correlation between errors and learning

[mobeets]

this should be better now that i have real behavior

[mobeets]

this is the correlation (all dates) between mean errors and cov errors, for the cloud-hab hypothesis

[mobeets]

using cursor progress:

[mobeets]

and now for cursor_progress:

[mobeets]

sorted within a day by the error by condition

positive trend means: the more learning (more negative), the less error! where "learning" is measured either by the improvement from beginning of perturbation (Lbest), or the performance hit compared to the mean of intuitive (Lmax)

and the two clumps? that's the two monkeys

[mobeets]

and now for the two cov errors

[mobeets]

(axes flipped so that more positive always means more learning)

rsq =

    0.5075    6.1823    0.0474    0.5056
    0.0312    0.1934    0.6755    1.0650
    0.0595    0.3796    0.5605    0.0838
    0.1102    0.7428    0.4219    0.6043
    0.5782    8.2258    0.0285    0.0796

where 1st col = rsq, 3rd col = p-val of F-test

and the trends are the same for habitual and cloud-hab

[mobeets]

from what i can see, 20120601 always appears to be somewhat of a renegade. it's always moving in the opposite direction of the other dates. this was true both for progress and for trial_index. though trial_index in general shows stronger trends, whatever the direction.

[mobeets]

here's cov orient error, trial_index, for cloud-hab