Wittgenstein's Logic of Language | Bibliography | Search

Gestalt Shift

Wittgenstein wanted to say "what is involved in seeing something as something", for example seeing a single drawing either as a drawing of a rabbit or as a drawing of a duck (the duck-rabbit), but he told Drury this is not easy to do. Wittgenstein was talking about logic, not about psychology, but I don't quite know what it was that he wanted to say philosophically about this phenomenon. There are many aspects to the discussion, however.

Philosophy as Gestalt shifts

A philosopher says: "Look at things this way!" (CV p. 61, a remark from 1947)

And, I think, You must see the world afresh, and that means seeing a new world. To understand a philosopher is to experience a universal Gestalt shift. But those are metaphors and we must test their limits.

What is this Gestalt shift in the case of Wittgenstein's later work, his logic of language? It is not only learning a new technique for looking at language [a new logic: method, metaphors [comparisons], definitions, rules for reasoning in philosophy], it is also the feeling that you no longer have to be the victim of language; that you can take control of language. In other words, the feeling that sense is determinate; that meaning is something it is possible to determine; that an objective distinction can be made between sense and nonsense (where 'nonsense' = 'undefined' and techniques are given for defining words, phrases, etc). Before the shift you feel yourself to be at the mercy of vagueness and confusion. Because before the shift, language, rather than you yourself, is in control of your thinking.

The old picture, the false grammatical account: that an abstract noun somehow has meaning in itself -- not because it is a word with a use in our life -- but because it denotes an essence that is independent of language -- whether anyone is able to say what that essence is or not (and therefore there are only opinions about what the "real" or "true" meanings of words are).

The term 'abstract words' or 'abstractions' comes from the expression 'theory of abstraction' -- i.e. it comes from a metaphysical theory, not from the facts that we know (but from an unverifiable thesis about an imagined reality hidden behind those facts). And that in itself is good reason not to use the term 'abstract words': "An unsuitable type of expression is a sure means of remaining in a state of confusion. It as it were bars the way out" (Philosophical Investigations § 339).

I should not expect to understand everything -- or, rather, to come to an understanding of everything -- that a philosopher says at first blush -- i.e. before its meaning is explained to me (or before I explain it to myself). But I should expect that an objective explanation of the philosopher's meaning can be given; that I can hold myself and others to the standard: "Every explanation I can give myself, I can give you too" (PI § 208).

Outline of this page ...

The following is a question ("search query") from my site's logs:

Query: Gestalt shift - what is it?

We can see the two-dimensional figure below as a [three-dimensional] cube. There are [at least] two ways to do this, and the switch between the two is called a 'Gestalt shift'. It is not a voluntary shift, although one can try to bring it about.

Cube example of a Gestalt shift

If I fix my eyes first on the corners a and only glance at b, a appears in front and b behind, and vice versa. (Tractatus Logico-Philosophicus 5.5423, tr. Ogden)

We [i.e. most people, or, It is the common experience that we] can see this figure as two differently oriented cubes. That is, I can see either face a, a, a, a as the face of the cube that is nearest me (at lower left), or I can see face b, b, b, b as the face of the cube that is nearest me (at upper right). The visual experience of seeing it first one way and then another way is called a 'Gestalt shift'. The shift is actually more like a snap; it is not gradual but instant; and it may shift back despite our unwillingness for it to do so.

In the case I have in mind, however, with respect to Wittgenstein's logic of language's new point of view, is that it seems impossible to shift back -- i.e. to see [experience] philosophical problems in the old way. Although any picture (e.g. "spirit versus matter") may awaken confusion in us, we nonetheless see the -- i.e. a -- way out of that confusion: "a method has been found".

The Duck-Rabbit

In the Philosophical Investigations Wittgenstein uses a figure called 'the duck-rabbit' (derived from Jastrow), which can be seen either as a duck or as a rabbit, depending on where you direct your attention (focus your eyes). In either case you see an eye, but then either a duck's beak and creased skull or a rabbit's ears and mouth (The rabbit is looking skyward, the duck straight ahead to the left).

