Ed Yong recently blogged about this paper testing newborn preferences for biological motion:

Simion, F., Regolin, L., Bulf, H. (2008). A predisposition for biological motion in the newborn baby. Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0707021105

My university doesn't have access to the journal, but Ed was nice enough to email me a copy of the article.

I can't remember if I've blogged about biological motion before, but basically it's a not-very-well-defined concept usually referring to point-light displays such as this one, which affix lights or other sensors to the joints of animals that exhibit semi-rigid motion (usually humans). Humans are very fast at identifying common motions (such as throwing, golfing, rowing) from such point-light displays, and have shown an ability to discriminate individuals, even though the person may only be represented by 10-15 dots of light.

I had done some work with biological motion for a class project on development, and had read the previous study referenced that showed a preference for biological motion in newly-hatched chicks. Chicks would even move toward motion displays of cats as opposed to random light patterns with the same number of light points.

Some work had been done with infants, and actually a study by Fox and McDaniel with 2-month-old babies had not found a difference in infants' preferences. I was kind of shocked to read about this new study, since the infants were between 1 and 3 days old. Of course, they're trying to determine whether a particular attribute is innate, but I'd never heard of a study using infants that young.

The methodology involved setting up two adjacent 19-inch computer screens on which to display the the two types of stimuli (random and biological), with a flashing red light between them to initially draw the infant's attention. To determine the infant's preference, they measured the amount of time the infant spent looking at each display. They randomized and balanced the left-right display of each set of patterns. Infants sat on a naive (meaning they didn't know the expected outcome of the experiment) experimenter's lap 30cm from the displays.

I was initially very skeptical, because my understanding is that 1) Newborn infants have very poor vision, and 2) Newborn infants have very underdeveloped neck muscles and are unable to even support their head in an upright posture.

However, after poking around a bit on-line, it seems that even though infants have immature vision, they are primarily near-sighted at birth, and can see things pretty clearly in the 8-15 inches range (the 30cm range used in the experiments is about 12 inches). Also, the infants did not need to move their heads. Because the displays were so close, they were simply able to move their eyes to look at either display.

The authors note that one reason Fox and McDaniel didn't find results was due to the short duration of their trials (~15s), while Simion et al. used durations up to and over a minute. This sounds plausible, assuming all the controls mentioned were strictly adhered to.

And not only did the infants show a preference for biological over random motion, but they showed a preference for displays of biological motion that were right-side-up over those that were upside-down. So I think they did a reasonable job of dispelling my initial skepticism and providing evidence for innate mechanisms for the detection and preferential devotion of attentional resources to semi-rigid motion.

Since the visual cortex is still very underdeveloped at birth, and the authors point out that this behavior is phylogenetically remote (which just means that it probably originated in a common ancestor a long time ago) because it is seen in a wide variety of animals, this points to the mechanism being subcortical, in the much older parts of our brain.

A couple of follow-up studies that I'd be interested in seeing:

1) As the authors point out, they used three classes of stimuli: random motion, right-side-up biological motion, and upside-down biological motion. They say no tests have been done with rigid object motion (e.g., point lights attached to the corners of a cube, which is then rotated). The mechanism that is detecting biological motion may not strictly be a biological motion detector. If they're comparing it with random motion, where there is no meaningful information to be extracted, it could just be a general pattern detector, whose function is to determine whether or not any meaningful information can be extracted from the display. My guess is that displays of rotating rigid bodies would be preferentially selected for over random displays, but not as preferred as biological motion.

2) Blindsight is a condition where a patient has the subjective experience of being blind, but in actuality they are able to unconsciously process visual data. The visual cortex is basically our new visual processing center, built on top of an older system that took care of our vision. The patient's eyes and optic nerve might be fine, but they say they can't see, probably because of damage to the primary visual cortex. However, there are still pathways to the extrastriate cortex and subcortical structures, and it is unclear exactly how the processing is being done, but even without the subjective experience of seeing, the subjects are still able to "see" when presented with visual stimuli and forced to choose where they think it might be. Basically I'd like to see a similar study done with patients exhibiting blindsight, to see if they are able to unconsciously process biological motion.

Anyway, interesting stuff. If you have questions or comments, fire away.