Waterfall illusion, or motion aftereffect, is an illusion of movement. It is experienced after watching a stimulus moving in one direction for some time, and then looking at a stationary scene. The stationary scene appears to have movement (in the opposite direction to the moving stimulus that one previously watched). This is called the “waterfall illusion”, as it can be experienced after watching the motion of the water in a waterfall, and then attending to a stationary scene, for example the rocks by the side of the waterfall. Robert Addams popularised this illusion in 1834 after a trip to the Falls of Foyers in Scotland with his florid writing:
“Having steadfastly looked for a few seconds at a particular part of the cascade, admiring the confluence and decussation of the currents forming the liquid drapery of waters, and then suddenly directed my eyes to the left to observe the vertical face of the sombre age-worn rocks immediately contiguous to the waterfall, I saw the rocky face as if in motion upwards, and with an apparent velocity equal to that of the descending water.” (1834, p. 373)
Illusions of this sort were known much before 19th century. In fact, Greek philosopher Aristotle (384 – 322 BC) reported such illusions more than 2000 years before Addams: “when persons turn away from looking at objects in motion, e.g., rivers, and especially those which flow very rapidly, they find that the visual stimulations still present themselves, for the things really at rest are then seen moving.” (Aristotle, citeed in Ross, 1931, p. 459b). The use of a spinning spiral to induce the effect can be traced back to the Belgian physicist Joseph Plateau in 1849.
Aristotle also noted, correctly, that the speed of the inducing motion affects the speed of the illusory motion experienced afterwards. This has now been verified experimentally by Wright and Johnson (1985). Deas et al. (2008) found that there was an auditory version of this illusion that also exhibited the same dependence of the experienced auditory motion on the perceived inducing motion. Surprisingly, Berger and Ehrsson (2016) found that the visual illusion can be induced cross-modally by auditory stimuli.
The physiological explanation of this illusion involves neurons becoming less sensitive at various sites through out the brain. This sometimes occurs because neurons become fatigued (so they change what is called their ‘response gain’). But it can also happen because neurons change their sensitivity (or ‘contrast gain’) to a stimulus. The difference in motion between two things is the ‘contrast’. And neurons can change what sorts of contrast they are more or less sensitive to. See Boynton (2005) for an excellent explanation of contrast gain. And see Kohn & Movshon (2003) for work on this topic on the waterfall illusion. According to this explanation, when you are watching the stimulus with motion (for example, the moving water in a waterfall), the neurons that detect continuous movement in one direction (e.g., downward) become less sensitive to motion at that speed in that direction. As a result, when you look away, neurons that detect movement in the opposite direction (e.g., upwards) are more active in comparison. This results in the appearance of the stationary object moving in the latter direction (upwards). It is thought that many properties that we experience are encoded in this way in the brain: by a comparison between the firing rates of different populations of neurons, rather than the particular rate of each.
The Waterfall Illusion is philosophically interesting for a number of reasons. First, as with many other visual illusions, there is the question as to why we experience a stationary figure as moving despite, in many instances, knowing that it is stationary. Those who believe that the mind is “modular” will cite illusions like the Waterfall Illusion to support their thesis. To explain: on the hypothesis that the mind is modular, a mental module is a kind of semi-independent department of the mind which deals with particular types of inputs, and gives particular types of outputs, and whose inner workings are not accessible to the conscious awareness of the person – all one can get access to are the relevant outputs. So, in the case of the Waterfall Illusion, a standard way of explaining why experience of the illusion persists even though one knows that one is experiencing an illusion is that the module, or modules, which constitute the visual system are ‘cognitively impenetrable’ to some degree – i.e. their inner workings and outputs cannot be influenced by conscious awareness. For a general discussion of cognitive penetration, see Macpherson (2012).
Philosophers have also been interested in what illusions like the Waterfall Illusion can tell us about the nature of experience. For example, in the case of experiencing the Waterfall Illusion, it would seem to be that one can know that the objects in the latter scene are stationary whilst at the same time one experiences them as moving. If so, then this might count against the claim the perceptual states are belief-like, because if perceptual states were belief like then, when experiencing the Waterfall Illusion one would simultaneously believe that the objects were, and were not, moving. This would seem to suggest that one was being irrational when experiencing the Waterfall Illusion (because one would simultaneously be holding contradictory beliefs, or belief-like states), which seems implausible – if one is experiencing a visual illusion, this is not obviously a case of irrationality. (For discussion of this general point about the theory that perceptions are like beliefs, see Crane & French 2016).
Perhaps the most interesting philosophical question that the Waterfall Illusion has raised is whether what the illusory experience presents is an impossible state of affairs or not. In the passages we have quoted above from Aristotle and Addams the effect was simply described as involving movement expereinced in the oppositive direction to the previusly seen moving stimulus. There was no mention of the effect involving an experience of an impossible state of affairs. However, in the 1960s and 1970s some psychologists started to describe the illusion as involving experiencing movement yet at the same time experiencing that the things seen moving are not changin location. For example, Frisby says, “although the after-effect gives a very clear illusion of movement, the apparently moving features nevertheless seem to stay still! That is, we are still aware of features remaining in their 'proper' locations even though they are seen as moving. What we see is logically impossible!” (Frisby, 1979, p. 101). Likewise, contemporary philosopher of mind Tim Crane interprets the Waterfall Illusion as involving the illusory experience of an impossible state of affairs (1988).
Whether this is right is a particularly interesting question, for if it is, then it may provide a troubling case for the sense-data theory of perception. According to the sense-data theory, in veridical perception, illusion and hallucination, one is directly aware of some mental object (a sense-datum) that has the properties it appears to have - and in vitue of so doing, when the right conditions obtain for percpetion, one can come to see the external world indirectly in virtue of directly seeing sense-data. This theory nicely explains appearances in the illusory and hallucinatory case. The sense-data theory is committed to the “phenomenal principle”: if it sensibly appears to a subject S that there is something which has a sensible quality F, then there is an object which has F that S directly perceives (Robinson 1994). For further discussion, again, see Crane & French (2016). Now, in the Waterfall Illusion, if an object appears to be both moving and not moving at the same time, then it appears to have an impossible property (the property of moving and not-moving at the same time). Given that there cannot be objects with impossible properties, then there cannot be such sense data - and so the sense-data cannot explain what our experience is like.
Recent psychological evidence suggests that there is a change in the perceived position of a stimulus perceived whilst undergoing the motion aftereffect. See for example, Snowden (1998), Nishida and Johnston (1999), and McGraw et al. (2004). Snowdon (1998) notes that the amount of displacement depends on the speed on the inducer, which matches nicely with the observation that the speed of the illusory movement depends on the speed on the inducer. However, although this is suggestive that things are seen as both moving and changing position, it is not conclusive. Strictly speaking, it only shows that things are experienced as being not in the position that they actually are.
In our opinion, the question of what it is like to undergo the Waterfall Illusion is still not settled. It could involve simply experiencing things moving in the opposite direction of the stimulus and changing position. It could involve experiencing things moving in the opposite direction of the stimulus and yet not changing position. Or it could involve something more complex. For example, it could involve experiencing things moving and changing position outside of the centre of the visual field but as not moving at the centre. Or it could involve experiencing things moving and changing position, but then jumping back into the original position again before changing position again. More research is required in order to settle this question. If you would like to participate in our research please take our Waterfall Illusion Survey.