Your complimentary articles

You’ve read one of your four complimentary articles for this month.

You can read four articles free per month. To have complete access to the thousands of philosophy articles on this site, please

SUBSCRIBE NOW

David Hume

Hume’s Image Problem

Marc Bobro scrutinizes how Hume thinks about thought.

David Hume believed that the mind represents the world by having contents that resemble it, such as having images of it. This way of thinking about thinking has been called imagism. But, as Bertrand Russell and his friend Ludwig Wittgenstein noted much later, the same images can resemble different things, and may be variously interpreted. Such potential image ambiguity could be a serious problem for Hume. My aim here is not to defend imagism as a theory, but to determine how, if at all, Hume could defend his imagism in response to this potential ambiguity.

Hume’s Imagism

Hume is what is called a mental representationalist. Like most of the philosophers of his era, Hume thinks that all perception is essentially representational. This is not to say that Hume thinks external objects are necessarily just like our perception of them. It is generally agreed among Hume scholars that according to Hume no perceptions represent external objects as they are in themselves, since even the senses “convey to us nothing but a single perception, and never give us the least intimation of anything beyond” (Treatise on Human Nature, p.189) and moreover, “nothing is ever really present to the mind, besides its own perceptions” (p.197, Oxford Clarendon edition, 1978). Yet for representation to take place, some form of resemblance between that which represents and that which is represented is required. For Hume, perceptions are fundamentally and essentially imagistic; so for Hume, resemblance is to be understood in terms of images (or as Hume sometimes says, copies). Thus the representational power in the mind is in the imagination, not in the intellect.

For Hume, every experience is either an impression or an idea. Impressions are the sorts of contents of the mind that are short-lived, and constantly succeeding one another, including sensations, passions, and emotions. They’re created by the mind seizing a passing moment and fixing it in images. Ideas have a special relationship to impressions to Hume: ideas are impressions reproduced in the mind so that we can think about objects in their absence. And since ideas are reproduced impressions, and impressions are imagistic, so too must ideas be imagistic.

The Problem of Image Ambiguity

Today, imagism attracts few followers. Philosophers from Frege to Wittgenstein to Jerry Fodor have presented what many perceive to be devastating objections to theories which identify thought with images. However, I will use only Wittgenstein’s objection here.

Wittgenstein does not deny that the mind’s eye can view mental images. But Wittgenstein does object to the claim that images on their own individuate concepts: that is, he rejects what I call representational determinacy – the idea that a mental image means one particular thing. Wittgenstein writes: “I see a picture: it represents an old man walking up a steep path leaning on a stick. How? Might it not have looked just the same if he had been sliding downhill in that position? Perhaps a Martian would describe the picture so.” (Philosophical Investigations, p.54).

If valid, Wittgenstein’s argument would undercut Hume’s imagism, since Hume apparently thinks that mental images determine what is being thought. Say I have a thought of an old man walking up a steep path, and that this thought consists of an image or set of images, as Hume claims. Can any image convey or determine such a thought unambiguously? Wittgenstein says no: single images can never unambiguously represent direction of movement. Suppose the image of the man walking included arrows meant to indicate direction of movement. Even this doesn’t serve to disambiguate the image, since arrows are also symbols that can be variously interpreted. Perhaps Martians would understand arrows in an entirely different way. So images seem to need something else to provide the interpretation, such as statements, propositions, or actions. To put Wittgenstein’s point another way: if I declare, “Let this image represent an old man sliding downhill!” the image itself need not change at all. It seems then that in order to function as the vehicle for thought, images require interpretation: by themselves they cannot determine the meaning of thoughts. It follows that even if God looks into my mind he would not see there exactly what I was thinking.

It doesn’t help to disambiguate one’s mental images by constructing a set of images of old men walking up steep paths. Not only can each one of these images can be interpreted in different ways, but the whole set together can be interpreted in different ways. Consider Man Ascending by Eadweard Muybridge:

The title notwithstanding, not only can each individual image here be interpreted either as a man ascending or as a man descending, it’s not obvious how one must order these images. Imagine what a Martian might say.

So, Wittgenstein’s objection is a powerful one: by attending to any one particular image or series of images, it seems there’s no way to determine the identity of what is supposedly being thought in those images, at least with regard to movement.