Duck-rabbit example of a Gestalt shift

Gestalt is a German word for 'shape' or 'form' in English. We can say that a "form shift" is: an involuntary perceptual jump from one shape [or, form or pattern] to another, e.g. from the duck-shape to the rabbit-shape. By 'Gestalt shift' we may mean: seeing a single pattern now as this, now as that: e.g. now the duck, now the rabbit; this is an involuntary visual shift from first one "aspect" (Wittgenstein-Anscombe's word) or form/shape to then another "aspect" or form/shape, without the pattern itself changing: all that changes is our perception of the pattern: asked to draw now the duck, now the rabbit, we will produce identical drawings.

I have written nothing about the duck-rabbit image (See PI II, xi, p. 193 ff.) because I find Wittgenstein's presentation of his ideas in this case so clear that nothing I might say would help make it clearer. But perhaps this is also because I am not perplexed by -- or not too interested in -- psychological phenomena: "Now you see a duck looking straight forward; now you see a rabbit looking skywards; the duck's bill has become rabbit's ears". But what is there that is philosophically interesting here -- except from the point of view of a possibly confusing metaphor, for example that of N.R. Hanson ("Seeing is theory laden") -- or mine in saying that to understand Wittgenstein's logic of language requires a Gestalt shift: "In a large class of cases [i.e. in those cases where the word is not defined ostensively] a word's meaning is its use in the language" versus "The meaning of a word is the thought or idea it stands for"? Look at the search query above: the question "What is it?" versus the question "What does it mean?" [i.e. "How do we, as common practice [acceptation], use the combination of words 'Gestalt shift'?"]

In 1949 Wittgenstein showed Drury the drawing of the duck-rabbit and said:

Now you try and say what is involved in seeing something as something; it is not easy. These thoughts I am now working at are as hard as granite. (Recollections p. 159)

He had earlier written to Rush Rhees:

My lectures aren't too terribly bad but they're are pretty poor. I'm talking about problems of Gestalt psychology and am frightfully unclear myself and unable to get the deep aspects of the matter. (Wittgenstein in Cambridge: Letters and Documents 1911-1951 (2008), ed. McGuinness, Document 348, Letter dated 7.2.1946, p. 395.

I myself was never able to see "the deep aspects" (cf. PI § 387). Perhaps I should be perplexed here, but I am not. The shift phenomenon, the way we are able to focus on an area of the duck-rabbit drawing and in this way have what we see the drawing as shift from one "aspect" to another -- i.e. "Now I see a duck, then suddenly I instead see a rabbit" -- does strike me as psychologically interesting. But philosophically? On the other hand, Wittgenstein later wrote about "concept blindness", which is a metaphor for the inability to see one or another aspect (e.g. being able to see only the rabbit, but not the duck [which would be "aspect blindness" with respect to the duck-aspect]). What exactly is to be done with this metaphor, I don't know. Maybe it is an example of philosophy as clarification by full-stop (M. O'C. Drury, The Danger of Words (1973), p. xii). N.R. Hanson reproduces a drawing which he says can be seen either as a young or as an old woman. I myself have never been able to see the drawing as a young woman.

But what philosophical implications that visual fact has -- i.e. that not everyone [or even that no member of the human species] can see every aspect or (acquire every concept) --, I don't know. But it is connected to the notion of "forms of life" and to "The limit of thought [in philosophy, science, mathematics] -- is concept formation". It is a powerful metaphor for that.

Old Woman-Young Woman

Drawing of an old or young woman?

But when Hanson adds that if one looks at the drawing "à la Toulouse-Lautrec" (Patterns of Discovery (1958), p. 11) then one sees a young woman ..... I wonder if in this case it is useful to speak of a Gestalt shift. Certainly if you add Toulouse-Lautrec, then you do encourage a shift of sorts -- i.e. a particular way of looking at things, at a human face e.g. ("Yes, now I can see her as a young woman") ..... I would not call this an example [instance] of Gestalt shift (even if there is a "family resemblance" to that, because ... as if resemblances were ever hard to find; indeed, a limitation of Wittgenstein's metaphor is that resemblances are only too easy to find). Hanson has provided a context -- whereas a Gestalt shift should be raw [spontaneous]? But by labeling the cube above, does not Wittgenstein also place that drawing in a context?

When I wrote the two paragraphs below I was experiencing "aspect blindness" or inability to see one aspect of the Gestalt. (How to see both aspects.) However, not everything in the next two paragraphs is incorrect; it is not clear why Hanson says what he does about the drawing.

