So the previous pointing distributions were short on any statistics. What happens when the numbers are tallied and cranks are turned?
First up: standard orthogonal bias (front, left, back, and right). How do they compare to each other? Usually, for pointing time & errors we see that forward is faster and more accurate than back which is faster and more accurate than left and right.
Well, as a reminder, we didn’t match that for latency
But are in the ballpark for error (sans left=backward)
But the biases are interesting as well.. These are separated by load (Control/WM-task) and position (Stay/Rotate). What’s plotted is the delta between observed and expected (0 would be no bias)
Not surprisingly, when subjects don’t move, there is little pointing bias (and this holds across the board). When they do move, it sky-rockets. So, what kind of significances do we get? RMANOVA 2×2x4 (load x position x direction) yields the following significances:
- load (p<0.02) WM-task shows least bias
- position (p<0.000) stay least bias
- direction (p<0.001) forward shows the greatest bias
- and a smattering of interactions that I wont get into for now.
Most interesting in this is that a) doing the 1back actually decreases ones pointing biases relative to control (wtf?) b) actual type of 1back has no effect (spatial or verbal). Weird.
Well, what about cardinal biases? These would be the biasing of the four cardinal directions (above) relative to the other directions. Here I’m just summing the deltas from above and subtracting the sum of the 45s. (i.e. sum(d0,d90,d180,d270) - sum(d45,d135,d225,d315)).
Here’s the gist:
- load (p<0.02) - same as above (C: 19 v WM:14, stderr 1.3)
- position (p<0.000) - same as above (S:6 v R:28, stderr 1.3)
- load x position (p<0.04) ..
So, after rotation, there is a significantly greater cardinal bias - but condition is irrelevant. Again, somehow, doing the 1back reduces the bias.
What the hell is going on? My brain and eyes are killing me right now, so I’ll leave this question open until I can think about it later. Weird.