Rescue Attempts

It’s not clear that all forms of imagism are susceptible to Wittgenstein’s criticism. Let’s consider three attempts to rescue Hume’s theory from the problem of image-ambiguity.

Attempt 1. Hume’s understanding of the derivation of ideas from impressions has been compared to the creation of photographs: ideas are copies of impressions in the way that a photograph is a copy of its original, whether singly or in series. To take another example, my idea of a running horse is thought to derive from a series of contiguous impressions, the way chronophotography or a film is derived from its original.

The point pertinent to Wittgenstein’s objection is that photographic snapshots take time. Even the fastest shutter speed is not instantaneous. So if we were to analyze a photograph of something moving, shouldn’t we thus be able to tell both that it was moving, and in what direction?

Consider the famous ‘Case of the Cottingley Fairies’. Cousins Frances Griffiths, who was nine-years-old, and Elsie Wright, who was sixteen, said they used Frances’ father’s camera to take photographs of the fairies, like this:

Notwithstanding the obvious fraud in the Cottingley fairy case, the image remains ambiguous. Might not a Martian see a fuzzy but stationary ‘waterfall’ in the background here? There’s nothing in the image itself that unambiguously indicates even that the exposure took time. Even if a Martian knew that the image took time to be exposed, it still seems that the image alone couldn’t indicate whether the water was falling or climbing during the time of the photo.

Now imagine a photo of an old man as he climbs a steep path. Even if the exposure were obviously lengthy, say three seconds, we couldn’t tell whether the shutter initially opened to the hiker at the lower point on the hill or a higher one: we couldn’t tell whether the old man was walking uphill or sliding downhill. So, this attempt to save Hume’s imagism by analogy with photography fails to disambiguate images to determine thoughts.

Attempt 2. Representations can make the static dynamic, and a single image can represent more than it depicts. Consider Marcel Duchamp’s Nude Descending Stairs, No. 2 (1912).

Duchamp wrote about this painting: “My aim was a static representation of movement – a static composition of indications of various positions taken by a form in movement.” Or consider Giacomo Balla’s Dynamism of a Dog on a Leash (right), painted in the same year. In these Futurist paintings, a single image conveys or represents more than one moment in time. Temporal duration is collapsed or captured in one static image. Does this help Hume?

Sure, images can represent more than is seen instantaneously. Cubist paintings, for example, collapse several spatial perspectives into one. But this simply misses Wittgenstein’s point. Even supposing that Duchamp’s nude represents movement, even descent, it surely doesn’t follow that it exclusively represents movement or descent. The painting alone doesn’t force a particular interpretation on us. It fails to unambiguously represent even one and the same nude enduring through time. A human, much less a Martian, might legitimately see six or seven nudes. Or take the Balla painting. Can’t it equally represent a mutant terrier with forty legs and eight tails?

The issue here is not with representation in general, but with a particular form of representation – depiction. Depiction is ‘unambiguous representation by means of resemblance’. The fact that a particular image is supposed to represent movement and direction of movement says nothing whatsoever about whether that particular image actually succeeds in depicting movement and direction of movement for any given viewing. Attempt 2 fails.

Attempt 3. The failure of the first two rescue attempts distill down to the same thing: nothing instantaneous, such as an image, can depict (that is, unambiguously represent) anything non-instantaneous, such as an old man walking. But must images be instantaneous? Imagine dynamic mental images, playing like clips of mental movies. That is to say, suppose that our ‘mental images’ are actually changing images. The intrinsic properties of such ‘images’ could then apparently determine their interpretation, at least in regard to movement, for such images would actually move in the mind’s eye. We can call this the dynamic imagist theory of thought.

As we’ve seen, a static picture or image can’t on its own depict a dynamic event such as an old man walking up a steep path leaning on a stick. But why can’t a dynamic image? A dynamic image can unambiguously represent a dynamic event, at least with regard to its movement and direction. So dynamic imagism doesn’t seem susceptible to Wittgenstein’s objection. But does Hume think of ideas (or some ideas) as dynamic images? Is he even open to the possibility?

Could Hume Be A Dynamic Imagist?

