Tuesday 19 October 2010

The Ames Room and the Bower Bird

A recent study (Endler et al, 2010) has shown that bowerbirds take advantage of forced perspective (the Ames Room effect) in the construction of their bowers. This study has a few interesting things to say on the topic of the origin of the Ames Room effect and the likelihood of equivalent configurations popping up by chance. (This article got a lot of popular press when it came out: Ed Yong at Not Exactly Rocket Science covered it best here.)

Bowerbirds construct bowers to demonstrate their genetic fitness to potential mates. These bowers are structures which contain long entrance avenues, and females therefore enter from one direction. Male bowerbirds take advantage of this and create a forced perspective illusion along the avenue by 'meticulously manipulating' the size and placement of stones along the surface. Specifically, large stones are placed at the back, and smaller stones placed towards the front. The net result is a static texture gradient in which each element subtends the same visual angle, instead of the usual circumstance in which the visual angle shrinks with distance (see figure).
From Endler et al (2010). The top panel shows how visual angle usually varies with distance, while the bottom panel shows how the visual angle can be kept constant if object sizes are carefully arranged in size order
An object placed on this carefully constructed texture gradient will subtend a visual angle that, relative to the gradient, suggests an object that is larger than it is (for a given distance). This information is contained within the motion-independent structure, and while it will conflict with all the other information available, it will still contribute to the overall perception of the scene; exactly as in the Ames Room scenario. The net result is an avenue in which, from the forced perspective from the single entrance, you would misperceive the size of, say, a male bowerbird as being larger than he actually is; hello, ladies!

Bowerbirds work hard to explicitly create this effect; when Endler et al (2010) moved the stones to reverse the intended gradient (placing large stones at the front) the male bird spent a couple of days reversing the reversal to restore the illusion. The gradient is also constructed with specific reference to the location of the entrance; the male birds understand to construct their gradients for a specific viewpoint. Male bowerbirds also vary in their skill at creating these effects, with consequences for their chances of mating, which suggests this behaviour has been selected for (rather than being a spandrel, or side effect of another selected behaviour).

The authors are unsure whether birds will have the same susceptibility as people to this illusion, but there is no reason to think they won't. In fact, bowerbirds have eyes mounted on the side, which provides 360° vision but little overlap in the visual fields, meaning little or no binocular vision. Their monocular perspective on the scene may make them less able than humans to resolve the conflict and more dependant on motion-independent information. Monocular vision still has access to motion-based information, but is simply less stable than binocular vision for certain tasks such as perceiving size and distance.

Some Conclusions
It is not clear how much cognition is needed to create the size-distance gradients. Males might explicitly place small objects close to the avenue entrances and larger objects farther away. This could result from complex cognition, be created through a simple inherited decision rule, or be the result of trial and error. Video recordings show that males spend 78% +/- 15% of their time at their bowers on bower ‘‘maintenance,’’ which involves moving gesso and other objects on their courts and avenue. They frequently show a cycle of looking at the court from the female’s avenue viewpoint and then moving court objects, and this is repeated throughout the day. Forced perspective could therefore be created by trial and error, in which the male moves the objects until the court looks ‘‘good’’ from the avenue.
Endler et al, 2010, p. 1683
It's quite clear what's happening: the bowerbird is placing stones, then going to look to see how it appears from the relevant viewing location, then going in and tweaking, until there's no more need to tweak. They are perceiving, and acting; actually, they are doing exactly what you or I would do faced with the same task. It also speaks to how such an ambiguous, 'equivalent configuration' is not all that likely by chance: it requires very specific intervention to occur and maintain, namely the kind of careful reverse engineering of a configuration of objects from a specific optical circumstance.

I find this quite fascinating, given the traditional accounts for the Ames Room illusion in humans. Essentially, the traditional account notes that there is an ambiguity in the proximal stimulus. There are two equivalent configurations in the world that lead to the same proximal stimulus, and yet we happily see just one (the square; in fact, we can't not see it, from the correct viewpoint). Traditional accounts suggest we have resolved the ambiguity with non-perceptual resources - specifically we have enriched our perceptual experience with assumptions about the common shape of rooms, etc. This is used as an example of what is supposed to be the normal state of affairs, namely ambiguity in the proximal stimulus. Runeson's analysis suggested instead that we are correctly perceiving what the (motion-independent) perceptual information contained in the Ames Room setup specifies, namely a square room, and the reason we can't detect the ambiguity is that our smart perceptual systems did not develop to handle the extraordinarily unlikely possibility of such an equivalent configuration. It could do this because the improbability isn't just luck, it's a fact of the 'prevailing physical and ecological constraints’ and smartness is all about taking full advantage of such constraints.

So what about the bowerbird? I find it unlikely that any psychologist would like to claim they have extensive experience of carpentered environments, the artificial environments in which squared off regions exist. Nature doesn't have a lot of squares in it; it does, however, have 'prevailing physical and ecological constraints’ that leads to texture gradients, reliable variation in visual angle with distance, etc. Smart visual systems attune to information, they don't make assumptions. There's no reason to invoke complex cognition to birds to explain this behaviour*, so it's mere hubris to invoke it for the same behaviour in humans.

All that said, it does raise the question of what counts as cognitive behaviour in another setting. As my partner in crime has been posting, it's not actually clear that the cognitive/non-cognitive distinction is buying us anything all that useful, and it certainly doesn't seem able to help us decide boundary cases like this one. In fact, I've been wondering recently what animal behaviour means for theories of cognition: are animals more cognitive than we tend to give them credit? Are we less? Is their behaviour non-cognitive, but similar behaviour in us cognitive? Is, in fact, the question not all that useful?
*Depressingly this didn't stop the authors of this paper speculating about how bowerbirds learned linear perspective before people did in the 1500s. I'm inclined to believe our visual systems beat out art on this one.

Endler, J., Endler, L., & Doerr, N. (2010). Great Bowerbirds Create Theaters with Forced Perspective When Seen by Their Audience Current Biology, 20 (18), 1679-1684 DOI: 10.1016/j.cub.2010.08.033

No comments:

Post a Comment