The Shrunken Finger Illusion

Ed Yong has a great write-up of an interesting little study in Current Biology (Ekroll, Sayim, Vander Hallen & Wagemans, 2016) that caught my eye. The study reports an illusion (the 'shrunken finger illusion') that shows how amodal volume completion can make you feel like your finger has shrunk, and everyone is very excited about how this shows our experience of the hidden back-sides of objects is "real".

In this post, I'll review the results, do a little ecological finger wagging about the breathless write-up (Ekroll's, not Ed's) and think about some studies the ecological reframing of the effect might motivate. Briefly, I think this effect is definitely real, and that we really do genuinely perceive hidden objects under certain circumstances. Of course, this has nothing to do with amodal mental representations of what we think is there and everything to do with the information the system is interacting with, but you know that of course because this is always the answer!
The Shrunken Finger Illusion
The set up is simple, and you should work along at home. Take one opaque, hemispherical plastic shell (say, half a ping pong ball) and balance it on the tip of your finger so that you cannot see that the ball is missing it's back half. Your line of sight should show you a sphere (see the figure from the paper below)

Figure 1. The set up and basic result

Now take a pointer (they used a 3mm knitting needle) and point to where 'it feels like the tip of your fingertip is'. People made errors as if their fingers were shorter, and the magnitude of the errors grew as the diameter of the sphere increased. The data (see Figure 2) suggest that people perceive that the hemisphere is actually a complete sphere, and they perceive that they are touching the far outside of a complete sphere, not the near inside of the incomplete sphere.

Figure 2. Data from the main experiment. The solid horizontal line at 0 is where the fingertip actually is; the solid diagonal line is where the bottom of the sphere would be if the ball was complete; the dashed diagonal line is where the bottom of a complete hemisphere would be if it were there

We read this paper in lab meeting and collected a little data of our own; see Figure 3.

Figure 3. An n=2 replication of the shrunken finger illusion

A few things to note:

The solid diagonal line is where the finger would be if it was actually sitting on the outside surface of the visually specified sphere. The data are nowhere near this; the gain (the magnitude of the illusion as a proportion of the full possible magnitude) was only 22%. I think the effect is small because viewing conditions were not carefully controlled; people just placed the spheres on their fingers, they could easily have it misaligned a little, etc. The low gain may suggest people are perceiving a different shape, perhaps the solid hemisphere indicated by the dashed line. The authors rule this out by noting that no one reported such a percept (and if all they can see is the spherical part, there's no information driving perception of a hemisphere). That said, the slope of the judgment data more closely matches the slope of of the dotted line than the solid line, so the way in which judgments were scaling with object size do not actually track a sphere. This definitely needs investigating.

The 'wait' conditions refer to the condition where people were asked to wait until they were definitely experiencing a complete sphere when looking at the ball; this did significantly raise the gain. This gets to the fact that perception takes time and has a dynamic (i.e. it develops in particular ways over time and will have some persistence to fluctuations in access to information). Lots of interesting questions about how this develops over time.

A Little Ecological Finger Wagging

This observation provides strong evidence for the controversial and counterintuitive idea that our experience of the hidden backsides of objects is shaped by genuine perceptual representations rather than mere cognitive guesswork
...
the shrunken finger illusion goes further by demonstrating that purely subjective mental representations of the invisible backsides of things can have comparable causal powers: although amodally volume-completed percepts refer to the hidden backsides of objects, our findings show that they are ‘‘real’’ in the sense that they can—even against better knowledge—affect the experience of our own body in a dramatic and almost bizarre way.

To be a little blunt, this is the kind of surprise you only experience if you are a cognitive psychologist who thinks 'knowledge' is the name of the game. If you think perception requires knowledge to fill in the blanks and make the whole thing work, you have to answer the question, 'where did that knowledge come from, if not from perception?'. There is no answer to this question, and what that means is that of course cognitive guesswork isn't playing a role. It can't, and so actually this set up of 'perceptual representations vs cognitive knowledge' is a red herring being eaten by a straw man.

