Perception, Language and Cognitive Success
Perception, Language and Cognitive Success


I compare three types o perception: perception X, perception as…, and perception that… to decide which type is cognitively the most promising. Commonly, perception that… is taken to be epistemically superior as both propositional and factive. I argue in favour of perception as… which implies revealing new aspects. I analyze two standard cases: the ‘bent stick’ illusion and the ‘duck-rabbit’ picture to explain how seeing new aspects engages language usage, transcends factivity and becomes heuristic.

Table of contents

    It is deceptive to claim that by perception people acquire information about the world since a source of information is one thing and an agent who is to consume information is another: produced information needs not to be consumed. By consuming information I mean achieving a cognitive goal and I want to decide which type of perception entails genuine consuming. I ask first if linguistic interpretation of perceptual information is the sufficient condition of a cognitive goal. This issue is obliterated by the controversy over whether visual perception represents the world and, therefore, brings information about it (Crane 2006). Even if we think that visual representation is passive, i.e., the agent is unable to modify the mental picture delivered by his eyes, and that passive representation gives him the optimal information about the world, from the fact that some information has appeared it does not follow that any information has been absorbed (Sosa 2007, 46).

    The opposition between non-epistemic and epistemic perception only minimally explains what is necessary for cognitive success. A typical distinction is that between conceptually loaded perception: seeing that p or seeing X as…, and conceptually empty perception: seeing X (O’Brien 2004). It leads immediately to interpretational difficulties. Propositional seeing that… is classified as factive. Whether non-propositional seeing X is factive, i.e., adequately represents the world, becomes dubious even if we assume that the world is ontologically stable, e.g., it consists of mentally independent facts.

    Suppose that opening his eyes person A acquires a picture of falling rain and the rain is actually falling. If A is a child unable to make the statement: “The rain is falling” we would diagnose that A sees X (X = falling rain) but A does not understand what he sees: A neither sees that p (p = the rain is falling), nor A sees X as X, i.e., falling rain as falling rain. His eyes have delivered a picture, but this non-interpreted picture remains cognitively useless. In a similar situation, of a competent speaker we would say either that he sees that the rain is falling or he sees falling rain as falling rain. The child’s visual picture merely represents the world, the speaker’s picture seems to be cognitively consumed. The question I ask is: Is such a consumption satisfactory?

    Considering the property of factivity helps to realize why in cases of simple perception the agent cognitively fails. Of propositional perception that… it is easy to say that it is factive, i.e., if A sees that it is raining, it is true that it is raining. But to say the same of perception X is problematic. If A’s perception of falling rain is non-propositional we should not predicate the truth about it. The conditional ‘If A sees falling rain, then it is true that the rain is falling’ is epistemically strange at least for proponents of the view that truth can be ascribed only to propositions. There are therefore two things: to see a fact and to see a fact propositionally. I think that the term ‘factivity’ should be limited to linguistically interpreted perception: perception X is then beyond factivity.

    Similar doubts concern fallibility. When we say that if A perceives, then A makes sometimes mistakes, i.e., A’s perception does not agree with facts, the question arises: What type of perception is fallible and how? Fallibility and perception that… exclude each other. Certainly, we can safely speak of perception unrelated to facts when we discuss cases of perception as… . There are then two things under suspicion: a visual picture and its linguistic description. Surprisingly, if we understand fallibility as the possibility of making a mistake the fallibility of perception X becomes dubious. We can perhaps say of a hallucinating child who passively acquires a picture of falling rain that he is mistaken but surely not that he makes a mistake. Of a person entirely passive we should not, therefore, say that his perception is fallible. Making mistakes requires some cognitive activity, e.g., linguistic interpretation. Thus, only perception as … allows for two types of fallibility: (1) making factual mistake by accepting as the representation of fact p a picture not related to p and (2) making linguistic mistake by interpreting the picture related to fact p as representing not-p.

    These proposals are themselves disputable as there are no clear ideas about passive components of perception. But I think that the possibility of making another kind of mistake is worth considering: I call it “the mistake of ontological interpretation” as it follows from the view that successful perception should adequately depict the world. I suggest below that we had better follow the hedonistic view that successful perception should help us to prosper. I am interested in situations where someone interprets visual representations truly but uselessly. I am going to stress that ontologically oriented epistemology often promotes making ontological mistakes by concentrating on cases of factive perception which do not entail cognitive success.