What I would say is that in the case of Hanson's "old woman-young woman" drawing, there are not two aspects, because both the young and old woman are absolutely identical -- i.e. you cannot show Hanson's distinction by pointing to specific places on the drawing as you can in Wittgenstein's examples: "Focus on 'a', now on 'b'" and "Here is a duck's bill; but now, there is a rabbit's ears". Instead, in Hanson's example you look at the drawing with Toulouse-Lautrec's posters in mind (and easily see a young woman), or without Toulouse-Lautrec (You see an old woman). That is, to call Hanson's example an instance of Gestalt shift has a blurring effect: it dissolves an otherwise sharp distinction.

In sum, if you allow yourself too much freedom in how you apply the expression 'Gestalt shift', then any drawing at all might be used as an example (You have only to provide a context as Hanson does). Thus, considered as a tool, it would be more useful to apply the expression 'Gestalt shift' only to cases where the viewer experiences a snap or jump from one aspect to another. In Hanson's "old woman-young woman" drawing, the viewer experiences no such snap.

How to See the Young-woman Aspect

With regard to the old woman-young woman, a correspondent wrote that "it is as most certainly a Gestalt shift as the duck-rabbit. The old woman's eye is the young woman's ear, the underside of the old woman's nose is the young woman's jaw line, and the old woman's smile the young woman's choker around her neck. The young woman is looking away, her neck exposed." (Anonymous, 30 September 2009)

Old woman-young woman example of a Gestalt shift

In other words: The end of the old woman's nose is the young woman's chin. The young woman's face is turned away from us, and the old woman's eye is the young woman's left ear, but the hair is the same for both aspect-women. (That, of course, assumes that the old woman is the aspect that resembles a poster by Toulouse-Lautrec.)

I myself could, at first, see that young woman only after I used a colored pencil (to outline her face and neck) and an eraser (to remove much of her hood and dress), but I can thus see her. Although then I do not see the point of Hanson's adding "à la Toulouse-Lautrec". It is possible, in any case, to simply see the drawing either as an old woman (in our normal style of drawing) or as a young woman (in the style of Toulouse-Lautrec) without any need for a Gestalt shift. And either way of looking at the drawing serves Hanson's purpose, I think, but there is some broadening of Hanson's point if the drawing is looked at in the latter way -- i.e. without any Gestalt shift.

Query: drawings where two different images can be seen but not at the same time.

That would belong to the defining characteristics of 'Gestalt shift', or, the criteria for classifying any particular image as an example of a Gestalt shift.

Query: Gestalt concepts overlap.

What might the query mean by the word 'overlap'? The duck and rabbit aspects do not overlap (1) in the sense of 'overlap' that one aspect partially covers up the other, unless that simply means that only one aspect can be seen at a time, or (2) in the sense of 'overlap' that they have something in common, unless that simply means that they are both aspects of the same image.

Is it not an important point that that they don't overlap? that they are different concepts, not inter-connected but only externally connected -- i.e. nothing would be lost by the duck-aspect if there were no rabbit-aspect, or any other aspect besides the duck-aspect, to be seen in the duck-rabbit image, just as it is logically possible that ducks might exist without rabbits also existing. (Cf. although the part of speech of both the words 'duck' and 'rabbit' is name-of-object, the objects that are pointed to when those two words are ostensively defined are different: the meaning [use] of one is not dependent on the meaning of the other).

Is a Gestalt shift image ambiguous in meaning? Maybe we could say that, if the old woman or the duck is one "meaning", and the young woman or the rabbit is another. And if the same image can have different "meanings", then it is an example of what we mean by the word 'ambiguous'. (Then maybe it would be the case that 'a Gestalt switch image' = 'an ambiguous image'.)

Query: in between Gestalt switches.

What characterizes this phenomenon, and is used to define it -- i.e. to define ("to limit", "set boundaries to") the concept 'Gestalt shift' -- is that there is no "in between", i.e. no transition from seeing one aspect to seeing the other; it happens in an instant (and thus expressions such as 'Gestalt snap' or maybe 'Gestalt flip' may be particularly apt). Note that 'Gestalt shift' is defined by psychological not physiological criteria, by testimony rather than by e.g. events in the central nervous system.

