Experiment #1 has been in the can since December. Now I’m up to my eyeballs in #2. Here’s the latest.

So I submitted a manuscript to Cogsci 2007 on the first experiment. The basic gist was that the alignment effect is influenced by set-size, but only for latency, and that this extends even to the case when the alignment is reversed after rotation.

Specifically, as the number of targets increases, the benefit for aligned judgments decreases for latency. Egocentric theories would have predicted the same effect for pointing error - but it’s not there. While this might just be a serial search effect, that would be a main effect and not an interaction.

Anyway.. on to the next experiment.

At the recommendation of Walt, I’ve taken a different tac. If spatial reasoning is dependent upon egocentric representations, then there must be a capacity limitation in the processing. The first study took the obvious route of increasing the target set sizes and met with moderate results. Walt’s recommendation was to attack it from a dual-task, WM manipulation perspective.

The new version has people engaging in egocentric spatial judgments (pointing to targets around them before and after rotation) within a slew of four target configurations, but also mixes in a 1back task. Two versions were generated: a verbal 1back and a spatial 1back. Both use the same auditory stimuli: three different numbers from three different locations (left, right, center). In the verbal 1back, subjects track the number; In the spatial, they track the location. Both 1back’s are continuous in that they respond to every cue presented.

The question here is how does the 1back affect pointing responses before and after rotation. The idea is that the 1back will fill up WM with extra crap. If verbal and spatial are separate entities, then their impacts on performance should be different.

Early numbers are showing that again, only latency is effected..

In other news, jACT-R upgrades are coming along nicely. The new module framework absolutely flies. I’m getting 100x real time (with cycle-skipping enabled, 10x w/o), enable module specific threading and it pretty much doubles. Very sweet.

The discovery of MINA has allowed me to gut my spaghetti networking code (a personal weakness of mine). The remote debugging now screams and can easily be adapted to new network topologies and protocols.

The upgrade of ACT-R/S is almost done and once I release the new IDE tools, I’ll release it as well.