    Typically, one distinguishes between perception that… and perception as… in order to proclaim inferiority of the latter. A case of illusion recalled in this context is that of a stick submerged in water. It is said of this situation, I think wrongly, that person A sees a straight stick as bent. In my opinion, all depends on whether he simply sees a bent stick, or he sees that stick as bent. In the first case he is only mistaken, in the second he also makes a mistake. In the first case we have passive perception, in the second we have active perception. An important question is which mistake can be overcome. I do not think that a passive mistake can be corrected. In the case of interpreted perception as … the mistake can be overcome only if A refers to an explanation involving his linguistic competence, i.e., being a result of inference or instruction. We are then justified to say of A that although A cannot correct his visual picture of the stick, he can correct his belief about its shape. I think it is essential that without an additional linguistic explanation A would be unable to correct his belief and the possession of this explanation does not change A’s simple seeing: A still has the picture of a bent stick, although A believes now that the stick straight. A is continuously deceived by the picture, though he ceases to make a doxastic mistake.

    We are in trouble, however, if we want to decide what to say of person A who believes that the stick is straight. Agreeing that he has a visual picture of a bent stick, we should say that he has a deformed picture of a straight stick. But what does his perceptual belief relate to? Can we say that it relates to A’s seeing that the stick is straight, or should we say that it relates to A’s seeing a bent stick as straight, or should we separate seeing from believing and say that A simply sees a bent stick but, thanks to an explanation, he believes that it is straight. It seems that only if the additional explanation is involved we can say of A that he sees that the stick is straight. Since it is important here that no additional explanation can correct A’s way of seeing the question arises whether A’s true belief that this stick is straight remains a genuinely perceptual belief.

    When we define perceptual beliefs as those concerning the perceptible features of the world we rely on the ontological assumption that the world has a content that can be causally transformed into a phenomenal content. The phenomenal content waits as it were for an adequate linguistic interpretation, interpretation that is in principle passive. For example, if person A sees falling rain, then he acquires the propositional perception that the rain is falling only if he depicts the fact of the falling rain by the proposition: ‘The rain is falling’. To destroy the myth of passive description philosophers sometimes refer to the ‘duck-rabbit’ picture (O’Brien 2004). The question asked is to what a degree A’s perception depends in this case on A’s linguistic competence. Traditional foundationalists are linked with the view that the content of perceptual experience can always be non-conceptual, i.e., pre-linguistic. In reaction, it is argued that the phenomenon of ‘seeing as’ sufficiently proves the conceptual nature of perceptual experience. The ‘duck-rabbit’ picture is tricky, though, as it is hard to decide (1) whether it allows for simple perception and (2) what follows from the fact that it does not allow.

    Can we reasonably ask: What does a pre-linguistic child see: a duck, a rabbit, changing aspects of a duck and a rabbit, or a duck-rabbit? If his perception is passive, must it be stable or can be dynamic? If stable, should his perception be monistic or dualistic. And so on. Further, is it possible to be mistaken in this case? Is it possible to make a mistake in this case? Can one obtain a perceptual success in this case? Are the above questions crucial, or just pathetic? The anti-foundationalist theory suggests that simple perception is an empty hypothesis and the lesson to be learnt from the ‘duck-rabbit’ picture is that every perception is conceptually loaded. It explains that only a person who has the concept of a rabbit can see the ‘duck-rabbit’ picture as a rabbit and, further, that the dynamic change of aspects is possible only if the person has both concepts: he must think of a duck to see the duck-rabbit it as a duck.

    Unfortunately, this alternative theory provokes analogous difficulties: Do I see the changing aspects only because I am unable to construct the concept of a duck-rabbit? If I had the dualistic concept, would my perception stabilize? Is there anything ontologically determined on a lower level of this picture, i.e., Don’t I see the same configuration of black marks? If seeing and interpreting are inseparable, can perception be cognitively unsuccessful? The thesis that we see only through concepts loses some of its attraction when we face them. To conclude, I do not think that the ‘duck-rabbit’ picture is ontologically illuminating. Nevertheless, I believe that epistemically it surely is if it helps to understand that ontological theses concerning perceptual representation are cognitively misleading. Philosophers dominated by the ontological perspective persistently ask about the possibility of representative seeing and then it seems to them, by analogy, that the basic epistemic question concerns the possibility of representative perceptual beliefs. To counter this perspective I would like once again to focus on two theses: (1 ) linguistically interpreted perception can still be cognitively useless and (2) the cognitive usefulness of perception as… is substantially higher than perception that… . Thus, I am going to switch from the ontological controversy over whether genuine perception is conceptual to the purely epistemic explanation in what way conceptual perception becomes cognitively successful.