Gestalt Shift as Simile

That was how I began this page, when I used the expression 'a universal Gestalt shift', with a comparison. The form of a simile: A is like B. But that comparison is undefined language until it is stated in what way A is like B. (And that statement may be true or false, because A may or may not be like B in that particular way. On the other hand, some similes are suggestive rather than precise, suggestive or too vague to make anything clearer.)

Query: seeing aspects, Wittgenstein, religion.

Gestalt shift would in this case be a metaphor, like 'concept-blindness' (Wittgenstein's remark about blindness to the concept 'God'). But how is that like the duck-rabbit? Suppose it were the case that some people could (as perhaps some really can) only see one or the other aspect. But how is being unable to see the rabbit-aspect (e.g.) to be compared with not being able to understand how any reasonable human being could seriously use the word 'God' (how the concept 'God' may have an important place in a reasonable man's thinking)? Can one be blind to the religious aspect of life?

Some similes are suggestive -- suggestive or too vague to make anything clearer. Which is this?

"Seeing is theory-laden"

Seeing a bird in the sky involves seeing that it will not suddenly do vertical snap rolls ... This is knowledge. (N.R. Hanson)

And not asking the geese in the park for directions and the time of day. We may perhaps say that this is knowledge (an example of what we call 'knowledge'), but what relationship has this to seeing? Would we say that there is any theory-laden seeing here: "If you see a goose, you also see that ..." or "If you see a goose, you see at the same time that ..." or "Seeing a goose is seeing that ..." Perhaps Hanson's point is that: there are not two acts here, only one.

Does making a connection to perception make something clearer here; what does Hanson accomplish by making this connection? Is the connection that normally we don't think about these things: I do not normally deduce: "That is a goose, and geese do not bite, and therefore I need not be afraid that this goose will bite me." Although I might, if asked, give this deduction as a justification for my behavior, my behavior can be compared to seeing in the sense that it can also be compared to instinct: I act without thinking. I do not say "This is a staircase and ... and therefore I must lift my foot if I wish to climb it."

When I see a goose I see an object about which I know many things -- 'I know' means that I have many expectations, and if I am on familiar territory, then I have a clear attitude toward anomalies ... or have I? I do not know what I would say if just any such-and-such [prodigy] occurred. We do not live in a world where such things happen very often (Indeed they are quite rare).

Was Hanson stating a metaphor [making a comparison, an analogy]? "In science we want to look without pre-conceptions." -- But conceptions and perceptions are not the same thing: they are different concepts. Hanson wanted to point to the areas where these concepts overlap, where there is no clear distinction between a conception and a perception. However, I do not believe you can say that these concepts overlap everywhere [i.e. in all cases where we apply the words 'percept' and 'concept'], as "seeing is theory-laden" seems to suggest.

Comparison between Kant and Hanson: but Hanson's pre-conceptions can change, whereas Kant's ["time", "space"] cannot (They belong to the essence of human nature). Is this correct?

Query: Gestalt switch in science.

A scientific theory is a selection of facts plus imagination: imagination supplies the way of organizing -- i.e. looking at -- a selection of data. I won't take the query in the sense that I spoke metaphorically of a Gestalt shift in philosophy (metaphorically because the shift is not visual), but ask whether at least sometimes a change in the basic model of scientific understanding is a non-metaphorical Gestalt shift. For example, did Darwin see animal life differently from those who used the model "that species were all independently created and are immutable", e.g. looking at fossils: seeing a fossil as a stage in the evolution of a species? Did Einstein see events in the sky differently from Newton with the model "relative space frames" rather than absolute space (what the eye of God would see, i.e. 'motion' as an absolute rather than a relational concept) -- 'see' in the visual sense? (The worth of our thinking here is only as high as the examples we can bring forward to defend or refute this thesis.)

Kant and Wittgenstein

The expression of a change of aspect is the expression of a new perception and at the same time of the perception's being unchanged. (PI II, xi, p. 196a)

To express this in Kant's jargon, if I understand Kant's jargon, which I my not, although its application appears clear to me in this case: In a Gestalt shift the concept changes, but the percept does not. Here, Wittgenstein's 'aspect' would correspond to Kant's 'concept', I think.

Thus the duck-rabbit drawing (which in itself would be a "bare percept"; cf. bare sign) does not itself show the concept-change. That change can only be shown ostensively -- i.e. by pointing to areas of the duck-rabbit drawing: Don't you see, these might be a rabbit's ears rather than a duck's bill, for example, and hoping that others will now see a rabbit rather than a duck where at first they saw only a duck.