Just as for Hume we cannot derive personal identity and causality from our impressions, he seems to also deny that we can derive ideas of time (duration) from single impressions, no matter how vivid the impression may be. He writes: “Whenever we have no successive perceptions, we have no notion of time… we may conclude that time cannot make its appearance to the mind, either alone or attended with a steady unchangeable object, but is always discovered by some perceivable succession of changeable objects” (Treatise, p.35). Yet if Hume did take single images to be possibly dynamic, that is, non-static, it seems that we could derive the ideas of time, as well as personal identity and causality, even from a single perception.

Could images for Hume be just dynamic enough to determine thought, but not changeable enough to provide us with ideas of time, identity, and causality? Probably not, since it’s hard to see how an image could endure just long enough for us to determine its interpretation, but be too temporally truncated for us to notice it changing, and thus derive the other ideas from it.

This idea is not only philosophically far-fetched, Hume gives no indication he approves such a position. Rather, he appears to reject it, arguing that “the opinion of a continued… existence never arises from the senses” (p.192), and all perceptions are “interrupted”, “broken”, and all impressions “perishing” and“perceptions… succeed each other with an inconceivable rapidity… in a perpetual flux and movement” (p.252).

So although in theory dynamic imagism seems to represent a possible response to Wittgenstein’s charge that images are representationally ambiguous, Hume cannot avail himself of it unless he’s willing to back off on his skeptical claims about the derivation of certain basic ideas from single impressions.

A Fresh Approach

We have seen that for Hume our thoughts are images, and that these images are static, but static images resemble far too many things and so may be variously interpreted, therefore, image ambiguity remains a serious problem for Hume. I want to finally question here whether it is indeed the case that for Hume images alone are supposed to fix their own meaning.

To do so, let’s first contrast Hume’s view with John Locke’s theory of general/abstract ideas. In An Essay Concerning Human Understanding, Locke writes: “Ideas become general by separating from them the circumstances of time and place, and any other ideas, that may determine them to this or that particular existence. By this way of abstraction [ideas] are made capable of representing more individuals than one; each of which [individual], having in it a conformity to that abstract idea, is (as we call it) of that sort” (§ III, 3, ¶6). Similarly for Hume, general ideas of triangles are images taken to represent all triangles. Hume writes, “abstract ideas are, therefore, in themselves, individual; however, they may become general in their representation. The image in the mind is only that of a particular object, though the application of it in our reasoning be the same, as if it were universal” (Treatise, p.20). He also says, “the general idea of a line, notwithstanding all our abstractions and refinements, has in its appearance in the mind a precise degree of quantity and quality; it may be made to represent others, which have different degrees of both” (Treatise, p.19). So even Hume’s imagism doesn’t presuppose unique image interpretation: images can be quantitatively and qualitatively determinate, yet representationally indeterminate, as they can be made by the mind to stand for either a particular or a general idea. So it appears that for Hume it is not essential that an image is of this [thought] and of nothing else. One and the same determinate image can represent both a particular object, say an individual triangle, and a universal, say the general idea of triangle.

But does this idea of the interpretability of ideas answer the problem of image ambiguity articulated so nicely by Wittgenstein? Superficially the answer might appear to be yes, as for Hume images can be made to represent different ideas. I’m afraid Wittgenstein’s objection stands, however, for a problem remains for Hume: neither the idea of an old man walking uphill nor that of an old man sliding downhill is a general idea.

One and the same image can for Hume represent both a particular object and a universal. But can one and the same image be made to represent more than one particular object? Hume seems to deny this possibility when he writes “The image in the mind is only that of a particular object, though the application of it in our reasoning be the same, as if it were universal.” This seems to imply that if one’s mental image is of an old man walking uphill, it can also be of the universal ‘old men who walk uphill’, but it cannot also be of a particular old man sliding downhill (or of that universal, either). So the mere fact that Hume admits some representational indeterminacy in mental images does not suffice to answer Wittgenstein’s objection, for Hume’s indeterminacy in the mind’s use of its images seems limited to the generation of general ideas. The problem of image ambiguity stands as a serious roadblock to the coherence of Hume’s theory of mental representation.

Marc Bobro teaches Philosophy at Santa Barbara City College.