A common phenom in the spatial reasoning literature is that there is a strong bias in spatial memory towards the cardinal directions (egocentrically, that is).
Do the current data support tha phenom?
The usual gross version of this is to look at latency and error as a function of the egocentric cardinal directions:
Ok, so pointing forward is somewhat faster than left, right or back, but definitely more accurate.
Can we get a better picture? Of course.. Let’s bin @ 45 degree increments (centered)
Latency is still marginal - but look at that pointing error. It’s the much loved jigsaw pattern that Mou&McNamara use to support the argument that there are intrinsic alignments. Well, these are egocentric pointing responses, which make me doubt that conclusion (that and the subjects studied the targets from the center of the configurations).
Simply put, point along the cardinal directions is more accurate. But what about the biases?
Let’s look at the distribution of targets in space, both real and reported:
Yoinks! Take a look at that. Now there’s a bias. Subjects are misrepresenting targets towards the cardinal directions, boy are they ever. And this holds across all subjects without much of any variation.
Allocentric explanations for this bias would have to resort to either response biases (easier to point along cardinals - which is certainly true physically, is much less a valid point when using a joystick that is potentiometer based (no haptically salient axes)) or egocentric mediation.
I will concede that with the discovery of grid cells, pure egocentric representations are looking less likely - but behaviorally, there is so little evidence for anything remotely like the metric allocentric networks proposed. I’m thinking Sholl’s egocentric mediation of allocentric networks is looking much more attractive.