However, in the example of the cube faces above, would we say that there is a concept change as well as a percept change, or only a percept change -- for either in case 'a a a a' or case 'b b b b', there is only one concept, namely, 'cube'. But yet in that example of a Gestalt shift, we do nonetheless see different aspects of the cube.

Query: shifting realities, Gestalts.

No, shifting perceptions of reality. Because is it reality that changes ("shifts") or is it instead that our perceptions of reality change? An example of a change of reality would be if a rabbit metamorphosed into a duck; but such a change would not be what we mean by the expression 'Gestalt shift'.

Whatever we mean by the word 'reality', we do not, as we normally use the word 'reality', mean mere conceptions or perceptions. Indeed, the words 'conception' and 'perception' are the antitheses of the word 'reality', because as we normally use that word, reality is what is independent of our perceptions and conceptions of "it", whatever "it" may be. But, of course, as we normally use the word 'reality' (i.e. setting Kant's metaphysics aside), we are able to say what reality is; e.g. in the case of the duck-rabbit, reality is the line drawing itself.

If I remember his work aright, Bishop Berkeley did not go so far as to say that perception = reality (i.e. 'perception' = 'reality'). To go that far, to make that identity would be nonsense, because undefined words is what results when a word is identified with or divorced from its antithesis: both words become nonsense. (Cf. 'hot' and 'cold', 'illusion' and 'reality', 'selfish' and 'selfless'. In metaphysics, philosophers often want to say that "All A is really B", but such a claim, if it follows the pattern I have described, is nonsense.)

Query: how does a picture of a duck-rabbit show the principle of perception?

That "percepts without concepts are blind"? For if there were a concept 'rabbit', but no concept 'duck' in our language [-- for language is the concern of logic, although there may be other uses for the word 'concept' than to mean 'rules for using a word' --], then the duck-rabbit would be an image of a rabbit only?

Query: categories, Kant; form of life, Wittgenstein.

The connection would, I think, be that Kant's categories are one kind of "form of life", the kind where there is no choice [the kind where we speak of "different life forms"] -- i.e. one that is involuntary. That is, some ways of life are shared by some human beings but not others; whereas Kant's categories are necessarily shared by all human beings, because they belong to a frame of reference that is essential to the human mind.

If we wanted to extend Kant's notion of categories, we can compare the categories to Plato's Forms ("Ideas"), that is to say, the meaning of a common name would be a category -- that is to say, the common nature will be, not in the things themselves, but in the human mind, just as common nature is not in particular things that are named by the common name but is in the Platonic Form; it is the Form that is the common nature.

The difference is that Wittgenstein's account is a description, not an explanation; whereas Kant's account, like Plato's account, is not a description, but instead a metaphysical theory. (This is if I know what I am talking about, and I may not know what I am talking about.)

Philosophy as Gestalt shift (Another example)

Query: Kant's Gestalt switch.

I imagine that you could apply the metaphor, if metaphor it is, I have applied to Wittgenstein's logic of language to the work of every philosopher: a new way of looking at things replaces our old way of looking at things. But, as I intend the metaphor to be understood, this is not simply seeing a new aspect of a familiar figure, but the replacement of e.g. the duck with the rabbit, not the retention of both.

Another example of a philosophical shift can, I think, be derived from Kant's notion of innate categories of perception, namely that there is no ultimate reality as such, but only a multitude of points of reference. (That is of course a grammatical remark, a remark about defined and undefined combinations of words.)

The notion of an ultimate reality independent of points of reference results from following a grammatical analogy from part to whole, to generalize the distinction we make between reality and illusion in the particular case -- to talk about "the whole of reality" as if that were a single thing, just as a part is a single thing of a whole. (Some philosophical problems are the constructs (PI § 118) of quite simple muddles like this, but that doesn't make them easy to resolve.)

That is the origin of the notion "ultimate reality". But the origin of the notion of an "absolute point of reference", in contrast to a relative point of reference more or less arbitrarily selected to be the point of orientation for a system of relative points (as the origin of a Cartesian graph is) is different. It belongs to our picture (both religious and metaphysical) "the eye of God"; we say that "man knows only relative points -- but God knows, because he himself is, the absolute point of view".

