of poetic creation (kāvya-hetu), the ‘fruits’ accruing from it (kāvya-phala) or the purpose served by it (kāvya-prayojana), and the nature of poetry (kāvya-lakṣaṇa).”

  5. I noticed that S.S. Barlingay distinguishes between two types of pratibhā in Indian aesthetics: creative or kārayitrī and cognitive or bhāvayitrī. It seems Chari is speaking to the former, but I could be mistaken. Is anyone familiar with this distinction and its possible relevance to this discussion in Indian aesthetic theory/philosophy of art?

  6. hi man – sorry to ask you this but could you remove my post – as am having old problems when i post on ultimate subjects – thanks

Leave a Reply

Your email address will not be published. Required fields are marked *