An underlying assumption here is that perception is derived from sensations. There are currently no sensations generated by the occluded back of the object, therefore any experience of the back of that object must be created as a 'purely subjective mental representation'. Perception, of course, has nothing to do with sensations, and is instead driven by information. This means you don't need rays of light from a surface to be hitting your eye for you to still be perceiving that surface (see Gibson on the informational basis governing the perception of dynamic occlusion events for his best example on this; Gibson et al, 1969).

This gets to my next wag of the finger, at the idea that this little illusion study is strong evidence for anything. I like the study, but if you want strong evidence for the causal role the perception of occluded surfaces can play in our experience, see Gibson et al (1969), and more recently related shape perception work from Bingham (e.g. Bingham & Lind, 2008; Lee & Bingham, 2010; Lee, Lind, Bingham & Bingham, 2012). These papers are the most recent in Geoff's 15 year programme investigating the informational basis of metric shape perception (how we are able to perceive actual sizes of things when the information is only about relative sizes). He finally cracked it after years of work by showing that veridical shape perception only becomes possible with large (greater than 45°) changes in perspective. In other words, you calibrate the angular visual information about shape using other information gathered as you, for example, walk into a room. So dear cognitive psychology, make sure you check the literature before getting super excited about your study with ping pong balls and knitting needles.

I don't want to be too snippy here. The point is that the results of this paper are only sexy and amazing and worthy of publication in Current Biology under this inaccurate framing about perception vs knowledge. From the perspective that perception is information based, not sensation based, and with the understanding that the information you are presenting the participant with visually specifies a complete sphere, the surprise about the recalibration of perceived finger length goes away.

An ecological reframing
This paper demonstrates an effect. The question is, what is the mechanism that is producing that effect? Behavioural mechanisms always involve perceptual information, and so experiments investigating the mechanism should begin by manipulating that information. 

• The effect takes a little time to develop; this is, I think, because people get to see the actual shape of the ball while they rest it on their finger, etc. Can you increase the effect by getting people to do the task while never seeing the fact the ball is cut in half? 
• The slope of the judgment data actually matches the expected slope if people were completing to a solid hemisphere; is this an artifact of the fact people have to rest the ball on their finger and spheres don't balance on fingers this easily? In other words, is information about the behaviour of the sphere as it balances being perceived and affecting the judgments?
• What about other shapes? Spheres are geometrically useful objects to test but does the illusion persist for other objects? What about the slope of the judgments?
• People don't perceive the sizes and shapes of things just for fun. We perceive these things in order to coordinate and control our actions with respect to those objects. This is why action measures of perception are better than judgments (Bingham & Pagano, 1998) although judgments are a perfectly acceptable place to start sometimes. Can we find evidence of the shrunken-finger using action measures, such as reaching-to-grasp?

Summary

This is a fun effect that I agree is definitely perceptual (just not representational!), and I'm looking forward to trying some of this out next year with my student RAs. There's definitely some simple extensions of the task that would be interesting to see the data on. 

References
Bingham, G. P., & Lind, M. (2008). Large continuous perspective transformations are necessary and sufficient for accurate perception of metric shape. Perception & Psychophysics, 70(3), 524-540.

Bingham, G.P. & Pagano, C.C. (1998). The necessity of a perception/action approach to definite distance perception: Monocular distance perception to guide reaching. Journal of Experimental Psychology: Human Perception and Performance, 24 , 145-168. 

Lee, Y. L., & Bingham, G. P. (2010). Large perspective changes yield perception of metric shape that allows accurate feedforward reaches-to-grasp and it persists after the optic flow has stopped!. Experimental Brain Research, 204(4), 559-573.

Lee, Y. L., Lind, M., Bingham, N., & Bingham, G. P. (2012). Object recognition using metric shape. Vision Research, 69, 23-31.

Ekroll, V., Sayim, B., Van der Hallen, R., & Wagemans, J. (2016). Illusory Visual Completion of an Object’s Invisible Backside Can Make Your Finger Feel Shorter. Current Biology, 26, 1-5.

Gibson, J. J., Kaplan, G. A., Reynolds, H. N., & Wheeler, K. (1969). The change from visible to invisible. Perception & Psychophysics, 5(2), 113-116.