[Cf.] here it is an important fact that I imagined a deity in order to imagine this. (ibid. § 346)

But did I in this case imagine, say, a god who sees what man cannot see? ... But even God can't see nonsense, i.e. an undefined combination of words is no picture at all, not a picture that only God can see. (Or do you think every combination of words must have a meaning, that even if mankind doesn't know what that meaning is, God does.)

Relative points contrast with one another, not with an absolute point as such.

And this is -- or would be -- an example of coming to one's own understanding of a philosopher's thought -- whether or not it is the philosopher's own understanding of his thought -- as experiencing a universal Gestalt shift in philosophy. In this case because the presumption since the pre-Socratics has been that there is an ultimate reality, a whole-thing-in-itself. (When Plato introduces the notion of "all things" in the Sophist [233e-234a], he tries to explain what he means as if it were the simplest thing in the world, but of course that is the point -- that "all things" isn't in the world, and that what applies to things in the world (the parts) may not apply to "the world" (the whole).)

[If every explanation I can give myself, I can give you too (PI §  210), then I do I know what I'm talking about here?]

Query: metaphors create a Gestalt shift in perception.

This proposition is itself a metaphor: "A metaphor creates a Gestalt shift." But a metaphor must be treated with skepticism, carefully restated in prose, the comparison's limits traced. But the query also suggests the question: is a Gestalt shift a shift in perception -- or a shift in conception? Both? (I don't mean 'perception' in the sense that experimental psychology studies perception; logic is a conceptual, not a scientific, investigation.)

Query: rabbit, duck, different view points, theory.

The duck-rabbit image can be used as a metaphor for two different points of view (or frames of reference): they may begin with the exact same data -- although whether or not they do begin that way cannot be judged within either frame of reference (duck or rabbit) nor is there of course an absolute frame of reference, but only by some third frame of reference or other -- but see entirely different things, make entirely different things of it.

But is the image duck-rabbit also a metaphor for two contrasting scientific theories, e.g. the geo- and helio-centric models of the solar system: different pictures, but exactly the same data? That is a vague, carelessly suggested notion, as the following shows:

Query: geocentric, heliocentric, percepts, concepts.

That does not apply in the case of these two theories, because it is not a matter of different Gestalts [Gestalten] (as with the duck-rabbit). Because although we organize the facts differently using these two models, we neither perceive nor conceive the facts we see (in the night sky) differently. Whereas with the duck-rabbit, the percepts are the same [the single drawing, the fact] but the concepts ['duck', 'rabbit'] are different. A model is not a concept in the sense of the word 'concept' that contrasts with 'percept'. (We use the word 'concept' many different ways, some quite vague, as when we call a scientific model a concept.)

There is also Gestalt shift, not as a metaphor, but as a picture that can be used to give a sense to 'Time is not real' (although it is a fantasy picture).

If I say, as I have, that "To understand Wittgenstein's logic of language is to experience a universal Gestalt shift", that shift is only in the domain of Logic. Compare the motto of the Philosophical Investigations:

It is the nature of every advance that it appears much greater than it actually is. -- Nestroy. (See Norman Malcolm's Ludwig Wittgenstein: a Memoir (1958), p. 60; Nestroy's words have been translated in various ways; however, M. O'C. Drury (DW p. 113-4) used this translation)

And indeed neither the Philosophical Investigations nor the rest of Wittgenstein's post-TLP work in any way responds to the eternal philosophical question of (in Plato's words) "we are discussing no small matter, but how to live" (although it may point away from false paths in the study of Ethics, as well as in Metaphysics).

Photograph Gestalt shift

Photograph example of a Gestalt shift, 20 KB

Photograph by E.M. Axelsson

We can't -- i.e. there is no criterion by which to -- say of the duck-rabbit that it is either really a duck or really a rabbit. But if we knew that the one who did the drawing intended to draw a duck, would that change this?

But we know that this photograph is in fact a photograph of a cat and not of a rat. So, then, is the rat-cat image not really an example of a Gestalt shift?

Site copyright © September 1998. Please send corrections and criticism of this page to Robert Wesley Angelo. Last revised: 10 October 2016 : 2016-10-10

The URL of this Web page:

Back to top of page

Wittgenstein's Logic of Language | Bibliography | Search | Site Map