    It is reasonable to think that an epistemic counterpart of the problem of simple perception is the problem of non-inferentially justified beliefs because foundationalists usually hope that perceptual beliefs are both non-inferential and sensitive, safe or at least apt (Sosa 2007, 98-105). They either maintain that phenomenal beliefs such as: ‘It looks to me as if the rain is falling’ are non-inferential and infallible or that objective beliefs such as: ‘I see that the rain is falling’ are non-inferential and prima facie justified. Critics reject the first position because it fails to explain how phenomenal beliefs can justify objective beliefs. The second position looks promising only for those who think that un-defeated perceptual beliefs are self-sufficient (Audi 2002, 83-87). Anyway, respect for perceptual beliefs, preferably factive, is quite common.

    If asked of the type of perception that should be particularly favoured, the foundationalists answer that perception that… is incomparable. Actually, in reaction to Gettier counterexamples, they are engaged in specifying conditions under which perception that… becomes knowledge, i.e., is effectively un-gettierized (Alston 2005, 21-28). As long as they automatically prefer perception that… to perception as… their thinking remains ontologically oriented: they value perception that p for its factivity and imply that perceptual belief that p is perfect when determined by the fact p.

    What disturbs me about the phrase ‘A sees that p’ is its easy reduction to banal ‘A sees X as X’. For example, ‘A sees that the rain is falling’ seems no more informative than ‘A sees falling rain as falling rain’. What is heuristic in seeing falling rain as falling rain? If nothing, then it is worth stressing that perception that… is in most cases non- heuristic. Heuristic perception must resemble the duck-rabbit perception insofar as it must reflect the agent’s switch to different aspects or his seeing new aspects. Only seeing object X as object Y, i.e., seeing X as something else, or seeing X as a sign of something else can be heuristic.

    To end with the case of person A seeing falling rain, it is intuitively obvious, I hope, that rain rarely is a neutral fact beyond positive or negative evaluation: as something pleasant, desirable or unpleasant, dangerous. If rain were neutral, heuristic seeing of rain would be inessential. Suppose that A lives by a river and the rain has been continuously falling for a week but every time when A looks out of the window he sees that the rain is falling and nothing else: he is unable to see the falling rain as a sign of coming floods. Or suppose that A has a garden, there has been no rain for weeks and when it starts raining at last all that he sees is that the rain is falling: he is unable to see the rain as a sign of the ending drought. Is then A’s perception that … a cognitive success? Can we say that A cognitively consumed some perceptual information? In such contexts, perception that… seems to be heuristic only if it is an abbreviation of prior perception as…, i.e., when A says: ‘I see that floods are coming’, or ‘I see that the drought is ending’.


    1. Alston, William 2005 Beyond “Justification”, Ithaca: Cornell UP
    2. Audi, Robert 2002, “The Sources of Knowledge”, in: Paul Moser (ed.), The Oxford Handbook of Epistemology, Oxford: Oxford UP, 71-94
    3. Crane, Tim 2007, “Is there a Perceptual Relation”, in: Tamar Szabo Gendler and John Hawthorne (eds.), Perceptual Experience, Oxford: Oxford UP, 126-146.
    4. O’Brien, Daniel 2004, “The Epistemology of Perception” The Internet Encyclopedia of Philosophy
    5. Sosa, Ernest 2007 A Virtue Epistemology, Oxford: Clarendon Press
    Tadeusz Czarnecki. Date: XML TEI markup by WAB (Rune J. Falch, Heinz W. Krüger, Alois Pichler, Deirdre C.P. Smith) 2011-13. Last change 18.12.2013.
    This page is made available under the Creative Commons General Public License "Attribution, Non-Commercial, Share-Alike", version 3.0 (CCPL BY-NC-SA)


    • There are currently no refbacks.