How very interesting. The second study is showing no 1back interference on the pointing task. Hawhat?
Details after the bump.
Ok, here’s the basic design:
1. Study phase (subjects learn the locations of targets and point to them)
2. Retention interval / 1back task
3. Blindfolded-pointing (stay)
4. Sighted-pointing (baseline)
5. Blindfolded guided rotation (& 1back task)
6. Blindfolded-pointing (rotate)
Subjects are assigned to two groups: spatial & verbal. Everyone does 10 test configurations (after training), half of which are control (no 1back task), half entail the respective 1back dual-task component.
The two 1backs (verbal & spatial) are exactly the same in terms of stimuli, base-rate, and pacing. Continuous response nBack, based on 9 audio stimuli: 3 numbers from 3 different locations. For the verbal 1back, subjects attend to the number. During the spatial 1back, subjects attend to the location they hear the cue from.
First the 1Back numbers:
Not surprisingly, asking subjects to track the objects (as they rotate) and do the 1back is harder than just doing the 1back. Interesting, there is no significant difference between the spatial and verbal 1backs.
I was anticipating some interference for the spatial 1back since the subjects were already engaged in spatial processing.
But what about the other interference direction?
Pointing numbers:
Fortunately, across the board, baseline=stay, as it should have been (roughly). For both latency and error, baseline=stay < rotate.
Interestingly, control=verbal=spatial. WTF? No spatial interference to be found anywhere.
It looks as though the subjects are all engaging in offline updating (if updating is actually a real phenom), despite the subject self-reports (as accurate as those can ever be).
I suspect that by using 4 target configurations, I am already exceeding the capacity of most subjects. If I were to drop down to 2 or 3 target configurations would I be able to see the interference predicted because participants would be more likely to engage in online updating?
Or is it that the 1back just isn’t difficult enough. At 95% accuracy (stay), there is certainly room for disimprovement ( 
). Unfortunately, the 2back is tough (90% average) - and likely too hard for a dual-task scenario. If the secondary task is too hard, the subjects will certainly fall back to an offline updating strategy.
Very interesting. Still debating what to do next.
